From c659b9f94fff29adfb2abe4f6b345710b65e8741 Mon Sep 17 00:00:00 2001 From: Hubert Chathi Date: Thu, 25 Jul 2019 11:08:24 -0400 Subject: allow uploading keys for cross-signing --- tests/handlers/test_e2e_keys.py | 63 +++++++++++++++++++++++++++++++++++++++++ 1 file changed, 63 insertions(+) (limited to 'tests') diff --git a/tests/handlers/test_e2e_keys.py b/tests/handlers/test_e2e_keys.py index 8dccc6826e..9ae4cb6ea2 100644 --- a/tests/handlers/test_e2e_keys.py +++ b/tests/handlers/test_e2e_keys.py @@ -1,5 +1,7 @@ # -*- coding: utf-8 -*- # Copyright 2016 OpenMarket Ltd +# Copyright 2019 New Vector Ltd +# Copyright 2019 The Matrix.org Foundation C.I.C. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -145,3 +147,64 @@ class E2eKeysHandlerTestCase(unittest.TestCase): "one_time_keys": {local_user: {device_id: {"alg1:k1": "key1"}}}, }, ) + + @defer.inlineCallbacks + def test_replace_master_key(self): + """uploading a new signing key should make the old signing key unavailable""" + local_user = "@boris:" + self.hs.hostname + keys1 = { + "master_key": { + # private key: 2lonYOM6xYKdEsO+6KrC766xBcHnYnim1x/4LFGF8B0 + "user_id": local_user, + "usage": ["master"], + "keys": { + "ed25519:nqOvzeuGWT/sRx3h7+MHoInYj3Uk2LD/unI9kDYcHwk": "nqOvzeuGWT/sRx3h7+MHoInYj3Uk2LD/unI9kDYcHwk" + }, + } + } + yield self.handler.upload_signing_keys_for_user(local_user, keys1) + + keys2 = { + "master_key": { + # private key: 4TL4AjRYwDVwD3pqQzcor+ez/euOB1/q78aTJ+czDNs + "user_id": local_user, + "usage": ["master"], + "keys": { + "ed25519:Hq6gL+utB4ET+UvD5ci0kgAwsX6qP/zvf8v6OInU5iw": "Hq6gL+utB4ET+UvD5ci0kgAwsX6qP/zvf8v6OInU5iw" + }, + } + } + yield self.handler.upload_signing_keys_for_user(local_user, keys2) + + devices = yield self.handler.query_devices({"device_keys": {local_user: []}}, 0, local_user) + self.assertDictEqual(devices["master_keys"], {local_user: keys2["master_key"]}) + + @defer.inlineCallbacks + def test_self_signing_key_doesnt_show_up_as_device(self): + """signing keys should be hidden when fetching a user's devices""" + local_user = "@boris:" + self.hs.hostname + keys1 = { + "master_key": { + # private key: 2lonYOM6xYKdEsO+6KrC766xBcHnYnim1x/4LFGF8B0 + "user_id": local_user, + "usage": ["master"], + "keys": { + "ed25519:nqOvzeuGWT/sRx3h7+MHoInYj3Uk2LD/unI9kDYcHwk": "nqOvzeuGWT/sRx3h7+MHoInYj3Uk2LD/unI9kDYcHwk" + }, + } + } + yield self.handler.upload_signing_keys_for_user(local_user, keys1) + + res = None + try: + yield self.hs.get_device_handler().check_device_registered( + user_id=local_user, + device_id="nqOvzeuGWT/sRx3h7+MHoInYj3Uk2LD/unI9kDYcHwk", + initial_device_display_name="new display name", + ) + except errors.SynapseError as e: + res = e.code + self.assertEqual(res, 400) + + res = yield self.handler.query_local_devices({local_user: None}) + self.assertDictEqual(res, {local_user: {}}) -- cgit 1.4.1 From 8c9adcc95dee892f90d6acbbe5c54acbf621720b Mon Sep 17 00:00:00 2001 From: Hubert Chathi Date: Thu, 1 Aug 2019 22:09:05 -0400 Subject: fix formatting --- changelog.d/5769.feature | 2 +- tests/handlers/test_e2e_keys.py | 4 +++- 2 files changed, 4 insertions(+), 2 deletions(-) (limited to 'tests') diff --git a/changelog.d/5769.feature b/changelog.d/5769.feature index c34257cb8f..bf994ca327 100644 --- a/changelog.d/5769.feature +++ b/changelog.d/5769.feature @@ -1 +1 @@ -allow uploading of cross-signing keys \ No newline at end of file +Allow uploading of cross-signing keys. \ No newline at end of file diff --git a/tests/handlers/test_e2e_keys.py b/tests/handlers/test_e2e_keys.py index 9ae4cb6ea2..a62c52eefa 100644 --- a/tests/handlers/test_e2e_keys.py +++ b/tests/handlers/test_e2e_keys.py @@ -176,7 +176,9 @@ class E2eKeysHandlerTestCase(unittest.TestCase): } yield self.handler.upload_signing_keys_for_user(local_user, keys2) - devices = yield self.handler.query_devices({"device_keys": {local_user: []}}, 0, local_user) + devices = yield self.handler.query_devices( + {"device_keys": {local_user: []}}, 0, local_user + ) self.assertDictEqual(devices["master_keys"], {local_user: keys2["master_key"]}) @defer.inlineCallbacks -- cgit 1.4.1 From ac4746ac4bb4d9371c5a25e94ecccd83effb8b9a Mon Sep 17 00:00:00 2001 From: Hubert Chathi Date: Wed, 17 Jul 2019 22:11:31 -0400 Subject: allow uploading signatures of master key signed by devices --- synapse/handlers/e2e_keys.py | 232 ++++++++++++++++++++++------------- synapse/rest/client/v2_alpha/keys.py | 2 +- synapse/storage/end_to_end_keys.py | 2 +- tests/handlers/test_e2e_keys.py | 227 +++++++++++++++++++++++++++++++++- 4 files changed, 378 insertions(+), 85 deletions(-) (limited to 'tests') diff --git a/synapse/handlers/e2e_keys.py b/synapse/handlers/e2e_keys.py index 9747b517ff..1148803c1e 100644 --- a/synapse/handlers/e2e_keys.py +++ b/synapse/handlers/e2e_keys.py @@ -20,7 +20,9 @@ import logging from six import iteritems from canonicaljson import encode_canonical_json, json +from signedjson.key import decode_verify_key_bytes from signedjson.sign import SignatureVerifyException, verify_signed_json +from unpaddedbase64 import decode_base64 from twisted.internet import defer @@ -619,8 +621,11 @@ class E2eKeysHandler(object): """ failures = {} - signature_list = [] # signatures to be stored - self_device_ids = [] # what devices have been updated, for notifying + # signatures to be stored. Each item will be a tuple of + # (signing_key_id, target_user_id, target_device_id, signature) + signature_list = [] + # what devices have been updated, for notifying + self_device_ids = [] # split between checking signatures for own user and signatures for # other users, since we verify them with different keys @@ -630,46 +635,107 @@ class E2eKeysHandler(object): self_device_ids = list(self_signatures.keys()) try: # get our self-signing key to verify the signatures - self_signing_key = yield self.store.get_e2e_cross_signing_key( - user_id, "self_signing" - ) - if self_signing_key is None: - raise SynapseError( - 404, - "No self-signing key found", - Codes.NOT_FOUND + self_signing_key, self_signing_key_id, self_signing_verify_key \ + = yield self._get_e2e_cross_signing_verify_key( + user_id, "self_signing" ) - self_signing_key_id, self_signing_verify_key \ - = get_verify_key_from_cross_signing_key(self_signing_key) + # get our master key, since it may be signed + master_key, master_key_id, master_verify_key \ + = yield self._get_e2e_cross_signing_verify_key( + user_id, "master" + ) - # fetch our stored devices so that we can compare with what was sent - user_devices = [] - for device in self_signatures.keys(): - user_devices.append((user_id, device)) - devices = yield self.store.get_e2e_device_keys(user_devices) + # fetch our stored devices. This is used to 1. verify + # signatures on the master key, and 2. to can compare with what + # was sent if the device was signed + devices = yield self.store.get_e2e_device_keys([(user_id, None)]) if user_id not in devices: raise SynapseError( - 404, - "No device key found", - Codes.NOT_FOUND + 404, "No device keys found", Codes.NOT_FOUND ) devices = devices[user_id] for device_id, device in self_signatures.items(): try: if ("signatures" not in device or - user_id not in device["signatures"] or - self_signing_key_id not in device["signatures"][user_id]): + user_id not in device["signatures"]): + # no signature was sent + raise SynapseError( + 400, "Invalid signature", Codes.INVALID_SIGNATURE + ) + + if device_id == master_verify_key.version: + # we have master key signed by devices: for each + # device that signed, check the signature. Since + # the "failures" property in the response only has + # granularity up to the signed device, either all + # of the signatures on the master key succeed, or + # all fail. So loop over the signatures and add + # them to a separate signature list. If everything + # works out, then add them all to the main + # signature list. (In practice, we're likely to + # only have only one signature anyways.) + master_key_signature_list = [] + for signing_key_id, signature in device["signatures"][user_id].items(): + alg, signing_device_id = signing_key_id.split(":", 1) + if (signing_device_id not in devices or + signing_key_id not in + devices[signing_device_id]["keys"]["keys"]): + # signed by an unknown device, or the + # device does not have the key + raise SynapseError( + 400, "Invalid signature", Codes.INVALID_SIGNATURE + ) + + sigs = device["signatures"] + del device["signatures"] + # use pop to avoid exception if key doesn't exist + device.pop("unsigned", None) + master_key.pop("signature", None) + master_key.pop("unsigned", None) + + if master_key != device: + raise SynapseError( + 400, "Key does not match" + ) + + # get the key and check the signature + pubkey = devices[signing_device_id]["keys"]["keys"][signing_key_id] + verify_key = decode_verify_key_bytes( + signing_key_id, decode_base64(pubkey) + ) + device["signatures"] = sigs + try: + verify_signed_json(device, user_id, verify_key) + except SignatureVerifyException: + raise SynapseError( + 400, "Invalid signature", Codes.INVALID_SIGNATURE + ) + + master_key_signature_list.append( + (signing_key_id, user_id, device_id, signature) + ) + + signature_list.extend(master_key_signature_list) + continue + + # at this point, we have a device that should be signed + # by the self-signing key + if self_signing_key_id not in device["signatures"][user_id]: # no signature was sent raise SynapseError( - 400, - "Invalid signature", - Codes.INVALID_SIGNATURE + 400, "Invalid signature", Codes.INVALID_SIGNATURE ) - stored_device = devices[device_id]["keys"] + stored_device = None + try: + stored_device = devices[device_id]["keys"] + except KeyError: + raise SynapseError( + 404, "Unknown device", Codes.NOT_FOUND + ) if self_signing_key_id in stored_device.get("signatures", {}) \ .get(user_id, {}): # we already have a signature on this device, so we @@ -698,69 +764,50 @@ class E2eKeysHandler(object): # if signatures isn't empty, then we have signatures for other # users. These signatures will be signed by the user signing key - # get our user-signing key to verify the signatures - user_signing_key = yield self.store.get_e2e_cross_signing_key( - user_id, "user_signing" - ) - if user_signing_key is None: - for user, devicemap in signatures.items(): - failures[user] = { - device: _exception_to_failure(SynapseError( - 404, - "No user-signing key found", - Codes.NOT_FOUND - )) - for device in devicemap.keys() - } - else: - user_signing_key_id, user_signing_verify_key \ - = get_verify_key_from_cross_signing_key(user_signing_key) + try: + # get our user-signing key to verify the signatures + user_signing_key, user_signing_key_id, user_signing_verify_key \ + = yield self._get_e2e_cross_signing_verify_key( + user_id, "user_signing" + ) for user, devicemap in signatures.items(): device_id = None try: # get the user's master key, to make sure it matches # what was sent - stored_key = yield self.store.get_e2e_cross_signing_key( - user, "master", user_id - ) - if stored_key is None: - logger.error( - "upload signature: no user key found for %s", user - ) - raise SynapseError( - 404, - "User's master key not found", - Codes.NOT_FOUND + stored_key, stored_key_id, _ \ + = yield self._get_e2e_cross_signing_verify_key( + user, "master", user_id ) # make sure that the user's master key is the one that # was signed (and no others) - device_id = get_verify_key_from_cross_signing_key(stored_key)[0] \ - .split(":", 1)[1] + device_id = stored_key_id.split(":", 1)[1] if device_id not in devicemap: + # set device to None so that the failure gets + # marked on all the signatures + device_id = None logger.error( "upload signature: wrong device: %s vs %s", device, devicemap ) raise SynapseError( - 404, - "Unknown device", - Codes.NOT_FOUND + 404, "Unknown device", Codes.NOT_FOUND ) - if len(devicemap) > 1: + key = devicemap[device_id] + del devicemap[device_id] + if len(devicemap) > 0: + # other devices were signed -- mark those as failures logger.error("upload signature: too many devices specified") + failure = _exception_to_failure(SynapseError( + 404, "Unknown device", Codes.NOT_FOUND + )) failures[user] = { - device: _exception_to_failure(SynapseError( - 404, - "Unknown device", - Codes.NOT_FOUND - )) + device: failure for device in devicemap.keys() } - key = devicemap[device_id] - if user_signing_key_id in stored_key.get("signatures", {}) \ .get(user_id, {}): # we already have the signature, so we can skip it @@ -770,25 +817,31 @@ class E2eKeysHandler(object): user_id, user_signing_verify_key, key, stored_key ) - signature = key["signatures"][user_id][user_signing_key_id] - signed_users.append(user) + signature = key["signatures"][user_id][user_signing_key_id] signature_list.append( (user_signing_key_id, user, device_id, signature) ) except SynapseError as e: + failure = _exception_to_failure(e) if device_id is None: failures[user] = { - device_id: _exception_to_failure(e) + device_id: failure for device_id in devicemap.keys() } else: - failures.setdefault(user, {})[device_id] \ - = _exception_to_failure(e) + failures.setdefault(user, {})[device_id] = failure + except SynapseError as e: + failure = _exception_to_failure(e) + for user, devicemap in signature.items(): + failures[user] = { + device_id: failure + for device_id in devicemap.keys() + } # store the signature, and send the appropriate notifications for sync logger.debug("upload signature failures: %r", failures) - yield self.store.store_e2e_device_signatures(user_id, signature_list) + yield self.store.store_e2e_cross_signing_signatures(user_id, signature_list) if len(self_device_ids): yield self.device_handler.notify_device_update(user_id, self_device_ids) @@ -797,6 +850,22 @@ class E2eKeysHandler(object): defer.returnValue({"failures": failures}) + @defer.inlineCallbacks + def _get_e2e_cross_signing_verify_key(self, user_id, key_type, from_user_id=None): + key = yield self.store.get_e2e_cross_signing_key( + user_id, key_type, from_user_id + ) + if key is None: + logger.error("no %s key found for %s", key_type, user_id) + raise SynapseError( + 404, + "No %s key found for %s" % (key_type, user_id), + Codes.NOT_FOUND + ) + key_id, verify_key = get_verify_key_from_cross_signing_key(key) + return key, key_id, verify_key + + def _check_cross_signing_key(key, user_id, key_type, signing_key=None): """Check a cross-signing key uploaded by a user. Performs some basic sanity checking, and ensures that it is signed, if a signature is required. @@ -851,21 +920,17 @@ def _check_device_signature(user_id, verify_key, signed_device, stored_device): # make sure that the device submitted matches what we have stored del signed_device["signatures"] - if "unsigned" in signed_device: - del signed_device["unsigned"] - if "signatures" in stored_device: - del stored_device["signatures"] - if "unsigned" in stored_device: - del stored_device["unsigned"] + # use pop to avoid exception if key doesn't exist + signed_device.pop("unsigned", None) + stored_device.pop("signatures", None) + stored_device.pop("unsigned", None) if signed_device != stored_device: logger.error( "upload signatures: key does not match %s vs %s", signed_device, stored_device ) raise SynapseError( - 400, - "Key does not match", - "M_MISMATCHED_KEY" + 400, "Key does not match", ) # check the signature @@ -887,6 +952,9 @@ def _check_device_signature(user_id, verify_key, signed_device, stored_device): def _exception_to_failure(e): + if isinstance(e, SynapseError): + return {"status": e.code, "errcode": e.errcode, "message": str(e)} + if isinstance(e, CodeMessageException): return {"status": e.code, "message": str(e)} diff --git a/synapse/rest/client/v2_alpha/keys.py b/synapse/rest/client/v2_alpha/keys.py index 5c288d48b7..cb3c52cb8e 100644 --- a/synapse/rest/client/v2_alpha/keys.py +++ b/synapse/rest/client/v2_alpha/keys.py @@ -303,7 +303,7 @@ class SignaturesUploadServlet(RestServlet): } } """ - PATTERNS = client_v2_patterns("/keys/signatures/upload$") + PATTERNS = client_patterns("/keys/signatures/upload$") def __init__(self, hs): """ diff --git a/synapse/storage/end_to_end_keys.py b/synapse/storage/end_to_end_keys.py index fe786f3093..e68ce318af 100644 --- a/synapse/storage/end_to_end_keys.py +++ b/synapse/storage/end_to_end_keys.py @@ -132,7 +132,7 @@ class EndToEndKeyWorkerStore(SQLBaseStore): # get signatures on the device signature_sql = ( "SELECT * " - " FROM e2e_device_signatures " + " FROM e2e_cross_signing_signatures " " WHERE %s" ) % ( " OR ".join("(" + q + ")" for q in signature_query_clauses) diff --git a/tests/handlers/test_e2e_keys.py b/tests/handlers/test_e2e_keys.py index a62c52eefa..b1d3a4cfae 100644 --- a/tests/handlers/test_e2e_keys.py +++ b/tests/handlers/test_e2e_keys.py @@ -17,9 +17,10 @@ import mock +import signedjson.key as key +import signedjson.sign as sign from twisted.internet import defer -import synapse.api.errors import synapse.handlers.e2e_keys import synapse.storage from synapse.api import errors @@ -210,3 +211,227 @@ class E2eKeysHandlerTestCase(unittest.TestCase): res = yield self.handler.query_local_devices({local_user: None}) self.assertDictEqual(res, {local_user: {}}) + + @defer.inlineCallbacks + def test_upload_signatures(self): + """should check signatures that are uploaded""" + # set up a user with cross-signing keys and a device. This user will + # try uploading signatures + local_user = "@boris:" + self.hs.hostname + device_id = "xyz" + # private key: OMkooTr76ega06xNvXIGPbgvvxAOzmQncN8VObS7aBA + device_pubkey = "NnHhnqiMFQkq969szYkooLaBAXW244ZOxgukCvm2ZeY" + device_key = { + "user_id": local_user, + "device_id": device_id, + "algorithms": ["m.olm.curve25519-aes-sha256", "m.megolm.v1.aes-sha"], + "keys": { + "curve25519:xyz": "curve25519+key", + "ed25519:xyz": device_pubkey + }, + "signatures": { + local_user: { + "ed25519:xyz": "something" + } + } + } + device_signing_key = key.decode_signing_key_base64( + "ed25519", + "xyz", + "OMkooTr76ega06xNvXIGPbgvvxAOzmQncN8VObS7aBA" + ) + + yield self.handler.upload_keys_for_user( + local_user, device_id, {"device_keys": device_key} + ) + + # private key: 2lonYOM6xYKdEsO+6KrC766xBcHnYnim1x/4LFGF8B0 + master_pubkey = "nqOvzeuGWT/sRx3h7+MHoInYj3Uk2LD/unI9kDYcHwk" + master_key = { + "user_id": local_user, + "usage": ["master"], + "keys": { + "ed25519:" + master_pubkey: master_pubkey + } + } + master_signing_key = key.decode_signing_key_base64( + "ed25519", master_pubkey, + "2lonYOM6xYKdEsO+6KrC766xBcHnYnim1x/4LFGF8B0" + ) + usersigning_pubkey = "Hq6gL+utB4ET+UvD5ci0kgAwsX6qP/zvf8v6OInU5iw" + usersigning_key = { + # private key: 4TL4AjRYwDVwD3pqQzcor+ez/euOB1/q78aTJ+czDNs + "user_id": local_user, + "usage": ["user_signing"], + "keys": { + "ed25519:" + usersigning_pubkey: usersigning_pubkey, + } + } + usersigning_signing_key = key.decode_signing_key_base64( + "ed25519", usersigning_pubkey, + "4TL4AjRYwDVwD3pqQzcor+ez/euOB1/q78aTJ+czDNs" + ) + sign.sign_json(usersigning_key, local_user, master_signing_key) + # private key: HvQBbU+hc2Zr+JP1sE0XwBe1pfZZEYtJNPJLZJtS+F8 + selfsigning_pubkey = "EmkqvokUn8p+vQAGZitOk4PWjp7Ukp3txV2TbMPEiBQ" + selfsigning_key = { + "user_id": local_user, + "usage": ["self_signing"], + "keys": { + "ed25519:" + selfsigning_pubkey: selfsigning_pubkey, + } + } + selfsigning_signing_key = key.decode_signing_key_base64( + "ed25519", selfsigning_pubkey, + "HvQBbU+hc2Zr+JP1sE0XwBe1pfZZEYtJNPJLZJtS+F8" + ) + sign.sign_json(selfsigning_key, local_user, master_signing_key) + cross_signing_keys = { + "master_key": master_key, + "user_signing_key": usersigning_key, + "self_signing_key": selfsigning_key, + } + yield self.handler.upload_signing_keys_for_user(local_user, cross_signing_keys) + + # set up another user with a master key. This user will be signed by + # the first user + other_user = "@otherboris:" + self.hs.hostname + other_master_pubkey = "fHZ3NPiKxoLQm5OoZbKa99SYxprOjNs4TwJUKP+twCM" + other_master_key = { + # private key: oyw2ZUx0O4GifbfFYM0nQvj9CL0b8B7cyN4FprtK8OI + "user_id": other_user, + "usage": ["master"], + "keys": { + "ed25519:" + other_master_pubkey: other_master_pubkey + } + } + yield self.handler.upload_signing_keys_for_user(other_user, { + "master_key": other_master_key + }) + + # test various signature failures (see below) + ret = yield self.handler.upload_signatures_for_device_keys( + local_user, + { + local_user: { + # fails because the signature is invalid + # should fail with INVALID_SIGNATURE + device_id: { + "user_id": local_user, + "device_id": device_id, + "algorithms": ["m.olm.curve25519-aes-sha256", "m.megolm.v1.aes-sha"], + "keys": { + "curve25519:xyz": "curve25519+key", + # private key: OMkooTr76ega06xNvXIGPbgvvxAOzmQncN8VObS7aBA + "ed25519:xyz": device_pubkey + }, + "signatures": { + local_user: { + "ed25519:" + selfsigning_pubkey: "something", + } + } + }, + # fails because device is unknown + # should fail with NOT_FOUND + "unknown": { + "user_id": local_user, + "device_id": "unknown", + "signatures": { + local_user: { + "ed25519:" + selfsigning_pubkey: "something", + } + } + }, + # fails because the signature is invalid + # should fail with INVALID_SIGNATURE + master_pubkey: { + "user_id": local_user, + "usage": ["master"], + "keys": { + "ed25519:" + master_pubkey: master_pubkey + }, + "signatures": { + local_user: { + "ed25519:" + device_pubkey: "something", + } + } + } + }, + other_user: { + # fails because the device is not the user's master-signing key + # should fail with NOT_FOUND + "unknown": { + "user_id": other_user, + "device_id": "unknown", + "signatures": { + local_user: { + "ed25519:" + usersigning_pubkey: "something", + } + } + }, + other_master_pubkey: { + # fails because the key doesn't match what the server has + # should fail with UNKNOWN + "user_id": other_user, + "usage": ["master"], + "keys": { + "ed25519:" + other_master_pubkey: other_master_pubkey + }, + "something": "random", + "signatures": { + local_user: { + "ed25519:" + usersigning_pubkey: "something", + } + } + } + } + } + ) + + user_failures = ret["failures"][local_user] + self.assertEqual(user_failures[device_id]["errcode"], errors.Codes.INVALID_SIGNATURE) + self.assertEqual(user_failures[master_pubkey]["errcode"], errors.Codes.INVALID_SIGNATURE) + self.assertEqual(user_failures["unknown"]["errcode"], errors.Codes.NOT_FOUND) + + other_user_failures = ret["failures"][other_user] + self.assertEqual(other_user_failures["unknown"]["errcode"], errors.Codes.NOT_FOUND) + self.assertEqual(other_user_failures[other_master_pubkey]["errcode"], errors.Codes.UNKNOWN) + + # test successful signatures + del device_key["signatures"] + sign.sign_json(device_key, local_user, selfsigning_signing_key) + sign.sign_json(master_key, local_user, device_signing_key) + sign.sign_json(other_master_key, local_user, usersigning_signing_key) + ret = yield self.handler.upload_signatures_for_device_keys( + local_user, + { + local_user: { + device_id: device_key, + master_pubkey: master_key + }, + other_user: { + other_master_pubkey: other_master_key + } + } + ) + + self.assertEqual(ret["failures"], {}) + + # fetch the signed keys/devices and make sure that the signatures are there + ret = yield self.handler.query_devices( + {"device_keys": {local_user: [], other_user: []}}, + 0, local_user + ) + + self.assertEqual( + ret["device_keys"][local_user]["xyz"]["signatures"][local_user]["ed25519:" + selfsigning_pubkey], + device_key["signatures"][local_user]["ed25519:" + selfsigning_pubkey] + ) + self.assertEqual( + ret["master_keys"][local_user]["signatures"][local_user]["ed25519:" + device_id], + master_key["signatures"][local_user]["ed25519:" + device_id] + ) + self.assertEqual( + ret["master_keys"][other_user]["signatures"][local_user]["ed25519:" + usersigning_pubkey], + other_master_key["signatures"][local_user]["ed25519:" + usersigning_pubkey] + ) -- cgit 1.4.1 From 7d6c70fc7ad08b94b8b577c537953a8d9b568562 Mon Sep 17 00:00:00 2001 From: Hubert Chathi Date: Mon, 22 Jul 2019 12:52:39 -0400 Subject: make black happy --- synapse/handlers/e2e_keys.py | 147 ++++++++++++++++------------------- synapse/rest/client/v2_alpha/keys.py | 1 + synapse/storage/end_to_end_keys.py | 24 +++--- tests/handlers/test_e2e_keys.py | 147 +++++++++++++++-------------------- 4 files changed, 141 insertions(+), 178 deletions(-) (limited to 'tests') diff --git a/synapse/handlers/e2e_keys.py b/synapse/handlers/e2e_keys.py index 1148803c1e..74bceddc46 100644 --- a/synapse/handlers/e2e_keys.py +++ b/synapse/handlers/e2e_keys.py @@ -635,16 +635,14 @@ class E2eKeysHandler(object): self_device_ids = list(self_signatures.keys()) try: # get our self-signing key to verify the signatures - self_signing_key, self_signing_key_id, self_signing_verify_key \ - = yield self._get_e2e_cross_signing_verify_key( - user_id, "self_signing" - ) + self_signing_key, self_signing_key_id, self_signing_verify_key = yield self._get_e2e_cross_signing_verify_key( + user_id, "self_signing" + ) # get our master key, since it may be signed - master_key, master_key_id, master_verify_key \ - = yield self._get_e2e_cross_signing_verify_key( - user_id, "master" - ) + master_key, master_key_id, master_verify_key = yield self._get_e2e_cross_signing_verify_key( + user_id, "master" + ) # fetch our stored devices. This is used to 1. verify # signatures on the master key, and 2. to can compare with what @@ -652,15 +650,15 @@ class E2eKeysHandler(object): devices = yield self.store.get_e2e_device_keys([(user_id, None)]) if user_id not in devices: - raise SynapseError( - 404, "No device keys found", Codes.NOT_FOUND - ) + raise SynapseError(404, "No device keys found", Codes.NOT_FOUND) devices = devices[user_id] for device_id, device in self_signatures.items(): try: - if ("signatures" not in device or - user_id not in device["signatures"]): + if ( + "signatures" not in device + or user_id not in device["signatures"] + ): # no signature was sent raise SynapseError( 400, "Invalid signature", Codes.INVALID_SIGNATURE @@ -678,15 +676,21 @@ class E2eKeysHandler(object): # signature list. (In practice, we're likely to # only have only one signature anyways.) master_key_signature_list = [] - for signing_key_id, signature in device["signatures"][user_id].items(): + for signing_key_id, signature in device["signatures"][ + user_id + ].items(): alg, signing_device_id = signing_key_id.split(":", 1) - if (signing_device_id not in devices or - signing_key_id not in - devices[signing_device_id]["keys"]["keys"]): + if ( + signing_device_id not in devices + or signing_key_id + not in devices[signing_device_id]["keys"]["keys"] + ): # signed by an unknown device, or the # device does not have the key raise SynapseError( - 400, "Invalid signature", Codes.INVALID_SIGNATURE + 400, + "Invalid signature", + Codes.INVALID_SIGNATURE, ) sigs = device["signatures"] @@ -697,12 +701,12 @@ class E2eKeysHandler(object): master_key.pop("unsigned", None) if master_key != device: - raise SynapseError( - 400, "Key does not match" - ) + raise SynapseError(400, "Key does not match") # get the key and check the signature - pubkey = devices[signing_device_id]["keys"]["keys"][signing_key_id] + pubkey = devices[signing_device_id]["keys"]["keys"][ + signing_key_id + ] verify_key = decode_verify_key_bytes( signing_key_id, decode_base64(pubkey) ) @@ -711,7 +715,9 @@ class E2eKeysHandler(object): verify_signed_json(device, user_id, verify_key) except SignatureVerifyException: raise SynapseError( - 400, "Invalid signature", Codes.INVALID_SIGNATURE + 400, + "Invalid signature", + Codes.INVALID_SIGNATURE, ) master_key_signature_list.append( @@ -733,11 +739,10 @@ class E2eKeysHandler(object): try: stored_device = devices[device_id]["keys"] except KeyError: - raise SynapseError( - 404, "Unknown device", Codes.NOT_FOUND - ) - if self_signing_key_id in stored_device.get("signatures", {}) \ - .get(user_id, {}): + raise SynapseError(404, "Unknown device", Codes.NOT_FOUND) + if self_signing_key_id in stored_device.get( + "signatures", {} + ).get(user_id, {}): # we already have a signature on this device, so we # can skip it, since it should be exactly the same continue @@ -751,8 +756,9 @@ class E2eKeysHandler(object): (self_signing_key_id, user_id, device_id, signature) ) except SynapseError as e: - failures.setdefault(user_id, {})[device_id] \ - = _exception_to_failure(e) + failures.setdefault(user_id, {})[ + device_id + ] = _exception_to_failure(e) except SynapseError as e: failures[user_id] = { device: _exception_to_failure(e) @@ -766,20 +772,18 @@ class E2eKeysHandler(object): try: # get our user-signing key to verify the signatures - user_signing_key, user_signing_key_id, user_signing_verify_key \ - = yield self._get_e2e_cross_signing_verify_key( - user_id, "user_signing" - ) + user_signing_key, user_signing_key_id, user_signing_verify_key = yield self._get_e2e_cross_signing_verify_key( + user_id, "user_signing" + ) for user, devicemap in signatures.items(): device_id = None try: # get the user's master key, to make sure it matches # what was sent - stored_key, stored_key_id, _ \ - = yield self._get_e2e_cross_signing_verify_key( - user, "master", user_id - ) + stored_key, stored_key_id, _ = yield self._get_e2e_cross_signing_verify_key( + user, "master", user_id + ) # make sure that the user's master key is the one that # was signed (and no others) @@ -790,26 +794,25 @@ class E2eKeysHandler(object): device_id = None logger.error( "upload signature: wrong device: %s vs %s", - device, devicemap - ) - raise SynapseError( - 404, "Unknown device", Codes.NOT_FOUND + device, + devicemap, ) + raise SynapseError(404, "Unknown device", Codes.NOT_FOUND) key = devicemap[device_id] del devicemap[device_id] if len(devicemap) > 0: # other devices were signed -- mark those as failures logger.error("upload signature: too many devices specified") - failure = _exception_to_failure(SynapseError( - 404, "Unknown device", Codes.NOT_FOUND - )) + failure = _exception_to_failure( + SynapseError(404, "Unknown device", Codes.NOT_FOUND) + ) failures[user] = { - device: failure - for device in devicemap.keys() + device: failure for device in devicemap.keys() } - if user_signing_key_id in stored_key.get("signatures", {}) \ - .get(user_id, {}): + if user_signing_key_id in stored_key.get("signatures", {}).get( + user_id, {} + ): # we already have the signature, so we can skip it continue @@ -826,8 +829,7 @@ class E2eKeysHandler(object): failure = _exception_to_failure(e) if device_id is None: failures[user] = { - device_id: failure - for device_id in devicemap.keys() + device_id: failure for device_id in devicemap.keys() } else: failures.setdefault(user, {})[device_id] = failure @@ -835,8 +837,7 @@ class E2eKeysHandler(object): failure = _exception_to_failure(e) for user, devicemap in signature.items(): failures[user] = { - device_id: failure - for device_id in devicemap.keys() + device_id: failure for device_id in devicemap.keys() } # store the signature, and send the appropriate notifications for sync @@ -846,7 +847,9 @@ class E2eKeysHandler(object): if len(self_device_ids): yield self.device_handler.notify_device_update(user_id, self_device_ids) if len(signed_users): - yield self.device_handler.notify_user_signature_update(user_id, signed_users) + yield self.device_handler.notify_user_signature_update( + user_id, signed_users + ) defer.returnValue({"failures": failures}) @@ -858,9 +861,7 @@ class E2eKeysHandler(object): if key is None: logger.error("no %s key found for %s", key_type, user_id) raise SynapseError( - 404, - "No %s key found for %s" % (key_type, user_id), - Codes.NOT_FOUND + 404, "No %s key found for %s" % (key_type, user_id), Codes.NOT_FOUND ) key_id, verify_key = get_verify_key_from_cross_signing_key(key) return key, key_id, verify_key @@ -907,14 +908,13 @@ def _check_device_signature(user_id, verify_key, signed_device, stored_device): key_id = "%s:%s" % (verify_key.alg, verify_key.version) # make sure the device is signed - if ("signatures" not in signed_device or user_id not in signed_device["signatures"] - or key_id not in signed_device["signatures"][user_id]): + if ( + "signatures" not in signed_device + or user_id not in signed_device["signatures"] + or key_id not in signed_device["signatures"][user_id] + ): logger.error("upload signature: user not found in signatures") - raise SynapseError( - 400, - "Invalid signature", - Codes.INVALID_SIGNATURE - ) + raise SynapseError(400, "Invalid signature", Codes.INVALID_SIGNATURE) signature = signed_device["signatures"][user_id][key_id] @@ -927,28 +927,19 @@ def _check_device_signature(user_id, verify_key, signed_device, stored_device): if signed_device != stored_device: logger.error( "upload signatures: key does not match %s vs %s", - signed_device, stored_device - ) - raise SynapseError( - 400, "Key does not match", + signed_device, + stored_device, ) + raise SynapseError(400, "Key does not match") # check the signature - signed_device["signatures"] = { - user_id: { - key_id: signature - } - } + signed_device["signatures"] = {user_id: {key_id: signature}} try: verify_signed_json(signed_device, user_id, verify_key) except SignatureVerifyException: logger.error("invalid signature on key") - raise SynapseError( - 400, - "Invalid signature", - Codes.INVALID_SIGNATURE - ) + raise SynapseError(400, "Invalid signature", Codes.INVALID_SIGNATURE) def _exception_to_failure(e): diff --git a/synapse/rest/client/v2_alpha/keys.py b/synapse/rest/client/v2_alpha/keys.py index cb3c52cb8e..a205281830 100644 --- a/synapse/rest/client/v2_alpha/keys.py +++ b/synapse/rest/client/v2_alpha/keys.py @@ -303,6 +303,7 @@ class SignaturesUploadServlet(RestServlet): } } """ + PATTERNS = client_patterns("/keys/signatures/upload$") def __init__(self, hs): diff --git a/synapse/storage/end_to_end_keys.py b/synapse/storage/end_to_end_keys.py index e68ce318af..258e8dcb47 100644 --- a/synapse/storage/end_to_end_keys.py +++ b/synapse/storage/end_to_end_keys.py @@ -62,9 +62,9 @@ class EndToEndKeyWorkerStore(SQLBaseStore): # add cross-signing signatures to the keys if "signatures" in device_info: for sig_user_id, sigs in device_info["signatures"].items(): - device_info["keys"].setdefault("signatures", {}) \ - .setdefault(sig_user_id, {}) \ - .update(sigs) + device_info["keys"].setdefault("signatures", {}).setdefault( + sig_user_id, {} + ).update(sigs) return results @@ -131,12 +131,8 @@ class EndToEndKeyWorkerStore(SQLBaseStore): # get signatures on the device signature_sql = ( - "SELECT * " - " FROM e2e_cross_signing_signatures " - " WHERE %s" - ) % ( - " OR ".join("(" + q + ")" for q in signature_query_clauses) - ) + "SELECT * " " FROM e2e_cross_signing_signatures " " WHERE %s" + ) % (" OR ".join("(" + q + ")" for q in signature_query_clauses)) txn.execute(signature_sql, signature_query_params) rows = self.cursor_to_dict(txn) @@ -144,12 +140,10 @@ class EndToEndKeyWorkerStore(SQLBaseStore): for row in rows: target_user_id = row["target_user_id"] target_device_id = row["target_device_id"] - if target_user_id in result \ - and target_device_id in result[target_user_id]: - result[target_user_id][target_device_id] \ - .setdefault("signatures", {}) \ - .setdefault(row["user_id"], {})[row["key_id"]] \ - = row["signature"] + if target_user_id in result and target_device_id in result[target_user_id]: + result[target_user_id][target_device_id].setdefault( + "signatures", {} + ).setdefault(row["user_id"], {})[row["key_id"]] = row["signature"] log_kv(result) return result diff --git a/tests/handlers/test_e2e_keys.py b/tests/handlers/test_e2e_keys.py index b1d3a4cfae..8c0ee3f7d3 100644 --- a/tests/handlers/test_e2e_keys.py +++ b/tests/handlers/test_e2e_keys.py @@ -225,20 +225,11 @@ class E2eKeysHandlerTestCase(unittest.TestCase): "user_id": local_user, "device_id": device_id, "algorithms": ["m.olm.curve25519-aes-sha256", "m.megolm.v1.aes-sha"], - "keys": { - "curve25519:xyz": "curve25519+key", - "ed25519:xyz": device_pubkey - }, - "signatures": { - local_user: { - "ed25519:xyz": "something" - } - } + "keys": {"curve25519:xyz": "curve25519+key", "ed25519:xyz": device_pubkey}, + "signatures": {local_user: {"ed25519:xyz": "something"}}, } device_signing_key = key.decode_signing_key_base64( - "ed25519", - "xyz", - "OMkooTr76ega06xNvXIGPbgvvxAOzmQncN8VObS7aBA" + "ed25519", "xyz", "OMkooTr76ega06xNvXIGPbgvvxAOzmQncN8VObS7aBA" ) yield self.handler.upload_keys_for_user( @@ -250,26 +241,20 @@ class E2eKeysHandlerTestCase(unittest.TestCase): master_key = { "user_id": local_user, "usage": ["master"], - "keys": { - "ed25519:" + master_pubkey: master_pubkey - } + "keys": {"ed25519:" + master_pubkey: master_pubkey}, } master_signing_key = key.decode_signing_key_base64( - "ed25519", master_pubkey, - "2lonYOM6xYKdEsO+6KrC766xBcHnYnim1x/4LFGF8B0" + "ed25519", master_pubkey, "2lonYOM6xYKdEsO+6KrC766xBcHnYnim1x/4LFGF8B0" ) usersigning_pubkey = "Hq6gL+utB4ET+UvD5ci0kgAwsX6qP/zvf8v6OInU5iw" usersigning_key = { # private key: 4TL4AjRYwDVwD3pqQzcor+ez/euOB1/q78aTJ+czDNs "user_id": local_user, "usage": ["user_signing"], - "keys": { - "ed25519:" + usersigning_pubkey: usersigning_pubkey, - } + "keys": {"ed25519:" + usersigning_pubkey: usersigning_pubkey}, } usersigning_signing_key = key.decode_signing_key_base64( - "ed25519", usersigning_pubkey, - "4TL4AjRYwDVwD3pqQzcor+ez/euOB1/q78aTJ+czDNs" + "ed25519", usersigning_pubkey, "4TL4AjRYwDVwD3pqQzcor+ez/euOB1/q78aTJ+czDNs" ) sign.sign_json(usersigning_key, local_user, master_signing_key) # private key: HvQBbU+hc2Zr+JP1sE0XwBe1pfZZEYtJNPJLZJtS+F8 @@ -277,13 +262,10 @@ class E2eKeysHandlerTestCase(unittest.TestCase): selfsigning_key = { "user_id": local_user, "usage": ["self_signing"], - "keys": { - "ed25519:" + selfsigning_pubkey: selfsigning_pubkey, - } + "keys": {"ed25519:" + selfsigning_pubkey: selfsigning_pubkey}, } selfsigning_signing_key = key.decode_signing_key_base64( - "ed25519", selfsigning_pubkey, - "HvQBbU+hc2Zr+JP1sE0XwBe1pfZZEYtJNPJLZJtS+F8" + "ed25519", selfsigning_pubkey, "HvQBbU+hc2Zr+JP1sE0XwBe1pfZZEYtJNPJLZJtS+F8" ) sign.sign_json(selfsigning_key, local_user, master_signing_key) cross_signing_keys = { @@ -301,13 +283,11 @@ class E2eKeysHandlerTestCase(unittest.TestCase): # private key: oyw2ZUx0O4GifbfFYM0nQvj9CL0b8B7cyN4FprtK8OI "user_id": other_user, "usage": ["master"], - "keys": { - "ed25519:" + other_master_pubkey: other_master_pubkey - } + "keys": {"ed25519:" + other_master_pubkey: other_master_pubkey}, } - yield self.handler.upload_signing_keys_for_user(other_user, { - "master_key": other_master_key - }) + yield self.handler.upload_signing_keys_for_user( + other_user, {"master_key": other_master_key} + ) # test various signature failures (see below) ret = yield self.handler.upload_signatures_for_device_keys( @@ -319,17 +299,18 @@ class E2eKeysHandlerTestCase(unittest.TestCase): device_id: { "user_id": local_user, "device_id": device_id, - "algorithms": ["m.olm.curve25519-aes-sha256", "m.megolm.v1.aes-sha"], + "algorithms": [ + "m.olm.curve25519-aes-sha256", + "m.megolm.v1.aes-sha", + ], "keys": { "curve25519:xyz": "curve25519+key", # private key: OMkooTr76ega06xNvXIGPbgvvxAOzmQncN8VObS7aBA - "ed25519:xyz": device_pubkey + "ed25519:xyz": device_pubkey, }, "signatures": { - local_user: { - "ed25519:" + selfsigning_pubkey: "something", - } - } + local_user: {"ed25519:" + selfsigning_pubkey: "something"} + }, }, # fails because device is unknown # should fail with NOT_FOUND @@ -337,25 +318,19 @@ class E2eKeysHandlerTestCase(unittest.TestCase): "user_id": local_user, "device_id": "unknown", "signatures": { - local_user: { - "ed25519:" + selfsigning_pubkey: "something", - } - } + local_user: {"ed25519:" + selfsigning_pubkey: "something"} + }, }, # fails because the signature is invalid # should fail with INVALID_SIGNATURE master_pubkey: { "user_id": local_user, "usage": ["master"], - "keys": { - "ed25519:" + master_pubkey: master_pubkey - }, + "keys": {"ed25519:" + master_pubkey: master_pubkey}, "signatures": { - local_user: { - "ed25519:" + device_pubkey: "something", - } - } - } + local_user: {"ed25519:" + device_pubkey: "something"} + }, + }, }, other_user: { # fails because the device is not the user's master-signing key @@ -364,38 +339,40 @@ class E2eKeysHandlerTestCase(unittest.TestCase): "user_id": other_user, "device_id": "unknown", "signatures": { - local_user: { - "ed25519:" + usersigning_pubkey: "something", - } - } + local_user: {"ed25519:" + usersigning_pubkey: "something"} + }, }, other_master_pubkey: { # fails because the key doesn't match what the server has # should fail with UNKNOWN "user_id": other_user, "usage": ["master"], - "keys": { - "ed25519:" + other_master_pubkey: other_master_pubkey - }, + "keys": {"ed25519:" + other_master_pubkey: other_master_pubkey}, "something": "random", "signatures": { - local_user: { - "ed25519:" + usersigning_pubkey: "something", - } - } - } - } - } + local_user: {"ed25519:" + usersigning_pubkey: "something"} + }, + }, + }, + }, ) user_failures = ret["failures"][local_user] - self.assertEqual(user_failures[device_id]["errcode"], errors.Codes.INVALID_SIGNATURE) - self.assertEqual(user_failures[master_pubkey]["errcode"], errors.Codes.INVALID_SIGNATURE) + self.assertEqual( + user_failures[device_id]["errcode"], errors.Codes.INVALID_SIGNATURE + ) + self.assertEqual( + user_failures[master_pubkey]["errcode"], errors.Codes.INVALID_SIGNATURE + ) self.assertEqual(user_failures["unknown"]["errcode"], errors.Codes.NOT_FOUND) other_user_failures = ret["failures"][other_user] - self.assertEqual(other_user_failures["unknown"]["errcode"], errors.Codes.NOT_FOUND) - self.assertEqual(other_user_failures[other_master_pubkey]["errcode"], errors.Codes.UNKNOWN) + self.assertEqual( + other_user_failures["unknown"]["errcode"], errors.Codes.NOT_FOUND + ) + self.assertEqual( + other_user_failures[other_master_pubkey]["errcode"], errors.Codes.UNKNOWN + ) # test successful signatures del device_key["signatures"] @@ -405,33 +382,33 @@ class E2eKeysHandlerTestCase(unittest.TestCase): ret = yield self.handler.upload_signatures_for_device_keys( local_user, { - local_user: { - device_id: device_key, - master_pubkey: master_key - }, - other_user: { - other_master_pubkey: other_master_key - } - } + local_user: {device_id: device_key, master_pubkey: master_key}, + other_user: {other_master_pubkey: other_master_key}, + }, ) self.assertEqual(ret["failures"], {}) # fetch the signed keys/devices and make sure that the signatures are there ret = yield self.handler.query_devices( - {"device_keys": {local_user: [], other_user: []}}, - 0, local_user + {"device_keys": {local_user: [], other_user: []}}, 0, local_user ) self.assertEqual( - ret["device_keys"][local_user]["xyz"]["signatures"][local_user]["ed25519:" + selfsigning_pubkey], - device_key["signatures"][local_user]["ed25519:" + selfsigning_pubkey] + ret["device_keys"][local_user]["xyz"]["signatures"][local_user][ + "ed25519:" + selfsigning_pubkey + ], + device_key["signatures"][local_user]["ed25519:" + selfsigning_pubkey], ) self.assertEqual( - ret["master_keys"][local_user]["signatures"][local_user]["ed25519:" + device_id], - master_key["signatures"][local_user]["ed25519:" + device_id] + ret["master_keys"][local_user]["signatures"][local_user][ + "ed25519:" + device_id + ], + master_key["signatures"][local_user]["ed25519:" + device_id], ) self.assertEqual( - ret["master_keys"][other_user]["signatures"][local_user]["ed25519:" + usersigning_pubkey], - other_master_key["signatures"][local_user]["ed25519:" + usersigning_pubkey] + ret["master_keys"][other_user]["signatures"][local_user][ + "ed25519:" + usersigning_pubkey + ], + other_master_key["signatures"][local_user]["ed25519:" + usersigning_pubkey], ) -- cgit 1.4.1 From 9061b4198af4b30bb99d98aab7ad227f8ed636f8 Mon Sep 17 00:00:00 2001 From: Hubert Chathi Date: Mon, 22 Jul 2019 12:58:04 -0400 Subject: make isort happy --- tests/handlers/test_e2e_keys.py | 1 + 1 file changed, 1 insertion(+) (limited to 'tests') diff --git a/tests/handlers/test_e2e_keys.py b/tests/handlers/test_e2e_keys.py index 8c0ee3f7d3..c900451e03 100644 --- a/tests/handlers/test_e2e_keys.py +++ b/tests/handlers/test_e2e_keys.py @@ -19,6 +19,7 @@ import mock import signedjson.key as key import signedjson.sign as sign + from twisted.internet import defer import synapse.handlers.e2e_keys -- cgit 1.4.1 From 5914fd09c725342d03f702a50ec1da6290e946a9 Mon Sep 17 00:00:00 2001 From: Hubert Chathi Date: Mon, 22 Jul 2019 13:01:10 -0400 Subject: add test --- tests/handlers/test_e2e_keys.py | 88 +++++++++++++++++++++++++++++++++++++++++ 1 file changed, 88 insertions(+) (limited to 'tests') diff --git a/tests/handlers/test_e2e_keys.py b/tests/handlers/test_e2e_keys.py index c900451e03..316dd6259d 100644 --- a/tests/handlers/test_e2e_keys.py +++ b/tests/handlers/test_e2e_keys.py @@ -183,6 +183,94 @@ class E2eKeysHandlerTestCase(unittest.TestCase): ) self.assertDictEqual(devices["master_keys"], {local_user: keys2["master_key"]}) + @defer.inlineCallbacks + def test_reupload_signatures(self): + """re-uploading a signature should not fail""" + local_user = "@boris:" + self.hs.hostname + keys1 = { + "master_key": { + # private key: HvQBbU+hc2Zr+JP1sE0XwBe1pfZZEYtJNPJLZJtS+F8 + "user_id": local_user, + "usage": ["master"], + "keys": { + "ed25519:EmkqvokUn8p+vQAGZitOk4PWjp7Ukp3txV2TbMPEiBQ": "EmkqvokUn8p+vQAGZitOk4PWjp7Ukp3txV2TbMPEiBQ" + }, + }, + "self_signing_key": { + # private key: 2lonYOM6xYKdEsO+6KrC766xBcHnYnim1x/4LFGF8B0 + "user_id": local_user, + "usage": ["self_signing"], + "keys": { + "ed25519:nqOvzeuGWT/sRx3h7+MHoInYj3Uk2LD/unI9kDYcHwk": "nqOvzeuGWT/sRx3h7+MHoInYj3Uk2LD/unI9kDYcHwk" + }, + }, + } + master_signing_key = key.decode_signing_key_base64( + "ed25519", + "EmkqvokUn8p+vQAGZitOk4PWjp7Ukp3txV2TbMPEiBQ", + "HvQBbU+hc2Zr+JP1sE0XwBe1pfZZEYtJNPJLZJtS+F8", + ) + sign.sign_json(keys1["self_signing_key"], local_user, master_signing_key) + signing_key = key.decode_signing_key_base64( + "ed25519", + "nqOvzeuGWT/sRx3h7+MHoInYj3Uk2LD/unI9kDYcHwk", + "2lonYOM6xYKdEsO+6KrC766xBcHnYnim1x/4LFGF8B0", + ) + yield self.handler.upload_signing_keys_for_user(local_user, keys1) + + # upload two device keys, which will be signed later by the self-signing key + device_key_1 = { + "user_id": local_user, + "device_id": "abc", + "algorithms": ["m.olm.curve25519-aes-sha256", "m.megolm.v1.aes-sha"], + "keys": { + "ed25519:abc": "base64+ed25519+key", + "curve25519:abc": "base64+curve25519+key", + }, + "signatures": {local_user: {"ed25519:abc": "base64+signature"}}, + } + device_key_2 = { + "user_id": local_user, + "device_id": "def", + "algorithms": ["m.olm.curve25519-aes-sha256", "m.megolm.v1.aes-sha"], + "keys": { + "ed25519:def": "base64+ed25519+key", + "curve25519:def": "base64+curve25519+key", + }, + "signatures": {local_user: {"ed25519:def": "base64+signature"}}, + } + + yield self.handler.upload_keys_for_user( + local_user, "abc", {"device_keys": device_key_1} + ) + yield self.handler.upload_keys_for_user( + local_user, "def", {"device_keys": device_key_2} + ) + + # sign the first device key and upload it + del device_key_1["signatures"] + sign.sign_json(device_key_1, local_user, signing_key) + yield self.handler.upload_signatures_for_device_keys( + local_user, {local_user: {"abc": device_key_1}} + ) + + # sign the second device key and upload both device keys. The server + # should ignore the first device key since it already has a valid + # signature for it + del device_key_2["signatures"] + sign.sign_json(device_key_2, local_user, signing_key) + yield self.handler.upload_signatures_for_device_keys( + local_user, {local_user: {"abc": device_key_1, "def": device_key_2}} + ) + + device_key_1["signatures"][local_user]["ed25519:abc"] = "base64+signature" + device_key_2["signatures"][local_user]["ed25519:def"] = "base64+signature" + devices = yield self.handler.query_devices({"device_keys": {local_user: []}}, 0) + del devices["device_keys"][local_user]["abc"]["unsigned"] + del devices["device_keys"][local_user]["def"]["unsigned"] + self.assertDictEqual(devices["device_keys"][local_user]["abc"], device_key_1) + self.assertDictEqual(devices["device_keys"][local_user]["def"], device_key_2) + @defer.inlineCallbacks def test_self_signing_key_doesnt_show_up_as_device(self): """signing keys should be hidden when fetching a user's devices""" -- cgit 1.4.1 From e47af0f086839c5d22a0de87a32a49386abef8df Mon Sep 17 00:00:00 2001 From: Hubert Chathi Date: Thu, 5 Sep 2019 17:03:14 -0400 Subject: fix test --- tests/handlers/test_e2e_keys.py | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'tests') diff --git a/tests/handlers/test_e2e_keys.py b/tests/handlers/test_e2e_keys.py index 316dd6259d..7a59ec5085 100644 --- a/tests/handlers/test_e2e_keys.py +++ b/tests/handlers/test_e2e_keys.py @@ -265,7 +265,9 @@ class E2eKeysHandlerTestCase(unittest.TestCase): device_key_1["signatures"][local_user]["ed25519:abc"] = "base64+signature" device_key_2["signatures"][local_user]["ed25519:def"] = "base64+signature" - devices = yield self.handler.query_devices({"device_keys": {local_user: []}}, 0) + devices = yield self.handler.query_devices( + {"device_keys": {local_user: []}}, 0, 0 + ) del devices["device_keys"][local_user]["abc"]["unsigned"] del devices["device_keys"][local_user]["def"]["unsigned"] self.assertDictEqual(devices["device_keys"][local_user]["abc"], device_key_1) -- cgit 1.4.1 From ab729e31cfca4d1a958937bb576010271b9c8044 Mon Sep 17 00:00:00 2001 From: Hubert Chathi Date: Fri, 6 Sep 2019 17:52:37 -0400 Subject: use something that's the right type for user_id --- tests/handlers/test_e2e_keys.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tests') diff --git a/tests/handlers/test_e2e_keys.py b/tests/handlers/test_e2e_keys.py index 7a59ec5085..854eb6c024 100644 --- a/tests/handlers/test_e2e_keys.py +++ b/tests/handlers/test_e2e_keys.py @@ -266,7 +266,7 @@ class E2eKeysHandlerTestCase(unittest.TestCase): device_key_1["signatures"][local_user]["ed25519:abc"] = "base64+signature" device_key_2["signatures"][local_user]["ed25519:def"] = "base64+signature" devices = yield self.handler.query_devices( - {"device_keys": {local_user: []}}, 0, 0 + {"device_keys": {local_user: []}}, 0, local_user ) del devices["device_keys"][local_user]["abc"]["unsigned"] del devices["device_keys"][local_user]["def"]["unsigned"] -- cgit 1.4.1 From 51d28272e20d799b2e35a8a14b3c1d9d5f555d10 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Mon, 23 Sep 2019 16:00:18 +0100 Subject: Query devices table for last seen info. This is a) simpler than querying user_ips directly and b) means we can purge older entries from user_ips without losing the required info. The storage functions now no longer return the access_token, since it was unused. --- synapse/storage/client_ips.py | 57 ++++++---------------------------------- tests/storage/test_client_ips.py | 1 - 2 files changed, 8 insertions(+), 50 deletions(-) (limited to 'tests') diff --git a/synapse/storage/client_ips.py b/synapse/storage/client_ips.py index 8839562269..a4e6d9dbe7 100644 --- a/synapse/storage/client_ips.py +++ b/synapse/storage/client_ips.py @@ -392,19 +392,14 @@ class ClientIpStore(background_updates.BackgroundUpdateStore): keys giving the column names """ - res = yield self.runInteraction( - "get_last_client_ip_by_device", - self._get_last_client_ip_by_device_txn, - user_id, - device_id, - retcols=( - "user_id", - "access_token", - "ip", - "user_agent", - "device_id", - "last_seen", - ), + keyvalues = {"user_id": user_id} + if device_id: + keyvalues["device_id"] = device_id + + res = yield self._simple_select_list( + table="devices", + keyvalues=keyvalues, + retcols=("user_id", "ip", "user_agent", "device_id", "last_seen"), ) ret = {(d["user_id"], d["device_id"]): d for d in res} @@ -423,42 +418,6 @@ class ClientIpStore(background_updates.BackgroundUpdateStore): } return ret - @classmethod - def _get_last_client_ip_by_device_txn(cls, txn, user_id, device_id, retcols): - where_clauses = [] - bindings = [] - if device_id is None: - where_clauses.append("user_id = ?") - bindings.extend((user_id,)) - else: - where_clauses.append("(user_id = ? AND device_id = ?)") - bindings.extend((user_id, device_id)) - - if not where_clauses: - return [] - - inner_select = ( - "SELECT MAX(last_seen) mls, user_id, device_id FROM user_ips " - "WHERE %(where)s " - "GROUP BY user_id, device_id" - ) % {"where": " OR ".join(where_clauses)} - - sql = ( - "SELECT %(retcols)s FROM user_ips " - "JOIN (%(inner_select)s) ips ON" - " user_ips.last_seen = ips.mls AND" - " user_ips.user_id = ips.user_id AND" - " (user_ips.device_id = ips.device_id OR" - " (user_ips.device_id IS NULL AND ips.device_id IS NULL)" - " )" - ) % { - "retcols": ",".join("user_ips." + c for c in retcols), - "inner_select": inner_select, - } - - txn.execute(sql, bindings) - return cls.cursor_to_dict(txn) - @defer.inlineCallbacks def get_user_ip_and_agents(self, user): user_id = user.to_string() diff --git a/tests/storage/test_client_ips.py b/tests/storage/test_client_ips.py index 09305c3bf1..6ac4654085 100644 --- a/tests/storage/test_client_ips.py +++ b/tests/storage/test_client_ips.py @@ -55,7 +55,6 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): { "user_id": user_id, "device_id": "device_id", - "access_token": "access_token", "ip": "ip", "user_agent": "user_agent", "last_seen": 12345678000, -- cgit 1.4.1 From acb62a7cc6973618397a868289b5881f1c3c1ec3 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Mon, 23 Sep 2019 16:50:31 +0100 Subject: Test background update --- tests/storage/test_client_ips.py | 79 ++++++++++++++++++++++++++++++++++++++++ 1 file changed, 79 insertions(+) (limited to 'tests') diff --git a/tests/storage/test_client_ips.py b/tests/storage/test_client_ips.py index 6ac4654085..76fe65b59e 100644 --- a/tests/storage/test_client_ips.py +++ b/tests/storage/test_client_ips.py @@ -200,6 +200,85 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): active = self.get_success(self.store.user_last_seen_monthly_active(user_id)) self.assertTrue(active) + def test_devices_last_seen_bg_update(self): + # First make sure we have completed all updates. + while not self.get_success(self.store.has_completed_background_updates()): + self.get_success(self.store.do_next_background_update(100), by=0.1) + + # Insert a user IP + user_id = "@user:id" + self.get_success( + self.store.insert_client_ip( + user_id, "access_token", "ip", "user_agent", "device_id" + ) + ) + + # Force persisting to disk + self.reactor.advance(200) + + # But clear the associated entry in devices table + self.get_success( + self.store._simple_update( + table="devices", + keyvalues={"user_id": user_id, "device_id": "device_id"}, + updatevalues={"last_seen": None, "ip": None, "user_agent": None}, + desc="test_devices_last_seen_bg_update", + ) + ) + + # We should now get nulls when querying + result = self.get_success( + self.store.get_last_client_ip_by_device(user_id, "device_id") + ) + + r = result[(user_id, "device_id")] + self.assertDictContainsSubset( + { + "user_id": user_id, + "device_id": "device_id", + "ip": None, + "user_agent": None, + "last_seen": None, + }, + r, + ) + + # Register the background update to run again. + self.get_success( + self.store._simple_insert( + table="background_updates", + values={ + "update_name": "devices_last_seen", + "progress_json": "{}", + "depends_on": None, + }, + ) + ) + + # ... and tell the DataStore that it hasn't finished all updates yet + self.store._all_done = False + + # Now let's actually drive the updates to completion + while not self.get_success(self.store.has_completed_background_updates()): + self.get_success(self.store.do_next_background_update(100), by=0.1) + + # We should now get the correct result again + result = self.get_success( + self.store.get_last_client_ip_by_device(user_id, "device_id") + ) + + r = result[(user_id, "device_id")] + self.assertDictContainsSubset( + { + "user_id": user_id, + "device_id": "device_id", + "ip": "ip", + "user_agent": "user_agent", + "last_seen": 0, + }, + r, + ) + class ClientIpAuthTestCase(unittest.HomeserverTestCase): -- cgit 1.4.1 From faac453f08046ddf00b39b90ba255f774b75c253 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Tue, 24 Sep 2019 15:51:42 +0100 Subject: Test that pruning of old user IPs works --- tests/storage/test_client_ips.py | 71 ++++++++++++++++++++++++++++++++++++++++ 1 file changed, 71 insertions(+) (limited to 'tests') diff --git a/tests/storage/test_client_ips.py b/tests/storage/test_client_ips.py index 76fe65b59e..afac5dec7f 100644 --- a/tests/storage/test_client_ips.py +++ b/tests/storage/test_client_ips.py @@ -279,6 +279,77 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): r, ) + def test_old_user_ips_pruned(self): + # First make sure we have completed all updates. + while not self.get_success(self.store.has_completed_background_updates()): + self.get_success(self.store.do_next_background_update(100), by=0.1) + + # Insert a user IP + user_id = "@user:id" + self.get_success( + self.store.insert_client_ip( + user_id, "access_token", "ip", "user_agent", "device_id" + ) + ) + + # Force persisting to disk + self.reactor.advance(200) + + # We should see that in the DB + result = self.get_success( + self.store._simple_select_list( + table="user_ips", + keyvalues={"user_id": user_id}, + retcols=["access_token", "ip", "user_agent", "device_id", "last_seen"], + desc="get_user_ip_and_agents", + ) + ) + + self.assertEqual( + result, + [ + { + "access_token": "access_token", + "ip": "ip", + "user_agent": "user_agent", + "device_id": "device_id", + "last_seen": 0, + } + ], + ) + + # Now advance by a couple of months + self.reactor.advance(60 * 24 * 60 * 60) + + # We should get no results. + result = self.get_success( + self.store._simple_select_list( + table="user_ips", + keyvalues={"user_id": user_id}, + retcols=["access_token", "ip", "user_agent", "device_id", "last_seen"], + desc="get_user_ip_and_agents", + ) + ) + + self.assertEqual(result, []) + + # But we should still get the correct values for the device + result = self.get_success( + self.store.get_last_client_ip_by_device(user_id, "device_id") + ) + + r = result[(user_id, "device_id")] + self.assertDictContainsSubset( + { + "user_id": user_id, + "device_id": "device_id", + "ip": "ip", + "user_agent": "user_agent", + "last_seen": 0, + }, + r, + ) + class ClientIpAuthTestCase(unittest.HomeserverTestCase): -- cgit 1.4.1 From 8004d6ca2faf0f2f843fcdcaf225d7bcab847503 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Wed, 25 Sep 2019 11:32:05 +0100 Subject: Refactor code for calculating registration flows (#6106) because, frankly, it looked like it was written by an axe-murderer. This should be a non-functional change, except that where `m.login.dummy` was previously advertised *before* `m.login.terms`, it will now be advertised afterwards. AFAICT that should have no effect, and will be more consistent with the flows that involve passing a 3pid. --- changelog.d/6106.misc | 1 + synapse/rest/client/v2_alpha/register.py | 124 ++++++++++++++-------------- tests/rest/client/v2_alpha/test_register.py | 79 +++++++++++++++--- tests/test_terms_auth.py | 24 ++++-- 4 files changed, 145 insertions(+), 83 deletions(-) create mode 100644 changelog.d/6106.misc (limited to 'tests') diff --git a/changelog.d/6106.misc b/changelog.d/6106.misc new file mode 100644 index 0000000000..d732091779 --- /dev/null +++ b/changelog.d/6106.misc @@ -0,0 +1 @@ +Refactor code for calculating registration flows. diff --git a/synapse/rest/client/v2_alpha/register.py b/synapse/rest/client/v2_alpha/register.py index 135a70808f..e3f3d9126f 100644 --- a/synapse/rest/client/v2_alpha/register.py +++ b/synapse/rest/client/v2_alpha/register.py @@ -16,6 +16,7 @@ import hmac import logging +from typing import List, Union from six import string_types @@ -31,8 +32,11 @@ from synapse.api.errors import ( ThreepidValidationError, UnrecognizedRequestError, ) +from synapse.config.captcha import CaptchaConfig +from synapse.config.consent_config import ConsentConfig from synapse.config.emailconfig import ThreepidBehaviour from synapse.config.ratelimiting import FederationRateLimitConfig +from synapse.config.registration import RegistrationConfig from synapse.config.server import is_threepid_reserved from synapse.http.server import finish_request from synapse.http.servlet import ( @@ -371,6 +375,8 @@ class RegisterRestServlet(RestServlet): self.ratelimiter = hs.get_registration_ratelimiter() self.clock = hs.get_clock() + self._registration_flows = _calculate_registration_flows(hs.config) + @interactive_auth_handler @defer.inlineCallbacks def on_POST(self, request): @@ -491,69 +497,8 @@ class RegisterRestServlet(RestServlet): assigned_user_id=registered_user_id, ) - # FIXME: need a better error than "no auth flow found" for scenarios - # where we required 3PID for registration but the user didn't give one - require_email = "email" in self.hs.config.registrations_require_3pid - require_msisdn = "msisdn" in self.hs.config.registrations_require_3pid - - show_msisdn = True - if self.hs.config.disable_msisdn_registration: - show_msisdn = False - require_msisdn = False - - flows = [] - if self.hs.config.enable_registration_captcha: - # only support 3PIDless registration if no 3PIDs are required - if not require_email and not require_msisdn: - # Also add a dummy flow here, otherwise if a client completes - # recaptcha first we'll assume they were going for this flow - # and complete the request, when they could have been trying to - # complete one of the flows with email/msisdn auth. - flows.extend([[LoginType.RECAPTCHA, LoginType.DUMMY]]) - # only support the email-only flow if we don't require MSISDN 3PIDs - if not require_msisdn: - flows.extend([[LoginType.RECAPTCHA, LoginType.EMAIL_IDENTITY]]) - - if show_msisdn: - # only support the MSISDN-only flow if we don't require email 3PIDs - if not require_email: - flows.extend([[LoginType.RECAPTCHA, LoginType.MSISDN]]) - # always let users provide both MSISDN & email - flows.extend( - [[LoginType.RECAPTCHA, LoginType.MSISDN, LoginType.EMAIL_IDENTITY]] - ) - else: - # only support 3PIDless registration if no 3PIDs are required - if not require_email and not require_msisdn: - flows.extend([[LoginType.DUMMY]]) - # only support the email-only flow if we don't require MSISDN 3PIDs - if not require_msisdn: - flows.extend([[LoginType.EMAIL_IDENTITY]]) - - if show_msisdn: - # only support the MSISDN-only flow if we don't require email 3PIDs - if not require_email or require_msisdn: - flows.extend([[LoginType.MSISDN]]) - # always let users provide both MSISDN & email - flows.extend([[LoginType.MSISDN, LoginType.EMAIL_IDENTITY]]) - - # Append m.login.terms to all flows if we're requiring consent - if self.hs.config.user_consent_at_registration: - new_flows = [] - for flow in flows: - inserted = False - # m.login.terms should go near the end but before msisdn or email auth - for i, stage in enumerate(flow): - if stage == LoginType.EMAIL_IDENTITY or stage == LoginType.MSISDN: - flow.insert(i, LoginType.TERMS) - inserted = True - break - if not inserted: - flow.append(LoginType.TERMS) - flows.extend(new_flows) - auth_result, params, session_id = yield self.auth_handler.check_auth( - flows, body, self.hs.get_ip_from_request(request) + self._registration_flows, body, self.hs.get_ip_from_request(request) ) # Check that we're not trying to register a denied 3pid. @@ -716,6 +661,61 @@ class RegisterRestServlet(RestServlet): ) +def _calculate_registration_flows( + # technically `config` has to provide *all* of these interfaces, not just one + config: Union[RegistrationConfig, ConsentConfig, CaptchaConfig], +) -> List[List[str]]: + """Get a suitable flows list for registration + + Args: + config: server configuration + + Returns: a list of supported flows + """ + # FIXME: need a better error than "no auth flow found" for scenarios + # where we required 3PID for registration but the user didn't give one + require_email = "email" in config.registrations_require_3pid + require_msisdn = "msisdn" in config.registrations_require_3pid + + show_msisdn = True + if config.disable_msisdn_registration: + show_msisdn = False + require_msisdn = False + + flows = [] + + # only support 3PIDless registration if no 3PIDs are required + if not require_email and not require_msisdn: + # Add a dummy step here, otherwise if a client completes + # recaptcha first we'll assume they were going for this flow + # and complete the request, when they could have been trying to + # complete one of the flows with email/msisdn auth. + flows.append([LoginType.DUMMY]) + + # only support the email-only flow if we don't require MSISDN 3PIDs + if not require_msisdn: + flows.append([LoginType.EMAIL_IDENTITY]) + + # only support the MSISDN-only flow if we don't require email 3PIDs + if show_msisdn and not require_email: + flows.append([LoginType.MSISDN]) + + if show_msisdn: + flows.append([LoginType.MSISDN, LoginType.EMAIL_IDENTITY]) + + # Prepend m.login.terms to all flows if we're requiring consent + if config.user_consent_at_registration: + for flow in flows: + flow.insert(0, LoginType.TERMS) + + # Prepend recaptcha to all flows if we're requiring captcha + if config.enable_registration_captcha: + for flow in flows: + flow.insert(0, LoginType.RECAPTCHA) + + return flows + + def register_servlets(hs, http_server): EmailRegisterRequestTokenRestServlet(hs).register(http_server) MsisdnRegisterRequestTokenRestServlet(hs).register(http_server) diff --git a/tests/rest/client/v2_alpha/test_register.py b/tests/rest/client/v2_alpha/test_register.py index ab4d7d70d0..bc2dc47973 100644 --- a/tests/rest/client/v2_alpha/test_register.py +++ b/tests/rest/client/v2_alpha/test_register.py @@ -34,19 +34,12 @@ from tests import unittest class RegisterRestServletTestCase(unittest.HomeserverTestCase): servlets = [register.register_servlets] + url = b"/_matrix/client/r0/register" - def make_homeserver(self, reactor, clock): - - self.url = b"/_matrix/client/r0/register" - - self.hs = self.setup_test_homeserver() - self.hs.config.enable_registration = True - self.hs.config.registrations_require_3pid = [] - self.hs.config.auto_join_rooms = [] - self.hs.config.enable_registration_captcha = False - self.hs.config.allow_guest_access = True - - return self.hs + def default_config(self, name="test"): + config = super().default_config(name) + config["allow_guest_access"] = True + return config def test_POST_appservice_registration_valid(self): user_id = "@as_user_kermit:test" @@ -199,6 +192,68 @@ class RegisterRestServletTestCase(unittest.HomeserverTestCase): self.assertEquals(channel.result["code"], b"200", channel.result) + def test_advertised_flows(self): + request, channel = self.make_request(b"POST", self.url, b"{}") + self.render(request) + self.assertEquals(channel.result["code"], b"401", channel.result) + flows = channel.json_body["flows"] + + # with the stock config, we expect all four combinations of 3pid + self.assertCountEqual( + [ + ["m.login.dummy"], + ["m.login.email.identity"], + ["m.login.msisdn"], + ["m.login.msisdn", "m.login.email.identity"], + ], + (f["stages"] for f in flows), + ) + + @unittest.override_config( + { + "enable_registration_captcha": True, + "user_consent": { + "version": "1", + "template_dir": "/", + "require_at_registration": True, + }, + } + ) + def test_advertised_flows_captcha_and_terms(self): + request, channel = self.make_request(b"POST", self.url, b"{}") + self.render(request) + self.assertEquals(channel.result["code"], b"401", channel.result) + flows = channel.json_body["flows"] + + self.assertCountEqual( + [ + ["m.login.recaptcha", "m.login.terms", "m.login.dummy"], + ["m.login.recaptcha", "m.login.terms", "m.login.email.identity"], + ["m.login.recaptcha", "m.login.terms", "m.login.msisdn"], + [ + "m.login.recaptcha", + "m.login.terms", + "m.login.msisdn", + "m.login.email.identity", + ], + ], + (f["stages"] for f in flows), + ) + + @unittest.override_config( + {"registrations_require_3pid": ["email"], "disable_msisdn_registration": True} + ) + def test_advertised_flows_no_msisdn_email_required(self): + request, channel = self.make_request(b"POST", self.url, b"{}") + self.render(request) + self.assertEquals(channel.result["code"], b"401", channel.result) + flows = channel.json_body["flows"] + + # with the stock config, we expect all four combinations of 3pid + self.assertCountEqual( + [["m.login.email.identity"]], (f["stages"] for f in flows) + ) + class AccountValidityTestCase(unittest.HomeserverTestCase): diff --git a/tests/test_terms_auth.py b/tests/test_terms_auth.py index 52739fbabc..5ec5d2b358 100644 --- a/tests/test_terms_auth.py +++ b/tests/test_terms_auth.py @@ -28,6 +28,21 @@ from tests import unittest class TermsTestCase(unittest.HomeserverTestCase): servlets = [register_servlets] + def default_config(self, name="test"): + config = super().default_config(name) + config.update( + { + "public_baseurl": "https://example.org/", + "user_consent": { + "version": "1.0", + "policy_name": "My Cool Privacy Policy", + "template_dir": "/", + "require_at_registration": True, + }, + } + ) + return config + def prepare(self, reactor, clock, hs): self.clock = MemoryReactorClock() self.hs_clock = Clock(self.clock) @@ -35,17 +50,8 @@ class TermsTestCase(unittest.HomeserverTestCase): self.registration_handler = Mock() self.auth_handler = Mock() self.device_handler = Mock() - hs.config.enable_registration = True - hs.config.registrations_require_3pid = [] - hs.config.auto_join_rooms = [] - hs.config.enable_registration_captcha = False def test_ui_auth(self): - self.hs.config.user_consent_at_registration = True - self.hs.config.user_consent_policy_name = "My Cool Privacy Policy" - self.hs.config.public_baseurl = "https://example.org/" - self.hs.config.user_consent_version = "1.0" - # Do a UI auth request request, channel = self.make_request(b"POST", self.url, b"{}") self.render(request) -- cgit 1.4.1 From 2cd98812ba338eefe83fee4ae2390d00f5499de9 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Wed, 25 Sep 2019 11:33:03 +0100 Subject: Refactor the user-interactive auth handling (#6105) Pull the checkers out to their own classes, rather than having them lost in a massive 1000-line class which does everything. This is also preparation for some more intelligent advertising of flows, as per #6100 --- changelog.d/6105.misc | 1 + synapse/handlers/auth.py | 141 ++------------------- synapse/handlers/ui_auth/__init__.py | 22 ++++ synapse/handlers/ui_auth/checkers.py | 216 ++++++++++++++++++++++++++++++++ tests/rest/client/v2_alpha/test_auth.py | 26 ++-- 5 files changed, 265 insertions(+), 141 deletions(-) create mode 100644 changelog.d/6105.misc create mode 100644 synapse/handlers/ui_auth/__init__.py create mode 100644 synapse/handlers/ui_auth/checkers.py (limited to 'tests') diff --git a/changelog.d/6105.misc b/changelog.d/6105.misc new file mode 100644 index 0000000000..2e838a35c6 --- /dev/null +++ b/changelog.d/6105.misc @@ -0,0 +1 @@ +Refactor the user-interactive auth handling. diff --git a/synapse/handlers/auth.py b/synapse/handlers/auth.py index 374372b69e..f920c2f6c1 100644 --- a/synapse/handlers/auth.py +++ b/synapse/handlers/auth.py @@ -21,10 +21,8 @@ import unicodedata import attr import bcrypt import pymacaroons -from canonicaljson import json from twisted.internet import defer -from twisted.web.client import PartialDownloadError import synapse.util.stringutils as stringutils from synapse.api.constants import LoginType @@ -38,7 +36,8 @@ from synapse.api.errors import ( UserDeactivatedError, ) from synapse.api.ratelimiting import Ratelimiter -from synapse.config.emailconfig import ThreepidBehaviour +from synapse.handlers.ui_auth import INTERACTIVE_AUTH_CHECKERS +from synapse.handlers.ui_auth.checkers import UserInteractiveAuthChecker from synapse.logging.context import defer_to_thread from synapse.module_api import ModuleApi from synapse.types import UserID @@ -58,13 +57,12 @@ class AuthHandler(BaseHandler): hs (synapse.server.HomeServer): """ super(AuthHandler, self).__init__(hs) - self.checkers = { - LoginType.RECAPTCHA: self._check_recaptcha, - LoginType.EMAIL_IDENTITY: self._check_email_identity, - LoginType.MSISDN: self._check_msisdn, - LoginType.DUMMY: self._check_dummy_auth, - LoginType.TERMS: self._check_terms_auth, - } + + self.checkers = {} # type: dict[str, UserInteractiveAuthChecker] + for auth_checker_class in INTERACTIVE_AUTH_CHECKERS: + inst = auth_checker_class(hs) + self.checkers[inst.AUTH_TYPE] = inst + self.bcrypt_rounds = hs.config.bcrypt_rounds # This is not a cache per se, but a store of all current sessions that @@ -292,7 +290,7 @@ class AuthHandler(BaseHandler): sess["creds"] = {} creds = sess["creds"] - result = yield self.checkers[stagetype](authdict, clientip) + result = yield self.checkers[stagetype].check_auth(authdict, clientip) if result: creds[stagetype] = result self._save_session(sess) @@ -363,7 +361,7 @@ class AuthHandler(BaseHandler): login_type = authdict["type"] checker = self.checkers.get(login_type) if checker is not None: - res = yield checker(authdict, clientip=clientip) + res = yield checker.check_auth(authdict, clientip=clientip) return res # build a v1-login-style dict out of the authdict and fall back to the @@ -376,125 +374,6 @@ class AuthHandler(BaseHandler): (canonical_id, callback) = yield self.validate_login(user_id, authdict) return canonical_id - @defer.inlineCallbacks - def _check_recaptcha(self, authdict, clientip, **kwargs): - try: - user_response = authdict["response"] - except KeyError: - # Client tried to provide captcha but didn't give the parameter: - # bad request. - raise LoginError( - 400, "Captcha response is required", errcode=Codes.CAPTCHA_NEEDED - ) - - logger.info( - "Submitting recaptcha response %s with remoteip %s", user_response, clientip - ) - - # TODO: get this from the homeserver rather than creating a new one for - # each request - try: - client = self.hs.get_simple_http_client() - resp_body = yield client.post_urlencoded_get_json( - self.hs.config.recaptcha_siteverify_api, - args={ - "secret": self.hs.config.recaptcha_private_key, - "response": user_response, - "remoteip": clientip, - }, - ) - except PartialDownloadError as pde: - # Twisted is silly - data = pde.response - resp_body = json.loads(data) - - if "success" in resp_body: - # Note that we do NOT check the hostname here: we explicitly - # intend the CAPTCHA to be presented by whatever client the - # user is using, we just care that they have completed a CAPTCHA. - logger.info( - "%s reCAPTCHA from hostname %s", - "Successful" if resp_body["success"] else "Failed", - resp_body.get("hostname"), - ) - if resp_body["success"]: - return True - raise LoginError(401, "", errcode=Codes.UNAUTHORIZED) - - def _check_email_identity(self, authdict, **kwargs): - return self._check_threepid("email", authdict, **kwargs) - - def _check_msisdn(self, authdict, **kwargs): - return self._check_threepid("msisdn", authdict) - - def _check_dummy_auth(self, authdict, **kwargs): - return defer.succeed(True) - - def _check_terms_auth(self, authdict, **kwargs): - return defer.succeed(True) - - @defer.inlineCallbacks - def _check_threepid(self, medium, authdict, **kwargs): - if "threepid_creds" not in authdict: - raise LoginError(400, "Missing threepid_creds", Codes.MISSING_PARAM) - - threepid_creds = authdict["threepid_creds"] - - identity_handler = self.hs.get_handlers().identity_handler - - logger.info("Getting validated threepid. threepidcreds: %r", (threepid_creds,)) - if self.hs.config.threepid_behaviour_email == ThreepidBehaviour.REMOTE: - if medium == "email": - threepid = yield identity_handler.threepid_from_creds( - self.hs.config.account_threepid_delegate_email, threepid_creds - ) - elif medium == "msisdn": - threepid = yield identity_handler.threepid_from_creds( - self.hs.config.account_threepid_delegate_msisdn, threepid_creds - ) - else: - raise SynapseError(400, "Unrecognized threepid medium: %s" % (medium,)) - elif self.hs.config.threepid_behaviour_email == ThreepidBehaviour.LOCAL: - row = yield self.store.get_threepid_validation_session( - medium, - threepid_creds["client_secret"], - sid=threepid_creds["sid"], - validated=True, - ) - - threepid = ( - { - "medium": row["medium"], - "address": row["address"], - "validated_at": row["validated_at"], - } - if row - else None - ) - - if row: - # Valid threepid returned, delete from the db - yield self.store.delete_threepid_session(threepid_creds["sid"]) - else: - raise SynapseError( - 400, "Password resets are not enabled on this homeserver" - ) - - if not threepid: - raise LoginError(401, "", errcode=Codes.UNAUTHORIZED) - - if threepid["medium"] != medium: - raise LoginError( - 401, - "Expecting threepid of type '%s', got '%s'" - % (medium, threepid["medium"]), - errcode=Codes.UNAUTHORIZED, - ) - - threepid["threepid_creds"] = authdict["threepid_creds"] - - return threepid - def _get_params_recaptcha(self): return {"public_key": self.hs.config.recaptcha_public_key} diff --git a/synapse/handlers/ui_auth/__init__.py b/synapse/handlers/ui_auth/__init__.py new file mode 100644 index 0000000000..824f37f8f8 --- /dev/null +++ b/synapse/handlers/ui_auth/__init__.py @@ -0,0 +1,22 @@ +# -*- coding: utf-8 -*- +# Copyright 2019 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +"""This module implements user-interactive auth verification. + +TODO: move more stuff out of AuthHandler in here. + +""" + +from synapse.handlers.ui_auth.checkers import INTERACTIVE_AUTH_CHECKERS # noqa: F401 diff --git a/synapse/handlers/ui_auth/checkers.py b/synapse/handlers/ui_auth/checkers.py new file mode 100644 index 0000000000..fd633b7b0e --- /dev/null +++ b/synapse/handlers/ui_auth/checkers.py @@ -0,0 +1,216 @@ +# -*- coding: utf-8 -*- +# Copyright 2019 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import logging + +from canonicaljson import json + +from twisted.internet import defer +from twisted.web.client import PartialDownloadError + +from synapse.api.constants import LoginType +from synapse.api.errors import Codes, LoginError, SynapseError +from synapse.config.emailconfig import ThreepidBehaviour + +logger = logging.getLogger(__name__) + + +class UserInteractiveAuthChecker: + """Abstract base class for an interactive auth checker""" + + def __init__(self, hs): + pass + + def check_auth(self, authdict, clientip): + """Given the authentication dict from the client, attempt to check this step + + Args: + authdict (dict): authentication dictionary from the client + clientip (str): The IP address of the client. + + Raises: + SynapseError if authentication failed + + Returns: + Deferred: the result of authentication (to pass back to the client?) + """ + raise NotImplementedError() + + +class DummyAuthChecker(UserInteractiveAuthChecker): + AUTH_TYPE = LoginType.DUMMY + + def check_auth(self, authdict, clientip): + return defer.succeed(True) + + +class TermsAuthChecker(UserInteractiveAuthChecker): + AUTH_TYPE = LoginType.TERMS + + def check_auth(self, authdict, clientip): + return defer.succeed(True) + + +class RecaptchaAuthChecker(UserInteractiveAuthChecker): + AUTH_TYPE = LoginType.RECAPTCHA + + def __init__(self, hs): + super().__init__(hs) + self._http_client = hs.get_simple_http_client() + self._url = hs.config.recaptcha_siteverify_api + self._secret = hs.config.recaptcha_private_key + + @defer.inlineCallbacks + def check_auth(self, authdict, clientip): + try: + user_response = authdict["response"] + except KeyError: + # Client tried to provide captcha but didn't give the parameter: + # bad request. + raise LoginError( + 400, "Captcha response is required", errcode=Codes.CAPTCHA_NEEDED + ) + + logger.info( + "Submitting recaptcha response %s with remoteip %s", user_response, clientip + ) + + # TODO: get this from the homeserver rather than creating a new one for + # each request + try: + resp_body = yield self._http_client.post_urlencoded_get_json( + self._url, + args={ + "secret": self._secret, + "response": user_response, + "remoteip": clientip, + }, + ) + except PartialDownloadError as pde: + # Twisted is silly + data = pde.response + resp_body = json.loads(data) + + if "success" in resp_body: + # Note that we do NOT check the hostname here: we explicitly + # intend the CAPTCHA to be presented by whatever client the + # user is using, we just care that they have completed a CAPTCHA. + logger.info( + "%s reCAPTCHA from hostname %s", + "Successful" if resp_body["success"] else "Failed", + resp_body.get("hostname"), + ) + if resp_body["success"]: + return True + raise LoginError(401, "", errcode=Codes.UNAUTHORIZED) + + +class _BaseThreepidAuthChecker: + def __init__(self, hs): + self.hs = hs + self.store = hs.get_datastore() + + @defer.inlineCallbacks + def _check_threepid(self, medium, authdict): + if "threepid_creds" not in authdict: + raise LoginError(400, "Missing threepid_creds", Codes.MISSING_PARAM) + + threepid_creds = authdict["threepid_creds"] + + identity_handler = self.hs.get_handlers().identity_handler + + logger.info("Getting validated threepid. threepidcreds: %r", (threepid_creds,)) + if self.hs.config.threepid_behaviour_email == ThreepidBehaviour.REMOTE: + if medium == "email": + threepid = yield identity_handler.threepid_from_creds( + self.hs.config.account_threepid_delegate_email, threepid_creds + ) + elif medium == "msisdn": + threepid = yield identity_handler.threepid_from_creds( + self.hs.config.account_threepid_delegate_msisdn, threepid_creds + ) + else: + raise SynapseError(400, "Unrecognized threepid medium: %s" % (medium,)) + elif self.hs.config.threepid_behaviour_email == ThreepidBehaviour.LOCAL: + row = yield self.store.get_threepid_validation_session( + medium, + threepid_creds["client_secret"], + sid=threepid_creds["sid"], + validated=True, + ) + + threepid = ( + { + "medium": row["medium"], + "address": row["address"], + "validated_at": row["validated_at"], + } + if row + else None + ) + + if row: + # Valid threepid returned, delete from the db + yield self.store.delete_threepid_session(threepid_creds["sid"]) + else: + raise SynapseError( + 400, "Password resets are not enabled on this homeserver" + ) + + if not threepid: + raise LoginError(401, "", errcode=Codes.UNAUTHORIZED) + + if threepid["medium"] != medium: + raise LoginError( + 401, + "Expecting threepid of type '%s', got '%s'" + % (medium, threepid["medium"]), + errcode=Codes.UNAUTHORIZED, + ) + + threepid["threepid_creds"] = authdict["threepid_creds"] + + return threepid + + +class EmailIdentityAuthChecker(UserInteractiveAuthChecker, _BaseThreepidAuthChecker): + AUTH_TYPE = LoginType.EMAIL_IDENTITY + + def __init__(self, hs): + UserInteractiveAuthChecker.__init__(self, hs) + _BaseThreepidAuthChecker.__init__(self, hs) + + def check_auth(self, authdict, clientip): + return self._check_threepid("email", authdict) + + +class MsisdnAuthChecker(UserInteractiveAuthChecker, _BaseThreepidAuthChecker): + AUTH_TYPE = LoginType.MSISDN + + def __init__(self, hs): + UserInteractiveAuthChecker.__init__(self, hs) + _BaseThreepidAuthChecker.__init__(self, hs) + + def check_auth(self, authdict, clientip): + return self._check_threepid("msisdn", authdict) + + +INTERACTIVE_AUTH_CHECKERS = [ + DummyAuthChecker, + TermsAuthChecker, + RecaptchaAuthChecker, + EmailIdentityAuthChecker, + MsisdnAuthChecker, +] +"""A list of UserInteractiveAuthChecker classes""" diff --git a/tests/rest/client/v2_alpha/test_auth.py b/tests/rest/client/v2_alpha/test_auth.py index b9ef46e8fb..b6df1396ad 100644 --- a/tests/rest/client/v2_alpha/test_auth.py +++ b/tests/rest/client/v2_alpha/test_auth.py @@ -18,11 +18,22 @@ from twisted.internet.defer import succeed import synapse.rest.admin from synapse.api.constants import LoginType +from synapse.handlers.ui_auth.checkers import UserInteractiveAuthChecker from synapse.rest.client.v2_alpha import auth, register from tests import unittest +class DummyRecaptchaChecker(UserInteractiveAuthChecker): + def __init__(self, hs): + super().__init__(hs) + self.recaptcha_attempts = [] + + def check_auth(self, authdict, clientip): + self.recaptcha_attempts.append((authdict, clientip)) + return succeed(True) + + class FallbackAuthTests(unittest.HomeserverTestCase): servlets = [ @@ -44,15 +55,9 @@ class FallbackAuthTests(unittest.HomeserverTestCase): return hs def prepare(self, reactor, clock, hs): + self.recaptcha_checker = DummyRecaptchaChecker(hs) auth_handler = hs.get_auth_handler() - - self.recaptcha_attempts = [] - - def _recaptcha(authdict, clientip): - self.recaptcha_attempts.append((authdict, clientip)) - return succeed(True) - - auth_handler.checkers[LoginType.RECAPTCHA] = _recaptcha + auth_handler.checkers[LoginType.RECAPTCHA] = self.recaptcha_checker @unittest.INFO def test_fallback_captcha(self): @@ -89,8 +94,9 @@ class FallbackAuthTests(unittest.HomeserverTestCase): self.assertEqual(request.code, 200) # The recaptcha handler is called with the response given - self.assertEqual(len(self.recaptcha_attempts), 1) - self.assertEqual(self.recaptcha_attempts[0][0]["response"], "a") + attempts = self.recaptcha_checker.recaptcha_attempts + self.assertEqual(len(attempts), 1) + self.assertEqual(attempts[0][0]["response"], "a") # also complete the dummy auth request, channel = self.make_request( -- cgit 1.4.1 From 990928abde4f3ccd7d43e6214abd7d36434953a9 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Wed, 25 Sep 2019 12:10:26 +0100 Subject: Stop advertising unsupported flows for registration (#6107) If email or msisdn verification aren't supported, let's stop advertising them for registration. Fixes #6100. --- changelog.d/6107.bugfix | 1 + synapse/handlers/auth.py | 11 +++++++++- synapse/handlers/ui_auth/checkers.py | 26 +++++++++++++++++++++++ synapse/rest/client/v2_alpha/register.py | 32 ++++++++++++++++++++++++++--- tests/rest/client/v2_alpha/test_register.py | 29 +++++++++++++++----------- 5 files changed, 83 insertions(+), 16 deletions(-) create mode 100644 changelog.d/6107.bugfix (limited to 'tests') diff --git a/changelog.d/6107.bugfix b/changelog.d/6107.bugfix new file mode 100644 index 0000000000..d4b9516ac7 --- /dev/null +++ b/changelog.d/6107.bugfix @@ -0,0 +1 @@ +Ensure that servers which are not configured to support email address verification do not offer it in the registration flows. \ No newline at end of file diff --git a/synapse/handlers/auth.py b/synapse/handlers/auth.py index f920c2f6c1..333eb30625 100644 --- a/synapse/handlers/auth.py +++ b/synapse/handlers/auth.py @@ -61,7 +61,8 @@ class AuthHandler(BaseHandler): self.checkers = {} # type: dict[str, UserInteractiveAuthChecker] for auth_checker_class in INTERACTIVE_AUTH_CHECKERS: inst = auth_checker_class(hs) - self.checkers[inst.AUTH_TYPE] = inst + if inst.is_enabled(): + self.checkers[inst.AUTH_TYPE] = inst self.bcrypt_rounds = hs.config.bcrypt_rounds @@ -156,6 +157,14 @@ class AuthHandler(BaseHandler): return params + def get_enabled_auth_types(self): + """Return the enabled user-interactive authentication types + + Returns the UI-Auth types which are supported by the homeserver's current + config. + """ + return self.checkers.keys() + @defer.inlineCallbacks def check_auth(self, flows, clientdict, clientip): """ diff --git a/synapse/handlers/ui_auth/checkers.py b/synapse/handlers/ui_auth/checkers.py index fd633b7b0e..ee69223243 100644 --- a/synapse/handlers/ui_auth/checkers.py +++ b/synapse/handlers/ui_auth/checkers.py @@ -32,6 +32,13 @@ class UserInteractiveAuthChecker: def __init__(self, hs): pass + def is_enabled(self): + """Check if the configuration of the homeserver allows this checker to work + + Returns: + bool: True if this login type is enabled. + """ + def check_auth(self, authdict, clientip): """Given the authentication dict from the client, attempt to check this step @@ -51,6 +58,9 @@ class UserInteractiveAuthChecker: class DummyAuthChecker(UserInteractiveAuthChecker): AUTH_TYPE = LoginType.DUMMY + def is_enabled(self): + return True + def check_auth(self, authdict, clientip): return defer.succeed(True) @@ -58,6 +68,9 @@ class DummyAuthChecker(UserInteractiveAuthChecker): class TermsAuthChecker(UserInteractiveAuthChecker): AUTH_TYPE = LoginType.TERMS + def is_enabled(self): + return True + def check_auth(self, authdict, clientip): return defer.succeed(True) @@ -67,10 +80,14 @@ class RecaptchaAuthChecker(UserInteractiveAuthChecker): def __init__(self, hs): super().__init__(hs) + self._enabled = bool(hs.config.recaptcha_private_key) self._http_client = hs.get_simple_http_client() self._url = hs.config.recaptcha_siteverify_api self._secret = hs.config.recaptcha_private_key + def is_enabled(self): + return self._enabled + @defer.inlineCallbacks def check_auth(self, authdict, clientip): try: @@ -191,6 +208,12 @@ class EmailIdentityAuthChecker(UserInteractiveAuthChecker, _BaseThreepidAuthChec UserInteractiveAuthChecker.__init__(self, hs) _BaseThreepidAuthChecker.__init__(self, hs) + def is_enabled(self): + return self.hs.config.threepid_behaviour_email in ( + ThreepidBehaviour.REMOTE, + ThreepidBehaviour.LOCAL, + ) + def check_auth(self, authdict, clientip): return self._check_threepid("email", authdict) @@ -202,6 +225,9 @@ class MsisdnAuthChecker(UserInteractiveAuthChecker, _BaseThreepidAuthChecker): UserInteractiveAuthChecker.__init__(self, hs) _BaseThreepidAuthChecker.__init__(self, hs) + def is_enabled(self): + return bool(self.hs.config.account_threepid_delegate_msisdn) + def check_auth(self, authdict, clientip): return self._check_threepid("msisdn", authdict) diff --git a/synapse/rest/client/v2_alpha/register.py b/synapse/rest/client/v2_alpha/register.py index e3f3d9126f..4f24a124a6 100644 --- a/synapse/rest/client/v2_alpha/register.py +++ b/synapse/rest/client/v2_alpha/register.py @@ -32,12 +32,14 @@ from synapse.api.errors import ( ThreepidValidationError, UnrecognizedRequestError, ) +from synapse.config import ConfigError from synapse.config.captcha import CaptchaConfig from synapse.config.consent_config import ConsentConfig from synapse.config.emailconfig import ThreepidBehaviour from synapse.config.ratelimiting import FederationRateLimitConfig from synapse.config.registration import RegistrationConfig from synapse.config.server import is_threepid_reserved +from synapse.handlers.auth import AuthHandler from synapse.http.server import finish_request from synapse.http.servlet import ( RestServlet, @@ -375,7 +377,9 @@ class RegisterRestServlet(RestServlet): self.ratelimiter = hs.get_registration_ratelimiter() self.clock = hs.get_clock() - self._registration_flows = _calculate_registration_flows(hs.config) + self._registration_flows = _calculate_registration_flows( + hs.config, self.auth_handler + ) @interactive_auth_handler @defer.inlineCallbacks @@ -664,11 +668,13 @@ class RegisterRestServlet(RestServlet): def _calculate_registration_flows( # technically `config` has to provide *all* of these interfaces, not just one config: Union[RegistrationConfig, ConsentConfig, CaptchaConfig], + auth_handler: AuthHandler, ) -> List[List[str]]: """Get a suitable flows list for registration Args: config: server configuration + auth_handler: authorization handler Returns: a list of supported flows """ @@ -678,10 +684,29 @@ def _calculate_registration_flows( require_msisdn = "msisdn" in config.registrations_require_3pid show_msisdn = True + show_email = True + if config.disable_msisdn_registration: show_msisdn = False require_msisdn = False + enabled_auth_types = auth_handler.get_enabled_auth_types() + if LoginType.EMAIL_IDENTITY not in enabled_auth_types: + show_email = False + if require_email: + raise ConfigError( + "Configuration requires email address at registration, but email " + "validation is not configured" + ) + + if LoginType.MSISDN not in enabled_auth_types: + show_msisdn = False + if require_msisdn: + raise ConfigError( + "Configuration requires msisdn at registration, but msisdn " + "validation is not configured" + ) + flows = [] # only support 3PIDless registration if no 3PIDs are required @@ -693,14 +718,15 @@ def _calculate_registration_flows( flows.append([LoginType.DUMMY]) # only support the email-only flow if we don't require MSISDN 3PIDs - if not require_msisdn: + if show_email and not require_msisdn: flows.append([LoginType.EMAIL_IDENTITY]) # only support the MSISDN-only flow if we don't require email 3PIDs if show_msisdn and not require_email: flows.append([LoginType.MSISDN]) - if show_msisdn: + if show_email and show_msisdn: + # always let users provide both MSISDN & email flows.append([LoginType.MSISDN, LoginType.EMAIL_IDENTITY]) # Prepend m.login.terms to all flows if we're requiring consent diff --git a/tests/rest/client/v2_alpha/test_register.py b/tests/rest/client/v2_alpha/test_register.py index bc2dc47973..dab87e5edf 100644 --- a/tests/rest/client/v2_alpha/test_register.py +++ b/tests/rest/client/v2_alpha/test_register.py @@ -198,16 +198,8 @@ class RegisterRestServletTestCase(unittest.HomeserverTestCase): self.assertEquals(channel.result["code"], b"401", channel.result) flows = channel.json_body["flows"] - # with the stock config, we expect all four combinations of 3pid - self.assertCountEqual( - [ - ["m.login.dummy"], - ["m.login.email.identity"], - ["m.login.msisdn"], - ["m.login.msisdn", "m.login.email.identity"], - ], - (f["stages"] for f in flows), - ) + # with the stock config, we only expect the dummy flow + self.assertCountEqual([["m.login.dummy"]], (f["stages"] for f in flows)) @unittest.override_config( { @@ -217,9 +209,13 @@ class RegisterRestServletTestCase(unittest.HomeserverTestCase): "template_dir": "/", "require_at_registration": True, }, + "account_threepid_delegates": { + "email": "https://id_server", + "msisdn": "https://id_server", + }, } ) - def test_advertised_flows_captcha_and_terms(self): + def test_advertised_flows_captcha_and_terms_and_3pids(self): request, channel = self.make_request(b"POST", self.url, b"{}") self.render(request) self.assertEquals(channel.result["code"], b"401", channel.result) @@ -241,7 +237,16 @@ class RegisterRestServletTestCase(unittest.HomeserverTestCase): ) @unittest.override_config( - {"registrations_require_3pid": ["email"], "disable_msisdn_registration": True} + { + "public_baseurl": "https://test_server", + "registrations_require_3pid": ["email"], + "disable_msisdn_registration": True, + "email": { + "smtp_host": "mail_server", + "smtp_port": 2525, + "notif_from": "sender@host", + }, + } ) def test_advertised_flows_no_msisdn_email_required(self): request, channel = self.make_request(b"POST", self.url, b"{}") -- cgit 1.4.1 From 034db2ba2115d935ce62b641b4051e477a454eac Mon Sep 17 00:00:00 2001 From: Neil Johnson Date: Thu, 26 Sep 2019 11:47:53 +0100 Subject: Fix dummy event insertion consent bug (#6053) Fixes #5905 --- changelog.d/6053.bugfix | 1 + synapse/handlers/message.py | 99 ++++++++++++++++------ synapse/storage/event_federation.py | 18 +++- tests/storage/test_cleanup_extrems.py | 147 +++++++++++++++++++++++++++++++-- tests/storage/test_event_federation.py | 40 +++++++++ 5 files changed, 266 insertions(+), 39 deletions(-) create mode 100644 changelog.d/6053.bugfix (limited to 'tests') diff --git a/changelog.d/6053.bugfix b/changelog.d/6053.bugfix new file mode 100644 index 0000000000..6311157bf6 --- /dev/null +++ b/changelog.d/6053.bugfix @@ -0,0 +1 @@ +Prevent exceptions being logged when extremity-cleanup events fail due to lack of user consent to the terms of service. diff --git a/synapse/handlers/message.py b/synapse/handlers/message.py index 1f8272784e..0f8cce8ffe 100644 --- a/synapse/handlers/message.py +++ b/synapse/handlers/message.py @@ -222,6 +222,13 @@ class MessageHandler(object): } +# The duration (in ms) after which rooms should be removed +# `_rooms_to_exclude_from_dummy_event_insertion` (with the effect that we will try +# to generate a dummy event for them once more) +# +_DUMMY_EVENT_ROOM_EXCLUSION_EXPIRY = 7 * 24 * 60 * 60 * 1000 + + class EventCreationHandler(object): def __init__(self, hs): self.hs = hs @@ -258,6 +265,13 @@ class EventCreationHandler(object): self.config.block_events_without_consent_error ) + # Rooms which should be excluded from dummy insertion. (For instance, + # those without local users who can send events into the room). + # + # map from room id to time-of-last-attempt. + # + self._rooms_to_exclude_from_dummy_event_insertion = {} # type: dict[str, int] + # we need to construct a ConsentURIBuilder here, as it checks that the necessary # config options, but *only* if we have a configuration for which we are # going to need it. @@ -888,9 +902,11 @@ class EventCreationHandler(object): """Background task to send dummy events into rooms that have a large number of extremities """ - + self._expire_rooms_to_exclude_from_dummy_event_insertion() room_ids = yield self.store.get_rooms_with_many_extremities( - min_count=10, limit=5 + min_count=10, + limit=5, + room_id_filter=self._rooms_to_exclude_from_dummy_event_insertion.keys(), ) for room_id in room_ids: @@ -904,32 +920,61 @@ class EventCreationHandler(object): members = yield self.state.get_current_users_in_room( room_id, latest_event_ids=latest_event_ids ) + dummy_event_sent = False + for user_id in members: + if not self.hs.is_mine_id(user_id): + continue + requester = create_requester(user_id) + try: + event, context = yield self.create_event( + requester, + { + "type": "org.matrix.dummy_event", + "content": {}, + "room_id": room_id, + "sender": user_id, + }, + prev_events_and_hashes=prev_events_and_hashes, + ) - user_id = None - for member in members: - if self.hs.is_mine_id(member): - user_id = member - break - - if not user_id: - # We don't have a joined user. - # TODO: We should do something here to stop the room from - # appearing next time. - continue + event.internal_metadata.proactively_send = False - requester = create_requester(user_id) + yield self.send_nonmember_event( + requester, event, context, ratelimit=False + ) + dummy_event_sent = True + break + except ConsentNotGivenError: + logger.info( + "Failed to send dummy event into room %s for user %s due to " + "lack of consent. Will try another user" % (room_id, user_id) + ) + except AuthError: + logger.info( + "Failed to send dummy event into room %s for user %s due to " + "lack of power. Will try another user" % (room_id, user_id) + ) - event, context = yield self.create_event( - requester, - { - "type": "org.matrix.dummy_event", - "content": {}, - "room_id": room_id, - "sender": user_id, - }, - prev_events_and_hashes=prev_events_and_hashes, + if not dummy_event_sent: + # Did not find a valid user in the room, so remove from future attempts + # Exclusion is time limited, so the room will be rechecked in the future + # dependent on _DUMMY_EVENT_ROOM_EXCLUSION_EXPIRY + logger.info( + "Failed to send dummy event into room %s. Will exclude it from " + "future attempts until cache expires" % (room_id,) + ) + now = self.clock.time_msec() + self._rooms_to_exclude_from_dummy_event_insertion[room_id] = now + + def _expire_rooms_to_exclude_from_dummy_event_insertion(self): + expire_before = self.clock.time_msec() - _DUMMY_EVENT_ROOM_EXCLUSION_EXPIRY + to_expire = set() + for room_id, time in self._rooms_to_exclude_from_dummy_event_insertion.items(): + if time < expire_before: + to_expire.add(room_id) + for room_id in to_expire: + logger.debug( + "Expiring room id %s from dummy event insertion exclusion cache", + room_id, ) - - event.internal_metadata.proactively_send = False - - yield self.send_nonmember_event(requester, event, context, ratelimit=False) + del self._rooms_to_exclude_from_dummy_event_insertion[room_id] diff --git a/synapse/storage/event_federation.py b/synapse/storage/event_federation.py index 4f500d893e..f5e8c39262 100644 --- a/synapse/storage/event_federation.py +++ b/synapse/storage/event_federation.py @@ -12,6 +12,7 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. +import itertools import logging import random @@ -190,12 +191,13 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBas room_id, ) - def get_rooms_with_many_extremities(self, min_count, limit): + def get_rooms_with_many_extremities(self, min_count, limit, room_id_filter): """Get the top rooms with at least N extremities. Args: min_count (int): The minimum number of extremities limit (int): The maximum number of rooms to return. + room_id_filter (iterable[str]): room_ids to exclude from the results Returns: Deferred[list]: At most `limit` room IDs that have at least @@ -203,15 +205,25 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBas """ def _get_rooms_with_many_extremities_txn(txn): + where_clause = "1=1" + if room_id_filter: + where_clause = "room_id NOT IN (%s)" % ( + ",".join("?" for _ in room_id_filter), + ) + sql = """ SELECT room_id FROM event_forward_extremities + WHERE %s GROUP BY room_id HAVING count(*) > ? ORDER BY count(*) DESC LIMIT ? - """ + """ % ( + where_clause, + ) - txn.execute(sql, (min_count, limit)) + query_args = list(itertools.chain(room_id_filter, [min_count, limit])) + txn.execute(sql, query_args) return [room_id for room_id, in txn] return self.runInteraction( diff --git a/tests/storage/test_cleanup_extrems.py b/tests/storage/test_cleanup_extrems.py index e9e2d5337c..34f9c72709 100644 --- a/tests/storage/test_cleanup_extrems.py +++ b/tests/storage/test_cleanup_extrems.py @@ -14,7 +14,13 @@ # limitations under the License. import os.path +from unittest.mock import patch +from mock import Mock + +import synapse.rest.admin +from synapse.api.constants import EventTypes +from synapse.rest.client.v1 import login, room from synapse.storage import prepare_database from synapse.types import Requester, UserID @@ -225,6 +231,14 @@ class CleanupExtremBackgroundUpdateStoreTestCase(HomeserverTestCase): class CleanupExtremDummyEventsTestCase(HomeserverTestCase): + CONSENT_VERSION = "1" + EXTREMITIES_COUNT = 50 + servlets = [ + synapse.rest.admin.register_servlets_for_client_rest_resource, + login.register_servlets, + room.register_servlets, + ] + def make_homeserver(self, reactor, clock): config = self.default_config() config["cleanup_extremities_with_dummy_events"] = True @@ -233,33 +247,148 @@ class CleanupExtremDummyEventsTestCase(HomeserverTestCase): def prepare(self, reactor, clock, homeserver): self.store = homeserver.get_datastore() self.room_creator = homeserver.get_room_creation_handler() + self.event_creator_handler = homeserver.get_event_creation_handler() # Create a test user and room - self.user = UserID("alice", "test") + self.user = UserID.from_string(self.register_user("user1", "password")) + self.token1 = self.login("user1", "password") self.requester = Requester(self.user, None, False, None, None) info = self.get_success(self.room_creator.create_room(self.requester, {})) self.room_id = info["room_id"] + self.event_creator = homeserver.get_event_creation_handler() + homeserver.config.user_consent_version = self.CONSENT_VERSION def test_send_dummy_event(self): - # Create a bushy graph with 50 extremities. - - event_id_start = self.create_and_send_event(self.room_id, self.user) + self._create_extremity_rich_graph() - for _ in range(50): - self.create_and_send_event( - self.room_id, self.user, prev_event_ids=[event_id_start] - ) + # Pump the reactor repeatedly so that the background updates have a + # chance to run. + self.pump(10 * 60) latest_event_ids = self.get_success( self.store.get_latest_event_ids_in_room(self.room_id) ) - self.assertEqual(len(latest_event_ids), 50) + self.assertTrue(len(latest_event_ids) < 10, len(latest_event_ids)) + @patch("synapse.handlers.message._DUMMY_EVENT_ROOM_EXCLUSION_EXPIRY", new=0) + def test_send_dummy_events_when_insufficient_power(self): + self._create_extremity_rich_graph() + # Criple power levels + self.helper.send_state( + self.room_id, + EventTypes.PowerLevels, + body={"users": {str(self.user): -1}}, + tok=self.token1, + ) # Pump the reactor repeatedly so that the background updates have a # chance to run. self.pump(10 * 60) + latest_event_ids = self.get_success( + self.store.get_latest_event_ids_in_room(self.room_id) + ) + # Check that the room has not been pruned + self.assertTrue(len(latest_event_ids) > 10) + + # New user with regular levels + user2 = self.register_user("user2", "password") + token2 = self.login("user2", "password") + self.helper.join(self.room_id, user2, tok=token2) + self.pump(10 * 60) + + latest_event_ids = self.get_success( + self.store.get_latest_event_ids_in_room(self.room_id) + ) + self.assertTrue(len(latest_event_ids) < 10, len(latest_event_ids)) + + @patch("synapse.handlers.message._DUMMY_EVENT_ROOM_EXCLUSION_EXPIRY", new=0) + def test_send_dummy_event_without_consent(self): + self._create_extremity_rich_graph() + self._enable_consent_checking() + + # Pump the reactor repeatedly so that the background updates have a + # chance to run. Attempt to add dummy event with user that has not consented + # Check that dummy event send fails. + self.pump(10 * 60) + latest_event_ids = self.get_success( + self.store.get_latest_event_ids_in_room(self.room_id) + ) + self.assertTrue(len(latest_event_ids) == self.EXTREMITIES_COUNT) + + # Create new user, and add consent + user2 = self.register_user("user2", "password") + token2 = self.login("user2", "password") + self.get_success( + self.store.user_set_consent_version(user2, self.CONSENT_VERSION) + ) + self.helper.join(self.room_id, user2, tok=token2) + + # Background updates should now cause a dummy event to be added to the graph + self.pump(10 * 60) + latest_event_ids = self.get_success( self.store.get_latest_event_ids_in_room(self.room_id) ) self.assertTrue(len(latest_event_ids) < 10, len(latest_event_ids)) + + @patch("synapse.handlers.message._DUMMY_EVENT_ROOM_EXCLUSION_EXPIRY", new=250) + def test_expiry_logic(self): + """Simple test to ensure that _expire_rooms_to_exclude_from_dummy_event_insertion() + expires old entries correctly. + """ + self.event_creator_handler._rooms_to_exclude_from_dummy_event_insertion[ + "1" + ] = 100000 + self.event_creator_handler._rooms_to_exclude_from_dummy_event_insertion[ + "2" + ] = 200000 + self.event_creator_handler._rooms_to_exclude_from_dummy_event_insertion[ + "3" + ] = 300000 + self.event_creator_handler._expire_rooms_to_exclude_from_dummy_event_insertion() + # All entries within time frame + self.assertEqual( + len( + self.event_creator_handler._rooms_to_exclude_from_dummy_event_insertion + ), + 3, + ) + # Oldest room to expire + self.pump(1) + self.event_creator_handler._expire_rooms_to_exclude_from_dummy_event_insertion() + self.assertEqual( + len( + self.event_creator_handler._rooms_to_exclude_from_dummy_event_insertion + ), + 2, + ) + # All rooms to expire + self.pump(2) + self.assertEqual( + len( + self.event_creator_handler._rooms_to_exclude_from_dummy_event_insertion + ), + 0, + ) + + def _create_extremity_rich_graph(self): + """Helper method to create bushy graph on demand""" + + event_id_start = self.create_and_send_event(self.room_id, self.user) + + for _ in range(self.EXTREMITIES_COUNT): + self.create_and_send_event( + self.room_id, self.user, prev_event_ids=[event_id_start] + ) + + latest_event_ids = self.get_success( + self.store.get_latest_event_ids_in_room(self.room_id) + ) + self.assertEqual(len(latest_event_ids), 50) + + def _enable_consent_checking(self): + """Helper method to enable consent checking""" + self.event_creator._block_events_without_consent_error = "No consent from user" + consent_uri_builder = Mock() + consent_uri_builder.build_user_consent_uri.return_value = "http://example.com" + self.event_creator._consent_uri_builder = consent_uri_builder diff --git a/tests/storage/test_event_federation.py b/tests/storage/test_event_federation.py index 86c7ac350d..b58386994e 100644 --- a/tests/storage/test_event_federation.py +++ b/tests/storage/test_event_federation.py @@ -75,3 +75,43 @@ class EventFederationWorkerStoreTestCase(tests.unittest.TestCase): el = r[i] depth = el[2] self.assertLessEqual(5, depth) + + @defer.inlineCallbacks + def test_get_rooms_with_many_extremities(self): + room1 = "#room1" + room2 = "#room2" + room3 = "#room3" + + def insert_event(txn, i, room_id): + event_id = "$event_%i:local" % i + txn.execute( + ( + "INSERT INTO event_forward_extremities (room_id, event_id) " + "VALUES (?, ?)" + ), + (room_id, event_id), + ) + + for i in range(0, 20): + yield self.store.runInteraction("insert", insert_event, i, room1) + yield self.store.runInteraction("insert", insert_event, i, room2) + yield self.store.runInteraction("insert", insert_event, i, room3) + + # Test simple case + r = yield self.store.get_rooms_with_many_extremities(5, 5, []) + self.assertEqual(len(r), 3) + + # Does filter work? + + r = yield self.store.get_rooms_with_many_extremities(5, 5, [room1]) + self.assertTrue(room2 in r) + self.assertTrue(room3 in r) + self.assertEqual(len(r), 2) + + r = yield self.store.get_rooms_with_many_extremities(5, 5, [room1, room2]) + self.assertEqual(r, [room3]) + + # Does filter and limit work? + + r = yield self.store.get_rooms_with_many_extremities(5, 1, [room1]) + self.assertTrue(r == [room2] or r == [room3]) -- cgit 1.4.1 From 132279a46fd76de8f767bc6977192900c450fec9 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Fri, 27 Sep 2019 15:11:14 +0100 Subject: Patch inlinecallbacks for log contexts --- synapse/__init__.py | 6 +++ synapse/handlers/room_member.py | 4 +- synapse/storage/_base.py | 7 ++-- synapse/storage/push_rule.py | 2 +- tests/patch_inline_callbacks.py | 86 +++++++++++++++++++++++++++++++++++++++-- 5 files changed, 95 insertions(+), 10 deletions(-) (limited to 'tests') diff --git a/synapse/__init__.py b/synapse/__init__.py index ddfe9ec542..4401fd52f0 100644 --- a/synapse/__init__.py +++ b/synapse/__init__.py @@ -17,8 +17,11 @@ """ This is a reference implementation of a Matrix home server. """ +import os import sys +from tests.patch_inline_callbacks import do_patch + # Check that we're not running on an unsupported Python version. if sys.version_info < (3, 5): print("Synapse requires Python 3.5 or above.") @@ -36,3 +39,6 @@ except ImportError: pass __version__ = "1.4.0rc1" + +if bool(os.environ.get("SYNAPSE_TEST_PATCH_LOG_CONTEXTS", False)): + do_patch() diff --git a/synapse/handlers/room_member.py b/synapse/handlers/room_member.py index 8abdb1b6e6..19e44b5460 100644 --- a/synapse/handlers/room_member.py +++ b/synapse/handlers/room_member.py @@ -213,11 +213,11 @@ class RoomMemberHandler(object): if predecessor: # It is an upgraded room. Copy over old tags - self.copy_room_tags_and_direct_to_room( + yield self.copy_room_tags_and_direct_to_room( predecessor["room_id"], room_id, user_id ) # Move over old push rules - self.store.move_push_rules_from_room_to_room_for_user( + yield self.store.move_push_rules_from_room_to_room_for_user( predecessor["room_id"], room_id, user_id ) elif event.membership == Membership.LEAVE: diff --git a/synapse/storage/_base.py b/synapse/storage/_base.py index abe16334ec..06cc14fcd1 100644 --- a/synapse/storage/_base.py +++ b/synapse/storage/_base.py @@ -30,7 +30,7 @@ from prometheus_client import Histogram from twisted.internet import defer from synapse.api.errors import StoreError -from synapse.logging.context import LoggingContext, PreserveLoggingContext +from synapse.logging.context import LoggingContext, make_deferred_yieldable from synapse.metrics.background_process_metrics import run_as_background_process from synapse.storage.engines import PostgresEngine, Sqlite3Engine from synapse.types import get_domain_from_id @@ -550,8 +550,9 @@ class SQLBaseStore(object): return func(conn, *args, **kwargs) - with PreserveLoggingContext(): - result = yield self._db_pool.runWithConnection(inner_func, *args, **kwargs) + result = yield make_deferred_yieldable( + self._db_pool.runWithConnection(inner_func, *args, **kwargs) + ) return result diff --git a/synapse/storage/push_rule.py b/synapse/storage/push_rule.py index a6517c4cf3..1f878e6575 100644 --- a/synapse/storage/push_rule.py +++ b/synapse/storage/push_rule.py @@ -235,7 +235,7 @@ class PushRulesWorkerStore( (c.get("key") == "room_id" and c.get("pattern") == old_room_id) for c in conditions ): - self.move_push_rule_from_room_to_room(new_room_id, user_id, rule) + yield self.move_push_rule_from_room_to_room(new_room_id, user_id, rule) @defer.inlineCallbacks def bulk_get_push_rules_for_room(self, event, context): diff --git a/tests/patch_inline_callbacks.py b/tests/patch_inline_callbacks.py index 220884311c..5ef0aff0c3 100644 --- a/tests/patch_inline_callbacks.py +++ b/tests/patch_inline_callbacks.py @@ -15,6 +15,7 @@ from __future__ import print_function +import inspect import functools import sys @@ -33,17 +34,19 @@ def do_patch(): orig_inline_callbacks = defer.inlineCallbacks def new_inline_callbacks(f): - - orig = orig_inline_callbacks(f) - @functools.wraps(f) def wrapped(*args, **kwargs): start_context = LoggingContext.current_context() + changes = [] + orig = orig_inline_callbacks(_check_yield_points(f, changes, start_context)) try: res = orig(*args, **kwargs) except Exception: if LoggingContext.current_context() != start_context: + for err in changes: + print(err, file=sys.stderr) + err = "%s changed context from %s to %s on exception" % ( f, start_context, @@ -55,7 +58,10 @@ def do_patch(): if not isinstance(res, Deferred) or res.called: if LoggingContext.current_context() != start_context: - err = "%s changed context from %s to %s" % ( + for err in changes: + print(err, file=sys.stderr) + + err = "Completed %s changed context from %s to %s" % ( f, start_context, LoggingContext.current_context(), @@ -76,6 +82,8 @@ def do_patch(): def check_ctx(r): if LoggingContext.current_context() != start_context: + for err in changes: + print(err, file=sys.stderr) err = "%s completion of %s changed context from %s to %s" % ( "Failure" if isinstance(r, Failure) else "Success", f, @@ -92,3 +100,73 @@ def do_patch(): return wrapped defer.inlineCallbacks = new_inline_callbacks + + +def _check_yield_points(f, changes, start_context): + from synapse.logging.context import LoggingContext + + @functools.wraps(f) + def check_yield_points_inner(*args, **kwargs): + gen = f(*args, **kwargs) + + last_yield_line_no = 1 + result = None + while True: + try: + isFailure = isinstance(result, Failure) + if isFailure: + d = result.throwExceptionIntoGenerator(gen) + else: + d = gen.send(result) + except (StopIteration, defer._DefGen_Return) as e: + if LoggingContext.current_context() != start_context: + # This happens when the context is lost sometime *after* the + # final yield and returning. E.g. we forgot to yield on a + # function that returns a deferred. + err = ( + "%s returned and changed context from %s to %s, in %s between %d and end of func" + % ( + f.__qualname__, + start_context, + LoggingContext.current_context(), + f.__code__.co_filename, + last_yield_line_no, + ) + ) + changes.append(err) + # print(err, file=sys.stderr) + # raise Exception(err) + return getattr(e, "value", None) + + try: + result = yield d + except Exception as e: + result = Failure(e) + + frame = gen.gi_frame + if frame.f_code.co_name == "check_yield_points_inner": + frame = inspect.getgeneratorlocals(gen)["gen"].gi_frame + + if LoggingContext.current_context() != start_context: + # This happens because the context is lost sometime *after* the + # previous yield and *after* the current yield. E.g. the + # deferred we waited on didn't follow the rules, or we forgot to + # yield on a function between the two yield points. + err = ( + "%s changed context from %s to %s, happened between lines %d and %d in %s" + % ( + frame.f_code.co_name, + start_context, + LoggingContext.current_context(), + last_yield_line_no, + frame.f_lineno, + frame.f_code.co_filename, + ) + ) + changes.append(err) + # print(err, file=sys.stderr) + # raise Exception(err) + + last_yield_line_no = frame.f_lineno + + return check_yield_points_inner -- cgit 1.4.1 From e94ff67903c3370fc5bc8b6c336433057e38ff05 Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Fri, 27 Sep 2019 15:14:02 +0100 Subject: Add test to validate the change --- tests/rest/client/v2_alpha/test_account.py | 70 ++++++++++++++++++++++++------ 1 file changed, 57 insertions(+), 13 deletions(-) (limited to 'tests') diff --git a/tests/rest/client/v2_alpha/test_account.py b/tests/rest/client/v2_alpha/test_account.py index 920de41de4..69c33dfd8a 100644 --- a/tests/rest/client/v2_alpha/test_account.py +++ b/tests/rest/client/v2_alpha/test_account.py @@ -23,8 +23,8 @@ from email.parser import Parser import pkg_resources import synapse.rest.admin -from synapse.api.constants import LoginType -from synapse.rest.client.v1 import login +from synapse.api.constants import LoginType, Membership +from synapse.rest.client.v1 import login, room from synapse.rest.client.v2_alpha import account, register from tests import unittest @@ -244,16 +244,69 @@ class DeactivateTestCase(unittest.HomeserverTestCase): synapse.rest.admin.register_servlets_for_client_rest_resource, login.register_servlets, account.register_servlets, + room.register_servlets, ] def make_homeserver(self, reactor, clock): - hs = self.setup_test_homeserver() - return hs + self.hs = self.setup_test_homeserver() + return self.hs def test_deactivate_account(self): user_id = self.register_user("kermit", "test") tok = self.login("kermit", "test") + self.deactivate(user_id, tok) + + store = self.hs.get_datastore() + + # Check that the user has been marked as deactivated. + self.assertTrue(self.get_success(store.get_user_deactivated_status(user_id))) + + # Check that this access token has been invalidated. + request, channel = self.make_request("GET", "account/whoami") + self.render(request) + self.assertEqual(request.code, 401) + + @unittest.INFO + def test_pending_invites(self): + """Tests that deactivating a user rejects every pending invite for them.""" + store = self.hs.get_datastore() + + inviter_id = self.register_user("inviter", "test") + inviter_tok = self.login("inviter", "test") + + invitee_id = self.register_user("invitee", "test") + invitee_tok = self.login("invitee", "test") + + # Make @inviter:test invite @invitee:test in a new room. + room_id = self.helper.create_room_as(inviter_id, tok=inviter_tok) + self.helper.invite( + room=room_id, + src=inviter_id, + targ=invitee_id, + tok=inviter_tok, + ) + + # Make sure the invite is here. + pending_invites = self.get_success(store.get_invited_rooms_for_user(invitee_id)) + self.assertEqual(len(pending_invites), 1, pending_invites) + self.assertEqual(pending_invites[0].room_id, room_id, pending_invites) + + # Deactivate @invitee:test. + self.deactivate(invitee_id, invitee_tok) + + # Check that the invite isn't there anymore. + pending_invites = self.get_success(store.get_invited_rooms_for_user(invitee_id)) + self.assertEqual(len(pending_invites), 0, pending_invites) + + # Check that the membership of @invitee:test in the room is now "leave". + memberships = self.get_success( + store.get_rooms_for_user_where_membership_is(invitee_id, [Membership.LEAVE]) + ) + self.assertEqual(len(memberships), 1, memberships) + self.assertEqual(memberships[0].room_id, room_id, memberships) + + def deactivate(self, user_id, tok): request_data = json.dumps( { "auth": { @@ -270,12 +323,3 @@ class DeactivateTestCase(unittest.HomeserverTestCase): self.render(request) self.assertEqual(request.code, 200) - store = self.hs.get_datastore() - - # Check that the user has been marked as deactivated. - self.assertTrue(self.get_success(store.get_user_deactivated_status(user_id))) - - # Check that this access token has been invalidated. - request, channel = self.make_request("GET", "account/whoami") - self.render(request) - self.assertEqual(request.code, 401) -- cgit 1.4.1 From 873fe7883cf0d7cf5346a9a55d40967a35848e33 Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Fri, 27 Sep 2019 15:21:03 +0100 Subject: Lint --- synapse/handlers/deactivate_account.py | 4 +--- tests/rest/client/v2_alpha/test_account.py | 8 +------- 2 files changed, 2 insertions(+), 10 deletions(-) (limited to 'tests') diff --git a/synapse/handlers/deactivate_account.py b/synapse/handlers/deactivate_account.py index 763fea3a24..148d1424ca 100644 --- a/synapse/handlers/deactivate_account.py +++ b/synapse/handlers/deactivate_account.py @@ -156,9 +156,7 @@ class DeactivateAccountHandler(BaseHandler): require_consent=False, ) logger.info( - "Rejected invite for user %r in room %r", - user_id, - room.room_id, + "Rejected invite for user %r in room %r", user_id, room.room_id ) except Exception: logger.exception( diff --git a/tests/rest/client/v2_alpha/test_account.py b/tests/rest/client/v2_alpha/test_account.py index 69c33dfd8a..434b730faf 100644 --- a/tests/rest/client/v2_alpha/test_account.py +++ b/tests/rest/client/v2_alpha/test_account.py @@ -280,12 +280,7 @@ class DeactivateTestCase(unittest.HomeserverTestCase): # Make @inviter:test invite @invitee:test in a new room. room_id = self.helper.create_room_as(inviter_id, tok=inviter_tok) - self.helper.invite( - room=room_id, - src=inviter_id, - targ=invitee_id, - tok=inviter_tok, - ) + self.helper.invite(room=room_id, src=inviter_id, targ=invitee_id, tok=inviter_tok) # Make sure the invite is here. pending_invites = self.get_success(store.get_invited_rooms_for_user(invitee_id)) @@ -322,4 +317,3 @@ class DeactivateTestCase(unittest.HomeserverTestCase): ) self.render(request) self.assertEqual(request.code, 200) - -- cgit 1.4.1 From fbb8ff3088abab48bd5815a1acaeb9243ada7431 Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Fri, 27 Sep 2019 15:23:07 +0100 Subject: ok --- tests/rest/client/v2_alpha/test_account.py | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'tests') diff --git a/tests/rest/client/v2_alpha/test_account.py b/tests/rest/client/v2_alpha/test_account.py index 434b730faf..0f51895b81 100644 --- a/tests/rest/client/v2_alpha/test_account.py +++ b/tests/rest/client/v2_alpha/test_account.py @@ -280,7 +280,9 @@ class DeactivateTestCase(unittest.HomeserverTestCase): # Make @inviter:test invite @invitee:test in a new room. room_id = self.helper.create_room_as(inviter_id, tok=inviter_tok) - self.helper.invite(room=room_id, src=inviter_id, targ=invitee_id, tok=inviter_tok) + self.helper.invite( + room=room_id, src=inviter_id, targ=invitee_id, tok=inviter_tok + ) # Make sure the invite is here. pending_invites = self.get_success(store.get_invited_rooms_for_user(invitee_id)) -- cgit 1.4.1 From 6374ca40c2ff3d2eaa41af5ebf4f8324522ecb84 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Fri, 27 Sep 2019 15:58:14 +0100 Subject: Update --- tests/patch_inline_callbacks.py | 23 +++++++++++++++-------- 1 file changed, 15 insertions(+), 8 deletions(-) (limited to 'tests') diff --git a/tests/patch_inline_callbacks.py b/tests/patch_inline_callbacks.py index 5ef0aff0c3..a35a1d3305 100644 --- a/tests/patch_inline_callbacks.py +++ b/tests/patch_inline_callbacks.py @@ -15,7 +15,6 @@ from __future__ import print_function -import inspect import functools import sys @@ -32,6 +31,8 @@ def do_patch(): from synapse.logging.context import LoggingContext orig_inline_callbacks = defer.inlineCallbacks + if hasattr(orig_inline_callbacks, "patched_by_synapse"): + return def new_inline_callbacks(f): @functools.wraps(f) @@ -100,13 +101,20 @@ def do_patch(): return wrapped defer.inlineCallbacks = new_inline_callbacks + new_inline_callbacks.patched_by_synapse = True def _check_yield_points(f, changes, start_context): + """Wraps a generator that is about to passed to defer.inlineCallbacks + checking that after every yield the log contexts are correct. + """ + from synapse.logging.context import LoggingContext @functools.wraps(f) def check_yield_points_inner(*args, **kwargs): + expected_context = start_context + gen = f(*args, **kwargs) last_yield_line_no = 1 @@ -119,12 +127,13 @@ def _check_yield_points(f, changes, start_context): else: d = gen.send(result) except (StopIteration, defer._DefGen_Return) as e: - if LoggingContext.current_context() != start_context: + if LoggingContext.current_context() != expected_context: # This happens when the context is lost sometime *after* the # final yield and returning. E.g. we forgot to yield on a # function that returns a deferred. err = ( - "%s returned and changed context from %s to %s, in %s between %d and end of func" + "Function %r returned and changed context from %s to %s," + " in %s between %d and end of func" % ( f.__qualname__, start_context, @@ -134,7 +143,6 @@ def _check_yield_points(f, changes, start_context): ) ) changes.append(err) - # print(err, file=sys.stderr) # raise Exception(err) return getattr(e, "value", None) @@ -144,10 +152,8 @@ def _check_yield_points(f, changes, start_context): result = Failure(e) frame = gen.gi_frame - if frame.f_code.co_name == "check_yield_points_inner": - frame = inspect.getgeneratorlocals(gen)["gen"].gi_frame - if LoggingContext.current_context() != start_context: + if LoggingContext.current_context() != expected_context: # This happens because the context is lost sometime *after* the # previous yield and *after* the current yield. E.g. the # deferred we waited on didn't follow the rules, or we forgot to @@ -164,9 +170,10 @@ def _check_yield_points(f, changes, start_context): ) ) changes.append(err) - # print(err, file=sys.stderr) # raise Exception(err) + expected_context = LoggingContext.current_context() + last_yield_line_no = frame.f_lineno return check_yield_points_inner -- cgit 1.4.1 From ce7a3e7e27ba4215ed86112d2967b0acca4cfb77 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 2 Oct 2019 10:14:01 +0100 Subject: Fix fetching censored redactions from DB Fetching a censored redactions caused an exception due to the code expecting redactions to have a `redact` key, which redacted redactions don't have. --- synapse/storage/events_worker.py | 14 ++++++++++++++ tests/storage/test_redaction.py | 36 ++++++++++++++++++++++++++++++++++++ 2 files changed, 50 insertions(+) (limited to 'tests') diff --git a/synapse/storage/events_worker.py b/synapse/storage/events_worker.py index c6fa7f82fd..57ce0304e9 100644 --- a/synapse/storage/events_worker.py +++ b/synapse/storage/events_worker.py @@ -238,6 +238,20 @@ class EventsWorkerStore(SQLBaseStore): # we have to recheck auth now. if not allow_rejected and entry.event.type == EventTypes.Redaction: + if not hasattr(entry.event, "redacts"): + # A redacted redaction doesn't have a `redacts` key, in + # which case lets just withhold the event. + # + # Note: Most of the time if the redactions has been + # redacted we still have the un-redacted event in the DB + # and so we'll still see the `redacts` key. However, this + # isn't always true e.g. if we have censored the event. + logger.debug( + "Withholding redaction event %s as we don't have redacts key", + event_id, + ) + continue + redacted_event_id = entry.event.redacts event_map = yield self._get_events_from_cache_or_db([redacted_event_id]) original_event_entry = event_map.get(redacted_event_id) diff --git a/tests/storage/test_redaction.py b/tests/storage/test_redaction.py index deecfad9fb..427d3c49c5 100644 --- a/tests/storage/test_redaction.py +++ b/tests/storage/test_redaction.py @@ -118,6 +118,8 @@ class RedactionTestCase(unittest.HomeserverTestCase): self.get_success(self.store.persist_event(event, context)) + return event + def test_redact(self): self.get_success( self.inject_room_member(self.room1, self.u_alice, Membership.JOIN) @@ -361,3 +363,37 @@ class RedactionTestCase(unittest.HomeserverTestCase): ) self.assert_dict({"content": {}}, json.loads(event_json)) + + def test_redact_redaction(self): + """Tests that we can redact a redaction and can fetch it again. + """ + + self.get_success( + self.inject_room_member(self.room1, self.u_alice, Membership.JOIN) + ) + + msg_event = self.get_success(self.inject_message(self.room1, self.u_alice, "t")) + + first_redact_event = self.get_success( + self.inject_redaction( + self.room1, msg_event.event_id, self.u_alice, "Redacting message" + ) + ) + + self.get_success( + self.inject_redaction( + self.room1, + first_redact_event.event_id, + self.u_alice, + "Redacting redaction", + ) + ) + + # Now lets jump to the future where we have censored the redaction event + # in the DB. + self.reactor.advance(60 * 60 * 24 * 31) + + # We just want to check that fetching the event doesn't raise an exception. + self.get_success( + self.store.get_event(first_redact_event.event_id, allow_none=True) + ) -- cgit 1.4.1 From f44f1d2e8374b7250a8a68cf3a49e6d1ac63b0fb Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 2 Oct 2019 10:36:27 +0100 Subject: Fix errors storing large retry intervals. We have set the max retry interval to a value larger than a postgres or sqlite int can hold, which caused exceptions when updating the destinations table. To fix postgres we need to change the column to a bigint, and for sqlite we lower the max interval to 2**62 (which is still incredibly long). --- .../56/destinations_retry_interval_type.sql.postgres | 18 ++++++++++++++++++ synapse/util/retryutils.py | 2 +- tests/storage/test_transactions.py | 11 +++++++++++ 3 files changed, 30 insertions(+), 1 deletion(-) create mode 100644 synapse/storage/schema/delta/56/destinations_retry_interval_type.sql.postgres (limited to 'tests') diff --git a/synapse/storage/schema/delta/56/destinations_retry_interval_type.sql.postgres b/synapse/storage/schema/delta/56/destinations_retry_interval_type.sql.postgres new file mode 100644 index 0000000000..b9bbb18a91 --- /dev/null +++ b/synapse/storage/schema/delta/56/destinations_retry_interval_type.sql.postgres @@ -0,0 +1,18 @@ +/* Copyright 2019 The Matrix.org Foundation C.I.C + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- We want to store large retry intervals so we upgrade the column from INT +-- to BIGINT. We don't need to do this on SQLite. +ALTER TABLE destinations ALTER retry_interval SET DATA TYPE BIGINT; diff --git a/synapse/util/retryutils.py b/synapse/util/retryutils.py index a5f2fbef5c..af69587196 100644 --- a/synapse/util/retryutils.py +++ b/synapse/util/retryutils.py @@ -29,7 +29,7 @@ MIN_RETRY_INTERVAL = 10 * 60 * 1000 RETRY_MULTIPLIER = 5 # a cap on the backoff. (Essentially none) -MAX_RETRY_INTERVAL = 2 ** 63 +MAX_RETRY_INTERVAL = 2 ** 62 class NotRetryingDestination(Exception): diff --git a/tests/storage/test_transactions.py b/tests/storage/test_transactions.py index a771d5af29..8e817e2c7f 100644 --- a/tests/storage/test_transactions.py +++ b/tests/storage/test_transactions.py @@ -13,6 +13,8 @@ # See the License for the specific language governing permissions and # limitations under the License. +from synapse.util.retryutils import MAX_RETRY_INTERVAL + from tests.unittest import HomeserverTestCase @@ -45,3 +47,12 @@ class TransactionStoreTestCase(HomeserverTestCase): """ d = self.store.set_destination_retry_timings("example.com", 1000, 50, 100) self.get_success(d) + + def test_large_destination_retry(self): + d = self.store.set_destination_retry_timings( + "example.com", MAX_RETRY_INTERVAL, MAX_RETRY_INTERVAL, MAX_RETRY_INTERVAL + ) + self.get_success(d) + + d = self.store.get_destination_retry_timings("example.com") + self.get_success(d) -- cgit 1.4.1 From a5166e4d5febc0e03ba9da9db99127a797a0bc4d Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 2 Oct 2019 14:08:35 +0100 Subject: Land improved room list based on room stats (#6019) Use room_stats and room_state for room directory search --- changelog.d/6019.misc | 1 + synapse/federation/transport/server.py | 8 + synapse/handlers/room_list.py | 323 ++++++--------------- synapse/rest/client/v1/room.py | 8 + synapse/storage/room.py | 228 ++++++++++----- .../schema/delta/56/public_room_list_idx.sql | 16 + tests/handlers/test_roomlist.py | 39 --- 7 files changed, 273 insertions(+), 350 deletions(-) create mode 100644 changelog.d/6019.misc create mode 100644 synapse/storage/schema/delta/56/public_room_list_idx.sql delete mode 100644 tests/handlers/test_roomlist.py (limited to 'tests') diff --git a/changelog.d/6019.misc b/changelog.d/6019.misc new file mode 100644 index 0000000000..dfee73c28f --- /dev/null +++ b/changelog.d/6019.misc @@ -0,0 +1 @@ +Improve performance of the public room list directory. diff --git a/synapse/federation/transport/server.py b/synapse/federation/transport/server.py index 7f8a16e355..0f16f21c2d 100644 --- a/synapse/federation/transport/server.py +++ b/synapse/federation/transport/server.py @@ -765,6 +765,10 @@ class PublicRoomList(BaseFederationServlet): else: network_tuple = ThirdPartyInstanceID(None, None) + if limit == 0: + # zero is a special value which corresponds to no limit. + limit = None + data = await maybeDeferred( self.handler.get_local_public_room_list, limit, @@ -800,6 +804,10 @@ class PublicRoomList(BaseFederationServlet): if search_filter is None: logger.warning("Nonefilter") + if limit == 0: + # zero is a special value which corresponds to no limit. + limit = None + data = await self.handler.get_local_public_room_list( limit=limit, since_token=since_token, diff --git a/synapse/handlers/room_list.py b/synapse/handlers/room_list.py index a7e55f00e5..4e1cc5460f 100644 --- a/synapse/handlers/room_list.py +++ b/synapse/handlers/room_list.py @@ -16,8 +16,7 @@ import logging from collections import namedtuple -from six import PY3, iteritems -from six.moves import range +from six import iteritems import msgpack from unpaddedbase64 import decode_base64, encode_base64 @@ -27,7 +26,6 @@ from twisted.internet import defer from synapse.api.constants import EventTypes, JoinRules from synapse.api.errors import Codes, HttpResponseException from synapse.types import ThirdPartyInstanceID -from synapse.util.async_helpers import concurrently_execute from synapse.util.caches.descriptors import cachedInlineCallbacks from synapse.util.caches.response_cache import ResponseCache @@ -37,7 +35,6 @@ logger = logging.getLogger(__name__) REMOTE_ROOM_LIST_POLL_INTERVAL = 60 * 1000 - # This is used to indicate we should only return rooms published to the main list. EMPTY_THIRD_PARTY_ID = ThirdPartyInstanceID(None, None) @@ -72,6 +69,8 @@ class RoomListHandler(BaseHandler): This can be (None, None) to indicate the main list, or a particular appservice and network id to use an appservice specific one. Setting to None returns all public rooms across all lists. + from_federation (bool): true iff the request comes from the federation + API """ if not self.enable_room_list_search: return defer.succeed({"chunk": [], "total_room_count_estimate": 0}) @@ -133,239 +132,109 @@ class RoomListHandler(BaseHandler): from_federation (bool): Whether this request originated from a federating server or a client. Used for room filtering. timeout (int|None): Amount of seconds to wait for a response before - timing out. + timing out. TODO """ - if since_token and since_token != "END": - since_token = RoomListNextBatch.from_token(since_token) - else: - since_token = None - rooms_to_order_value = {} - rooms_to_num_joined = {} + # Pagination tokens work by storing the room ID sent in the last batch, + # plus the direction (forwards or backwards). Next batch tokens always + # go forwards, prev batch tokens always go backwards. - newly_visible = [] - newly_unpublished = [] if since_token: - stream_token = since_token.stream_ordering - current_public_id = yield self.store.get_current_public_room_stream_id() - public_room_stream_id = since_token.public_room_stream_id - newly_visible, newly_unpublished = yield self.store.get_public_room_changes( - public_room_stream_id, current_public_id, network_tuple=network_tuple - ) - else: - stream_token = yield self.store.get_room_max_stream_ordering() - public_room_stream_id = yield self.store.get_current_public_room_stream_id() - - room_ids = yield self.store.get_public_room_ids_at_stream_id( - public_room_stream_id, network_tuple=network_tuple - ) - - # We want to return rooms in a particular order: the number of joined - # users. We then arbitrarily use the room_id as a tie breaker. - - @defer.inlineCallbacks - def get_order_for_room(room_id): - # Most of the rooms won't have changed between the since token and - # now (especially if the since token is "now"). So, we can ask what - # the current users are in a room (that will hit a cache) and then - # check if the room has changed since the since token. (We have to - # do it in that order to avoid races). - # If things have changed then fall back to getting the current state - # at the since token. - joined_users = yield self.store.get_users_in_room(room_id) - if self.store.has_room_changed_since(room_id, stream_token): - latest_event_ids = yield self.store.get_forward_extremeties_for_room( - room_id, stream_token - ) - - if not latest_event_ids: - return + batch_token = RoomListNextBatch.from_token(since_token) - joined_users = yield self.state_handler.get_current_users_in_room( - room_id, latest_event_ids - ) - - num_joined_users = len(joined_users) - rooms_to_num_joined[room_id] = num_joined_users + last_room_id = batch_token.last_room_id + forwards = batch_token.direction_is_forward + else: + batch_token = None - if num_joined_users == 0: - return + last_room_id = None + forwards = True - # We want larger rooms to be first, hence negating num_joined_users - rooms_to_order_value[room_id] = (-num_joined_users, room_id) + # we request one more than wanted to see if there are more pages to come + probing_limit = limit + 1 if limit is not None else None - logger.info( - "Getting ordering for %i rooms since %s", len(room_ids), stream_token + results = yield self.store.get_largest_public_rooms( + network_tuple, + search_filter, + probing_limit, + last_room_id=last_room_id, + forwards=forwards, + ignore_non_federatable=from_federation, ) - yield concurrently_execute(get_order_for_room, room_ids, 10) - sorted_entries = sorted(rooms_to_order_value.items(), key=lambda e: e[1]) - sorted_rooms = [room_id for room_id, _ in sorted_entries] + def build_room_entry(room): + entry = { + "room_id": room["room_id"], + "name": room["name"], + "topic": room["topic"], + "canonical_alias": room["canonical_alias"], + "num_joined_members": room["joined_members"], + "avatar_url": room["avatar"], + "world_readable": room["history_visibility"] == "world_readable", + "guest_can_join": room["guest_access"] == "can_join", + } - # `sorted_rooms` should now be a list of all public room ids that is - # stable across pagination. Therefore, we can use indices into this - # list as our pagination tokens. + # Filter out Nones – rather omit the field altogether + return {k: v for k, v in entry.items() if v is not None} - # Filter out rooms that we don't want to return - rooms_to_scan = [ - r - for r in sorted_rooms - if r not in newly_unpublished and rooms_to_num_joined[r] > 0 - ] + results = [build_room_entry(r) for r in results] - total_room_count = len(rooms_to_scan) + response = {} + num_results = len(results) + if limit is not None: + more_to_come = num_results == probing_limit - if since_token: - # Filter out rooms we've already returned previously - # `since_token.current_limit` is the index of the last room we - # sent down, so we exclude it and everything before/after it. - if since_token.direction_is_forward: - rooms_to_scan = rooms_to_scan[since_token.current_limit + 1 :] + # Depending on direction we trim either the front or back. + if forwards: + results = results[:limit] else: - rooms_to_scan = rooms_to_scan[: since_token.current_limit] - rooms_to_scan.reverse() - - logger.info("After sorting and filtering, %i rooms remain", len(rooms_to_scan)) - - # _append_room_entry_to_chunk will append to chunk but will stop if - # len(chunk) > limit - # - # Normally we will generate enough results on the first iteration here, - # but if there is a search filter, _append_room_entry_to_chunk may - # filter some results out, in which case we loop again. - # - # We don't want to scan over the entire range either as that - # would potentially waste a lot of work. - # - # XXX if there is no limit, we may end up DoSing the server with - # calls to get_current_state_ids for every single room on the - # server. Surely we should cap this somehow? - # - if limit: - step = limit + 1 + results = results[-limit:] else: - # step cannot be zero - step = len(rooms_to_scan) if len(rooms_to_scan) != 0 else 1 - - chunk = [] - for i in range(0, len(rooms_to_scan), step): - if timeout and self.clock.time() > timeout: - raise Exception("Timed out searching room directory") - - batch = rooms_to_scan[i : i + step] - logger.info("Processing %i rooms for result", len(batch)) - yield concurrently_execute( - lambda r: self._append_room_entry_to_chunk( - r, - rooms_to_num_joined[r], - chunk, - limit, - search_filter, - from_federation=from_federation, - ), - batch, - 5, - ) - logger.info("Now %i rooms in result", len(chunk)) - if len(chunk) >= limit + 1: - break - - chunk.sort(key=lambda e: (-e["num_joined_members"], e["room_id"])) - - # Work out the new limit of the batch for pagination, or None if we - # know there are no more results that would be returned. - # i.e., [since_token.current_limit..new_limit] is the batch of rooms - # we've returned (or the reverse if we paginated backwards) - # We tried to pull out limit + 1 rooms above, so if we have <= limit - # then we know there are no more results to return - new_limit = None - if chunk and (not limit or len(chunk) > limit): - - if not since_token or since_token.direction_is_forward: - if limit: - chunk = chunk[:limit] - last_room_id = chunk[-1]["room_id"] + more_to_come = False + + if num_results > 0: + final_room_id = results[-1]["room_id"] + initial_room_id = results[0]["room_id"] + + if forwards: + if batch_token: + # If there was a token given then we assume that there + # must be previous results. + response["prev_batch"] = RoomListNextBatch( + last_room_id=initial_room_id, direction_is_forward=False + ).to_token() + + if more_to_come: + response["next_batch"] = RoomListNextBatch( + last_room_id=final_room_id, direction_is_forward=True + ).to_token() else: - if limit: - chunk = chunk[-limit:] - last_room_id = chunk[0]["room_id"] - - new_limit = sorted_rooms.index(last_room_id) - - results = {"chunk": chunk, "total_room_count_estimate": total_room_count} - - if since_token: - results["new_rooms"] = bool(newly_visible) - - if not since_token or since_token.direction_is_forward: - if new_limit is not None: - results["next_batch"] = RoomListNextBatch( - stream_ordering=stream_token, - public_room_stream_id=public_room_stream_id, - current_limit=new_limit, - direction_is_forward=True, - ).to_token() - - if since_token: - results["prev_batch"] = since_token.copy_and_replace( - direction_is_forward=False, - current_limit=since_token.current_limit + 1, - ).to_token() - else: - if new_limit is not None: - results["prev_batch"] = RoomListNextBatch( - stream_ordering=stream_token, - public_room_stream_id=public_room_stream_id, - current_limit=new_limit, - direction_is_forward=False, - ).to_token() - - if since_token: - results["next_batch"] = since_token.copy_and_replace( - direction_is_forward=True, - current_limit=since_token.current_limit - 1, - ).to_token() - - return results - - @defer.inlineCallbacks - def _append_room_entry_to_chunk( - self, - room_id, - num_joined_users, - chunk, - limit, - search_filter, - from_federation=False, - ): - """Generate the entry for a room in the public room list and append it - to the `chunk` if it matches the search filter - - Args: - room_id (str): The ID of the room. - num_joined_users (int): The number of joined users in the room. - chunk (list) - limit (int|None): Maximum amount of rooms to display. Function will - return if length of chunk is greater than limit + 1. - search_filter (dict|None) - from_federation (bool): Whether this request originated from a - federating server or a client. Used for room filtering. - """ - if limit and len(chunk) > limit + 1: - # We've already got enough, so lets just drop it. - return + if batch_token: + response["next_batch"] = RoomListNextBatch( + last_room_id=final_room_id, direction_is_forward=True + ).to_token() + + if more_to_come: + response["prev_batch"] = RoomListNextBatch( + last_room_id=initial_room_id, direction_is_forward=False + ).to_token() + + for room in results: + # populate search result entries with additional fields, namely + # 'aliases' + room_id = room["room_id"] + + aliases = yield self.store.get_aliases_for_room(room_id) + if aliases: + room["aliases"] = aliases - result = yield self.generate_room_entry(room_id, num_joined_users) - if not result: - return + response["chunk"] = results - if from_federation and not result.get("m.federate", True): - # This is a room that other servers cannot join. Do not show them - # this room. - return + response["total_room_count_estimate"] = yield self.store.count_public_rooms( + network_tuple, ignore_non_federatable=from_federation + ) - if _matches_room_entry(result, search_filter): - chunk.append(result) + return response @cachedInlineCallbacks(num_args=1, cache_context=True) def generate_room_entry( @@ -580,32 +449,18 @@ class RoomListNextBatch( namedtuple( "RoomListNextBatch", ( - "stream_ordering", # stream_ordering of the first public room list - "public_room_stream_id", # public room stream id for first public room list - "current_limit", # The number of previous rooms returned + "last_room_id", # The room_id to get rooms after/before "direction_is_forward", # Bool if this is a next_batch, false if prev_batch ), ) ): - - KEY_DICT = { - "stream_ordering": "s", - "public_room_stream_id": "p", - "current_limit": "n", - "direction_is_forward": "d", - } + KEY_DICT = {"last_room_id": "r", "direction_is_forward": "d"} REVERSE_KEY_DICT = {v: k for k, v in KEY_DICT.items()} @classmethod def from_token(cls, token): - if PY3: - # The argument raw=False is only available on new versions of - # msgpack, and only really needed on Python 3. Gate it behind - # a PY3 check to avoid causing issues on Debian-packaged versions. - decoded = msgpack.loads(decode_base64(token), raw=False) - else: - decoded = msgpack.loads(decode_base64(token)) + decoded = msgpack.loads(decode_base64(token), raw=False) return RoomListNextBatch( **{cls.REVERSE_KEY_DICT[key]: val for key, val in decoded.items()} ) diff --git a/synapse/rest/client/v1/room.py b/synapse/rest/client/v1/room.py index 6bf924dedc..9c1d41421c 100644 --- a/synapse/rest/client/v1/room.py +++ b/synapse/rest/client/v1/room.py @@ -361,6 +361,10 @@ class PublicRoomListRestServlet(TransactionRestServlet): limit = parse_integer(request, "limit", 0) since_token = parse_string(request, "since", None) + if limit == 0: + # zero is a special value which corresponds to no limit. + limit = None + handler = self.hs.get_room_list_handler() if server: data = yield handler.get_remote_public_room_list( @@ -398,6 +402,10 @@ class PublicRoomListRestServlet(TransactionRestServlet): else: network_tuple = ThirdPartyInstanceID.from_string(third_party_instance_id) + if limit == 0: + # zero is a special value which corresponds to no limit. + limit = None + handler = self.hs.get_room_list_handler() if server: data = yield handler.get_remote_public_room_list( diff --git a/synapse/storage/room.py b/synapse/storage/room.py index 08e13f3a3b..c02787a73d 100644 --- a/synapse/storage/room.py +++ b/synapse/storage/room.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2019 The Matrix.org Foundation C.I.C. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -63,103 +64,176 @@ class RoomWorkerStore(SQLBaseStore): desc="get_public_room_ids", ) - @cached(num_args=2, max_entries=100) - def get_public_room_ids_at_stream_id(self, stream_id, network_tuple): - """Get pulbic rooms for a particular list, or across all lists. + def count_public_rooms(self, network_tuple, ignore_non_federatable): + """Counts the number of public rooms as tracked in the room_stats_current + and room_stats_state table. Args: - stream_id (int) - network_tuple (ThirdPartyInstanceID): The list to use (None, None) - means the main list, None means all lsits. + network_tuple (ThirdPartyInstanceID|None) + ignore_non_federatable (bool): If true filters out non-federatable rooms """ - return self.runInteraction( - "get_public_room_ids_at_stream_id", - self.get_public_room_ids_at_stream_id_txn, - stream_id, - network_tuple=network_tuple, - ) - - def get_public_room_ids_at_stream_id_txn(self, txn, stream_id, network_tuple): - return { - rm - for rm, vis in self.get_published_at_stream_id_txn( - txn, stream_id, network_tuple=network_tuple - ).items() - if vis - } - def get_published_at_stream_id_txn(self, txn, stream_id, network_tuple): - if network_tuple: - # We want to get from a particular list. No aggregation required. + def _count_public_rooms_txn(txn): + query_args = [] + + if network_tuple: + if network_tuple.appservice_id: + published_sql = """ + SELECT room_id from appservice_room_list + WHERE appservice_id = ? AND network_id = ? + """ + query_args.append(network_tuple.appservice_id) + query_args.append(network_tuple.network_id) + else: + published_sql = """ + SELECT room_id FROM rooms WHERE is_public + """ + else: + published_sql = """ + SELECT room_id FROM rooms WHERE is_public + UNION SELECT room_id from appservice_room_list + """ sql = """ - SELECT room_id, visibility FROM public_room_list_stream - INNER JOIN ( - SELECT room_id, max(stream_id) AS stream_id - FROM public_room_list_stream - WHERE stream_id <= ? %s - GROUP BY room_id - ) grouped USING (room_id, stream_id) - """ + SELECT + COALESCE(COUNT(*), 0) + FROM ( + %(published_sql)s + ) published + INNER JOIN room_stats_state USING (room_id) + INNER JOIN room_stats_current USING (room_id) + WHERE + ( + join_rules = 'public' OR history_visibility = 'world_readable' + ) + AND joined_members > 0 + """ % { + "published_sql": published_sql + } - if network_tuple.appservice_id is not None: - txn.execute( - sql % ("AND appservice_id = ? AND network_id = ?",), - (stream_id, network_tuple.appservice_id, network_tuple.network_id), - ) - else: - txn.execute(sql % ("AND appservice_id IS NULL",), (stream_id,)) - return dict(txn) - else: - # We want to get from all lists, so we need to aggregate the results + txn.execute(sql, query_args) + return txn.fetchone()[0] - logger.info("Executing full list") + return self.runInteraction("count_public_rooms", _count_public_rooms_txn) - sql = """ - SELECT room_id, visibility - FROM public_room_list_stream - INNER JOIN ( - SELECT - room_id, max(stream_id) AS stream_id, appservice_id, - network_id - FROM public_room_list_stream - WHERE stream_id <= ? - GROUP BY room_id, appservice_id, network_id - ) grouped USING (room_id, stream_id) - """ + @defer.inlineCallbacks + def get_largest_public_rooms( + self, + network_tuple, + search_filter, + limit, + last_room_id, + forwards, + ignore_non_federatable=False, + ): + """Gets the largest public rooms (where largest is in terms of joined + members, as tracked in the statistics table). - txn.execute(sql, (stream_id,)) + Args: + network_tuple (ThirdPartyInstanceID|None): + search_filter (dict|None): + limit (int|None): Maxmimum number of rows to return, unlimited otherwise. + last_room_id (str|None): if present, a room ID which bounds the + result set, and is always *excluded* from the result set. + forwards (bool): true iff going forwards, going backwards otherwise + ignore_non_federatable (bool): If true filters out non-federatable rooms. - results = {} - # A room is visible if its visible on any list. - for room_id, visibility in txn: - results[room_id] = bool(visibility) or results.get(room_id, False) + Returns: + Rooms in order: biggest number of joined users first. + We then arbitrarily use the room_id as a tie breaker. - return results + """ - def get_public_room_changes(self, prev_stream_id, new_stream_id, network_tuple): - def get_public_room_changes_txn(txn): - then_rooms = self.get_public_room_ids_at_stream_id_txn( - txn, prev_stream_id, network_tuple - ) + where_clauses = [] + query_args = [] - now_rooms_dict = self.get_published_at_stream_id_txn( - txn, new_stream_id, network_tuple - ) + if last_room_id: + if forwards: + where_clauses.append("room_id < ?") + else: + where_clauses.append("? < room_id") - now_rooms_visible = set(rm for rm, vis in now_rooms_dict.items() if vis) - now_rooms_not_visible = set( - rm for rm, vis in now_rooms_dict.items() if not vis + query_args += [last_room_id] + + if search_filter and search_filter.get("generic_search_term", None): + search_term = "%" + search_filter["generic_search_term"] + "%" + + where_clauses.append( + """ + ( + name LIKE ? + OR topic LIKE ? + OR canonical_alias LIKE ? + ) + """ ) + query_args += [search_term, search_term, search_term] + + if network_tuple: + if network_tuple.appservice_id: + published_sql = """ + SELECT room_id from appservice_room_list + WHERE appservice_id = ? AND network_id = ? + """ + query_args.append(network_tuple.appservice_id) + query_args.append(network_tuple.network_id) + else: + published_sql = """ + SELECT room_id FROM rooms WHERE is_public + """ + else: + published_sql = """ + SELECT room_id FROM rooms WHERE is_public + UNION SELECT room_id from appservice_room_list + """ - newly_visible = now_rooms_visible - then_rooms - newly_unpublished = now_rooms_not_visible & then_rooms + where_clause = "" + if where_clauses: + where_clause = " AND " + " AND ".join(where_clauses) + + sql = """ + SELECT + room_id, name, topic, canonical_alias, joined_members, + avatar, history_visibility, joined_members, guest_access + FROM ( + %(published_sql)s + ) published + INNER JOIN room_stats_state USING (room_id) + INNER JOIN room_stats_current USING (room_id) + WHERE + ( + join_rules = 'public' OR history_visibility = 'world_readable' + ) + AND joined_members > 0 + %(where_clause)s + ORDER BY joined_members %(dir)s, room_id %(dir)s + """ % { + "published_sql": published_sql, + "where_clause": where_clause, + "dir": "DESC" if forwards else "ASC", + } - return newly_visible, newly_unpublished + if limit is not None: + query_args.append(limit) - return self.runInteraction( - "get_public_room_changes", get_public_room_changes_txn + sql += """ + LIMIT ? + """ + + def _get_largest_public_rooms_txn(txn): + txn.execute(sql, query_args) + + results = self.cursor_to_dict(txn) + + if not forwards: + results.reverse() + + return results + + ret_val = yield self.runInteraction( + "get_largest_public_rooms", _get_largest_public_rooms_txn ) + defer.returnValue(ret_val) @cached(max_entries=10000) def is_room_blocked(self, room_id): diff --git a/synapse/storage/schema/delta/56/public_room_list_idx.sql b/synapse/storage/schema/delta/56/public_room_list_idx.sql new file mode 100644 index 0000000000..7be31ffebb --- /dev/null +++ b/synapse/storage/schema/delta/56/public_room_list_idx.sql @@ -0,0 +1,16 @@ +/* Copyright 2019 The Matrix.org Foundation C.I.C. + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE INDEX public_room_list_stream_network ON public_room_list_stream (appservice_id, network_id, room_id); diff --git a/tests/handlers/test_roomlist.py b/tests/handlers/test_roomlist.py deleted file mode 100644 index 61eebb6985..0000000000 --- a/tests/handlers/test_roomlist.py +++ /dev/null @@ -1,39 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2018 New Vector Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -from synapse.handlers.room_list import RoomListNextBatch - -import tests.unittest -import tests.utils - - -class RoomListTestCase(tests.unittest.TestCase): - """ Tests RoomList's RoomListNextBatch. """ - - def setUp(self): - pass - - def test_check_read_batch_tokens(self): - batch_token = RoomListNextBatch( - stream_ordering="abcdef", - public_room_stream_id="123", - current_limit=20, - direction_is_forward=True, - ).to_token() - next_batch = RoomListNextBatch.from_token(batch_token) - self.assertEquals(next_batch.stream_ordering, "abcdef") - self.assertEquals(next_batch.public_room_stream_id, "123") - self.assertEquals(next_batch.current_limit, 20) - self.assertEquals(next_batch.direction_is_forward, True) -- cgit 1.4.1 From 6527fa18c1e6f9bcb22318916b5e9534c91c84c1 Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Wed, 2 Oct 2019 14:44:58 +0100 Subject: Add test case --- synapse/handlers/federation.py | 2 +- tests/handlers/test_federation.py | 83 +++++++++++++++++++++++++++++++++++++++ 2 files changed, 84 insertions(+), 1 deletion(-) create mode 100644 tests/handlers/test_federation.py (limited to 'tests') diff --git a/synapse/handlers/federation.py b/synapse/handlers/federation.py index 75d79bb8e4..91f3a69298 100644 --- a/synapse/handlers/federation.py +++ b/synapse/handlers/federation.py @@ -2570,7 +2570,7 @@ class FederationHandler(BaseHandler): ) try: - self.auth.check_from_context(room_version, event, context) + yield self.auth.check_from_context(room_version, event, context) except AuthError as e: logger.warn("Denying third party invite %r because %s", event, e) raise e diff --git a/tests/handlers/test_federation.py b/tests/handlers/test_federation.py new file mode 100644 index 0000000000..20416a0142 --- /dev/null +++ b/tests/handlers/test_federation.py @@ -0,0 +1,83 @@ +# -*- coding: utf-8 -*- +# Copyright 2019 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +from synapse.api.constants import EventTypes +from synapse.api.errors import AuthError, Codes +from synapse.rest import admin +from synapse.rest.client.v1 import login, room + +from tests import unittest + + +class FederationTestCase(unittest.HomeserverTestCase): + servlets = [ + admin.register_servlets, + login.register_servlets, + room.register_servlets, + ] + + def make_homeserver(self, reactor, clock): + hs = self.setup_test_homeserver(http_client=None) + self.handler = hs.get_handlers().federation_handler + self.store = hs.get_datastore() + return hs + + def test_exchange_revoked_invite(self): + user_id = self.register_user("kermit", "test") + tok = self.login("kermit", "test") + + room_id = self.helper.create_room_as( + room_creator=user_id, tok=tok + ) + + # Send a 3PID invite event with an empty body so it's considered as a revoked one. + invite_token = "sometoken" + self.helper.send_state( + room_id=room_id, + event_type=EventTypes.ThirdPartyInvite, + state_key=invite_token, + body={}, + tok=tok, + ) + + d = self.handler.on_exchange_third_party_invite_request( + room_id=room_id, + event_dict={ + "type": EventTypes.Member, + "room_id": room_id, + "sender": user_id, + "state_key": "@someone:example.org", + "content": { + "membership": "invite", + "third_party_invite": { + "display_name": "alice", + "signed": { + "mxid": "@alice:localhost", + "token": invite_token, + "signatures": { + "magic.forest": { + "ed25519:3": "fQpGIW1Snz+pwLZu6sTy2aHy/DYWWTspTJRPyNp0PKkymfIsNffysMl6ObMMFdIJhk6g6pwlIqZ54rxo8SLmAg" + } + } + } + } + } + } + ) + + failure = self.get_failure(d, AuthError).value + + self.assertEqual(failure.code, 403, failure) + self.assertEqual(failure.errcode, Codes.FORBIDDEN, failure) + self.assertEqual(failure.msg, "You are not invited to this room.") -- cgit 1.4.1 From ebcb6a30d7b1bdb859a1fd22d567b163a1488763 Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Thu, 3 Oct 2019 11:29:07 +0100 Subject: Lint --- tests/handlers/test_federation.py | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) (limited to 'tests') diff --git a/tests/handlers/test_federation.py b/tests/handlers/test_federation.py index 20416a0142..a18dfc0e96 100644 --- a/tests/handlers/test_federation.py +++ b/tests/handlers/test_federation.py @@ -69,11 +69,11 @@ class FederationTestCase(unittest.HomeserverTestCase): "magic.forest": { "ed25519:3": "fQpGIW1Snz+pwLZu6sTy2aHy/DYWWTspTJRPyNp0PKkymfIsNffysMl6ObMMFdIJhk6g6pwlIqZ54rxo8SLmAg" } - } - } - } - } - } + }, + }, + }, + }, + }, ) failure = self.get_failure(d, AuthError).value -- cgit 1.4.1 From 8a5e8e829b98687ea274fae47db3aa801b6f97d3 Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Thu, 3 Oct 2019 11:30:43 +0100 Subject: Lint (again) --- tests/handlers/test_federation.py | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) (limited to 'tests') diff --git a/tests/handlers/test_federation.py b/tests/handlers/test_federation.py index a18dfc0e96..d56220f403 100644 --- a/tests/handlers/test_federation.py +++ b/tests/handlers/test_federation.py @@ -37,9 +37,7 @@ class FederationTestCase(unittest.HomeserverTestCase): user_id = self.register_user("kermit", "test") tok = self.login("kermit", "test") - room_id = self.helper.create_room_as( - room_creator=user_id, tok=tok - ) + room_id = self.helper.create_room_as(room_creator=user_id, tok=tok) # Send a 3PID invite event with an empty body so it's considered as a revoked one. invite_token = "sometoken" -- cgit 1.4.1 From 4535a07f4af0854eed0cfd171e4032bdd3f39cbb Mon Sep 17 00:00:00 2001 From: Hubert Chathi Date: Wed, 9 Oct 2019 17:54:03 -0400 Subject: make version optional in body of e2e backup version update to agree with latest version of the MSC --- synapse/handlers/e2e_room_keys.py | 4 +-- synapse/rest/client/v2_alpha/room_keys.py | 2 +- tests/handlers/test_e2e_room_keys.py | 47 ++++++++++++++++++++----------- 3 files changed, 34 insertions(+), 19 deletions(-) (limited to 'tests') diff --git a/synapse/handlers/e2e_room_keys.py b/synapse/handlers/e2e_room_keys.py index a9d80f708c..0cea445f0d 100644 --- a/synapse/handlers/e2e_room_keys.py +++ b/synapse/handlers/e2e_room_keys.py @@ -352,8 +352,8 @@ class E2eRoomKeysHandler(object): A deferred of an empty dict. """ if "version" not in version_info: - raise SynapseError(400, "Missing version in body", Codes.MISSING_PARAM) - if version_info["version"] != version: + version_info["version"] = version + elif version_info["version"] != version: raise SynapseError( 400, "Version in body does not match", Codes.INVALID_PARAM ) diff --git a/synapse/rest/client/v2_alpha/room_keys.py b/synapse/rest/client/v2_alpha/room_keys.py index df4f44cd36..d596786430 100644 --- a/synapse/rest/client/v2_alpha/room_keys.py +++ b/synapse/rest/client/v2_alpha/room_keys.py @@ -375,7 +375,7 @@ class RoomKeysVersionServlet(RestServlet): "ed25519:something": "hijklmnop" } }, - "version": "42" + "version": "12345" } HTTP/1.1 200 OK diff --git a/tests/handlers/test_e2e_room_keys.py b/tests/handlers/test_e2e_room_keys.py index c4503c1611..c700a2fad1 100644 --- a/tests/handlers/test_e2e_room_keys.py +++ b/tests/handlers/test_e2e_room_keys.py @@ -187,9 +187,8 @@ class E2eRoomKeysHandlerTestCase(unittest.TestCase): self.assertEqual(res, 404) @defer.inlineCallbacks - def test_update_bad_version(self): - """Check that we get a 400 if the version in the body is missing or - doesn't match + def test_update_missing_version(self): + """Check that the update succeeds if the version is missing from the body """ version = yield self.handler.create_version( self.local_user, @@ -197,19 +196,35 @@ class E2eRoomKeysHandlerTestCase(unittest.TestCase): ) self.assertEqual(version, "1") - res = None - try: - yield self.handler.update_version( - self.local_user, - version, - { - "algorithm": "m.megolm_backup.v1", - "auth_data": "revised_first_version_auth_data", - }, - ) - except errors.SynapseError as e: - res = e.code - self.assertEqual(res, 400) + yield self.handler.update_version( + self.local_user, + version, + { + "algorithm": "m.megolm_backup.v1", + "auth_data": "revised_first_version_auth_data", + }, + ) + + # check we can retrieve it as the current version + res = yield self.handler.get_version_info(self.local_user) + self.assertDictEqual( + res, + { + "algorithm": "m.megolm_backup.v1", + "auth_data": "revised_first_version_auth_data", + "version": version, + }, + ) + + @defer.inlineCallbacks + def test_update_bad_version(self): + """Check that we get a 400 if the version in the body doesn't match + """ + version = yield self.handler.create_version( + self.local_user, + {"algorithm": "m.megolm_backup.v1", "auth_data": "first_version_auth_data"}, + ) + self.assertEqual(version, "1") res = None try: -- cgit 1.4.1 From f743108a94658eb1dbaf168d39874272f756a386 Mon Sep 17 00:00:00 2001 From: Amber Brown Date: Thu, 10 Oct 2019 09:39:35 +0100 Subject: Refactor HomeserverConfig so it can be typechecked (#6137) --- changelog.d/6137.misc | 1 + mypy.ini | 16 ++- synapse/config/_base.py | 191 +++++++++++++++++++++++------- synapse/config/_base.pyi | 135 +++++++++++++++++++++ synapse/config/api.py | 2 + synapse/config/appservice.py | 2 + synapse/config/captcha.py | 2 + synapse/config/cas.py | 2 + synapse/config/consent_config.py | 3 + synapse/config/database.py | 2 + synapse/config/emailconfig.py | 2 + synapse/config/groups.py | 2 + synapse/config/homeserver.py | 68 +++++------ synapse/config/jwt_config.py | 2 + synapse/config/key.py | 2 + synapse/config/logger.py | 2 + synapse/config/metrics.py | 2 + synapse/config/password.py | 2 + synapse/config/password_auth_providers.py | 2 + synapse/config/push.py | 2 + synapse/config/ratelimiting.py | 2 + synapse/config/registration.py | 4 + synapse/config/repository.py | 2 + synapse/config/room_directory.py | 2 + synapse/config/saml2_config.py | 2 + synapse/config/server.py | 2 + synapse/config/server_notices_config.py | 2 + synapse/config/spam_checker.py | 2 + synapse/config/stats.py | 2 + synapse/config/third_party_event_rules.py | 2 + synapse/config/tls.py | 9 +- synapse/config/tracer.py | 2 + synapse/config/user_directory.py | 2 + synapse/config/voip.py | 2 + synapse/config/workers.py | 2 + tests/config/test_tls.py | 25 ++-- tox.ini | 3 +- 37 files changed, 415 insertions(+), 94 deletions(-) create mode 100644 changelog.d/6137.misc create mode 100644 synapse/config/_base.pyi (limited to 'tests') diff --git a/changelog.d/6137.misc b/changelog.d/6137.misc new file mode 100644 index 0000000000..92a02e71c3 --- /dev/null +++ b/changelog.d/6137.misc @@ -0,0 +1 @@ +Refactor configuration loading to allow better typechecking. diff --git a/mypy.ini b/mypy.ini index 8788574ee3..ffadaddc0b 100644 --- a/mypy.ini +++ b/mypy.ini @@ -4,10 +4,6 @@ plugins=mypy_zope:plugin follow_imports=skip mypy_path=stubs -[mypy-synapse.config.homeserver] -# this is a mess because of the metaclass shenanigans -ignore_errors = True - [mypy-zope] ignore_missing_imports = True @@ -52,3 +48,15 @@ ignore_missing_imports = True [mypy-signedjson.*] ignore_missing_imports = True + +[mypy-prometheus_client.*] +ignore_missing_imports = True + +[mypy-service_identity.*] +ignore_missing_imports = True + +[mypy-daemonize] +ignore_missing_imports = True + +[mypy-sentry_sdk] +ignore_missing_imports = True diff --git a/synapse/config/_base.py b/synapse/config/_base.py index 31f6530978..08619404bb 100644 --- a/synapse/config/_base.py +++ b/synapse/config/_base.py @@ -18,7 +18,9 @@ import argparse import errno import os +from collections import OrderedDict from textwrap import dedent +from typing import Any, MutableMapping, Optional from six import integer_types @@ -51,7 +53,56 @@ Missing mandatory `server_name` config option. """ +def path_exists(file_path): + """Check if a file exists + + Unlike os.path.exists, this throws an exception if there is an error + checking if the file exists (for example, if there is a perms error on + the parent dir). + + Returns: + bool: True if the file exists; False if not. + """ + try: + os.stat(file_path) + return True + except OSError as e: + if e.errno != errno.ENOENT: + raise e + return False + + class Config(object): + """ + A configuration section, containing configuration keys and values. + + Attributes: + section (str): The section title of this config object, such as + "tls" or "logger". This is used to refer to it on the root + logger (for example, `config.tls.some_option`). Must be + defined in subclasses. + """ + + section = None + + def __init__(self, root_config=None): + self.root = root_config + + def __getattr__(self, item: str) -> Any: + """ + Try and fetch a configuration option that does not exist on this class. + + This is so that existing configs that rely on `self.value`, where value + is actually from a different config section, continue to work. + """ + if item in ["generate_config_section", "read_config"]: + raise AttributeError(item) + + if self.root is None: + raise AttributeError(item) + else: + return self.root._get_unclassed_config(self.section, item) + @staticmethod def parse_size(value): if isinstance(value, integer_types): @@ -88,22 +139,7 @@ class Config(object): @classmethod def path_exists(cls, file_path): - """Check if a file exists - - Unlike os.path.exists, this throws an exception if there is an error - checking if the file exists (for example, if there is a perms error on - the parent dir). - - Returns: - bool: True if the file exists; False if not. - """ - try: - os.stat(file_path) - return True - except OSError as e: - if e.errno != errno.ENOENT: - raise e - return False + return path_exists(file_path) @classmethod def check_file(cls, file_path, config_name): @@ -136,42 +172,106 @@ class Config(object): with open(file_path) as file_stream: return file_stream.read() - def invoke_all(self, name, *args, **kargs): - """Invoke all instance methods with the given name and arguments in the - class's MRO. + +class RootConfig(object): + """ + Holder of an application's configuration. + + What configuration this object holds is defined by `config_classes`, a list + of Config classes that will be instantiated and given the contents of a + configuration file to read. They can then be accessed on this class by their + section name, defined in the Config or dynamically set to be the name of the + class, lower-cased and with "Config" removed. + """ + + config_classes = [] + + def __init__(self): + self._configs = OrderedDict() + + for config_class in self.config_classes: + if config_class.section is None: + raise ValueError("%r requires a section name" % (config_class,)) + + try: + conf = config_class(self) + except Exception as e: + raise Exception("Failed making %s: %r" % (config_class.section, e)) + self._configs[config_class.section] = conf + + def __getattr__(self, item: str) -> Any: + """ + Redirect lookups on this object either to config objects, or values on + config objects, so that `config.tls.blah` works, as well as legacy uses + of things like `config.server_name`. It will first look up the config + section name, and then values on those config classes. + """ + if item in self._configs.keys(): + return self._configs[item] + + return self._get_unclassed_config(None, item) + + def _get_unclassed_config(self, asking_section: Optional[str], item: str): + """ + Fetch a config value from one of the instantiated config classes that + has not been fetched directly. + + Args: + asking_section: If this check is coming from a Config child, which + one? This section will not be asked if it has the value. + item: The configuration value key. + + Raises: + AttributeError if no config classes have the config key. The body + will contain what sections were checked. + """ + for key, val in self._configs.items(): + if key == asking_section: + continue + + if item in dir(val): + return getattr(val, item) + + raise AttributeError(item, "not found in %s" % (list(self._configs.keys()),)) + + def invoke_all(self, func_name: str, *args, **kwargs) -> MutableMapping[str, Any]: + """ + Invoke a function on all instantiated config objects this RootConfig is + configured to use. Args: - name (str): Name of function to invoke + func_name: Name of function to invoke *args **kwargs - Returns: - list: The list of the return values from each method called + ordered dictionary of config section name and the result of the + function from it. """ - results = [] - for cls in type(self).mro(): - if name in cls.__dict__: - results.append(getattr(cls, name)(self, *args, **kargs)) - return results + res = OrderedDict() + + for name, config in self._configs.items(): + if hasattr(config, func_name): + res[name] = getattr(config, func_name)(*args, **kwargs) + + return res @classmethod - def invoke_all_static(cls, name, *args, **kargs): - """Invoke all static methods with the given name and arguments in the - class's MRO. + def invoke_all_static(cls, func_name: str, *args, **kwargs): + """ + Invoke a static function on config objects this RootConfig is + configured to use. Args: - name (str): Name of function to invoke + func_name: Name of function to invoke *args **kwargs - Returns: - list: The list of the return values from each method called + ordered dictionary of config section name and the result of the + function from it. """ - results = [] - for c in cls.mro(): - if name in c.__dict__: - results.append(getattr(c, name)(*args, **kargs)) - return results + for config in cls.config_classes: + if hasattr(config, func_name): + getattr(config, func_name)(*args, **kwargs) def generate_config( self, @@ -187,7 +287,8 @@ class Config(object): tls_private_key_path=None, acme_domain=None, ): - """Build a default configuration file + """ + Build a default configuration file This is used when the user explicitly asks us to generate a config file (eg with --generate_config). @@ -242,6 +343,7 @@ class Config(object): Returns: str: the yaml config file """ + return "\n\n".join( dedent(conf) for conf in self.invoke_all( @@ -257,7 +359,7 @@ class Config(object): tls_certificate_path=tls_certificate_path, tls_private_key_path=tls_private_key_path, acme_domain=acme_domain, - ) + ).values() ) @classmethod @@ -444,7 +546,7 @@ class Config(object): ) (config_path,) = config_files - if not cls.path_exists(config_path): + if not path_exists(config_path): print("Generating config file %s" % (config_path,)) if config_args.data_directory: @@ -469,7 +571,7 @@ class Config(object): open_private_ports=config_args.open_private_ports, ) - if not cls.path_exists(config_dir_path): + if not path_exists(config_dir_path): os.makedirs(config_dir_path) with open(config_path, "w") as config_file: config_file.write("# vim:ft=yaml\n\n") @@ -518,7 +620,7 @@ class Config(object): return obj - def parse_config_dict(self, config_dict, config_dir_path, data_dir_path): + def parse_config_dict(self, config_dict, config_dir_path=None, data_dir_path=None): """Read the information from the config dict into this Config object. Args: @@ -607,3 +709,6 @@ def find_config_files(search_paths): else: config_files.append(config_path) return config_files + + +__all__ = ["Config", "RootConfig"] diff --git a/synapse/config/_base.pyi b/synapse/config/_base.pyi new file mode 100644 index 0000000000..86bc965ee4 --- /dev/null +++ b/synapse/config/_base.pyi @@ -0,0 +1,135 @@ +from typing import Any, List, Optional + +from synapse.config import ( + api, + appservice, + captcha, + cas, + consent_config, + database, + emailconfig, + groups, + jwt_config, + key, + logger, + metrics, + password, + password_auth_providers, + push, + ratelimiting, + registration, + repository, + room_directory, + saml2_config, + server, + server_notices_config, + spam_checker, + stats, + third_party_event_rules, + tls, + tracer, + user_directory, + voip, + workers, +) + +class ConfigError(Exception): ... + +MISSING_REPORT_STATS_CONFIG_INSTRUCTIONS: str +MISSING_REPORT_STATS_SPIEL: str +MISSING_SERVER_NAME: str + +def path_exists(file_path: str): ... + +class RootConfig: + server: server.ServerConfig + tls: tls.TlsConfig + database: database.DatabaseConfig + logging: logger.LoggingConfig + ratelimit: ratelimiting.RatelimitConfig + media: repository.ContentRepositoryConfig + captcha: captcha.CaptchaConfig + voip: voip.VoipConfig + registration: registration.RegistrationConfig + metrics: metrics.MetricsConfig + api: api.ApiConfig + appservice: appservice.AppServiceConfig + key: key.KeyConfig + saml2: saml2_config.SAML2Config + cas: cas.CasConfig + jwt: jwt_config.JWTConfig + password: password.PasswordConfig + email: emailconfig.EmailConfig + worker: workers.WorkerConfig + authproviders: password_auth_providers.PasswordAuthProviderConfig + push: push.PushConfig + spamchecker: spam_checker.SpamCheckerConfig + groups: groups.GroupsConfig + userdirectory: user_directory.UserDirectoryConfig + consent: consent_config.ConsentConfig + stats: stats.StatsConfig + servernotices: server_notices_config.ServerNoticesConfig + roomdirectory: room_directory.RoomDirectoryConfig + thirdpartyrules: third_party_event_rules.ThirdPartyRulesConfig + tracer: tracer.TracerConfig + + config_classes: List = ... + def __init__(self) -> None: ... + def invoke_all(self, func_name: str, *args: Any, **kwargs: Any): ... + @classmethod + def invoke_all_static(cls, func_name: str, *args: Any, **kwargs: Any) -> None: ... + def __getattr__(self, item: str): ... + def parse_config_dict( + self, + config_dict: Any, + config_dir_path: Optional[Any] = ..., + data_dir_path: Optional[Any] = ..., + ) -> None: ... + read_config: Any = ... + def generate_config( + self, + config_dir_path: str, + data_dir_path: str, + server_name: str, + generate_secrets: bool = ..., + report_stats: Optional[str] = ..., + open_private_ports: bool = ..., + listeners: Optional[Any] = ..., + database_conf: Optional[Any] = ..., + tls_certificate_path: Optional[str] = ..., + tls_private_key_path: Optional[str] = ..., + acme_domain: Optional[str] = ..., + ): ... + @classmethod + def load_or_generate_config(cls, description: Any, argv: Any): ... + @classmethod + def load_config(cls, description: Any, argv: Any): ... + @classmethod + def add_arguments_to_parser(cls, config_parser: Any) -> None: ... + @classmethod + def load_config_with_parser(cls, parser: Any, argv: Any): ... + def generate_missing_files( + self, config_dict: dict, config_dir_path: str + ) -> None: ... + +class Config: + root: RootConfig + def __init__(self, root_config: Optional[RootConfig] = ...) -> None: ... + def __getattr__(self, item: str, from_root: bool = ...): ... + @staticmethod + def parse_size(value: Any): ... + @staticmethod + def parse_duration(value: Any): ... + @staticmethod + def abspath(file_path: Optional[str]): ... + @classmethod + def path_exists(cls, file_path: str): ... + @classmethod + def check_file(cls, file_path: str, config_name: str): ... + @classmethod + def ensure_directory(cls, dir_path: str): ... + @classmethod + def read_file(cls, file_path: str, config_name: str): ... + +def read_config_files(config_files: List[str]): ... +def find_config_files(search_paths: List[str]): ... diff --git a/synapse/config/api.py b/synapse/config/api.py index dddea79a8a..74cd53a8ed 100644 --- a/synapse/config/api.py +++ b/synapse/config/api.py @@ -18,6 +18,8 @@ from ._base import Config class ApiConfig(Config): + section = "api" + def read_config(self, config, **kwargs): self.room_invite_state_types = config.get( "room_invite_state_types", diff --git a/synapse/config/appservice.py b/synapse/config/appservice.py index 28d36b1bc3..9b4682222d 100644 --- a/synapse/config/appservice.py +++ b/synapse/config/appservice.py @@ -30,6 +30,8 @@ logger = logging.getLogger(__name__) class AppServiceConfig(Config): + section = "appservice" + def read_config(self, config, **kwargs): self.app_service_config_files = config.get("app_service_config_files", []) self.notify_appservices = config.get("notify_appservices", True) diff --git a/synapse/config/captcha.py b/synapse/config/captcha.py index 8dac8152cf..44bd5c6799 100644 --- a/synapse/config/captcha.py +++ b/synapse/config/captcha.py @@ -16,6 +16,8 @@ from ._base import Config class CaptchaConfig(Config): + section = "captcha" + def read_config(self, config, **kwargs): self.recaptcha_private_key = config.get("recaptcha_private_key") self.recaptcha_public_key = config.get("recaptcha_public_key") diff --git a/synapse/config/cas.py b/synapse/config/cas.py index ebe34d933b..b916c3aa66 100644 --- a/synapse/config/cas.py +++ b/synapse/config/cas.py @@ -22,6 +22,8 @@ class CasConfig(Config): cas_server_url: URL of CAS server """ + section = "cas" + def read_config(self, config, **kwargs): cas_config = config.get("cas_config", None) if cas_config: diff --git a/synapse/config/consent_config.py b/synapse/config/consent_config.py index 48976e17b1..62c4c44d60 100644 --- a/synapse/config/consent_config.py +++ b/synapse/config/consent_config.py @@ -73,6 +73,9 @@ DEFAULT_CONFIG = """\ class ConsentConfig(Config): + + section = "consent" + def __init__(self, *args): super(ConsentConfig, self).__init__(*args) diff --git a/synapse/config/database.py b/synapse/config/database.py index 118aafbd4a..0e2509f0b1 100644 --- a/synapse/config/database.py +++ b/synapse/config/database.py @@ -21,6 +21,8 @@ from ._base import Config class DatabaseConfig(Config): + section = "database" + def read_config(self, config, **kwargs): self.event_cache_size = self.parse_size(config.get("event_cache_size", "10K")) diff --git a/synapse/config/emailconfig.py b/synapse/config/emailconfig.py index d9b43de660..658897a77e 100644 --- a/synapse/config/emailconfig.py +++ b/synapse/config/emailconfig.py @@ -28,6 +28,8 @@ from ._base import Config, ConfigError class EmailConfig(Config): + section = "email" + def read_config(self, config, **kwargs): # TODO: We should separate better the email configuration from the notification # and account validity config. diff --git a/synapse/config/groups.py b/synapse/config/groups.py index 2a522b5f44..d6862d9a64 100644 --- a/synapse/config/groups.py +++ b/synapse/config/groups.py @@ -17,6 +17,8 @@ from ._base import Config class GroupsConfig(Config): + section = "groups" + def read_config(self, config, **kwargs): self.enable_group_creation = config.get("enable_group_creation", False) self.group_creation_prefix = config.get("group_creation_prefix", "") diff --git a/synapse/config/homeserver.py b/synapse/config/homeserver.py index 72acad4f18..6e348671c7 100644 --- a/synapse/config/homeserver.py +++ b/synapse/config/homeserver.py @@ -14,6 +14,7 @@ # See the License for the specific language governing permissions and # limitations under the License. +from ._base import RootConfig from .api import ApiConfig from .appservice import AppServiceConfig from .captcha import CaptchaConfig @@ -46,36 +47,37 @@ from .voip import VoipConfig from .workers import WorkerConfig -class HomeServerConfig( - ServerConfig, - TlsConfig, - DatabaseConfig, - LoggingConfig, - RatelimitConfig, - ContentRepositoryConfig, - CaptchaConfig, - VoipConfig, - RegistrationConfig, - MetricsConfig, - ApiConfig, - AppServiceConfig, - KeyConfig, - SAML2Config, - CasConfig, - JWTConfig, - PasswordConfig, - EmailConfig, - WorkerConfig, - PasswordAuthProviderConfig, - PushConfig, - SpamCheckerConfig, - GroupsConfig, - UserDirectoryConfig, - ConsentConfig, - StatsConfig, - ServerNoticesConfig, - RoomDirectoryConfig, - ThirdPartyRulesConfig, - TracerConfig, -): - pass +class HomeServerConfig(RootConfig): + + config_classes = [ + ServerConfig, + TlsConfig, + DatabaseConfig, + LoggingConfig, + RatelimitConfig, + ContentRepositoryConfig, + CaptchaConfig, + VoipConfig, + RegistrationConfig, + MetricsConfig, + ApiConfig, + AppServiceConfig, + KeyConfig, + SAML2Config, + CasConfig, + JWTConfig, + PasswordConfig, + EmailConfig, + WorkerConfig, + PasswordAuthProviderConfig, + PushConfig, + SpamCheckerConfig, + GroupsConfig, + UserDirectoryConfig, + ConsentConfig, + StatsConfig, + ServerNoticesConfig, + RoomDirectoryConfig, + ThirdPartyRulesConfig, + TracerConfig, + ] diff --git a/synapse/config/jwt_config.py b/synapse/config/jwt_config.py index 36d87cef03..a568726985 100644 --- a/synapse/config/jwt_config.py +++ b/synapse/config/jwt_config.py @@ -23,6 +23,8 @@ MISSING_JWT = """Missing jwt library. This is required for jwt login. class JWTConfig(Config): + section = "jwt" + def read_config(self, config, **kwargs): jwt_config = config.get("jwt_config", None) if jwt_config: diff --git a/synapse/config/key.py b/synapse/config/key.py index f039f96e9c..ec5d430afb 100644 --- a/synapse/config/key.py +++ b/synapse/config/key.py @@ -92,6 +92,8 @@ class TrustedKeyServer(object): class KeyConfig(Config): + section = "key" + def read_config(self, config, config_dir_path, **kwargs): # the signing key can be specified inline or in a separate file if "signing_key" in config: diff --git a/synapse/config/logger.py b/synapse/config/logger.py index 767ecfdf09..d609ec111b 100644 --- a/synapse/config/logger.py +++ b/synapse/config/logger.py @@ -84,6 +84,8 @@ root: class LoggingConfig(Config): + section = "logging" + def read_config(self, config, **kwargs): self.log_config = self.abspath(config.get("log_config")) self.no_redirect_stdio = config.get("no_redirect_stdio", False) diff --git a/synapse/config/metrics.py b/synapse/config/metrics.py index ec35a6b868..282a43bddb 100644 --- a/synapse/config/metrics.py +++ b/synapse/config/metrics.py @@ -34,6 +34,8 @@ class MetricsFlags(object): class MetricsConfig(Config): + section = "metrics" + def read_config(self, config, **kwargs): self.enable_metrics = config.get("enable_metrics", False) self.report_stats = config.get("report_stats", None) diff --git a/synapse/config/password.py b/synapse/config/password.py index d5b5953f2f..2a634ac751 100644 --- a/synapse/config/password.py +++ b/synapse/config/password.py @@ -20,6 +20,8 @@ class PasswordConfig(Config): """Password login configuration """ + section = "password" + def read_config(self, config, **kwargs): password_config = config.get("password_config", {}) if password_config is None: diff --git a/synapse/config/password_auth_providers.py b/synapse/config/password_auth_providers.py index c50e244394..9746bbc681 100644 --- a/synapse/config/password_auth_providers.py +++ b/synapse/config/password_auth_providers.py @@ -23,6 +23,8 @@ LDAP_PROVIDER = "ldap_auth_provider.LdapAuthProvider" class PasswordAuthProviderConfig(Config): + section = "authproviders" + def read_config(self, config, **kwargs): self.password_providers = [] # type: List[Any] providers = [] diff --git a/synapse/config/push.py b/synapse/config/push.py index 1b932722a5..0910958649 100644 --- a/synapse/config/push.py +++ b/synapse/config/push.py @@ -18,6 +18,8 @@ from ._base import Config class PushConfig(Config): + section = "push" + def read_config(self, config, **kwargs): push_config = config.get("push", {}) self.push_include_content = push_config.get("include_content", True) diff --git a/synapse/config/ratelimiting.py b/synapse/config/ratelimiting.py index 587e2862b7..947f653e03 100644 --- a/synapse/config/ratelimiting.py +++ b/synapse/config/ratelimiting.py @@ -36,6 +36,8 @@ class FederationRateLimitConfig(object): class RatelimitConfig(Config): + section = "ratelimiting" + def read_config(self, config, **kwargs): # Load the new-style messages config if it exists. Otherwise fall back diff --git a/synapse/config/registration.py b/synapse/config/registration.py index bef89e2bf4..b3e3e6dda2 100644 --- a/synapse/config/registration.py +++ b/synapse/config/registration.py @@ -24,6 +24,8 @@ from synapse.util.stringutils import random_string_with_symbols class AccountValidityConfig(Config): + section = "accountvalidity" + def __init__(self, config, synapse_config): self.enabled = config.get("enabled", False) self.renew_by_email_enabled = "renew_at" in config @@ -77,6 +79,8 @@ class AccountValidityConfig(Config): class RegistrationConfig(Config): + section = "registration" + def read_config(self, config, **kwargs): self.enable_registration = bool( strtobool(str(config.get("enable_registration", False))) diff --git a/synapse/config/repository.py b/synapse/config/repository.py index 14740891f3..d0205e14b9 100644 --- a/synapse/config/repository.py +++ b/synapse/config/repository.py @@ -78,6 +78,8 @@ def parse_thumbnail_requirements(thumbnail_sizes): class ContentRepositoryConfig(Config): + section = "media" + def read_config(self, config, **kwargs): # Only enable the media repo if either the media repo is enabled or the diff --git a/synapse/config/room_directory.py b/synapse/config/room_directory.py index a92693017b..7c9f05bde4 100644 --- a/synapse/config/room_directory.py +++ b/synapse/config/room_directory.py @@ -19,6 +19,8 @@ from ._base import Config, ConfigError class RoomDirectoryConfig(Config): + section = "roomdirectory" + def read_config(self, config, **kwargs): self.enable_room_list_search = config.get("enable_room_list_search", True) diff --git a/synapse/config/saml2_config.py b/synapse/config/saml2_config.py index ab34b41ca8..c407e13680 100644 --- a/synapse/config/saml2_config.py +++ b/synapse/config/saml2_config.py @@ -55,6 +55,8 @@ def _dict_merge(merge_dict, into_dict): class SAML2Config(Config): + section = "saml2" + def read_config(self, config, **kwargs): self.saml2_enabled = False diff --git a/synapse/config/server.py b/synapse/config/server.py index 709bd387e5..afc4d6a4ab 100644 --- a/synapse/config/server.py +++ b/synapse/config/server.py @@ -58,6 +58,8 @@ on how to configure the new listener. class ServerConfig(Config): + section = "server" + def read_config(self, config, **kwargs): self.server_name = config["server_name"] self.server_context = config.get("server_context", None) diff --git a/synapse/config/server_notices_config.py b/synapse/config/server_notices_config.py index 6d4285ef93..6ea2ea8869 100644 --- a/synapse/config/server_notices_config.py +++ b/synapse/config/server_notices_config.py @@ -59,6 +59,8 @@ class ServerNoticesConfig(Config): None if server notices are not enabled. """ + section = "servernotices" + def __init__(self, *args): super(ServerNoticesConfig, self).__init__(*args) self.server_notices_mxid = None diff --git a/synapse/config/spam_checker.py b/synapse/config/spam_checker.py index e40797ab50..36e0ddab5c 100644 --- a/synapse/config/spam_checker.py +++ b/synapse/config/spam_checker.py @@ -19,6 +19,8 @@ from ._base import Config class SpamCheckerConfig(Config): + section = "spamchecker" + def read_config(self, config, **kwargs): self.spam_checker = None diff --git a/synapse/config/stats.py b/synapse/config/stats.py index b18ddbd1fa..62485189ea 100644 --- a/synapse/config/stats.py +++ b/synapse/config/stats.py @@ -25,6 +25,8 @@ class StatsConfig(Config): Configuration for the behaviour of synapse's stats engine """ + section = "stats" + def read_config(self, config, **kwargs): self.stats_enabled = True self.stats_bucket_size = 86400 * 1000 diff --git a/synapse/config/third_party_event_rules.py b/synapse/config/third_party_event_rules.py index b3431441b9..10a99c792e 100644 --- a/synapse/config/third_party_event_rules.py +++ b/synapse/config/third_party_event_rules.py @@ -19,6 +19,8 @@ from ._base import Config class ThirdPartyRulesConfig(Config): + section = "thirdpartyrules" + def read_config(self, config, **kwargs): self.third_party_event_rules = None diff --git a/synapse/config/tls.py b/synapse/config/tls.py index fc47ba3e9a..f06341eb67 100644 --- a/synapse/config/tls.py +++ b/synapse/config/tls.py @@ -18,6 +18,7 @@ import os import warnings from datetime import datetime from hashlib import sha256 +from typing import List import six @@ -33,7 +34,9 @@ logger = logging.getLogger(__name__) class TlsConfig(Config): - def read_config(self, config, config_dir_path, **kwargs): + section = "tls" + + def read_config(self, config: dict, config_dir_path: str, **kwargs): acme_config = config.get("acme", None) if acme_config is None: @@ -57,7 +60,7 @@ class TlsConfig(Config): self.tls_certificate_file = self.abspath(config.get("tls_certificate_path")) self.tls_private_key_file = self.abspath(config.get("tls_private_key_path")) - if self.has_tls_listener(): + if self.root.server.has_tls_listener(): if not self.tls_certificate_file: raise ConfigError( "tls_certificate_path must be specified if TLS-enabled listeners are " @@ -108,7 +111,7 @@ class TlsConfig(Config): ) # Support globs (*) in whitelist values - self.federation_certificate_verification_whitelist = [] + self.federation_certificate_verification_whitelist = [] # type: List[str] for entry in fed_whitelist_entries: try: entry_regex = glob_to_regex(entry.encode("ascii").decode("ascii")) diff --git a/synapse/config/tracer.py b/synapse/config/tracer.py index 85d99a3166..8be1346113 100644 --- a/synapse/config/tracer.py +++ b/synapse/config/tracer.py @@ -19,6 +19,8 @@ from ._base import Config, ConfigError class TracerConfig(Config): + section = "tracing" + def read_config(self, config, **kwargs): opentracing_config = config.get("opentracing") if opentracing_config is None: diff --git a/synapse/config/user_directory.py b/synapse/config/user_directory.py index f6313e17d4..c8d19c5d6b 100644 --- a/synapse/config/user_directory.py +++ b/synapse/config/user_directory.py @@ -21,6 +21,8 @@ class UserDirectoryConfig(Config): Configuration for the behaviour of the /user_directory API """ + section = "userdirectory" + def read_config(self, config, **kwargs): self.user_directory_search_enabled = True self.user_directory_search_all_users = False diff --git a/synapse/config/voip.py b/synapse/config/voip.py index 2ca0e1cf70..a68a3068aa 100644 --- a/synapse/config/voip.py +++ b/synapse/config/voip.py @@ -16,6 +16,8 @@ from ._base import Config class VoipConfig(Config): + section = "voip" + def read_config(self, config, **kwargs): self.turn_uris = config.get("turn_uris", []) self.turn_shared_secret = config.get("turn_shared_secret") diff --git a/synapse/config/workers.py b/synapse/config/workers.py index 1ec4998625..fef72ed974 100644 --- a/synapse/config/workers.py +++ b/synapse/config/workers.py @@ -21,6 +21,8 @@ class WorkerConfig(Config): They have their own pid_file and listener configuration. They use the replication_url to talk to the main synapse process.""" + section = "worker" + def read_config(self, config, **kwargs): self.worker_app = config.get("worker_app") diff --git a/tests/config/test_tls.py b/tests/config/test_tls.py index b02780772a..1be6ff563b 100644 --- a/tests/config/test_tls.py +++ b/tests/config/test_tls.py @@ -21,17 +21,24 @@ import yaml from OpenSSL import SSL +from synapse.config._base import Config, RootConfig from synapse.config.tls import ConfigError, TlsConfig from synapse.crypto.context_factory import ClientTLSOptionsFactory from tests.unittest import TestCase -class TestConfig(TlsConfig): +class FakeServer(Config): + section = "server" + def has_tls_listener(self): return False +class TestConfig(RootConfig): + config_classes = [FakeServer, TlsConfig] + + class TLSConfigTests(TestCase): def test_warn_self_signed(self): """ @@ -202,13 +209,13 @@ s4niecZKPBizL6aucT59CsunNmmb5Glq8rlAcU+1ZTZZzGYqVYhF6axB9Qg= conf = TestConfig() conf.read_config( yaml.safe_load( - TestConfig().generate_config_section( + TestConfig().generate_config( "/config_dir_path", "my_super_secure_server", "/data_dir_path", - "/tls_cert_path", - "tls_private_key", - None, # This is the acme_domain + tls_certificate_path="/tls_cert_path", + tls_private_key_path="tls_private_key", + acme_domain=None, # This is the acme_domain ) ), "/config_dir_path", @@ -223,13 +230,13 @@ s4niecZKPBizL6aucT59CsunNmmb5Glq8rlAcU+1ZTZZzGYqVYhF6axB9Qg= conf = TestConfig() conf.read_config( yaml.safe_load( - TestConfig().generate_config_section( + TestConfig().generate_config( "/config_dir_path", "my_super_secure_server", "/data_dir_path", - "/tls_cert_path", - "tls_private_key", - "my_supe_secure_server", # This is the acme_domain + tls_certificate_path="/tls_cert_path", + tls_private_key_path="tls_private_key", + acme_domain="my_supe_secure_server", # This is the acme_domain ) ), "/config_dir_path", diff --git a/tox.ini b/tox.ini index 1bce10a4ce..367cc2ccf2 100644 --- a/tox.ini +++ b/tox.ini @@ -163,10 +163,9 @@ deps = {[base]deps} mypy mypy-zope - typeshed env = MYPYPATH = stubs/ extras = all -commands = mypy --show-traceback \ +commands = mypy --show-traceback --check-untyped-defs --show-error-codes --follow-imports=normal \ synapse/logging/ \ synapse/config/ -- cgit 1.4.1 From 1d6dd1c2944c22147258dda8ccf2777c68b38fba Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Thu, 10 Oct 2019 10:53:06 +0100 Subject: Move patch_inline_callbacks into synapse/ --- synapse/__init__.py | 2 +- synapse/util/patch_inline_callbacks.py | 179 +++++++++++++++++++++++++++++++++ tests/__init__.py | 4 +- tests/patch_inline_callbacks.py | 179 --------------------------------- 4 files changed, 182 insertions(+), 182 deletions(-) create mode 100644 synapse/util/patch_inline_callbacks.py delete mode 100644 tests/patch_inline_callbacks.py (limited to 'tests') diff --git a/synapse/__init__.py b/synapse/__init__.py index 1055f54e00..bf102244a9 100644 --- a/synapse/__init__.py +++ b/synapse/__init__.py @@ -20,7 +20,7 @@ import os import sys -from tests.patch_inline_callbacks import do_patch +from synapse.util.patch_inline_callbacks import do_patch # Check that we're not running on an unsupported Python version. if sys.version_info < (3, 5): diff --git a/synapse/util/patch_inline_callbacks.py b/synapse/util/patch_inline_callbacks.py new file mode 100644 index 0000000000..4fb49b0b2b --- /dev/null +++ b/synapse/util/patch_inline_callbacks.py @@ -0,0 +1,179 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import print_function + +import functools +import sys + +from twisted.internet import defer +from twisted.internet.defer import Deferred +from twisted.python.failure import Failure + + +def do_patch(): + """ + Patch defer.inlineCallbacks so that it checks the state of the logcontext on exit + """ + + from synapse.logging.context import LoggingContext + + orig_inline_callbacks = defer.inlineCallbacks + if hasattr(orig_inline_callbacks, "patched_by_synapse"): + return + + def new_inline_callbacks(f): + @functools.wraps(f) + def wrapped(*args, **kwargs): + start_context = LoggingContext.current_context() + changes = [] + orig = orig_inline_callbacks(_check_yield_points(f, changes, start_context)) + + try: + res = orig(*args, **kwargs) + except Exception: + if LoggingContext.current_context() != start_context: + for err in changes: + print(err, file=sys.stderr) + + err = "%s changed context from %s to %s on exception" % ( + f, + start_context, + LoggingContext.current_context(), + ) + print(err, file=sys.stderr) + raise Exception(err) + raise + + if not isinstance(res, Deferred) or res.called: + if LoggingContext.current_context() != start_context: + for err in changes: + print(err, file=sys.stderr) + + err = "Completed %s changed context from %s to %s" % ( + f, + start_context, + LoggingContext.current_context(), + ) + # print the error to stderr because otherwise all we + # see in travis-ci is the 500 error + print(err, file=sys.stderr) + raise Exception(err) + return res + + if LoggingContext.current_context() != LoggingContext.sentinel: + err = ( + "%s returned incomplete deferred in non-sentinel context " + "%s (start was %s)" + ) % (f, LoggingContext.current_context(), start_context) + print(err, file=sys.stderr) + raise Exception(err) + + def check_ctx(r): + if LoggingContext.current_context() != start_context: + for err in changes: + print(err, file=sys.stderr) + err = "%s completion of %s changed context from %s to %s" % ( + "Failure" if isinstance(r, Failure) else "Success", + f, + start_context, + LoggingContext.current_context(), + ) + print(err, file=sys.stderr) + raise Exception(err) + return r + + res.addBoth(check_ctx) + return res + + return wrapped + + defer.inlineCallbacks = new_inline_callbacks + new_inline_callbacks.patched_by_synapse = True + + +def _check_yield_points(f, changes, start_context): + """Wraps a generator that is about to be passed to defer.inlineCallbacks + checking that after every yield the log contexts are correct. + """ + + from synapse.logging.context import LoggingContext + + @functools.wraps(f) + def check_yield_points_inner(*args, **kwargs): + expected_context = start_context + + gen = f(*args, **kwargs) + + last_yield_line_no = 1 + result = None + while True: + try: + isFailure = isinstance(result, Failure) + if isFailure: + d = result.throwExceptionIntoGenerator(gen) + else: + d = gen.send(result) + except (StopIteration, defer._DefGen_Return) as e: + if LoggingContext.current_context() != expected_context: + # This happens when the context is lost sometime *after* the + # final yield and returning. E.g. we forgot to yield on a + # function that returns a deferred. + err = ( + "Function %r returned and changed context from %s to %s," + " in %s between %d and end of func" + % ( + f.__qualname__, + start_context, + LoggingContext.current_context(), + f.__code__.co_filename, + last_yield_line_no, + ) + ) + changes.append(err) + # raise Exception(err) + return getattr(e, "value", None) + + try: + result = yield d + except Exception as e: + result = Failure(e) + + frame = gen.gi_frame + + if LoggingContext.current_context() != expected_context: + # This happens because the context is lost sometime *after* the + # previous yield and *after* the current yield. E.g. the + # deferred we waited on didn't follow the rules, or we forgot to + # yield on a function between the two yield points. + err = ( + "%s changed context from %s to %s, happened between lines %d and %d in %s" + % ( + frame.f_code.co_name, + start_context, + LoggingContext.current_context(), + last_yield_line_no, + frame.f_lineno, + frame.f_code.co_filename, + ) + ) + changes.append(err) + # raise Exception(err) + + expected_context = LoggingContext.current_context() + + last_yield_line_no = frame.f_lineno + + return check_yield_points_inner diff --git a/tests/__init__.py b/tests/__init__.py index f7fc502f01..ed805db1c2 100644 --- a/tests/__init__.py +++ b/tests/__init__.py @@ -16,9 +16,9 @@ from twisted.trial import util -import tests.patch_inline_callbacks +from synapse.util.patch_inline_callbacks import do_patch # attempt to do the patch before we load any synapse code -tests.patch_inline_callbacks.do_patch() +do_patch() util.DEFAULT_TIMEOUT_DURATION = 20 diff --git a/tests/patch_inline_callbacks.py b/tests/patch_inline_callbacks.py deleted file mode 100644 index a35a1d3305..0000000000 --- a/tests/patch_inline_callbacks.py +++ /dev/null @@ -1,179 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2018 New Vector Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -from __future__ import print_function - -import functools -import sys - -from twisted.internet import defer -from twisted.internet.defer import Deferred -from twisted.python.failure import Failure - - -def do_patch(): - """ - Patch defer.inlineCallbacks so that it checks the state of the logcontext on exit - """ - - from synapse.logging.context import LoggingContext - - orig_inline_callbacks = defer.inlineCallbacks - if hasattr(orig_inline_callbacks, "patched_by_synapse"): - return - - def new_inline_callbacks(f): - @functools.wraps(f) - def wrapped(*args, **kwargs): - start_context = LoggingContext.current_context() - changes = [] - orig = orig_inline_callbacks(_check_yield_points(f, changes, start_context)) - - try: - res = orig(*args, **kwargs) - except Exception: - if LoggingContext.current_context() != start_context: - for err in changes: - print(err, file=sys.stderr) - - err = "%s changed context from %s to %s on exception" % ( - f, - start_context, - LoggingContext.current_context(), - ) - print(err, file=sys.stderr) - raise Exception(err) - raise - - if not isinstance(res, Deferred) or res.called: - if LoggingContext.current_context() != start_context: - for err in changes: - print(err, file=sys.stderr) - - err = "Completed %s changed context from %s to %s" % ( - f, - start_context, - LoggingContext.current_context(), - ) - # print the error to stderr because otherwise all we - # see in travis-ci is the 500 error - print(err, file=sys.stderr) - raise Exception(err) - return res - - if LoggingContext.current_context() != LoggingContext.sentinel: - err = ( - "%s returned incomplete deferred in non-sentinel context " - "%s (start was %s)" - ) % (f, LoggingContext.current_context(), start_context) - print(err, file=sys.stderr) - raise Exception(err) - - def check_ctx(r): - if LoggingContext.current_context() != start_context: - for err in changes: - print(err, file=sys.stderr) - err = "%s completion of %s changed context from %s to %s" % ( - "Failure" if isinstance(r, Failure) else "Success", - f, - start_context, - LoggingContext.current_context(), - ) - print(err, file=sys.stderr) - raise Exception(err) - return r - - res.addBoth(check_ctx) - return res - - return wrapped - - defer.inlineCallbacks = new_inline_callbacks - new_inline_callbacks.patched_by_synapse = True - - -def _check_yield_points(f, changes, start_context): - """Wraps a generator that is about to passed to defer.inlineCallbacks - checking that after every yield the log contexts are correct. - """ - - from synapse.logging.context import LoggingContext - - @functools.wraps(f) - def check_yield_points_inner(*args, **kwargs): - expected_context = start_context - - gen = f(*args, **kwargs) - - last_yield_line_no = 1 - result = None - while True: - try: - isFailure = isinstance(result, Failure) - if isFailure: - d = result.throwExceptionIntoGenerator(gen) - else: - d = gen.send(result) - except (StopIteration, defer._DefGen_Return) as e: - if LoggingContext.current_context() != expected_context: - # This happens when the context is lost sometime *after* the - # final yield and returning. E.g. we forgot to yield on a - # function that returns a deferred. - err = ( - "Function %r returned and changed context from %s to %s," - " in %s between %d and end of func" - % ( - f.__qualname__, - start_context, - LoggingContext.current_context(), - f.__code__.co_filename, - last_yield_line_no, - ) - ) - changes.append(err) - # raise Exception(err) - return getattr(e, "value", None) - - try: - result = yield d - except Exception as e: - result = Failure(e) - - frame = gen.gi_frame - - if LoggingContext.current_context() != expected_context: - # This happens because the context is lost sometime *after* the - # previous yield and *after* the current yield. E.g. the - # deferred we waited on didn't follow the rules, or we forgot to - # yield on a function between the two yield points. - err = ( - "%s changed context from %s to %s, happened between lines %d and %d in %s" - % ( - frame.f_code.co_name, - start_context, - LoggingContext.current_context(), - last_yield_line_no, - frame.f_lineno, - frame.f_code.co_filename, - ) - ) - changes.append(err) - # raise Exception(err) - - expected_context = LoggingContext.current_context() - - last_yield_line_no = frame.f_lineno - - return check_yield_points_inner -- cgit 1.4.1 From a139420a3cfda6a4a4ee4750611b31dd71fc33f3 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Thu, 10 Oct 2019 11:29:01 +0100 Subject: Fix races in room stats (and other) updates. (#6187) Hopefully this will fix the occasional failures we were seeing in the room directory. The problem was that events are not necessarily persisted (and `current_state_delta_stream` updated) in the same order as their stream_id. So for instance current_state_delta 9 might be persisted *before* current_state_delta 8. Then, when the room stats saw stream_id 9, it assumed it had done everything up to 9, and never came back to do stream_id 8. We can solve this easily by only processing up to the stream_id where we know all events have been persisted. --- changelog.d/6187.bugfix | 1 + synapse/handlers/presence.py | 16 ++++++++++++---- synapse/handlers/stats.py | 12 +++++++----- synapse/handlers/user_directory.py | 17 ++++++++++++----- synapse/storage/state_deltas.py | 38 +++++++++++++++++++++++++++++--------- tests/handlers/test_typing.py | 2 +- tests/rest/admin/test_admin.py | 2 +- 7 files changed, 63 insertions(+), 25 deletions(-) create mode 100644 changelog.d/6187.bugfix (limited to 'tests') diff --git a/changelog.d/6187.bugfix b/changelog.d/6187.bugfix new file mode 100644 index 0000000000..6142c5b98d --- /dev/null +++ b/changelog.d/6187.bugfix @@ -0,0 +1 @@ +Fix occasional missed updates in the room and user directories. \ No newline at end of file diff --git a/synapse/handlers/presence.py b/synapse/handlers/presence.py index 053cf66b28..2a5f1a007d 100644 --- a/synapse/handlers/presence.py +++ b/synapse/handlers/presence.py @@ -803,17 +803,25 @@ class PresenceHandler(object): # Loop round handling deltas until we're up to date while True: with Measure(self.clock, "presence_delta"): - deltas = yield self.store.get_current_state_deltas(self._event_pos) - if not deltas: + room_max_stream_ordering = self.store.get_room_max_stream_ordering() + if self._event_pos == room_max_stream_ordering: return + logger.debug( + "Processing presence stats %s->%s", + self._event_pos, + room_max_stream_ordering, + ) + max_pos, deltas = yield self.store.get_current_state_deltas( + self._event_pos, room_max_stream_ordering + ) yield self._handle_state_delta(deltas) - self._event_pos = deltas[-1]["stream_id"] + self._event_pos = max_pos # Expose current event processing position to prometheus synapse.metrics.event_processing_positions.labels("presence").set( - self._event_pos + max_pos ) @defer.inlineCallbacks diff --git a/synapse/handlers/stats.py b/synapse/handlers/stats.py index c62b113115..466daf9202 100644 --- a/synapse/handlers/stats.py +++ b/synapse/handlers/stats.py @@ -87,21 +87,23 @@ class StatsHandler(StateDeltasHandler): # Be sure to read the max stream_ordering *before* checking if there are any outstanding # deltas, since there is otherwise a chance that we could miss updates which arrive # after we check the deltas. - room_max_stream_ordering = yield self.store.get_room_max_stream_ordering() + room_max_stream_ordering = self.store.get_room_max_stream_ordering() if self.pos == room_max_stream_ordering: break - deltas = yield self.store.get_current_state_deltas(self.pos) + logger.debug( + "Processing room stats %s->%s", self.pos, room_max_stream_ordering + ) + max_pos, deltas = yield self.store.get_current_state_deltas( + self.pos, room_max_stream_ordering + ) if deltas: logger.debug("Handling %d state deltas", len(deltas)) room_deltas, user_deltas = yield self._handle_deltas(deltas) - - max_pos = deltas[-1]["stream_id"] else: room_deltas = {} user_deltas = {} - max_pos = room_max_stream_ordering # Then count deltas for total_events and total_event_bytes. room_count, user_count = yield self.store.get_changes_room_total_events_and_bytes( diff --git a/synapse/handlers/user_directory.py b/synapse/handlers/user_directory.py index e53669e40d..624f05ab5b 100644 --- a/synapse/handlers/user_directory.py +++ b/synapse/handlers/user_directory.py @@ -138,21 +138,28 @@ class UserDirectoryHandler(StateDeltasHandler): # Loop round handling deltas until we're up to date while True: with Measure(self.clock, "user_dir_delta"): - deltas = yield self.store.get_current_state_deltas(self.pos) - if not deltas: + room_max_stream_ordering = self.store.get_room_max_stream_ordering() + if self.pos == room_max_stream_ordering: return + logger.debug( + "Processing user stats %s->%s", self.pos, room_max_stream_ordering + ) + max_pos, deltas = yield self.store.get_current_state_deltas( + self.pos, room_max_stream_ordering + ) + logger.info("Handling %d state deltas", len(deltas)) yield self._handle_deltas(deltas) - self.pos = deltas[-1]["stream_id"] + self.pos = max_pos # Expose current event processing position to prometheus synapse.metrics.event_processing_positions.labels("user_dir").set( - self.pos + max_pos ) - yield self.store.update_user_directory_stream_pos(self.pos) + yield self.store.update_user_directory_stream_pos(max_pos) @defer.inlineCallbacks def _handle_deltas(self, deltas): diff --git a/synapse/storage/state_deltas.py b/synapse/storage/state_deltas.py index 5fdb442104..28f33ec18f 100644 --- a/synapse/storage/state_deltas.py +++ b/synapse/storage/state_deltas.py @@ -21,7 +21,7 @@ logger = logging.getLogger(__name__) class StateDeltasStore(SQLBaseStore): - def get_current_state_deltas(self, prev_stream_id): + def get_current_state_deltas(self, prev_stream_id: int, max_stream_id: int): """Fetch a list of room state changes since the given stream id Each entry in the result contains the following fields: @@ -36,15 +36,27 @@ class StateDeltasStore(SQLBaseStore): Args: prev_stream_id (int): point to get changes since (exclusive) + max_stream_id (int): the point that we know has been correctly persisted + - ie, an upper limit to return changes from. Returns: - Deferred[list[dict]]: results + Deferred[tuple[int, list[dict]]: A tuple consisting of: + - the stream id which these results go up to + - list of current_state_delta_stream rows. If it is empty, we are + up to date. """ prev_stream_id = int(prev_stream_id) + + # check we're not going backwards + assert prev_stream_id <= max_stream_id + if not self._curr_state_delta_stream_cache.has_any_entity_changed( prev_stream_id ): - return [] + # if the CSDs haven't changed between prev_stream_id and now, we + # know for certain that they haven't changed between prev_stream_id and + # max_stream_id. + return max_stream_id, [] def get_current_state_deltas_txn(txn): # First we calculate the max stream id that will give us less than @@ -54,21 +66,29 @@ class StateDeltasStore(SQLBaseStore): sql = """ SELECT stream_id, count(*) FROM current_state_delta_stream - WHERE stream_id > ? + WHERE stream_id > ? AND stream_id <= ? GROUP BY stream_id ORDER BY stream_id ASC LIMIT 100 """ - txn.execute(sql, (prev_stream_id,)) + txn.execute(sql, (prev_stream_id, max_stream_id)) total = 0 - max_stream_id = prev_stream_id - for max_stream_id, count in txn: + + for stream_id, count in txn: total += count if total > 100: # We arbitarily limit to 100 entries to ensure we don't # select toooo many. + logger.debug( + "Clipping current_state_delta_stream rows to stream_id %i", + stream_id, + ) + clipped_stream_id = stream_id break + else: + # if there's no problem, we may as well go right up to the max_stream_id + clipped_stream_id = max_stream_id # Now actually get the deltas sql = """ @@ -77,8 +97,8 @@ class StateDeltasStore(SQLBaseStore): WHERE ? < stream_id AND stream_id <= ? ORDER BY stream_id ASC """ - txn.execute(sql, (prev_stream_id, max_stream_id)) - return self.cursor_to_dict(txn) + txn.execute(sql, (prev_stream_id, clipped_stream_id)) + return clipped_stream_id, self.cursor_to_dict(txn) return self.runInteraction( "get_current_state_deltas", get_current_state_deltas_txn diff --git a/tests/handlers/test_typing.py b/tests/handlers/test_typing.py index 1f2ef5d01f..67f1013051 100644 --- a/tests/handlers/test_typing.py +++ b/tests/handlers/test_typing.py @@ -139,7 +139,7 @@ class TypingNotificationsTestCase(unittest.HomeserverTestCase): defer.succeed(1) ) - self.datastore.get_current_state_deltas.return_value = None + self.datastore.get_current_state_deltas.return_value = (0, None) self.datastore.get_to_device_stream_token = lambda: 0 self.datastore.get_new_device_msgs_for_remote = lambda *args, **kargs: ([], 0) diff --git a/tests/rest/admin/test_admin.py b/tests/rest/admin/test_admin.py index 5877bb2133..d3a4f717f7 100644 --- a/tests/rest/admin/test_admin.py +++ b/tests/rest/admin/test_admin.py @@ -62,7 +62,7 @@ class UserRegisterTestCase(unittest.HomeserverTestCase): self.device_handler.check_device_registered = Mock(return_value="FAKE") self.datastore = Mock(return_value=Mock()) - self.datastore.get_current_state_deltas = Mock(return_value=[]) + self.datastore.get_current_state_deltas = Mock(return_value=(0, [])) self.secrets = Mock() -- cgit 1.4.1 From 2efd050c9db2e96fd96535dc9b1c6f54acbd163d Mon Sep 17 00:00:00 2001 From: krombel Date: Thu, 10 Oct 2019 13:59:55 +0200 Subject: send 404 as http-status when filter-id is unknown to the server (#2380) This fixed the weirdness of 400 vs 404 as http status code in the case the filter id is not known by the server. As e.g. matrix-js-sdk expects 404 to catch this situation this leads to unwanted behaviour. --- changelog.d/2380.bugfix | 1 + synapse/rest/client/v2_alpha/filter.py | 12 +++++---- synapse/rest/client/v2_alpha/sync.py | 41 ++++++++++++++++++------------- tests/rest/client/v2_alpha/test_filter.py | 2 +- 4 files changed, 33 insertions(+), 23 deletions(-) create mode 100644 changelog.d/2380.bugfix (limited to 'tests') diff --git a/changelog.d/2380.bugfix b/changelog.d/2380.bugfix new file mode 100644 index 0000000000..eae3206031 --- /dev/null +++ b/changelog.d/2380.bugfix @@ -0,0 +1 @@ +Return an HTTP 404 instead of 400 when requesting a filter by ID that is unknown to the server. Thanks to @krombel for contributing this! diff --git a/synapse/rest/client/v2_alpha/filter.py b/synapse/rest/client/v2_alpha/filter.py index c6ddf24c8d..17a8bc7366 100644 --- a/synapse/rest/client/v2_alpha/filter.py +++ b/synapse/rest/client/v2_alpha/filter.py @@ -17,7 +17,7 @@ import logging from twisted.internet import defer -from synapse.api.errors import AuthError, Codes, StoreError, SynapseError +from synapse.api.errors import AuthError, NotFoundError, StoreError, SynapseError from synapse.http.servlet import RestServlet, parse_json_object_from_request from synapse.types import UserID @@ -52,13 +52,15 @@ class GetFilterRestServlet(RestServlet): raise SynapseError(400, "Invalid filter_id") try: - filter = yield self.filtering.get_user_filter( + filter_collection = yield self.filtering.get_user_filter( user_localpart=target_user.localpart, filter_id=filter_id ) + except StoreError as e: + if e.code != 404: + raise + raise NotFoundError("No such filter") - return 200, filter.get_filter_json() - except (KeyError, StoreError): - raise SynapseError(400, "No such filter", errcode=Codes.NOT_FOUND) + return 200, filter_collection.get_filter_json() class CreateFilterRestServlet(RestServlet): diff --git a/synapse/rest/client/v2_alpha/sync.py b/synapse/rest/client/v2_alpha/sync.py index c98c5a3802..a883c8adda 100644 --- a/synapse/rest/client/v2_alpha/sync.py +++ b/synapse/rest/client/v2_alpha/sync.py @@ -21,7 +21,7 @@ from canonicaljson import json from twisted.internet import defer from synapse.api.constants import PresenceState -from synapse.api.errors import SynapseError +from synapse.api.errors import Codes, StoreError, SynapseError from synapse.api.filtering import DEFAULT_FILTER_COLLECTION, FilterCollection from synapse.events.utils import ( format_event_for_client_v2_without_room_id, @@ -119,25 +119,32 @@ class SyncRestServlet(RestServlet): request_key = (user, timeout, since, filter_id, full_state, device_id) - if filter_id: - if filter_id.startswith("{"): - try: - filter_object = json.loads(filter_id) - set_timeline_upper_limit( - filter_object, self.hs.config.filter_timeline_limit - ) - except Exception: - raise SynapseError(400, "Invalid filter JSON") - self.filtering.check_valid_filter(filter_object) - filter = FilterCollection(filter_object) - else: - filter = yield self.filtering.get_user_filter(user.localpart, filter_id) + if filter_id is None: + filter_collection = DEFAULT_FILTER_COLLECTION + elif filter_id.startswith("{"): + try: + filter_object = json.loads(filter_id) + set_timeline_upper_limit( + filter_object, self.hs.config.filter_timeline_limit + ) + except Exception: + raise SynapseError(400, "Invalid filter JSON") + self.filtering.check_valid_filter(filter_object) + filter_collection = FilterCollection(filter_object) else: - filter = DEFAULT_FILTER_COLLECTION + try: + filter_collection = yield self.filtering.get_user_filter( + user.localpart, filter_id + ) + except StoreError as err: + if err.code != 404: + raise + # fix up the description and errcode to be more useful + raise SynapseError(400, "No such filter", errcode=Codes.INVALID_PARAM) sync_config = SyncConfig( user=user, - filter_collection=filter, + filter_collection=filter_collection, is_guest=requester.is_guest, request_key=request_key, device_id=device_id, @@ -171,7 +178,7 @@ class SyncRestServlet(RestServlet): time_now = self.clock.time_msec() response_content = yield self.encode_response( - time_now, sync_result, requester.access_token_id, filter + time_now, sync_result, requester.access_token_id, filter_collection ) return 200, response_content diff --git a/tests/rest/client/v2_alpha/test_filter.py b/tests/rest/client/v2_alpha/test_filter.py index f42a8efbf4..e0e9e94fbf 100644 --- a/tests/rest/client/v2_alpha/test_filter.py +++ b/tests/rest/client/v2_alpha/test_filter.py @@ -92,7 +92,7 @@ class FilterTestCase(unittest.HomeserverTestCase): ) self.render(request) - self.assertEqual(channel.result["code"], b"400") + self.assertEqual(channel.result["code"], b"404") self.assertEquals(channel.json_body["errcode"], Codes.NOT_FOUND) # Currently invalid params do not have an appropriate errcode -- cgit 1.4.1 From b5b03b7079a9baa34a25915d6a569e383e8307c3 Mon Sep 17 00:00:00 2001 From: werner291 Date: Thu, 10 Oct 2019 14:05:48 +0200 Subject: Add domain validation when creating room with list of invitees (#6121) --- changelog.d/4088.bugfix | 1 + synapse/handlers/room.py | 4 +++- tests/rest/client/v1/test_rooms.py | 9 +++++++++ 3 files changed, 13 insertions(+), 1 deletion(-) create mode 100644 changelog.d/4088.bugfix (limited to 'tests') diff --git a/changelog.d/4088.bugfix b/changelog.d/4088.bugfix new file mode 100644 index 0000000000..61722b6224 --- /dev/null +++ b/changelog.d/4088.bugfix @@ -0,0 +1 @@ +Added domain validation when including a list of invitees upon room creation. \ No newline at end of file diff --git a/synapse/handlers/room.py b/synapse/handlers/room.py index 970be3c846..2816bd8f87 100644 --- a/synapse/handlers/room.py +++ b/synapse/handlers/room.py @@ -28,6 +28,7 @@ from twisted.internet import defer from synapse.api.constants import EventTypes, JoinRules, RoomCreationPreset from synapse.api.errors import AuthError, Codes, NotFoundError, StoreError, SynapseError from synapse.api.room_versions import KNOWN_ROOM_VERSIONS +from synapse.http.endpoint import parse_and_validate_server_name from synapse.storage.state import StateFilter from synapse.types import RoomAlias, RoomID, RoomStreamToken, StreamToken, UserID from synapse.util import stringutils @@ -554,7 +555,8 @@ class RoomCreationHandler(BaseHandler): invite_list = config.get("invite", []) for i in invite_list: try: - UserID.from_string(i) + uid = UserID.from_string(i) + parse_and_validate_server_name(uid.domain) except Exception: raise SynapseError(400, "Invalid user_id: %s" % (i,)) diff --git a/tests/rest/client/v1/test_rooms.py b/tests/rest/client/v1/test_rooms.py index fe741637f5..2f2ca74611 100644 --- a/tests/rest/client/v1/test_rooms.py +++ b/tests/rest/client/v1/test_rooms.py @@ -484,6 +484,15 @@ class RoomsCreateTestCase(RoomBase): self.render(request) self.assertEquals(400, channel.code) + def test_post_room_invitees_invalid_mxid(self): + # POST with invalid invitee, see https://github.com/matrix-org/synapse/issues/4088 + # Note the trailing space in the MXID here! + request, channel = self.make_request( + "POST", "/createRoom", b'{"invite":["@alice:example.com "]}' + ) + self.render(request) + self.assertEquals(400, channel.code) + class RoomTopicTestCase(RoomBase): """ Tests /rooms/$room_id/topic REST events. """ -- cgit 1.4.1 From 5373de6cced56c983098c82872cf17c311abdb96 Mon Sep 17 00:00:00 2001 From: Hubert Chathi Date: Thu, 10 Oct 2019 08:54:07 -0400 Subject: change test name to be unique --- tests/handlers/test_e2e_room_keys.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tests') diff --git a/tests/handlers/test_e2e_room_keys.py b/tests/handlers/test_e2e_room_keys.py index c700a2fad1..0bb96674a2 100644 --- a/tests/handlers/test_e2e_room_keys.py +++ b/tests/handlers/test_e2e_room_keys.py @@ -187,7 +187,7 @@ class E2eRoomKeysHandlerTestCase(unittest.TestCase): self.assertEqual(res, 404) @defer.inlineCallbacks - def test_update_missing_version(self): + def test_update_omitted_version(self): """Check that the update succeeds if the version is missing from the body """ version = yield self.handler.create_version( -- cgit 1.4.1 From bc244627ac759bbd4691a2a20ac16383b4c2348c Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Thu, 10 Oct 2019 15:37:53 +0100 Subject: Fix postgres unit tests --- tests/storage/test_event_federation.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tests') diff --git a/tests/storage/test_event_federation.py b/tests/storage/test_event_federation.py index b58386994e..2fe50377f8 100644 --- a/tests/storage/test_event_federation.py +++ b/tests/storage/test_event_federation.py @@ -57,7 +57,7 @@ class EventFederationWorkerStoreTestCase(tests.unittest.TestCase): "(event_id, algorithm, hash) " "VALUES (?, 'sha256', ?)" ), - (event_id, b"ffff"), + (event_id, bytearray(b"ffff")), ) for i in range(0, 11): -- cgit 1.4.1 From 4908fb3b30ac007fda5993521448804067751a6d Mon Sep 17 00:00:00 2001 From: Hubert Chathi Date: Thu, 10 Oct 2019 15:56:00 -0400 Subject: make storage layer in charge of interpreting the device key data --- synapse/handlers/e2e_keys.py | 11 ----------- synapse/storage/end_to_end_keys.py | 11 +++++++++-- tests/storage/test_end_to_end_keys.py | 12 ++++++------ 3 files changed, 15 insertions(+), 19 deletions(-) (limited to 'tests') diff --git a/synapse/handlers/e2e_keys.py b/synapse/handlers/e2e_keys.py index 056fb97acb..6708d983ac 100644 --- a/synapse/handlers/e2e_keys.py +++ b/synapse/handlers/e2e_keys.py @@ -248,17 +248,6 @@ class E2eKeysHandler(object): results = yield self.store.get_e2e_device_keys(local_query) - # Build the result structure, un-jsonify the results, and add the - # "unsigned" section - for user_id, device_keys in results.items(): - for device_id, device_info in device_keys.items(): - r = dict(device_info["keys"]) - r["unsigned"] = {} - display_name = device_info["device_display_name"] - if display_name is not None: - r["unsigned"]["device_display_name"] = display_name - result_dict[user_id][device_id] = r - log_kv(results) return result_dict diff --git a/synapse/storage/end_to_end_keys.py b/synapse/storage/end_to_end_keys.py index 33e3a84933..d802d7a485 100644 --- a/synapse/storage/end_to_end_keys.py +++ b/synapse/storage/end_to_end_keys.py @@ -40,7 +40,7 @@ class EndToEndKeyWorkerStore(SQLBaseStore): This option only takes effect if include_all_devices is true. Returns: Dict mapping from user-id to dict mapping from device_id to - dict containing "key_json", "device_display_name". + key data. """ set_tag("query_list", query_list) if not query_list: @@ -54,9 +54,16 @@ class EndToEndKeyWorkerStore(SQLBaseStore): include_deleted_devices, ) + # Build the result structure, un-jsonify the results, and add the + # "unsigned" section for user_id, device_keys in iteritems(results): for device_id, device_info in iteritems(device_keys): - device_info["keys"] = db_to_json(device_info.pop("key_json")) + r = db_to_json(device_info.pop("key_json")) + r["unsigned"] = {} + display_name = device_info["device_display_name"] + if display_name is not None: + r["unsigned"]["device_display_name"] = display_name + results[user_id][device_id] = r return results diff --git a/tests/storage/test_end_to_end_keys.py b/tests/storage/test_end_to_end_keys.py index c8ece15284..398d546280 100644 --- a/tests/storage/test_end_to_end_keys.py +++ b/tests/storage/test_end_to_end_keys.py @@ -38,7 +38,7 @@ class EndToEndKeyStoreTestCase(tests.unittest.TestCase): self.assertIn("user", res) self.assertIn("device", res["user"]) dev = res["user"]["device"] - self.assertDictContainsSubset({"keys": json, "device_display_name": None}, dev) + self.assertDictContainsSubset(json, dev) @defer.inlineCallbacks def test_reupload_key(self): @@ -68,7 +68,7 @@ class EndToEndKeyStoreTestCase(tests.unittest.TestCase): self.assertIn("device", res["user"]) dev = res["user"]["device"] self.assertDictContainsSubset( - {"keys": json, "device_display_name": "display_name"}, dev + {"key": "value", "unsigned": {"device_display_name": "display_name"}}, dev ) @defer.inlineCallbacks @@ -80,10 +80,10 @@ class EndToEndKeyStoreTestCase(tests.unittest.TestCase): yield self.store.store_device("user2", "device1", None) yield self.store.store_device("user2", "device2", None) - yield self.store.set_e2e_device_keys("user1", "device1", now, "json11") - yield self.store.set_e2e_device_keys("user1", "device2", now, "json12") - yield self.store.set_e2e_device_keys("user2", "device1", now, "json21") - yield self.store.set_e2e_device_keys("user2", "device2", now, "json22") + yield self.store.set_e2e_device_keys("user1", "device1", now, {"key": "json11"}) + yield self.store.set_e2e_device_keys("user1", "device2", now, {"key": "json12"}) + yield self.store.set_e2e_device_keys("user2", "device1", now, {"key": "json21"}) + yield self.store.set_e2e_device_keys("user2", "device2", now, {"key": "json22"}) res = yield self.store.get_e2e_device_keys( (("user1", "device1"), ("user2", "device2")) -- cgit 1.4.1 From a0d0ba7862e38588aa0d0ac29a720fdf06f1ab8d Mon Sep 17 00:00:00 2001 From: Neil Johnson Date: Fri, 11 Oct 2019 09:38:26 +0100 Subject: Fix MAU reaping where reserved users are specified. (#6168) --- changelog.d/6168.bugfix | 1 + synapse/app/homeserver.py | 6 +- synapse/storage/monthly_active_users.py | 101 ++++++++++++++++++----------- tests/storage/test_monthly_active_users.py | 58 ++++++++++++++--- 4 files changed, 115 insertions(+), 51 deletions(-) create mode 100644 changelog.d/6168.bugfix (limited to 'tests') diff --git a/changelog.d/6168.bugfix b/changelog.d/6168.bugfix new file mode 100644 index 0000000000..39e8e9d019 --- /dev/null +++ b/changelog.d/6168.bugfix @@ -0,0 +1 @@ +Fix monthly active user reaping where reserved users are specified. diff --git a/synapse/app/homeserver.py b/synapse/app/homeserver.py index 774326dff9..eb54f56853 100644 --- a/synapse/app/homeserver.py +++ b/synapse/app/homeserver.py @@ -605,13 +605,13 @@ def run(hs): @defer.inlineCallbacks def generate_monthly_active_users(): current_mau_count = 0 - reserved_count = 0 + reserved_users = () store = hs.get_datastore() if hs.config.limit_usage_by_mau or hs.config.mau_stats_only: current_mau_count = yield store.get_monthly_active_count() - reserved_count = yield store.get_registered_reserved_users_count() + reserved_users = yield store.get_registered_reserved_users() current_mau_gauge.set(float(current_mau_count)) - registered_reserved_users_mau_gauge.set(float(reserved_count)) + registered_reserved_users_mau_gauge.set(float(len(reserved_users))) max_mau_gauge.set(float(hs.config.max_mau_value)) def start_generate_monthly_active_users(): diff --git a/synapse/storage/monthly_active_users.py b/synapse/storage/monthly_active_users.py index 752e9788a2..3803604be7 100644 --- a/synapse/storage/monthly_active_users.py +++ b/synapse/storage/monthly_active_users.py @@ -32,7 +32,6 @@ class MonthlyActiveUsersStore(SQLBaseStore): super(MonthlyActiveUsersStore, self).__init__(None, hs) self._clock = hs.get_clock() self.hs = hs - self.reserved_users = () # Do not add more reserved users than the total allowable number self._new_transaction( dbconn, @@ -51,7 +50,6 @@ class MonthlyActiveUsersStore(SQLBaseStore): txn (cursor): threepids (list[dict]): List of threepid dicts to reserve """ - reserved_user_list = [] for tp in threepids: user_id = self.get_user_id_by_threepid_txn(txn, tp["medium"], tp["address"]) @@ -60,10 +58,8 @@ class MonthlyActiveUsersStore(SQLBaseStore): is_support = self.is_support_user_txn(txn, user_id) if not is_support: self.upsert_monthly_active_user_txn(txn, user_id) - reserved_user_list.append(user_id) else: logger.warning("mau limit reserved threepid %s not found in db" % tp) - self.reserved_users = tuple(reserved_user_list) @defer.inlineCallbacks def reap_monthly_active_users(self): @@ -74,8 +70,11 @@ class MonthlyActiveUsersStore(SQLBaseStore): Deferred[] """ - def _reap_users(txn): - # Purge stale users + def _reap_users(txn, reserved_users): + """ + Args: + reserved_users (tuple): reserved users to preserve + """ thirty_days_ago = int(self._clock.time_msec()) - (1000 * 60 * 60 * 24 * 30) query_args = [thirty_days_ago] @@ -83,20 +82,19 @@ class MonthlyActiveUsersStore(SQLBaseStore): # Need if/else since 'AND user_id NOT IN ({})' fails on Postgres # when len(reserved_users) == 0. Works fine on sqlite. - if len(self.reserved_users) > 0: + if len(reserved_users) > 0: # questionmarks is a hack to overcome sqlite not supporting # tuples in 'WHERE IN %s' - questionmarks = "?" * len(self.reserved_users) + question_marks = ",".join("?" * len(reserved_users)) - query_args.extend(self.reserved_users) - sql = base_sql + """ AND user_id NOT IN ({})""".format( - ",".join(questionmarks) - ) + query_args.extend(reserved_users) + sql = base_sql + " AND user_id NOT IN ({})".format(question_marks) else: sql = base_sql txn.execute(sql, query_args) + max_mau_value = self.hs.config.max_mau_value if self.hs.config.limit_usage_by_mau: # If MAU user count still exceeds the MAU threshold, then delete on # a least recently active basis. @@ -106,31 +104,52 @@ class MonthlyActiveUsersStore(SQLBaseStore): # While Postgres does not require 'LIMIT', but also does not support # negative LIMIT values. So there is no way to write it that both can # support - safe_guard = self.hs.config.max_mau_value - len(self.reserved_users) - # Must be greater than zero for postgres - safe_guard = safe_guard if safe_guard > 0 else 0 - query_args = [safe_guard] - - base_sql = """ - DELETE FROM monthly_active_users - WHERE user_id NOT IN ( - SELECT user_id FROM monthly_active_users - ORDER BY timestamp DESC - LIMIT ? + if len(reserved_users) == 0: + sql = """ + DELETE FROM monthly_active_users + WHERE user_id NOT IN ( + SELECT user_id FROM monthly_active_users + ORDER BY timestamp DESC + LIMIT ? ) - """ + """ + txn.execute(sql, (max_mau_value,)) # Need if/else since 'AND user_id NOT IN ({})' fails on Postgres # when len(reserved_users) == 0. Works fine on sqlite. - if len(self.reserved_users) > 0: - query_args.extend(self.reserved_users) - sql = base_sql + """ AND user_id NOT IN ({})""".format( - ",".join(questionmarks) - ) else: - sql = base_sql - txn.execute(sql, query_args) + # Must be >= 0 for postgres + num_of_non_reserved_users_to_remove = max( + max_mau_value - len(reserved_users), 0 + ) + + # It is important to filter reserved users twice to guard + # against the case where the reserved user is present in the + # SELECT, meaning that a legitmate mau is deleted. + sql = """ + DELETE FROM monthly_active_users + WHERE user_id NOT IN ( + SELECT user_id FROM monthly_active_users + WHERE user_id NOT IN ({}) + ORDER BY timestamp DESC + LIMIT ? + ) + AND user_id NOT IN ({}) + """.format( + question_marks, question_marks + ) + + query_args = [ + *reserved_users, + num_of_non_reserved_users_to_remove, + *reserved_users, + ] - yield self.runInteraction("reap_monthly_active_users", _reap_users) + txn.execute(sql, query_args) + + reserved_users = yield self.get_registered_reserved_users() + yield self.runInteraction( + "reap_monthly_active_users", _reap_users, reserved_users + ) # It seems poor to invalidate the whole cache, Postgres supports # 'Returning' which would allow me to invalidate only the # specific users, but sqlite has no way to do this and instead @@ -159,21 +178,25 @@ class MonthlyActiveUsersStore(SQLBaseStore): return self.runInteraction("count_users", _count_users) @defer.inlineCallbacks - def get_registered_reserved_users_count(self): - """Of the reserved threepids defined in config, how many are associated + def get_registered_reserved_users(self): + """Of the reserved threepids defined in config, which are associated with registered users? Returns: - Defered[int]: Number of real reserved users + Defered[list]: Real reserved users """ - count = 0 - for tp in self.hs.config.mau_limits_reserved_threepids: + users = [] + + for tp in self.hs.config.mau_limits_reserved_threepids[ + : self.hs.config.max_mau_value + ]: user_id = yield self.hs.get_datastore().get_user_id_by_threepid( tp["medium"], tp["address"] ) if user_id: - count = count + 1 - return count + users.append(user_id) + + return users @defer.inlineCallbacks def upsert_monthly_active_user(self, user_id): diff --git a/tests/storage/test_monthly_active_users.py b/tests/storage/test_monthly_active_users.py index 1494650d10..90a63dc477 100644 --- a/tests/storage/test_monthly_active_users.py +++ b/tests/storage/test_monthly_active_users.py @@ -50,6 +50,7 @@ class MonthlyActiveUsersTestCase(unittest.HomeserverTestCase): {"medium": "email", "address": user2_email}, {"medium": "email", "address": user3_email}, ] + self.hs.config.mau_limits_reserved_threepids = threepids # -1 because user3 is a support user and does not count user_num = len(threepids) - 1 @@ -84,6 +85,7 @@ class MonthlyActiveUsersTestCase(unittest.HomeserverTestCase): self.hs.config.max_mau_value = 0 self.reactor.advance(FORTY_DAYS) + self.hs.config.max_mau_value = 5 self.store.reap_monthly_active_users() self.pump() @@ -147,9 +149,7 @@ class MonthlyActiveUsersTestCase(unittest.HomeserverTestCase): self.store.reap_monthly_active_users() self.pump() count = self.store.get_monthly_active_count() - self.assertEquals( - self.get_success(count), initial_users - self.hs.config.max_mau_value - ) + self.assertEquals(self.get_success(count), self.hs.config.max_mau_value) self.reactor.advance(FORTY_DAYS) self.store.reap_monthly_active_users() @@ -158,6 +158,44 @@ class MonthlyActiveUsersTestCase(unittest.HomeserverTestCase): count = self.store.get_monthly_active_count() self.assertEquals(self.get_success(count), 0) + def test_reap_monthly_active_users_reserved_users(self): + """ Tests that reaping correctly handles reaping where reserved users are + present""" + + self.hs.config.max_mau_value = 5 + initial_users = 5 + reserved_user_number = initial_users - 1 + threepids = [] + for i in range(initial_users): + user = "@user%d:server" % i + email = "user%d@example.com" % i + self.get_success(self.store.upsert_monthly_active_user(user)) + threepids.append({"medium": "email", "address": email}) + # Need to ensure that the most recent entries in the + # monthly_active_users table are reserved + now = int(self.hs.get_clock().time_msec()) + if i != 0: + self.get_success( + self.store.register_user(user_id=user, password_hash=None) + ) + self.get_success( + self.store.user_add_threepid(user, "email", email, now, now) + ) + + self.hs.config.mau_limits_reserved_threepids = threepids + self.store.runInteraction( + "initialise", self.store._initialise_reserved_users, threepids + ) + count = self.store.get_monthly_active_count() + self.assertTrue(self.get_success(count), initial_users) + + users = self.store.get_registered_reserved_users() + self.assertEquals(len(self.get_success(users)), reserved_user_number) + + self.get_success(self.store.reap_monthly_active_users()) + count = self.store.get_monthly_active_count() + self.assertEquals(self.get_success(count), self.hs.config.max_mau_value) + def test_populate_monthly_users_is_guest(self): # Test that guest users are not added to mau list user_id = "@user_id:host" @@ -192,12 +230,13 @@ class MonthlyActiveUsersTestCase(unittest.HomeserverTestCase): def test_get_reserved_real_user_account(self): # Test no reserved users, or reserved threepids - count = self.store.get_registered_reserved_users_count() - self.assertEquals(self.get_success(count), 0) + users = self.get_success(self.store.get_registered_reserved_users()) + self.assertEquals(len(users), 0) # Test reserved users but no registered users user1 = "@user1:example.com" user2 = "@user2:example.com" + user1_email = "user1@example.com" user2_email = "user2@example.com" threepids = [ @@ -210,8 +249,8 @@ class MonthlyActiveUsersTestCase(unittest.HomeserverTestCase): ) self.pump() - count = self.store.get_registered_reserved_users_count() - self.assertEquals(self.get_success(count), 0) + users = self.get_success(self.store.get_registered_reserved_users()) + self.assertEquals(len(users), 0) # Test reserved registed users self.store.register_user(user_id=user1, password_hash=None) @@ -221,8 +260,9 @@ class MonthlyActiveUsersTestCase(unittest.HomeserverTestCase): now = int(self.hs.get_clock().time_msec()) self.store.user_add_threepid(user1, "email", user1_email, now, now) self.store.user_add_threepid(user2, "email", user2_email, now, now) - count = self.store.get_registered_reserved_users_count() - self.assertEquals(self.get_success(count), len(threepids)) + + users = self.get_success(self.store.get_registered_reserved_users()) + self.assertEquals(len(users), len(threepids)) def test_support_user_not_add_to_mau_limits(self): support_user_id = "@support:test" -- cgit 1.4.1 From 5859a5c569c03f3b7c578fe4dbf2274e37af03bb Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Fri, 18 Oct 2019 07:42:26 +0200 Subject: Fix presence timeouts when synchrotron restarts. (#6212) * Fix presence timeouts when synchrotron restarts. Handling timeouts would fail if there was an external process that had timed out, e.g. a synchrotron restarting. This was due to a couple of variable name typoes. Fixes #3715. --- changelog.d/6212.bugfix | 1 + synapse/handlers/presence.py | 13 +++++++++---- tests/handlers/test_presence.py | 39 +++++++++++++++++++++++++++++++++++++++ tox.ini | 2 +- 4 files changed, 50 insertions(+), 5 deletions(-) create mode 100644 changelog.d/6212.bugfix (limited to 'tests') diff --git a/changelog.d/6212.bugfix b/changelog.d/6212.bugfix new file mode 100644 index 0000000000..918755fee0 --- /dev/null +++ b/changelog.d/6212.bugfix @@ -0,0 +1 @@ +Fix bug where presence would not get timed out correctly if a synchrotron worker is used and restarted. diff --git a/synapse/handlers/presence.py b/synapse/handlers/presence.py index 2a5f1a007d..eda15bc623 100644 --- a/synapse/handlers/presence.py +++ b/synapse/handlers/presence.py @@ -24,6 +24,7 @@ The methods that define policy are: import logging from contextlib import contextmanager +from typing import Dict, Set from six import iteritems, itervalues @@ -179,8 +180,9 @@ class PresenceHandler(object): # we assume that all the sync requests on that process have stopped. # Stored as a dict from process_id to set of user_id, and a dict of # process_id to millisecond timestamp last updated. - self.external_process_to_current_syncs = {} - self.external_process_last_updated_ms = {} + self.external_process_to_current_syncs = {} # type: Dict[int, Set[str]] + self.external_process_last_updated_ms = {} # type: Dict[int, int] + self.external_sync_linearizer = Linearizer(name="external_sync_linearizer") # Start a LoopingCall in 30s that fires every 5s. @@ -349,10 +351,13 @@ class PresenceHandler(object): if now - last_update > EXTERNAL_PROCESS_EXPIRY ] for process_id in expired_process_ids: + # For each expired process drop tracking info and check the users + # that were syncing on that process to see if they need to be timed + # out. users_to_check.update( - self.external_process_last_updated_ms.pop(process_id, ()) + self.external_process_to_current_syncs.pop(process_id, ()) ) - self.external_process_last_update.pop(process_id) + self.external_process_last_updated_ms.pop(process_id) states = [ self.user_to_current_state.get(user_id, UserPresenceState.default(user_id)) diff --git a/tests/handlers/test_presence.py b/tests/handlers/test_presence.py index f70c6e7d65..d4293b4312 100644 --- a/tests/handlers/test_presence.py +++ b/tests/handlers/test_presence.py @@ -22,6 +22,7 @@ from synapse.api.constants import EventTypes, Membership, PresenceState from synapse.events import room_version_to_event_format from synapse.events.builder import EventBuilder from synapse.handlers.presence import ( + EXTERNAL_PROCESS_EXPIRY, FEDERATION_PING_INTERVAL, FEDERATION_TIMEOUT, IDLE_TIMER, @@ -413,6 +414,44 @@ class PresenceTimeoutTestCase(unittest.TestCase): self.assertEquals(state, new_state) +class PresenceHandlerTestCase(unittest.HomeserverTestCase): + def prepare(self, reactor, clock, hs): + self.presence_handler = hs.get_presence_handler() + self.clock = hs.get_clock() + + def test_external_process_timeout(self): + """Test that if an external process doesn't update the records for a while + we time out their syncing users presence. + """ + process_id = 1 + user_id = "@test:server" + + # Notify handler that a user is now syncing. + self.get_success( + self.presence_handler.update_external_syncs_row( + process_id, user_id, True, self.clock.time_msec() + ) + ) + + # Check that if we wait a while without telling the handler the user has + # stopped syncing that their presence state doesn't get timed out. + self.reactor.advance(EXTERNAL_PROCESS_EXPIRY / 2) + + state = self.get_success( + self.presence_handler.get_state(UserID.from_string(user_id)) + ) + self.assertEqual(state.state, PresenceState.ONLINE) + + # Check that if the external process timeout fires, then the syncing + # user gets timed out + self.reactor.advance(EXTERNAL_PROCESS_EXPIRY) + + state = self.get_success( + self.presence_handler.get_state(UserID.from_string(user_id)) + ) + self.assertEqual(state.state, PresenceState.OFFLINE) + + class PresenceJoinTestCase(unittest.HomeserverTestCase): """Tests remote servers get told about presence of users in the room when they join and when new local users join. diff --git a/tox.ini b/tox.ini index 367cc2ccf2..7ba6f6339f 100644 --- a/tox.ini +++ b/tox.ini @@ -161,7 +161,7 @@ basepython = python3.7 skip_install = True deps = {[base]deps} - mypy + mypy==0.730 mypy-zope env = MYPYPATH = stubs/ -- cgit 1.4.1 From c66a06ac6b69b0a03f5c6284ded980399e9df94e Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Mon, 21 Oct 2019 12:56:42 +0100 Subject: Move storage classes into a main "data store". This is in preparation for having multiple data stores that offer different functionality, e.g. splitting out state or event storage. --- synapse/app/event_creator.py | 2 +- synapse/app/media_repository.py | 2 +- synapse/app/synchrotron.py | 2 +- synapse/app/user_dir.py | 2 +- synapse/federation/sender/per_destination_queue.py | 2 +- synapse/replication/slave/storage/account_data.py | 4 +- synapse/replication/slave/storage/appservice.py | 2 +- synapse/replication/slave/storage/client_ips.py | 2 +- synapse/replication/slave/storage/deviceinbox.py | 2 +- synapse/replication/slave/storage/devices.py | 4 +- synapse/replication/slave/storage/directory.py | 2 +- synapse/replication/slave/storage/events.py | 20 +- synapse/replication/slave/storage/filtering.py | 2 +- synapse/replication/slave/storage/keys.py | 2 +- synapse/replication/slave/storage/presence.py | 2 +- synapse/replication/slave/storage/profile.py | 2 +- synapse/replication/slave/storage/push_rule.py | 2 +- synapse/replication/slave/storage/pushers.py | 2 +- synapse/replication/slave/storage/receipts.py | 2 +- synapse/replication/slave/storage/registration.py | 2 +- synapse/replication/slave/storage/room.py | 2 +- synapse/replication/slave/storage/transactions.py | 2 +- synapse/storage/__init__.py | 504 +--- synapse/storage/account_data.py | 391 --- synapse/storage/appservice.py | 368 --- synapse/storage/client_ips.py | 581 ----- synapse/storage/data_stores/__init__.py | 14 + synapse/storage/data_stores/main/__init__.py | 524 +++++ synapse/storage/data_stores/main/account_data.py | 391 +++ synapse/storage/data_stores/main/appservice.py | 367 +++ synapse/storage/data_stores/main/client_ips.py | 580 +++++ synapse/storage/data_stores/main/deviceinbox.py | 457 ++++ synapse/storage/data_stores/main/devices.py | 937 ++++++++ synapse/storage/data_stores/main/directory.py | 173 ++ synapse/storage/data_stores/main/e2e_room_keys.py | 336 +++ .../storage/data_stores/main/end_to_end_keys.py | 322 +++ .../storage/data_stores/main/event_federation.py | 672 ++++++ .../storage/data_stores/main/event_push_actions.py | 960 ++++++++ synapse/storage/data_stores/main/events.py | 2489 ++++++++++++++++++++ .../storage/data_stores/main/events_bg_updates.py | 505 ++++ synapse/storage/data_stores/main/events_worker.py | 882 +++++++ synapse/storage/data_stores/main/filtering.py | 74 + synapse/storage/data_stores/main/group_server.py | 1180 ++++++++++ synapse/storage/data_stores/main/keys.py | 214 ++ .../storage/data_stores/main/media_repository.py | 378 +++ .../data_stores/main/monthly_active_users.py | 328 +++ synapse/storage/data_stores/main/openid.py | 31 + synapse/storage/data_stores/main/presence.py | 150 ++ synapse/storage/data_stores/main/profile.py | 178 ++ synapse/storage/data_stores/main/push_rule.py | 713 ++++++ synapse/storage/data_stores/main/pusher.py | 371 +++ synapse/storage/data_stores/main/receipts.py | 536 +++++ synapse/storage/data_stores/main/registration.py | 1499 ++++++++++++ synapse/storage/data_stores/main/rejections.py | 42 + synapse/storage/data_stores/main/relations.py | 385 +++ synapse/storage/data_stores/main/room.py | 681 ++++++ synapse/storage/data_stores/main/roommember.py | 1145 +++++++++ .../data_stores/main/schema/delta/12/v12.sql | 63 + .../data_stores/main/schema/delta/13/v13.sql | 19 + .../data_stores/main/schema/delta/14/v14.sql | 23 + .../main/schema/delta/15/appservice_txns.sql | 31 + .../main/schema/delta/15/presence_indices.sql | 2 + .../data_stores/main/schema/delta/15/v15.sql | 24 + .../main/schema/delta/16/events_order_index.sql | 4 + .../schema/delta/16/remote_media_cache_index.sql | 2 + .../main/schema/delta/16/remove_duplicates.sql | 9 + .../main/schema/delta/16/room_alias_index.sql | 3 + .../main/schema/delta/16/unique_constraints.sql | 72 + .../data_stores/main/schema/delta/16/users.sql | 56 + .../main/schema/delta/17/drop_indexes.sql | 18 + .../main/schema/delta/17/server_keys.sql | 24 + .../main/schema/delta/17/user_threepids.sql | 9 + .../schema/delta/18/server_keys_bigger_ints.sql | 32 + .../main/schema/delta/19/event_index.sql | 19 + .../data_stores/main/schema/delta/20/dummy.sql | 1 + .../data_stores/main/schema/delta/20/pushers.py | 88 + .../main/schema/delta/21/end_to_end_keys.sql | 34 + .../data_stores/main/schema/delta/21/receipts.sql | 38 + .../main/schema/delta/22/receipts_index.sql | 22 + .../main/schema/delta/22/user_threepids_unique.sql | 19 + .../main/schema/delta/23/drop_state_index.sql | 16 + .../main/schema/delta/24/stats_reporting.sql | 18 + .../main/schema/delta/25/00background_updates.sql | 21 + .../data_stores/main/schema/delta/25/fts.py | 82 + .../main/schema/delta/25/guest_access.sql | 25 + .../main/schema/delta/25/history_visibility.sql | 25 + .../data_stores/main/schema/delta/25/tags.sql | 38 + .../main/schema/delta/26/account_data.sql | 17 + .../main/schema/delta/27/account_data.sql | 36 + .../main/schema/delta/27/forgotten_memberships.sql | 26 + .../storage/data_stores/main/schema/delta/27/ts.py | 61 + .../main/schema/delta/28/event_push_actions.sql | 27 + .../main/schema/delta/28/events_room_stream.sql | 20 + .../main/schema/delta/28/public_roms_index.sql | 20 + .../schema/delta/28/receipts_user_id_index.sql | 22 + .../main/schema/delta/28/upgrade_times.sql | 21 + .../main/schema/delta/28/users_is_guest.sql | 22 + .../main/schema/delta/29/push_actions.sql | 35 + .../main/schema/delta/30/alias_creator.sql | 16 + .../data_stores/main/schema/delta/30/as_users.py | 69 + .../main/schema/delta/30/deleted_pushers.sql | 25 + .../main/schema/delta/30/presence_stream.sql | 30 + .../main/schema/delta/30/public_rooms.sql | 23 + .../main/schema/delta/30/push_rule_stream.sql | 38 + .../main/schema/delta/30/state_stream.sql | 33 + .../delta/30/threepid_guest_access_tokens.sql | 24 + .../data_stores/main/schema/delta/31/invites.sql | 42 + .../delta/31/local_media_repository_url_cache.sql | 27 + .../data_stores/main/schema/delta/31/pushers.py | 87 + .../main/schema/delta/31/pushers_index.sql | 22 + .../main/schema/delta/31/search_update.py | 66 + .../data_stores/main/schema/delta/32/events.sql | 16 + .../data_stores/main/schema/delta/32/openid.sql | 9 + .../main/schema/delta/32/pusher_throttle.sql | 23 + .../main/schema/delta/32/remove_indices.sql | 34 + .../data_stores/main/schema/delta/32/reports.sql | 25 + .../schema/delta/33/access_tokens_device_index.sql | 17 + .../data_stores/main/schema/delta/33/devices.sql | 21 + .../main/schema/delta/33/devices_for_e2e_keys.sql | 19 + .../devices_for_e2e_keys_clear_unknown_device.sql | 20 + .../main/schema/delta/33/event_fields.py | 61 + .../main/schema/delta/33/remote_media_ts.py | 30 + .../main/schema/delta/33/user_ips_index.sql | 17 + .../main/schema/delta/34/appservice_stream.sql | 23 + .../main/schema/delta/34/cache_stream.py | 46 + .../main/schema/delta/34/device_inbox.sql | 24 + .../schema/delta/34/push_display_name_rename.sql | 20 + .../main/schema/delta/34/received_txn_purge.py | 32 + .../main/schema/delta/35/add_state_index.sql | 17 + .../main/schema/delta/35/contains_url.sql | 17 + .../main/schema/delta/35/device_outbox.sql | 39 + .../main/schema/delta/35/device_stream_id.sql | 21 + .../schema/delta/35/event_push_actions_index.sql | 17 + .../delta/35/public_room_list_change_stream.sql | 33 + .../data_stores/main/schema/delta/35/state.sql | 22 + .../main/schema/delta/35/state_dedupe.sql | 17 + .../schema/delta/35/stream_order_to_extrem.sql | 37 + .../main/schema/delta/36/readd_public_rooms.sql | 26 + .../main/schema/delta/37/remove_auth_idx.py | 85 + .../main/schema/delta/37/user_threepids.sql | 52 + .../main/schema/delta/38/postgres_fts_gist.sql | 19 + .../main/schema/delta/39/appservice_room_list.sql | 29 + .../delta/39/device_federation_stream_idx.sql | 16 + .../main/schema/delta/39/event_push_index.sql | 17 + .../schema/delta/39/federation_out_position.sql | 22 + .../main/schema/delta/39/membership_profile.sql | 20 + .../main/schema/delta/40/current_state_idx.sql | 17 + .../main/schema/delta/40/device_inbox.sql | 21 + .../main/schema/delta/40/device_list_streams.sql | 60 + .../main/schema/delta/40/event_push_summary.sql | 37 + .../data_stores/main/schema/delta/40/pushers.sql | 39 + .../schema/delta/41/device_list_stream_idx.sql | 17 + .../main/schema/delta/41/device_outbound_index.sql | 16 + .../schema/delta/41/event_search_event_id_idx.sql | 17 + .../data_stores/main/schema/delta/41/ratelimit.sql | 22 + .../main/schema/delta/42/current_state_delta.sql | 26 + .../main/schema/delta/42/device_list_last_id.sql | 33 + .../main/schema/delta/42/event_auth_state_only.sql | 17 + .../data_stores/main/schema/delta/42/user_dir.py | 84 + .../main/schema/delta/43/blocked_rooms.sql | 21 + .../main/schema/delta/43/quarantine_media.sql | 17 + .../data_stores/main/schema/delta/43/url_cache.sql | 16 + .../main/schema/delta/43/user_share.sql | 33 + .../main/schema/delta/44/expire_url_cache.sql | 41 + .../main/schema/delta/45/group_server.sql | 167 ++ .../main/schema/delta/45/profile_cache.sql | 28 + .../main/schema/delta/46/drop_refresh_tokens.sql | 17 + .../delta/46/drop_unique_deleted_pushers.sql | 35 + .../main/schema/delta/46/group_server.sql | 32 + .../delta/46/local_media_repository_url_idx.sql | 24 + .../schema/delta/46/user_dir_null_room_ids.sql | 35 + .../main/schema/delta/46/user_dir_typos.sql | 24 + .../main/schema/delta/47/last_access_media.sql | 16 + .../main/schema/delta/47/postgres_fts_gin.sql | 17 + .../main/schema/delta/47/push_actions_staging.sql | 28 + .../main/schema/delta/47/state_group_seq.py | 34 + .../main/schema/delta/48/add_user_consent.sql | 18 + .../delta/48/add_user_ips_last_seen_index.sql | 17 + .../main/schema/delta/48/deactivated_users.sql | 25 + .../main/schema/delta/48/group_unique_indexes.py | 63 + .../main/schema/delta/48/groups_joinable.sql | 22 + .../49/add_user_consent_server_notice_sent.sql | 20 + .../main/schema/delta/49/add_user_daily_visits.sql | 21 + .../delta/49/add_user_ips_last_seen_only_index.sql | 17 + .../delta/50/add_creation_ts_users_index.sql | 19 + .../main/schema/delta/50/erasure_store.sql | 21 + .../schema/delta/50/make_event_content_nullable.py | 96 + .../main/schema/delta/51/e2e_room_keys.sql | 39 + .../main/schema/delta/51/monthly_active_users.sql | 27 + .../delta/52/add_event_to_state_group_index.sql | 19 + .../delta/52/device_list_streams_unique_idx.sql | 36 + .../main/schema/delta/52/e2e_room_keys.sql | 53 + .../schema/delta/53/add_user_type_to_users.sql | 19 + .../schema/delta/53/drop_sent_transactions.sql | 16 + .../main/schema/delta/53/event_format_version.sql | 16 + .../main/schema/delta/53/user_dir_populate.sql | 30 + .../main/schema/delta/53/user_ips_index.sql | 30 + .../main/schema/delta/53/user_share.sql | 44 + .../main/schema/delta/53/user_threepid_id.sql | 29 + .../main/schema/delta/53/users_in_public_rooms.sql | 28 + .../delta/54/account_validity_with_renewal.sql | 30 + .../delta/54/add_validity_to_server_keys.sql | 23 + .../schema/delta/54/delete_forward_extremities.sql | 23 + .../main/schema/delta/54/drop_legacy_tables.sql | 30 + .../main/schema/delta/54/drop_presence_list.sql | 16 + .../data_stores/main/schema/delta/54/relations.sql | 27 + .../data_stores/main/schema/delta/54/stats.sql | 80 + .../data_stores/main/schema/delta/54/stats2.sql | 28 + .../main/schema/delta/55/access_token_expiry.sql | 18 + .../schema/delta/55/track_threepid_validations.sql | 31 + .../schema/delta/55/users_alter_deactivated.sql | 19 + .../schema/delta/56/add_spans_to_device_lists.sql | 20 + .../delta/56/current_state_events_membership.sql | 22 + .../56/current_state_events_membership_mk2.sql | 24 + .../schema/delta/56/destinations_failure_ts.sql | 25 + .../destinations_retry_interval_type.sql.postgres | 18 + .../main/schema/delta/56/devices_last_seen.sql | 24 + .../schema/delta/56/drop_unused_event_tables.sql | 20 + .../main/schema/delta/56/fix_room_keys_index.sql | 18 + .../main/schema/delta/56/public_room_list_idx.sql | 16 + .../main/schema/delta/56/redaction_censor.sql | 17 + .../main/schema/delta/56/redaction_censor2.sql | 20 + .../56/redaction_censor3_fix_update.sql.postgres | 25 + .../main/schema/delta/56/room_membership_idx.sql | 18 + .../main/schema/delta/56/stats_separated.sql | 152 ++ .../schema/delta/56/unique_user_filter_index.py | 52 + .../main/schema/delta/56/user_external_ids.sql | 24 + .../schema/delta/56/users_in_public_rooms_idx.sql | 17 + .../full_schemas/16/application_services.sql | 37 + .../main/schema/full_schemas/16/event_edges.sql | 70 + .../schema/full_schemas/16/event_signatures.sql | 38 + .../data_stores/main/schema/full_schemas/16/im.sql | 120 + .../main/schema/full_schemas/16/keys.sql | 26 + .../schema/full_schemas/16/media_repository.sql | 68 + .../main/schema/full_schemas/16/presence.sql | 32 + .../main/schema/full_schemas/16/profiles.sql | 20 + .../main/schema/full_schemas/16/push.sql | 74 + .../main/schema/full_schemas/16/redactions.sql | 22 + .../main/schema/full_schemas/16/room_aliases.sql | 29 + .../main/schema/full_schemas/16/state.sql | 40 + .../main/schema/full_schemas/16/transactions.sql | 44 + .../main/schema/full_schemas/16/users.sql | 42 + .../main/schema/full_schemas/54/full.sql.postgres | 2035 ++++++++++++++++ .../main/schema/full_schemas/54/full.sql.sqlite | 259 ++ .../schema/full_schemas/54/stream_positions.sql | 7 + .../main/schema/full_schemas/README.txt | 19 + synapse/storage/data_stores/main/search.py | 712 ++++++ synapse/storage/data_stores/main/signatures.py | 101 + synapse/storage/data_stores/main/state.py | 1244 ++++++++++ synapse/storage/data_stores/main/state_deltas.py | 119 + synapse/storage/data_stores/main/stats.py | 881 +++++++ synapse/storage/data_stores/main/stream.py | 948 ++++++++ synapse/storage/data_stores/main/tags.py | 265 +++ synapse/storage/data_stores/main/transactions.py | 273 +++ synapse/storage/data_stores/main/user_directory.py | 827 +++++++ .../storage/data_stores/main/user_erasure_store.py | 91 + synapse/storage/deviceinbox.py | 457 ---- synapse/storage/devices.py | 937 -------- synapse/storage/directory.py | 174 -- synapse/storage/e2e_room_keys.py | 337 --- synapse/storage/end_to_end_keys.py | 323 --- synapse/storage/event_federation.py | 672 ------ synapse/storage/event_push_actions.py | 960 -------- synapse/storage/events.py | 2489 -------------------- synapse/storage/events_bg_updates.py | 505 ---- synapse/storage/events_worker.py | 882 ------- synapse/storage/filtering.py | 75 - synapse/storage/group_server.py | 1181 ---------- synapse/storage/keys.py | 194 -- synapse/storage/media_repository.py | 378 --- synapse/storage/monthly_active_users.py | 329 --- synapse/storage/openid.py | 31 - synapse/storage/presence.py | 134 -- synapse/storage/profile.py | 179 -- synapse/storage/push_rule.py | 698 ------ synapse/storage/pusher.py | 372 --- synapse/storage/receipts.py | 536 ----- synapse/storage/registration.py | 1499 ------------ synapse/storage/rejections.py | 42 - synapse/storage/relations.py | 359 --- synapse/storage/room.py | 681 ------ synapse/storage/roommember.py | 1119 --------- synapse/storage/schema/delta/12/v12.sql | 63 - synapse/storage/schema/delta/13/v13.sql | 19 - synapse/storage/schema/delta/14/v14.sql | 23 - .../storage/schema/delta/15/appservice_txns.sql | 31 - .../storage/schema/delta/15/presence_indices.sql | 2 - synapse/storage/schema/delta/15/v15.sql | 24 - .../storage/schema/delta/16/events_order_index.sql | 4 - .../schema/delta/16/remote_media_cache_index.sql | 2 - .../storage/schema/delta/16/remove_duplicates.sql | 9 - .../storage/schema/delta/16/room_alias_index.sql | 3 - .../storage/schema/delta/16/unique_constraints.sql | 72 - synapse/storage/schema/delta/16/users.sql | 56 - synapse/storage/schema/delta/17/drop_indexes.sql | 18 - synapse/storage/schema/delta/17/server_keys.sql | 24 - synapse/storage/schema/delta/17/user_threepids.sql | 9 - .../schema/delta/18/server_keys_bigger_ints.sql | 32 - synapse/storage/schema/delta/19/event_index.sql | 19 - synapse/storage/schema/delta/20/dummy.sql | 1 - synapse/storage/schema/delta/20/pushers.py | 88 - .../storage/schema/delta/21/end_to_end_keys.sql | 34 - synapse/storage/schema/delta/21/receipts.sql | 38 - synapse/storage/schema/delta/22/receipts_index.sql | 22 - .../schema/delta/22/user_threepids_unique.sql | 19 - .../storage/schema/delta/23/drop_state_index.sql | 16 - .../storage/schema/delta/24/stats_reporting.sql | 18 - synapse/storage/schema/delta/25/fts.py | 82 - synapse/storage/schema/delta/25/guest_access.sql | 25 - .../storage/schema/delta/25/history_visibility.sql | 25 - synapse/storage/schema/delta/25/tags.sql | 38 - synapse/storage/schema/delta/26/account_data.sql | 17 - synapse/storage/schema/delta/27/account_data.sql | 36 - .../schema/delta/27/forgotten_memberships.sql | 26 - synapse/storage/schema/delta/27/ts.py | 61 - .../storage/schema/delta/28/event_push_actions.sql | 27 - .../storage/schema/delta/28/events_room_stream.sql | 20 - .../storage/schema/delta/28/public_roms_index.sql | 20 - .../schema/delta/28/receipts_user_id_index.sql | 22 - synapse/storage/schema/delta/28/upgrade_times.sql | 21 - synapse/storage/schema/delta/28/users_is_guest.sql | 22 - synapse/storage/schema/delta/29/push_actions.sql | 35 - synapse/storage/schema/delta/30/alias_creator.sql | 16 - synapse/storage/schema/delta/30/as_users.py | 69 - .../storage/schema/delta/30/deleted_pushers.sql | 25 - .../storage/schema/delta/30/presence_stream.sql | 30 - synapse/storage/schema/delta/30/public_rooms.sql | 23 - .../storage/schema/delta/30/push_rule_stream.sql | 38 - synapse/storage/schema/delta/30/state_stream.sql | 33 - .../delta/30/threepid_guest_access_tokens.sql | 24 - synapse/storage/schema/delta/31/invites.sql | 42 - .../delta/31/local_media_repository_url_cache.sql | 27 - synapse/storage/schema/delta/31/pushers.py | 87 - synapse/storage/schema/delta/31/pushers_index.sql | 22 - synapse/storage/schema/delta/31/search_update.py | 66 - synapse/storage/schema/delta/32/events.sql | 16 - synapse/storage/schema/delta/32/openid.sql | 9 - .../storage/schema/delta/32/pusher_throttle.sql | 23 - synapse/storage/schema/delta/32/remove_indices.sql | 34 - synapse/storage/schema/delta/32/reports.sql | 25 - .../schema/delta/33/access_tokens_device_index.sql | 17 - synapse/storage/schema/delta/33/devices.sql | 21 - .../schema/delta/33/devices_for_e2e_keys.sql | 19 - .../devices_for_e2e_keys_clear_unknown_device.sql | 20 - synapse/storage/schema/delta/33/event_fields.py | 61 - synapse/storage/schema/delta/33/remote_media_ts.py | 30 - synapse/storage/schema/delta/33/user_ips_index.sql | 17 - .../storage/schema/delta/34/appservice_stream.sql | 23 - synapse/storage/schema/delta/34/cache_stream.py | 46 - synapse/storage/schema/delta/34/device_inbox.sql | 24 - .../schema/delta/34/push_display_name_rename.sql | 20 - .../storage/schema/delta/34/received_txn_purge.py | 32 - .../delta/35/00background_updates_add_col.sql | 17 + .../storage/schema/delta/35/add_state_index.sql | 20 - synapse/storage/schema/delta/35/contains_url.sql | 17 - synapse/storage/schema/delta/35/device_outbox.sql | 39 - .../storage/schema/delta/35/device_stream_id.sql | 21 - .../schema/delta/35/event_push_actions_index.sql | 17 - .../delta/35/public_room_list_change_stream.sql | 33 - synapse/storage/schema/delta/35/state.sql | 22 - synapse/storage/schema/delta/35/state_dedupe.sql | 17 - .../schema/delta/35/stream_order_to_extrem.sql | 37 - .../storage/schema/delta/36/readd_public_rooms.sql | 26 - synapse/storage/schema/delta/37/remove_auth_idx.py | 85 - synapse/storage/schema/delta/37/user_threepids.sql | 52 - .../storage/schema/delta/38/postgres_fts_gist.sql | 19 - .../schema/delta/39/appservice_room_list.sql | 29 - .../delta/39/device_federation_stream_idx.sql | 16 - .../storage/schema/delta/39/event_push_index.sql | 17 - .../schema/delta/39/federation_out_position.sql | 22 - .../storage/schema/delta/39/membership_profile.sql | 20 - .../storage/schema/delta/40/current_state_idx.sql | 17 - synapse/storage/schema/delta/40/device_inbox.sql | 21 - .../schema/delta/40/device_list_streams.sql | 60 - .../storage/schema/delta/40/event_push_summary.sql | 37 - synapse/storage/schema/delta/40/pushers.sql | 39 - .../schema/delta/41/device_list_stream_idx.sql | 17 - .../schema/delta/41/device_outbound_index.sql | 16 - .../schema/delta/41/event_search_event_id_idx.sql | 17 - synapse/storage/schema/delta/41/ratelimit.sql | 22 - .../schema/delta/42/current_state_delta.sql | 26 - .../schema/delta/42/device_list_last_id.sql | 33 - .../schema/delta/42/event_auth_state_only.sql | 17 - synapse/storage/schema/delta/42/user_dir.py | 84 - synapse/storage/schema/delta/43/blocked_rooms.sql | 21 - .../storage/schema/delta/43/quarantine_media.sql | 17 - synapse/storage/schema/delta/43/url_cache.sql | 16 - synapse/storage/schema/delta/43/user_share.sql | 33 - .../storage/schema/delta/44/expire_url_cache.sql | 41 - synapse/storage/schema/delta/45/group_server.sql | 167 -- synapse/storage/schema/delta/45/profile_cache.sql | 28 - .../schema/delta/46/drop_refresh_tokens.sql | 17 - .../delta/46/drop_unique_deleted_pushers.sql | 35 - synapse/storage/schema/delta/46/group_server.sql | 32 - .../delta/46/local_media_repository_url_idx.sql | 24 - .../schema/delta/46/user_dir_null_room_ids.sql | 35 - synapse/storage/schema/delta/46/user_dir_typos.sql | 24 - .../storage/schema/delta/47/last_access_media.sql | 16 - .../storage/schema/delta/47/postgres_fts_gin.sql | 17 - .../schema/delta/47/push_actions_staging.sql | 28 - synapse/storage/schema/delta/47/state_group_seq.py | 34 - .../storage/schema/delta/48/add_user_consent.sql | 18 - .../delta/48/add_user_ips_last_seen_index.sql | 17 - .../storage/schema/delta/48/deactivated_users.sql | 25 - .../schema/delta/48/group_unique_indexes.py | 63 - .../storage/schema/delta/48/groups_joinable.sql | 22 - .../49/add_user_consent_server_notice_sent.sql | 20 - .../schema/delta/49/add_user_daily_visits.sql | 21 - .../delta/49/add_user_ips_last_seen_only_index.sql | 17 - .../delta/50/add_creation_ts_users_index.sql | 19 - synapse/storage/schema/delta/50/erasure_store.sql | 21 - .../schema/delta/50/make_event_content_nullable.py | 96 - synapse/storage/schema/delta/51/e2e_room_keys.sql | 39 - .../schema/delta/51/monthly_active_users.sql | 27 - .../delta/52/add_event_to_state_group_index.sql | 19 - .../delta/52/device_list_streams_unique_idx.sql | 36 - synapse/storage/schema/delta/52/e2e_room_keys.sql | 53 - .../schema/delta/53/add_user_type_to_users.sql | 19 - .../schema/delta/53/drop_sent_transactions.sql | 16 - .../schema/delta/53/event_format_version.sql | 16 - .../storage/schema/delta/53/user_dir_populate.sql | 30 - synapse/storage/schema/delta/53/user_ips_index.sql | 30 - synapse/storage/schema/delta/53/user_share.sql | 44 - .../storage/schema/delta/53/user_threepid_id.sql | 29 - .../schema/delta/53/users_in_public_rooms.sql | 28 - .../delta/54/account_validity_with_renewal.sql | 30 - .../delta/54/add_validity_to_server_keys.sql | 23 - .../schema/delta/54/delete_forward_extremities.sql | 23 - .../storage/schema/delta/54/drop_legacy_tables.sql | 30 - .../storage/schema/delta/54/drop_presence_list.sql | 16 - synapse/storage/schema/delta/54/relations.sql | 27 - synapse/storage/schema/delta/54/stats.sql | 80 - synapse/storage/schema/delta/54/stats2.sql | 28 - .../schema/delta/55/access_token_expiry.sql | 18 - .../schema/delta/55/track_threepid_validations.sql | 31 - .../schema/delta/55/users_alter_deactivated.sql | 19 - .../schema/delta/56/add_spans_to_device_lists.sql | 20 - .../delta/56/current_state_events_membership.sql | 22 - .../56/current_state_events_membership_mk2.sql | 24 - .../schema/delta/56/destinations_failure_ts.sql | 25 - .../destinations_retry_interval_type.sql.postgres | 18 - .../storage/schema/delta/56/devices_last_seen.sql | 24 - .../schema/delta/56/drop_unused_event_tables.sql | 20 - .../schema/delta/56/fix_room_keys_index.sql | 18 - .../schema/delta/56/public_room_list_idx.sql | 16 - .../storage/schema/delta/56/redaction_censor.sql | 17 - .../storage/schema/delta/56/redaction_censor2.sql | 20 - .../56/redaction_censor3_fix_update.sql.postgres | 25 - .../schema/delta/56/room_membership_idx.sql | 18 - .../storage/schema/delta/56/stats_separated.sql | 152 -- .../schema/delta/56/unique_user_filter_index.py | 52 - .../storage/schema/delta/56/user_external_ids.sql | 24 - .../schema/delta/56/users_in_public_rooms_idx.sql | 17 - .../full_schemas/16/application_services.sql | 37 - .../storage/schema/full_schemas/16/event_edges.sql | 70 - .../schema/full_schemas/16/event_signatures.sql | 38 - synapse/storage/schema/full_schemas/16/im.sql | 120 - synapse/storage/schema/full_schemas/16/keys.sql | 26 - .../schema/full_schemas/16/media_repository.sql | 68 - .../storage/schema/full_schemas/16/presence.sql | 32 - .../storage/schema/full_schemas/16/profiles.sql | 20 - synapse/storage/schema/full_schemas/16/push.sql | 74 - .../storage/schema/full_schemas/16/redactions.sql | 22 - .../schema/full_schemas/16/room_aliases.sql | 29 - synapse/storage/schema/full_schemas/16/state.sql | 40 - .../schema/full_schemas/16/transactions.sql | 44 - synapse/storage/schema/full_schemas/16/users.sql | 42 - synapse/storage/schema/full_schemas/54/full.sql | 8 + .../schema/full_schemas/54/full.sql.postgres | 2052 ---------------- .../storage/schema/full_schemas/54/full.sql.sqlite | 260 -- .../schema/full_schemas/54/stream_positions.sql | 7 - synapse/storage/schema/full_schemas/README.txt | 19 - synapse/storage/search.py | 713 ------ synapse/storage/signatures.py | 102 - synapse/storage/state.py | 1221 ---------- synapse/storage/state_deltas.py | 119 - synapse/storage/stats.py | 881 ------- synapse/storage/stream.py | 948 -------- synapse/storage/tags.py | 265 --- synapse/storage/transactions.py | 274 --- synapse/storage/user_directory.py | 827 ------- synapse/storage/user_erasure_store.py | 91 - tests/handlers/test_stats.py | 8 +- tests/storage/test_appservice.py | 2 +- tests/storage/test_cleanup_extrems.py | 2 + tests/storage/test_profile.py | 2 +- tests/storage/test_user_directory.py | 2 +- 487 files changed, 31168 insertions(+), 30990 deletions(-) delete mode 100644 synapse/storage/account_data.py delete mode 100644 synapse/storage/appservice.py delete mode 100644 synapse/storage/client_ips.py create mode 100644 synapse/storage/data_stores/__init__.py create mode 100644 synapse/storage/data_stores/main/__init__.py create mode 100644 synapse/storage/data_stores/main/account_data.py create mode 100644 synapse/storage/data_stores/main/appservice.py create mode 100644 synapse/storage/data_stores/main/client_ips.py create mode 100644 synapse/storage/data_stores/main/deviceinbox.py create mode 100644 synapse/storage/data_stores/main/devices.py create mode 100644 synapse/storage/data_stores/main/directory.py create mode 100644 synapse/storage/data_stores/main/e2e_room_keys.py create mode 100644 synapse/storage/data_stores/main/end_to_end_keys.py create mode 100644 synapse/storage/data_stores/main/event_federation.py create mode 100644 synapse/storage/data_stores/main/event_push_actions.py create mode 100644 synapse/storage/data_stores/main/events.py create mode 100644 synapse/storage/data_stores/main/events_bg_updates.py create mode 100644 synapse/storage/data_stores/main/events_worker.py create mode 100644 synapse/storage/data_stores/main/filtering.py create mode 100644 synapse/storage/data_stores/main/group_server.py create mode 100644 synapse/storage/data_stores/main/keys.py create mode 100644 synapse/storage/data_stores/main/media_repository.py create mode 100644 synapse/storage/data_stores/main/monthly_active_users.py create mode 100644 synapse/storage/data_stores/main/openid.py create mode 100644 synapse/storage/data_stores/main/presence.py create mode 100644 synapse/storage/data_stores/main/profile.py create mode 100644 synapse/storage/data_stores/main/push_rule.py create mode 100644 synapse/storage/data_stores/main/pusher.py create mode 100644 synapse/storage/data_stores/main/receipts.py create mode 100644 synapse/storage/data_stores/main/registration.py create mode 100644 synapse/storage/data_stores/main/rejections.py create mode 100644 synapse/storage/data_stores/main/relations.py create mode 100644 synapse/storage/data_stores/main/room.py create mode 100644 synapse/storage/data_stores/main/roommember.py create mode 100644 synapse/storage/data_stores/main/schema/delta/12/v12.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/13/v13.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/14/v14.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/15/appservice_txns.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/15/presence_indices.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/15/v15.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/16/events_order_index.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/16/remote_media_cache_index.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/16/remove_duplicates.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/16/room_alias_index.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/16/unique_constraints.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/16/users.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/17/drop_indexes.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/17/server_keys.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/17/user_threepids.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/18/server_keys_bigger_ints.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/19/event_index.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/20/dummy.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/20/pushers.py create mode 100644 synapse/storage/data_stores/main/schema/delta/21/end_to_end_keys.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/21/receipts.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/22/receipts_index.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/22/user_threepids_unique.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/23/drop_state_index.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/24/stats_reporting.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/25/00background_updates.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/25/fts.py create mode 100644 synapse/storage/data_stores/main/schema/delta/25/guest_access.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/25/history_visibility.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/25/tags.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/26/account_data.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/27/account_data.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/27/forgotten_memberships.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/27/ts.py create mode 100644 synapse/storage/data_stores/main/schema/delta/28/event_push_actions.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/28/events_room_stream.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/28/public_roms_index.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/28/receipts_user_id_index.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/28/upgrade_times.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/28/users_is_guest.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/29/push_actions.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/30/alias_creator.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/30/as_users.py create mode 100644 synapse/storage/data_stores/main/schema/delta/30/deleted_pushers.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/30/presence_stream.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/30/public_rooms.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/30/push_rule_stream.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/30/state_stream.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/30/threepid_guest_access_tokens.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/31/invites.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/31/local_media_repository_url_cache.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/31/pushers.py create mode 100644 synapse/storage/data_stores/main/schema/delta/31/pushers_index.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/31/search_update.py create mode 100644 synapse/storage/data_stores/main/schema/delta/32/events.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/32/openid.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/32/pusher_throttle.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/32/remove_indices.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/32/reports.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/33/access_tokens_device_index.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/33/devices.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/33/devices_for_e2e_keys.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/33/devices_for_e2e_keys_clear_unknown_device.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/33/event_fields.py create mode 100644 synapse/storage/data_stores/main/schema/delta/33/remote_media_ts.py create mode 100644 synapse/storage/data_stores/main/schema/delta/33/user_ips_index.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/34/appservice_stream.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/34/cache_stream.py create mode 100644 synapse/storage/data_stores/main/schema/delta/34/device_inbox.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/34/push_display_name_rename.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/34/received_txn_purge.py create mode 100644 synapse/storage/data_stores/main/schema/delta/35/add_state_index.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/35/contains_url.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/35/device_outbox.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/35/device_stream_id.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/35/event_push_actions_index.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/35/public_room_list_change_stream.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/35/state.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/35/state_dedupe.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/35/stream_order_to_extrem.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/36/readd_public_rooms.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/37/remove_auth_idx.py create mode 100644 synapse/storage/data_stores/main/schema/delta/37/user_threepids.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/38/postgres_fts_gist.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/39/appservice_room_list.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/39/device_federation_stream_idx.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/39/event_push_index.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/39/federation_out_position.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/39/membership_profile.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/40/current_state_idx.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/40/device_inbox.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/40/device_list_streams.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/40/event_push_summary.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/40/pushers.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/41/device_list_stream_idx.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/41/device_outbound_index.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/41/event_search_event_id_idx.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/41/ratelimit.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/42/current_state_delta.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/42/device_list_last_id.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/42/event_auth_state_only.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/42/user_dir.py create mode 100644 synapse/storage/data_stores/main/schema/delta/43/blocked_rooms.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/43/quarantine_media.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/43/url_cache.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/43/user_share.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/44/expire_url_cache.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/45/group_server.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/45/profile_cache.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/46/drop_refresh_tokens.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/46/drop_unique_deleted_pushers.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/46/group_server.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/46/local_media_repository_url_idx.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/46/user_dir_null_room_ids.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/46/user_dir_typos.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/47/last_access_media.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/47/postgres_fts_gin.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/47/push_actions_staging.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/47/state_group_seq.py create mode 100644 synapse/storage/data_stores/main/schema/delta/48/add_user_consent.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/48/add_user_ips_last_seen_index.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/48/deactivated_users.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/48/group_unique_indexes.py create mode 100644 synapse/storage/data_stores/main/schema/delta/48/groups_joinable.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/49/add_user_consent_server_notice_sent.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/49/add_user_daily_visits.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/49/add_user_ips_last_seen_only_index.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/50/add_creation_ts_users_index.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/50/erasure_store.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/50/make_event_content_nullable.py create mode 100644 synapse/storage/data_stores/main/schema/delta/51/e2e_room_keys.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/51/monthly_active_users.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/52/add_event_to_state_group_index.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/52/device_list_streams_unique_idx.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/52/e2e_room_keys.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/53/add_user_type_to_users.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/53/drop_sent_transactions.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/53/event_format_version.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/53/user_dir_populate.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/53/user_ips_index.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/53/user_share.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/53/user_threepid_id.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/53/users_in_public_rooms.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/54/account_validity_with_renewal.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/54/add_validity_to_server_keys.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/54/delete_forward_extremities.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/54/drop_legacy_tables.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/54/drop_presence_list.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/54/relations.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/54/stats.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/54/stats2.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/55/access_token_expiry.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/55/track_threepid_validations.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/55/users_alter_deactivated.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/56/add_spans_to_device_lists.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/56/current_state_events_membership.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/56/current_state_events_membership_mk2.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/56/destinations_failure_ts.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/56/destinations_retry_interval_type.sql.postgres create mode 100644 synapse/storage/data_stores/main/schema/delta/56/devices_last_seen.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/56/drop_unused_event_tables.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/56/fix_room_keys_index.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/56/public_room_list_idx.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/56/redaction_censor.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/56/redaction_censor2.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/56/redaction_censor3_fix_update.sql.postgres create mode 100644 synapse/storage/data_stores/main/schema/delta/56/room_membership_idx.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/56/stats_separated.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/56/unique_user_filter_index.py create mode 100644 synapse/storage/data_stores/main/schema/delta/56/user_external_ids.sql create mode 100644 synapse/storage/data_stores/main/schema/delta/56/users_in_public_rooms_idx.sql create mode 100644 synapse/storage/data_stores/main/schema/full_schemas/16/application_services.sql create mode 100644 synapse/storage/data_stores/main/schema/full_schemas/16/event_edges.sql create mode 100644 synapse/storage/data_stores/main/schema/full_schemas/16/event_signatures.sql create mode 100644 synapse/storage/data_stores/main/schema/full_schemas/16/im.sql create mode 100644 synapse/storage/data_stores/main/schema/full_schemas/16/keys.sql create mode 100644 synapse/storage/data_stores/main/schema/full_schemas/16/media_repository.sql create mode 100644 synapse/storage/data_stores/main/schema/full_schemas/16/presence.sql create mode 100644 synapse/storage/data_stores/main/schema/full_schemas/16/profiles.sql create mode 100644 synapse/storage/data_stores/main/schema/full_schemas/16/push.sql create mode 100644 synapse/storage/data_stores/main/schema/full_schemas/16/redactions.sql create mode 100644 synapse/storage/data_stores/main/schema/full_schemas/16/room_aliases.sql create mode 100644 synapse/storage/data_stores/main/schema/full_schemas/16/state.sql create mode 100644 synapse/storage/data_stores/main/schema/full_schemas/16/transactions.sql create mode 100644 synapse/storage/data_stores/main/schema/full_schemas/16/users.sql create mode 100644 synapse/storage/data_stores/main/schema/full_schemas/54/full.sql.postgres create mode 100644 synapse/storage/data_stores/main/schema/full_schemas/54/full.sql.sqlite create mode 100644 synapse/storage/data_stores/main/schema/full_schemas/54/stream_positions.sql create mode 100644 synapse/storage/data_stores/main/schema/full_schemas/README.txt create mode 100644 synapse/storage/data_stores/main/search.py create mode 100644 synapse/storage/data_stores/main/signatures.py create mode 100644 synapse/storage/data_stores/main/state.py create mode 100644 synapse/storage/data_stores/main/state_deltas.py create mode 100644 synapse/storage/data_stores/main/stats.py create mode 100644 synapse/storage/data_stores/main/stream.py create mode 100644 synapse/storage/data_stores/main/tags.py create mode 100644 synapse/storage/data_stores/main/transactions.py create mode 100644 synapse/storage/data_stores/main/user_directory.py create mode 100644 synapse/storage/data_stores/main/user_erasure_store.py delete mode 100644 synapse/storage/deviceinbox.py delete mode 100644 synapse/storage/devices.py delete mode 100644 synapse/storage/directory.py delete mode 100644 synapse/storage/e2e_room_keys.py delete mode 100644 synapse/storage/end_to_end_keys.py delete mode 100644 synapse/storage/event_federation.py delete mode 100644 synapse/storage/event_push_actions.py delete mode 100644 synapse/storage/events.py delete mode 100644 synapse/storage/events_bg_updates.py delete mode 100644 synapse/storage/events_worker.py delete mode 100644 synapse/storage/filtering.py delete mode 100644 synapse/storage/group_server.py delete mode 100644 synapse/storage/media_repository.py delete mode 100644 synapse/storage/monthly_active_users.py delete mode 100644 synapse/storage/openid.py delete mode 100644 synapse/storage/profile.py delete mode 100644 synapse/storage/pusher.py delete mode 100644 synapse/storage/receipts.py delete mode 100644 synapse/storage/registration.py delete mode 100644 synapse/storage/rejections.py delete mode 100644 synapse/storage/room.py delete mode 100644 synapse/storage/schema/delta/12/v12.sql delete mode 100644 synapse/storage/schema/delta/13/v13.sql delete mode 100644 synapse/storage/schema/delta/14/v14.sql delete mode 100644 synapse/storage/schema/delta/15/appservice_txns.sql delete mode 100644 synapse/storage/schema/delta/15/presence_indices.sql delete mode 100644 synapse/storage/schema/delta/15/v15.sql delete mode 100644 synapse/storage/schema/delta/16/events_order_index.sql delete mode 100644 synapse/storage/schema/delta/16/remote_media_cache_index.sql delete mode 100644 synapse/storage/schema/delta/16/remove_duplicates.sql delete mode 100644 synapse/storage/schema/delta/16/room_alias_index.sql delete mode 100644 synapse/storage/schema/delta/16/unique_constraints.sql delete mode 100644 synapse/storage/schema/delta/16/users.sql delete mode 100644 synapse/storage/schema/delta/17/drop_indexes.sql delete mode 100644 synapse/storage/schema/delta/17/server_keys.sql delete mode 100644 synapse/storage/schema/delta/17/user_threepids.sql delete mode 100644 synapse/storage/schema/delta/18/server_keys_bigger_ints.sql delete mode 100644 synapse/storage/schema/delta/19/event_index.sql delete mode 100644 synapse/storage/schema/delta/20/dummy.sql delete mode 100644 synapse/storage/schema/delta/20/pushers.py delete mode 100644 synapse/storage/schema/delta/21/end_to_end_keys.sql delete mode 100644 synapse/storage/schema/delta/21/receipts.sql delete mode 100644 synapse/storage/schema/delta/22/receipts_index.sql delete mode 100644 synapse/storage/schema/delta/22/user_threepids_unique.sql delete mode 100644 synapse/storage/schema/delta/23/drop_state_index.sql delete mode 100644 synapse/storage/schema/delta/24/stats_reporting.sql delete mode 100644 synapse/storage/schema/delta/25/fts.py delete mode 100644 synapse/storage/schema/delta/25/guest_access.sql delete mode 100644 synapse/storage/schema/delta/25/history_visibility.sql delete mode 100644 synapse/storage/schema/delta/25/tags.sql delete mode 100644 synapse/storage/schema/delta/26/account_data.sql delete mode 100644 synapse/storage/schema/delta/27/account_data.sql delete mode 100644 synapse/storage/schema/delta/27/forgotten_memberships.sql delete mode 100644 synapse/storage/schema/delta/27/ts.py delete mode 100644 synapse/storage/schema/delta/28/event_push_actions.sql delete mode 100644 synapse/storage/schema/delta/28/events_room_stream.sql delete mode 100644 synapse/storage/schema/delta/28/public_roms_index.sql delete mode 100644 synapse/storage/schema/delta/28/receipts_user_id_index.sql delete mode 100644 synapse/storage/schema/delta/28/upgrade_times.sql delete mode 100644 synapse/storage/schema/delta/28/users_is_guest.sql delete mode 100644 synapse/storage/schema/delta/29/push_actions.sql delete mode 100644 synapse/storage/schema/delta/30/alias_creator.sql delete mode 100644 synapse/storage/schema/delta/30/as_users.py delete mode 100644 synapse/storage/schema/delta/30/deleted_pushers.sql delete mode 100644 synapse/storage/schema/delta/30/presence_stream.sql delete mode 100644 synapse/storage/schema/delta/30/public_rooms.sql delete mode 100644 synapse/storage/schema/delta/30/push_rule_stream.sql delete mode 100644 synapse/storage/schema/delta/30/state_stream.sql delete mode 100644 synapse/storage/schema/delta/30/threepid_guest_access_tokens.sql delete mode 100644 synapse/storage/schema/delta/31/invites.sql delete mode 100644 synapse/storage/schema/delta/31/local_media_repository_url_cache.sql delete mode 100644 synapse/storage/schema/delta/31/pushers.py delete mode 100644 synapse/storage/schema/delta/31/pushers_index.sql delete mode 100644 synapse/storage/schema/delta/31/search_update.py delete mode 100644 synapse/storage/schema/delta/32/events.sql delete mode 100644 synapse/storage/schema/delta/32/openid.sql delete mode 100644 synapse/storage/schema/delta/32/pusher_throttle.sql delete mode 100644 synapse/storage/schema/delta/32/remove_indices.sql delete mode 100644 synapse/storage/schema/delta/32/reports.sql delete mode 100644 synapse/storage/schema/delta/33/access_tokens_device_index.sql delete mode 100644 synapse/storage/schema/delta/33/devices.sql delete mode 100644 synapse/storage/schema/delta/33/devices_for_e2e_keys.sql delete mode 100644 synapse/storage/schema/delta/33/devices_for_e2e_keys_clear_unknown_device.sql delete mode 100644 synapse/storage/schema/delta/33/event_fields.py delete mode 100644 synapse/storage/schema/delta/33/remote_media_ts.py delete mode 100644 synapse/storage/schema/delta/33/user_ips_index.sql delete mode 100644 synapse/storage/schema/delta/34/appservice_stream.sql delete mode 100644 synapse/storage/schema/delta/34/cache_stream.py delete mode 100644 synapse/storage/schema/delta/34/device_inbox.sql delete mode 100644 synapse/storage/schema/delta/34/push_display_name_rename.sql delete mode 100644 synapse/storage/schema/delta/34/received_txn_purge.py create mode 100644 synapse/storage/schema/delta/35/00background_updates_add_col.sql delete mode 100644 synapse/storage/schema/delta/35/add_state_index.sql delete mode 100644 synapse/storage/schema/delta/35/contains_url.sql delete mode 100644 synapse/storage/schema/delta/35/device_outbox.sql delete mode 100644 synapse/storage/schema/delta/35/device_stream_id.sql delete mode 100644 synapse/storage/schema/delta/35/event_push_actions_index.sql delete mode 100644 synapse/storage/schema/delta/35/public_room_list_change_stream.sql delete mode 100644 synapse/storage/schema/delta/35/state.sql delete mode 100644 synapse/storage/schema/delta/35/state_dedupe.sql delete mode 100644 synapse/storage/schema/delta/35/stream_order_to_extrem.sql delete mode 100644 synapse/storage/schema/delta/36/readd_public_rooms.sql delete mode 100644 synapse/storage/schema/delta/37/remove_auth_idx.py delete mode 100644 synapse/storage/schema/delta/37/user_threepids.sql delete mode 100644 synapse/storage/schema/delta/38/postgres_fts_gist.sql delete mode 100644 synapse/storage/schema/delta/39/appservice_room_list.sql delete mode 100644 synapse/storage/schema/delta/39/device_federation_stream_idx.sql delete mode 100644 synapse/storage/schema/delta/39/event_push_index.sql delete mode 100644 synapse/storage/schema/delta/39/federation_out_position.sql delete mode 100644 synapse/storage/schema/delta/39/membership_profile.sql delete mode 100644 synapse/storage/schema/delta/40/current_state_idx.sql delete mode 100644 synapse/storage/schema/delta/40/device_inbox.sql delete mode 100644 synapse/storage/schema/delta/40/device_list_streams.sql delete mode 100644 synapse/storage/schema/delta/40/event_push_summary.sql delete mode 100644 synapse/storage/schema/delta/40/pushers.sql delete mode 100644 synapse/storage/schema/delta/41/device_list_stream_idx.sql delete mode 100644 synapse/storage/schema/delta/41/device_outbound_index.sql delete mode 100644 synapse/storage/schema/delta/41/event_search_event_id_idx.sql delete mode 100644 synapse/storage/schema/delta/41/ratelimit.sql delete mode 100644 synapse/storage/schema/delta/42/current_state_delta.sql delete mode 100644 synapse/storage/schema/delta/42/device_list_last_id.sql delete mode 100644 synapse/storage/schema/delta/42/event_auth_state_only.sql delete mode 100644 synapse/storage/schema/delta/42/user_dir.py delete mode 100644 synapse/storage/schema/delta/43/blocked_rooms.sql delete mode 100644 synapse/storage/schema/delta/43/quarantine_media.sql delete mode 100644 synapse/storage/schema/delta/43/url_cache.sql delete mode 100644 synapse/storage/schema/delta/43/user_share.sql delete mode 100644 synapse/storage/schema/delta/44/expire_url_cache.sql delete mode 100644 synapse/storage/schema/delta/45/group_server.sql delete mode 100644 synapse/storage/schema/delta/45/profile_cache.sql delete mode 100644 synapse/storage/schema/delta/46/drop_refresh_tokens.sql delete mode 100644 synapse/storage/schema/delta/46/drop_unique_deleted_pushers.sql delete mode 100644 synapse/storage/schema/delta/46/group_server.sql delete mode 100644 synapse/storage/schema/delta/46/local_media_repository_url_idx.sql delete mode 100644 synapse/storage/schema/delta/46/user_dir_null_room_ids.sql delete mode 100644 synapse/storage/schema/delta/46/user_dir_typos.sql delete mode 100644 synapse/storage/schema/delta/47/last_access_media.sql delete mode 100644 synapse/storage/schema/delta/47/postgres_fts_gin.sql delete mode 100644 synapse/storage/schema/delta/47/push_actions_staging.sql delete mode 100644 synapse/storage/schema/delta/47/state_group_seq.py delete mode 100644 synapse/storage/schema/delta/48/add_user_consent.sql delete mode 100644 synapse/storage/schema/delta/48/add_user_ips_last_seen_index.sql delete mode 100644 synapse/storage/schema/delta/48/deactivated_users.sql delete mode 100644 synapse/storage/schema/delta/48/group_unique_indexes.py delete mode 100644 synapse/storage/schema/delta/48/groups_joinable.sql delete mode 100644 synapse/storage/schema/delta/49/add_user_consent_server_notice_sent.sql delete mode 100644 synapse/storage/schema/delta/49/add_user_daily_visits.sql delete mode 100644 synapse/storage/schema/delta/49/add_user_ips_last_seen_only_index.sql delete mode 100644 synapse/storage/schema/delta/50/add_creation_ts_users_index.sql delete mode 100644 synapse/storage/schema/delta/50/erasure_store.sql delete mode 100644 synapse/storage/schema/delta/50/make_event_content_nullable.py delete mode 100644 synapse/storage/schema/delta/51/e2e_room_keys.sql delete mode 100644 synapse/storage/schema/delta/51/monthly_active_users.sql delete mode 100644 synapse/storage/schema/delta/52/add_event_to_state_group_index.sql delete mode 100644 synapse/storage/schema/delta/52/device_list_streams_unique_idx.sql delete mode 100644 synapse/storage/schema/delta/52/e2e_room_keys.sql delete mode 100644 synapse/storage/schema/delta/53/add_user_type_to_users.sql delete mode 100644 synapse/storage/schema/delta/53/drop_sent_transactions.sql delete mode 100644 synapse/storage/schema/delta/53/event_format_version.sql delete mode 100644 synapse/storage/schema/delta/53/user_dir_populate.sql delete mode 100644 synapse/storage/schema/delta/53/user_ips_index.sql delete mode 100644 synapse/storage/schema/delta/53/user_share.sql delete mode 100644 synapse/storage/schema/delta/53/user_threepid_id.sql delete mode 100644 synapse/storage/schema/delta/53/users_in_public_rooms.sql delete mode 100644 synapse/storage/schema/delta/54/account_validity_with_renewal.sql delete mode 100644 synapse/storage/schema/delta/54/add_validity_to_server_keys.sql delete mode 100644 synapse/storage/schema/delta/54/delete_forward_extremities.sql delete mode 100644 synapse/storage/schema/delta/54/drop_legacy_tables.sql delete mode 100644 synapse/storage/schema/delta/54/drop_presence_list.sql delete mode 100644 synapse/storage/schema/delta/54/relations.sql delete mode 100644 synapse/storage/schema/delta/54/stats.sql delete mode 100644 synapse/storage/schema/delta/54/stats2.sql delete mode 100644 synapse/storage/schema/delta/55/access_token_expiry.sql delete mode 100644 synapse/storage/schema/delta/55/track_threepid_validations.sql delete mode 100644 synapse/storage/schema/delta/55/users_alter_deactivated.sql delete mode 100644 synapse/storage/schema/delta/56/add_spans_to_device_lists.sql delete mode 100644 synapse/storage/schema/delta/56/current_state_events_membership.sql delete mode 100644 synapse/storage/schema/delta/56/current_state_events_membership_mk2.sql delete mode 100644 synapse/storage/schema/delta/56/destinations_failure_ts.sql delete mode 100644 synapse/storage/schema/delta/56/destinations_retry_interval_type.sql.postgres delete mode 100644 synapse/storage/schema/delta/56/devices_last_seen.sql delete mode 100644 synapse/storage/schema/delta/56/drop_unused_event_tables.sql delete mode 100644 synapse/storage/schema/delta/56/fix_room_keys_index.sql delete mode 100644 synapse/storage/schema/delta/56/public_room_list_idx.sql delete mode 100644 synapse/storage/schema/delta/56/redaction_censor.sql delete mode 100644 synapse/storage/schema/delta/56/redaction_censor2.sql delete mode 100644 synapse/storage/schema/delta/56/redaction_censor3_fix_update.sql.postgres delete mode 100644 synapse/storage/schema/delta/56/room_membership_idx.sql delete mode 100644 synapse/storage/schema/delta/56/stats_separated.sql delete mode 100644 synapse/storage/schema/delta/56/unique_user_filter_index.py delete mode 100644 synapse/storage/schema/delta/56/user_external_ids.sql delete mode 100644 synapse/storage/schema/delta/56/users_in_public_rooms_idx.sql delete mode 100644 synapse/storage/schema/full_schemas/16/application_services.sql delete mode 100644 synapse/storage/schema/full_schemas/16/event_edges.sql delete mode 100644 synapse/storage/schema/full_schemas/16/event_signatures.sql delete mode 100644 synapse/storage/schema/full_schemas/16/im.sql delete mode 100644 synapse/storage/schema/full_schemas/16/keys.sql delete mode 100644 synapse/storage/schema/full_schemas/16/media_repository.sql delete mode 100644 synapse/storage/schema/full_schemas/16/presence.sql delete mode 100644 synapse/storage/schema/full_schemas/16/profiles.sql delete mode 100644 synapse/storage/schema/full_schemas/16/push.sql delete mode 100644 synapse/storage/schema/full_schemas/16/redactions.sql delete mode 100644 synapse/storage/schema/full_schemas/16/room_aliases.sql delete mode 100644 synapse/storage/schema/full_schemas/16/state.sql delete mode 100644 synapse/storage/schema/full_schemas/16/transactions.sql delete mode 100644 synapse/storage/schema/full_schemas/16/users.sql create mode 100644 synapse/storage/schema/full_schemas/54/full.sql delete mode 100644 synapse/storage/schema/full_schemas/54/full.sql.postgres delete mode 100644 synapse/storage/schema/full_schemas/54/full.sql.sqlite delete mode 100644 synapse/storage/schema/full_schemas/54/stream_positions.sql delete mode 100644 synapse/storage/schema/full_schemas/README.txt delete mode 100644 synapse/storage/search.py delete mode 100644 synapse/storage/signatures.py delete mode 100644 synapse/storage/state_deltas.py delete mode 100644 synapse/storage/stats.py delete mode 100644 synapse/storage/stream.py delete mode 100644 synapse/storage/tags.py delete mode 100644 synapse/storage/transactions.py delete mode 100644 synapse/storage/user_directory.py delete mode 100644 synapse/storage/user_erasure_store.py (limited to 'tests') diff --git a/synapse/app/event_creator.py b/synapse/app/event_creator.py index c67fe69a50..f20d810ece 100644 --- a/synapse/app/event_creator.py +++ b/synapse/app/event_creator.py @@ -56,8 +56,8 @@ from synapse.rest.client.v1.room import ( RoomStateEventRestServlet, ) from synapse.server import HomeServer +from synapse.storage.data_stores.main.user_directory import UserDirectoryStore from synapse.storage.engines import create_engine -from synapse.storage.user_directory import UserDirectoryStore from synapse.util.httpresourcetree import create_resource_tree from synapse.util.manhole import manhole from synapse.util.versionstring import get_version_string diff --git a/synapse/app/media_repository.py b/synapse/app/media_repository.py index 2ac783ffa3..6bc7202f33 100644 --- a/synapse/app/media_repository.py +++ b/synapse/app/media_repository.py @@ -39,8 +39,8 @@ from synapse.replication.tcp.client import ReplicationClientHandler from synapse.rest.admin import register_servlets_for_media_repo from synapse.rest.media.v0.content_repository import ContentRepoResource from synapse.server import HomeServer +from synapse.storage.data_stores.main.media_repository import MediaRepositoryStore from synapse.storage.engines import create_engine -from synapse.storage.media_repository import MediaRepositoryStore from synapse.util.httpresourcetree import create_resource_tree from synapse.util.manhole import manhole from synapse.util.versionstring import get_version_string diff --git a/synapse/app/synchrotron.py b/synapse/app/synchrotron.py index 473026fce5..6a7e2fa707 100644 --- a/synapse/app/synchrotron.py +++ b/synapse/app/synchrotron.py @@ -54,8 +54,8 @@ from synapse.rest.client.v1.initial_sync import InitialSyncRestServlet from synapse.rest.client.v1.room import RoomInitialSyncRestServlet from synapse.rest.client.v2_alpha import sync from synapse.server import HomeServer +from synapse.storage.data_stores.main.presence import UserPresenceState from synapse.storage.engines import create_engine -from synapse.storage.presence import UserPresenceState from synapse.util.httpresourcetree import create_resource_tree from synapse.util.manhole import manhole from synapse.util.stringutils import random_string diff --git a/synapse/app/user_dir.py b/synapse/app/user_dir.py index e01afb39f2..a5d6dc7915 100644 --- a/synapse/app/user_dir.py +++ b/synapse/app/user_dir.py @@ -42,8 +42,8 @@ from synapse.replication.tcp.streams.events import ( ) from synapse.rest.client.v2_alpha import user_directory from synapse.server import HomeServer +from synapse.storage.data_stores.main.user_directory import UserDirectoryStore from synapse.storage.engines import create_engine -from synapse.storage.user_directory import UserDirectoryStore from synapse.util.caches.stream_change_cache import StreamChangeCache from synapse.util.httpresourcetree import create_resource_tree from synapse.util.manhole import manhole diff --git a/synapse/federation/sender/per_destination_queue.py b/synapse/federation/sender/per_destination_queue.py index fad980b893..cc75c39476 100644 --- a/synapse/federation/sender/per_destination_queue.py +++ b/synapse/federation/sender/per_destination_queue.py @@ -30,7 +30,7 @@ from synapse.federation.units import Edu from synapse.handlers.presence import format_user_presence_state from synapse.metrics import sent_transactions_counter from synapse.metrics.background_process_metrics import run_as_background_process -from synapse.storage import UserPresenceState +from synapse.storage.presence import UserPresenceState from synapse.util.retryutils import NotRetryingDestination, get_retry_limiter # This is defined in the Matrix spec and enforced by the receiver. diff --git a/synapse/replication/slave/storage/account_data.py b/synapse/replication/slave/storage/account_data.py index 3c44d1d48d..bc2f6a12ae 100644 --- a/synapse/replication/slave/storage/account_data.py +++ b/synapse/replication/slave/storage/account_data.py @@ -16,8 +16,8 @@ from synapse.replication.slave.storage._base import BaseSlavedStore from synapse.replication.slave.storage._slaved_id_tracker import SlavedIdTracker -from synapse.storage.account_data import AccountDataWorkerStore -from synapse.storage.tags import TagsWorkerStore +from synapse.storage.data_stores.main.account_data import AccountDataWorkerStore +from synapse.storage.data_stores.main.tags import TagsWorkerStore class SlavedAccountDataStore(TagsWorkerStore, AccountDataWorkerStore, BaseSlavedStore): diff --git a/synapse/replication/slave/storage/appservice.py b/synapse/replication/slave/storage/appservice.py index cda12ea70d..a67fbeffb7 100644 --- a/synapse/replication/slave/storage/appservice.py +++ b/synapse/replication/slave/storage/appservice.py @@ -14,7 +14,7 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.storage.appservice import ( +from synapse.storage.data_stores.main.appservice import ( ApplicationServiceTransactionWorkerStore, ApplicationServiceWorkerStore, ) diff --git a/synapse/replication/slave/storage/client_ips.py b/synapse/replication/slave/storage/client_ips.py index 14ced32333..b4f58cea19 100644 --- a/synapse/replication/slave/storage/client_ips.py +++ b/synapse/replication/slave/storage/client_ips.py @@ -13,7 +13,7 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.storage.client_ips import LAST_SEEN_GRANULARITY +from synapse.storage.data_stores.main.client_ips import LAST_SEEN_GRANULARITY from synapse.util.caches import CACHE_SIZE_FACTOR from synapse.util.caches.descriptors import Cache diff --git a/synapse/replication/slave/storage/deviceinbox.py b/synapse/replication/slave/storage/deviceinbox.py index 284fd30d89..9fb6c5c6ff 100644 --- a/synapse/replication/slave/storage/deviceinbox.py +++ b/synapse/replication/slave/storage/deviceinbox.py @@ -15,7 +15,7 @@ from synapse.replication.slave.storage._base import BaseSlavedStore from synapse.replication.slave.storage._slaved_id_tracker import SlavedIdTracker -from synapse.storage.deviceinbox import DeviceInboxWorkerStore +from synapse.storage.data_stores.main.deviceinbox import DeviceInboxWorkerStore from synapse.util.caches.expiringcache import ExpiringCache from synapse.util.caches.stream_change_cache import StreamChangeCache diff --git a/synapse/replication/slave/storage/devices.py b/synapse/replication/slave/storage/devices.py index d9300fce33..f856c72d84 100644 --- a/synapse/replication/slave/storage/devices.py +++ b/synapse/replication/slave/storage/devices.py @@ -15,8 +15,8 @@ from synapse.replication.slave.storage._base import BaseSlavedStore from synapse.replication.slave.storage._slaved_id_tracker import SlavedIdTracker -from synapse.storage.devices import DeviceWorkerStore -from synapse.storage.end_to_end_keys import EndToEndKeyWorkerStore +from synapse.storage.data_stores.main.devices import DeviceWorkerStore +from synapse.storage.data_stores.main.end_to_end_keys import EndToEndKeyWorkerStore from synapse.util.caches.stream_change_cache import StreamChangeCache diff --git a/synapse/replication/slave/storage/directory.py b/synapse/replication/slave/storage/directory.py index 1d1d48709a..8b9717c46f 100644 --- a/synapse/replication/slave/storage/directory.py +++ b/synapse/replication/slave/storage/directory.py @@ -13,7 +13,7 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.storage.directory import DirectoryWorkerStore +from synapse.storage.data_stores.main.directory import DirectoryWorkerStore from ._base import BaseSlavedStore diff --git a/synapse/replication/slave/storage/events.py b/synapse/replication/slave/storage/events.py index ab5937e638..d0a0eaf75b 100644 --- a/synapse/replication/slave/storage/events.py +++ b/synapse/replication/slave/storage/events.py @@ -20,15 +20,17 @@ from synapse.replication.tcp.streams.events import ( EventsStreamCurrentStateRow, EventsStreamEventRow, ) -from synapse.storage.event_federation import EventFederationWorkerStore -from synapse.storage.event_push_actions import EventPushActionsWorkerStore -from synapse.storage.events_worker import EventsWorkerStore -from synapse.storage.relations import RelationsWorkerStore -from synapse.storage.roommember import RoomMemberWorkerStore -from synapse.storage.signatures import SignatureWorkerStore -from synapse.storage.state import StateGroupWorkerStore -from synapse.storage.stream import StreamWorkerStore -from synapse.storage.user_erasure_store import UserErasureWorkerStore +from synapse.storage.data_stores.main.event_federation import EventFederationWorkerStore +from synapse.storage.data_stores.main.event_push_actions import ( + EventPushActionsWorkerStore, +) +from synapse.storage.data_stores.main.events_worker import EventsWorkerStore +from synapse.storage.data_stores.main.relations import RelationsWorkerStore +from synapse.storage.data_stores.main.roommember import RoomMemberWorkerStore +from synapse.storage.data_stores.main.signatures import SignatureWorkerStore +from synapse.storage.data_stores.main.state import StateGroupWorkerStore +from synapse.storage.data_stores.main.stream import StreamWorkerStore +from synapse.storage.data_stores.main.user_erasure_store import UserErasureWorkerStore from ._base import BaseSlavedStore from ._slaved_id_tracker import SlavedIdTracker diff --git a/synapse/replication/slave/storage/filtering.py b/synapse/replication/slave/storage/filtering.py index 456a14cd5c..5c84ebd125 100644 --- a/synapse/replication/slave/storage/filtering.py +++ b/synapse/replication/slave/storage/filtering.py @@ -13,7 +13,7 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.storage.filtering import FilteringStore +from synapse.storage.data_stores.main.filtering import FilteringStore from ._base import BaseSlavedStore diff --git a/synapse/replication/slave/storage/keys.py b/synapse/replication/slave/storage/keys.py index cc6f7f009f..3def367ae9 100644 --- a/synapse/replication/slave/storage/keys.py +++ b/synapse/replication/slave/storage/keys.py @@ -13,7 +13,7 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.storage import KeyStore +from synapse.storage.data_stores.main.keys import KeyStore # KeyStore isn't really safe to use from a worker, but for now we do so and hope that # the races it creates aren't too bad. diff --git a/synapse/replication/slave/storage/presence.py b/synapse/replication/slave/storage/presence.py index 82d808af4c..747ced0c84 100644 --- a/synapse/replication/slave/storage/presence.py +++ b/synapse/replication/slave/storage/presence.py @@ -14,7 +14,7 @@ # limitations under the License. from synapse.storage import DataStore -from synapse.storage.presence import PresenceStore +from synapse.storage.data_stores.main.presence import PresenceStore from synapse.util.caches.stream_change_cache import StreamChangeCache from ._base import BaseSlavedStore, __func__ diff --git a/synapse/replication/slave/storage/profile.py b/synapse/replication/slave/storage/profile.py index 46c28d4171..28c508aad3 100644 --- a/synapse/replication/slave/storage/profile.py +++ b/synapse/replication/slave/storage/profile.py @@ -14,7 +14,7 @@ # limitations under the License. from synapse.replication.slave.storage._base import BaseSlavedStore -from synapse.storage.profile import ProfileWorkerStore +from synapse.storage.data_stores.main.profile import ProfileWorkerStore class SlavedProfileStore(ProfileWorkerStore, BaseSlavedStore): diff --git a/synapse/replication/slave/storage/push_rule.py b/synapse/replication/slave/storage/push_rule.py index af7012702e..3655f05e54 100644 --- a/synapse/replication/slave/storage/push_rule.py +++ b/synapse/replication/slave/storage/push_rule.py @@ -14,7 +14,7 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.storage.push_rule import PushRulesWorkerStore +from synapse.storage.data_stores.main.push_rule import PushRulesWorkerStore from ._slaved_id_tracker import SlavedIdTracker from .events import SlavedEventStore diff --git a/synapse/replication/slave/storage/pushers.py b/synapse/replication/slave/storage/pushers.py index 8eeb267d61..b4331d0799 100644 --- a/synapse/replication/slave/storage/pushers.py +++ b/synapse/replication/slave/storage/pushers.py @@ -14,7 +14,7 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.storage.pusher import PusherWorkerStore +from synapse.storage.data_stores.main.pusher import PusherWorkerStore from ._base import BaseSlavedStore from ._slaved_id_tracker import SlavedIdTracker diff --git a/synapse/replication/slave/storage/receipts.py b/synapse/replication/slave/storage/receipts.py index 91afa5a72b..43d823c601 100644 --- a/synapse/replication/slave/storage/receipts.py +++ b/synapse/replication/slave/storage/receipts.py @@ -14,7 +14,7 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.storage.receipts import ReceiptsWorkerStore +from synapse.storage.data_stores.main.receipts import ReceiptsWorkerStore from ._base import BaseSlavedStore from ._slaved_id_tracker import SlavedIdTracker diff --git a/synapse/replication/slave/storage/registration.py b/synapse/replication/slave/storage/registration.py index 408d91df1c..4b8553e250 100644 --- a/synapse/replication/slave/storage/registration.py +++ b/synapse/replication/slave/storage/registration.py @@ -13,7 +13,7 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.storage.registration import RegistrationWorkerStore +from synapse.storage.data_stores.main.registration import RegistrationWorkerStore from ._base import BaseSlavedStore diff --git a/synapse/replication/slave/storage/room.py b/synapse/replication/slave/storage/room.py index f68b3378e3..d9ad386b28 100644 --- a/synapse/replication/slave/storage/room.py +++ b/synapse/replication/slave/storage/room.py @@ -13,7 +13,7 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.storage.room import RoomWorkerStore +from synapse.storage.data_stores.main.room import RoomWorkerStore from ._base import BaseSlavedStore from ._slaved_id_tracker import SlavedIdTracker diff --git a/synapse/replication/slave/storage/transactions.py b/synapse/replication/slave/storage/transactions.py index 3527beb3c9..ac88e6b8c3 100644 --- a/synapse/replication/slave/storage/transactions.py +++ b/synapse/replication/slave/storage/transactions.py @@ -13,7 +13,7 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.storage.transactions import TransactionStore +from synapse.storage.data_stores.main.transactions import TransactionStore from ._base import BaseSlavedStore diff --git a/synapse/storage/__init__.py b/synapse/storage/__init__.py index e7f6ea7286..e42fba45a1 100644 --- a/synapse/storage/__init__.py +++ b/synapse/storage/__init__.py @@ -14,509 +14,7 @@ # See the License for the specific language governing permissions and # limitations under the License. -import calendar -import logging -import time - -from twisted.internet import defer - -from synapse.api.constants import PresenceState -from synapse.storage.devices import DeviceStore -from synapse.storage.user_erasure_store import UserErasureStore -from synapse.util.caches.stream_change_cache import StreamChangeCache - -from .account_data import AccountDataStore -from .appservice import ApplicationServiceStore, ApplicationServiceTransactionStore -from .client_ips import ClientIpStore -from .deviceinbox import DeviceInboxStore -from .directory import DirectoryStore -from .e2e_room_keys import EndToEndRoomKeyStore -from .end_to_end_keys import EndToEndKeyStore -from .engines import PostgresEngine -from .event_federation import EventFederationStore -from .event_push_actions import EventPushActionsStore -from .events import EventsStore -from .events_bg_updates import EventsBackgroundUpdatesStore -from .filtering import FilteringStore -from .group_server import GroupServerStore -from .keys import KeyStore -from .media_repository import MediaRepositoryStore -from .monthly_active_users import MonthlyActiveUsersStore -from .openid import OpenIdStore -from .presence import PresenceStore, UserPresenceState -from .profile import ProfileStore -from .push_rule import PushRuleStore -from .pusher import PusherStore -from .receipts import ReceiptsStore -from .registration import RegistrationStore -from .rejections import RejectionsStore -from .relations import RelationsStore -from .room import RoomStore -from .roommember import RoomMemberStore -from .search import SearchStore -from .signatures import SignatureStore -from .state import StateStore -from .stats import StatsStore -from .stream import StreamStore -from .tags import TagsStore -from .transactions import TransactionStore -from .user_directory import UserDirectoryStore -from .util.id_generators import ChainedIdGenerator, IdGenerator, StreamIdGenerator - -logger = logging.getLogger(__name__) - - -class DataStore( - EventsBackgroundUpdatesStore, - RoomMemberStore, - RoomStore, - RegistrationStore, - StreamStore, - ProfileStore, - PresenceStore, - TransactionStore, - DirectoryStore, - KeyStore, - StateStore, - SignatureStore, - ApplicationServiceStore, - EventsStore, - EventFederationStore, - MediaRepositoryStore, - RejectionsStore, - FilteringStore, - PusherStore, - PushRuleStore, - ApplicationServiceTransactionStore, - ReceiptsStore, - EndToEndKeyStore, - EndToEndRoomKeyStore, - SearchStore, - TagsStore, - AccountDataStore, - EventPushActionsStore, - OpenIdStore, - ClientIpStore, - DeviceStore, - DeviceInboxStore, - UserDirectoryStore, - GroupServerStore, - UserErasureStore, - MonthlyActiveUsersStore, - StatsStore, - RelationsStore, -): - def __init__(self, db_conn, hs): - self.hs = hs - self._clock = hs.get_clock() - self.database_engine = hs.database_engine - - self._stream_id_gen = StreamIdGenerator( - db_conn, - "events", - "stream_ordering", - extra_tables=[("local_invites", "stream_id")], - ) - self._backfill_id_gen = StreamIdGenerator( - db_conn, - "events", - "stream_ordering", - step=-1, - extra_tables=[("ex_outlier_stream", "event_stream_ordering")], - ) - self._presence_id_gen = StreamIdGenerator( - db_conn, "presence_stream", "stream_id" - ) - self._device_inbox_id_gen = StreamIdGenerator( - db_conn, "device_max_stream_id", "stream_id" - ) - self._public_room_id_gen = StreamIdGenerator( - db_conn, "public_room_list_stream", "stream_id" - ) - self._device_list_id_gen = StreamIdGenerator( - db_conn, "device_lists_stream", "stream_id" - ) - - self._access_tokens_id_gen = IdGenerator(db_conn, "access_tokens", "id") - self._event_reports_id_gen = IdGenerator(db_conn, "event_reports", "id") - self._push_rule_id_gen = IdGenerator(db_conn, "push_rules", "id") - self._push_rules_enable_id_gen = IdGenerator(db_conn, "push_rules_enable", "id") - self._push_rules_stream_id_gen = ChainedIdGenerator( - self._stream_id_gen, db_conn, "push_rules_stream", "stream_id" - ) - self._pushers_id_gen = StreamIdGenerator( - db_conn, "pushers", "id", extra_tables=[("deleted_pushers", "stream_id")] - ) - self._group_updates_id_gen = StreamIdGenerator( - db_conn, "local_group_updates", "stream_id" - ) - - if isinstance(self.database_engine, PostgresEngine): - self._cache_id_gen = StreamIdGenerator( - db_conn, "cache_invalidation_stream", "stream_id" - ) - else: - self._cache_id_gen = None - - self._presence_on_startup = self._get_active_presence(db_conn) - - presence_cache_prefill, min_presence_val = self._get_cache_dict( - db_conn, - "presence_stream", - entity_column="user_id", - stream_column="stream_id", - max_value=self._presence_id_gen.get_current_token(), - ) - self.presence_stream_cache = StreamChangeCache( - "PresenceStreamChangeCache", - min_presence_val, - prefilled_cache=presence_cache_prefill, - ) - - max_device_inbox_id = self._device_inbox_id_gen.get_current_token() - device_inbox_prefill, min_device_inbox_id = self._get_cache_dict( - db_conn, - "device_inbox", - entity_column="user_id", - stream_column="stream_id", - max_value=max_device_inbox_id, - limit=1000, - ) - self._device_inbox_stream_cache = StreamChangeCache( - "DeviceInboxStreamChangeCache", - min_device_inbox_id, - prefilled_cache=device_inbox_prefill, - ) - # The federation outbox and the local device inbox uses the same - # stream_id generator. - device_outbox_prefill, min_device_outbox_id = self._get_cache_dict( - db_conn, - "device_federation_outbox", - entity_column="destination", - stream_column="stream_id", - max_value=max_device_inbox_id, - limit=1000, - ) - self._device_federation_outbox_stream_cache = StreamChangeCache( - "DeviceFederationOutboxStreamChangeCache", - min_device_outbox_id, - prefilled_cache=device_outbox_prefill, - ) - - device_list_max = self._device_list_id_gen.get_current_token() - self._device_list_stream_cache = StreamChangeCache( - "DeviceListStreamChangeCache", device_list_max - ) - self._device_list_federation_stream_cache = StreamChangeCache( - "DeviceListFederationStreamChangeCache", device_list_max - ) - - events_max = self._stream_id_gen.get_current_token() - curr_state_delta_prefill, min_curr_state_delta_id = self._get_cache_dict( - db_conn, - "current_state_delta_stream", - entity_column="room_id", - stream_column="stream_id", - max_value=events_max, # As we share the stream id with events token - limit=1000, - ) - self._curr_state_delta_stream_cache = StreamChangeCache( - "_curr_state_delta_stream_cache", - min_curr_state_delta_id, - prefilled_cache=curr_state_delta_prefill, - ) - - _group_updates_prefill, min_group_updates_id = self._get_cache_dict( - db_conn, - "local_group_updates", - entity_column="user_id", - stream_column="stream_id", - max_value=self._group_updates_id_gen.get_current_token(), - limit=1000, - ) - self._group_updates_stream_cache = StreamChangeCache( - "_group_updates_stream_cache", - min_group_updates_id, - prefilled_cache=_group_updates_prefill, - ) - - self._stream_order_on_start = self.get_room_max_stream_ordering() - self._min_stream_order_on_start = self.get_room_min_stream_ordering() - - # Used in _generate_user_daily_visits to keep track of progress - self._last_user_visit_update = self._get_start_of_day() - - super(DataStore, self).__init__(db_conn, hs) - - def take_presence_startup_info(self): - active_on_startup = self._presence_on_startup - self._presence_on_startup = None - return active_on_startup - - def _get_active_presence(self, db_conn): - """Fetch non-offline presence from the database so that we can register - the appropriate time outs. - """ - - sql = ( - "SELECT user_id, state, last_active_ts, last_federation_update_ts," - " last_user_sync_ts, status_msg, currently_active FROM presence_stream" - " WHERE state != ?" - ) - sql = self.database_engine.convert_param_style(sql) - - txn = db_conn.cursor() - txn.execute(sql, (PresenceState.OFFLINE,)) - rows = self.cursor_to_dict(txn) - txn.close() - - for row in rows: - row["currently_active"] = bool(row["currently_active"]) - - return [UserPresenceState(**row) for row in rows] - - def count_daily_users(self): - """ - Counts the number of users who used this homeserver in the last 24 hours. - """ - yesterday = int(self._clock.time_msec()) - (1000 * 60 * 60 * 24) - return self.runInteraction("count_daily_users", self._count_users, yesterday) - - def count_monthly_users(self): - """ - Counts the number of users who used this homeserver in the last 30 days. - Note this method is intended for phonehome metrics only and is different - from the mau figure in synapse.storage.monthly_active_users which, - amongst other things, includes a 3 day grace period before a user counts. - """ - thirty_days_ago = int(self._clock.time_msec()) - (1000 * 60 * 60 * 24 * 30) - return self.runInteraction( - "count_monthly_users", self._count_users, thirty_days_ago - ) - - def _count_users(self, txn, time_from): - """ - Returns number of users seen in the past time_from period - """ - sql = """ - SELECT COALESCE(count(*), 0) FROM ( - SELECT user_id FROM user_ips - WHERE last_seen > ? - GROUP BY user_id - ) u - """ - txn.execute(sql, (time_from,)) - count, = txn.fetchone() - return count - - def count_r30_users(self): - """ - Counts the number of 30 day retained users, defined as:- - * Users who have created their accounts more than 30 days ago - * Where last seen at most 30 days ago - * Where account creation and last_seen are > 30 days apart - - Returns counts globaly for a given user as well as breaking - by platform - """ - - def _count_r30_users(txn): - thirty_days_in_secs = 86400 * 30 - now = int(self._clock.time()) - thirty_days_ago_in_secs = now - thirty_days_in_secs - - sql = """ - SELECT platform, COALESCE(count(*), 0) FROM ( - SELECT - users.name, platform, users.creation_ts * 1000, - MAX(uip.last_seen) - FROM users - INNER JOIN ( - SELECT - user_id, - last_seen, - CASE - WHEN user_agent LIKE '%%Android%%' THEN 'android' - WHEN user_agent LIKE '%%iOS%%' THEN 'ios' - WHEN user_agent LIKE '%%Electron%%' THEN 'electron' - WHEN user_agent LIKE '%%Mozilla%%' THEN 'web' - WHEN user_agent LIKE '%%Gecko%%' THEN 'web' - ELSE 'unknown' - END - AS platform - FROM user_ips - ) uip - ON users.name = uip.user_id - AND users.appservice_id is NULL - AND users.creation_ts < ? - AND uip.last_seen/1000 > ? - AND (uip.last_seen/1000) - users.creation_ts > 86400 * 30 - GROUP BY users.name, platform, users.creation_ts - ) u GROUP BY platform - """ - - results = {} - txn.execute(sql, (thirty_days_ago_in_secs, thirty_days_ago_in_secs)) - - for row in txn: - if row[0] == "unknown": - pass - results[row[0]] = row[1] - - sql = """ - SELECT COALESCE(count(*), 0) FROM ( - SELECT users.name, users.creation_ts * 1000, - MAX(uip.last_seen) - FROM users - INNER JOIN ( - SELECT - user_id, - last_seen - FROM user_ips - ) uip - ON users.name = uip.user_id - AND appservice_id is NULL - AND users.creation_ts < ? - AND uip.last_seen/1000 > ? - AND (uip.last_seen/1000) - users.creation_ts > 86400 * 30 - GROUP BY users.name, users.creation_ts - ) u - """ - - txn.execute(sql, (thirty_days_ago_in_secs, thirty_days_ago_in_secs)) - - count, = txn.fetchone() - results["all"] = count - - return results - - return self.runInteraction("count_r30_users", _count_r30_users) - - def _get_start_of_day(self): - """ - Returns millisecond unixtime for start of UTC day. - """ - now = time.gmtime() - today_start = calendar.timegm((now.tm_year, now.tm_mon, now.tm_mday, 0, 0, 0)) - return today_start * 1000 - - def generate_user_daily_visits(self): - """ - Generates daily visit data for use in cohort/ retention analysis - """ - - def _generate_user_daily_visits(txn): - logger.info("Calling _generate_user_daily_visits") - today_start = self._get_start_of_day() - a_day_in_milliseconds = 24 * 60 * 60 * 1000 - now = self.clock.time_msec() - - sql = """ - INSERT INTO user_daily_visits (user_id, device_id, timestamp) - SELECT u.user_id, u.device_id, ? - FROM user_ips AS u - LEFT JOIN ( - SELECT user_id, device_id, timestamp FROM user_daily_visits - WHERE timestamp = ? - ) udv - ON u.user_id = udv.user_id AND u.device_id=udv.device_id - INNER JOIN users ON users.name=u.user_id - WHERE last_seen > ? AND last_seen <= ? - AND udv.timestamp IS NULL AND users.is_guest=0 - AND users.appservice_id IS NULL - GROUP BY u.user_id, u.device_id - """ - - # This means that the day has rolled over but there could still - # be entries from the previous day. There is an edge case - # where if the user logs in at 23:59 and overwrites their - # last_seen at 00:01 then they will not be counted in the - # previous day's stats - it is important that the query is run - # often to minimise this case. - if today_start > self._last_user_visit_update: - yesterday_start = today_start - a_day_in_milliseconds - txn.execute( - sql, - ( - yesterday_start, - yesterday_start, - self._last_user_visit_update, - today_start, - ), - ) - self._last_user_visit_update = today_start - - txn.execute( - sql, (today_start, today_start, self._last_user_visit_update, now) - ) - # Update _last_user_visit_update to now. The reason to do this - # rather just clamping to the beginning of the day is to limit - # the size of the join - meaning that the query can be run more - # frequently - self._last_user_visit_update = now - - return self.runInteraction( - "generate_user_daily_visits", _generate_user_daily_visits - ) - - def get_users(self): - """Function to reterive a list of users in users table. - - Args: - Returns: - defer.Deferred: resolves to list[dict[str, Any]] - """ - return self._simple_select_list( - table="users", - keyvalues={}, - retcols=["name", "password_hash", "is_guest", "admin", "user_type"], - desc="get_users", - ) - - @defer.inlineCallbacks - def get_users_paginate(self, order, start, limit): - """Function to reterive a paginated list of users from - users list. This will return a json object, which contains - list of users and the total number of users in users table. - - Args: - order (str): column name to order the select by this column - start (int): start number to begin the query from - limit (int): number of rows to reterive - Returns: - defer.Deferred: resolves to json object {list[dict[str, Any]], count} - """ - users = yield self.runInteraction( - "get_users_paginate", - self._simple_select_list_paginate_txn, - table="users", - keyvalues={"is_guest": False}, - orderby=order, - start=start, - limit=limit, - retcols=["name", "password_hash", "is_guest", "admin", "user_type"], - ) - count = yield self.runInteraction("get_users_paginate", self.get_user_count_txn) - retval = {"users": users, "total": count} - return retval - - def search_users(self, term): - """Function to search users list for one or more users with - the matched term. - - Args: - term (str): search term - col (str): column to query term should be matched to - Returns: - defer.Deferred: resolves to list[dict[str, Any]] - """ - return self._simple_search_list( - table="users", - term=term, - col="name", - retcols=["name", "password_hash", "is_guest", "admin", "user_type"], - desc="search_users", - ) +from synapse.storage.data_stores.main import DataStore # noqa: F401 def are_all_users_on_domain(txn, database_engine, domain): diff --git a/synapse/storage/account_data.py b/synapse/storage/account_data.py deleted file mode 100644 index 6afbfc0d74..0000000000 --- a/synapse/storage/account_data.py +++ /dev/null @@ -1,391 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2014-2016 OpenMarket Ltd -# Copyright 2018 New Vector Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import abc -import logging - -from canonicaljson import json - -from twisted.internet import defer - -from synapse.storage._base import SQLBaseStore -from synapse.storage.util.id_generators import StreamIdGenerator -from synapse.util.caches.descriptors import cached, cachedInlineCallbacks -from synapse.util.caches.stream_change_cache import StreamChangeCache - -logger = logging.getLogger(__name__) - - -class AccountDataWorkerStore(SQLBaseStore): - """This is an abstract base class where subclasses must implement - `get_max_account_data_stream_id` which can be called in the initializer. - """ - - # This ABCMeta metaclass ensures that we cannot be instantiated without - # the abstract methods being implemented. - __metaclass__ = abc.ABCMeta - - def __init__(self, db_conn, hs): - account_max = self.get_max_account_data_stream_id() - self._account_data_stream_cache = StreamChangeCache( - "AccountDataAndTagsChangeCache", account_max - ) - - super(AccountDataWorkerStore, self).__init__(db_conn, hs) - - @abc.abstractmethod - def get_max_account_data_stream_id(self): - """Get the current max stream ID for account data stream - - Returns: - int - """ - raise NotImplementedError() - - @cached() - def get_account_data_for_user(self, user_id): - """Get all the client account_data for a user. - - Args: - user_id(str): The user to get the account_data for. - Returns: - A deferred pair of a dict of global account_data and a dict - mapping from room_id string to per room account_data dicts. - """ - - def get_account_data_for_user_txn(txn): - rows = self._simple_select_list_txn( - txn, - "account_data", - {"user_id": user_id}, - ["account_data_type", "content"], - ) - - global_account_data = { - row["account_data_type"]: json.loads(row["content"]) for row in rows - } - - rows = self._simple_select_list_txn( - txn, - "room_account_data", - {"user_id": user_id}, - ["room_id", "account_data_type", "content"], - ) - - by_room = {} - for row in rows: - room_data = by_room.setdefault(row["room_id"], {}) - room_data[row["account_data_type"]] = json.loads(row["content"]) - - return global_account_data, by_room - - return self.runInteraction( - "get_account_data_for_user", get_account_data_for_user_txn - ) - - @cachedInlineCallbacks(num_args=2, max_entries=5000) - def get_global_account_data_by_type_for_user(self, data_type, user_id): - """ - Returns: - Deferred: A dict - """ - result = yield self._simple_select_one_onecol( - table="account_data", - keyvalues={"user_id": user_id, "account_data_type": data_type}, - retcol="content", - desc="get_global_account_data_by_type_for_user", - allow_none=True, - ) - - if result: - return json.loads(result) - else: - return None - - @cached(num_args=2) - def get_account_data_for_room(self, user_id, room_id): - """Get all the client account_data for a user for a room. - - Args: - user_id(str): The user to get the account_data for. - room_id(str): The room to get the account_data for. - Returns: - A deferred dict of the room account_data - """ - - def get_account_data_for_room_txn(txn): - rows = self._simple_select_list_txn( - txn, - "room_account_data", - {"user_id": user_id, "room_id": room_id}, - ["account_data_type", "content"], - ) - - return { - row["account_data_type"]: json.loads(row["content"]) for row in rows - } - - return self.runInteraction( - "get_account_data_for_room", get_account_data_for_room_txn - ) - - @cached(num_args=3, max_entries=5000) - def get_account_data_for_room_and_type(self, user_id, room_id, account_data_type): - """Get the client account_data of given type for a user for a room. - - Args: - user_id(str): The user to get the account_data for. - room_id(str): The room to get the account_data for. - account_data_type (str): The account data type to get. - Returns: - A deferred of the room account_data for that type, or None if - there isn't any set. - """ - - def get_account_data_for_room_and_type_txn(txn): - content_json = self._simple_select_one_onecol_txn( - txn, - table="room_account_data", - keyvalues={ - "user_id": user_id, - "room_id": room_id, - "account_data_type": account_data_type, - }, - retcol="content", - allow_none=True, - ) - - return json.loads(content_json) if content_json else None - - return self.runInteraction( - "get_account_data_for_room_and_type", get_account_data_for_room_and_type_txn - ) - - def get_all_updated_account_data( - self, last_global_id, last_room_id, current_id, limit - ): - """Get all the client account_data that has changed on the server - Args: - last_global_id(int): The position to fetch from for top level data - last_room_id(int): The position to fetch from for per room data - current_id(int): The position to fetch up to. - Returns: - A deferred pair of lists of tuples of stream_id int, user_id string, - room_id string, type string, and content string. - """ - if last_room_id == current_id and last_global_id == current_id: - return defer.succeed(([], [])) - - def get_updated_account_data_txn(txn): - sql = ( - "SELECT stream_id, user_id, account_data_type, content" - " FROM account_data WHERE ? < stream_id AND stream_id <= ?" - " ORDER BY stream_id ASC LIMIT ?" - ) - txn.execute(sql, (last_global_id, current_id, limit)) - global_results = txn.fetchall() - - sql = ( - "SELECT stream_id, user_id, room_id, account_data_type, content" - " FROM room_account_data WHERE ? < stream_id AND stream_id <= ?" - " ORDER BY stream_id ASC LIMIT ?" - ) - txn.execute(sql, (last_room_id, current_id, limit)) - room_results = txn.fetchall() - return global_results, room_results - - return self.runInteraction( - "get_all_updated_account_data_txn", get_updated_account_data_txn - ) - - def get_updated_account_data_for_user(self, user_id, stream_id): - """Get all the client account_data for a that's changed for a user - - Args: - user_id(str): The user to get the account_data for. - stream_id(int): The point in the stream since which to get updates - Returns: - A deferred pair of a dict of global account_data and a dict - mapping from room_id string to per room account_data dicts. - """ - - def get_updated_account_data_for_user_txn(txn): - sql = ( - "SELECT account_data_type, content FROM account_data" - " WHERE user_id = ? AND stream_id > ?" - ) - - txn.execute(sql, (user_id, stream_id)) - - global_account_data = {row[0]: json.loads(row[1]) for row in txn} - - sql = ( - "SELECT room_id, account_data_type, content FROM room_account_data" - " WHERE user_id = ? AND stream_id > ?" - ) - - txn.execute(sql, (user_id, stream_id)) - - account_data_by_room = {} - for row in txn: - room_account_data = account_data_by_room.setdefault(row[0], {}) - room_account_data[row[1]] = json.loads(row[2]) - - return global_account_data, account_data_by_room - - changed = self._account_data_stream_cache.has_entity_changed( - user_id, int(stream_id) - ) - if not changed: - return {}, {} - - return self.runInteraction( - "get_updated_account_data_for_user", get_updated_account_data_for_user_txn - ) - - @cachedInlineCallbacks(num_args=2, cache_context=True, max_entries=5000) - def is_ignored_by(self, ignored_user_id, ignorer_user_id, cache_context): - ignored_account_data = yield self.get_global_account_data_by_type_for_user( - "m.ignored_user_list", - ignorer_user_id, - on_invalidate=cache_context.invalidate, - ) - if not ignored_account_data: - return False - - return ignored_user_id in ignored_account_data.get("ignored_users", {}) - - -class AccountDataStore(AccountDataWorkerStore): - def __init__(self, db_conn, hs): - self._account_data_id_gen = StreamIdGenerator( - db_conn, "account_data_max_stream_id", "stream_id" - ) - - super(AccountDataStore, self).__init__(db_conn, hs) - - def get_max_account_data_stream_id(self): - """Get the current max stream id for the private user data stream - - Returns: - A deferred int. - """ - return self._account_data_id_gen.get_current_token() - - @defer.inlineCallbacks - def add_account_data_to_room(self, user_id, room_id, account_data_type, content): - """Add some account_data to a room for a user. - Args: - user_id(str): The user to add a tag for. - room_id(str): The room to add a tag for. - account_data_type(str): The type of account_data to add. - content(dict): A json object to associate with the tag. - Returns: - A deferred that completes once the account_data has been added. - """ - content_json = json.dumps(content) - - with self._account_data_id_gen.get_next() as next_id: - # no need to lock here as room_account_data has a unique constraint - # on (user_id, room_id, account_data_type) so _simple_upsert will - # retry if there is a conflict. - yield self._simple_upsert( - desc="add_room_account_data", - table="room_account_data", - keyvalues={ - "user_id": user_id, - "room_id": room_id, - "account_data_type": account_data_type, - }, - values={"stream_id": next_id, "content": content_json}, - lock=False, - ) - - # it's theoretically possible for the above to succeed and the - # below to fail - in which case we might reuse a stream id on - # restart, and the above update might not get propagated. That - # doesn't sound any worse than the whole update getting lost, - # which is what would happen if we combined the two into one - # transaction. - yield self._update_max_stream_id(next_id) - - self._account_data_stream_cache.entity_has_changed(user_id, next_id) - self.get_account_data_for_user.invalidate((user_id,)) - self.get_account_data_for_room.invalidate((user_id, room_id)) - self.get_account_data_for_room_and_type.prefill( - (user_id, room_id, account_data_type), content - ) - - result = self._account_data_id_gen.get_current_token() - return result - - @defer.inlineCallbacks - def add_account_data_for_user(self, user_id, account_data_type, content): - """Add some account_data to a room for a user. - Args: - user_id(str): The user to add a tag for. - account_data_type(str): The type of account_data to add. - content(dict): A json object to associate with the tag. - Returns: - A deferred that completes once the account_data has been added. - """ - content_json = json.dumps(content) - - with self._account_data_id_gen.get_next() as next_id: - # no need to lock here as account_data has a unique constraint on - # (user_id, account_data_type) so _simple_upsert will retry if - # there is a conflict. - yield self._simple_upsert( - desc="add_user_account_data", - table="account_data", - keyvalues={"user_id": user_id, "account_data_type": account_data_type}, - values={"stream_id": next_id, "content": content_json}, - lock=False, - ) - - # it's theoretically possible for the above to succeed and the - # below to fail - in which case we might reuse a stream id on - # restart, and the above update might not get propagated. That - # doesn't sound any worse than the whole update getting lost, - # which is what would happen if we combined the two into one - # transaction. - yield self._update_max_stream_id(next_id) - - self._account_data_stream_cache.entity_has_changed(user_id, next_id) - self.get_account_data_for_user.invalidate((user_id,)) - self.get_global_account_data_by_type_for_user.invalidate( - (account_data_type, user_id) - ) - - result = self._account_data_id_gen.get_current_token() - return result - - def _update_max_stream_id(self, next_id): - """Update the max stream_id - - Args: - next_id(int): The the revision to advance to. - """ - - def _update(txn): - update_max_id_sql = ( - "UPDATE account_data_max_stream_id" - " SET stream_id = ?" - " WHERE stream_id < ?" - ) - txn.execute(update_max_id_sql, (next_id, next_id)) - - return self.runInteraction("update_account_data_max_stream_id", _update) diff --git a/synapse/storage/appservice.py b/synapse/storage/appservice.py deleted file mode 100644 index 435b2acd4d..0000000000 --- a/synapse/storage/appservice.py +++ /dev/null @@ -1,368 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2015, 2016 OpenMarket Ltd -# Copyright 2018 New Vector Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. -import logging -import re - -from canonicaljson import json - -from twisted.internet import defer - -from synapse.appservice import AppServiceTransaction -from synapse.config.appservice import load_appservices -from synapse.storage.events_worker import EventsWorkerStore - -from ._base import SQLBaseStore - -logger = logging.getLogger(__name__) - - -def _make_exclusive_regex(services_cache): - # We precompie a regex constructed from all the regexes that the AS's - # have registered for exclusive users. - exclusive_user_regexes = [ - regex.pattern - for service in services_cache - for regex in service.get_exlusive_user_regexes() - ] - if exclusive_user_regexes: - exclusive_user_regex = "|".join("(" + r + ")" for r in exclusive_user_regexes) - exclusive_user_regex = re.compile(exclusive_user_regex) - else: - # We handle this case specially otherwise the constructed regex - # will always match - exclusive_user_regex = None - - return exclusive_user_regex - - -class ApplicationServiceWorkerStore(SQLBaseStore): - def __init__(self, db_conn, hs): - self.services_cache = load_appservices( - hs.hostname, hs.config.app_service_config_files - ) - self.exclusive_user_regex = _make_exclusive_regex(self.services_cache) - - super(ApplicationServiceWorkerStore, self).__init__(db_conn, hs) - - def get_app_services(self): - return self.services_cache - - def get_if_app_services_interested_in_user(self, user_id): - """Check if the user is one associated with an app service (exclusively) - """ - if self.exclusive_user_regex: - return bool(self.exclusive_user_regex.match(user_id)) - else: - return False - - def get_app_service_by_user_id(self, user_id): - """Retrieve an application service from their user ID. - - All application services have associated with them a particular user ID. - There is no distinguishing feature on the user ID which indicates it - represents an application service. This function allows you to map from - a user ID to an application service. - - Args: - user_id(str): The user ID to see if it is an application service. - Returns: - synapse.appservice.ApplicationService or None. - """ - for service in self.services_cache: - if service.sender == user_id: - return service - return None - - def get_app_service_by_token(self, token): - """Get the application service with the given appservice token. - - Args: - token (str): The application service token. - Returns: - synapse.appservice.ApplicationService or None. - """ - for service in self.services_cache: - if service.token == token: - return service - return None - - def get_app_service_by_id(self, as_id): - """Get the application service with the given appservice ID. - - Args: - as_id (str): The application service ID. - Returns: - synapse.appservice.ApplicationService or None. - """ - for service in self.services_cache: - if service.id == as_id: - return service - return None - - -class ApplicationServiceStore(ApplicationServiceWorkerStore): - # This is currently empty due to there not being any AS storage functions - # that can't be run on the workers. Since this may change in future, and - # to keep consistency with the other stores, we keep this empty class for - # now. - pass - - -class ApplicationServiceTransactionWorkerStore( - ApplicationServiceWorkerStore, EventsWorkerStore -): - @defer.inlineCallbacks - def get_appservices_by_state(self, state): - """Get a list of application services based on their state. - - Args: - state(ApplicationServiceState): The state to filter on. - Returns: - A Deferred which resolves to a list of ApplicationServices, which - may be empty. - """ - results = yield self._simple_select_list( - "application_services_state", dict(state=state), ["as_id"] - ) - # NB: This assumes this class is linked with ApplicationServiceStore - as_list = self.get_app_services() - services = [] - - for res in results: - for service in as_list: - if service.id == res["as_id"]: - services.append(service) - return services - - @defer.inlineCallbacks - def get_appservice_state(self, service): - """Get the application service state. - - Args: - service(ApplicationService): The service whose state to set. - Returns: - A Deferred which resolves to ApplicationServiceState. - """ - result = yield self._simple_select_one( - "application_services_state", - dict(as_id=service.id), - ["state"], - allow_none=True, - desc="get_appservice_state", - ) - if result: - return result.get("state") - return None - - def set_appservice_state(self, service, state): - """Set the application service state. - - Args: - service(ApplicationService): The service whose state to set. - state(ApplicationServiceState): The connectivity state to apply. - Returns: - A Deferred which resolves when the state was set successfully. - """ - return self._simple_upsert( - "application_services_state", dict(as_id=service.id), dict(state=state) - ) - - def create_appservice_txn(self, service, events): - """Atomically creates a new transaction for this application service - with the given list of events. - - Args: - service(ApplicationService): The service who the transaction is for. - events(list): A list of events to put in the transaction. - Returns: - AppServiceTransaction: A new transaction. - """ - - def _create_appservice_txn(txn): - # work out new txn id (highest txn id for this service += 1) - # The highest id may be the last one sent (in which case it is last_txn) - # or it may be the highest in the txns list (which are waiting to be/are - # being sent) - last_txn_id = self._get_last_txn(txn, service.id) - - txn.execute( - "SELECT MAX(txn_id) FROM application_services_txns WHERE as_id=?", - (service.id,), - ) - highest_txn_id = txn.fetchone()[0] - if highest_txn_id is None: - highest_txn_id = 0 - - new_txn_id = max(highest_txn_id, last_txn_id) + 1 - - # Insert new txn into txn table - event_ids = json.dumps([e.event_id for e in events]) - txn.execute( - "INSERT INTO application_services_txns(as_id, txn_id, event_ids) " - "VALUES(?,?,?)", - (service.id, new_txn_id, event_ids), - ) - return AppServiceTransaction(service=service, id=new_txn_id, events=events) - - return self.runInteraction("create_appservice_txn", _create_appservice_txn) - - def complete_appservice_txn(self, txn_id, service): - """Completes an application service transaction. - - Args: - txn_id(str): The transaction ID being completed. - service(ApplicationService): The application service which was sent - this transaction. - Returns: - A Deferred which resolves if this transaction was stored - successfully. - """ - txn_id = int(txn_id) - - def _complete_appservice_txn(txn): - # Debugging query: Make sure the txn being completed is EXACTLY +1 from - # what was there before. If it isn't, we've got problems (e.g. the AS - # has probably missed some events), so whine loudly but still continue, - # since it shouldn't fail completion of the transaction. - last_txn_id = self._get_last_txn(txn, service.id) - if (last_txn_id + 1) != txn_id: - logger.error( - "appservice: Completing a transaction which has an ID > 1 from " - "the last ID sent to this AS. We've either dropped events or " - "sent it to the AS out of order. FIX ME. last_txn=%s " - "completing_txn=%s service_id=%s", - last_txn_id, - txn_id, - service.id, - ) - - # Set current txn_id for AS to 'txn_id' - self._simple_upsert_txn( - txn, - "application_services_state", - dict(as_id=service.id), - dict(last_txn=txn_id), - ) - - # Delete txn - self._simple_delete_txn( - txn, "application_services_txns", dict(txn_id=txn_id, as_id=service.id) - ) - - return self.runInteraction("complete_appservice_txn", _complete_appservice_txn) - - @defer.inlineCallbacks - def get_oldest_unsent_txn(self, service): - """Get the oldest transaction which has not been sent for this - service. - - Args: - service(ApplicationService): The app service to get the oldest txn. - Returns: - A Deferred which resolves to an AppServiceTransaction or - None. - """ - - def _get_oldest_unsent_txn(txn): - # Monotonically increasing txn ids, so just select the smallest - # one in the txns table (we delete them when they are sent) - txn.execute( - "SELECT * FROM application_services_txns WHERE as_id=?" - " ORDER BY txn_id ASC LIMIT 1", - (service.id,), - ) - rows = self.cursor_to_dict(txn) - if not rows: - return None - - entry = rows[0] - - return entry - - entry = yield self.runInteraction( - "get_oldest_unsent_appservice_txn", _get_oldest_unsent_txn - ) - - if not entry: - return None - - event_ids = json.loads(entry["event_ids"]) - - events = yield self.get_events_as_list(event_ids) - - return AppServiceTransaction(service=service, id=entry["txn_id"], events=events) - - def _get_last_txn(self, txn, service_id): - txn.execute( - "SELECT last_txn FROM application_services_state WHERE as_id=?", - (service_id,), - ) - last_txn_id = txn.fetchone() - if last_txn_id is None or last_txn_id[0] is None: # no row exists - return 0 - else: - return int(last_txn_id[0]) # select 'last_txn' col - - def set_appservice_last_pos(self, pos): - def set_appservice_last_pos_txn(txn): - txn.execute( - "UPDATE appservice_stream_position SET stream_ordering = ?", (pos,) - ) - - return self.runInteraction( - "set_appservice_last_pos", set_appservice_last_pos_txn - ) - - @defer.inlineCallbacks - def get_new_events_for_appservice(self, current_id, limit): - """Get all new evnets""" - - def get_new_events_for_appservice_txn(txn): - sql = ( - "SELECT e.stream_ordering, e.event_id" - " FROM events AS e" - " WHERE" - " (SELECT stream_ordering FROM appservice_stream_position)" - " < e.stream_ordering" - " AND e.stream_ordering <= ?" - " ORDER BY e.stream_ordering ASC" - " LIMIT ?" - ) - - txn.execute(sql, (current_id, limit)) - rows = txn.fetchall() - - upper_bound = current_id - if len(rows) == limit: - upper_bound = rows[-1][0] - - return upper_bound, [row[1] for row in rows] - - upper_bound, event_ids = yield self.runInteraction( - "get_new_events_for_appservice", get_new_events_for_appservice_txn - ) - - events = yield self.get_events_as_list(event_ids) - - return upper_bound, events - - -class ApplicationServiceTransactionStore(ApplicationServiceTransactionWorkerStore): - # This is currently empty due to there not being any AS storage functions - # that can't be run on the workers. Since this may change in future, and - # to keep consistency with the other stores, we keep this empty class for - # now. - pass diff --git a/synapse/storage/client_ips.py b/synapse/storage/client_ips.py deleted file mode 100644 index 067820a5da..0000000000 --- a/synapse/storage/client_ips.py +++ /dev/null @@ -1,581 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import logging - -from six import iteritems - -from twisted.internet import defer - -from synapse.metrics.background_process_metrics import wrap_as_background_process -from synapse.util.caches import CACHE_SIZE_FACTOR - -from . import background_updates -from ._base import Cache - -logger = logging.getLogger(__name__) - -# Number of msec of granularity to store the user IP 'last seen' time. Smaller -# times give more inserts into the database even for readonly API hits -# 120 seconds == 2 minutes -LAST_SEEN_GRANULARITY = 120 * 1000 - - -class ClientIpBackgroundUpdateStore(background_updates.BackgroundUpdateStore): - def __init__(self, db_conn, hs): - super(ClientIpBackgroundUpdateStore, self).__init__(db_conn, hs) - - self.register_background_index_update( - "user_ips_device_index", - index_name="user_ips_device_id", - table="user_ips", - columns=["user_id", "device_id", "last_seen"], - ) - - self.register_background_index_update( - "user_ips_last_seen_index", - index_name="user_ips_last_seen", - table="user_ips", - columns=["user_id", "last_seen"], - ) - - self.register_background_index_update( - "user_ips_last_seen_only_index", - index_name="user_ips_last_seen_only", - table="user_ips", - columns=["last_seen"], - ) - - self.register_background_update_handler( - "user_ips_analyze", self._analyze_user_ip - ) - - self.register_background_update_handler( - "user_ips_remove_dupes", self._remove_user_ip_dupes - ) - - # Register a unique index - self.register_background_index_update( - "user_ips_device_unique_index", - index_name="user_ips_user_token_ip_unique_index", - table="user_ips", - columns=["user_id", "access_token", "ip"], - unique=True, - ) - - # Drop the old non-unique index - self.register_background_update_handler( - "user_ips_drop_nonunique_index", self._remove_user_ip_nonunique - ) - - # Update the last seen info in devices. - self.register_background_update_handler( - "devices_last_seen", self._devices_last_seen_update - ) - - @defer.inlineCallbacks - def _remove_user_ip_nonunique(self, progress, batch_size): - def f(conn): - txn = conn.cursor() - txn.execute("DROP INDEX IF EXISTS user_ips_user_ip") - txn.close() - - yield self.runWithConnection(f) - yield self._end_background_update("user_ips_drop_nonunique_index") - return 1 - - @defer.inlineCallbacks - def _analyze_user_ip(self, progress, batch_size): - # Background update to analyze user_ips table before we run the - # deduplication background update. The table may not have been analyzed - # for ages due to the table locks. - # - # This will lock out the naive upserts to user_ips while it happens, but - # the analyze should be quick (28GB table takes ~10s) - def user_ips_analyze(txn): - txn.execute("ANALYZE user_ips") - - yield self.runInteraction("user_ips_analyze", user_ips_analyze) - - yield self._end_background_update("user_ips_analyze") - - return 1 - - @defer.inlineCallbacks - def _remove_user_ip_dupes(self, progress, batch_size): - # This works function works by scanning the user_ips table in batches - # based on `last_seen`. For each row in a batch it searches the rest of - # the table to see if there are any duplicates, if there are then they - # are removed and replaced with a suitable row. - - # Fetch the start of the batch - begin_last_seen = progress.get("last_seen", 0) - - def get_last_seen(txn): - txn.execute( - """ - SELECT last_seen FROM user_ips - WHERE last_seen > ? - ORDER BY last_seen - LIMIT 1 - OFFSET ? - """, - (begin_last_seen, batch_size), - ) - row = txn.fetchone() - if row: - return row[0] - else: - return None - - # Get a last seen that has roughly `batch_size` since `begin_last_seen` - end_last_seen = yield self.runInteraction( - "user_ips_dups_get_last_seen", get_last_seen - ) - - # If it returns None, then we're processing the last batch - last = end_last_seen is None - - logger.info( - "Scanning for duplicate 'user_ips' rows in range: %s <= last_seen < %s", - begin_last_seen, - end_last_seen, - ) - - def remove(txn): - # This works by looking at all entries in the given time span, and - # then for each (user_id, access_token, ip) tuple in that range - # checking for any duplicates in the rest of the table (via a join). - # It then only returns entries which have duplicates, and the max - # last_seen across all duplicates, which can the be used to delete - # all other duplicates. - # It is efficient due to the existence of (user_id, access_token, - # ip) and (last_seen) indices. - - # Define the search space, which requires handling the last batch in - # a different way - if last: - clause = "? <= last_seen" - args = (begin_last_seen,) - else: - clause = "? <= last_seen AND last_seen < ?" - args = (begin_last_seen, end_last_seen) - - # (Note: The DISTINCT in the inner query is important to ensure that - # the COUNT(*) is accurate, otherwise double counting may happen due - # to the join effectively being a cross product) - txn.execute( - """ - SELECT user_id, access_token, ip, - MAX(device_id), MAX(user_agent), MAX(last_seen), - COUNT(*) - FROM ( - SELECT DISTINCT user_id, access_token, ip - FROM user_ips - WHERE {} - ) c - INNER JOIN user_ips USING (user_id, access_token, ip) - GROUP BY user_id, access_token, ip - HAVING count(*) > 1 - """.format( - clause - ), - args, - ) - res = txn.fetchall() - - # We've got some duplicates - for i in res: - user_id, access_token, ip, device_id, user_agent, last_seen, count = i - - # We want to delete the duplicates so we end up with only a - # single row. - # - # The naive way of doing this would be just to delete all rows - # and reinsert a constructed row. However, if there are a lot of - # duplicate rows this can cause the table to grow a lot, which - # can be problematic in two ways: - # 1. If user_ips is already large then this can cause the - # table to rapidly grow, potentially filling the disk. - # 2. Reinserting a lot of rows can confuse the table - # statistics for postgres, causing it to not use the - # correct indices for the query above, resulting in a full - # table scan. This is incredibly slow for large tables and - # can kill database performance. (This seems to mainly - # happen for the last query where the clause is simply `? < - # last_seen`) - # - # So instead we want to delete all but *one* of the duplicate - # rows. That is hard to do reliably, so we cheat and do a two - # step process: - # 1. Delete all rows with a last_seen strictly less than the - # max last_seen. This hopefully results in deleting all but - # one row the majority of the time, but there may be - # duplicate last_seen - # 2. If multiple rows remain, we fall back to the naive method - # and simply delete all rows and reinsert. - # - # Note that this relies on no new duplicate rows being inserted, - # but if that is happening then this entire process is futile - # anyway. - - # Do step 1: - - txn.execute( - """ - DELETE FROM user_ips - WHERE user_id = ? AND access_token = ? AND ip = ? AND last_seen < ? - """, - (user_id, access_token, ip, last_seen), - ) - if txn.rowcount == count - 1: - # We deleted all but one of the duplicate rows, i.e. there - # is exactly one remaining and so there is nothing left to - # do. - continue - elif txn.rowcount >= count: - raise Exception( - "We deleted more duplicate rows from 'user_ips' than expected" - ) - - # The previous step didn't delete enough rows, so we fallback to - # step 2: - - # Drop all the duplicates - txn.execute( - """ - DELETE FROM user_ips - WHERE user_id = ? AND access_token = ? AND ip = ? - """, - (user_id, access_token, ip), - ) - - # Add in one to be the last_seen - txn.execute( - """ - INSERT INTO user_ips - (user_id, access_token, ip, device_id, user_agent, last_seen) - VALUES (?, ?, ?, ?, ?, ?) - """, - (user_id, access_token, ip, device_id, user_agent, last_seen), - ) - - self._background_update_progress_txn( - txn, "user_ips_remove_dupes", {"last_seen": end_last_seen} - ) - - yield self.runInteraction("user_ips_dups_remove", remove) - - if last: - yield self._end_background_update("user_ips_remove_dupes") - - return batch_size - - @defer.inlineCallbacks - def _devices_last_seen_update(self, progress, batch_size): - """Background update to insert last seen info into devices table - """ - - last_user_id = progress.get("last_user_id", "") - last_device_id = progress.get("last_device_id", "") - - def _devices_last_seen_update_txn(txn): - # This consists of two queries: - # - # 1. The sub-query searches for the next N devices and joins - # against user_ips to find the max last_seen associated with - # that device. - # 2. The outer query then joins again against user_ips on - # user/device/last_seen. This *should* hopefully only - # return one row, but if it does return more than one then - # we'll just end up updating the same device row multiple - # times, which is fine. - - if self.database_engine.supports_tuple_comparison: - where_clause = "(user_id, device_id) > (?, ?)" - where_args = [last_user_id, last_device_id] - else: - # We explicitly do a `user_id >= ? AND (...)` here to ensure - # that an index is used, as doing `user_id > ? OR (user_id = ? AND ...)` - # makes it hard for query optimiser to tell that it can use the - # index on user_id - where_clause = "user_id >= ? AND (user_id > ? OR device_id > ?)" - where_args = [last_user_id, last_user_id, last_device_id] - - sql = """ - SELECT - last_seen, ip, user_agent, user_id, device_id - FROM ( - SELECT - user_id, device_id, MAX(u.last_seen) AS last_seen - FROM devices - INNER JOIN user_ips AS u USING (user_id, device_id) - WHERE %(where_clause)s - GROUP BY user_id, device_id - ORDER BY user_id ASC, device_id ASC - LIMIT ? - ) c - INNER JOIN user_ips AS u USING (user_id, device_id, last_seen) - """ % { - "where_clause": where_clause - } - txn.execute(sql, where_args + [batch_size]) - - rows = txn.fetchall() - if not rows: - return 0 - - sql = """ - UPDATE devices - SET last_seen = ?, ip = ?, user_agent = ? - WHERE user_id = ? AND device_id = ? - """ - txn.execute_batch(sql, rows) - - _, _, _, user_id, device_id = rows[-1] - self._background_update_progress_txn( - txn, - "devices_last_seen", - {"last_user_id": user_id, "last_device_id": device_id}, - ) - - return len(rows) - - updated = yield self.runInteraction( - "_devices_last_seen_update", _devices_last_seen_update_txn - ) - - if not updated: - yield self._end_background_update("devices_last_seen") - - return updated - - -class ClientIpStore(ClientIpBackgroundUpdateStore): - def __init__(self, db_conn, hs): - - self.client_ip_last_seen = Cache( - name="client_ip_last_seen", keylen=4, max_entries=50000 * CACHE_SIZE_FACTOR - ) - - super(ClientIpStore, self).__init__(db_conn, hs) - - self.user_ips_max_age = hs.config.user_ips_max_age - - # (user_id, access_token, ip,) -> (user_agent, device_id, last_seen) - self._batch_row_update = {} - - self._client_ip_looper = self._clock.looping_call( - self._update_client_ips_batch, 5 * 1000 - ) - self.hs.get_reactor().addSystemEventTrigger( - "before", "shutdown", self._update_client_ips_batch - ) - - if self.user_ips_max_age: - self._clock.looping_call(self._prune_old_user_ips, 5 * 1000) - - @defer.inlineCallbacks - def insert_client_ip( - self, user_id, access_token, ip, user_agent, device_id, now=None - ): - if not now: - now = int(self._clock.time_msec()) - key = (user_id, access_token, ip) - - try: - last_seen = self.client_ip_last_seen.get(key) - except KeyError: - last_seen = None - yield self.populate_monthly_active_users(user_id) - # Rate-limited inserts - if last_seen is not None and (now - last_seen) < LAST_SEEN_GRANULARITY: - return - - self.client_ip_last_seen.prefill(key, now) - - self._batch_row_update[key] = (user_agent, device_id, now) - - @wrap_as_background_process("update_client_ips") - def _update_client_ips_batch(self): - - # If the DB pool has already terminated, don't try updating - if not self.hs.get_db_pool().running: - return - - to_update = self._batch_row_update - self._batch_row_update = {} - - return self.runInteraction( - "_update_client_ips_batch", self._update_client_ips_batch_txn, to_update - ) - - def _update_client_ips_batch_txn(self, txn, to_update): - if "user_ips" in self._unsafe_to_upsert_tables or ( - not self.database_engine.can_native_upsert - ): - self.database_engine.lock_table(txn, "user_ips") - - for entry in iteritems(to_update): - (user_id, access_token, ip), (user_agent, device_id, last_seen) = entry - - try: - self._simple_upsert_txn( - txn, - table="user_ips", - keyvalues={ - "user_id": user_id, - "access_token": access_token, - "ip": ip, - }, - values={ - "user_agent": user_agent, - "device_id": device_id, - "last_seen": last_seen, - }, - lock=False, - ) - - # Technically an access token might not be associated with - # a device so we need to check. - if device_id: - self._simple_upsert_txn( - txn, - table="devices", - keyvalues={"user_id": user_id, "device_id": device_id}, - values={ - "user_agent": user_agent, - "last_seen": last_seen, - "ip": ip, - }, - lock=False, - ) - except Exception as e: - # Failed to upsert, log and continue - logger.error("Failed to insert client IP %r: %r", entry, e) - - @defer.inlineCallbacks - def get_last_client_ip_by_device(self, user_id, device_id): - """For each device_id listed, give the user_ip it was last seen on - - Args: - user_id (str) - device_id (str): If None fetches all devices for the user - - Returns: - defer.Deferred: resolves to a dict, where the keys - are (user_id, device_id) tuples. The values are also dicts, with - keys giving the column names - """ - - keyvalues = {"user_id": user_id} - if device_id is not None: - keyvalues["device_id"] = device_id - - res = yield self._simple_select_list( - table="devices", - keyvalues=keyvalues, - retcols=("user_id", "ip", "user_agent", "device_id", "last_seen"), - ) - - ret = {(d["user_id"], d["device_id"]): d for d in res} - for key in self._batch_row_update: - uid, access_token, ip = key - if uid == user_id: - user_agent, did, last_seen = self._batch_row_update[key] - if not device_id or did == device_id: - ret[(user_id, device_id)] = { - "user_id": user_id, - "access_token": access_token, - "ip": ip, - "user_agent": user_agent, - "device_id": did, - "last_seen": last_seen, - } - return ret - - @defer.inlineCallbacks - def get_user_ip_and_agents(self, user): - user_id = user.to_string() - results = {} - - for key in self._batch_row_update: - uid, access_token, ip, = key - if uid == user_id: - user_agent, _, last_seen = self._batch_row_update[key] - results[(access_token, ip)] = (user_agent, last_seen) - - rows = yield self._simple_select_list( - table="user_ips", - keyvalues={"user_id": user_id}, - retcols=["access_token", "ip", "user_agent", "last_seen"], - desc="get_user_ip_and_agents", - ) - - results.update( - ((row["access_token"], row["ip"]), (row["user_agent"], row["last_seen"])) - for row in rows - ) - return list( - { - "access_token": access_token, - "ip": ip, - "user_agent": user_agent, - "last_seen": last_seen, - } - for (access_token, ip), (user_agent, last_seen) in iteritems(results) - ) - - @wrap_as_background_process("prune_old_user_ips") - async def _prune_old_user_ips(self): - """Removes entries in user IPs older than the configured period. - """ - - if self.user_ips_max_age is None: - # Nothing to do - return - - if not await self.has_completed_background_update("devices_last_seen"): - # Only start pruning if we have finished populating the devices - # last seen info. - return - - # We do a slightly funky SQL delete to ensure we don't try and delete - # too much at once (as the table may be very large from before we - # started pruning). - # - # This works by finding the max last_seen that is less than the given - # time, but has no more than N rows before it, deleting all rows with - # a lesser last_seen time. (We COALESCE so that the sub-SELECT always - # returns exactly one row). - sql = """ - DELETE FROM user_ips - WHERE last_seen <= ( - SELECT COALESCE(MAX(last_seen), -1) - FROM ( - SELECT last_seen FROM user_ips - WHERE last_seen <= ? - ORDER BY last_seen ASC - LIMIT 5000 - ) AS u - ) - """ - - timestamp = self.clock.time_msec() - self.user_ips_max_age - - def _prune_old_user_ips_txn(txn): - txn.execute(sql, (timestamp,)) - - await self.runInteraction("_prune_old_user_ips", _prune_old_user_ips_txn) diff --git a/synapse/storage/data_stores/__init__.py b/synapse/storage/data_stores/__init__.py new file mode 100644 index 0000000000..56094078ed --- /dev/null +++ b/synapse/storage/data_stores/__init__.py @@ -0,0 +1,14 @@ +# -*- coding: utf-8 -*- +# Copyright 2019 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. diff --git a/synapse/storage/data_stores/main/__init__.py b/synapse/storage/data_stores/main/__init__.py new file mode 100644 index 0000000000..d29135588f --- /dev/null +++ b/synapse/storage/data_stores/main/__init__.py @@ -0,0 +1,524 @@ +# -*- coding: utf-8 -*- +# Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd +# Copyright 2019 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import calendar +import logging +import time + +from twisted.internet import defer + +from synapse.api.constants import PresenceState +from synapse.storage.engines import PostgresEngine +from synapse.storage.util.id_generators import ( + ChainedIdGenerator, + IdGenerator, + StreamIdGenerator, +) +from synapse.util.caches.stream_change_cache import StreamChangeCache + +from .account_data import AccountDataStore +from .appservice import ApplicationServiceStore, ApplicationServiceTransactionStore +from .client_ips import ClientIpStore +from .deviceinbox import DeviceInboxStore +from .devices import DeviceStore +from .directory import DirectoryStore +from .e2e_room_keys import EndToEndRoomKeyStore +from .end_to_end_keys import EndToEndKeyStore +from .event_federation import EventFederationStore +from .event_push_actions import EventPushActionsStore +from .events import EventsStore +from .events_bg_updates import EventsBackgroundUpdatesStore +from .filtering import FilteringStore +from .group_server import GroupServerStore +from .keys import KeyStore +from .media_repository import MediaRepositoryStore +from .monthly_active_users import MonthlyActiveUsersStore +from .openid import OpenIdStore +from .presence import PresenceStore, UserPresenceState +from .profile import ProfileStore +from .push_rule import PushRuleStore +from .pusher import PusherStore +from .receipts import ReceiptsStore +from .registration import RegistrationStore +from .rejections import RejectionsStore +from .relations import RelationsStore +from .room import RoomStore +from .roommember import RoomMemberStore +from .search import SearchStore +from .signatures import SignatureStore +from .state import StateStore +from .stats import StatsStore +from .stream import StreamStore +from .tags import TagsStore +from .transactions import TransactionStore +from .user_directory import UserDirectoryStore +from .user_erasure_store import UserErasureStore + +logger = logging.getLogger(__name__) + + +class DataStore( + EventsBackgroundUpdatesStore, + RoomMemberStore, + RoomStore, + RegistrationStore, + StreamStore, + ProfileStore, + PresenceStore, + TransactionStore, + DirectoryStore, + KeyStore, + StateStore, + SignatureStore, + ApplicationServiceStore, + EventsStore, + EventFederationStore, + MediaRepositoryStore, + RejectionsStore, + FilteringStore, + PusherStore, + PushRuleStore, + ApplicationServiceTransactionStore, + ReceiptsStore, + EndToEndKeyStore, + EndToEndRoomKeyStore, + SearchStore, + TagsStore, + AccountDataStore, + EventPushActionsStore, + OpenIdStore, + ClientIpStore, + DeviceStore, + DeviceInboxStore, + UserDirectoryStore, + GroupServerStore, + UserErasureStore, + MonthlyActiveUsersStore, + StatsStore, + RelationsStore, +): + def __init__(self, db_conn, hs): + self.hs = hs + self._clock = hs.get_clock() + self.database_engine = hs.database_engine + + self._stream_id_gen = StreamIdGenerator( + db_conn, + "events", + "stream_ordering", + extra_tables=[("local_invites", "stream_id")], + ) + self._backfill_id_gen = StreamIdGenerator( + db_conn, + "events", + "stream_ordering", + step=-1, + extra_tables=[("ex_outlier_stream", "event_stream_ordering")], + ) + self._presence_id_gen = StreamIdGenerator( + db_conn, "presence_stream", "stream_id" + ) + self._device_inbox_id_gen = StreamIdGenerator( + db_conn, "device_max_stream_id", "stream_id" + ) + self._public_room_id_gen = StreamIdGenerator( + db_conn, "public_room_list_stream", "stream_id" + ) + self._device_list_id_gen = StreamIdGenerator( + db_conn, "device_lists_stream", "stream_id" + ) + + self._access_tokens_id_gen = IdGenerator(db_conn, "access_tokens", "id") + self._event_reports_id_gen = IdGenerator(db_conn, "event_reports", "id") + self._push_rule_id_gen = IdGenerator(db_conn, "push_rules", "id") + self._push_rules_enable_id_gen = IdGenerator(db_conn, "push_rules_enable", "id") + self._push_rules_stream_id_gen = ChainedIdGenerator( + self._stream_id_gen, db_conn, "push_rules_stream", "stream_id" + ) + self._pushers_id_gen = StreamIdGenerator( + db_conn, "pushers", "id", extra_tables=[("deleted_pushers", "stream_id")] + ) + self._group_updates_id_gen = StreamIdGenerator( + db_conn, "local_group_updates", "stream_id" + ) + + if isinstance(self.database_engine, PostgresEngine): + self._cache_id_gen = StreamIdGenerator( + db_conn, "cache_invalidation_stream", "stream_id" + ) + else: + self._cache_id_gen = None + + self._presence_on_startup = self._get_active_presence(db_conn) + + presence_cache_prefill, min_presence_val = self._get_cache_dict( + db_conn, + "presence_stream", + entity_column="user_id", + stream_column="stream_id", + max_value=self._presence_id_gen.get_current_token(), + ) + self.presence_stream_cache = StreamChangeCache( + "PresenceStreamChangeCache", + min_presence_val, + prefilled_cache=presence_cache_prefill, + ) + + max_device_inbox_id = self._device_inbox_id_gen.get_current_token() + device_inbox_prefill, min_device_inbox_id = self._get_cache_dict( + db_conn, + "device_inbox", + entity_column="user_id", + stream_column="stream_id", + max_value=max_device_inbox_id, + limit=1000, + ) + self._device_inbox_stream_cache = StreamChangeCache( + "DeviceInboxStreamChangeCache", + min_device_inbox_id, + prefilled_cache=device_inbox_prefill, + ) + # The federation outbox and the local device inbox uses the same + # stream_id generator. + device_outbox_prefill, min_device_outbox_id = self._get_cache_dict( + db_conn, + "device_federation_outbox", + entity_column="destination", + stream_column="stream_id", + max_value=max_device_inbox_id, + limit=1000, + ) + self._device_federation_outbox_stream_cache = StreamChangeCache( + "DeviceFederationOutboxStreamChangeCache", + min_device_outbox_id, + prefilled_cache=device_outbox_prefill, + ) + + device_list_max = self._device_list_id_gen.get_current_token() + self._device_list_stream_cache = StreamChangeCache( + "DeviceListStreamChangeCache", device_list_max + ) + self._device_list_federation_stream_cache = StreamChangeCache( + "DeviceListFederationStreamChangeCache", device_list_max + ) + + events_max = self._stream_id_gen.get_current_token() + curr_state_delta_prefill, min_curr_state_delta_id = self._get_cache_dict( + db_conn, + "current_state_delta_stream", + entity_column="room_id", + stream_column="stream_id", + max_value=events_max, # As we share the stream id with events token + limit=1000, + ) + self._curr_state_delta_stream_cache = StreamChangeCache( + "_curr_state_delta_stream_cache", + min_curr_state_delta_id, + prefilled_cache=curr_state_delta_prefill, + ) + + _group_updates_prefill, min_group_updates_id = self._get_cache_dict( + db_conn, + "local_group_updates", + entity_column="user_id", + stream_column="stream_id", + max_value=self._group_updates_id_gen.get_current_token(), + limit=1000, + ) + self._group_updates_stream_cache = StreamChangeCache( + "_group_updates_stream_cache", + min_group_updates_id, + prefilled_cache=_group_updates_prefill, + ) + + self._stream_order_on_start = self.get_room_max_stream_ordering() + self._min_stream_order_on_start = self.get_room_min_stream_ordering() + + # Used in _generate_user_daily_visits to keep track of progress + self._last_user_visit_update = self._get_start_of_day() + + super(DataStore, self).__init__(db_conn, hs) + + def take_presence_startup_info(self): + active_on_startup = self._presence_on_startup + self._presence_on_startup = None + return active_on_startup + + def _get_active_presence(self, db_conn): + """Fetch non-offline presence from the database so that we can register + the appropriate time outs. + """ + + sql = ( + "SELECT user_id, state, last_active_ts, last_federation_update_ts," + " last_user_sync_ts, status_msg, currently_active FROM presence_stream" + " WHERE state != ?" + ) + sql = self.database_engine.convert_param_style(sql) + + txn = db_conn.cursor() + txn.execute(sql, (PresenceState.OFFLINE,)) + rows = self.cursor_to_dict(txn) + txn.close() + + for row in rows: + row["currently_active"] = bool(row["currently_active"]) + + return [UserPresenceState(**row) for row in rows] + + def count_daily_users(self): + """ + Counts the number of users who used this homeserver in the last 24 hours. + """ + yesterday = int(self._clock.time_msec()) - (1000 * 60 * 60 * 24) + return self.runInteraction("count_daily_users", self._count_users, yesterday) + + def count_monthly_users(self): + """ + Counts the number of users who used this homeserver in the last 30 days. + Note this method is intended for phonehome metrics only and is different + from the mau figure in synapse.storage.monthly_active_users which, + amongst other things, includes a 3 day grace period before a user counts. + """ + thirty_days_ago = int(self._clock.time_msec()) - (1000 * 60 * 60 * 24 * 30) + return self.runInteraction( + "count_monthly_users", self._count_users, thirty_days_ago + ) + + def _count_users(self, txn, time_from): + """ + Returns number of users seen in the past time_from period + """ + sql = """ + SELECT COALESCE(count(*), 0) FROM ( + SELECT user_id FROM user_ips + WHERE last_seen > ? + GROUP BY user_id + ) u + """ + txn.execute(sql, (time_from,)) + count, = txn.fetchone() + return count + + def count_r30_users(self): + """ + Counts the number of 30 day retained users, defined as:- + * Users who have created their accounts more than 30 days ago + * Where last seen at most 30 days ago + * Where account creation and last_seen are > 30 days apart + + Returns counts globaly for a given user as well as breaking + by platform + """ + + def _count_r30_users(txn): + thirty_days_in_secs = 86400 * 30 + now = int(self._clock.time()) + thirty_days_ago_in_secs = now - thirty_days_in_secs + + sql = """ + SELECT platform, COALESCE(count(*), 0) FROM ( + SELECT + users.name, platform, users.creation_ts * 1000, + MAX(uip.last_seen) + FROM users + INNER JOIN ( + SELECT + user_id, + last_seen, + CASE + WHEN user_agent LIKE '%%Android%%' THEN 'android' + WHEN user_agent LIKE '%%iOS%%' THEN 'ios' + WHEN user_agent LIKE '%%Electron%%' THEN 'electron' + WHEN user_agent LIKE '%%Mozilla%%' THEN 'web' + WHEN user_agent LIKE '%%Gecko%%' THEN 'web' + ELSE 'unknown' + END + AS platform + FROM user_ips + ) uip + ON users.name = uip.user_id + AND users.appservice_id is NULL + AND users.creation_ts < ? + AND uip.last_seen/1000 > ? + AND (uip.last_seen/1000) - users.creation_ts > 86400 * 30 + GROUP BY users.name, platform, users.creation_ts + ) u GROUP BY platform + """ + + results = {} + txn.execute(sql, (thirty_days_ago_in_secs, thirty_days_ago_in_secs)) + + for row in txn: + if row[0] == "unknown": + pass + results[row[0]] = row[1] + + sql = """ + SELECT COALESCE(count(*), 0) FROM ( + SELECT users.name, users.creation_ts * 1000, + MAX(uip.last_seen) + FROM users + INNER JOIN ( + SELECT + user_id, + last_seen + FROM user_ips + ) uip + ON users.name = uip.user_id + AND appservice_id is NULL + AND users.creation_ts < ? + AND uip.last_seen/1000 > ? + AND (uip.last_seen/1000) - users.creation_ts > 86400 * 30 + GROUP BY users.name, users.creation_ts + ) u + """ + + txn.execute(sql, (thirty_days_ago_in_secs, thirty_days_ago_in_secs)) + + count, = txn.fetchone() + results["all"] = count + + return results + + return self.runInteraction("count_r30_users", _count_r30_users) + + def _get_start_of_day(self): + """ + Returns millisecond unixtime for start of UTC day. + """ + now = time.gmtime() + today_start = calendar.timegm((now.tm_year, now.tm_mon, now.tm_mday, 0, 0, 0)) + return today_start * 1000 + + def generate_user_daily_visits(self): + """ + Generates daily visit data for use in cohort/ retention analysis + """ + + def _generate_user_daily_visits(txn): + logger.info("Calling _generate_user_daily_visits") + today_start = self._get_start_of_day() + a_day_in_milliseconds = 24 * 60 * 60 * 1000 + now = self.clock.time_msec() + + sql = """ + INSERT INTO user_daily_visits (user_id, device_id, timestamp) + SELECT u.user_id, u.device_id, ? + FROM user_ips AS u + LEFT JOIN ( + SELECT user_id, device_id, timestamp FROM user_daily_visits + WHERE timestamp = ? + ) udv + ON u.user_id = udv.user_id AND u.device_id=udv.device_id + INNER JOIN users ON users.name=u.user_id + WHERE last_seen > ? AND last_seen <= ? + AND udv.timestamp IS NULL AND users.is_guest=0 + AND users.appservice_id IS NULL + GROUP BY u.user_id, u.device_id + """ + + # This means that the day has rolled over but there could still + # be entries from the previous day. There is an edge case + # where if the user logs in at 23:59 and overwrites their + # last_seen at 00:01 then they will not be counted in the + # previous day's stats - it is important that the query is run + # often to minimise this case. + if today_start > self._last_user_visit_update: + yesterday_start = today_start - a_day_in_milliseconds + txn.execute( + sql, + ( + yesterday_start, + yesterday_start, + self._last_user_visit_update, + today_start, + ), + ) + self._last_user_visit_update = today_start + + txn.execute( + sql, (today_start, today_start, self._last_user_visit_update, now) + ) + # Update _last_user_visit_update to now. The reason to do this + # rather just clamping to the beginning of the day is to limit + # the size of the join - meaning that the query can be run more + # frequently + self._last_user_visit_update = now + + return self.runInteraction( + "generate_user_daily_visits", _generate_user_daily_visits + ) + + def get_users(self): + """Function to reterive a list of users in users table. + + Args: + Returns: + defer.Deferred: resolves to list[dict[str, Any]] + """ + return self._simple_select_list( + table="users", + keyvalues={}, + retcols=["name", "password_hash", "is_guest", "admin", "user_type"], + desc="get_users", + ) + + @defer.inlineCallbacks + def get_users_paginate(self, order, start, limit): + """Function to reterive a paginated list of users from + users list. This will return a json object, which contains + list of users and the total number of users in users table. + + Args: + order (str): column name to order the select by this column + start (int): start number to begin the query from + limit (int): number of rows to reterive + Returns: + defer.Deferred: resolves to json object {list[dict[str, Any]], count} + """ + users = yield self.runInteraction( + "get_users_paginate", + self._simple_select_list_paginate_txn, + table="users", + keyvalues={"is_guest": False}, + orderby=order, + start=start, + limit=limit, + retcols=["name", "password_hash", "is_guest", "admin", "user_type"], + ) + count = yield self.runInteraction("get_users_paginate", self.get_user_count_txn) + retval = {"users": users, "total": count} + return retval + + def search_users(self, term): + """Function to search users list for one or more users with + the matched term. + + Args: + term (str): search term + col (str): column to query term should be matched to + Returns: + defer.Deferred: resolves to list[dict[str, Any]] + """ + return self._simple_search_list( + table="users", + term=term, + col="name", + retcols=["name", "password_hash", "is_guest", "admin", "user_type"], + desc="search_users", + ) diff --git a/synapse/storage/data_stores/main/account_data.py b/synapse/storage/data_stores/main/account_data.py new file mode 100644 index 0000000000..6afbfc0d74 --- /dev/null +++ b/synapse/storage/data_stores/main/account_data.py @@ -0,0 +1,391 @@ +# -*- coding: utf-8 -*- +# Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import abc +import logging + +from canonicaljson import json + +from twisted.internet import defer + +from synapse.storage._base import SQLBaseStore +from synapse.storage.util.id_generators import StreamIdGenerator +from synapse.util.caches.descriptors import cached, cachedInlineCallbacks +from synapse.util.caches.stream_change_cache import StreamChangeCache + +logger = logging.getLogger(__name__) + + +class AccountDataWorkerStore(SQLBaseStore): + """This is an abstract base class where subclasses must implement + `get_max_account_data_stream_id` which can be called in the initializer. + """ + + # This ABCMeta metaclass ensures that we cannot be instantiated without + # the abstract methods being implemented. + __metaclass__ = abc.ABCMeta + + def __init__(self, db_conn, hs): + account_max = self.get_max_account_data_stream_id() + self._account_data_stream_cache = StreamChangeCache( + "AccountDataAndTagsChangeCache", account_max + ) + + super(AccountDataWorkerStore, self).__init__(db_conn, hs) + + @abc.abstractmethod + def get_max_account_data_stream_id(self): + """Get the current max stream ID for account data stream + + Returns: + int + """ + raise NotImplementedError() + + @cached() + def get_account_data_for_user(self, user_id): + """Get all the client account_data for a user. + + Args: + user_id(str): The user to get the account_data for. + Returns: + A deferred pair of a dict of global account_data and a dict + mapping from room_id string to per room account_data dicts. + """ + + def get_account_data_for_user_txn(txn): + rows = self._simple_select_list_txn( + txn, + "account_data", + {"user_id": user_id}, + ["account_data_type", "content"], + ) + + global_account_data = { + row["account_data_type"]: json.loads(row["content"]) for row in rows + } + + rows = self._simple_select_list_txn( + txn, + "room_account_data", + {"user_id": user_id}, + ["room_id", "account_data_type", "content"], + ) + + by_room = {} + for row in rows: + room_data = by_room.setdefault(row["room_id"], {}) + room_data[row["account_data_type"]] = json.loads(row["content"]) + + return global_account_data, by_room + + return self.runInteraction( + "get_account_data_for_user", get_account_data_for_user_txn + ) + + @cachedInlineCallbacks(num_args=2, max_entries=5000) + def get_global_account_data_by_type_for_user(self, data_type, user_id): + """ + Returns: + Deferred: A dict + """ + result = yield self._simple_select_one_onecol( + table="account_data", + keyvalues={"user_id": user_id, "account_data_type": data_type}, + retcol="content", + desc="get_global_account_data_by_type_for_user", + allow_none=True, + ) + + if result: + return json.loads(result) + else: + return None + + @cached(num_args=2) + def get_account_data_for_room(self, user_id, room_id): + """Get all the client account_data for a user for a room. + + Args: + user_id(str): The user to get the account_data for. + room_id(str): The room to get the account_data for. + Returns: + A deferred dict of the room account_data + """ + + def get_account_data_for_room_txn(txn): + rows = self._simple_select_list_txn( + txn, + "room_account_data", + {"user_id": user_id, "room_id": room_id}, + ["account_data_type", "content"], + ) + + return { + row["account_data_type"]: json.loads(row["content"]) for row in rows + } + + return self.runInteraction( + "get_account_data_for_room", get_account_data_for_room_txn + ) + + @cached(num_args=3, max_entries=5000) + def get_account_data_for_room_and_type(self, user_id, room_id, account_data_type): + """Get the client account_data of given type for a user for a room. + + Args: + user_id(str): The user to get the account_data for. + room_id(str): The room to get the account_data for. + account_data_type (str): The account data type to get. + Returns: + A deferred of the room account_data for that type, or None if + there isn't any set. + """ + + def get_account_data_for_room_and_type_txn(txn): + content_json = self._simple_select_one_onecol_txn( + txn, + table="room_account_data", + keyvalues={ + "user_id": user_id, + "room_id": room_id, + "account_data_type": account_data_type, + }, + retcol="content", + allow_none=True, + ) + + return json.loads(content_json) if content_json else None + + return self.runInteraction( + "get_account_data_for_room_and_type", get_account_data_for_room_and_type_txn + ) + + def get_all_updated_account_data( + self, last_global_id, last_room_id, current_id, limit + ): + """Get all the client account_data that has changed on the server + Args: + last_global_id(int): The position to fetch from for top level data + last_room_id(int): The position to fetch from for per room data + current_id(int): The position to fetch up to. + Returns: + A deferred pair of lists of tuples of stream_id int, user_id string, + room_id string, type string, and content string. + """ + if last_room_id == current_id and last_global_id == current_id: + return defer.succeed(([], [])) + + def get_updated_account_data_txn(txn): + sql = ( + "SELECT stream_id, user_id, account_data_type, content" + " FROM account_data WHERE ? < stream_id AND stream_id <= ?" + " ORDER BY stream_id ASC LIMIT ?" + ) + txn.execute(sql, (last_global_id, current_id, limit)) + global_results = txn.fetchall() + + sql = ( + "SELECT stream_id, user_id, room_id, account_data_type, content" + " FROM room_account_data WHERE ? < stream_id AND stream_id <= ?" + " ORDER BY stream_id ASC LIMIT ?" + ) + txn.execute(sql, (last_room_id, current_id, limit)) + room_results = txn.fetchall() + return global_results, room_results + + return self.runInteraction( + "get_all_updated_account_data_txn", get_updated_account_data_txn + ) + + def get_updated_account_data_for_user(self, user_id, stream_id): + """Get all the client account_data for a that's changed for a user + + Args: + user_id(str): The user to get the account_data for. + stream_id(int): The point in the stream since which to get updates + Returns: + A deferred pair of a dict of global account_data and a dict + mapping from room_id string to per room account_data dicts. + """ + + def get_updated_account_data_for_user_txn(txn): + sql = ( + "SELECT account_data_type, content FROM account_data" + " WHERE user_id = ? AND stream_id > ?" + ) + + txn.execute(sql, (user_id, stream_id)) + + global_account_data = {row[0]: json.loads(row[1]) for row in txn} + + sql = ( + "SELECT room_id, account_data_type, content FROM room_account_data" + " WHERE user_id = ? AND stream_id > ?" + ) + + txn.execute(sql, (user_id, stream_id)) + + account_data_by_room = {} + for row in txn: + room_account_data = account_data_by_room.setdefault(row[0], {}) + room_account_data[row[1]] = json.loads(row[2]) + + return global_account_data, account_data_by_room + + changed = self._account_data_stream_cache.has_entity_changed( + user_id, int(stream_id) + ) + if not changed: + return {}, {} + + return self.runInteraction( + "get_updated_account_data_for_user", get_updated_account_data_for_user_txn + ) + + @cachedInlineCallbacks(num_args=2, cache_context=True, max_entries=5000) + def is_ignored_by(self, ignored_user_id, ignorer_user_id, cache_context): + ignored_account_data = yield self.get_global_account_data_by_type_for_user( + "m.ignored_user_list", + ignorer_user_id, + on_invalidate=cache_context.invalidate, + ) + if not ignored_account_data: + return False + + return ignored_user_id in ignored_account_data.get("ignored_users", {}) + + +class AccountDataStore(AccountDataWorkerStore): + def __init__(self, db_conn, hs): + self._account_data_id_gen = StreamIdGenerator( + db_conn, "account_data_max_stream_id", "stream_id" + ) + + super(AccountDataStore, self).__init__(db_conn, hs) + + def get_max_account_data_stream_id(self): + """Get the current max stream id for the private user data stream + + Returns: + A deferred int. + """ + return self._account_data_id_gen.get_current_token() + + @defer.inlineCallbacks + def add_account_data_to_room(self, user_id, room_id, account_data_type, content): + """Add some account_data to a room for a user. + Args: + user_id(str): The user to add a tag for. + room_id(str): The room to add a tag for. + account_data_type(str): The type of account_data to add. + content(dict): A json object to associate with the tag. + Returns: + A deferred that completes once the account_data has been added. + """ + content_json = json.dumps(content) + + with self._account_data_id_gen.get_next() as next_id: + # no need to lock here as room_account_data has a unique constraint + # on (user_id, room_id, account_data_type) so _simple_upsert will + # retry if there is a conflict. + yield self._simple_upsert( + desc="add_room_account_data", + table="room_account_data", + keyvalues={ + "user_id": user_id, + "room_id": room_id, + "account_data_type": account_data_type, + }, + values={"stream_id": next_id, "content": content_json}, + lock=False, + ) + + # it's theoretically possible for the above to succeed and the + # below to fail - in which case we might reuse a stream id on + # restart, and the above update might not get propagated. That + # doesn't sound any worse than the whole update getting lost, + # which is what would happen if we combined the two into one + # transaction. + yield self._update_max_stream_id(next_id) + + self._account_data_stream_cache.entity_has_changed(user_id, next_id) + self.get_account_data_for_user.invalidate((user_id,)) + self.get_account_data_for_room.invalidate((user_id, room_id)) + self.get_account_data_for_room_and_type.prefill( + (user_id, room_id, account_data_type), content + ) + + result = self._account_data_id_gen.get_current_token() + return result + + @defer.inlineCallbacks + def add_account_data_for_user(self, user_id, account_data_type, content): + """Add some account_data to a room for a user. + Args: + user_id(str): The user to add a tag for. + account_data_type(str): The type of account_data to add. + content(dict): A json object to associate with the tag. + Returns: + A deferred that completes once the account_data has been added. + """ + content_json = json.dumps(content) + + with self._account_data_id_gen.get_next() as next_id: + # no need to lock here as account_data has a unique constraint on + # (user_id, account_data_type) so _simple_upsert will retry if + # there is a conflict. + yield self._simple_upsert( + desc="add_user_account_data", + table="account_data", + keyvalues={"user_id": user_id, "account_data_type": account_data_type}, + values={"stream_id": next_id, "content": content_json}, + lock=False, + ) + + # it's theoretically possible for the above to succeed and the + # below to fail - in which case we might reuse a stream id on + # restart, and the above update might not get propagated. That + # doesn't sound any worse than the whole update getting lost, + # which is what would happen if we combined the two into one + # transaction. + yield self._update_max_stream_id(next_id) + + self._account_data_stream_cache.entity_has_changed(user_id, next_id) + self.get_account_data_for_user.invalidate((user_id,)) + self.get_global_account_data_by_type_for_user.invalidate( + (account_data_type, user_id) + ) + + result = self._account_data_id_gen.get_current_token() + return result + + def _update_max_stream_id(self, next_id): + """Update the max stream_id + + Args: + next_id(int): The the revision to advance to. + """ + + def _update(txn): + update_max_id_sql = ( + "UPDATE account_data_max_stream_id" + " SET stream_id = ?" + " WHERE stream_id < ?" + ) + txn.execute(update_max_id_sql, (next_id, next_id)) + + return self.runInteraction("update_account_data_max_stream_id", _update) diff --git a/synapse/storage/data_stores/main/appservice.py b/synapse/storage/data_stores/main/appservice.py new file mode 100644 index 0000000000..81babf2029 --- /dev/null +++ b/synapse/storage/data_stores/main/appservice.py @@ -0,0 +1,367 @@ +# -*- coding: utf-8 -*- +# Copyright 2015, 2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import logging +import re + +from canonicaljson import json + +from twisted.internet import defer + +from synapse.appservice import AppServiceTransaction +from synapse.config.appservice import load_appservices +from synapse.storage._base import SQLBaseStore +from synapse.storage.data_stores.main.events_worker import EventsWorkerStore + +logger = logging.getLogger(__name__) + + +def _make_exclusive_regex(services_cache): + # We precompie a regex constructed from all the regexes that the AS's + # have registered for exclusive users. + exclusive_user_regexes = [ + regex.pattern + for service in services_cache + for regex in service.get_exlusive_user_regexes() + ] + if exclusive_user_regexes: + exclusive_user_regex = "|".join("(" + r + ")" for r in exclusive_user_regexes) + exclusive_user_regex = re.compile(exclusive_user_regex) + else: + # We handle this case specially otherwise the constructed regex + # will always match + exclusive_user_regex = None + + return exclusive_user_regex + + +class ApplicationServiceWorkerStore(SQLBaseStore): + def __init__(self, db_conn, hs): + self.services_cache = load_appservices( + hs.hostname, hs.config.app_service_config_files + ) + self.exclusive_user_regex = _make_exclusive_regex(self.services_cache) + + super(ApplicationServiceWorkerStore, self).__init__(db_conn, hs) + + def get_app_services(self): + return self.services_cache + + def get_if_app_services_interested_in_user(self, user_id): + """Check if the user is one associated with an app service (exclusively) + """ + if self.exclusive_user_regex: + return bool(self.exclusive_user_regex.match(user_id)) + else: + return False + + def get_app_service_by_user_id(self, user_id): + """Retrieve an application service from their user ID. + + All application services have associated with them a particular user ID. + There is no distinguishing feature on the user ID which indicates it + represents an application service. This function allows you to map from + a user ID to an application service. + + Args: + user_id(str): The user ID to see if it is an application service. + Returns: + synapse.appservice.ApplicationService or None. + """ + for service in self.services_cache: + if service.sender == user_id: + return service + return None + + def get_app_service_by_token(self, token): + """Get the application service with the given appservice token. + + Args: + token (str): The application service token. + Returns: + synapse.appservice.ApplicationService or None. + """ + for service in self.services_cache: + if service.token == token: + return service + return None + + def get_app_service_by_id(self, as_id): + """Get the application service with the given appservice ID. + + Args: + as_id (str): The application service ID. + Returns: + synapse.appservice.ApplicationService or None. + """ + for service in self.services_cache: + if service.id == as_id: + return service + return None + + +class ApplicationServiceStore(ApplicationServiceWorkerStore): + # This is currently empty due to there not being any AS storage functions + # that can't be run on the workers. Since this may change in future, and + # to keep consistency with the other stores, we keep this empty class for + # now. + pass + + +class ApplicationServiceTransactionWorkerStore( + ApplicationServiceWorkerStore, EventsWorkerStore +): + @defer.inlineCallbacks + def get_appservices_by_state(self, state): + """Get a list of application services based on their state. + + Args: + state(ApplicationServiceState): The state to filter on. + Returns: + A Deferred which resolves to a list of ApplicationServices, which + may be empty. + """ + results = yield self._simple_select_list( + "application_services_state", dict(state=state), ["as_id"] + ) + # NB: This assumes this class is linked with ApplicationServiceStore + as_list = self.get_app_services() + services = [] + + for res in results: + for service in as_list: + if service.id == res["as_id"]: + services.append(service) + return services + + @defer.inlineCallbacks + def get_appservice_state(self, service): + """Get the application service state. + + Args: + service(ApplicationService): The service whose state to set. + Returns: + A Deferred which resolves to ApplicationServiceState. + """ + result = yield self._simple_select_one( + "application_services_state", + dict(as_id=service.id), + ["state"], + allow_none=True, + desc="get_appservice_state", + ) + if result: + return result.get("state") + return None + + def set_appservice_state(self, service, state): + """Set the application service state. + + Args: + service(ApplicationService): The service whose state to set. + state(ApplicationServiceState): The connectivity state to apply. + Returns: + A Deferred which resolves when the state was set successfully. + """ + return self._simple_upsert( + "application_services_state", dict(as_id=service.id), dict(state=state) + ) + + def create_appservice_txn(self, service, events): + """Atomically creates a new transaction for this application service + with the given list of events. + + Args: + service(ApplicationService): The service who the transaction is for. + events(list): A list of events to put in the transaction. + Returns: + AppServiceTransaction: A new transaction. + """ + + def _create_appservice_txn(txn): + # work out new txn id (highest txn id for this service += 1) + # The highest id may be the last one sent (in which case it is last_txn) + # or it may be the highest in the txns list (which are waiting to be/are + # being sent) + last_txn_id = self._get_last_txn(txn, service.id) + + txn.execute( + "SELECT MAX(txn_id) FROM application_services_txns WHERE as_id=?", + (service.id,), + ) + highest_txn_id = txn.fetchone()[0] + if highest_txn_id is None: + highest_txn_id = 0 + + new_txn_id = max(highest_txn_id, last_txn_id) + 1 + + # Insert new txn into txn table + event_ids = json.dumps([e.event_id for e in events]) + txn.execute( + "INSERT INTO application_services_txns(as_id, txn_id, event_ids) " + "VALUES(?,?,?)", + (service.id, new_txn_id, event_ids), + ) + return AppServiceTransaction(service=service, id=new_txn_id, events=events) + + return self.runInteraction("create_appservice_txn", _create_appservice_txn) + + def complete_appservice_txn(self, txn_id, service): + """Completes an application service transaction. + + Args: + txn_id(str): The transaction ID being completed. + service(ApplicationService): The application service which was sent + this transaction. + Returns: + A Deferred which resolves if this transaction was stored + successfully. + """ + txn_id = int(txn_id) + + def _complete_appservice_txn(txn): + # Debugging query: Make sure the txn being completed is EXACTLY +1 from + # what was there before. If it isn't, we've got problems (e.g. the AS + # has probably missed some events), so whine loudly but still continue, + # since it shouldn't fail completion of the transaction. + last_txn_id = self._get_last_txn(txn, service.id) + if (last_txn_id + 1) != txn_id: + logger.error( + "appservice: Completing a transaction which has an ID > 1 from " + "the last ID sent to this AS. We've either dropped events or " + "sent it to the AS out of order. FIX ME. last_txn=%s " + "completing_txn=%s service_id=%s", + last_txn_id, + txn_id, + service.id, + ) + + # Set current txn_id for AS to 'txn_id' + self._simple_upsert_txn( + txn, + "application_services_state", + dict(as_id=service.id), + dict(last_txn=txn_id), + ) + + # Delete txn + self._simple_delete_txn( + txn, "application_services_txns", dict(txn_id=txn_id, as_id=service.id) + ) + + return self.runInteraction("complete_appservice_txn", _complete_appservice_txn) + + @defer.inlineCallbacks + def get_oldest_unsent_txn(self, service): + """Get the oldest transaction which has not been sent for this + service. + + Args: + service(ApplicationService): The app service to get the oldest txn. + Returns: + A Deferred which resolves to an AppServiceTransaction or + None. + """ + + def _get_oldest_unsent_txn(txn): + # Monotonically increasing txn ids, so just select the smallest + # one in the txns table (we delete them when they are sent) + txn.execute( + "SELECT * FROM application_services_txns WHERE as_id=?" + " ORDER BY txn_id ASC LIMIT 1", + (service.id,), + ) + rows = self.cursor_to_dict(txn) + if not rows: + return None + + entry = rows[0] + + return entry + + entry = yield self.runInteraction( + "get_oldest_unsent_appservice_txn", _get_oldest_unsent_txn + ) + + if not entry: + return None + + event_ids = json.loads(entry["event_ids"]) + + events = yield self.get_events_as_list(event_ids) + + return AppServiceTransaction(service=service, id=entry["txn_id"], events=events) + + def _get_last_txn(self, txn, service_id): + txn.execute( + "SELECT last_txn FROM application_services_state WHERE as_id=?", + (service_id,), + ) + last_txn_id = txn.fetchone() + if last_txn_id is None or last_txn_id[0] is None: # no row exists + return 0 + else: + return int(last_txn_id[0]) # select 'last_txn' col + + def set_appservice_last_pos(self, pos): + def set_appservice_last_pos_txn(txn): + txn.execute( + "UPDATE appservice_stream_position SET stream_ordering = ?", (pos,) + ) + + return self.runInteraction( + "set_appservice_last_pos", set_appservice_last_pos_txn + ) + + @defer.inlineCallbacks + def get_new_events_for_appservice(self, current_id, limit): + """Get all new evnets""" + + def get_new_events_for_appservice_txn(txn): + sql = ( + "SELECT e.stream_ordering, e.event_id" + " FROM events AS e" + " WHERE" + " (SELECT stream_ordering FROM appservice_stream_position)" + " < e.stream_ordering" + " AND e.stream_ordering <= ?" + " ORDER BY e.stream_ordering ASC" + " LIMIT ?" + ) + + txn.execute(sql, (current_id, limit)) + rows = txn.fetchall() + + upper_bound = current_id + if len(rows) == limit: + upper_bound = rows[-1][0] + + return upper_bound, [row[1] for row in rows] + + upper_bound, event_ids = yield self.runInteraction( + "get_new_events_for_appservice", get_new_events_for_appservice_txn + ) + + events = yield self.get_events_as_list(event_ids) + + return upper_bound, events + + +class ApplicationServiceTransactionStore(ApplicationServiceTransactionWorkerStore): + # This is currently empty due to there not being any AS storage functions + # that can't be run on the workers. Since this may change in future, and + # to keep consistency with the other stores, we keep this empty class for + # now. + pass diff --git a/synapse/storage/data_stores/main/client_ips.py b/synapse/storage/data_stores/main/client_ips.py new file mode 100644 index 0000000000..706c6a1f3f --- /dev/null +++ b/synapse/storage/data_stores/main/client_ips.py @@ -0,0 +1,580 @@ +# -*- coding: utf-8 -*- +# Copyright 2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging + +from six import iteritems + +from twisted.internet import defer + +from synapse.metrics.background_process_metrics import wrap_as_background_process +from synapse.storage import background_updates +from synapse.storage._base import Cache +from synapse.util.caches import CACHE_SIZE_FACTOR + +logger = logging.getLogger(__name__) + +# Number of msec of granularity to store the user IP 'last seen' time. Smaller +# times give more inserts into the database even for readonly API hits +# 120 seconds == 2 minutes +LAST_SEEN_GRANULARITY = 120 * 1000 + + +class ClientIpBackgroundUpdateStore(background_updates.BackgroundUpdateStore): + def __init__(self, db_conn, hs): + super(ClientIpBackgroundUpdateStore, self).__init__(db_conn, hs) + + self.register_background_index_update( + "user_ips_device_index", + index_name="user_ips_device_id", + table="user_ips", + columns=["user_id", "device_id", "last_seen"], + ) + + self.register_background_index_update( + "user_ips_last_seen_index", + index_name="user_ips_last_seen", + table="user_ips", + columns=["user_id", "last_seen"], + ) + + self.register_background_index_update( + "user_ips_last_seen_only_index", + index_name="user_ips_last_seen_only", + table="user_ips", + columns=["last_seen"], + ) + + self.register_background_update_handler( + "user_ips_analyze", self._analyze_user_ip + ) + + self.register_background_update_handler( + "user_ips_remove_dupes", self._remove_user_ip_dupes + ) + + # Register a unique index + self.register_background_index_update( + "user_ips_device_unique_index", + index_name="user_ips_user_token_ip_unique_index", + table="user_ips", + columns=["user_id", "access_token", "ip"], + unique=True, + ) + + # Drop the old non-unique index + self.register_background_update_handler( + "user_ips_drop_nonunique_index", self._remove_user_ip_nonunique + ) + + # Update the last seen info in devices. + self.register_background_update_handler( + "devices_last_seen", self._devices_last_seen_update + ) + + @defer.inlineCallbacks + def _remove_user_ip_nonunique(self, progress, batch_size): + def f(conn): + txn = conn.cursor() + txn.execute("DROP INDEX IF EXISTS user_ips_user_ip") + txn.close() + + yield self.runWithConnection(f) + yield self._end_background_update("user_ips_drop_nonunique_index") + return 1 + + @defer.inlineCallbacks + def _analyze_user_ip(self, progress, batch_size): + # Background update to analyze user_ips table before we run the + # deduplication background update. The table may not have been analyzed + # for ages due to the table locks. + # + # This will lock out the naive upserts to user_ips while it happens, but + # the analyze should be quick (28GB table takes ~10s) + def user_ips_analyze(txn): + txn.execute("ANALYZE user_ips") + + yield self.runInteraction("user_ips_analyze", user_ips_analyze) + + yield self._end_background_update("user_ips_analyze") + + return 1 + + @defer.inlineCallbacks + def _remove_user_ip_dupes(self, progress, batch_size): + # This works function works by scanning the user_ips table in batches + # based on `last_seen`. For each row in a batch it searches the rest of + # the table to see if there are any duplicates, if there are then they + # are removed and replaced with a suitable row. + + # Fetch the start of the batch + begin_last_seen = progress.get("last_seen", 0) + + def get_last_seen(txn): + txn.execute( + """ + SELECT last_seen FROM user_ips + WHERE last_seen > ? + ORDER BY last_seen + LIMIT 1 + OFFSET ? + """, + (begin_last_seen, batch_size), + ) + row = txn.fetchone() + if row: + return row[0] + else: + return None + + # Get a last seen that has roughly `batch_size` since `begin_last_seen` + end_last_seen = yield self.runInteraction( + "user_ips_dups_get_last_seen", get_last_seen + ) + + # If it returns None, then we're processing the last batch + last = end_last_seen is None + + logger.info( + "Scanning for duplicate 'user_ips' rows in range: %s <= last_seen < %s", + begin_last_seen, + end_last_seen, + ) + + def remove(txn): + # This works by looking at all entries in the given time span, and + # then for each (user_id, access_token, ip) tuple in that range + # checking for any duplicates in the rest of the table (via a join). + # It then only returns entries which have duplicates, and the max + # last_seen across all duplicates, which can the be used to delete + # all other duplicates. + # It is efficient due to the existence of (user_id, access_token, + # ip) and (last_seen) indices. + + # Define the search space, which requires handling the last batch in + # a different way + if last: + clause = "? <= last_seen" + args = (begin_last_seen,) + else: + clause = "? <= last_seen AND last_seen < ?" + args = (begin_last_seen, end_last_seen) + + # (Note: The DISTINCT in the inner query is important to ensure that + # the COUNT(*) is accurate, otherwise double counting may happen due + # to the join effectively being a cross product) + txn.execute( + """ + SELECT user_id, access_token, ip, + MAX(device_id), MAX(user_agent), MAX(last_seen), + COUNT(*) + FROM ( + SELECT DISTINCT user_id, access_token, ip + FROM user_ips + WHERE {} + ) c + INNER JOIN user_ips USING (user_id, access_token, ip) + GROUP BY user_id, access_token, ip + HAVING count(*) > 1 + """.format( + clause + ), + args, + ) + res = txn.fetchall() + + # We've got some duplicates + for i in res: + user_id, access_token, ip, device_id, user_agent, last_seen, count = i + + # We want to delete the duplicates so we end up with only a + # single row. + # + # The naive way of doing this would be just to delete all rows + # and reinsert a constructed row. However, if there are a lot of + # duplicate rows this can cause the table to grow a lot, which + # can be problematic in two ways: + # 1. If user_ips is already large then this can cause the + # table to rapidly grow, potentially filling the disk. + # 2. Reinserting a lot of rows can confuse the table + # statistics for postgres, causing it to not use the + # correct indices for the query above, resulting in a full + # table scan. This is incredibly slow for large tables and + # can kill database performance. (This seems to mainly + # happen for the last query where the clause is simply `? < + # last_seen`) + # + # So instead we want to delete all but *one* of the duplicate + # rows. That is hard to do reliably, so we cheat and do a two + # step process: + # 1. Delete all rows with a last_seen strictly less than the + # max last_seen. This hopefully results in deleting all but + # one row the majority of the time, but there may be + # duplicate last_seen + # 2. If multiple rows remain, we fall back to the naive method + # and simply delete all rows and reinsert. + # + # Note that this relies on no new duplicate rows being inserted, + # but if that is happening then this entire process is futile + # anyway. + + # Do step 1: + + txn.execute( + """ + DELETE FROM user_ips + WHERE user_id = ? AND access_token = ? AND ip = ? AND last_seen < ? + """, + (user_id, access_token, ip, last_seen), + ) + if txn.rowcount == count - 1: + # We deleted all but one of the duplicate rows, i.e. there + # is exactly one remaining and so there is nothing left to + # do. + continue + elif txn.rowcount >= count: + raise Exception( + "We deleted more duplicate rows from 'user_ips' than expected" + ) + + # The previous step didn't delete enough rows, so we fallback to + # step 2: + + # Drop all the duplicates + txn.execute( + """ + DELETE FROM user_ips + WHERE user_id = ? AND access_token = ? AND ip = ? + """, + (user_id, access_token, ip), + ) + + # Add in one to be the last_seen + txn.execute( + """ + INSERT INTO user_ips + (user_id, access_token, ip, device_id, user_agent, last_seen) + VALUES (?, ?, ?, ?, ?, ?) + """, + (user_id, access_token, ip, device_id, user_agent, last_seen), + ) + + self._background_update_progress_txn( + txn, "user_ips_remove_dupes", {"last_seen": end_last_seen} + ) + + yield self.runInteraction("user_ips_dups_remove", remove) + + if last: + yield self._end_background_update("user_ips_remove_dupes") + + return batch_size + + @defer.inlineCallbacks + def _devices_last_seen_update(self, progress, batch_size): + """Background update to insert last seen info into devices table + """ + + last_user_id = progress.get("last_user_id", "") + last_device_id = progress.get("last_device_id", "") + + def _devices_last_seen_update_txn(txn): + # This consists of two queries: + # + # 1. The sub-query searches for the next N devices and joins + # against user_ips to find the max last_seen associated with + # that device. + # 2. The outer query then joins again against user_ips on + # user/device/last_seen. This *should* hopefully only + # return one row, but if it does return more than one then + # we'll just end up updating the same device row multiple + # times, which is fine. + + if self.database_engine.supports_tuple_comparison: + where_clause = "(user_id, device_id) > (?, ?)" + where_args = [last_user_id, last_device_id] + else: + # We explicitly do a `user_id >= ? AND (...)` here to ensure + # that an index is used, as doing `user_id > ? OR (user_id = ? AND ...)` + # makes it hard for query optimiser to tell that it can use the + # index on user_id + where_clause = "user_id >= ? AND (user_id > ? OR device_id > ?)" + where_args = [last_user_id, last_user_id, last_device_id] + + sql = """ + SELECT + last_seen, ip, user_agent, user_id, device_id + FROM ( + SELECT + user_id, device_id, MAX(u.last_seen) AS last_seen + FROM devices + INNER JOIN user_ips AS u USING (user_id, device_id) + WHERE %(where_clause)s + GROUP BY user_id, device_id + ORDER BY user_id ASC, device_id ASC + LIMIT ? + ) c + INNER JOIN user_ips AS u USING (user_id, device_id, last_seen) + """ % { + "where_clause": where_clause + } + txn.execute(sql, where_args + [batch_size]) + + rows = txn.fetchall() + if not rows: + return 0 + + sql = """ + UPDATE devices + SET last_seen = ?, ip = ?, user_agent = ? + WHERE user_id = ? AND device_id = ? + """ + txn.execute_batch(sql, rows) + + _, _, _, user_id, device_id = rows[-1] + self._background_update_progress_txn( + txn, + "devices_last_seen", + {"last_user_id": user_id, "last_device_id": device_id}, + ) + + return len(rows) + + updated = yield self.runInteraction( + "_devices_last_seen_update", _devices_last_seen_update_txn + ) + + if not updated: + yield self._end_background_update("devices_last_seen") + + return updated + + +class ClientIpStore(ClientIpBackgroundUpdateStore): + def __init__(self, db_conn, hs): + + self.client_ip_last_seen = Cache( + name="client_ip_last_seen", keylen=4, max_entries=50000 * CACHE_SIZE_FACTOR + ) + + super(ClientIpStore, self).__init__(db_conn, hs) + + self.user_ips_max_age = hs.config.user_ips_max_age + + # (user_id, access_token, ip,) -> (user_agent, device_id, last_seen) + self._batch_row_update = {} + + self._client_ip_looper = self._clock.looping_call( + self._update_client_ips_batch, 5 * 1000 + ) + self.hs.get_reactor().addSystemEventTrigger( + "before", "shutdown", self._update_client_ips_batch + ) + + if self.user_ips_max_age: + self._clock.looping_call(self._prune_old_user_ips, 5 * 1000) + + @defer.inlineCallbacks + def insert_client_ip( + self, user_id, access_token, ip, user_agent, device_id, now=None + ): + if not now: + now = int(self._clock.time_msec()) + key = (user_id, access_token, ip) + + try: + last_seen = self.client_ip_last_seen.get(key) + except KeyError: + last_seen = None + yield self.populate_monthly_active_users(user_id) + # Rate-limited inserts + if last_seen is not None and (now - last_seen) < LAST_SEEN_GRANULARITY: + return + + self.client_ip_last_seen.prefill(key, now) + + self._batch_row_update[key] = (user_agent, device_id, now) + + @wrap_as_background_process("update_client_ips") + def _update_client_ips_batch(self): + + # If the DB pool has already terminated, don't try updating + if not self.hs.get_db_pool().running: + return + + to_update = self._batch_row_update + self._batch_row_update = {} + + return self.runInteraction( + "_update_client_ips_batch", self._update_client_ips_batch_txn, to_update + ) + + def _update_client_ips_batch_txn(self, txn, to_update): + if "user_ips" in self._unsafe_to_upsert_tables or ( + not self.database_engine.can_native_upsert + ): + self.database_engine.lock_table(txn, "user_ips") + + for entry in iteritems(to_update): + (user_id, access_token, ip), (user_agent, device_id, last_seen) = entry + + try: + self._simple_upsert_txn( + txn, + table="user_ips", + keyvalues={ + "user_id": user_id, + "access_token": access_token, + "ip": ip, + }, + values={ + "user_agent": user_agent, + "device_id": device_id, + "last_seen": last_seen, + }, + lock=False, + ) + + # Technically an access token might not be associated with + # a device so we need to check. + if device_id: + self._simple_upsert_txn( + txn, + table="devices", + keyvalues={"user_id": user_id, "device_id": device_id}, + values={ + "user_agent": user_agent, + "last_seen": last_seen, + "ip": ip, + }, + lock=False, + ) + except Exception as e: + # Failed to upsert, log and continue + logger.error("Failed to insert client IP %r: %r", entry, e) + + @defer.inlineCallbacks + def get_last_client_ip_by_device(self, user_id, device_id): + """For each device_id listed, give the user_ip it was last seen on + + Args: + user_id (str) + device_id (str): If None fetches all devices for the user + + Returns: + defer.Deferred: resolves to a dict, where the keys + are (user_id, device_id) tuples. The values are also dicts, with + keys giving the column names + """ + + keyvalues = {"user_id": user_id} + if device_id is not None: + keyvalues["device_id"] = device_id + + res = yield self._simple_select_list( + table="devices", + keyvalues=keyvalues, + retcols=("user_id", "ip", "user_agent", "device_id", "last_seen"), + ) + + ret = {(d["user_id"], d["device_id"]): d for d in res} + for key in self._batch_row_update: + uid, access_token, ip = key + if uid == user_id: + user_agent, did, last_seen = self._batch_row_update[key] + if not device_id or did == device_id: + ret[(user_id, device_id)] = { + "user_id": user_id, + "access_token": access_token, + "ip": ip, + "user_agent": user_agent, + "device_id": did, + "last_seen": last_seen, + } + return ret + + @defer.inlineCallbacks + def get_user_ip_and_agents(self, user): + user_id = user.to_string() + results = {} + + for key in self._batch_row_update: + uid, access_token, ip, = key + if uid == user_id: + user_agent, _, last_seen = self._batch_row_update[key] + results[(access_token, ip)] = (user_agent, last_seen) + + rows = yield self._simple_select_list( + table="user_ips", + keyvalues={"user_id": user_id}, + retcols=["access_token", "ip", "user_agent", "last_seen"], + desc="get_user_ip_and_agents", + ) + + results.update( + ((row["access_token"], row["ip"]), (row["user_agent"], row["last_seen"])) + for row in rows + ) + return list( + { + "access_token": access_token, + "ip": ip, + "user_agent": user_agent, + "last_seen": last_seen, + } + for (access_token, ip), (user_agent, last_seen) in iteritems(results) + ) + + @wrap_as_background_process("prune_old_user_ips") + async def _prune_old_user_ips(self): + """Removes entries in user IPs older than the configured period. + """ + + if self.user_ips_max_age is None: + # Nothing to do + return + + if not await self.has_completed_background_update("devices_last_seen"): + # Only start pruning if we have finished populating the devices + # last seen info. + return + + # We do a slightly funky SQL delete to ensure we don't try and delete + # too much at once (as the table may be very large from before we + # started pruning). + # + # This works by finding the max last_seen that is less than the given + # time, but has no more than N rows before it, deleting all rows with + # a lesser last_seen time. (We COALESCE so that the sub-SELECT always + # returns exactly one row). + sql = """ + DELETE FROM user_ips + WHERE last_seen <= ( + SELECT COALESCE(MAX(last_seen), -1) + FROM ( + SELECT last_seen FROM user_ips + WHERE last_seen <= ? + ORDER BY last_seen ASC + LIMIT 5000 + ) AS u + ) + """ + + timestamp = self.clock.time_msec() - self.user_ips_max_age + + def _prune_old_user_ips_txn(txn): + txn.execute(sql, (timestamp,)) + + await self.runInteraction("_prune_old_user_ips", _prune_old_user_ips_txn) diff --git a/synapse/storage/data_stores/main/deviceinbox.py b/synapse/storage/data_stores/main/deviceinbox.py new file mode 100644 index 0000000000..f04aad0743 --- /dev/null +++ b/synapse/storage/data_stores/main/deviceinbox.py @@ -0,0 +1,457 @@ +# -*- coding: utf-8 -*- +# Copyright 2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging + +from canonicaljson import json + +from twisted.internet import defer + +from synapse.logging.opentracing import log_kv, set_tag, trace +from synapse.storage._base import SQLBaseStore, make_in_list_sql_clause +from synapse.storage.background_updates import BackgroundUpdateStore +from synapse.util.caches.expiringcache import ExpiringCache + +logger = logging.getLogger(__name__) + + +class DeviceInboxWorkerStore(SQLBaseStore): + def get_to_device_stream_token(self): + return self._device_inbox_id_gen.get_current_token() + + def get_new_messages_for_device( + self, user_id, device_id, last_stream_id, current_stream_id, limit=100 + ): + """ + Args: + user_id(str): The recipient user_id. + device_id(str): The recipient device_id. + current_stream_id(int): The current position of the to device + message stream. + Returns: + Deferred ([dict], int): List of messages for the device and where + in the stream the messages got to. + """ + has_changed = self._device_inbox_stream_cache.has_entity_changed( + user_id, last_stream_id + ) + if not has_changed: + return defer.succeed(([], current_stream_id)) + + def get_new_messages_for_device_txn(txn): + sql = ( + "SELECT stream_id, message_json FROM device_inbox" + " WHERE user_id = ? AND device_id = ?" + " AND ? < stream_id AND stream_id <= ?" + " ORDER BY stream_id ASC" + " LIMIT ?" + ) + txn.execute( + sql, (user_id, device_id, last_stream_id, current_stream_id, limit) + ) + messages = [] + for row in txn: + stream_pos = row[0] + messages.append(json.loads(row[1])) + if len(messages) < limit: + stream_pos = current_stream_id + return messages, stream_pos + + return self.runInteraction( + "get_new_messages_for_device", get_new_messages_for_device_txn + ) + + @trace + @defer.inlineCallbacks + def delete_messages_for_device(self, user_id, device_id, up_to_stream_id): + """ + Args: + user_id(str): The recipient user_id. + device_id(str): The recipient device_id. + up_to_stream_id(int): Where to delete messages up to. + Returns: + A deferred that resolves to the number of messages deleted. + """ + # If we have cached the last stream id we've deleted up to, we can + # check if there is likely to be anything that needs deleting + last_deleted_stream_id = self._last_device_delete_cache.get( + (user_id, device_id), None + ) + + set_tag("last_deleted_stream_id", last_deleted_stream_id) + + if last_deleted_stream_id: + has_changed = self._device_inbox_stream_cache.has_entity_changed( + user_id, last_deleted_stream_id + ) + if not has_changed: + log_kv({"message": "No changes in cache since last check"}) + return 0 + + def delete_messages_for_device_txn(txn): + sql = ( + "DELETE FROM device_inbox" + " WHERE user_id = ? AND device_id = ?" + " AND stream_id <= ?" + ) + txn.execute(sql, (user_id, device_id, up_to_stream_id)) + return txn.rowcount + + count = yield self.runInteraction( + "delete_messages_for_device", delete_messages_for_device_txn + ) + + log_kv( + {"message": "deleted {} messages for device".format(count), "count": count} + ) + + # Update the cache, ensuring that we only ever increase the value + last_deleted_stream_id = self._last_device_delete_cache.get( + (user_id, device_id), 0 + ) + self._last_device_delete_cache[(user_id, device_id)] = max( + last_deleted_stream_id, up_to_stream_id + ) + + return count + + @trace + def get_new_device_msgs_for_remote( + self, destination, last_stream_id, current_stream_id, limit + ): + """ + Args: + destination(str): The name of the remote server. + last_stream_id(int|long): The last position of the device message stream + that the server sent up to. + current_stream_id(int|long): The current position of the device + message stream. + Returns: + Deferred ([dict], int|long): List of messages for the device and where + in the stream the messages got to. + """ + + set_tag("destination", destination) + set_tag("last_stream_id", last_stream_id) + set_tag("current_stream_id", current_stream_id) + set_tag("limit", limit) + + has_changed = self._device_federation_outbox_stream_cache.has_entity_changed( + destination, last_stream_id + ) + if not has_changed or last_stream_id == current_stream_id: + log_kv({"message": "No new messages in stream"}) + return defer.succeed(([], current_stream_id)) + + if limit <= 0: + # This can happen if we run out of room for EDUs in the transaction. + return defer.succeed(([], last_stream_id)) + + @trace + def get_new_messages_for_remote_destination_txn(txn): + sql = ( + "SELECT stream_id, messages_json FROM device_federation_outbox" + " WHERE destination = ?" + " AND ? < stream_id AND stream_id <= ?" + " ORDER BY stream_id ASC" + " LIMIT ?" + ) + txn.execute(sql, (destination, last_stream_id, current_stream_id, limit)) + messages = [] + for row in txn: + stream_pos = row[0] + messages.append(json.loads(row[1])) + if len(messages) < limit: + log_kv({"message": "Set stream position to current position"}) + stream_pos = current_stream_id + return messages, stream_pos + + return self.runInteraction( + "get_new_device_msgs_for_remote", + get_new_messages_for_remote_destination_txn, + ) + + @trace + def delete_device_msgs_for_remote(self, destination, up_to_stream_id): + """Used to delete messages when the remote destination acknowledges + their receipt. + + Args: + destination(str): The destination server_name + up_to_stream_id(int): Where to delete messages up to. + Returns: + A deferred that resolves when the messages have been deleted. + """ + + def delete_messages_for_remote_destination_txn(txn): + sql = ( + "DELETE FROM device_federation_outbox" + " WHERE destination = ?" + " AND stream_id <= ?" + ) + txn.execute(sql, (destination, up_to_stream_id)) + + return self.runInteraction( + "delete_device_msgs_for_remote", delete_messages_for_remote_destination_txn + ) + + +class DeviceInboxBackgroundUpdateStore(BackgroundUpdateStore): + DEVICE_INBOX_STREAM_ID = "device_inbox_stream_drop" + + def __init__(self, db_conn, hs): + super(DeviceInboxBackgroundUpdateStore, self).__init__(db_conn, hs) + + self.register_background_index_update( + "device_inbox_stream_index", + index_name="device_inbox_stream_id_user_id", + table="device_inbox", + columns=["stream_id", "user_id"], + ) + + self.register_background_update_handler( + self.DEVICE_INBOX_STREAM_ID, self._background_drop_index_device_inbox + ) + + @defer.inlineCallbacks + def _background_drop_index_device_inbox(self, progress, batch_size): + def reindex_txn(conn): + txn = conn.cursor() + txn.execute("DROP INDEX IF EXISTS device_inbox_stream_id") + txn.close() + + yield self.runWithConnection(reindex_txn) + + yield self._end_background_update(self.DEVICE_INBOX_STREAM_ID) + + return 1 + + +class DeviceInboxStore(DeviceInboxWorkerStore, DeviceInboxBackgroundUpdateStore): + DEVICE_INBOX_STREAM_ID = "device_inbox_stream_drop" + + def __init__(self, db_conn, hs): + super(DeviceInboxStore, self).__init__(db_conn, hs) + + # Map of (user_id, device_id) to the last stream_id that has been + # deleted up to. This is so that we can no op deletions. + self._last_device_delete_cache = ExpiringCache( + cache_name="last_device_delete_cache", + clock=self._clock, + max_len=10000, + expiry_ms=30 * 60 * 1000, + ) + + @trace + @defer.inlineCallbacks + def add_messages_to_device_inbox( + self, local_messages_by_user_then_device, remote_messages_by_destination + ): + """Used to send messages from this server. + + Args: + sender_user_id(str): The ID of the user sending these messages. + local_messages_by_user_and_device(dict): + Dictionary of user_id to device_id to message. + remote_messages_by_destination(dict): + Dictionary of destination server_name to the EDU JSON to send. + Returns: + A deferred stream_id that resolves when the messages have been + inserted. + """ + + def add_messages_txn(txn, now_ms, stream_id): + # Add the local messages directly to the local inbox. + self._add_messages_to_local_device_inbox_txn( + txn, stream_id, local_messages_by_user_then_device + ) + + # Add the remote messages to the federation outbox. + # We'll send them to a remote server when we next send a + # federation transaction to that destination. + sql = ( + "INSERT INTO device_federation_outbox" + " (destination, stream_id, queued_ts, messages_json)" + " VALUES (?,?,?,?)" + ) + rows = [] + for destination, edu in remote_messages_by_destination.items(): + edu_json = json.dumps(edu) + rows.append((destination, stream_id, now_ms, edu_json)) + txn.executemany(sql, rows) + + with self._device_inbox_id_gen.get_next() as stream_id: + now_ms = self.clock.time_msec() + yield self.runInteraction( + "add_messages_to_device_inbox", add_messages_txn, now_ms, stream_id + ) + for user_id in local_messages_by_user_then_device.keys(): + self._device_inbox_stream_cache.entity_has_changed(user_id, stream_id) + for destination in remote_messages_by_destination.keys(): + self._device_federation_outbox_stream_cache.entity_has_changed( + destination, stream_id + ) + + return self._device_inbox_id_gen.get_current_token() + + @defer.inlineCallbacks + def add_messages_from_remote_to_device_inbox( + self, origin, message_id, local_messages_by_user_then_device + ): + def add_messages_txn(txn, now_ms, stream_id): + # Check if we've already inserted a matching message_id for that + # origin. This can happen if the origin doesn't receive our + # acknowledgement from the first time we received the message. + already_inserted = self._simple_select_one_txn( + txn, + table="device_federation_inbox", + keyvalues={"origin": origin, "message_id": message_id}, + retcols=("message_id",), + allow_none=True, + ) + if already_inserted is not None: + return + + # Add an entry for this message_id so that we know we've processed + # it. + self._simple_insert_txn( + txn, + table="device_federation_inbox", + values={ + "origin": origin, + "message_id": message_id, + "received_ts": now_ms, + }, + ) + + # Add the messages to the approriate local device inboxes so that + # they'll be sent to the devices when they next sync. + self._add_messages_to_local_device_inbox_txn( + txn, stream_id, local_messages_by_user_then_device + ) + + with self._device_inbox_id_gen.get_next() as stream_id: + now_ms = self.clock.time_msec() + yield self.runInteraction( + "add_messages_from_remote_to_device_inbox", + add_messages_txn, + now_ms, + stream_id, + ) + for user_id in local_messages_by_user_then_device.keys(): + self._device_inbox_stream_cache.entity_has_changed(user_id, stream_id) + + return stream_id + + def _add_messages_to_local_device_inbox_txn( + self, txn, stream_id, messages_by_user_then_device + ): + sql = "UPDATE device_max_stream_id" " SET stream_id = ?" " WHERE stream_id < ?" + txn.execute(sql, (stream_id, stream_id)) + + local_by_user_then_device = {} + for user_id, messages_by_device in messages_by_user_then_device.items(): + messages_json_for_user = {} + devices = list(messages_by_device.keys()) + if len(devices) == 1 and devices[0] == "*": + # Handle wildcard device_ids. + sql = "SELECT device_id FROM devices" " WHERE user_id = ?" + txn.execute(sql, (user_id,)) + message_json = json.dumps(messages_by_device["*"]) + for row in txn: + # Add the message for all devices for this user on this + # server. + device = row[0] + messages_json_for_user[device] = message_json + else: + if not devices: + continue + + clause, args = make_in_list_sql_clause( + txn.database_engine, "device_id", devices + ) + sql = "SELECT device_id FROM devices WHERE user_id = ? AND " + clause + + # TODO: Maybe this needs to be done in batches if there are + # too many local devices for a given user. + txn.execute(sql, [user_id] + list(args)) + for row in txn: + # Only insert into the local inbox if the device exists on + # this server + device = row[0] + message_json = json.dumps(messages_by_device[device]) + messages_json_for_user[device] = message_json + + if messages_json_for_user: + local_by_user_then_device[user_id] = messages_json_for_user + + if not local_by_user_then_device: + return + + sql = ( + "INSERT INTO device_inbox" + " (user_id, device_id, stream_id, message_json)" + " VALUES (?,?,?,?)" + ) + rows = [] + for user_id, messages_by_device in local_by_user_then_device.items(): + for device_id, message_json in messages_by_device.items(): + rows.append((user_id, device_id, stream_id, message_json)) + + txn.executemany(sql, rows) + + def get_all_new_device_messages(self, last_pos, current_pos, limit): + """ + Args: + last_pos(int): + current_pos(int): + limit(int): + Returns: + A deferred list of rows from the device inbox + """ + if last_pos == current_pos: + return defer.succeed([]) + + def get_all_new_device_messages_txn(txn): + # We limit like this as we might have multiple rows per stream_id, and + # we want to make sure we always get all entries for any stream_id + # we return. + upper_pos = min(current_pos, last_pos + limit) + sql = ( + "SELECT max(stream_id), user_id" + " FROM device_inbox" + " WHERE ? < stream_id AND stream_id <= ?" + " GROUP BY user_id" + ) + txn.execute(sql, (last_pos, upper_pos)) + rows = txn.fetchall() + + sql = ( + "SELECT max(stream_id), destination" + " FROM device_federation_outbox" + " WHERE ? < stream_id AND stream_id <= ?" + " GROUP BY destination" + ) + txn.execute(sql, (last_pos, upper_pos)) + rows.extend(txn) + + # Order by ascending stream ordering + rows.sort() + + return rows + + return self.runInteraction( + "get_all_new_device_messages", get_all_new_device_messages_txn + ) diff --git a/synapse/storage/data_stores/main/devices.py b/synapse/storage/data_stores/main/devices.py new file mode 100644 index 0000000000..ac5239e50a --- /dev/null +++ b/synapse/storage/data_stores/main/devices.py @@ -0,0 +1,937 @@ +# -*- coding: utf-8 -*- +# Copyright 2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import logging + +from six import iteritems + +from canonicaljson import json + +from twisted.internet import defer + +from synapse.api.errors import StoreError +from synapse.logging.opentracing import ( + get_active_span_text_map, + set_tag, + trace, + whitelisted_homeserver, +) +from synapse.metrics.background_process_metrics import run_as_background_process +from synapse.storage._base import ( + Cache, + SQLBaseStore, + db_to_json, + make_in_list_sql_clause, +) +from synapse.storage.background_updates import BackgroundUpdateStore +from synapse.util import batch_iter +from synapse.util.caches.descriptors import cached, cachedInlineCallbacks, cachedList + +logger = logging.getLogger(__name__) + +DROP_DEVICE_LIST_STREAMS_NON_UNIQUE_INDEXES = ( + "drop_device_list_streams_non_unique_indexes" +) + + +class DeviceWorkerStore(SQLBaseStore): + def get_device(self, user_id, device_id): + """Retrieve a device. + + Args: + user_id (str): The ID of the user which owns the device + device_id (str): The ID of the device to retrieve + Returns: + defer.Deferred for a dict containing the device information + Raises: + StoreError: if the device is not found + """ + return self._simple_select_one( + table="devices", + keyvalues={"user_id": user_id, "device_id": device_id}, + retcols=("user_id", "device_id", "display_name"), + desc="get_device", + ) + + @defer.inlineCallbacks + def get_devices_by_user(self, user_id): + """Retrieve all of a user's registered devices. + + Args: + user_id (str): + Returns: + defer.Deferred: resolves to a dict from device_id to a dict + containing "device_id", "user_id" and "display_name" for each + device. + """ + devices = yield self._simple_select_list( + table="devices", + keyvalues={"user_id": user_id}, + retcols=("user_id", "device_id", "display_name"), + desc="get_devices_by_user", + ) + + return {d["device_id"]: d for d in devices} + + @trace + @defer.inlineCallbacks + def get_devices_by_remote(self, destination, from_stream_id, limit): + """Get stream of updates to send to remote servers + + Returns: + Deferred[tuple[int, list[dict]]]: + current stream id (ie, the stream id of the last update included in the + response), and the list of updates + """ + now_stream_id = self._device_list_id_gen.get_current_token() + + has_changed = self._device_list_federation_stream_cache.has_entity_changed( + destination, int(from_stream_id) + ) + if not has_changed: + return now_stream_id, [] + + # We retrieve n+1 devices from the list of outbound pokes where n is + # our outbound device update limit. We then check if the very last + # device has the same stream_id as the second-to-last device. If so, + # then we ignore all devices with that stream_id and only send the + # devices with a lower stream_id. + # + # If when culling the list we end up with no devices afterwards, we + # consider the device update to be too large, and simply skip the + # stream_id; the rationale being that such a large device list update + # is likely an error. + updates = yield self.runInteraction( + "get_devices_by_remote", + self._get_devices_by_remote_txn, + destination, + from_stream_id, + now_stream_id, + limit + 1, + ) + + # Return an empty list if there are no updates + if not updates: + return now_stream_id, [] + + # if we have exceeded the limit, we need to exclude any results with the + # same stream_id as the last row. + if len(updates) > limit: + stream_id_cutoff = updates[-1][2] + now_stream_id = stream_id_cutoff - 1 + else: + stream_id_cutoff = None + + # Perform the equivalent of a GROUP BY + # + # Iterate through the updates list and copy non-duplicate + # (user_id, device_id) entries into a map, with the value being + # the max stream_id across each set of duplicate entries + # + # maps (user_id, device_id) -> (stream_id, opentracing_context) + # as long as their stream_id does not match that of the last row + # + # opentracing_context contains the opentracing metadata for the request + # that created the poke + # + # The most recent request's opentracing_context is used as the + # context which created the Edu. + + query_map = {} + for update in updates: + if stream_id_cutoff is not None and update[2] >= stream_id_cutoff: + # Stop processing updates + break + + key = (update[0], update[1]) + + update_context = update[3] + update_stream_id = update[2] + + previous_update_stream_id, _ = query_map.get(key, (0, None)) + + if update_stream_id > previous_update_stream_id: + query_map[key] = (update_stream_id, update_context) + + # If we didn't find any updates with a stream_id lower than the cutoff, it + # means that there are more than limit updates all of which have the same + # steam_id. + + # That should only happen if a client is spamming the server with new + # devices, in which case E2E isn't going to work well anyway. We'll just + # skip that stream_id and return an empty list, and continue with the next + # stream_id next time. + if not query_map: + return stream_id_cutoff, [] + + results = yield self._get_device_update_edus_by_remote( + destination, from_stream_id, query_map + ) + + return now_stream_id, results + + def _get_devices_by_remote_txn( + self, txn, destination, from_stream_id, now_stream_id, limit + ): + """Return device update information for a given remote destination + + Args: + txn (LoggingTransaction): The transaction to execute + destination (str): The host the device updates are intended for + from_stream_id (int): The minimum stream_id to filter updates by, exclusive + now_stream_id (int): The maximum stream_id to filter updates by, inclusive + limit (int): Maximum number of device updates to return + + Returns: + List: List of device updates + """ + sql = """ + SELECT user_id, device_id, stream_id, opentracing_context FROM device_lists_outbound_pokes + WHERE destination = ? AND ? < stream_id AND stream_id <= ? AND sent = ? + ORDER BY stream_id + LIMIT ? + """ + txn.execute(sql, (destination, from_stream_id, now_stream_id, False, limit)) + + return list(txn) + + @defer.inlineCallbacks + def _get_device_update_edus_by_remote(self, destination, from_stream_id, query_map): + """Returns a list of device update EDUs as well as E2EE keys + + Args: + destination (str): The host the device updates are intended for + from_stream_id (int): The minimum stream_id to filter updates by, exclusive + query_map (Dict[(str, str): (int, str|None)]): Dictionary mapping + user_id/device_id to update stream_id and the relevent json-encoded + opentracing context + + Returns: + List[Dict]: List of objects representing an device update EDU + + """ + devices = yield self.runInteraction( + "_get_e2e_device_keys_txn", + self._get_e2e_device_keys_txn, + query_map.keys(), + include_all_devices=True, + include_deleted_devices=True, + ) + + results = [] + for user_id, user_devices in iteritems(devices): + # The prev_id for the first row is always the last row before + # `from_stream_id` + prev_id = yield self._get_last_device_update_for_remote_user( + destination, user_id, from_stream_id + ) + for device_id, device in iteritems(user_devices): + stream_id, opentracing_context = query_map[(user_id, device_id)] + result = { + "user_id": user_id, + "device_id": device_id, + "prev_id": [prev_id] if prev_id else [], + "stream_id": stream_id, + "org.matrix.opentracing_context": opentracing_context, + } + + prev_id = stream_id + + if device is not None: + key_json = device.get("key_json", None) + if key_json: + result["keys"] = db_to_json(key_json) + device_display_name = device.get("device_display_name", None) + if device_display_name: + result["device_display_name"] = device_display_name + else: + result["deleted"] = True + + results.append(result) + + return results + + def _get_last_device_update_for_remote_user( + self, destination, user_id, from_stream_id + ): + def f(txn): + prev_sent_id_sql = """ + SELECT coalesce(max(stream_id), 0) as stream_id + FROM device_lists_outbound_last_success + WHERE destination = ? AND user_id = ? AND stream_id <= ? + """ + txn.execute(prev_sent_id_sql, (destination, user_id, from_stream_id)) + rows = txn.fetchall() + return rows[0][0] + + return self.runInteraction("get_last_device_update_for_remote_user", f) + + def mark_as_sent_devices_by_remote(self, destination, stream_id): + """Mark that updates have successfully been sent to the destination. + """ + return self.runInteraction( + "mark_as_sent_devices_by_remote", + self._mark_as_sent_devices_by_remote_txn, + destination, + stream_id, + ) + + def _mark_as_sent_devices_by_remote_txn(self, txn, destination, stream_id): + # We update the device_lists_outbound_last_success with the successfully + # poked users. We do the join to see which users need to be inserted and + # which updated. + sql = """ + SELECT user_id, coalesce(max(o.stream_id), 0), (max(s.stream_id) IS NOT NULL) + FROM device_lists_outbound_pokes as o + LEFT JOIN device_lists_outbound_last_success as s + USING (destination, user_id) + WHERE destination = ? AND o.stream_id <= ? + GROUP BY user_id + """ + txn.execute(sql, (destination, stream_id)) + rows = txn.fetchall() + + sql = """ + UPDATE device_lists_outbound_last_success + SET stream_id = ? + WHERE destination = ? AND user_id = ? + """ + txn.executemany(sql, ((row[1], destination, row[0]) for row in rows if row[2])) + + sql = """ + INSERT INTO device_lists_outbound_last_success + (destination, user_id, stream_id) VALUES (?, ?, ?) + """ + txn.executemany( + sql, ((destination, row[0], row[1]) for row in rows if not row[2]) + ) + + # Delete all sent outbound pokes + sql = """ + DELETE FROM device_lists_outbound_pokes + WHERE destination = ? AND stream_id <= ? + """ + txn.execute(sql, (destination, stream_id)) + + def get_device_stream_token(self): + return self._device_list_id_gen.get_current_token() + + @trace + @defer.inlineCallbacks + def get_user_devices_from_cache(self, query_list): + """Get the devices (and keys if any) for remote users from the cache. + + Args: + query_list(list): List of (user_id, device_ids), if device_ids is + falsey then return all device ids for that user. + + Returns: + (user_ids_not_in_cache, results_map), where user_ids_not_in_cache is + a set of user_ids and results_map is a mapping of + user_id -> device_id -> device_info + """ + user_ids = set(user_id for user_id, _ in query_list) + user_map = yield self.get_device_list_last_stream_id_for_remotes(list(user_ids)) + user_ids_in_cache = set( + user_id for user_id, stream_id in user_map.items() if stream_id + ) + user_ids_not_in_cache = user_ids - user_ids_in_cache + + results = {} + for user_id, device_id in query_list: + if user_id not in user_ids_in_cache: + continue + + if device_id: + device = yield self._get_cached_user_device(user_id, device_id) + results.setdefault(user_id, {})[device_id] = device + else: + results[user_id] = yield self._get_cached_devices_for_user(user_id) + + set_tag("in_cache", results) + set_tag("not_in_cache", user_ids_not_in_cache) + + return user_ids_not_in_cache, results + + @cachedInlineCallbacks(num_args=2, tree=True) + def _get_cached_user_device(self, user_id, device_id): + content = yield self._simple_select_one_onecol( + table="device_lists_remote_cache", + keyvalues={"user_id": user_id, "device_id": device_id}, + retcol="content", + desc="_get_cached_user_device", + ) + return db_to_json(content) + + @cachedInlineCallbacks() + def _get_cached_devices_for_user(self, user_id): + devices = yield self._simple_select_list( + table="device_lists_remote_cache", + keyvalues={"user_id": user_id}, + retcols=("device_id", "content"), + desc="_get_cached_devices_for_user", + ) + return { + device["device_id"]: db_to_json(device["content"]) for device in devices + } + + def get_devices_with_keys_by_user(self, user_id): + """Get all devices (with any device keys) for a user + + Returns: + (stream_id, devices) + """ + return self.runInteraction( + "get_devices_with_keys_by_user", + self._get_devices_with_keys_by_user_txn, + user_id, + ) + + def _get_devices_with_keys_by_user_txn(self, txn, user_id): + now_stream_id = self._device_list_id_gen.get_current_token() + + devices = self._get_e2e_device_keys_txn( + txn, [(user_id, None)], include_all_devices=True + ) + + if devices: + user_devices = devices[user_id] + results = [] + for device_id, device in iteritems(user_devices): + result = {"device_id": device_id} + + key_json = device.get("key_json", None) + if key_json: + result["keys"] = db_to_json(key_json) + device_display_name = device.get("device_display_name", None) + if device_display_name: + result["device_display_name"] = device_display_name + + results.append(result) + + return now_stream_id, results + + return now_stream_id, [] + + def get_users_whose_devices_changed(self, from_key, user_ids): + """Get set of users whose devices have changed since `from_key` that + are in the given list of user_ids. + + Args: + from_key (str): The device lists stream token + user_ids (Iterable[str]) + + Returns: + Deferred[set[str]]: The set of user_ids whose devices have changed + since `from_key` + """ + from_key = int(from_key) + + # Get set of users who *may* have changed. Users not in the returned + # list have definitely not changed. + to_check = list( + self._device_list_stream_cache.get_entities_changed(user_ids, from_key) + ) + + if not to_check: + return defer.succeed(set()) + + def _get_users_whose_devices_changed_txn(txn): + changes = set() + + sql = """ + SELECT DISTINCT user_id FROM device_lists_stream + WHERE stream_id > ? + AND + """ + + for chunk in batch_iter(to_check, 100): + clause, args = make_in_list_sql_clause( + txn.database_engine, "user_id", chunk + ) + txn.execute(sql + clause, (from_key,) + tuple(args)) + changes.update(user_id for user_id, in txn) + + return changes + + return self.runInteraction( + "get_users_whose_devices_changed", _get_users_whose_devices_changed_txn + ) + + def get_all_device_list_changes_for_remotes(self, from_key, to_key): + """Return a list of `(stream_id, user_id, destination)` which is the + combined list of changes to devices, and which destinations need to be + poked. `destination` may be None if no destinations need to be poked. + """ + # We do a group by here as there can be a large number of duplicate + # entries, since we throw away device IDs. + sql = """ + SELECT MAX(stream_id) AS stream_id, user_id, destination + FROM device_lists_stream + LEFT JOIN device_lists_outbound_pokes USING (stream_id, user_id, device_id) + WHERE ? < stream_id AND stream_id <= ? + GROUP BY user_id, destination + """ + return self._execute( + "get_all_device_list_changes_for_remotes", None, sql, from_key, to_key + ) + + @cached(max_entries=10000) + def get_device_list_last_stream_id_for_remote(self, user_id): + """Get the last stream_id we got for a user. May be None if we haven't + got any information for them. + """ + return self._simple_select_one_onecol( + table="device_lists_remote_extremeties", + keyvalues={"user_id": user_id}, + retcol="stream_id", + desc="get_device_list_last_stream_id_for_remote", + allow_none=True, + ) + + @cachedList( + cached_method_name="get_device_list_last_stream_id_for_remote", + list_name="user_ids", + inlineCallbacks=True, + ) + def get_device_list_last_stream_id_for_remotes(self, user_ids): + rows = yield self._simple_select_many_batch( + table="device_lists_remote_extremeties", + column="user_id", + iterable=user_ids, + retcols=("user_id", "stream_id"), + desc="get_device_list_last_stream_id_for_remotes", + ) + + results = {user_id: None for user_id in user_ids} + results.update({row["user_id"]: row["stream_id"] for row in rows}) + + return results + + +class DeviceBackgroundUpdateStore(BackgroundUpdateStore): + def __init__(self, db_conn, hs): + super(DeviceBackgroundUpdateStore, self).__init__(db_conn, hs) + + self.register_background_index_update( + "device_lists_stream_idx", + index_name="device_lists_stream_user_id", + table="device_lists_stream", + columns=["user_id", "device_id"], + ) + + # create a unique index on device_lists_remote_cache + self.register_background_index_update( + "device_lists_remote_cache_unique_idx", + index_name="device_lists_remote_cache_unique_id", + table="device_lists_remote_cache", + columns=["user_id", "device_id"], + unique=True, + ) + + # And one on device_lists_remote_extremeties + self.register_background_index_update( + "device_lists_remote_extremeties_unique_idx", + index_name="device_lists_remote_extremeties_unique_idx", + table="device_lists_remote_extremeties", + columns=["user_id"], + unique=True, + ) + + # once they complete, we can remove the old non-unique indexes. + self.register_background_update_handler( + DROP_DEVICE_LIST_STREAMS_NON_UNIQUE_INDEXES, + self._drop_device_list_streams_non_unique_indexes, + ) + + @defer.inlineCallbacks + def _drop_device_list_streams_non_unique_indexes(self, progress, batch_size): + def f(conn): + txn = conn.cursor() + txn.execute("DROP INDEX IF EXISTS device_lists_remote_cache_id") + txn.execute("DROP INDEX IF EXISTS device_lists_remote_extremeties_id") + txn.close() + + yield self.runWithConnection(f) + yield self._end_background_update(DROP_DEVICE_LIST_STREAMS_NON_UNIQUE_INDEXES) + return 1 + + +class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): + def __init__(self, db_conn, hs): + super(DeviceStore, self).__init__(db_conn, hs) + + # Map of (user_id, device_id) -> bool. If there is an entry that implies + # the device exists. + self.device_id_exists_cache = Cache( + name="device_id_exists", keylen=2, max_entries=10000 + ) + + self._clock.looping_call(self._prune_old_outbound_device_pokes, 60 * 60 * 1000) + + @defer.inlineCallbacks + def store_device(self, user_id, device_id, initial_device_display_name): + """Ensure the given device is known; add it to the store if not + + Args: + user_id (str): id of user associated with the device + device_id (str): id of device + initial_device_display_name (str): initial displayname of the + device. Ignored if device exists. + Returns: + defer.Deferred: boolean whether the device was inserted or an + existing device existed with that ID. + """ + key = (user_id, device_id) + if self.device_id_exists_cache.get(key, None): + return False + + try: + inserted = yield self._simple_insert( + "devices", + values={ + "user_id": user_id, + "device_id": device_id, + "display_name": initial_device_display_name, + }, + desc="store_device", + or_ignore=True, + ) + self.device_id_exists_cache.prefill(key, True) + return inserted + except Exception as e: + logger.error( + "store_device with device_id=%s(%r) user_id=%s(%r)" + " display_name=%s(%r) failed: %s", + type(device_id).__name__, + device_id, + type(user_id).__name__, + user_id, + type(initial_device_display_name).__name__, + initial_device_display_name, + e, + ) + raise StoreError(500, "Problem storing device.") + + @defer.inlineCallbacks + def delete_device(self, user_id, device_id): + """Delete a device. + + Args: + user_id (str): The ID of the user which owns the device + device_id (str): The ID of the device to delete + Returns: + defer.Deferred + """ + yield self._simple_delete_one( + table="devices", + keyvalues={"user_id": user_id, "device_id": device_id}, + desc="delete_device", + ) + + self.device_id_exists_cache.invalidate((user_id, device_id)) + + @defer.inlineCallbacks + def delete_devices(self, user_id, device_ids): + """Deletes several devices. + + Args: + user_id (str): The ID of the user which owns the devices + device_ids (list): The IDs of the devices to delete + Returns: + defer.Deferred + """ + yield self._simple_delete_many( + table="devices", + column="device_id", + iterable=device_ids, + keyvalues={"user_id": user_id}, + desc="delete_devices", + ) + for device_id in device_ids: + self.device_id_exists_cache.invalidate((user_id, device_id)) + + def update_device(self, user_id, device_id, new_display_name=None): + """Update a device. + + Args: + user_id (str): The ID of the user which owns the device + device_id (str): The ID of the device to update + new_display_name (str|None): new displayname for device; None + to leave unchanged + Raises: + StoreError: if the device is not found + Returns: + defer.Deferred + """ + updates = {} + if new_display_name is not None: + updates["display_name"] = new_display_name + if not updates: + return defer.succeed(None) + return self._simple_update_one( + table="devices", + keyvalues={"user_id": user_id, "device_id": device_id}, + updatevalues=updates, + desc="update_device", + ) + + @defer.inlineCallbacks + def mark_remote_user_device_list_as_unsubscribed(self, user_id): + """Mark that we no longer track device lists for remote user. + """ + yield self._simple_delete( + table="device_lists_remote_extremeties", + keyvalues={"user_id": user_id}, + desc="mark_remote_user_device_list_as_unsubscribed", + ) + self.get_device_list_last_stream_id_for_remote.invalidate((user_id,)) + + def update_remote_device_list_cache_entry( + self, user_id, device_id, content, stream_id + ): + """Updates a single device in the cache of a remote user's devicelist. + + Note: assumes that we are the only thread that can be updating this user's + device list. + + Args: + user_id (str): User to update device list for + device_id (str): ID of decivice being updated + content (dict): new data on this device + stream_id (int): the version of the device list + + Returns: + Deferred[None] + """ + return self.runInteraction( + "update_remote_device_list_cache_entry", + self._update_remote_device_list_cache_entry_txn, + user_id, + device_id, + content, + stream_id, + ) + + def _update_remote_device_list_cache_entry_txn( + self, txn, user_id, device_id, content, stream_id + ): + if content.get("deleted"): + self._simple_delete_txn( + txn, + table="device_lists_remote_cache", + keyvalues={"user_id": user_id, "device_id": device_id}, + ) + + txn.call_after(self.device_id_exists_cache.invalidate, (user_id, device_id)) + else: + self._simple_upsert_txn( + txn, + table="device_lists_remote_cache", + keyvalues={"user_id": user_id, "device_id": device_id}, + values={"content": json.dumps(content)}, + # we don't need to lock, because we assume we are the only thread + # updating this user's devices. + lock=False, + ) + + txn.call_after(self._get_cached_user_device.invalidate, (user_id, device_id)) + txn.call_after(self._get_cached_devices_for_user.invalidate, (user_id,)) + txn.call_after( + self.get_device_list_last_stream_id_for_remote.invalidate, (user_id,) + ) + + self._simple_upsert_txn( + txn, + table="device_lists_remote_extremeties", + keyvalues={"user_id": user_id}, + values={"stream_id": stream_id}, + # again, we can assume we are the only thread updating this user's + # extremity. + lock=False, + ) + + def update_remote_device_list_cache(self, user_id, devices, stream_id): + """Replace the entire cache of the remote user's devices. + + Note: assumes that we are the only thread that can be updating this user's + device list. + + Args: + user_id (str): User to update device list for + devices (list[dict]): list of device objects supplied over federation + stream_id (int): the version of the device list + + Returns: + Deferred[None] + """ + return self.runInteraction( + "update_remote_device_list_cache", + self._update_remote_device_list_cache_txn, + user_id, + devices, + stream_id, + ) + + def _update_remote_device_list_cache_txn(self, txn, user_id, devices, stream_id): + self._simple_delete_txn( + txn, table="device_lists_remote_cache", keyvalues={"user_id": user_id} + ) + + self._simple_insert_many_txn( + txn, + table="device_lists_remote_cache", + values=[ + { + "user_id": user_id, + "device_id": content["device_id"], + "content": json.dumps(content), + } + for content in devices + ], + ) + + txn.call_after(self._get_cached_devices_for_user.invalidate, (user_id,)) + txn.call_after(self._get_cached_user_device.invalidate_many, (user_id,)) + txn.call_after( + self.get_device_list_last_stream_id_for_remote.invalidate, (user_id,) + ) + + self._simple_upsert_txn( + txn, + table="device_lists_remote_extremeties", + keyvalues={"user_id": user_id}, + values={"stream_id": stream_id}, + # we don't need to lock, because we can assume we are the only thread + # updating this user's extremity. + lock=False, + ) + + @defer.inlineCallbacks + def add_device_change_to_streams(self, user_id, device_ids, hosts): + """Persist that a user's devices have been updated, and which hosts + (if any) should be poked. + """ + with self._device_list_id_gen.get_next() as stream_id: + yield self.runInteraction( + "add_device_change_to_streams", + self._add_device_change_txn, + user_id, + device_ids, + hosts, + stream_id, + ) + return stream_id + + def _add_device_change_txn(self, txn, user_id, device_ids, hosts, stream_id): + now = self._clock.time_msec() + + txn.call_after( + self._device_list_stream_cache.entity_has_changed, user_id, stream_id + ) + for host in hosts: + txn.call_after( + self._device_list_federation_stream_cache.entity_has_changed, + host, + stream_id, + ) + + # Delete older entries in the table, as we really only care about + # when the latest change happened. + txn.executemany( + """ + DELETE FROM device_lists_stream + WHERE user_id = ? AND device_id = ? AND stream_id < ? + """, + [(user_id, device_id, stream_id) for device_id in device_ids], + ) + + self._simple_insert_many_txn( + txn, + table="device_lists_stream", + values=[ + {"stream_id": stream_id, "user_id": user_id, "device_id": device_id} + for device_id in device_ids + ], + ) + + context = get_active_span_text_map() + + self._simple_insert_many_txn( + txn, + table="device_lists_outbound_pokes", + values=[ + { + "destination": destination, + "stream_id": stream_id, + "user_id": user_id, + "device_id": device_id, + "sent": False, + "ts": now, + "opentracing_context": json.dumps(context) + if whitelisted_homeserver(destination) + else "{}", + } + for destination in hosts + for device_id in device_ids + ], + ) + + def _prune_old_outbound_device_pokes(self): + """Delete old entries out of the device_lists_outbound_pokes to ensure + that we don't fill up due to dead servers. We keep one entry per + (destination, user_id) tuple to ensure that the prev_ids remain correct + if the server does come back. + """ + yesterday = self._clock.time_msec() - 24 * 60 * 60 * 1000 + + def _prune_txn(txn): + select_sql = """ + SELECT destination, user_id, max(stream_id) as stream_id + FROM device_lists_outbound_pokes + GROUP BY destination, user_id + HAVING min(ts) < ? AND count(*) > 1 + """ + + txn.execute(select_sql, (yesterday,)) + rows = txn.fetchall() + + if not rows: + return + + delete_sql = """ + DELETE FROM device_lists_outbound_pokes + WHERE ts < ? AND destination = ? AND user_id = ? AND stream_id < ? + """ + + txn.executemany( + delete_sql, ((yesterday, row[0], row[1], row[2]) for row in rows) + ) + + # Since we've deleted unsent deltas, we need to remove the entry + # of last successful sent so that the prev_ids are correctly set. + sql = """ + DELETE FROM device_lists_outbound_last_success + WHERE destination = ? AND user_id = ? + """ + txn.executemany(sql, ((row[0], row[1]) for row in rows)) + + logger.info("Pruned %d device list outbound pokes", txn.rowcount) + + return run_as_background_process( + "prune_old_outbound_device_pokes", + self.runInteraction, + "_prune_old_outbound_device_pokes", + _prune_txn, + ) diff --git a/synapse/storage/data_stores/main/directory.py b/synapse/storage/data_stores/main/directory.py new file mode 100644 index 0000000000..297966d9f4 --- /dev/null +++ b/synapse/storage/data_stores/main/directory.py @@ -0,0 +1,173 @@ +# -*- coding: utf-8 -*- +# Copyright 2014-2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from collections import namedtuple + +from twisted.internet import defer + +from synapse.api.errors import SynapseError +from synapse.storage._base import SQLBaseStore +from synapse.util.caches.descriptors import cached + +RoomAliasMapping = namedtuple("RoomAliasMapping", ("room_id", "room_alias", "servers")) + + +class DirectoryWorkerStore(SQLBaseStore): + @defer.inlineCallbacks + def get_association_from_room_alias(self, room_alias): + """ Get's the room_id and server list for a given room_alias + + Args: + room_alias (RoomAlias) + + Returns: + Deferred: results in namedtuple with keys "room_id" and + "servers" or None if no association can be found + """ + room_id = yield self._simple_select_one_onecol( + "room_aliases", + {"room_alias": room_alias.to_string()}, + "room_id", + allow_none=True, + desc="get_association_from_room_alias", + ) + + if not room_id: + return None + + servers = yield self._simple_select_onecol( + "room_alias_servers", + {"room_alias": room_alias.to_string()}, + "server", + desc="get_association_from_room_alias", + ) + + if not servers: + return None + + return RoomAliasMapping(room_id, room_alias.to_string(), servers) + + def get_room_alias_creator(self, room_alias): + return self._simple_select_one_onecol( + table="room_aliases", + keyvalues={"room_alias": room_alias}, + retcol="creator", + desc="get_room_alias_creator", + ) + + @cached(max_entries=5000) + def get_aliases_for_room(self, room_id): + return self._simple_select_onecol( + "room_aliases", + {"room_id": room_id}, + "room_alias", + desc="get_aliases_for_room", + ) + + +class DirectoryStore(DirectoryWorkerStore): + @defer.inlineCallbacks + def create_room_alias_association(self, room_alias, room_id, servers, creator=None): + """ Creates an association between a room alias and room_id/servers + + Args: + room_alias (RoomAlias) + room_id (str) + servers (list) + creator (str): Optional user_id of creator. + + Returns: + Deferred + """ + + def alias_txn(txn): + self._simple_insert_txn( + txn, + "room_aliases", + { + "room_alias": room_alias.to_string(), + "room_id": room_id, + "creator": creator, + }, + ) + + self._simple_insert_many_txn( + txn, + table="room_alias_servers", + values=[ + {"room_alias": room_alias.to_string(), "server": server} + for server in servers + ], + ) + + self._invalidate_cache_and_stream( + txn, self.get_aliases_for_room, (room_id,) + ) + + try: + ret = yield self.runInteraction("create_room_alias_association", alias_txn) + except self.database_engine.module.IntegrityError: + raise SynapseError( + 409, "Room alias %s already exists" % room_alias.to_string() + ) + return ret + + @defer.inlineCallbacks + def delete_room_alias(self, room_alias): + room_id = yield self.runInteraction( + "delete_room_alias", self._delete_room_alias_txn, room_alias + ) + + return room_id + + def _delete_room_alias_txn(self, txn, room_alias): + txn.execute( + "SELECT room_id FROM room_aliases WHERE room_alias = ?", + (room_alias.to_string(),), + ) + + res = txn.fetchone() + if res: + room_id = res[0] + else: + return None + + txn.execute( + "DELETE FROM room_aliases WHERE room_alias = ?", (room_alias.to_string(),) + ) + + txn.execute( + "DELETE FROM room_alias_servers WHERE room_alias = ?", + (room_alias.to_string(),), + ) + + self._invalidate_cache_and_stream(txn, self.get_aliases_for_room, (room_id,)) + + return room_id + + def update_aliases_for_room(self, old_room_id, new_room_id, creator): + def _update_aliases_for_room_txn(txn): + sql = "UPDATE room_aliases SET room_id = ?, creator = ? WHERE room_id = ?" + txn.execute(sql, (new_room_id, creator, old_room_id)) + self._invalidate_cache_and_stream( + txn, self.get_aliases_for_room, (old_room_id,) + ) + self._invalidate_cache_and_stream( + txn, self.get_aliases_for_room, (new_room_id,) + ) + + return self.runInteraction( + "_update_aliases_for_room_txn", _update_aliases_for_room_txn + ) diff --git a/synapse/storage/data_stores/main/e2e_room_keys.py b/synapse/storage/data_stores/main/e2e_room_keys.py new file mode 100644 index 0000000000..ef88e79293 --- /dev/null +++ b/synapse/storage/data_stores/main/e2e_room_keys.py @@ -0,0 +1,336 @@ +# -*- coding: utf-8 -*- +# Copyright 2017 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import json + +from twisted.internet import defer + +from synapse.api.errors import StoreError +from synapse.logging.opentracing import log_kv, trace +from synapse.storage._base import SQLBaseStore + + +class EndToEndRoomKeyStore(SQLBaseStore): + @defer.inlineCallbacks + def get_e2e_room_key(self, user_id, version, room_id, session_id): + """Get the encrypted E2E room key for a given session from a given + backup version of room_keys. We only store the 'best' room key for a given + session at a given time, as determined by the handler. + + Args: + user_id(str): the user whose backup we're querying + version(str): the version ID of the backup for the set of keys we're querying + room_id(str): the ID of the room whose keys we're querying. + This is a bit redundant as it's implied by the session_id, but + we include for consistency with the rest of the API. + session_id(str): the session whose room_key we're querying. + + Returns: + A deferred dict giving the session_data and message metadata for + this room key. + """ + + row = yield self._simple_select_one( + table="e2e_room_keys", + keyvalues={ + "user_id": user_id, + "version": version, + "room_id": room_id, + "session_id": session_id, + }, + retcols=( + "first_message_index", + "forwarded_count", + "is_verified", + "session_data", + ), + desc="get_e2e_room_key", + ) + + row["session_data"] = json.loads(row["session_data"]) + + return row + + @defer.inlineCallbacks + def set_e2e_room_key(self, user_id, version, room_id, session_id, room_key): + """Replaces or inserts the encrypted E2E room key for a given session in + a given backup + + Args: + user_id(str): the user whose backup we're setting + version(str): the version ID of the backup we're updating + room_id(str): the ID of the room whose keys we're setting + session_id(str): the session whose room_key we're setting + room_key(dict): the room_key being set + Raises: + StoreError + """ + + yield self._simple_upsert( + table="e2e_room_keys", + keyvalues={ + "user_id": user_id, + "version": version, + "room_id": room_id, + "session_id": session_id, + }, + values={ + "first_message_index": room_key["first_message_index"], + "forwarded_count": room_key["forwarded_count"], + "is_verified": room_key["is_verified"], + "session_data": json.dumps(room_key["session_data"]), + }, + lock=False, + ) + log_kv( + { + "message": "Set room key", + "room_id": room_id, + "session_id": session_id, + "room_key": room_key, + } + ) + + @trace + @defer.inlineCallbacks + def get_e2e_room_keys(self, user_id, version, room_id=None, session_id=None): + """Bulk get the E2E room keys for a given backup, optionally filtered to a given + room, or a given session. + + Args: + user_id(str): the user whose backup we're querying + version(str): the version ID of the backup for the set of keys we're querying + room_id(str): Optional. the ID of the room whose keys we're querying, if any. + If not specified, we return the keys for all the rooms in the backup. + session_id(str): Optional. the session whose room_key we're querying, if any. + If specified, we also require the room_id to be specified. + If not specified, we return all the keys in this version of + the backup (or for the specified room) + + Returns: + A deferred list of dicts giving the session_data and message metadata for + these room keys. + """ + + try: + version = int(version) + except ValueError: + return {"rooms": {}} + + keyvalues = {"user_id": user_id, "version": version} + if room_id: + keyvalues["room_id"] = room_id + if session_id: + keyvalues["session_id"] = session_id + + rows = yield self._simple_select_list( + table="e2e_room_keys", + keyvalues=keyvalues, + retcols=( + "user_id", + "room_id", + "session_id", + "first_message_index", + "forwarded_count", + "is_verified", + "session_data", + ), + desc="get_e2e_room_keys", + ) + + sessions = {"rooms": {}} + for row in rows: + room_entry = sessions["rooms"].setdefault(row["room_id"], {"sessions": {}}) + room_entry["sessions"][row["session_id"]] = { + "first_message_index": row["first_message_index"], + "forwarded_count": row["forwarded_count"], + "is_verified": row["is_verified"], + "session_data": json.loads(row["session_data"]), + } + + return sessions + + @trace + @defer.inlineCallbacks + def delete_e2e_room_keys(self, user_id, version, room_id=None, session_id=None): + """Bulk delete the E2E room keys for a given backup, optionally filtered to a given + room or a given session. + + Args: + user_id(str): the user whose backup we're deleting from + version(str): the version ID of the backup for the set of keys we're deleting + room_id(str): Optional. the ID of the room whose keys we're deleting, if any. + If not specified, we delete the keys for all the rooms in the backup. + session_id(str): Optional. the session whose room_key we're querying, if any. + If specified, we also require the room_id to be specified. + If not specified, we delete all the keys in this version of + the backup (or for the specified room) + + Returns: + A deferred of the deletion transaction + """ + + keyvalues = {"user_id": user_id, "version": int(version)} + if room_id: + keyvalues["room_id"] = room_id + if session_id: + keyvalues["session_id"] = session_id + + yield self._simple_delete( + table="e2e_room_keys", keyvalues=keyvalues, desc="delete_e2e_room_keys" + ) + + @staticmethod + def _get_current_version(txn, user_id): + txn.execute( + "SELECT MAX(version) FROM e2e_room_keys_versions " + "WHERE user_id=? AND deleted=0", + (user_id,), + ) + row = txn.fetchone() + if not row: + raise StoreError(404, "No current backup version") + return row[0] + + def get_e2e_room_keys_version_info(self, user_id, version=None): + """Get info metadata about a version of our room_keys backup. + + Args: + user_id(str): the user whose backup we're querying + version(str): Optional. the version ID of the backup we're querying about + If missing, we return the information about the current version. + Raises: + StoreError: with code 404 if there are no e2e_room_keys_versions present + Returns: + A deferred dict giving the info metadata for this backup version, with + fields including: + version(str) + algorithm(str) + auth_data(object): opaque dict supplied by the client + """ + + def _get_e2e_room_keys_version_info_txn(txn): + if version is None: + this_version = self._get_current_version(txn, user_id) + else: + try: + this_version = int(version) + except ValueError: + # Our versions are all ints so if we can't convert it to an integer, + # it isn't there. + raise StoreError(404, "No row found") + + result = self._simple_select_one_txn( + txn, + table="e2e_room_keys_versions", + keyvalues={"user_id": user_id, "version": this_version, "deleted": 0}, + retcols=("version", "algorithm", "auth_data"), + ) + result["auth_data"] = json.loads(result["auth_data"]) + result["version"] = str(result["version"]) + return result + + return self.runInteraction( + "get_e2e_room_keys_version_info", _get_e2e_room_keys_version_info_txn + ) + + @trace + def create_e2e_room_keys_version(self, user_id, info): + """Atomically creates a new version of this user's e2e_room_keys store + with the given version info. + + Args: + user_id(str): the user whose backup we're creating a version + info(dict): the info about the backup version to be created + + Returns: + A deferred string for the newly created version ID + """ + + def _create_e2e_room_keys_version_txn(txn): + txn.execute( + "SELECT MAX(version) FROM e2e_room_keys_versions WHERE user_id=?", + (user_id,), + ) + current_version = txn.fetchone()[0] + if current_version is None: + current_version = "0" + + new_version = str(int(current_version) + 1) + + self._simple_insert_txn( + txn, + table="e2e_room_keys_versions", + values={ + "user_id": user_id, + "version": new_version, + "algorithm": info["algorithm"], + "auth_data": json.dumps(info["auth_data"]), + }, + ) + + return new_version + + return self.runInteraction( + "create_e2e_room_keys_version_txn", _create_e2e_room_keys_version_txn + ) + + @trace + def update_e2e_room_keys_version(self, user_id, version, info): + """Update a given backup version + + Args: + user_id(str): the user whose backup version we're updating + version(str): the version ID of the backup version we're updating + info(dict): the new backup version info to store + """ + + return self._simple_update( + table="e2e_room_keys_versions", + keyvalues={"user_id": user_id, "version": version}, + updatevalues={"auth_data": json.dumps(info["auth_data"])}, + desc="update_e2e_room_keys_version", + ) + + @trace + def delete_e2e_room_keys_version(self, user_id, version=None): + """Delete a given backup version of the user's room keys. + Doesn't delete their actual key data. + + Args: + user_id(str): the user whose backup version we're deleting + version(str): Optional. the version ID of the backup version we're deleting + If missing, we delete the current backup version info. + Raises: + StoreError: with code 404 if there are no e2e_room_keys_versions present, + or if the version requested doesn't exist. + """ + + def _delete_e2e_room_keys_version_txn(txn): + if version is None: + this_version = self._get_current_version(txn, user_id) + else: + this_version = version + + return self._simple_update_one_txn( + txn, + table="e2e_room_keys_versions", + keyvalues={"user_id": user_id, "version": this_version}, + updatevalues={"deleted": 1}, + ) + + return self.runInteraction( + "delete_e2e_room_keys_version", _delete_e2e_room_keys_version_txn + ) diff --git a/synapse/storage/data_stores/main/end_to_end_keys.py b/synapse/storage/data_stores/main/end_to_end_keys.py new file mode 100644 index 0000000000..7e44f41046 --- /dev/null +++ b/synapse/storage/data_stores/main/end_to_end_keys.py @@ -0,0 +1,322 @@ +# -*- coding: utf-8 -*- +# Copyright 2015, 2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +from six import iteritems + +from canonicaljson import encode_canonical_json + +from twisted.internet import defer + +from synapse.logging.opentracing import log_kv, set_tag, trace +from synapse.storage._base import SQLBaseStore, db_to_json +from synapse.util.caches.descriptors import cached + + +class EndToEndKeyWorkerStore(SQLBaseStore): + @trace + @defer.inlineCallbacks + def get_e2e_device_keys( + self, query_list, include_all_devices=False, include_deleted_devices=False + ): + """Fetch a list of device keys. + Args: + query_list(list): List of pairs of user_ids and device_ids. + include_all_devices (bool): whether to include entries for devices + that don't have device keys + include_deleted_devices (bool): whether to include null entries for + devices which no longer exist (but were in the query_list). + This option only takes effect if include_all_devices is true. + Returns: + Dict mapping from user-id to dict mapping from device_id to + key data. The key data will be a dict in the same format as the + DeviceKeys type returned by POST /_matrix/client/r0/keys/query. + """ + set_tag("query_list", query_list) + if not query_list: + return {} + + results = yield self.runInteraction( + "get_e2e_device_keys", + self._get_e2e_device_keys_txn, + query_list, + include_all_devices, + include_deleted_devices, + ) + + # Build the result structure, un-jsonify the results, and add the + # "unsigned" section + rv = {} + for user_id, device_keys in iteritems(results): + rv[user_id] = {} + for device_id, device_info in iteritems(device_keys): + r = db_to_json(device_info.pop("key_json")) + r["unsigned"] = {} + display_name = device_info["device_display_name"] + if display_name is not None: + r["unsigned"]["device_display_name"] = display_name + rv[user_id][device_id] = r + + return rv + + @trace + def _get_e2e_device_keys_txn( + self, txn, query_list, include_all_devices=False, include_deleted_devices=False + ): + set_tag("include_all_devices", include_all_devices) + set_tag("include_deleted_devices", include_deleted_devices) + + query_clauses = [] + query_params = [] + + if include_all_devices is False: + include_deleted_devices = False + + if include_deleted_devices: + deleted_devices = set(query_list) + + for (user_id, device_id) in query_list: + query_clause = "user_id = ?" + query_params.append(user_id) + + if device_id is not None: + query_clause += " AND device_id = ?" + query_params.append(device_id) + + query_clauses.append(query_clause) + + sql = ( + "SELECT user_id, device_id, " + " d.display_name AS device_display_name, " + " k.key_json" + " FROM devices d" + " %s JOIN e2e_device_keys_json k USING (user_id, device_id)" + " WHERE %s" + ) % ( + "LEFT" if include_all_devices else "INNER", + " OR ".join("(" + q + ")" for q in query_clauses), + ) + + txn.execute(sql, query_params) + rows = self.cursor_to_dict(txn) + + result = {} + for row in rows: + if include_deleted_devices: + deleted_devices.remove((row["user_id"], row["device_id"])) + result.setdefault(row["user_id"], {})[row["device_id"]] = row + + if include_deleted_devices: + for user_id, device_id in deleted_devices: + result.setdefault(user_id, {})[device_id] = None + + log_kv(result) + return result + + @defer.inlineCallbacks + def get_e2e_one_time_keys(self, user_id, device_id, key_ids): + """Retrieve a number of one-time keys for a user + + Args: + user_id(str): id of user to get keys for + device_id(str): id of device to get keys for + key_ids(list[str]): list of key ids (excluding algorithm) to + retrieve + + Returns: + deferred resolving to Dict[(str, str), str]: map from (algorithm, + key_id) to json string for key + """ + + rows = yield self._simple_select_many_batch( + table="e2e_one_time_keys_json", + column="key_id", + iterable=key_ids, + retcols=("algorithm", "key_id", "key_json"), + keyvalues={"user_id": user_id, "device_id": device_id}, + desc="add_e2e_one_time_keys_check", + ) + result = {(row["algorithm"], row["key_id"]): row["key_json"] for row in rows} + log_kv({"message": "Fetched one time keys for user", "one_time_keys": result}) + return result + + @defer.inlineCallbacks + def add_e2e_one_time_keys(self, user_id, device_id, time_now, new_keys): + """Insert some new one time keys for a device. Errors if any of the + keys already exist. + + Args: + user_id(str): id of user to get keys for + device_id(str): id of device to get keys for + time_now(long): insertion time to record (ms since epoch) + new_keys(iterable[(str, str, str)]: keys to add - each a tuple of + (algorithm, key_id, key json) + """ + + def _add_e2e_one_time_keys(txn): + set_tag("user_id", user_id) + set_tag("device_id", device_id) + set_tag("new_keys", new_keys) + # We are protected from race between lookup and insertion due to + # a unique constraint. If there is a race of two calls to + # `add_e2e_one_time_keys` then they'll conflict and we will only + # insert one set. + self._simple_insert_many_txn( + txn, + table="e2e_one_time_keys_json", + values=[ + { + "user_id": user_id, + "device_id": device_id, + "algorithm": algorithm, + "key_id": key_id, + "ts_added_ms": time_now, + "key_json": json_bytes, + } + for algorithm, key_id, json_bytes in new_keys + ], + ) + self._invalidate_cache_and_stream( + txn, self.count_e2e_one_time_keys, (user_id, device_id) + ) + + yield self.runInteraction( + "add_e2e_one_time_keys_insert", _add_e2e_one_time_keys + ) + + @cached(max_entries=10000) + def count_e2e_one_time_keys(self, user_id, device_id): + """ Count the number of one time keys the server has for a device + Returns: + Dict mapping from algorithm to number of keys for that algorithm. + """ + + def _count_e2e_one_time_keys(txn): + sql = ( + "SELECT algorithm, COUNT(key_id) FROM e2e_one_time_keys_json" + " WHERE user_id = ? AND device_id = ?" + " GROUP BY algorithm" + ) + txn.execute(sql, (user_id, device_id)) + result = {} + for algorithm, key_count in txn: + result[algorithm] = key_count + return result + + return self.runInteraction("count_e2e_one_time_keys", _count_e2e_one_time_keys) + + +class EndToEndKeyStore(EndToEndKeyWorkerStore, SQLBaseStore): + def set_e2e_device_keys(self, user_id, device_id, time_now, device_keys): + """Stores device keys for a device. Returns whether there was a change + or the keys were already in the database. + """ + + def _set_e2e_device_keys_txn(txn): + set_tag("user_id", user_id) + set_tag("device_id", device_id) + set_tag("time_now", time_now) + set_tag("device_keys", device_keys) + + old_key_json = self._simple_select_one_onecol_txn( + txn, + table="e2e_device_keys_json", + keyvalues={"user_id": user_id, "device_id": device_id}, + retcol="key_json", + allow_none=True, + ) + + # In py3 we need old_key_json to match new_key_json type. The DB + # returns unicode while encode_canonical_json returns bytes. + new_key_json = encode_canonical_json(device_keys).decode("utf-8") + + if old_key_json == new_key_json: + log_kv({"Message": "Device key already stored."}) + return False + + self._simple_upsert_txn( + txn, + table="e2e_device_keys_json", + keyvalues={"user_id": user_id, "device_id": device_id}, + values={"ts_added_ms": time_now, "key_json": new_key_json}, + ) + log_kv({"message": "Device keys stored."}) + return True + + return self.runInteraction("set_e2e_device_keys", _set_e2e_device_keys_txn) + + def claim_e2e_one_time_keys(self, query_list): + """Take a list of one time keys out of the database""" + + @trace + def _claim_e2e_one_time_keys(txn): + sql = ( + "SELECT key_id, key_json FROM e2e_one_time_keys_json" + " WHERE user_id = ? AND device_id = ? AND algorithm = ?" + " LIMIT 1" + ) + result = {} + delete = [] + for user_id, device_id, algorithm in query_list: + user_result = result.setdefault(user_id, {}) + device_result = user_result.setdefault(device_id, {}) + txn.execute(sql, (user_id, device_id, algorithm)) + for key_id, key_json in txn: + device_result[algorithm + ":" + key_id] = key_json + delete.append((user_id, device_id, algorithm, key_id)) + sql = ( + "DELETE FROM e2e_one_time_keys_json" + " WHERE user_id = ? AND device_id = ? AND algorithm = ?" + " AND key_id = ?" + ) + for user_id, device_id, algorithm, key_id in delete: + log_kv( + { + "message": "Executing claim e2e_one_time_keys transaction on database." + } + ) + txn.execute(sql, (user_id, device_id, algorithm, key_id)) + log_kv({"message": "finished executing and invalidating cache"}) + self._invalidate_cache_and_stream( + txn, self.count_e2e_one_time_keys, (user_id, device_id) + ) + return result + + return self.runInteraction("claim_e2e_one_time_keys", _claim_e2e_one_time_keys) + + def delete_e2e_keys_by_device(self, user_id, device_id): + def delete_e2e_keys_by_device_txn(txn): + log_kv( + { + "message": "Deleting keys for device", + "device_id": device_id, + "user_id": user_id, + } + ) + self._simple_delete_txn( + txn, + table="e2e_device_keys_json", + keyvalues={"user_id": user_id, "device_id": device_id}, + ) + self._simple_delete_txn( + txn, + table="e2e_one_time_keys_json", + keyvalues={"user_id": user_id, "device_id": device_id}, + ) + self._invalidate_cache_and_stream( + txn, self.count_e2e_one_time_keys, (user_id, device_id) + ) + + return self.runInteraction( + "delete_e2e_keys_by_device", delete_e2e_keys_by_device_txn + ) diff --git a/synapse/storage/data_stores/main/event_federation.py b/synapse/storage/data_stores/main/event_federation.py new file mode 100644 index 0000000000..a470a48e0f --- /dev/null +++ b/synapse/storage/data_stores/main/event_federation.py @@ -0,0 +1,672 @@ +# -*- coding: utf-8 -*- +# Copyright 2014-2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import itertools +import logging +import random + +from six.moves import range +from six.moves.queue import Empty, PriorityQueue + +from unpaddedbase64 import encode_base64 + +from twisted.internet import defer + +from synapse.api.errors import StoreError +from synapse.metrics.background_process_metrics import run_as_background_process +from synapse.storage._base import SQLBaseStore, make_in_list_sql_clause +from synapse.storage.data_stores.main.events_worker import EventsWorkerStore +from synapse.storage.data_stores.main.signatures import SignatureWorkerStore +from synapse.util.caches.descriptors import cached + +logger = logging.getLogger(__name__) + + +class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBaseStore): + def get_auth_chain(self, event_ids, include_given=False): + """Get auth events for given event_ids. The events *must* be state events. + + Args: + event_ids (list): state events + include_given (bool): include the given events in result + + Returns: + list of events + """ + return self.get_auth_chain_ids( + event_ids, include_given=include_given + ).addCallback(self.get_events_as_list) + + def get_auth_chain_ids(self, event_ids, include_given=False): + """Get auth events for given event_ids. The events *must* be state events. + + Args: + event_ids (list): state events + include_given (bool): include the given events in result + + Returns: + list of event_ids + """ + return self.runInteraction( + "get_auth_chain_ids", self._get_auth_chain_ids_txn, event_ids, include_given + ) + + def _get_auth_chain_ids_txn(self, txn, event_ids, include_given): + if include_given: + results = set(event_ids) + else: + results = set() + + base_sql = "SELECT auth_id FROM event_auth WHERE " + + front = set(event_ids) + while front: + new_front = set() + front_list = list(front) + chunks = [front_list[x : x + 100] for x in range(0, len(front), 100)] + for chunk in chunks: + clause, args = make_in_list_sql_clause( + txn.database_engine, "event_id", chunk + ) + txn.execute(base_sql + clause, list(args)) + new_front.update([r[0] for r in txn]) + + new_front -= results + + front = new_front + results.update(front) + + return list(results) + + def get_oldest_events_in_room(self, room_id): + return self.runInteraction( + "get_oldest_events_in_room", self._get_oldest_events_in_room_txn, room_id + ) + + def get_oldest_events_with_depth_in_room(self, room_id): + return self.runInteraction( + "get_oldest_events_with_depth_in_room", + self.get_oldest_events_with_depth_in_room_txn, + room_id, + ) + + def get_oldest_events_with_depth_in_room_txn(self, txn, room_id): + sql = ( + "SELECT b.event_id, MAX(e.depth) FROM events as e" + " INNER JOIN event_edges as g" + " ON g.event_id = e.event_id" + " INNER JOIN event_backward_extremities as b" + " ON g.prev_event_id = b.event_id" + " WHERE b.room_id = ? AND g.is_state is ?" + " GROUP BY b.event_id" + ) + + txn.execute(sql, (room_id, False)) + + return dict(txn) + + @defer.inlineCallbacks + def get_max_depth_of(self, event_ids): + """Returns the max depth of a set of event IDs + + Args: + event_ids (list[str]) + + Returns + Deferred[int] + """ + rows = yield self._simple_select_many_batch( + table="events", + column="event_id", + iterable=event_ids, + retcols=("depth",), + desc="get_max_depth_of", + ) + + if not rows: + return 0 + else: + return max(row["depth"] for row in rows) + + def _get_oldest_events_in_room_txn(self, txn, room_id): + return self._simple_select_onecol_txn( + txn, + table="event_backward_extremities", + keyvalues={"room_id": room_id}, + retcol="event_id", + ) + + @defer.inlineCallbacks + def get_prev_events_for_room(self, room_id): + """ + Gets a subset of the current forward extremities in the given room. + + Limits the result to 10 extremities, so that we can avoid creating + events which refer to hundreds of prev_events. + + Args: + room_id (str): room_id + + Returns: + Deferred[list[(str, dict[str, str], int)]] + for each event, a tuple of (event_id, hashes, depth) + where *hashes* is a map from algorithm to hash. + """ + res = yield self.get_latest_event_ids_and_hashes_in_room(room_id) + if len(res) > 10: + # Sort by reverse depth, so we point to the most recent. + res.sort(key=lambda a: -a[2]) + + # we use half of the limit for the actual most recent events, and + # the other half to randomly point to some of the older events, to + # make sure that we don't completely ignore the older events. + res = res[0:5] + random.sample(res[5:], 5) + + return res + + def get_latest_event_ids_and_hashes_in_room(self, room_id): + """ + Gets the current forward extremities in the given room + + Args: + room_id (str): room_id + + Returns: + Deferred[list[(str, dict[str, str], int)]] + for each event, a tuple of (event_id, hashes, depth) + where *hashes* is a map from algorithm to hash. + """ + + return self.runInteraction( + "get_latest_event_ids_and_hashes_in_room", + self._get_latest_event_ids_and_hashes_in_room, + room_id, + ) + + def get_rooms_with_many_extremities(self, min_count, limit, room_id_filter): + """Get the top rooms with at least N extremities. + + Args: + min_count (int): The minimum number of extremities + limit (int): The maximum number of rooms to return. + room_id_filter (iterable[str]): room_ids to exclude from the results + + Returns: + Deferred[list]: At most `limit` room IDs that have at least + `min_count` extremities, sorted by extremity count. + """ + + def _get_rooms_with_many_extremities_txn(txn): + where_clause = "1=1" + if room_id_filter: + where_clause = "room_id NOT IN (%s)" % ( + ",".join("?" for _ in room_id_filter), + ) + + sql = """ + SELECT room_id FROM event_forward_extremities + WHERE %s + GROUP BY room_id + HAVING count(*) > ? + ORDER BY count(*) DESC + LIMIT ? + """ % ( + where_clause, + ) + + query_args = list(itertools.chain(room_id_filter, [min_count, limit])) + txn.execute(sql, query_args) + return [room_id for room_id, in txn] + + return self.runInteraction( + "get_rooms_with_many_extremities", _get_rooms_with_many_extremities_txn + ) + + @cached(max_entries=5000, iterable=True) + def get_latest_event_ids_in_room(self, room_id): + return self._simple_select_onecol( + table="event_forward_extremities", + keyvalues={"room_id": room_id}, + retcol="event_id", + desc="get_latest_event_ids_in_room", + ) + + def _get_latest_event_ids_and_hashes_in_room(self, txn, room_id): + sql = ( + "SELECT e.event_id, e.depth FROM events as e " + "INNER JOIN event_forward_extremities as f " + "ON e.event_id = f.event_id " + "AND e.room_id = f.room_id " + "WHERE f.room_id = ?" + ) + + txn.execute(sql, (room_id,)) + + results = [] + for event_id, depth in txn.fetchall(): + hashes = self._get_event_reference_hashes_txn(txn, event_id) + prev_hashes = { + k: encode_base64(v) for k, v in hashes.items() if k == "sha256" + } + results.append((event_id, prev_hashes, depth)) + + return results + + def get_min_depth(self, room_id): + """ For hte given room, get the minimum depth we have seen for it. + """ + return self.runInteraction( + "get_min_depth", self._get_min_depth_interaction, room_id + ) + + def _get_min_depth_interaction(self, txn, room_id): + min_depth = self._simple_select_one_onecol_txn( + txn, + table="room_depth", + keyvalues={"room_id": room_id}, + retcol="min_depth", + allow_none=True, + ) + + return int(min_depth) if min_depth is not None else None + + def get_forward_extremeties_for_room(self, room_id, stream_ordering): + """For a given room_id and stream_ordering, return the forward + extremeties of the room at that point in "time". + + Throws a StoreError if we have since purged the index for + stream_orderings from that point. + + Args: + room_id (str): + stream_ordering (int): + + Returns: + deferred, which resolves to a list of event_ids + """ + # We want to make the cache more effective, so we clamp to the last + # change before the given ordering. + last_change = self._events_stream_cache.get_max_pos_of_last_change(room_id) + + # We don't always have a full stream_to_exterm_id table, e.g. after + # the upgrade that introduced it, so we make sure we never ask for a + # stream_ordering from before a restart + last_change = max(self._stream_order_on_start, last_change) + + # provided the last_change is recent enough, we now clamp the requested + # stream_ordering to it. + if last_change > self.stream_ordering_month_ago: + stream_ordering = min(last_change, stream_ordering) + + return self._get_forward_extremeties_for_room(room_id, stream_ordering) + + @cached(max_entries=5000, num_args=2) + def _get_forward_extremeties_for_room(self, room_id, stream_ordering): + """For a given room_id and stream_ordering, return the forward + extremeties of the room at that point in "time". + + Throws a StoreError if we have since purged the index for + stream_orderings from that point. + """ + + if stream_ordering <= self.stream_ordering_month_ago: + raise StoreError(400, "stream_ordering too old") + + sql = """ + SELECT event_id FROM stream_ordering_to_exterm + INNER JOIN ( + SELECT room_id, MAX(stream_ordering) AS stream_ordering + FROM stream_ordering_to_exterm + WHERE stream_ordering <= ? GROUP BY room_id + ) AS rms USING (room_id, stream_ordering) + WHERE room_id = ? + """ + + def get_forward_extremeties_for_room_txn(txn): + txn.execute(sql, (stream_ordering, room_id)) + return [event_id for event_id, in txn] + + return self.runInteraction( + "get_forward_extremeties_for_room", get_forward_extremeties_for_room_txn + ) + + def get_backfill_events(self, room_id, event_list, limit): + """Get a list of Events for a given topic that occurred before (and + including) the events in event_list. Return a list of max size `limit` + + Args: + txn + room_id (str) + event_list (list) + limit (int) + """ + return ( + self.runInteraction( + "get_backfill_events", + self._get_backfill_events, + room_id, + event_list, + limit, + ) + .addCallback(self.get_events_as_list) + .addCallback(lambda l: sorted(l, key=lambda e: -e.depth)) + ) + + def _get_backfill_events(self, txn, room_id, event_list, limit): + logger.debug( + "_get_backfill_events: %s, %s, %s", room_id, repr(event_list), limit + ) + + event_results = set() + + # We want to make sure that we do a breadth-first, "depth" ordered + # search. + + query = ( + "SELECT depth, prev_event_id FROM event_edges" + " INNER JOIN events" + " ON prev_event_id = events.event_id" + " WHERE event_edges.event_id = ?" + " AND event_edges.is_state = ?" + " LIMIT ?" + ) + + queue = PriorityQueue() + + for event_id in event_list: + depth = self._simple_select_one_onecol_txn( + txn, + table="events", + keyvalues={"event_id": event_id, "room_id": room_id}, + retcol="depth", + allow_none=True, + ) + + if depth: + queue.put((-depth, event_id)) + + while not queue.empty() and len(event_results) < limit: + try: + _, event_id = queue.get_nowait() + except Empty: + break + + if event_id in event_results: + continue + + event_results.add(event_id) + + txn.execute(query, (event_id, False, limit - len(event_results))) + + for row in txn: + if row[1] not in event_results: + queue.put((-row[0], row[1])) + + return event_results + + @defer.inlineCallbacks + def get_missing_events(self, room_id, earliest_events, latest_events, limit): + ids = yield self.runInteraction( + "get_missing_events", + self._get_missing_events, + room_id, + earliest_events, + latest_events, + limit, + ) + events = yield self.get_events_as_list(ids) + return events + + def _get_missing_events(self, txn, room_id, earliest_events, latest_events, limit): + + seen_events = set(earliest_events) + front = set(latest_events) - seen_events + event_results = [] + + query = ( + "SELECT prev_event_id FROM event_edges " + "WHERE room_id = ? AND event_id = ? AND is_state = ? " + "LIMIT ?" + ) + + while front and len(event_results) < limit: + new_front = set() + for event_id in front: + txn.execute( + query, (room_id, event_id, False, limit - len(event_results)) + ) + + new_results = set(t[0] for t in txn) - seen_events + + new_front |= new_results + seen_events |= new_results + event_results.extend(new_results) + + front = new_front + + # we built the list working backwards from latest_events; we now need to + # reverse it so that the events are approximately chronological. + event_results.reverse() + return event_results + + @defer.inlineCallbacks + def get_successor_events(self, event_ids): + """Fetch all events that have the given events as a prev event + + Args: + event_ids (iterable[str]) + + Returns: + Deferred[list[str]] + """ + rows = yield self._simple_select_many_batch( + table="event_edges", + column="prev_event_id", + iterable=event_ids, + retcols=("event_id",), + desc="get_successor_events", + ) + + return [row["event_id"] for row in rows] + + +class EventFederationStore(EventFederationWorkerStore): + """ Responsible for storing and serving up the various graphs associated + with an event. Including the main event graph and the auth chains for an + event. + + Also has methods for getting the front (latest) and back (oldest) edges + of the event graphs. These are used to generate the parents for new events + and backfilling from another server respectively. + """ + + EVENT_AUTH_STATE_ONLY = "event_auth_state_only" + + def __init__(self, db_conn, hs): + super(EventFederationStore, self).__init__(db_conn, hs) + + self.register_background_update_handler( + self.EVENT_AUTH_STATE_ONLY, self._background_delete_non_state_event_auth + ) + + hs.get_clock().looping_call( + self._delete_old_forward_extrem_cache, 60 * 60 * 1000 + ) + + def _update_min_depth_for_room_txn(self, txn, room_id, depth): + min_depth = self._get_min_depth_interaction(txn, room_id) + + if min_depth and depth >= min_depth: + return + + self._simple_upsert_txn( + txn, + table="room_depth", + keyvalues={"room_id": room_id}, + values={"min_depth": depth}, + ) + + def _handle_mult_prev_events(self, txn, events): + """ + For the given event, update the event edges table and forward and + backward extremities tables. + """ + self._simple_insert_many_txn( + txn, + table="event_edges", + values=[ + { + "event_id": ev.event_id, + "prev_event_id": e_id, + "room_id": ev.room_id, + "is_state": False, + } + for ev in events + for e_id in ev.prev_event_ids() + ], + ) + + self._update_backward_extremeties(txn, events) + + def _update_backward_extremeties(self, txn, events): + """Updates the event_backward_extremities tables based on the new/updated + events being persisted. + + This is called for new events *and* for events that were outliers, but + are now being persisted as non-outliers. + + Forward extremities are handled when we first start persisting the events. + """ + events_by_room = {} + for ev in events: + events_by_room.setdefault(ev.room_id, []).append(ev) + + query = ( + "INSERT INTO event_backward_extremities (event_id, room_id)" + " SELECT ?, ? WHERE NOT EXISTS (" + " SELECT 1 FROM event_backward_extremities" + " WHERE event_id = ? AND room_id = ?" + " )" + " AND NOT EXISTS (" + " SELECT 1 FROM events WHERE event_id = ? AND room_id = ? " + " AND outlier = ?" + " )" + ) + + txn.executemany( + query, + [ + (e_id, ev.room_id, e_id, ev.room_id, e_id, ev.room_id, False) + for ev in events + for e_id in ev.prev_event_ids() + if not ev.internal_metadata.is_outlier() + ], + ) + + query = ( + "DELETE FROM event_backward_extremities" + " WHERE event_id = ? AND room_id = ?" + ) + txn.executemany( + query, + [ + (ev.event_id, ev.room_id) + for ev in events + if not ev.internal_metadata.is_outlier() + ], + ) + + def _delete_old_forward_extrem_cache(self): + def _delete_old_forward_extrem_cache_txn(txn): + # Delete entries older than a month, while making sure we don't delete + # the only entries for a room. + sql = """ + DELETE FROM stream_ordering_to_exterm + WHERE + room_id IN ( + SELECT room_id + FROM stream_ordering_to_exterm + WHERE stream_ordering > ? + ) AND stream_ordering < ? + """ + txn.execute( + sql, (self.stream_ordering_month_ago, self.stream_ordering_month_ago) + ) + + return run_as_background_process( + "delete_old_forward_extrem_cache", + self.runInteraction, + "_delete_old_forward_extrem_cache", + _delete_old_forward_extrem_cache_txn, + ) + + def clean_room_for_join(self, room_id): + return self.runInteraction( + "clean_room_for_join", self._clean_room_for_join_txn, room_id + ) + + def _clean_room_for_join_txn(self, txn, room_id): + query = "DELETE FROM event_forward_extremities WHERE room_id = ?" + + txn.execute(query, (room_id,)) + txn.call_after(self.get_latest_event_ids_in_room.invalidate, (room_id,)) + + @defer.inlineCallbacks + def _background_delete_non_state_event_auth(self, progress, batch_size): + def delete_event_auth(txn): + target_min_stream_id = progress.get("target_min_stream_id_inclusive") + max_stream_id = progress.get("max_stream_id_exclusive") + + if not target_min_stream_id or not max_stream_id: + txn.execute("SELECT COALESCE(MIN(stream_ordering), 0) FROM events") + rows = txn.fetchall() + target_min_stream_id = rows[0][0] + + txn.execute("SELECT COALESCE(MAX(stream_ordering), 0) FROM events") + rows = txn.fetchall() + max_stream_id = rows[0][0] + + min_stream_id = max_stream_id - batch_size + + sql = """ + DELETE FROM event_auth + WHERE event_id IN ( + SELECT event_id FROM events + LEFT JOIN state_events USING (room_id, event_id) + WHERE ? <= stream_ordering AND stream_ordering < ? + AND state_key IS null + ) + """ + + txn.execute(sql, (min_stream_id, max_stream_id)) + + new_progress = { + "target_min_stream_id_inclusive": target_min_stream_id, + "max_stream_id_exclusive": min_stream_id, + } + + self._background_update_progress_txn( + txn, self.EVENT_AUTH_STATE_ONLY, new_progress + ) + + return min_stream_id >= target_min_stream_id + + result = yield self.runInteraction( + self.EVENT_AUTH_STATE_ONLY, delete_event_auth + ) + + if not result: + yield self._end_background_update(self.EVENT_AUTH_STATE_ONLY) + + return batch_size diff --git a/synapse/storage/data_stores/main/event_push_actions.py b/synapse/storage/data_stores/main/event_push_actions.py new file mode 100644 index 0000000000..22025effbc --- /dev/null +++ b/synapse/storage/data_stores/main/event_push_actions.py @@ -0,0 +1,960 @@ +# -*- coding: utf-8 -*- +# Copyright 2015 OpenMarket Ltd +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging + +from six import iteritems + +from canonicaljson import json + +from twisted.internet import defer + +from synapse.metrics.background_process_metrics import run_as_background_process +from synapse.storage._base import LoggingTransaction, SQLBaseStore +from synapse.util.caches.descriptors import cachedInlineCallbacks + +logger = logging.getLogger(__name__) + + +DEFAULT_NOTIF_ACTION = ["notify", {"set_tweak": "highlight", "value": False}] +DEFAULT_HIGHLIGHT_ACTION = [ + "notify", + {"set_tweak": "sound", "value": "default"}, + {"set_tweak": "highlight"}, +] + + +def _serialize_action(actions, is_highlight): + """Custom serializer for actions. This allows us to "compress" common actions. + + We use the fact that most users have the same actions for notifs (and for + highlights). + We store these default actions as the empty string rather than the full JSON. + Since the empty string isn't valid JSON there is no risk of this clashing with + any real JSON actions + """ + if is_highlight: + if actions == DEFAULT_HIGHLIGHT_ACTION: + return "" # We use empty string as the column is non-NULL + else: + if actions == DEFAULT_NOTIF_ACTION: + return "" + return json.dumps(actions) + + +def _deserialize_action(actions, is_highlight): + """Custom deserializer for actions. This allows us to "compress" common actions + """ + if actions: + return json.loads(actions) + + if is_highlight: + return DEFAULT_HIGHLIGHT_ACTION + else: + return DEFAULT_NOTIF_ACTION + + +class EventPushActionsWorkerStore(SQLBaseStore): + def __init__(self, db_conn, hs): + super(EventPushActionsWorkerStore, self).__init__(db_conn, hs) + + # These get correctly set by _find_stream_orderings_for_times_txn + self.stream_ordering_month_ago = None + self.stream_ordering_day_ago = None + + cur = LoggingTransaction( + db_conn.cursor(), + name="_find_stream_orderings_for_times_txn", + database_engine=self.database_engine, + ) + self._find_stream_orderings_for_times_txn(cur) + cur.close() + + self.find_stream_orderings_looping_call = self._clock.looping_call( + self._find_stream_orderings_for_times, 10 * 60 * 1000 + ) + self._rotate_delay = 3 + self._rotate_count = 10000 + + @cachedInlineCallbacks(num_args=3, tree=True, max_entries=5000) + def get_unread_event_push_actions_by_room_for_user( + self, room_id, user_id, last_read_event_id + ): + ret = yield self.runInteraction( + "get_unread_event_push_actions_by_room", + self._get_unread_counts_by_receipt_txn, + room_id, + user_id, + last_read_event_id, + ) + return ret + + def _get_unread_counts_by_receipt_txn( + self, txn, room_id, user_id, last_read_event_id + ): + sql = ( + "SELECT stream_ordering" + " FROM events" + " WHERE room_id = ? AND event_id = ?" + ) + txn.execute(sql, (room_id, last_read_event_id)) + results = txn.fetchall() + if len(results) == 0: + return {"notify_count": 0, "highlight_count": 0} + + stream_ordering = results[0][0] + + return self._get_unread_counts_by_pos_txn( + txn, room_id, user_id, stream_ordering + ) + + def _get_unread_counts_by_pos_txn(self, txn, room_id, user_id, stream_ordering): + + # First get number of notifications. + # We don't need to put a notif=1 clause as all rows always have + # notif=1 + sql = ( + "SELECT count(*)" + " FROM event_push_actions ea" + " WHERE" + " user_id = ?" + " AND room_id = ?" + " AND stream_ordering > ?" + ) + + txn.execute(sql, (user_id, room_id, stream_ordering)) + row = txn.fetchone() + notify_count = row[0] if row else 0 + + txn.execute( + """ + SELECT notif_count FROM event_push_summary + WHERE room_id = ? AND user_id = ? AND stream_ordering > ? + """, + (room_id, user_id, stream_ordering), + ) + rows = txn.fetchall() + if rows: + notify_count += rows[0][0] + + # Now get the number of highlights + sql = ( + "SELECT count(*)" + " FROM event_push_actions ea" + " WHERE" + " highlight = 1" + " AND user_id = ?" + " AND room_id = ?" + " AND stream_ordering > ?" + ) + + txn.execute(sql, (user_id, room_id, stream_ordering)) + row = txn.fetchone() + highlight_count = row[0] if row else 0 + + return {"notify_count": notify_count, "highlight_count": highlight_count} + + @defer.inlineCallbacks + def get_push_action_users_in_range(self, min_stream_ordering, max_stream_ordering): + def f(txn): + sql = ( + "SELECT DISTINCT(user_id) FROM event_push_actions WHERE" + " stream_ordering >= ? AND stream_ordering <= ?" + ) + txn.execute(sql, (min_stream_ordering, max_stream_ordering)) + return [r[0] for r in txn] + + ret = yield self.runInteraction("get_push_action_users_in_range", f) + return ret + + @defer.inlineCallbacks + def get_unread_push_actions_for_user_in_range_for_http( + self, user_id, min_stream_ordering, max_stream_ordering, limit=20 + ): + """Get a list of the most recent unread push actions for a given user, + within the given stream ordering range. Called by the httppusher. + + Args: + user_id (str): The user to fetch push actions for. + min_stream_ordering(int): The exclusive lower bound on the + stream ordering of event push actions to fetch. + max_stream_ordering(int): The inclusive upper bound on the + stream ordering of event push actions to fetch. + limit (int): The maximum number of rows to return. + Returns: + A promise which resolves to a list of dicts with the keys "event_id", + "room_id", "stream_ordering", "actions". + The list will be ordered by ascending stream_ordering. + The list will have between 0~limit entries. + """ + # find rooms that have a read receipt in them and return the next + # push actions + def get_after_receipt(txn): + # find rooms that have a read receipt in them and return the next + # push actions + sql = ( + "SELECT ep.event_id, ep.room_id, ep.stream_ordering, ep.actions," + " ep.highlight " + " FROM (" + " SELECT room_id," + " MAX(stream_ordering) as stream_ordering" + " FROM events" + " INNER JOIN receipts_linearized USING (room_id, event_id)" + " WHERE receipt_type = 'm.read' AND user_id = ?" + " GROUP BY room_id" + ") AS rl," + " event_push_actions AS ep" + " WHERE" + " ep.room_id = rl.room_id" + " AND ep.stream_ordering > rl.stream_ordering" + " AND ep.user_id = ?" + " AND ep.stream_ordering > ?" + " AND ep.stream_ordering <= ?" + " ORDER BY ep.stream_ordering ASC LIMIT ?" + ) + args = [user_id, user_id, min_stream_ordering, max_stream_ordering, limit] + txn.execute(sql, args) + return txn.fetchall() + + after_read_receipt = yield self.runInteraction( + "get_unread_push_actions_for_user_in_range_http_arr", get_after_receipt + ) + + # There are rooms with push actions in them but you don't have a read receipt in + # them e.g. rooms you've been invited to, so get push actions for rooms which do + # not have read receipts in them too. + def get_no_receipt(txn): + sql = ( + "SELECT ep.event_id, ep.room_id, ep.stream_ordering, ep.actions," + " ep.highlight " + " FROM event_push_actions AS ep" + " INNER JOIN events AS e USING (room_id, event_id)" + " WHERE" + " ep.room_id NOT IN (" + " SELECT room_id FROM receipts_linearized" + " WHERE receipt_type = 'm.read' AND user_id = ?" + " GROUP BY room_id" + " )" + " AND ep.user_id = ?" + " AND ep.stream_ordering > ?" + " AND ep.stream_ordering <= ?" + " ORDER BY ep.stream_ordering ASC LIMIT ?" + ) + args = [user_id, user_id, min_stream_ordering, max_stream_ordering, limit] + txn.execute(sql, args) + return txn.fetchall() + + no_read_receipt = yield self.runInteraction( + "get_unread_push_actions_for_user_in_range_http_nrr", get_no_receipt + ) + + notifs = [ + { + "event_id": row[0], + "room_id": row[1], + "stream_ordering": row[2], + "actions": _deserialize_action(row[3], row[4]), + } + for row in after_read_receipt + no_read_receipt + ] + + # Now sort it so it's ordered correctly, since currently it will + # contain results from the first query, correctly ordered, followed + # by results from the second query, but we want them all ordered + # by stream_ordering, oldest first. + notifs.sort(key=lambda r: r["stream_ordering"]) + + # Take only up to the limit. We have to stop at the limit because + # one of the subqueries may have hit the limit. + return notifs[:limit] + + @defer.inlineCallbacks + def get_unread_push_actions_for_user_in_range_for_email( + self, user_id, min_stream_ordering, max_stream_ordering, limit=20 + ): + """Get a list of the most recent unread push actions for a given user, + within the given stream ordering range. Called by the emailpusher + + Args: + user_id (str): The user to fetch push actions for. + min_stream_ordering(int): The exclusive lower bound on the + stream ordering of event push actions to fetch. + max_stream_ordering(int): The inclusive upper bound on the + stream ordering of event push actions to fetch. + limit (int): The maximum number of rows to return. + Returns: + A promise which resolves to a list of dicts with the keys "event_id", + "room_id", "stream_ordering", "actions", "received_ts". + The list will be ordered by descending received_ts. + The list will have between 0~limit entries. + """ + # find rooms that have a read receipt in them and return the most recent + # push actions + def get_after_receipt(txn): + sql = ( + "SELECT ep.event_id, ep.room_id, ep.stream_ordering, ep.actions," + " ep.highlight, e.received_ts" + " FROM (" + " SELECT room_id," + " MAX(stream_ordering) as stream_ordering" + " FROM events" + " INNER JOIN receipts_linearized USING (room_id, event_id)" + " WHERE receipt_type = 'm.read' AND user_id = ?" + " GROUP BY room_id" + ") AS rl," + " event_push_actions AS ep" + " INNER JOIN events AS e USING (room_id, event_id)" + " WHERE" + " ep.room_id = rl.room_id" + " AND ep.stream_ordering > rl.stream_ordering" + " AND ep.user_id = ?" + " AND ep.stream_ordering > ?" + " AND ep.stream_ordering <= ?" + " ORDER BY ep.stream_ordering DESC LIMIT ?" + ) + args = [user_id, user_id, min_stream_ordering, max_stream_ordering, limit] + txn.execute(sql, args) + return txn.fetchall() + + after_read_receipt = yield self.runInteraction( + "get_unread_push_actions_for_user_in_range_email_arr", get_after_receipt + ) + + # There are rooms with push actions in them but you don't have a read receipt in + # them e.g. rooms you've been invited to, so get push actions for rooms which do + # not have read receipts in them too. + def get_no_receipt(txn): + sql = ( + "SELECT ep.event_id, ep.room_id, ep.stream_ordering, ep.actions," + " ep.highlight, e.received_ts" + " FROM event_push_actions AS ep" + " INNER JOIN events AS e USING (room_id, event_id)" + " WHERE" + " ep.room_id NOT IN (" + " SELECT room_id FROM receipts_linearized" + " WHERE receipt_type = 'm.read' AND user_id = ?" + " GROUP BY room_id" + " )" + " AND ep.user_id = ?" + " AND ep.stream_ordering > ?" + " AND ep.stream_ordering <= ?" + " ORDER BY ep.stream_ordering DESC LIMIT ?" + ) + args = [user_id, user_id, min_stream_ordering, max_stream_ordering, limit] + txn.execute(sql, args) + return txn.fetchall() + + no_read_receipt = yield self.runInteraction( + "get_unread_push_actions_for_user_in_range_email_nrr", get_no_receipt + ) + + # Make a list of dicts from the two sets of results. + notifs = [ + { + "event_id": row[0], + "room_id": row[1], + "stream_ordering": row[2], + "actions": _deserialize_action(row[3], row[4]), + "received_ts": row[5], + } + for row in after_read_receipt + no_read_receipt + ] + + # Now sort it so it's ordered correctly, since currently it will + # contain results from the first query, correctly ordered, followed + # by results from the second query, but we want them all ordered + # by received_ts (most recent first) + notifs.sort(key=lambda r: -(r["received_ts"] or 0)) + + # Now return the first `limit` + return notifs[:limit] + + def get_if_maybe_push_in_range_for_user(self, user_id, min_stream_ordering): + """A fast check to see if there might be something to push for the + user since the given stream ordering. May return false positives. + + Useful to know whether to bother starting a pusher on start up or not. + + Args: + user_id (str) + min_stream_ordering (int) + + Returns: + Deferred[bool]: True if there may be push to process, False if + there definitely isn't. + """ + + def _get_if_maybe_push_in_range_for_user_txn(txn): + sql = """ + SELECT 1 FROM event_push_actions + WHERE user_id = ? AND stream_ordering > ? + LIMIT 1 + """ + + txn.execute(sql, (user_id, min_stream_ordering)) + return bool(txn.fetchone()) + + return self.runInteraction( + "get_if_maybe_push_in_range_for_user", + _get_if_maybe_push_in_range_for_user_txn, + ) + + def add_push_actions_to_staging(self, event_id, user_id_actions): + """Add the push actions for the event to the push action staging area. + + Args: + event_id (str) + user_id_actions (dict[str, list[dict|str])]): A dictionary mapping + user_id to list of push actions, where an action can either be + a string or dict. + + Returns: + Deferred + """ + + if not user_id_actions: + return + + # This is a helper function for generating the necessary tuple that + # can be used to inert into the `event_push_actions_staging` table. + def _gen_entry(user_id, actions): + is_highlight = 1 if _action_has_highlight(actions) else 0 + return ( + event_id, # event_id column + user_id, # user_id column + _serialize_action(actions, is_highlight), # actions column + 1, # notif column + is_highlight, # highlight column + ) + + def _add_push_actions_to_staging_txn(txn): + # We don't use _simple_insert_many here to avoid the overhead + # of generating lists of dicts. + + sql = """ + INSERT INTO event_push_actions_staging + (event_id, user_id, actions, notif, highlight) + VALUES (?, ?, ?, ?, ?) + """ + + txn.executemany( + sql, + ( + _gen_entry(user_id, actions) + for user_id, actions in iteritems(user_id_actions) + ), + ) + + return self.runInteraction( + "add_push_actions_to_staging", _add_push_actions_to_staging_txn + ) + + @defer.inlineCallbacks + def remove_push_actions_from_staging(self, event_id): + """Called if we failed to persist the event to ensure that stale push + actions don't build up in the DB + + Args: + event_id (str) + """ + + try: + res = yield self._simple_delete( + table="event_push_actions_staging", + keyvalues={"event_id": event_id}, + desc="remove_push_actions_from_staging", + ) + return res + except Exception: + # this method is called from an exception handler, so propagating + # another exception here really isn't helpful - there's nothing + # the caller can do about it. Just log the exception and move on. + logger.exception( + "Error removing push actions after event persistence failure" + ) + + def _find_stream_orderings_for_times(self): + return run_as_background_process( + "event_push_action_stream_orderings", + self.runInteraction, + "_find_stream_orderings_for_times", + self._find_stream_orderings_for_times_txn, + ) + + def _find_stream_orderings_for_times_txn(self, txn): + logger.info("Searching for stream ordering 1 month ago") + self.stream_ordering_month_ago = self._find_first_stream_ordering_after_ts_txn( + txn, self._clock.time_msec() - 30 * 24 * 60 * 60 * 1000 + ) + logger.info( + "Found stream ordering 1 month ago: it's %d", self.stream_ordering_month_ago + ) + logger.info("Searching for stream ordering 1 day ago") + self.stream_ordering_day_ago = self._find_first_stream_ordering_after_ts_txn( + txn, self._clock.time_msec() - 24 * 60 * 60 * 1000 + ) + logger.info( + "Found stream ordering 1 day ago: it's %d", self.stream_ordering_day_ago + ) + + def find_first_stream_ordering_after_ts(self, ts): + """Gets the stream ordering corresponding to a given timestamp. + + Specifically, finds the stream_ordering of the first event that was + received on or after the timestamp. This is done by a binary search on + the events table, since there is no index on received_ts, so is + relatively slow. + + Args: + ts (int): timestamp in millis + + Returns: + Deferred[int]: stream ordering of the first event received on/after + the timestamp + """ + return self.runInteraction( + "_find_first_stream_ordering_after_ts_txn", + self._find_first_stream_ordering_after_ts_txn, + ts, + ) + + @staticmethod + def _find_first_stream_ordering_after_ts_txn(txn, ts): + """ + Find the stream_ordering of the first event that was received on or + after a given timestamp. This is relatively slow as there is no index + on received_ts but we can then use this to delete push actions before + this. + + received_ts must necessarily be in the same order as stream_ordering + and stream_ordering is indexed, so we manually binary search using + stream_ordering + + Args: + txn (twisted.enterprise.adbapi.Transaction): + ts (int): timestamp to search for + + Returns: + int: stream ordering + """ + txn.execute("SELECT MAX(stream_ordering) FROM events") + max_stream_ordering = txn.fetchone()[0] + + if max_stream_ordering is None: + return 0 + + # We want the first stream_ordering in which received_ts is greater + # than or equal to ts. Call this point X. + # + # We maintain the invariants: + # + # range_start <= X <= range_end + # + range_start = 0 + range_end = max_stream_ordering + 1 + + # Given a stream_ordering, look up the timestamp at that + # stream_ordering. + # + # The array may be sparse (we may be missing some stream_orderings). + # We treat the gaps as the same as having the same value as the + # preceding entry, because we will pick the lowest stream_ordering + # which satisfies our requirement of received_ts >= ts. + # + # For example, if our array of events indexed by stream_ordering is + # [10, , 20], we should treat this as being equivalent to + # [10, 10, 20]. + # + sql = ( + "SELECT received_ts FROM events" + " WHERE stream_ordering <= ?" + " ORDER BY stream_ordering DESC" + " LIMIT 1" + ) + + while range_end - range_start > 0: + middle = (range_end + range_start) // 2 + txn.execute(sql, (middle,)) + row = txn.fetchone() + if row is None: + # no rows with stream_ordering<=middle + range_start = middle + 1 + continue + + middle_ts = row[0] + if ts > middle_ts: + # we got a timestamp lower than the one we were looking for. + # definitely need to look higher: X > middle. + range_start = middle + 1 + else: + # we got a timestamp higher than (or the same as) the one we + # were looking for. We aren't yet sure about the point we + # looked up, but we can be sure that X <= middle. + range_end = middle + + return range_end + + +class EventPushActionsStore(EventPushActionsWorkerStore): + EPA_HIGHLIGHT_INDEX = "epa_highlight_index" + + def __init__(self, db_conn, hs): + super(EventPushActionsStore, self).__init__(db_conn, hs) + + self.register_background_index_update( + self.EPA_HIGHLIGHT_INDEX, + index_name="event_push_actions_u_highlight", + table="event_push_actions", + columns=["user_id", "stream_ordering"], + ) + + self.register_background_index_update( + "event_push_actions_highlights_index", + index_name="event_push_actions_highlights_index", + table="event_push_actions", + columns=["user_id", "room_id", "topological_ordering", "stream_ordering"], + where_clause="highlight=1", + ) + + self._doing_notif_rotation = False + self._rotate_notif_loop = self._clock.looping_call( + self._start_rotate_notifs, 30 * 60 * 1000 + ) + + def _set_push_actions_for_event_and_users_txn( + self, txn, events_and_contexts, all_events_and_contexts + ): + """Handles moving push actions from staging table to main + event_push_actions table for all events in `events_and_contexts`. + + Also ensures that all events in `all_events_and_contexts` are removed + from the push action staging area. + + Args: + events_and_contexts (list[(EventBase, EventContext)]): events + we are persisting + all_events_and_contexts (list[(EventBase, EventContext)]): all + events that we were going to persist. This includes events + we've already persisted, etc, that wouldn't appear in + events_and_context. + """ + + sql = """ + INSERT INTO event_push_actions ( + room_id, event_id, user_id, actions, stream_ordering, + topological_ordering, notif, highlight + ) + SELECT ?, event_id, user_id, actions, ?, ?, notif, highlight + FROM event_push_actions_staging + WHERE event_id = ? + """ + + if events_and_contexts: + txn.executemany( + sql, + ( + ( + event.room_id, + event.internal_metadata.stream_ordering, + event.depth, + event.event_id, + ) + for event, _ in events_and_contexts + ), + ) + + for event, _ in events_and_contexts: + user_ids = self._simple_select_onecol_txn( + txn, + table="event_push_actions_staging", + keyvalues={"event_id": event.event_id}, + retcol="user_id", + ) + + for uid in user_ids: + txn.call_after( + self.get_unread_event_push_actions_by_room_for_user.invalidate_many, + (event.room_id, uid), + ) + + # Now we delete the staging area for *all* events that were being + # persisted. + txn.executemany( + "DELETE FROM event_push_actions_staging WHERE event_id = ?", + ((event.event_id,) for event, _ in all_events_and_contexts), + ) + + @defer.inlineCallbacks + def get_push_actions_for_user( + self, user_id, before=None, limit=50, only_highlight=False + ): + def f(txn): + before_clause = "" + if before: + before_clause = "AND epa.stream_ordering < ?" + args = [user_id, before, limit] + else: + args = [user_id, limit] + + if only_highlight: + if len(before_clause) > 0: + before_clause += " " + before_clause += "AND epa.highlight = 1" + + # NB. This assumes event_ids are globally unique since + # it makes the query easier to index + sql = ( + "SELECT epa.event_id, epa.room_id," + " epa.stream_ordering, epa.topological_ordering," + " epa.actions, epa.highlight, epa.profile_tag, e.received_ts" + " FROM event_push_actions epa, events e" + " WHERE epa.event_id = e.event_id" + " AND epa.user_id = ? %s" + " ORDER BY epa.stream_ordering DESC" + " LIMIT ?" % (before_clause,) + ) + txn.execute(sql, args) + return self.cursor_to_dict(txn) + + push_actions = yield self.runInteraction("get_push_actions_for_user", f) + for pa in push_actions: + pa["actions"] = _deserialize_action(pa["actions"], pa["highlight"]) + return push_actions + + @defer.inlineCallbacks + def get_time_of_last_push_action_before(self, stream_ordering): + def f(txn): + sql = ( + "SELECT e.received_ts" + " FROM event_push_actions AS ep" + " JOIN events e ON ep.room_id = e.room_id AND ep.event_id = e.event_id" + " WHERE ep.stream_ordering > ?" + " ORDER BY ep.stream_ordering ASC" + " LIMIT 1" + ) + txn.execute(sql, (stream_ordering,)) + return txn.fetchone() + + result = yield self.runInteraction("get_time_of_last_push_action_before", f) + return result[0] if result else None + + @defer.inlineCallbacks + def get_latest_push_action_stream_ordering(self): + def f(txn): + txn.execute("SELECT MAX(stream_ordering) FROM event_push_actions") + return txn.fetchone() + + result = yield self.runInteraction("get_latest_push_action_stream_ordering", f) + return result[0] or 0 + + def _remove_push_actions_for_event_id_txn(self, txn, room_id, event_id): + # Sad that we have to blow away the cache for the whole room here + txn.call_after( + self.get_unread_event_push_actions_by_room_for_user.invalidate_many, + (room_id,), + ) + txn.execute( + "DELETE FROM event_push_actions WHERE room_id = ? AND event_id = ?", + (room_id, event_id), + ) + + def _remove_old_push_actions_before_txn( + self, txn, room_id, user_id, stream_ordering + ): + """ + Purges old push actions for a user and room before a given + stream_ordering. + + We however keep a months worth of highlighted notifications, so that + users can still get a list of recent highlights. + + Args: + txn: The transcation + room_id: Room ID to delete from + user_id: user ID to delete for + stream_ordering: The lowest stream ordering which will + not be deleted. + """ + txn.call_after( + self.get_unread_event_push_actions_by_room_for_user.invalidate_many, + (room_id, user_id), + ) + + # We need to join on the events table to get the received_ts for + # event_push_actions and sqlite won't let us use a join in a delete so + # we can't just delete where received_ts < x. Furthermore we can + # only identify event_push_actions by a tuple of room_id, event_id + # we we can't use a subquery. + # Instead, we look up the stream ordering for the last event in that + # room received before the threshold time and delete event_push_actions + # in the room with a stream_odering before that. + txn.execute( + "DELETE FROM event_push_actions " + " WHERE user_id = ? AND room_id = ? AND " + " stream_ordering <= ?" + " AND ((stream_ordering < ? AND highlight = 1) or highlight = 0)", + (user_id, room_id, stream_ordering, self.stream_ordering_month_ago), + ) + + txn.execute( + """ + DELETE FROM event_push_summary + WHERE room_id = ? AND user_id = ? AND stream_ordering <= ? + """, + (room_id, user_id, stream_ordering), + ) + + def _start_rotate_notifs(self): + return run_as_background_process("rotate_notifs", self._rotate_notifs) + + @defer.inlineCallbacks + def _rotate_notifs(self): + if self._doing_notif_rotation or self.stream_ordering_day_ago is None: + return + self._doing_notif_rotation = True + + try: + while True: + logger.info("Rotating notifications") + + caught_up = yield self.runInteraction( + "_rotate_notifs", self._rotate_notifs_txn + ) + if caught_up: + break + yield self.hs.get_clock().sleep(self._rotate_delay) + finally: + self._doing_notif_rotation = False + + def _rotate_notifs_txn(self, txn): + """Archives older notifications into event_push_summary. Returns whether + the archiving process has caught up or not. + """ + + old_rotate_stream_ordering = self._simple_select_one_onecol_txn( + txn, + table="event_push_summary_stream_ordering", + keyvalues={}, + retcol="stream_ordering", + ) + + # We don't to try and rotate millions of rows at once, so we cap the + # maximum stream ordering we'll rotate before. + txn.execute( + """ + SELECT stream_ordering FROM event_push_actions + WHERE stream_ordering > ? + ORDER BY stream_ordering ASC LIMIT 1 OFFSET ? + """, + (old_rotate_stream_ordering, self._rotate_count), + ) + stream_row = txn.fetchone() + if stream_row: + offset_stream_ordering, = stream_row + rotate_to_stream_ordering = min( + self.stream_ordering_day_ago, offset_stream_ordering + ) + caught_up = offset_stream_ordering >= self.stream_ordering_day_ago + else: + rotate_to_stream_ordering = self.stream_ordering_day_ago + caught_up = True + + logger.info("Rotating notifications up to: %s", rotate_to_stream_ordering) + + self._rotate_notifs_before_txn(txn, rotate_to_stream_ordering) + + # We have caught up iff we were limited by `stream_ordering_day_ago` + return caught_up + + def _rotate_notifs_before_txn(self, txn, rotate_to_stream_ordering): + old_rotate_stream_ordering = self._simple_select_one_onecol_txn( + txn, + table="event_push_summary_stream_ordering", + keyvalues={}, + retcol="stream_ordering", + ) + + # Calculate the new counts that should be upserted into event_push_summary + sql = """ + SELECT user_id, room_id, + coalesce(old.notif_count, 0) + upd.notif_count, + upd.stream_ordering, + old.user_id + FROM ( + SELECT user_id, room_id, count(*) as notif_count, + max(stream_ordering) as stream_ordering + FROM event_push_actions + WHERE ? <= stream_ordering AND stream_ordering < ? + AND highlight = 0 + GROUP BY user_id, room_id + ) AS upd + LEFT JOIN event_push_summary AS old USING (user_id, room_id) + """ + + txn.execute(sql, (old_rotate_stream_ordering, rotate_to_stream_ordering)) + rows = txn.fetchall() + + logger.info("Rotating notifications, handling %d rows", len(rows)) + + # If the `old.user_id` above is NULL then we know there isn't already an + # entry in the table, so we simply insert it. Otherwise we update the + # existing table. + self._simple_insert_many_txn( + txn, + table="event_push_summary", + values=[ + { + "user_id": row[0], + "room_id": row[1], + "notif_count": row[2], + "stream_ordering": row[3], + } + for row in rows + if row[4] is None + ], + ) + + txn.executemany( + """ + UPDATE event_push_summary SET notif_count = ?, stream_ordering = ? + WHERE user_id = ? AND room_id = ? + """, + ((row[2], row[3], row[0], row[1]) for row in rows if row[4] is not None), + ) + + txn.execute( + "DELETE FROM event_push_actions" + " WHERE ? <= stream_ordering AND stream_ordering < ? AND highlight = 0", + (old_rotate_stream_ordering, rotate_to_stream_ordering), + ) + + logger.info("Rotating notifications, deleted %s push actions", txn.rowcount) + + txn.execute( + "UPDATE event_push_summary_stream_ordering SET stream_ordering = ?", + (rotate_to_stream_ordering,), + ) + + +def _action_has_highlight(actions): + for action in actions: + try: + if action.get("set_tweak", None) == "highlight": + return action.get("value", True) + except AttributeError: + pass + + return False diff --git a/synapse/storage/data_stores/main/events.py b/synapse/storage/data_stores/main/events.py new file mode 100644 index 0000000000..03b5111c5d --- /dev/null +++ b/synapse/storage/data_stores/main/events.py @@ -0,0 +1,2489 @@ +# -*- coding: utf-8 -*- +# Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018-2019 New Vector Ltd +# Copyright 2019 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import itertools +import logging +from collections import Counter as c_counter, OrderedDict, deque, namedtuple +from functools import wraps + +from six import iteritems, text_type +from six.moves import range + +from canonicaljson import json +from prometheus_client import Counter, Histogram + +from twisted.internet import defer + +import synapse.metrics +from synapse.api.constants import EventTypes +from synapse.api.errors import SynapseError +from synapse.events import EventBase # noqa: F401 +from synapse.events.snapshot import EventContext # noqa: F401 +from synapse.events.utils import prune_event_dict +from synapse.logging.context import PreserveLoggingContext, make_deferred_yieldable +from synapse.logging.utils import log_function +from synapse.metrics import BucketCollector +from synapse.metrics.background_process_metrics import run_as_background_process +from synapse.state import StateResolutionStore +from synapse.storage._base import make_in_list_sql_clause +from synapse.storage.background_updates import BackgroundUpdateStore +from synapse.storage.data_stores.main.event_federation import EventFederationStore +from synapse.storage.data_stores.main.events_worker import EventsWorkerStore +from synapse.storage.data_stores.main.state import StateGroupWorkerStore +from synapse.types import RoomStreamToken, get_domain_from_id +from synapse.util import batch_iter +from synapse.util.async_helpers import ObservableDeferred +from synapse.util.caches.descriptors import cached, cachedInlineCallbacks +from synapse.util.frozenutils import frozendict_json_encoder +from synapse.util.metrics import Measure + +logger = logging.getLogger(__name__) + +persist_event_counter = Counter("synapse_storage_events_persisted_events", "") +event_counter = Counter( + "synapse_storage_events_persisted_events_sep", + "", + ["type", "origin_type", "origin_entity"], +) + +# The number of times we are recalculating the current state +state_delta_counter = Counter("synapse_storage_events_state_delta", "") + +# The number of times we are recalculating state when there is only a +# single forward extremity +state_delta_single_event_counter = Counter( + "synapse_storage_events_state_delta_single_event", "" +) + +# The number of times we are reculating state when we could have resonably +# calculated the delta when we calculated the state for an event we were +# persisting. +state_delta_reuse_delta_counter = Counter( + "synapse_storage_events_state_delta_reuse_delta", "" +) + +# The number of forward extremities for each new event. +forward_extremities_counter = Histogram( + "synapse_storage_events_forward_extremities_persisted", + "Number of forward extremities for each new event", + buckets=(1, 2, 3, 5, 7, 10, 15, 20, 50, 100, 200, 500, "+Inf"), +) + +# The number of stale forward extremities for each new event. Stale extremities +# are those that were in the previous set of extremities as well as the new. +stale_forward_extremities_counter = Histogram( + "synapse_storage_events_stale_forward_extremities_persisted", + "Number of unchanged forward extremities for each new event", + buckets=(0, 1, 2, 3, 5, 7, 10, 15, 20, 50, 100, 200, 500, "+Inf"), +) + + +def encode_json(json_object): + """ + Encode a Python object as JSON and return it in a Unicode string. + """ + out = frozendict_json_encoder.encode(json_object) + if isinstance(out, bytes): + out = out.decode("utf8") + return out + + +class _EventPeristenceQueue(object): + """Queues up events so that they can be persisted in bulk with only one + concurrent transaction per room. + """ + + _EventPersistQueueItem = namedtuple( + "_EventPersistQueueItem", ("events_and_contexts", "backfilled", "deferred") + ) + + def __init__(self): + self._event_persist_queues = {} + self._currently_persisting_rooms = set() + + def add_to_queue(self, room_id, events_and_contexts, backfilled): + """Add events to the queue, with the given persist_event options. + + NB: due to the normal usage pattern of this method, it does *not* + follow the synapse logcontext rules, and leaves the logcontext in + place whether or not the returned deferred is ready. + + Args: + room_id (str): + events_and_contexts (list[(EventBase, EventContext)]): + backfilled (bool): + + Returns: + defer.Deferred: a deferred which will resolve once the events are + persisted. Runs its callbacks *without* a logcontext. + """ + queue = self._event_persist_queues.setdefault(room_id, deque()) + if queue: + # if the last item in the queue has the same `backfilled` setting, + # we can just add these new events to that item. + end_item = queue[-1] + if end_item.backfilled == backfilled: + end_item.events_and_contexts.extend(events_and_contexts) + return end_item.deferred.observe() + + deferred = ObservableDeferred(defer.Deferred(), consumeErrors=True) + + queue.append( + self._EventPersistQueueItem( + events_and_contexts=events_and_contexts, + backfilled=backfilled, + deferred=deferred, + ) + ) + + return deferred.observe() + + def handle_queue(self, room_id, per_item_callback): + """Attempts to handle the queue for a room if not already being handled. + + The given callback will be invoked with for each item in the queue, + of type _EventPersistQueueItem. The per_item_callback will continuously + be called with new items, unless the queue becomnes empty. The return + value of the function will be given to the deferreds waiting on the item, + exceptions will be passed to the deferreds as well. + + This function should therefore be called whenever anything is added + to the queue. + + If another callback is currently handling the queue then it will not be + invoked. + """ + + if room_id in self._currently_persisting_rooms: + return + + self._currently_persisting_rooms.add(room_id) + + @defer.inlineCallbacks + def handle_queue_loop(): + try: + queue = self._get_drainining_queue(room_id) + for item in queue: + try: + ret = yield per_item_callback(item) + except Exception: + with PreserveLoggingContext(): + item.deferred.errback() + else: + with PreserveLoggingContext(): + item.deferred.callback(ret) + finally: + queue = self._event_persist_queues.pop(room_id, None) + if queue: + self._event_persist_queues[room_id] = queue + self._currently_persisting_rooms.discard(room_id) + + # set handle_queue_loop off in the background + run_as_background_process("persist_events", handle_queue_loop) + + def _get_drainining_queue(self, room_id): + queue = self._event_persist_queues.setdefault(room_id, deque()) + + try: + while True: + yield queue.popleft() + except IndexError: + # Queue has been drained. + pass + + +_EventCacheEntry = namedtuple("_EventCacheEntry", ("event", "redacted_event")) + + +def _retry_on_integrity_error(func): + """Wraps a database function so that it gets retried on IntegrityError, + with `delete_existing=True` passed in. + + Args: + func: function that returns a Deferred and accepts a `delete_existing` arg + """ + + @wraps(func) + @defer.inlineCallbacks + def f(self, *args, **kwargs): + try: + res = yield func(self, *args, **kwargs) + except self.database_engine.module.IntegrityError: + logger.exception("IntegrityError, retrying.") + res = yield func(self, *args, delete_existing=True, **kwargs) + return res + + return f + + +# inherits from EventFederationStore so that we can call _update_backward_extremities +# and _handle_mult_prev_events (though arguably those could both be moved in here) +class EventsStore( + StateGroupWorkerStore, + EventFederationStore, + EventsWorkerStore, + BackgroundUpdateStore, +): + def __init__(self, db_conn, hs): + super(EventsStore, self).__init__(db_conn, hs) + + self._event_persist_queue = _EventPeristenceQueue() + self._state_resolution_handler = hs.get_state_resolution_handler() + + # Collect metrics on the number of forward extremities that exist. + # Counter of number of extremities to count + self._current_forward_extremities_amount = c_counter() + + BucketCollector( + "synapse_forward_extremities", + lambda: self._current_forward_extremities_amount, + buckets=[1, 2, 3, 5, 7, 10, 15, 20, 50, 100, 200, 500, "+Inf"], + ) + + # Read the extrems every 60 minutes + def read_forward_extremities(): + # run as a background process to make sure that the database transactions + # have a logcontext to report to + return run_as_background_process( + "read_forward_extremities", self._read_forward_extremities + ) + + hs.get_clock().looping_call(read_forward_extremities, 60 * 60 * 1000) + + def _censor_redactions(): + return run_as_background_process( + "_censor_redactions", self._censor_redactions + ) + + if self.hs.config.redaction_retention_period is not None: + hs.get_clock().looping_call(_censor_redactions, 5 * 60 * 1000) + + @defer.inlineCallbacks + def _read_forward_extremities(self): + def fetch(txn): + txn.execute( + """ + select count(*) c from event_forward_extremities + group by room_id + """ + ) + return txn.fetchall() + + res = yield self.runInteraction("read_forward_extremities", fetch) + self._current_forward_extremities_amount = c_counter(list(x[0] for x in res)) + + @defer.inlineCallbacks + def persist_events(self, events_and_contexts, backfilled=False): + """ + Write events to the database + Args: + events_and_contexts: list of tuples of (event, context) + backfilled (bool): Whether the results are retrieved from federation + via backfill or not. Used to determine if they're "new" events + which might update the current state etc. + + Returns: + Deferred[int]: the stream ordering of the latest persisted event + """ + partitioned = {} + for event, ctx in events_and_contexts: + partitioned.setdefault(event.room_id, []).append((event, ctx)) + + deferreds = [] + for room_id, evs_ctxs in iteritems(partitioned): + d = self._event_persist_queue.add_to_queue( + room_id, evs_ctxs, backfilled=backfilled + ) + deferreds.append(d) + + for room_id in partitioned: + self._maybe_start_persisting(room_id) + + yield make_deferred_yieldable( + defer.gatherResults(deferreds, consumeErrors=True) + ) + + max_persisted_id = yield self._stream_id_gen.get_current_token() + + return max_persisted_id + + @defer.inlineCallbacks + @log_function + def persist_event(self, event, context, backfilled=False): + """ + + Args: + event (EventBase): + context (EventContext): + backfilled (bool): + + Returns: + Deferred: resolves to (int, int): the stream ordering of ``event``, + and the stream ordering of the latest persisted event + """ + deferred = self._event_persist_queue.add_to_queue( + event.room_id, [(event, context)], backfilled=backfilled + ) + + self._maybe_start_persisting(event.room_id) + + yield make_deferred_yieldable(deferred) + + max_persisted_id = yield self._stream_id_gen.get_current_token() + return (event.internal_metadata.stream_ordering, max_persisted_id) + + def _maybe_start_persisting(self, room_id): + @defer.inlineCallbacks + def persisting_queue(item): + with Measure(self._clock, "persist_events"): + yield self._persist_events( + item.events_and_contexts, backfilled=item.backfilled + ) + + self._event_persist_queue.handle_queue(room_id, persisting_queue) + + @_retry_on_integrity_error + @defer.inlineCallbacks + def _persist_events( + self, events_and_contexts, backfilled=False, delete_existing=False + ): + """Persist events to db + + Args: + events_and_contexts (list[(EventBase, EventContext)]): + backfilled (bool): + delete_existing (bool): + + Returns: + Deferred: resolves when the events have been persisted + """ + if not events_and_contexts: + return + + chunks = [ + events_and_contexts[x : x + 100] + for x in range(0, len(events_and_contexts), 100) + ] + + for chunk in chunks: + # We can't easily parallelize these since different chunks + # might contain the same event. :( + + # NB: Assumes that we are only persisting events for one room + # at a time. + + # map room_id->list[event_ids] giving the new forward + # extremities in each room + new_forward_extremeties = {} + + # map room_id->(type,state_key)->event_id tracking the full + # state in each room after adding these events. + # This is simply used to prefill the get_current_state_ids + # cache + current_state_for_room = {} + + # map room_id->(to_delete, to_insert) where to_delete is a list + # of type/state keys to remove from current state, and to_insert + # is a map (type,key)->event_id giving the state delta in each + # room + state_delta_for_room = {} + + if not backfilled: + with Measure(self._clock, "_calculate_state_and_extrem"): + # Work out the new "current state" for each room. + # We do this by working out what the new extremities are and then + # calculating the state from that. + events_by_room = {} + for event, context in chunk: + events_by_room.setdefault(event.room_id, []).append( + (event, context) + ) + + for room_id, ev_ctx_rm in iteritems(events_by_room): + latest_event_ids = yield self.get_latest_event_ids_in_room( + room_id + ) + new_latest_event_ids = yield self._calculate_new_extremities( + room_id, ev_ctx_rm, latest_event_ids + ) + + latest_event_ids = set(latest_event_ids) + if new_latest_event_ids == latest_event_ids: + # No change in extremities, so no change in state + continue + + # there should always be at least one forward extremity. + # (except during the initial persistence of the send_join + # results, in which case there will be no existing + # extremities, so we'll `continue` above and skip this bit.) + assert new_latest_event_ids, "No forward extremities left!" + + new_forward_extremeties[room_id] = new_latest_event_ids + + len_1 = ( + len(latest_event_ids) == 1 + and len(new_latest_event_ids) == 1 + ) + if len_1: + all_single_prev_not_state = all( + len(event.prev_event_ids()) == 1 + and not event.is_state() + for event, ctx in ev_ctx_rm + ) + # Don't bother calculating state if they're just + # a long chain of single ancestor non-state events. + if all_single_prev_not_state: + continue + + state_delta_counter.inc() + if len(new_latest_event_ids) == 1: + state_delta_single_event_counter.inc() + + # This is a fairly handwavey check to see if we could + # have guessed what the delta would have been when + # processing one of these events. + # What we're interested in is if the latest extremities + # were the same when we created the event as they are + # now. When this server creates a new event (as opposed + # to receiving it over federation) it will use the + # forward extremities as the prev_events, so we can + # guess this by looking at the prev_events and checking + # if they match the current forward extremities. + for ev, _ in ev_ctx_rm: + prev_event_ids = set(ev.prev_event_ids()) + if latest_event_ids == prev_event_ids: + state_delta_reuse_delta_counter.inc() + break + + logger.info("Calculating state delta for room %s", room_id) + with Measure( + self._clock, "persist_events.get_new_state_after_events" + ): + res = yield self._get_new_state_after_events( + room_id, + ev_ctx_rm, + latest_event_ids, + new_latest_event_ids, + ) + current_state, delta_ids = res + + # If either are not None then there has been a change, + # and we need to work out the delta (or use that + # given) + if delta_ids is not None: + # If there is a delta we know that we've + # only added or replaced state, never + # removed keys entirely. + state_delta_for_room[room_id] = ([], delta_ids) + elif current_state is not None: + with Measure( + self._clock, "persist_events.calculate_state_delta" + ): + delta = yield self._calculate_state_delta( + room_id, current_state + ) + state_delta_for_room[room_id] = delta + + # If we have the current_state then lets prefill + # the cache with it. + if current_state is not None: + current_state_for_room[room_id] = current_state + + # We want to calculate the stream orderings as late as possible, as + # we only notify after all events with a lesser stream ordering have + # been persisted. I.e. if we spend 10s inside the with block then + # that will delay all subsequent events from being notified about. + # Hence why we do it down here rather than wrapping the entire + # function. + # + # Its safe to do this after calculating the state deltas etc as we + # only need to protect the *persistence* of the events. This is to + # ensure that queries of the form "fetch events since X" don't + # return events and stream positions after events that are still in + # flight, as otherwise subsequent requests "fetch event since Y" + # will not return those events. + # + # Note: Multiple instances of this function cannot be in flight at + # the same time for the same room. + if backfilled: + stream_ordering_manager = self._backfill_id_gen.get_next_mult( + len(chunk) + ) + else: + stream_ordering_manager = self._stream_id_gen.get_next_mult(len(chunk)) + + with stream_ordering_manager as stream_orderings: + for (event, context), stream in zip(chunk, stream_orderings): + event.internal_metadata.stream_ordering = stream + + yield self.runInteraction( + "persist_events", + self._persist_events_txn, + events_and_contexts=chunk, + backfilled=backfilled, + delete_existing=delete_existing, + state_delta_for_room=state_delta_for_room, + new_forward_extremeties=new_forward_extremeties, + ) + persist_event_counter.inc(len(chunk)) + + if not backfilled: + # backfilled events have negative stream orderings, so we don't + # want to set the event_persisted_position to that. + synapse.metrics.event_persisted_position.set( + chunk[-1][0].internal_metadata.stream_ordering + ) + + for event, context in chunk: + if context.app_service: + origin_type = "local" + origin_entity = context.app_service.id + elif self.hs.is_mine_id(event.sender): + origin_type = "local" + origin_entity = "*client*" + else: + origin_type = "remote" + origin_entity = get_domain_from_id(event.sender) + + event_counter.labels(event.type, origin_type, origin_entity).inc() + + for room_id, new_state in iteritems(current_state_for_room): + self.get_current_state_ids.prefill((room_id,), new_state) + + for room_id, latest_event_ids in iteritems(new_forward_extremeties): + self.get_latest_event_ids_in_room.prefill( + (room_id,), list(latest_event_ids) + ) + + @defer.inlineCallbacks + def _calculate_new_extremities(self, room_id, event_contexts, latest_event_ids): + """Calculates the new forward extremities for a room given events to + persist. + + Assumes that we are only persisting events for one room at a time. + """ + + # we're only interested in new events which aren't outliers and which aren't + # being rejected. + new_events = [ + event + for event, ctx in event_contexts + if not event.internal_metadata.is_outlier() + and not ctx.rejected + and not event.internal_metadata.is_soft_failed() + ] + + latest_event_ids = set(latest_event_ids) + + # start with the existing forward extremities + result = set(latest_event_ids) + + # add all the new events to the list + result.update(event.event_id for event in new_events) + + # Now remove all events which are prev_events of any of the new events + result.difference_update( + e_id for event in new_events for e_id in event.prev_event_ids() + ) + + # Remove any events which are prev_events of any existing events. + existing_prevs = yield self._get_events_which_are_prevs(result) + result.difference_update(existing_prevs) + + # Finally handle the case where the new events have soft-failed prev + # events. If they do we need to remove them and their prev events, + # otherwise we end up with dangling extremities. + existing_prevs = yield self._get_prevs_before_rejected( + e_id for event in new_events for e_id in event.prev_event_ids() + ) + result.difference_update(existing_prevs) + + # We only update metrics for events that change forward extremities + # (e.g. we ignore backfill/outliers/etc) + if result != latest_event_ids: + forward_extremities_counter.observe(len(result)) + stale = latest_event_ids & result + stale_forward_extremities_counter.observe(len(stale)) + + return result + + @defer.inlineCallbacks + def _get_events_which_are_prevs(self, event_ids): + """Filter the supplied list of event_ids to get those which are prev_events of + existing (non-outlier/rejected) events. + + Args: + event_ids (Iterable[str]): event ids to filter + + Returns: + Deferred[List[str]]: filtered event ids + """ + results = [] + + def _get_events_which_are_prevs_txn(txn, batch): + sql = """ + SELECT prev_event_id, internal_metadata + FROM event_edges + INNER JOIN events USING (event_id) + LEFT JOIN rejections USING (event_id) + LEFT JOIN event_json USING (event_id) + WHERE + NOT events.outlier + AND rejections.event_id IS NULL + AND + """ + + clause, args = make_in_list_sql_clause( + self.database_engine, "prev_event_id", batch + ) + + txn.execute(sql + clause, args) + results.extend(r[0] for r in txn if not json.loads(r[1]).get("soft_failed")) + + for chunk in batch_iter(event_ids, 100): + yield self.runInteraction( + "_get_events_which_are_prevs", _get_events_which_are_prevs_txn, chunk + ) + + return results + + @defer.inlineCallbacks + def _get_prevs_before_rejected(self, event_ids): + """Get soft-failed ancestors to remove from the extremities. + + Given a set of events, find all those that have been soft-failed or + rejected. Returns those soft failed/rejected events and their prev + events (whether soft-failed/rejected or not), and recurses up the + prev-event graph until it finds no more soft-failed/rejected events. + + This is used to find extremities that are ancestors of new events, but + are separated by soft failed events. + + Args: + event_ids (Iterable[str]): Events to find prev events for. Note + that these must have already been persisted. + + Returns: + Deferred[set[str]] + """ + + # The set of event_ids to return. This includes all soft-failed events + # and their prev events. + existing_prevs = set() + + def _get_prevs_before_rejected_txn(txn, batch): + to_recursively_check = batch + + while to_recursively_check: + sql = """ + SELECT + event_id, prev_event_id, internal_metadata, + rejections.event_id IS NOT NULL + FROM event_edges + INNER JOIN events USING (event_id) + LEFT JOIN rejections USING (event_id) + LEFT JOIN event_json USING (event_id) + WHERE + NOT events.outlier + AND + """ + + clause, args = make_in_list_sql_clause( + self.database_engine, "event_id", to_recursively_check + ) + + txn.execute(sql + clause, args) + to_recursively_check = [] + + for event_id, prev_event_id, metadata, rejected in txn: + if prev_event_id in existing_prevs: + continue + + soft_failed = json.loads(metadata).get("soft_failed") + if soft_failed or rejected: + to_recursively_check.append(prev_event_id) + existing_prevs.add(prev_event_id) + + for chunk in batch_iter(event_ids, 100): + yield self.runInteraction( + "_get_prevs_before_rejected", _get_prevs_before_rejected_txn, chunk + ) + + return existing_prevs + + @defer.inlineCallbacks + def _get_new_state_after_events( + self, room_id, events_context, old_latest_event_ids, new_latest_event_ids + ): + """Calculate the current state dict after adding some new events to + a room + + Args: + room_id (str): + room to which the events are being added. Used for logging etc + + events_context (list[(EventBase, EventContext)]): + events and contexts which are being added to the room + + old_latest_event_ids (iterable[str]): + the old forward extremities for the room. + + new_latest_event_ids (iterable[str]): + the new forward extremities for the room. + + Returns: + Deferred[tuple[dict[(str,str), str]|None, dict[(str,str), str]|None]]: + Returns a tuple of two state maps, the first being the full new current + state and the second being the delta to the existing current state. + If both are None then there has been no change. + + If there has been a change then we only return the delta if its + already been calculated. Conversely if we do know the delta then + the new current state is only returned if we've already calculated + it. + """ + # map from state_group to ((type, key) -> event_id) state map + state_groups_map = {} + + # Map from (prev state group, new state group) -> delta state dict + state_group_deltas = {} + + for ev, ctx in events_context: + if ctx.state_group is None: + # This should only happen for outlier events. + if not ev.internal_metadata.is_outlier(): + raise Exception( + "Context for new event %s has no state " + "group" % (ev.event_id,) + ) + continue + + if ctx.state_group in state_groups_map: + continue + + # We're only interested in pulling out state that has already + # been cached in the context. We'll pull stuff out of the DB later + # if necessary. + current_state_ids = ctx.get_cached_current_state_ids() + if current_state_ids is not None: + state_groups_map[ctx.state_group] = current_state_ids + + if ctx.prev_group: + state_group_deltas[(ctx.prev_group, ctx.state_group)] = ctx.delta_ids + + # We need to map the event_ids to their state groups. First, let's + # check if the event is one we're persisting, in which case we can + # pull the state group from its context. + # Otherwise we need to pull the state group from the database. + + # Set of events we need to fetch groups for. (We know none of the old + # extremities are going to be in events_context). + missing_event_ids = set(old_latest_event_ids) + + event_id_to_state_group = {} + for event_id in new_latest_event_ids: + # First search in the list of new events we're adding. + for ev, ctx in events_context: + if event_id == ev.event_id and ctx.state_group is not None: + event_id_to_state_group[event_id] = ctx.state_group + break + else: + # If we couldn't find it, then we'll need to pull + # the state from the database + missing_event_ids.add(event_id) + + if missing_event_ids: + # Now pull out the state groups for any missing events from DB + event_to_groups = yield self._get_state_group_for_events(missing_event_ids) + event_id_to_state_group.update(event_to_groups) + + # State groups of old_latest_event_ids + old_state_groups = set( + event_id_to_state_group[evid] for evid in old_latest_event_ids + ) + + # State groups of new_latest_event_ids + new_state_groups = set( + event_id_to_state_group[evid] for evid in new_latest_event_ids + ) + + # If they old and new groups are the same then we don't need to do + # anything. + if old_state_groups == new_state_groups: + return None, None + + if len(new_state_groups) == 1 and len(old_state_groups) == 1: + # If we're going from one state group to another, lets check if + # we have a delta for that transition. If we do then we can just + # return that. + + new_state_group = next(iter(new_state_groups)) + old_state_group = next(iter(old_state_groups)) + + delta_ids = state_group_deltas.get((old_state_group, new_state_group), None) + if delta_ids is not None: + # We have a delta from the existing to new current state, + # so lets just return that. If we happen to already have + # the current state in memory then lets also return that, + # but it doesn't matter if we don't. + new_state = state_groups_map.get(new_state_group) + return new_state, delta_ids + + # Now that we have calculated new_state_groups we need to get + # their state IDs so we can resolve to a single state set. + missing_state = new_state_groups - set(state_groups_map) + if missing_state: + group_to_state = yield self._get_state_for_groups(missing_state) + state_groups_map.update(group_to_state) + + if len(new_state_groups) == 1: + # If there is only one state group, then we know what the current + # state is. + return state_groups_map[new_state_groups.pop()], None + + # Ok, we need to defer to the state handler to resolve our state sets. + + state_groups = {sg: state_groups_map[sg] for sg in new_state_groups} + + events_map = {ev.event_id: ev for ev, _ in events_context} + + # We need to get the room version, which is in the create event. + # Normally that'd be in the database, but its also possible that we're + # currently trying to persist it. + room_version = None + for ev, _ in events_context: + if ev.type == EventTypes.Create and ev.state_key == "": + room_version = ev.content.get("room_version", "1") + break + + if not room_version: + room_version = yield self.get_room_version(room_id) + + logger.debug("calling resolve_state_groups from preserve_events") + res = yield self._state_resolution_handler.resolve_state_groups( + room_id, + room_version, + state_groups, + events_map, + state_res_store=StateResolutionStore(self), + ) + + return res.state, None + + @defer.inlineCallbacks + def _calculate_state_delta(self, room_id, current_state): + """Calculate the new state deltas for a room. + + Assumes that we are only persisting events for one room at a time. + + Returns: + tuple[list, dict] (to_delete, to_insert): where to_delete are the + type/state_keys to remove from current_state_events and `to_insert` + are the updates to current_state_events. + """ + existing_state = yield self.get_current_state_ids(room_id) + + to_delete = [key for key in existing_state if key not in current_state] + + to_insert = { + key: ev_id + for key, ev_id in iteritems(current_state) + if ev_id != existing_state.get(key) + } + + return to_delete, to_insert + + @log_function + def _persist_events_txn( + self, + txn, + events_and_contexts, + backfilled, + delete_existing=False, + state_delta_for_room={}, + new_forward_extremeties={}, + ): + """Insert some number of room events into the necessary database tables. + + Rejected events are only inserted into the events table, the events_json table, + and the rejections table. Things reading from those table will need to check + whether the event was rejected. + + Args: + txn (twisted.enterprise.adbapi.Connection): db connection + events_and_contexts (list[(EventBase, EventContext)]): + events to persist + backfilled (bool): True if the events were backfilled + delete_existing (bool): True to purge existing table rows for the + events from the database. This is useful when retrying due to + IntegrityError. + state_delta_for_room (dict[str, (list, dict)]): + The current-state delta for each room. For each room, a tuple + (to_delete, to_insert), being a list of type/state keys to be + removed from the current state, and a state set to be added to + the current state. + new_forward_extremeties (dict[str, list[str]]): + The new forward extremities for each room. For each room, a + list of the event ids which are the forward extremities. + + """ + all_events_and_contexts = events_and_contexts + + min_stream_order = events_and_contexts[0][0].internal_metadata.stream_ordering + max_stream_order = events_and_contexts[-1][0].internal_metadata.stream_ordering + + self._update_forward_extremities_txn( + txn, + new_forward_extremities=new_forward_extremeties, + max_stream_order=max_stream_order, + ) + + # Ensure that we don't have the same event twice. + events_and_contexts = self._filter_events_and_contexts_for_duplicates( + events_and_contexts + ) + + self._update_room_depths_txn( + txn, events_and_contexts=events_and_contexts, backfilled=backfilled + ) + + # _update_outliers_txn filters out any events which have already been + # persisted, and returns the filtered list. + events_and_contexts = self._update_outliers_txn( + txn, events_and_contexts=events_and_contexts + ) + + # From this point onwards the events are only events that we haven't + # seen before. + + if delete_existing: + # For paranoia reasons, we go and delete all the existing entries + # for these events so we can reinsert them. + # This gets around any problems with some tables already having + # entries. + self._delete_existing_rows_txn(txn, events_and_contexts=events_and_contexts) + + self._store_event_txn(txn, events_and_contexts=events_and_contexts) + + # Insert into event_to_state_groups. + self._store_event_state_mappings_txn(txn, events_and_contexts) + + # We want to store event_auth mappings for rejected events, as they're + # used in state res v2. + # This is only necessary if the rejected event appears in an accepted + # event's auth chain, but its easier for now just to store them (and + # it doesn't take much storage compared to storing the entire event + # anyway). + self._simple_insert_many_txn( + txn, + table="event_auth", + values=[ + { + "event_id": event.event_id, + "room_id": event.room_id, + "auth_id": auth_id, + } + for event, _ in events_and_contexts + for auth_id in event.auth_event_ids() + if event.is_state() + ], + ) + + # _store_rejected_events_txn filters out any events which were + # rejected, and returns the filtered list. + events_and_contexts = self._store_rejected_events_txn( + txn, events_and_contexts=events_and_contexts + ) + + # From this point onwards the events are only ones that weren't + # rejected. + + self._update_metadata_tables_txn( + txn, + events_and_contexts=events_and_contexts, + all_events_and_contexts=all_events_and_contexts, + backfilled=backfilled, + ) + + # We call this last as it assumes we've inserted the events into + # room_memberships, where applicable. + self._update_current_state_txn(txn, state_delta_for_room, min_stream_order) + + def _update_current_state_txn(self, txn, state_delta_by_room, stream_id): + for room_id, current_state_tuple in iteritems(state_delta_by_room): + to_delete, to_insert = current_state_tuple + + # First we add entries to the current_state_delta_stream. We + # do this before updating the current_state_events table so + # that we can use it to calculate the `prev_event_id`. (This + # allows us to not have to pull out the existing state + # unnecessarily). + # + # The stream_id for the update is chosen to be the minimum of the stream_ids + # for the batch of the events that we are persisting; that means we do not + # end up in a situation where workers see events before the + # current_state_delta updates. + # + sql = """ + INSERT INTO current_state_delta_stream + (stream_id, room_id, type, state_key, event_id, prev_event_id) + SELECT ?, ?, ?, ?, ?, ( + SELECT event_id FROM current_state_events + WHERE room_id = ? AND type = ? AND state_key = ? + ) + """ + txn.executemany( + sql, + ( + ( + stream_id, + room_id, + etype, + state_key, + None, + room_id, + etype, + state_key, + ) + for etype, state_key in to_delete + # We sanity check that we're deleting rather than updating + if (etype, state_key) not in to_insert + ), + ) + txn.executemany( + sql, + ( + ( + stream_id, + room_id, + etype, + state_key, + ev_id, + room_id, + etype, + state_key, + ) + for (etype, state_key), ev_id in iteritems(to_insert) + ), + ) + + # Now we actually update the current_state_events table + + txn.executemany( + "DELETE FROM current_state_events" + " WHERE room_id = ? AND type = ? AND state_key = ?", + ( + (room_id, etype, state_key) + for etype, state_key in itertools.chain(to_delete, to_insert) + ), + ) + + # We include the membership in the current state table, hence we do + # a lookup when we insert. This assumes that all events have already + # been inserted into room_memberships. + txn.executemany( + """INSERT INTO current_state_events + (room_id, type, state_key, event_id, membership) + VALUES (?, ?, ?, ?, (SELECT membership FROM room_memberships WHERE event_id = ?)) + """, + [ + (room_id, key[0], key[1], ev_id, ev_id) + for key, ev_id in iteritems(to_insert) + ], + ) + + txn.call_after( + self._curr_state_delta_stream_cache.entity_has_changed, + room_id, + stream_id, + ) + + # Invalidate the various caches + + # Figure out the changes of membership to invalidate the + # `get_rooms_for_user` cache. + # We find out which membership events we may have deleted + # and which we have added, then we invlidate the caches for all + # those users. + members_changed = set( + state_key + for ev_type, state_key in itertools.chain(to_delete, to_insert) + if ev_type == EventTypes.Member + ) + + for member in members_changed: + txn.call_after( + self.get_rooms_for_user_with_stream_ordering.invalidate, (member,) + ) + + self._invalidate_state_caches_and_stream(txn, room_id, members_changed) + + def _update_forward_extremities_txn( + self, txn, new_forward_extremities, max_stream_order + ): + for room_id, new_extrem in iteritems(new_forward_extremities): + self._simple_delete_txn( + txn, table="event_forward_extremities", keyvalues={"room_id": room_id} + ) + txn.call_after(self.get_latest_event_ids_in_room.invalidate, (room_id,)) + + self._simple_insert_many_txn( + txn, + table="event_forward_extremities", + values=[ + {"event_id": ev_id, "room_id": room_id} + for room_id, new_extrem in iteritems(new_forward_extremities) + for ev_id in new_extrem + ], + ) + # We now insert into stream_ordering_to_exterm a mapping from room_id, + # new stream_ordering to new forward extremeties in the room. + # This allows us to later efficiently look up the forward extremeties + # for a room before a given stream_ordering + self._simple_insert_many_txn( + txn, + table="stream_ordering_to_exterm", + values=[ + { + "room_id": room_id, + "event_id": event_id, + "stream_ordering": max_stream_order, + } + for room_id, new_extrem in iteritems(new_forward_extremities) + for event_id in new_extrem + ], + ) + + @classmethod + def _filter_events_and_contexts_for_duplicates(cls, events_and_contexts): + """Ensure that we don't have the same event twice. + + Pick the earliest non-outlier if there is one, else the earliest one. + + Args: + events_and_contexts (list[(EventBase, EventContext)]): + Returns: + list[(EventBase, EventContext)]: filtered list + """ + new_events_and_contexts = OrderedDict() + for event, context in events_and_contexts: + prev_event_context = new_events_and_contexts.get(event.event_id) + if prev_event_context: + if not event.internal_metadata.is_outlier(): + if prev_event_context[0].internal_metadata.is_outlier(): + # To ensure correct ordering we pop, as OrderedDict is + # ordered by first insertion. + new_events_and_contexts.pop(event.event_id, None) + new_events_and_contexts[event.event_id] = (event, context) + else: + new_events_and_contexts[event.event_id] = (event, context) + return list(new_events_and_contexts.values()) + + def _update_room_depths_txn(self, txn, events_and_contexts, backfilled): + """Update min_depth for each room + + Args: + txn (twisted.enterprise.adbapi.Connection): db connection + events_and_contexts (list[(EventBase, EventContext)]): events + we are persisting + backfilled (bool): True if the events were backfilled + """ + depth_updates = {} + for event, context in events_and_contexts: + # Remove the any existing cache entries for the event_ids + txn.call_after(self._invalidate_get_event_cache, event.event_id) + if not backfilled: + txn.call_after( + self._events_stream_cache.entity_has_changed, + event.room_id, + event.internal_metadata.stream_ordering, + ) + + if not event.internal_metadata.is_outlier() and not context.rejected: + depth_updates[event.room_id] = max( + event.depth, depth_updates.get(event.room_id, event.depth) + ) + + for room_id, depth in iteritems(depth_updates): + self._update_min_depth_for_room_txn(txn, room_id, depth) + + def _update_outliers_txn(self, txn, events_and_contexts): + """Update any outliers with new event info. + + This turns outliers into ex-outliers (unless the new event was + rejected). + + Args: + txn (twisted.enterprise.adbapi.Connection): db connection + events_and_contexts (list[(EventBase, EventContext)]): events + we are persisting + + Returns: + list[(EventBase, EventContext)] new list, without events which + are already in the events table. + """ + txn.execute( + "SELECT event_id, outlier FROM events WHERE event_id in (%s)" + % (",".join(["?"] * len(events_and_contexts)),), + [event.event_id for event, _ in events_and_contexts], + ) + + have_persisted = {event_id: outlier for event_id, outlier in txn} + + to_remove = set() + for event, context in events_and_contexts: + if event.event_id not in have_persisted: + continue + + to_remove.add(event) + + if context.rejected: + # If the event is rejected then we don't care if the event + # was an outlier or not. + continue + + outlier_persisted = have_persisted[event.event_id] + if not event.internal_metadata.is_outlier() and outlier_persisted: + # We received a copy of an event that we had already stored as + # an outlier in the database. We now have some state at that + # so we need to update the state_groups table with that state. + + # insert into event_to_state_groups. + try: + self._store_event_state_mappings_txn(txn, ((event, context),)) + except Exception: + logger.exception("") + raise + + metadata_json = encode_json(event.internal_metadata.get_dict()) + + sql = ( + "UPDATE event_json SET internal_metadata = ?" " WHERE event_id = ?" + ) + txn.execute(sql, (metadata_json, event.event_id)) + + # Add an entry to the ex_outlier_stream table to replicate the + # change in outlier status to our workers. + stream_order = event.internal_metadata.stream_ordering + state_group_id = context.state_group + self._simple_insert_txn( + txn, + table="ex_outlier_stream", + values={ + "event_stream_ordering": stream_order, + "event_id": event.event_id, + "state_group": state_group_id, + }, + ) + + sql = "UPDATE events SET outlier = ?" " WHERE event_id = ?" + txn.execute(sql, (False, event.event_id)) + + # Update the event_backward_extremities table now that this + # event isn't an outlier any more. + self._update_backward_extremeties(txn, [event]) + + return [ec for ec in events_and_contexts if ec[0] not in to_remove] + + @classmethod + def _delete_existing_rows_txn(cls, txn, events_and_contexts): + if not events_and_contexts: + # nothing to do here + return + + logger.info("Deleting existing") + + for table in ( + "events", + "event_auth", + "event_json", + "event_edges", + "event_forward_extremities", + "event_reference_hashes", + "event_search", + "event_to_state_groups", + "local_invites", + "state_events", + "rejections", + "redactions", + "room_memberships", + ): + txn.executemany( + "DELETE FROM %s WHERE event_id = ?" % (table,), + [(ev.event_id,) for ev, _ in events_and_contexts], + ) + + for table in ("event_push_actions",): + txn.executemany( + "DELETE FROM %s WHERE room_id = ? AND event_id = ?" % (table,), + [(ev.room_id, ev.event_id) for ev, _ in events_and_contexts], + ) + + def _store_event_txn(self, txn, events_and_contexts): + """Insert new events into the event and event_json tables + + Args: + txn (twisted.enterprise.adbapi.Connection): db connection + events_and_contexts (list[(EventBase, EventContext)]): events + we are persisting + """ + + if not events_and_contexts: + # nothing to do here + return + + def event_dict(event): + d = event.get_dict() + d.pop("redacted", None) + d.pop("redacted_because", None) + return d + + self._simple_insert_many_txn( + txn, + table="event_json", + values=[ + { + "event_id": event.event_id, + "room_id": event.room_id, + "internal_metadata": encode_json( + event.internal_metadata.get_dict() + ), + "json": encode_json(event_dict(event)), + "format_version": event.format_version, + } + for event, _ in events_and_contexts + ], + ) + + self._simple_insert_many_txn( + txn, + table="events", + values=[ + { + "stream_ordering": event.internal_metadata.stream_ordering, + "topological_ordering": event.depth, + "depth": event.depth, + "event_id": event.event_id, + "room_id": event.room_id, + "type": event.type, + "processed": True, + "outlier": event.internal_metadata.is_outlier(), + "origin_server_ts": int(event.origin_server_ts), + "received_ts": self._clock.time_msec(), + "sender": event.sender, + "contains_url": ( + "url" in event.content + and isinstance(event.content["url"], text_type) + ), + } + for event, _ in events_and_contexts + ], + ) + + for event, _ in events_and_contexts: + if not event.internal_metadata.is_redacted(): + # If we're persisting an unredacted event we go and ensure + # that we mark any redactions that reference this event as + # requiring censoring. + self._simple_update_txn( + txn, + table="redactions", + keyvalues={"redacts": event.event_id}, + updatevalues={"have_censored": False}, + ) + + def _store_rejected_events_txn(self, txn, events_and_contexts): + """Add rows to the 'rejections' table for received events which were + rejected + + Args: + txn (twisted.enterprise.adbapi.Connection): db connection + events_and_contexts (list[(EventBase, EventContext)]): events + we are persisting + + Returns: + list[(EventBase, EventContext)] new list, without the rejected + events. + """ + # Remove the rejected events from the list now that we've added them + # to the events table and the events_json table. + to_remove = set() + for event, context in events_and_contexts: + if context.rejected: + # Insert the event_id into the rejections table + self._store_rejections_txn(txn, event.event_id, context.rejected) + to_remove.add(event) + + return [ec for ec in events_and_contexts if ec[0] not in to_remove] + + def _update_metadata_tables_txn( + self, txn, events_and_contexts, all_events_and_contexts, backfilled + ): + """Update all the miscellaneous tables for new events + + Args: + txn (twisted.enterprise.adbapi.Connection): db connection + events_and_contexts (list[(EventBase, EventContext)]): events + we are persisting + all_events_and_contexts (list[(EventBase, EventContext)]): all + events that we were going to persist. This includes events + we've already persisted, etc, that wouldn't appear in + events_and_context. + backfilled (bool): True if the events were backfilled + """ + + # Insert all the push actions into the event_push_actions table. + self._set_push_actions_for_event_and_users_txn( + txn, + events_and_contexts=events_and_contexts, + all_events_and_contexts=all_events_and_contexts, + ) + + if not events_and_contexts: + # nothing to do here + return + + for event, context in events_and_contexts: + if event.type == EventTypes.Redaction and event.redacts is not None: + # Remove the entries in the event_push_actions table for the + # redacted event. + self._remove_push_actions_for_event_id_txn( + txn, event.room_id, event.redacts + ) + + # Remove from relations table. + self._handle_redaction(txn, event.redacts) + + # Update the event_forward_extremities, event_backward_extremities and + # event_edges tables. + self._handle_mult_prev_events( + txn, events=[event for event, _ in events_and_contexts] + ) + + for event, _ in events_and_contexts: + if event.type == EventTypes.Name: + # Insert into the event_search table. + self._store_room_name_txn(txn, event) + elif event.type == EventTypes.Topic: + # Insert into the event_search table. + self._store_room_topic_txn(txn, event) + elif event.type == EventTypes.Message: + # Insert into the event_search table. + self._store_room_message_txn(txn, event) + elif event.type == EventTypes.Redaction: + # Insert into the redactions table. + self._store_redaction(txn, event) + + self._handle_event_relations(txn, event) + + # Insert into the room_memberships table. + self._store_room_members_txn( + txn, + [ + event + for event, _ in events_and_contexts + if event.type == EventTypes.Member + ], + backfilled=backfilled, + ) + + # Insert event_reference_hashes table. + self._store_event_reference_hashes_txn( + txn, [event for event, _ in events_and_contexts] + ) + + state_events_and_contexts = [ + ec for ec in events_and_contexts if ec[0].is_state() + ] + + state_values = [] + for event, context in state_events_and_contexts: + vals = { + "event_id": event.event_id, + "room_id": event.room_id, + "type": event.type, + "state_key": event.state_key, + } + + # TODO: How does this work with backfilling? + if hasattr(event, "replaces_state"): + vals["prev_state"] = event.replaces_state + + state_values.append(vals) + + self._simple_insert_many_txn(txn, table="state_events", values=state_values) + + # Prefill the event cache + self._add_to_cache(txn, events_and_contexts) + + def _add_to_cache(self, txn, events_and_contexts): + to_prefill = [] + + rows = [] + N = 200 + for i in range(0, len(events_and_contexts), N): + ev_map = {e[0].event_id: e[0] for e in events_and_contexts[i : i + N]} + if not ev_map: + break + + sql = ( + "SELECT " + " e.event_id as event_id, " + " r.redacts as redacts," + " rej.event_id as rejects " + " FROM events as e" + " LEFT JOIN rejections as rej USING (event_id)" + " LEFT JOIN redactions as r ON e.event_id = r.redacts" + " WHERE " + ) + + clause, args = make_in_list_sql_clause( + self.database_engine, "e.event_id", list(ev_map) + ) + + txn.execute(sql + clause, args) + rows = self.cursor_to_dict(txn) + for row in rows: + event = ev_map[row["event_id"]] + if not row["rejects"] and not row["redacts"]: + to_prefill.append( + _EventCacheEntry(event=event, redacted_event=None) + ) + + def prefill(): + for cache_entry in to_prefill: + self._get_event_cache.prefill((cache_entry[0].event_id,), cache_entry) + + txn.call_after(prefill) + + def _store_redaction(self, txn, event): + # invalidate the cache for the redacted event + txn.call_after(self._invalidate_get_event_cache, event.redacts) + + self._simple_insert_txn( + txn, + table="redactions", + values={ + "event_id": event.event_id, + "redacts": event.redacts, + "received_ts": self._clock.time_msec(), + }, + ) + + @defer.inlineCallbacks + def _censor_redactions(self): + """Censors all redactions older than the configured period that haven't + been censored yet. + + By censor we mean update the event_json table with the redacted event. + + Returns: + Deferred + """ + + if self.hs.config.redaction_retention_period is None: + return + + before_ts = self._clock.time_msec() - self.hs.config.redaction_retention_period + + # We fetch all redactions that: + # 1. point to an event we have, + # 2. has a received_ts from before the cut off, and + # 3. we haven't yet censored. + # + # This is limited to 100 events to ensure that we don't try and do too + # much at once. We'll get called again so this should eventually catch + # up. + sql = """ + SELECT redactions.event_id, redacts FROM redactions + LEFT JOIN events AS original_event ON ( + redacts = original_event.event_id + ) + WHERE NOT have_censored + AND redactions.received_ts <= ? + ORDER BY redactions.received_ts ASC + LIMIT ? + """ + + rows = yield self._execute( + "_censor_redactions_fetch", None, sql, before_ts, 100 + ) + + updates = [] + + for redaction_id, event_id in rows: + redaction_event = yield self.get_event(redaction_id, allow_none=True) + original_event = yield self.get_event( + event_id, allow_rejected=True, allow_none=True + ) + + # The SQL above ensures that we have both the redaction and + # original event, so if the `get_event` calls return None it + # means that the redaction wasn't allowed. Either way we know that + # the result won't change so we mark the fact that we've checked. + if ( + redaction_event + and original_event + and original_event.internal_metadata.is_redacted() + ): + # Redaction was allowed + pruned_json = encode_json(prune_event_dict(original_event.get_dict())) + else: + # Redaction wasn't allowed + pruned_json = None + + updates.append((redaction_id, event_id, pruned_json)) + + def _update_censor_txn(txn): + for redaction_id, event_id, pruned_json in updates: + if pruned_json: + self._simple_update_one_txn( + txn, + table="event_json", + keyvalues={"event_id": event_id}, + updatevalues={"json": pruned_json}, + ) + + self._simple_update_one_txn( + txn, + table="redactions", + keyvalues={"event_id": redaction_id}, + updatevalues={"have_censored": True}, + ) + + yield self.runInteraction("_update_censor_txn", _update_censor_txn) + + @defer.inlineCallbacks + def count_daily_messages(self): + """ + Returns an estimate of the number of messages sent in the last day. + + If it has been significantly less or more than one day since the last + call to this function, it will return None. + """ + + def _count_messages(txn): + sql = """ + SELECT COALESCE(COUNT(*), 0) FROM events + WHERE type = 'm.room.message' + AND stream_ordering > ? + """ + txn.execute(sql, (self.stream_ordering_day_ago,)) + count, = txn.fetchone() + return count + + ret = yield self.runInteraction("count_messages", _count_messages) + return ret + + @defer.inlineCallbacks + def count_daily_sent_messages(self): + def _count_messages(txn): + # This is good enough as if you have silly characters in your own + # hostname then thats your own fault. + like_clause = "%:" + self.hs.hostname + + sql = """ + SELECT COALESCE(COUNT(*), 0) FROM events + WHERE type = 'm.room.message' + AND sender LIKE ? + AND stream_ordering > ? + """ + + txn.execute(sql, (like_clause, self.stream_ordering_day_ago)) + count, = txn.fetchone() + return count + + ret = yield self.runInteraction("count_daily_sent_messages", _count_messages) + return ret + + @defer.inlineCallbacks + def count_daily_active_rooms(self): + def _count(txn): + sql = """ + SELECT COALESCE(COUNT(DISTINCT room_id), 0) FROM events + WHERE type = 'm.room.message' + AND stream_ordering > ? + """ + txn.execute(sql, (self.stream_ordering_day_ago,)) + count, = txn.fetchone() + return count + + ret = yield self.runInteraction("count_daily_active_rooms", _count) + return ret + + def get_current_backfill_token(self): + """The current minimum token that backfilled events have reached""" + return -self._backfill_id_gen.get_current_token() + + def get_current_events_token(self): + """The current maximum token that events have reached""" + return self._stream_id_gen.get_current_token() + + def get_all_new_forward_event_rows(self, last_id, current_id, limit): + if last_id == current_id: + return defer.succeed([]) + + def get_all_new_forward_event_rows(txn): + sql = ( + "SELECT e.stream_ordering, e.event_id, e.room_id, e.type," + " state_key, redacts, relates_to_id" + " FROM events AS e" + " LEFT JOIN redactions USING (event_id)" + " LEFT JOIN state_events USING (event_id)" + " LEFT JOIN event_relations USING (event_id)" + " WHERE ? < stream_ordering AND stream_ordering <= ?" + " ORDER BY stream_ordering ASC" + " LIMIT ?" + ) + txn.execute(sql, (last_id, current_id, limit)) + new_event_updates = txn.fetchall() + + if len(new_event_updates) == limit: + upper_bound = new_event_updates[-1][0] + else: + upper_bound = current_id + + sql = ( + "SELECT event_stream_ordering, e.event_id, e.room_id, e.type," + " state_key, redacts, relates_to_id" + " FROM events AS e" + " INNER JOIN ex_outlier_stream USING (event_id)" + " LEFT JOIN redactions USING (event_id)" + " LEFT JOIN state_events USING (event_id)" + " LEFT JOIN event_relations USING (event_id)" + " WHERE ? < event_stream_ordering" + " AND event_stream_ordering <= ?" + " ORDER BY event_stream_ordering DESC" + ) + txn.execute(sql, (last_id, upper_bound)) + new_event_updates.extend(txn) + + return new_event_updates + + return self.runInteraction( + "get_all_new_forward_event_rows", get_all_new_forward_event_rows + ) + + def get_all_new_backfill_event_rows(self, last_id, current_id, limit): + if last_id == current_id: + return defer.succeed([]) + + def get_all_new_backfill_event_rows(txn): + sql = ( + "SELECT -e.stream_ordering, e.event_id, e.room_id, e.type," + " state_key, redacts, relates_to_id" + " FROM events AS e" + " LEFT JOIN redactions USING (event_id)" + " LEFT JOIN state_events USING (event_id)" + " LEFT JOIN event_relations USING (event_id)" + " WHERE ? > stream_ordering AND stream_ordering >= ?" + " ORDER BY stream_ordering ASC" + " LIMIT ?" + ) + txn.execute(sql, (-last_id, -current_id, limit)) + new_event_updates = txn.fetchall() + + if len(new_event_updates) == limit: + upper_bound = new_event_updates[-1][0] + else: + upper_bound = current_id + + sql = ( + "SELECT -event_stream_ordering, e.event_id, e.room_id, e.type," + " state_key, redacts, relates_to_id" + " FROM events AS e" + " INNER JOIN ex_outlier_stream USING (event_id)" + " LEFT JOIN redactions USING (event_id)" + " LEFT JOIN state_events USING (event_id)" + " LEFT JOIN event_relations USING (event_id)" + " WHERE ? > event_stream_ordering" + " AND event_stream_ordering >= ?" + " ORDER BY event_stream_ordering DESC" + ) + txn.execute(sql, (-last_id, -upper_bound)) + new_event_updates.extend(txn.fetchall()) + + return new_event_updates + + return self.runInteraction( + "get_all_new_backfill_event_rows", get_all_new_backfill_event_rows + ) + + @cached(num_args=5, max_entries=10) + def get_all_new_events( + self, + last_backfill_id, + last_forward_id, + current_backfill_id, + current_forward_id, + limit, + ): + """Get all the new events that have arrived at the server either as + new events or as backfilled events""" + have_backfill_events = last_backfill_id != current_backfill_id + have_forward_events = last_forward_id != current_forward_id + + if not have_backfill_events and not have_forward_events: + return defer.succeed(AllNewEventsResult([], [], [], [], [])) + + def get_all_new_events_txn(txn): + sql = ( + "SELECT e.stream_ordering, e.event_id, e.room_id, e.type," + " state_key, redacts" + " FROM events AS e" + " LEFT JOIN redactions USING (event_id)" + " LEFT JOIN state_events USING (event_id)" + " WHERE ? < stream_ordering AND stream_ordering <= ?" + " ORDER BY stream_ordering ASC" + " LIMIT ?" + ) + if have_forward_events: + txn.execute(sql, (last_forward_id, current_forward_id, limit)) + new_forward_events = txn.fetchall() + + if len(new_forward_events) == limit: + upper_bound = new_forward_events[-1][0] + else: + upper_bound = current_forward_id + + sql = ( + "SELECT event_stream_ordering, event_id, state_group" + " FROM ex_outlier_stream" + " WHERE ? > event_stream_ordering" + " AND event_stream_ordering >= ?" + " ORDER BY event_stream_ordering DESC" + ) + txn.execute(sql, (last_forward_id, upper_bound)) + forward_ex_outliers = txn.fetchall() + else: + new_forward_events = [] + forward_ex_outliers = [] + + sql = ( + "SELECT -e.stream_ordering, e.event_id, e.room_id, e.type," + " state_key, redacts" + " FROM events AS e" + " LEFT JOIN redactions USING (event_id)" + " LEFT JOIN state_events USING (event_id)" + " WHERE ? > stream_ordering AND stream_ordering >= ?" + " ORDER BY stream_ordering DESC" + " LIMIT ?" + ) + if have_backfill_events: + txn.execute(sql, (-last_backfill_id, -current_backfill_id, limit)) + new_backfill_events = txn.fetchall() + + if len(new_backfill_events) == limit: + upper_bound = new_backfill_events[-1][0] + else: + upper_bound = current_backfill_id + + sql = ( + "SELECT -event_stream_ordering, event_id, state_group" + " FROM ex_outlier_stream" + " WHERE ? > event_stream_ordering" + " AND event_stream_ordering >= ?" + " ORDER BY event_stream_ordering DESC" + ) + txn.execute(sql, (-last_backfill_id, -upper_bound)) + backward_ex_outliers = txn.fetchall() + else: + new_backfill_events = [] + backward_ex_outliers = [] + + return AllNewEventsResult( + new_forward_events, + new_backfill_events, + forward_ex_outliers, + backward_ex_outliers, + ) + + return self.runInteraction("get_all_new_events", get_all_new_events_txn) + + def purge_history(self, room_id, token, delete_local_events): + """Deletes room history before a certain point + + Args: + room_id (str): + + token (str): A topological token to delete events before + + delete_local_events (bool): + if True, we will delete local events as well as remote ones + (instead of just marking them as outliers and deleting their + state groups). + """ + + return self.runInteraction( + "purge_history", + self._purge_history_txn, + room_id, + token, + delete_local_events, + ) + + def _purge_history_txn(self, txn, room_id, token_str, delete_local_events): + token = RoomStreamToken.parse(token_str) + + # Tables that should be pruned: + # event_auth + # event_backward_extremities + # event_edges + # event_forward_extremities + # event_json + # event_push_actions + # event_reference_hashes + # event_search + # event_to_state_groups + # events + # rejections + # room_depth + # state_groups + # state_groups_state + + # we will build a temporary table listing the events so that we don't + # have to keep shovelling the list back and forth across the + # connection. Annoyingly the python sqlite driver commits the + # transaction on CREATE, so let's do this first. + # + # furthermore, we might already have the table from a previous (failed) + # purge attempt, so let's drop the table first. + + txn.execute("DROP TABLE IF EXISTS events_to_purge") + + txn.execute( + "CREATE TEMPORARY TABLE events_to_purge (" + " event_id TEXT NOT NULL," + " should_delete BOOLEAN NOT NULL" + ")" + ) + + # First ensure that we're not about to delete all the forward extremeties + txn.execute( + "SELECT e.event_id, e.depth FROM events as e " + "INNER JOIN event_forward_extremities as f " + "ON e.event_id = f.event_id " + "AND e.room_id = f.room_id " + "WHERE f.room_id = ?", + (room_id,), + ) + rows = txn.fetchall() + max_depth = max(row[1] for row in rows) + + if max_depth < token.topological: + # We need to ensure we don't delete all the events from the database + # otherwise we wouldn't be able to send any events (due to not + # having any backwards extremeties) + raise SynapseError( + 400, "topological_ordering is greater than forward extremeties" + ) + + logger.info("[purge] looking for events to delete") + + should_delete_expr = "state_key IS NULL" + should_delete_params = () + if not delete_local_events: + should_delete_expr += " AND event_id NOT LIKE ?" + + # We include the parameter twice since we use the expression twice + should_delete_params += ("%:" + self.hs.hostname, "%:" + self.hs.hostname) + + should_delete_params += (room_id, token.topological) + + # Note that we insert events that are outliers and aren't going to be + # deleted, as nothing will happen to them. + txn.execute( + "INSERT INTO events_to_purge" + " SELECT event_id, %s" + " FROM events AS e LEFT JOIN state_events USING (event_id)" + " WHERE (NOT outlier OR (%s)) AND e.room_id = ? AND topological_ordering < ?" + % (should_delete_expr, should_delete_expr), + should_delete_params, + ) + + # We create the indices *after* insertion as that's a lot faster. + + # create an index on should_delete because later we'll be looking for + # the should_delete / shouldn't_delete subsets + txn.execute( + "CREATE INDEX events_to_purge_should_delete" + " ON events_to_purge(should_delete)" + ) + + # We do joins against events_to_purge for e.g. calculating state + # groups to purge, etc., so lets make an index. + txn.execute("CREATE INDEX events_to_purge_id" " ON events_to_purge(event_id)") + + txn.execute("SELECT event_id, should_delete FROM events_to_purge") + event_rows = txn.fetchall() + logger.info( + "[purge] found %i events before cutoff, of which %i can be deleted", + len(event_rows), + sum(1 for e in event_rows if e[1]), + ) + + logger.info("[purge] Finding new backward extremities") + + # We calculate the new entries for the backward extremeties by finding + # events to be purged that are pointed to by events we're not going to + # purge. + txn.execute( + "SELECT DISTINCT e.event_id FROM events_to_purge AS e" + " INNER JOIN event_edges AS ed ON e.event_id = ed.prev_event_id" + " LEFT JOIN events_to_purge AS ep2 ON ed.event_id = ep2.event_id" + " WHERE ep2.event_id IS NULL" + ) + new_backwards_extrems = txn.fetchall() + + logger.info("[purge] replacing backward extremities: %r", new_backwards_extrems) + + txn.execute( + "DELETE FROM event_backward_extremities WHERE room_id = ?", (room_id,) + ) + + # Update backward extremeties + txn.executemany( + "INSERT INTO event_backward_extremities (room_id, event_id)" + " VALUES (?, ?)", + [(room_id, event_id) for event_id, in new_backwards_extrems], + ) + + logger.info("[purge] finding redundant state groups") + + # Get all state groups that are referenced by events that are to be + # deleted. We then go and check if they are referenced by other events + # or state groups, and if not we delete them. + txn.execute( + """ + SELECT DISTINCT state_group FROM events_to_purge + INNER JOIN event_to_state_groups USING (event_id) + """ + ) + + referenced_state_groups = set(sg for sg, in txn) + logger.info( + "[purge] found %i referenced state groups", len(referenced_state_groups) + ) + + logger.info("[purge] finding state groups that can be deleted") + + _ = self._find_unreferenced_groups_during_purge(txn, referenced_state_groups) + state_groups_to_delete, remaining_state_groups = _ + + logger.info( + "[purge] found %i state groups to delete", len(state_groups_to_delete) + ) + + logger.info( + "[purge] de-delta-ing %i remaining state groups", + len(remaining_state_groups), + ) + + # Now we turn the state groups that reference to-be-deleted state + # groups to non delta versions. + for sg in remaining_state_groups: + logger.info("[purge] de-delta-ing remaining state group %s", sg) + curr_state = self._get_state_groups_from_groups_txn(txn, [sg]) + curr_state = curr_state[sg] + + self._simple_delete_txn( + txn, table="state_groups_state", keyvalues={"state_group": sg} + ) + + self._simple_delete_txn( + txn, table="state_group_edges", keyvalues={"state_group": sg} + ) + + self._simple_insert_many_txn( + txn, + table="state_groups_state", + values=[ + { + "state_group": sg, + "room_id": room_id, + "type": key[0], + "state_key": key[1], + "event_id": state_id, + } + for key, state_id in iteritems(curr_state) + ], + ) + + logger.info("[purge] removing redundant state groups") + txn.executemany( + "DELETE FROM state_groups_state WHERE state_group = ?", + ((sg,) for sg in state_groups_to_delete), + ) + txn.executemany( + "DELETE FROM state_groups WHERE id = ?", + ((sg,) for sg in state_groups_to_delete), + ) + + logger.info("[purge] removing events from event_to_state_groups") + txn.execute( + "DELETE FROM event_to_state_groups " + "WHERE event_id IN (SELECT event_id from events_to_purge)" + ) + for event_id, _ in event_rows: + txn.call_after(self._get_state_group_for_event.invalidate, (event_id,)) + + # Delete all remote non-state events + for table in ( + "events", + "event_json", + "event_auth", + "event_edges", + "event_forward_extremities", + "event_reference_hashes", + "event_search", + "rejections", + ): + logger.info("[purge] removing events from %s", table) + + txn.execute( + "DELETE FROM %s WHERE event_id IN (" + " SELECT event_id FROM events_to_purge WHERE should_delete" + ")" % (table,) + ) + + # event_push_actions lacks an index on event_id, and has one on + # (room_id, event_id) instead. + for table in ("event_push_actions",): + logger.info("[purge] removing events from %s", table) + + txn.execute( + "DELETE FROM %s WHERE room_id = ? AND event_id IN (" + " SELECT event_id FROM events_to_purge WHERE should_delete" + ")" % (table,), + (room_id,), + ) + + # Mark all state and own events as outliers + logger.info("[purge] marking remaining events as outliers") + txn.execute( + "UPDATE events SET outlier = ?" + " WHERE event_id IN (" + " SELECT event_id FROM events_to_purge " + " WHERE NOT should_delete" + ")", + (True,), + ) + + # synapse tries to take out an exclusive lock on room_depth whenever it + # persists events (because upsert), and once we run this update, we + # will block that for the rest of our transaction. + # + # So, let's stick it at the end so that we don't block event + # persistence. + # + # We do this by calculating the minimum depth of the backwards + # extremities. However, the events in event_backward_extremities + # are ones we don't have yet so we need to look at the events that + # point to it via event_edges table. + txn.execute( + """ + SELECT COALESCE(MIN(depth), 0) + FROM event_backward_extremities AS eb + INNER JOIN event_edges AS eg ON eg.prev_event_id = eb.event_id + INNER JOIN events AS e ON e.event_id = eg.event_id + WHERE eb.room_id = ? + """, + (room_id,), + ) + min_depth, = txn.fetchone() + + logger.info("[purge] updating room_depth to %d", min_depth) + + txn.execute( + "UPDATE room_depth SET min_depth = ? WHERE room_id = ?", + (min_depth, room_id), + ) + + # finally, drop the temp table. this will commit the txn in sqlite, + # so make sure to keep this actually last. + txn.execute("DROP TABLE events_to_purge") + + logger.info("[purge] done") + + def _find_unreferenced_groups_during_purge(self, txn, state_groups): + """Used when purging history to figure out which state groups can be + deleted and which need to be de-delta'ed (due to one of its prev groups + being scheduled for deletion). + + Args: + txn + state_groups (set[int]): Set of state groups referenced by events + that are going to be deleted. + + Returns: + tuple[set[int], set[int]]: The set of state groups that can be + deleted and the set of state groups that need to be de-delta'ed + """ + # Graph of state group -> previous group + graph = {} + + # Set of events that we have found to be referenced by events + referenced_groups = set() + + # Set of state groups we've already seen + state_groups_seen = set(state_groups) + + # Set of state groups to handle next. + next_to_search = set(state_groups) + while next_to_search: + # We bound size of groups we're looking up at once, to stop the + # SQL query getting too big + if len(next_to_search) < 100: + current_search = next_to_search + next_to_search = set() + else: + current_search = set(itertools.islice(next_to_search, 100)) + next_to_search -= current_search + + # Check if state groups are referenced + sql = """ + SELECT DISTINCT state_group FROM event_to_state_groups + LEFT JOIN events_to_purge AS ep USING (event_id) + WHERE ep.event_id IS NULL AND + """ + clause, args = make_in_list_sql_clause( + txn.database_engine, "state_group", current_search + ) + txn.execute(sql + clause, list(args)) + + referenced = set(sg for sg, in txn) + referenced_groups |= referenced + + # We don't continue iterating up the state group graphs for state + # groups that are referenced. + current_search -= referenced + + rows = self._simple_select_many_txn( + txn, + table="state_group_edges", + column="prev_state_group", + iterable=current_search, + keyvalues={}, + retcols=("prev_state_group", "state_group"), + ) + + prevs = set(row["state_group"] for row in rows) + # We don't bother re-handling groups we've already seen + prevs -= state_groups_seen + next_to_search |= prevs + state_groups_seen |= prevs + + for row in rows: + # Note: Each state group can have at most one prev group + graph[row["state_group"]] = row["prev_state_group"] + + to_delete = state_groups_seen - referenced_groups + + to_dedelta = set() + for sg in referenced_groups: + prev_sg = graph.get(sg) + if prev_sg and prev_sg in to_delete: + to_dedelta.add(sg) + + return to_delete, to_dedelta + + def purge_room(self, room_id): + """Deletes all record of a room + + Args: + room_id (str): + """ + + return self.runInteraction("purge_room", self._purge_room_txn, room_id) + + def _purge_room_txn(self, txn, room_id): + # first we have to delete the state groups states + logger.info("[purge] removing %s from state_groups_state", room_id) + + txn.execute( + """ + DELETE FROM state_groups_state WHERE state_group IN ( + SELECT state_group FROM events JOIN event_to_state_groups USING(event_id) + WHERE events.room_id=? + ) + """, + (room_id,), + ) + + # ... and the state group edges + logger.info("[purge] removing %s from state_group_edges", room_id) + + txn.execute( + """ + DELETE FROM state_group_edges WHERE state_group IN ( + SELECT state_group FROM events JOIN event_to_state_groups USING(event_id) + WHERE events.room_id=? + ) + """, + (room_id,), + ) + + # ... and the state groups + logger.info("[purge] removing %s from state_groups", room_id) + + txn.execute( + """ + DELETE FROM state_groups WHERE id IN ( + SELECT state_group FROM events JOIN event_to_state_groups USING(event_id) + WHERE events.room_id=? + ) + """, + (room_id,), + ) + + # and then tables which lack an index on room_id but have one on event_id + for table in ( + "event_auth", + "event_edges", + "event_push_actions_staging", + "event_reference_hashes", + "event_relations", + "event_to_state_groups", + "redactions", + "rejections", + "state_events", + ): + logger.info("[purge] removing %s from %s", room_id, table) + + txn.execute( + """ + DELETE FROM %s WHERE event_id IN ( + SELECT event_id FROM events WHERE room_id=? + ) + """ + % (table,), + (room_id,), + ) + + # and finally, the tables with an index on room_id (or no useful index) + for table in ( + "current_state_events", + "event_backward_extremities", + "event_forward_extremities", + "event_json", + "event_push_actions", + "event_search", + "events", + "group_rooms", + "public_room_list_stream", + "receipts_graph", + "receipts_linearized", + "room_aliases", + "room_depth", + "room_memberships", + "room_stats_state", + "room_stats_current", + "room_stats_historical", + "room_stats_earliest_token", + "rooms", + "stream_ordering_to_exterm", + "topics", + "users_in_public_rooms", + "users_who_share_private_rooms", + # no useful index, but let's clear them anyway + "appservice_room_list", + "e2e_room_keys", + "event_push_summary", + "pusher_throttle", + "group_summary_rooms", + "local_invites", + "room_account_data", + "room_tags", + ): + logger.info("[purge] removing %s from %s", room_id, table) + txn.execute("DELETE FROM %s WHERE room_id=?" % (table,), (room_id,)) + + # Other tables we do NOT need to clear out: + # + # - blocked_rooms + # This is important, to make sure that we don't accidentally rejoin a blocked + # room after it was purged + # + # - user_directory + # This has a room_id column, but it is unused + # + + # Other tables that we might want to consider clearing out include: + # + # - event_reports + # Given that these are intended for abuse management my initial + # inclination is to leave them in place. + # + # - current_state_delta_stream + # - ex_outlier_stream + # - room_tags_revisions + # The problem with these is that they are largeish and there is no room_id + # index on them. In any case we should be clearing out 'stream' tables + # periodically anyway (#5888) + + # TODO: we could probably usefully do a bunch of cache invalidation here + + logger.info("[purge] done") + + @defer.inlineCallbacks + def is_event_after(self, event_id1, event_id2): + """Returns True if event_id1 is after event_id2 in the stream + """ + to_1, so_1 = yield self._get_event_ordering(event_id1) + to_2, so_2 = yield self._get_event_ordering(event_id2) + return (to_1, so_1) > (to_2, so_2) + + @cachedInlineCallbacks(max_entries=5000) + def _get_event_ordering(self, event_id): + res = yield self._simple_select_one( + table="events", + retcols=["topological_ordering", "stream_ordering"], + keyvalues={"event_id": event_id}, + allow_none=True, + ) + + if not res: + raise SynapseError(404, "Could not find event %s" % (event_id,)) + + return (int(res["topological_ordering"]), int(res["stream_ordering"])) + + def get_all_updated_current_state_deltas(self, from_token, to_token, limit): + def get_all_updated_current_state_deltas_txn(txn): + sql = """ + SELECT stream_id, room_id, type, state_key, event_id + FROM current_state_delta_stream + WHERE ? < stream_id AND stream_id <= ? + ORDER BY stream_id ASC LIMIT ? + """ + txn.execute(sql, (from_token, to_token, limit)) + return txn.fetchall() + + return self.runInteraction( + "get_all_updated_current_state_deltas", + get_all_updated_current_state_deltas_txn, + ) + + +AllNewEventsResult = namedtuple( + "AllNewEventsResult", + [ + "new_forward_events", + "new_backfill_events", + "forward_ex_outliers", + "backward_ex_outliers", + ], +) diff --git a/synapse/storage/data_stores/main/events_bg_updates.py b/synapse/storage/data_stores/main/events_bg_updates.py new file mode 100644 index 0000000000..31ea6f917f --- /dev/null +++ b/synapse/storage/data_stores/main/events_bg_updates.py @@ -0,0 +1,505 @@ +# -*- coding: utf-8 -*- +# Copyright 2019 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging + +from six import text_type + +from canonicaljson import json + +from twisted.internet import defer + +from synapse.storage._base import make_in_list_sql_clause +from synapse.storage.background_updates import BackgroundUpdateStore + +logger = logging.getLogger(__name__) + + +class EventsBackgroundUpdatesStore(BackgroundUpdateStore): + + EVENT_ORIGIN_SERVER_TS_NAME = "event_origin_server_ts" + EVENT_FIELDS_SENDER_URL_UPDATE_NAME = "event_fields_sender_url" + DELETE_SOFT_FAILED_EXTREMITIES = "delete_soft_failed_extremities" + + def __init__(self, db_conn, hs): + super(EventsBackgroundUpdatesStore, self).__init__(db_conn, hs) + + self.register_background_update_handler( + self.EVENT_ORIGIN_SERVER_TS_NAME, self._background_reindex_origin_server_ts + ) + self.register_background_update_handler( + self.EVENT_FIELDS_SENDER_URL_UPDATE_NAME, + self._background_reindex_fields_sender, + ) + + self.register_background_index_update( + "event_contains_url_index", + index_name="event_contains_url_index", + table="events", + columns=["room_id", "topological_ordering", "stream_ordering"], + where_clause="contains_url = true AND outlier = false", + ) + + # an event_id index on event_search is useful for the purge_history + # api. Plus it means we get to enforce some integrity with a UNIQUE + # clause + self.register_background_index_update( + "event_search_event_id_idx", + index_name="event_search_event_id_idx", + table="event_search", + columns=["event_id"], + unique=True, + psql_only=True, + ) + + self.register_background_update_handler( + self.DELETE_SOFT_FAILED_EXTREMITIES, self._cleanup_extremities_bg_update + ) + + self.register_background_update_handler( + "redactions_received_ts", self._redactions_received_ts + ) + + # This index gets deleted in `event_fix_redactions_bytes` update + self.register_background_index_update( + "event_fix_redactions_bytes_create_index", + index_name="redactions_censored_redacts", + table="redactions", + columns=["redacts"], + where_clause="have_censored", + ) + + self.register_background_update_handler( + "event_fix_redactions_bytes", self._event_fix_redactions_bytes + ) + + @defer.inlineCallbacks + def _background_reindex_fields_sender(self, progress, batch_size): + target_min_stream_id = progress["target_min_stream_id_inclusive"] + max_stream_id = progress["max_stream_id_exclusive"] + rows_inserted = progress.get("rows_inserted", 0) + + INSERT_CLUMP_SIZE = 1000 + + def reindex_txn(txn): + sql = ( + "SELECT stream_ordering, event_id, json FROM events" + " INNER JOIN event_json USING (event_id)" + " WHERE ? <= stream_ordering AND stream_ordering < ?" + " ORDER BY stream_ordering DESC" + " LIMIT ?" + ) + + txn.execute(sql, (target_min_stream_id, max_stream_id, batch_size)) + + rows = txn.fetchall() + if not rows: + return 0 + + min_stream_id = rows[-1][0] + + update_rows = [] + for row in rows: + try: + event_id = row[1] + event_json = json.loads(row[2]) + sender = event_json["sender"] + content = event_json["content"] + + contains_url = "url" in content + if contains_url: + contains_url &= isinstance(content["url"], text_type) + except (KeyError, AttributeError): + # If the event is missing a necessary field then + # skip over it. + continue + + update_rows.append((sender, contains_url, event_id)) + + sql = "UPDATE events SET sender = ?, contains_url = ? WHERE event_id = ?" + + for index in range(0, len(update_rows), INSERT_CLUMP_SIZE): + clump = update_rows[index : index + INSERT_CLUMP_SIZE] + txn.executemany(sql, clump) + + progress = { + "target_min_stream_id_inclusive": target_min_stream_id, + "max_stream_id_exclusive": min_stream_id, + "rows_inserted": rows_inserted + len(rows), + } + + self._background_update_progress_txn( + txn, self.EVENT_FIELDS_SENDER_URL_UPDATE_NAME, progress + ) + + return len(rows) + + result = yield self.runInteraction( + self.EVENT_FIELDS_SENDER_URL_UPDATE_NAME, reindex_txn + ) + + if not result: + yield self._end_background_update(self.EVENT_FIELDS_SENDER_URL_UPDATE_NAME) + + return result + + @defer.inlineCallbacks + def _background_reindex_origin_server_ts(self, progress, batch_size): + target_min_stream_id = progress["target_min_stream_id_inclusive"] + max_stream_id = progress["max_stream_id_exclusive"] + rows_inserted = progress.get("rows_inserted", 0) + + INSERT_CLUMP_SIZE = 1000 + + def reindex_search_txn(txn): + sql = ( + "SELECT stream_ordering, event_id FROM events" + " WHERE ? <= stream_ordering AND stream_ordering < ?" + " ORDER BY stream_ordering DESC" + " LIMIT ?" + ) + + txn.execute(sql, (target_min_stream_id, max_stream_id, batch_size)) + + rows = txn.fetchall() + if not rows: + return 0 + + min_stream_id = rows[-1][0] + event_ids = [row[1] for row in rows] + + rows_to_update = [] + + chunks = [event_ids[i : i + 100] for i in range(0, len(event_ids), 100)] + for chunk in chunks: + ev_rows = self._simple_select_many_txn( + txn, + table="event_json", + column="event_id", + iterable=chunk, + retcols=["event_id", "json"], + keyvalues={}, + ) + + for row in ev_rows: + event_id = row["event_id"] + event_json = json.loads(row["json"]) + try: + origin_server_ts = event_json["origin_server_ts"] + except (KeyError, AttributeError): + # If the event is missing a necessary field then + # skip over it. + continue + + rows_to_update.append((origin_server_ts, event_id)) + + sql = "UPDATE events SET origin_server_ts = ? WHERE event_id = ?" + + for index in range(0, len(rows_to_update), INSERT_CLUMP_SIZE): + clump = rows_to_update[index : index + INSERT_CLUMP_SIZE] + txn.executemany(sql, clump) + + progress = { + "target_min_stream_id_inclusive": target_min_stream_id, + "max_stream_id_exclusive": min_stream_id, + "rows_inserted": rows_inserted + len(rows_to_update), + } + + self._background_update_progress_txn( + txn, self.EVENT_ORIGIN_SERVER_TS_NAME, progress + ) + + return len(rows_to_update) + + result = yield self.runInteraction( + self.EVENT_ORIGIN_SERVER_TS_NAME, reindex_search_txn + ) + + if not result: + yield self._end_background_update(self.EVENT_ORIGIN_SERVER_TS_NAME) + + return result + + @defer.inlineCallbacks + def _cleanup_extremities_bg_update(self, progress, batch_size): + """Background update to clean out extremities that should have been + deleted previously. + + Mainly used to deal with the aftermath of #5269. + """ + + # This works by first copying all existing forward extremities into the + # `_extremities_to_check` table at start up, and then checking each + # event in that table whether we have any descendants that are not + # soft-failed/rejected. If that is the case then we delete that event + # from the forward extremities table. + # + # For efficiency, we do this in batches by recursively pulling out all + # descendants of a batch until we find the non soft-failed/rejected + # events, i.e. the set of descendants whose chain of prev events back + # to the batch of extremities are all soft-failed or rejected. + # Typically, we won't find any such events as extremities will rarely + # have any descendants, but if they do then we should delete those + # extremities. + + def _cleanup_extremities_bg_update_txn(txn): + # The set of extremity event IDs that we're checking this round + original_set = set() + + # A dict[str, set[str]] of event ID to their prev events. + graph = {} + + # The set of descendants of the original set that are not rejected + # nor soft-failed. Ancestors of these events should be removed + # from the forward extremities table. + non_rejected_leaves = set() + + # Set of event IDs that have been soft failed, and for which we + # should check if they have descendants which haven't been soft + # failed. + soft_failed_events_to_lookup = set() + + # First, we get `batch_size` events from the table, pulling out + # their successor events, if any, and the successor events' + # rejection status. + txn.execute( + """SELECT prev_event_id, event_id, internal_metadata, + rejections.event_id IS NOT NULL, events.outlier + FROM ( + SELECT event_id AS prev_event_id + FROM _extremities_to_check + LIMIT ? + ) AS f + LEFT JOIN event_edges USING (prev_event_id) + LEFT JOIN events USING (event_id) + LEFT JOIN event_json USING (event_id) + LEFT JOIN rejections USING (event_id) + """, + (batch_size,), + ) + + for prev_event_id, event_id, metadata, rejected, outlier in txn: + original_set.add(prev_event_id) + + if not event_id or outlier: + # Common case where the forward extremity doesn't have any + # descendants. + continue + + graph.setdefault(event_id, set()).add(prev_event_id) + + soft_failed = False + if metadata: + soft_failed = json.loads(metadata).get("soft_failed") + + if soft_failed or rejected: + soft_failed_events_to_lookup.add(event_id) + else: + non_rejected_leaves.add(event_id) + + # Now we recursively check all the soft-failed descendants we + # found above in the same way, until we have nothing left to + # check. + while soft_failed_events_to_lookup: + # We only want to do 100 at a time, so we split given list + # into two. + batch = list(soft_failed_events_to_lookup) + to_check, to_defer = batch[:100], batch[100:] + soft_failed_events_to_lookup = set(to_defer) + + sql = """SELECT prev_event_id, event_id, internal_metadata, + rejections.event_id IS NOT NULL + FROM event_edges + INNER JOIN events USING (event_id) + INNER JOIN event_json USING (event_id) + LEFT JOIN rejections USING (event_id) + WHERE + NOT events.outlier + AND + """ + clause, args = make_in_list_sql_clause( + self.database_engine, "prev_event_id", to_check + ) + txn.execute(sql + clause, list(args)) + + for prev_event_id, event_id, metadata, rejected in txn: + if event_id in graph: + # Already handled this event previously, but we still + # want to record the edge. + graph[event_id].add(prev_event_id) + continue + + graph[event_id] = {prev_event_id} + + soft_failed = json.loads(metadata).get("soft_failed") + if soft_failed or rejected: + soft_failed_events_to_lookup.add(event_id) + else: + non_rejected_leaves.add(event_id) + + # We have a set of non-soft-failed descendants, so we recurse up + # the graph to find all ancestors and add them to the set of event + # IDs that we can delete from forward extremities table. + to_delete = set() + while non_rejected_leaves: + event_id = non_rejected_leaves.pop() + prev_event_ids = graph.get(event_id, set()) + non_rejected_leaves.update(prev_event_ids) + to_delete.update(prev_event_ids) + + to_delete.intersection_update(original_set) + + deleted = self._simple_delete_many_txn( + txn=txn, + table="event_forward_extremities", + column="event_id", + iterable=to_delete, + keyvalues={}, + ) + + logger.info( + "Deleted %d forward extremities of %d checked, to clean up #5269", + deleted, + len(original_set), + ) + + if deleted: + # We now need to invalidate the caches of these rooms + rows = self._simple_select_many_txn( + txn, + table="events", + column="event_id", + iterable=to_delete, + keyvalues={}, + retcols=("room_id",), + ) + room_ids = set(row["room_id"] for row in rows) + for room_id in room_ids: + txn.call_after( + self.get_latest_event_ids_in_room.invalidate, (room_id,) + ) + + self._simple_delete_many_txn( + txn=txn, + table="_extremities_to_check", + column="event_id", + iterable=original_set, + keyvalues={}, + ) + + return len(original_set) + + num_handled = yield self.runInteraction( + "_cleanup_extremities_bg_update", _cleanup_extremities_bg_update_txn + ) + + if not num_handled: + yield self._end_background_update(self.DELETE_SOFT_FAILED_EXTREMITIES) + + def _drop_table_txn(txn): + txn.execute("DROP TABLE _extremities_to_check") + + yield self.runInteraction( + "_cleanup_extremities_bg_update_drop_table", _drop_table_txn + ) + + return num_handled + + @defer.inlineCallbacks + def _redactions_received_ts(self, progress, batch_size): + """Handles filling out the `received_ts` column in redactions. + """ + last_event_id = progress.get("last_event_id", "") + + def _redactions_received_ts_txn(txn): + # Fetch the set of event IDs that we want to update + sql = """ + SELECT event_id FROM redactions + WHERE event_id > ? + ORDER BY event_id ASC + LIMIT ? + """ + + txn.execute(sql, (last_event_id, batch_size)) + + rows = txn.fetchall() + if not rows: + return 0 + + upper_event_id, = rows[-1] + + # Update the redactions with the received_ts. + # + # Note: Not all events have an associated received_ts, so we + # fallback to using origin_server_ts. If we for some reason don't + # have an origin_server_ts, lets just use the current timestamp. + # + # We don't want to leave it null, as then we'll never try and + # censor those redactions. + sql = """ + UPDATE redactions + SET received_ts = ( + SELECT COALESCE(received_ts, origin_server_ts, ?) FROM events + WHERE events.event_id = redactions.event_id + ) + WHERE ? <= event_id AND event_id <= ? + """ + + txn.execute(sql, (self._clock.time_msec(), last_event_id, upper_event_id)) + + self._background_update_progress_txn( + txn, "redactions_received_ts", {"last_event_id": upper_event_id} + ) + + return len(rows) + + count = yield self.runInteraction( + "_redactions_received_ts", _redactions_received_ts_txn + ) + + if not count: + yield self._end_background_update("redactions_received_ts") + + return count + + @defer.inlineCallbacks + def _event_fix_redactions_bytes(self, progress, batch_size): + """Undoes hex encoded censored redacted event JSON. + """ + + def _event_fix_redactions_bytes_txn(txn): + # This update is quite fast due to new index. + txn.execute( + """ + UPDATE event_json + SET + json = convert_from(json::bytea, 'utf8') + FROM redactions + WHERE + redactions.have_censored + AND event_json.event_id = redactions.redacts + AND json NOT LIKE '{%'; + """ + ) + + txn.execute("DROP INDEX redactions_censored_redacts") + + yield self.runInteraction( + "_event_fix_redactions_bytes", _event_fix_redactions_bytes_txn + ) + + yield self._end_background_update("event_fix_redactions_bytes") + + return 1 diff --git a/synapse/storage/data_stores/main/events_worker.py b/synapse/storage/data_stores/main/events_worker.py new file mode 100644 index 0000000000..4c4b76bd93 --- /dev/null +++ b/synapse/storage/data_stores/main/events_worker.py @@ -0,0 +1,882 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import division + +import itertools +import logging +from collections import namedtuple + +from canonicaljson import json + +from twisted.internet import defer + +from synapse.api.constants import EventTypes +from synapse.api.errors import NotFoundError +from synapse.api.room_versions import EventFormatVersions +from synapse.events import FrozenEvent, event_type_from_format_version # noqa: F401 +from synapse.events.snapshot import EventContext # noqa: F401 +from synapse.events.utils import prune_event +from synapse.logging.context import LoggingContext, PreserveLoggingContext +from synapse.metrics.background_process_metrics import run_as_background_process +from synapse.storage._base import SQLBaseStore, make_in_list_sql_clause +from synapse.types import get_domain_from_id +from synapse.util import batch_iter +from synapse.util.metrics import Measure + +logger = logging.getLogger(__name__) + + +# These values are used in the `enqueus_event` and `_do_fetch` methods to +# control how we batch/bulk fetch events from the database. +# The values are plucked out of thing air to make initial sync run faster +# on jki.re +# TODO: Make these configurable. +EVENT_QUEUE_THREADS = 3 # Max number of threads that will fetch events +EVENT_QUEUE_ITERATIONS = 3 # No. times we block waiting for requests for events +EVENT_QUEUE_TIMEOUT_S = 0.1 # Timeout when waiting for requests for events + + +_EventCacheEntry = namedtuple("_EventCacheEntry", ("event", "redacted_event")) + + +class EventsWorkerStore(SQLBaseStore): + def get_received_ts(self, event_id): + """Get received_ts (when it was persisted) for the event. + + Raises an exception for unknown events. + + Args: + event_id (str) + + Returns: + Deferred[int|None]: Timestamp in milliseconds, or None for events + that were persisted before received_ts was implemented. + """ + return self._simple_select_one_onecol( + table="events", + keyvalues={"event_id": event_id}, + retcol="received_ts", + desc="get_received_ts", + ) + + def get_received_ts_by_stream_pos(self, stream_ordering): + """Given a stream ordering get an approximate timestamp of when it + happened. + + This is done by simply taking the received ts of the first event that + has a stream ordering greater than or equal to the given stream pos. + If none exists returns the current time, on the assumption that it must + have happened recently. + + Args: + stream_ordering (int) + + Returns: + Deferred[int] + """ + + def _get_approximate_received_ts_txn(txn): + sql = """ + SELECT received_ts FROM events + WHERE stream_ordering >= ? + LIMIT 1 + """ + + txn.execute(sql, (stream_ordering,)) + row = txn.fetchone() + if row and row[0]: + ts = row[0] + else: + ts = self.clock.time_msec() + + return ts + + return self.runInteraction( + "get_approximate_received_ts", _get_approximate_received_ts_txn + ) + + @defer.inlineCallbacks + def get_event( + self, + event_id, + check_redacted=True, + get_prev_content=False, + allow_rejected=False, + allow_none=False, + check_room_id=None, + ): + """Get an event from the database by event_id. + + Args: + event_id (str): The event_id of the event to fetch + check_redacted (bool): If True, check if event has been redacted + and redact it. + get_prev_content (bool): If True and event is a state event, + include the previous states content in the unsigned field. + allow_rejected (bool): If True return rejected events. + allow_none (bool): If True, return None if no event found, if + False throw a NotFoundError + check_room_id (str|None): if not None, check the room of the found event. + If there is a mismatch, behave as per allow_none. + + Returns: + Deferred[EventBase|None] + """ + if not isinstance(event_id, str): + raise TypeError("Invalid event event_id %r" % (event_id,)) + + events = yield self.get_events_as_list( + [event_id], + check_redacted=check_redacted, + get_prev_content=get_prev_content, + allow_rejected=allow_rejected, + ) + + event = events[0] if events else None + + if event is not None and check_room_id is not None: + if event.room_id != check_room_id: + event = None + + if event is None and not allow_none: + raise NotFoundError("Could not find event %s" % (event_id,)) + + return event + + @defer.inlineCallbacks + def get_events( + self, + event_ids, + check_redacted=True, + get_prev_content=False, + allow_rejected=False, + ): + """Get events from the database + + Args: + event_ids (list): The event_ids of the events to fetch + check_redacted (bool): If True, check if event has been redacted + and redact it. + get_prev_content (bool): If True and event is a state event, + include the previous states content in the unsigned field. + allow_rejected (bool): If True return rejected events. + + Returns: + Deferred : Dict from event_id to event. + """ + events = yield self.get_events_as_list( + event_ids, + check_redacted=check_redacted, + get_prev_content=get_prev_content, + allow_rejected=allow_rejected, + ) + + return {e.event_id: e for e in events} + + @defer.inlineCallbacks + def get_events_as_list( + self, + event_ids, + check_redacted=True, + get_prev_content=False, + allow_rejected=False, + ): + """Get events from the database and return in a list in the same order + as given by `event_ids` arg. + + Args: + event_ids (list): The event_ids of the events to fetch + check_redacted (bool): If True, check if event has been redacted + and redact it. + get_prev_content (bool): If True and event is a state event, + include the previous states content in the unsigned field. + allow_rejected (bool): If True return rejected events. + + Returns: + Deferred[list[EventBase]]: List of events fetched from the database. The + events are in the same order as `event_ids` arg. + + Note that the returned list may be smaller than the list of event + IDs if not all events could be fetched. + """ + + if not event_ids: + return [] + + # there may be duplicates so we cast the list to a set + event_entry_map = yield self._get_events_from_cache_or_db( + set(event_ids), allow_rejected=allow_rejected + ) + + events = [] + for event_id in event_ids: + entry = event_entry_map.get(event_id, None) + if not entry: + continue + + if not allow_rejected: + assert not entry.event.rejected_reason, ( + "rejected event returned from _get_events_from_cache_or_db despite " + "allow_rejected=False" + ) + + # We may not have had the original event when we received a redaction, so + # we have to recheck auth now. + + if not allow_rejected and entry.event.type == EventTypes.Redaction: + if not hasattr(entry.event, "redacts"): + # A redacted redaction doesn't have a `redacts` key, in + # which case lets just withhold the event. + # + # Note: Most of the time if the redactions has been + # redacted we still have the un-redacted event in the DB + # and so we'll still see the `redacts` key. However, this + # isn't always true e.g. if we have censored the event. + logger.debug( + "Withholding redaction event %s as we don't have redacts key", + event_id, + ) + continue + + redacted_event_id = entry.event.redacts + event_map = yield self._get_events_from_cache_or_db([redacted_event_id]) + original_event_entry = event_map.get(redacted_event_id) + if not original_event_entry: + # we don't have the redacted event (or it was rejected). + # + # We assume that the redaction isn't authorized for now; if the + # redacted event later turns up, the redaction will be re-checked, + # and if it is found valid, the original will get redacted before it + # is served to the client. + logger.debug( + "Withholding redaction event %s since we don't (yet) have the " + "original %s", + event_id, + redacted_event_id, + ) + continue + + original_event = original_event_entry.event + if original_event.type == EventTypes.Create: + # we never serve redactions of Creates to clients. + logger.info( + "Withholding redaction %s of create event %s", + event_id, + redacted_event_id, + ) + continue + + if original_event.room_id != entry.event.room_id: + logger.info( + "Withholding redaction %s of event %s from a different room", + event_id, + redacted_event_id, + ) + continue + + if entry.event.internal_metadata.need_to_check_redaction(): + original_domain = get_domain_from_id(original_event.sender) + redaction_domain = get_domain_from_id(entry.event.sender) + if original_domain != redaction_domain: + # the senders don't match, so this is forbidden + logger.info( + "Withholding redaction %s whose sender domain %s doesn't " + "match that of redacted event %s %s", + event_id, + redaction_domain, + redacted_event_id, + original_domain, + ) + continue + + # Update the cache to save doing the checks again. + entry.event.internal_metadata.recheck_redaction = False + + if check_redacted and entry.redacted_event: + event = entry.redacted_event + else: + event = entry.event + + events.append(event) + + if get_prev_content: + if "replaces_state" in event.unsigned: + prev = yield self.get_event( + event.unsigned["replaces_state"], + get_prev_content=False, + allow_none=True, + ) + if prev: + event.unsigned = dict(event.unsigned) + event.unsigned["prev_content"] = prev.content + event.unsigned["prev_sender"] = prev.sender + + return events + + @defer.inlineCallbacks + def _get_events_from_cache_or_db(self, event_ids, allow_rejected=False): + """Fetch a bunch of events from the cache or the database. + + If events are pulled from the database, they will be cached for future lookups. + + Args: + event_ids (Iterable[str]): The event_ids of the events to fetch + allow_rejected (bool): Whether to include rejected events + + Returns: + Deferred[Dict[str, _EventCacheEntry]]: + map from event id to result + """ + event_entry_map = self._get_events_from_cache( + event_ids, allow_rejected=allow_rejected + ) + + missing_events_ids = [e for e in event_ids if e not in event_entry_map] + + if missing_events_ids: + log_ctx = LoggingContext.current_context() + log_ctx.record_event_fetch(len(missing_events_ids)) + + # Note that _get_events_from_db is also responsible for turning db rows + # into FrozenEvents (via _get_event_from_row), which involves seeing if + # the events have been redacted, and if so pulling the redaction event out + # of the database to check it. + # + missing_events = yield self._get_events_from_db( + missing_events_ids, allow_rejected=allow_rejected + ) + + event_entry_map.update(missing_events) + + return event_entry_map + + def _invalidate_get_event_cache(self, event_id): + self._get_event_cache.invalidate((event_id,)) + + def _get_events_from_cache(self, events, allow_rejected, update_metrics=True): + """Fetch events from the caches + + Args: + events (Iterable[str]): list of event_ids to fetch + allow_rejected (bool): Whether to return events that were rejected + update_metrics (bool): Whether to update the cache hit ratio metrics + + Returns: + dict of event_id -> _EventCacheEntry for each event_id in cache. If + allow_rejected is `False` then there will still be an entry but it + will be `None` + """ + event_map = {} + + for event_id in events: + ret = self._get_event_cache.get( + (event_id,), None, update_metrics=update_metrics + ) + if not ret: + continue + + if allow_rejected or not ret.event.rejected_reason: + event_map[event_id] = ret + else: + event_map[event_id] = None + + return event_map + + def _do_fetch(self, conn): + """Takes a database connection and waits for requests for events from + the _event_fetch_list queue. + """ + i = 0 + while True: + with self._event_fetch_lock: + event_list = self._event_fetch_list + self._event_fetch_list = [] + + if not event_list: + single_threaded = self.database_engine.single_threaded + if single_threaded or i > EVENT_QUEUE_ITERATIONS: + self._event_fetch_ongoing -= 1 + return + else: + self._event_fetch_lock.wait(EVENT_QUEUE_TIMEOUT_S) + i += 1 + continue + i = 0 + + self._fetch_event_list(conn, event_list) + + def _fetch_event_list(self, conn, event_list): + """Handle a load of requests from the _event_fetch_list queue + + Args: + conn (twisted.enterprise.adbapi.Connection): database connection + + event_list (list[Tuple[list[str], Deferred]]): + The fetch requests. Each entry consists of a list of event + ids to be fetched, and a deferred to be completed once the + events have been fetched. + + The deferreds are callbacked with a dictionary mapping from event id + to event row. Note that it may well contain additional events that + were not part of this request. + """ + with Measure(self._clock, "_fetch_event_list"): + try: + events_to_fetch = set( + event_id for events, _ in event_list for event_id in events + ) + + row_dict = self._new_transaction( + conn, "do_fetch", [], [], self._fetch_event_rows, events_to_fetch + ) + + # We only want to resolve deferreds from the main thread + def fire(): + for _, d in event_list: + d.callback(row_dict) + + with PreserveLoggingContext(): + self.hs.get_reactor().callFromThread(fire) + except Exception as e: + logger.exception("do_fetch") + + # We only want to resolve deferreds from the main thread + def fire(evs, exc): + for _, d in evs: + if not d.called: + with PreserveLoggingContext(): + d.errback(exc) + + with PreserveLoggingContext(): + self.hs.get_reactor().callFromThread(fire, event_list, e) + + @defer.inlineCallbacks + def _get_events_from_db(self, event_ids, allow_rejected=False): + """Fetch a bunch of events from the database. + + Returned events will be added to the cache for future lookups. + + Args: + event_ids (Iterable[str]): The event_ids of the events to fetch + allow_rejected (bool): Whether to include rejected events + + Returns: + Deferred[Dict[str, _EventCacheEntry]]: + map from event id to result. May return extra events which + weren't asked for. + """ + fetched_events = {} + events_to_fetch = event_ids + + while events_to_fetch: + row_map = yield self._enqueue_events(events_to_fetch) + + # we need to recursively fetch any redactions of those events + redaction_ids = set() + for event_id in events_to_fetch: + row = row_map.get(event_id) + fetched_events[event_id] = row + if row: + redaction_ids.update(row["redactions"]) + + events_to_fetch = redaction_ids.difference(fetched_events.keys()) + if events_to_fetch: + logger.debug("Also fetching redaction events %s", events_to_fetch) + + # build a map from event_id to EventBase + event_map = {} + for event_id, row in fetched_events.items(): + if not row: + continue + assert row["event_id"] == event_id + + rejected_reason = row["rejected_reason"] + + if not allow_rejected and rejected_reason: + continue + + d = json.loads(row["json"]) + internal_metadata = json.loads(row["internal_metadata"]) + + format_version = row["format_version"] + if format_version is None: + # This means that we stored the event before we had the concept + # of a event format version, so it must be a V1 event. + format_version = EventFormatVersions.V1 + + original_ev = event_type_from_format_version(format_version)( + event_dict=d, + internal_metadata_dict=internal_metadata, + rejected_reason=rejected_reason, + ) + + event_map[event_id] = original_ev + + # finally, we can decide whether each one nededs redacting, and build + # the cache entries. + result_map = {} + for event_id, original_ev in event_map.items(): + redactions = fetched_events[event_id]["redactions"] + redacted_event = self._maybe_redact_event_row( + original_ev, redactions, event_map + ) + + cache_entry = _EventCacheEntry( + event=original_ev, redacted_event=redacted_event + ) + + self._get_event_cache.prefill((event_id,), cache_entry) + result_map[event_id] = cache_entry + + return result_map + + @defer.inlineCallbacks + def _enqueue_events(self, events): + """Fetches events from the database using the _event_fetch_list. This + allows batch and bulk fetching of events - it allows us to fetch events + without having to create a new transaction for each request for events. + + Args: + events (Iterable[str]): events to be fetched. + + Returns: + Deferred[Dict[str, Dict]]: map from event id to row data from the database. + May contain events that weren't requested. + """ + + events_d = defer.Deferred() + with self._event_fetch_lock: + self._event_fetch_list.append((events, events_d)) + + self._event_fetch_lock.notify() + + if self._event_fetch_ongoing < EVENT_QUEUE_THREADS: + self._event_fetch_ongoing += 1 + should_start = True + else: + should_start = False + + if should_start: + run_as_background_process( + "fetch_events", self.runWithConnection, self._do_fetch + ) + + logger.debug("Loading %d events: %s", len(events), events) + with PreserveLoggingContext(): + row_map = yield events_d + logger.debug("Loaded %d events (%d rows)", len(events), len(row_map)) + + return row_map + + def _fetch_event_rows(self, txn, event_ids): + """Fetch event rows from the database + + Events which are not found are omitted from the result. + + The returned per-event dicts contain the following keys: + + * event_id (str) + + * json (str): json-encoded event structure + + * internal_metadata (str): json-encoded internal metadata dict + + * format_version (int|None): The format of the event. Hopefully one + of EventFormatVersions. 'None' means the event predates + EventFormatVersions (so the event is format V1). + + * rejected_reason (str|None): if the event was rejected, the reason + why. + + * redactions (List[str]): a list of event-ids which (claim to) redact + this event. + + Args: + txn (twisted.enterprise.adbapi.Connection): + event_ids (Iterable[str]): event IDs to fetch + + Returns: + Dict[str, Dict]: a map from event id to event info. + """ + event_dict = {} + for evs in batch_iter(event_ids, 200): + sql = ( + "SELECT " + " e.event_id, " + " e.internal_metadata," + " e.json," + " e.format_version, " + " rej.reason " + " FROM event_json as e" + " LEFT JOIN rejections as rej USING (event_id)" + " WHERE " + ) + + clause, args = make_in_list_sql_clause( + txn.database_engine, "e.event_id", evs + ) + + txn.execute(sql + clause, args) + + for row in txn: + event_id = row[0] + event_dict[event_id] = { + "event_id": event_id, + "internal_metadata": row[1], + "json": row[2], + "format_version": row[3], + "rejected_reason": row[4], + "redactions": [], + } + + # check for redactions + redactions_sql = "SELECT event_id, redacts FROM redactions WHERE " + + clause, args = make_in_list_sql_clause(txn.database_engine, "redacts", evs) + + txn.execute(redactions_sql + clause, args) + + for (redacter, redacted) in txn: + d = event_dict.get(redacted) + if d: + d["redactions"].append(redacter) + + return event_dict + + def _maybe_redact_event_row(self, original_ev, redactions, event_map): + """Given an event object and a list of possible redacting event ids, + determine whether to honour any of those redactions and if so return a redacted + event. + + Args: + original_ev (EventBase): + redactions (iterable[str]): list of event ids of potential redaction events + event_map (dict[str, EventBase]): other events which have been fetched, in + which we can look up the redaaction events. Map from event id to event. + + Returns: + Deferred[EventBase|None]: if the event should be redacted, a pruned + event object. Otherwise, None. + """ + if original_ev.type == "m.room.create": + # we choose to ignore redactions of m.room.create events. + return None + + for redaction_id in redactions: + redaction_event = event_map.get(redaction_id) + if not redaction_event or redaction_event.rejected_reason: + # we don't have the redaction event, or the redaction event was not + # authorized. + logger.debug( + "%s was redacted by %s but redaction not found/authed", + original_ev.event_id, + redaction_id, + ) + continue + + if redaction_event.room_id != original_ev.room_id: + logger.debug( + "%s was redacted by %s but redaction was in a different room!", + original_ev.event_id, + redaction_id, + ) + continue + + # Starting in room version v3, some redactions need to be + # rechecked if we didn't have the redacted event at the + # time, so we recheck on read instead. + if redaction_event.internal_metadata.need_to_check_redaction(): + expected_domain = get_domain_from_id(original_ev.sender) + if get_domain_from_id(redaction_event.sender) == expected_domain: + # This redaction event is allowed. Mark as not needing a recheck. + redaction_event.internal_metadata.recheck_redaction = False + else: + # Senders don't match, so the event isn't actually redacted + logger.debug( + "%s was redacted by %s but the senders don't match", + original_ev.event_id, + redaction_id, + ) + continue + + logger.debug("Redacting %s due to %s", original_ev.event_id, redaction_id) + + # we found a good redaction event. Redact! + redacted_event = prune_event(original_ev) + redacted_event.unsigned["redacted_by"] = redaction_id + + # It's fine to add the event directly, since get_pdu_json + # will serialise this field correctly + redacted_event.unsigned["redacted_because"] = redaction_event + + return redacted_event + + # no valid redaction found for this event + return None + + @defer.inlineCallbacks + def have_events_in_timeline(self, event_ids): + """Given a list of event ids, check if we have already processed and + stored them as non outliers. + """ + rows = yield self._simple_select_many_batch( + table="events", + retcols=("event_id",), + column="event_id", + iterable=list(event_ids), + keyvalues={"outlier": False}, + desc="have_events_in_timeline", + ) + + return set(r["event_id"] for r in rows) + + @defer.inlineCallbacks + def have_seen_events(self, event_ids): + """Given a list of event ids, check if we have already processed them. + + Args: + event_ids (iterable[str]): + + Returns: + Deferred[set[str]]: The events we have already seen. + """ + results = set() + + def have_seen_events_txn(txn, chunk): + sql = "SELECT event_id FROM events as e WHERE " + clause, args = make_in_list_sql_clause( + txn.database_engine, "e.event_id", chunk + ) + txn.execute(sql + clause, args) + for (event_id,) in txn: + results.add(event_id) + + # break the input up into chunks of 100 + input_iterator = iter(event_ids) + for chunk in iter(lambda: list(itertools.islice(input_iterator, 100)), []): + yield self.runInteraction("have_seen_events", have_seen_events_txn, chunk) + return results + + def get_seen_events_with_rejections(self, event_ids): + """Given a list of event ids, check if we rejected them. + + Args: + event_ids (list[str]) + + Returns: + Deferred[dict[str, str|None): + Has an entry for each event id we already have seen. Maps to + the rejected reason string if we rejected the event, else maps + to None. + """ + if not event_ids: + return defer.succeed({}) + + def f(txn): + sql = ( + "SELECT e.event_id, reason FROM events as e " + "LEFT JOIN rejections as r ON e.event_id = r.event_id " + "WHERE e.event_id = ?" + ) + + res = {} + for event_id in event_ids: + txn.execute(sql, (event_id,)) + row = txn.fetchone() + if row: + _, rejected = row + res[event_id] = rejected + + return res + + return self.runInteraction("get_seen_events_with_rejections", f) + + def _get_total_state_event_counts_txn(self, txn, room_id): + """ + See get_total_state_event_counts. + """ + # We join against the events table as that has an index on room_id + sql = """ + SELECT COUNT(*) FROM state_events + INNER JOIN events USING (room_id, event_id) + WHERE room_id=? + """ + txn.execute(sql, (room_id,)) + row = txn.fetchone() + return row[0] if row else 0 + + def get_total_state_event_counts(self, room_id): + """ + Gets the total number of state events in a room. + + Args: + room_id (str) + + Returns: + Deferred[int] + """ + return self.runInteraction( + "get_total_state_event_counts", + self._get_total_state_event_counts_txn, + room_id, + ) + + def _get_current_state_event_counts_txn(self, txn, room_id): + """ + See get_current_state_event_counts. + """ + sql = "SELECT COUNT(*) FROM current_state_events WHERE room_id=?" + txn.execute(sql, (room_id,)) + row = txn.fetchone() + return row[0] if row else 0 + + def get_current_state_event_counts(self, room_id): + """ + Gets the current number of state events in a room. + + Args: + room_id (str) + + Returns: + Deferred[int] + """ + return self.runInteraction( + "get_current_state_event_counts", + self._get_current_state_event_counts_txn, + room_id, + ) + + @defer.inlineCallbacks + def get_room_complexity(self, room_id): + """ + Get a rough approximation of the complexity of the room. This is used by + remote servers to decide whether they wish to join the room or not. + Higher complexity value indicates that being in the room will consume + more resources. + + Args: + room_id (str) + + Returns: + Deferred[dict[str:int]] of complexity version to complexity. + """ + state_events = yield self.get_current_state_event_counts(room_id) + + # Call this one "v1", so we can introduce new ones as we want to develop + # it. + complexity_v1 = round(state_events / 500, 2) + + return {"v1": complexity_v1} diff --git a/synapse/storage/data_stores/main/filtering.py b/synapse/storage/data_stores/main/filtering.py new file mode 100644 index 0000000000..a2a2a67927 --- /dev/null +++ b/synapse/storage/data_stores/main/filtering.py @@ -0,0 +1,74 @@ +# -*- coding: utf-8 -*- +# Copyright 2015, 2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from canonicaljson import encode_canonical_json + +from synapse.api.errors import Codes, SynapseError +from synapse.storage._base import SQLBaseStore, db_to_json +from synapse.util.caches.descriptors import cachedInlineCallbacks + + +class FilteringStore(SQLBaseStore): + @cachedInlineCallbacks(num_args=2) + def get_user_filter(self, user_localpart, filter_id): + # filter_id is BIGINT UNSIGNED, so if it isn't a number, fail + # with a coherent error message rather than 500 M_UNKNOWN. + try: + int(filter_id) + except ValueError: + raise SynapseError(400, "Invalid filter ID", Codes.INVALID_PARAM) + + def_json = yield self._simple_select_one_onecol( + table="user_filters", + keyvalues={"user_id": user_localpart, "filter_id": filter_id}, + retcol="filter_json", + allow_none=False, + desc="get_user_filter", + ) + + return db_to_json(def_json) + + def add_user_filter(self, user_localpart, user_filter): + def_json = encode_canonical_json(user_filter) + + # Need an atomic transaction to SELECT the maximal ID so far then + # INSERT a new one + def _do_txn(txn): + sql = ( + "SELECT filter_id FROM user_filters " + "WHERE user_id = ? AND filter_json = ?" + ) + txn.execute(sql, (user_localpart, bytearray(def_json))) + filter_id_response = txn.fetchone() + if filter_id_response is not None: + return filter_id_response[0] + + sql = "SELECT MAX(filter_id) FROM user_filters " "WHERE user_id = ?" + txn.execute(sql, (user_localpart,)) + max_id = txn.fetchone()[0] + if max_id is None: + filter_id = 0 + else: + filter_id = max_id + 1 + + sql = ( + "INSERT INTO user_filters (user_id, filter_id, filter_json)" + "VALUES(?, ?, ?)" + ) + txn.execute(sql, (user_localpart, filter_id, bytearray(def_json))) + + return filter_id + + return self.runInteraction("add_user_filter", _do_txn) diff --git a/synapse/storage/data_stores/main/group_server.py b/synapse/storage/data_stores/main/group_server.py new file mode 100644 index 0000000000..aeae5a2b28 --- /dev/null +++ b/synapse/storage/data_stores/main/group_server.py @@ -0,0 +1,1180 @@ +# -*- coding: utf-8 -*- +# Copyright 2017 Vector Creations Ltd +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from canonicaljson import json + +from twisted.internet import defer + +from synapse.api.errors import SynapseError +from synapse.storage._base import SQLBaseStore + +# The category ID for the "default" category. We don't store as null in the +# database to avoid the fun of null != null +_DEFAULT_CATEGORY_ID = "" +_DEFAULT_ROLE_ID = "" + + +class GroupServerStore(SQLBaseStore): + def set_group_join_policy(self, group_id, join_policy): + """Set the join policy of a group. + + join_policy can be one of: + * "invite" + * "open" + """ + return self._simple_update_one( + table="groups", + keyvalues={"group_id": group_id}, + updatevalues={"join_policy": join_policy}, + desc="set_group_join_policy", + ) + + def get_group(self, group_id): + return self._simple_select_one( + table="groups", + keyvalues={"group_id": group_id}, + retcols=( + "name", + "short_description", + "long_description", + "avatar_url", + "is_public", + "join_policy", + ), + allow_none=True, + desc="get_group", + ) + + def get_users_in_group(self, group_id, include_private=False): + # TODO: Pagination + + keyvalues = {"group_id": group_id} + if not include_private: + keyvalues["is_public"] = True + + return self._simple_select_list( + table="group_users", + keyvalues=keyvalues, + retcols=("user_id", "is_public", "is_admin"), + desc="get_users_in_group", + ) + + def get_invited_users_in_group(self, group_id): + # TODO: Pagination + + return self._simple_select_onecol( + table="group_invites", + keyvalues={"group_id": group_id}, + retcol="user_id", + desc="get_invited_users_in_group", + ) + + def get_rooms_in_group(self, group_id, include_private=False): + # TODO: Pagination + + keyvalues = {"group_id": group_id} + if not include_private: + keyvalues["is_public"] = True + + return self._simple_select_list( + table="group_rooms", + keyvalues=keyvalues, + retcols=("room_id", "is_public"), + desc="get_rooms_in_group", + ) + + def get_rooms_for_summary_by_category(self, group_id, include_private=False): + """Get the rooms and categories that should be included in a summary request + + Returns ([rooms], [categories]) + """ + + def _get_rooms_for_summary_txn(txn): + keyvalues = {"group_id": group_id} + if not include_private: + keyvalues["is_public"] = True + + sql = """ + SELECT room_id, is_public, category_id, room_order + FROM group_summary_rooms + WHERE group_id = ? + """ + + if not include_private: + sql += " AND is_public = ?" + txn.execute(sql, (group_id, True)) + else: + txn.execute(sql, (group_id,)) + + rooms = [ + { + "room_id": row[0], + "is_public": row[1], + "category_id": row[2] if row[2] != _DEFAULT_CATEGORY_ID else None, + "order": row[3], + } + for row in txn + ] + + sql = """ + SELECT category_id, is_public, profile, cat_order + FROM group_summary_room_categories + INNER JOIN group_room_categories USING (group_id, category_id) + WHERE group_id = ? + """ + + if not include_private: + sql += " AND is_public = ?" + txn.execute(sql, (group_id, True)) + else: + txn.execute(sql, (group_id,)) + + categories = { + row[0]: { + "is_public": row[1], + "profile": json.loads(row[2]), + "order": row[3], + } + for row in txn + } + + return rooms, categories + + return self.runInteraction("get_rooms_for_summary", _get_rooms_for_summary_txn) + + def add_room_to_summary(self, group_id, room_id, category_id, order, is_public): + return self.runInteraction( + "add_room_to_summary", + self._add_room_to_summary_txn, + group_id, + room_id, + category_id, + order, + is_public, + ) + + def _add_room_to_summary_txn( + self, txn, group_id, room_id, category_id, order, is_public + ): + """Add (or update) room's entry in summary. + + Args: + group_id (str) + room_id (str) + category_id (str): If not None then adds the category to the end of + the summary if its not already there. [Optional] + order (int): If not None inserts the room at that position, e.g. + an order of 1 will put the room first. Otherwise, the room gets + added to the end. + """ + room_in_group = self._simple_select_one_onecol_txn( + txn, + table="group_rooms", + keyvalues={"group_id": group_id, "room_id": room_id}, + retcol="room_id", + allow_none=True, + ) + if not room_in_group: + raise SynapseError(400, "room not in group") + + if category_id is None: + category_id = _DEFAULT_CATEGORY_ID + else: + cat_exists = self._simple_select_one_onecol_txn( + txn, + table="group_room_categories", + keyvalues={"group_id": group_id, "category_id": category_id}, + retcol="group_id", + allow_none=True, + ) + if not cat_exists: + raise SynapseError(400, "Category doesn't exist") + + # TODO: Check category is part of summary already + cat_exists = self._simple_select_one_onecol_txn( + txn, + table="group_summary_room_categories", + keyvalues={"group_id": group_id, "category_id": category_id}, + retcol="group_id", + allow_none=True, + ) + if not cat_exists: + # If not, add it with an order larger than all others + txn.execute( + """ + INSERT INTO group_summary_room_categories + (group_id, category_id, cat_order) + SELECT ?, ?, COALESCE(MAX(cat_order), 0) + 1 + FROM group_summary_room_categories + WHERE group_id = ? AND category_id = ? + """, + (group_id, category_id, group_id, category_id), + ) + + existing = self._simple_select_one_txn( + txn, + table="group_summary_rooms", + keyvalues={ + "group_id": group_id, + "room_id": room_id, + "category_id": category_id, + }, + retcols=("room_order", "is_public"), + allow_none=True, + ) + + if order is not None: + # Shuffle other room orders that come after the given order + sql = """ + UPDATE group_summary_rooms SET room_order = room_order + 1 + WHERE group_id = ? AND category_id = ? AND room_order >= ? + """ + txn.execute(sql, (group_id, category_id, order)) + elif not existing: + sql = """ + SELECT COALESCE(MAX(room_order), 0) + 1 FROM group_summary_rooms + WHERE group_id = ? AND category_id = ? + """ + txn.execute(sql, (group_id, category_id)) + order, = txn.fetchone() + + if existing: + to_update = {} + if order is not None: + to_update["room_order"] = order + if is_public is not None: + to_update["is_public"] = is_public + self._simple_update_txn( + txn, + table="group_summary_rooms", + keyvalues={ + "group_id": group_id, + "category_id": category_id, + "room_id": room_id, + }, + values=to_update, + ) + else: + if is_public is None: + is_public = True + + self._simple_insert_txn( + txn, + table="group_summary_rooms", + values={ + "group_id": group_id, + "category_id": category_id, + "room_id": room_id, + "room_order": order, + "is_public": is_public, + }, + ) + + def remove_room_from_summary(self, group_id, room_id, category_id): + if category_id is None: + category_id = _DEFAULT_CATEGORY_ID + + return self._simple_delete( + table="group_summary_rooms", + keyvalues={ + "group_id": group_id, + "category_id": category_id, + "room_id": room_id, + }, + desc="remove_room_from_summary", + ) + + @defer.inlineCallbacks + def get_group_categories(self, group_id): + rows = yield self._simple_select_list( + table="group_room_categories", + keyvalues={"group_id": group_id}, + retcols=("category_id", "is_public", "profile"), + desc="get_group_categories", + ) + + return { + row["category_id"]: { + "is_public": row["is_public"], + "profile": json.loads(row["profile"]), + } + for row in rows + } + + @defer.inlineCallbacks + def get_group_category(self, group_id, category_id): + category = yield self._simple_select_one( + table="group_room_categories", + keyvalues={"group_id": group_id, "category_id": category_id}, + retcols=("is_public", "profile"), + desc="get_group_category", + ) + + category["profile"] = json.loads(category["profile"]) + + return category + + def upsert_group_category(self, group_id, category_id, profile, is_public): + """Add/update room category for group + """ + insertion_values = {} + update_values = {"category_id": category_id} # This cannot be empty + + if profile is None: + insertion_values["profile"] = "{}" + else: + update_values["profile"] = json.dumps(profile) + + if is_public is None: + insertion_values["is_public"] = True + else: + update_values["is_public"] = is_public + + return self._simple_upsert( + table="group_room_categories", + keyvalues={"group_id": group_id, "category_id": category_id}, + values=update_values, + insertion_values=insertion_values, + desc="upsert_group_category", + ) + + def remove_group_category(self, group_id, category_id): + return self._simple_delete( + table="group_room_categories", + keyvalues={"group_id": group_id, "category_id": category_id}, + desc="remove_group_category", + ) + + @defer.inlineCallbacks + def get_group_roles(self, group_id): + rows = yield self._simple_select_list( + table="group_roles", + keyvalues={"group_id": group_id}, + retcols=("role_id", "is_public", "profile"), + desc="get_group_roles", + ) + + return { + row["role_id"]: { + "is_public": row["is_public"], + "profile": json.loads(row["profile"]), + } + for row in rows + } + + @defer.inlineCallbacks + def get_group_role(self, group_id, role_id): + role = yield self._simple_select_one( + table="group_roles", + keyvalues={"group_id": group_id, "role_id": role_id}, + retcols=("is_public", "profile"), + desc="get_group_role", + ) + + role["profile"] = json.loads(role["profile"]) + + return role + + def upsert_group_role(self, group_id, role_id, profile, is_public): + """Add/remove user role + """ + insertion_values = {} + update_values = {"role_id": role_id} # This cannot be empty + + if profile is None: + insertion_values["profile"] = "{}" + else: + update_values["profile"] = json.dumps(profile) + + if is_public is None: + insertion_values["is_public"] = True + else: + update_values["is_public"] = is_public + + return self._simple_upsert( + table="group_roles", + keyvalues={"group_id": group_id, "role_id": role_id}, + values=update_values, + insertion_values=insertion_values, + desc="upsert_group_role", + ) + + def remove_group_role(self, group_id, role_id): + return self._simple_delete( + table="group_roles", + keyvalues={"group_id": group_id, "role_id": role_id}, + desc="remove_group_role", + ) + + def add_user_to_summary(self, group_id, user_id, role_id, order, is_public): + return self.runInteraction( + "add_user_to_summary", + self._add_user_to_summary_txn, + group_id, + user_id, + role_id, + order, + is_public, + ) + + def _add_user_to_summary_txn( + self, txn, group_id, user_id, role_id, order, is_public + ): + """Add (or update) user's entry in summary. + + Args: + group_id (str) + user_id (str) + role_id (str): If not None then adds the role to the end of + the summary if its not already there. [Optional] + order (int): If not None inserts the user at that position, e.g. + an order of 1 will put the user first. Otherwise, the user gets + added to the end. + """ + user_in_group = self._simple_select_one_onecol_txn( + txn, + table="group_users", + keyvalues={"group_id": group_id, "user_id": user_id}, + retcol="user_id", + allow_none=True, + ) + if not user_in_group: + raise SynapseError(400, "user not in group") + + if role_id is None: + role_id = _DEFAULT_ROLE_ID + else: + role_exists = self._simple_select_one_onecol_txn( + txn, + table="group_roles", + keyvalues={"group_id": group_id, "role_id": role_id}, + retcol="group_id", + allow_none=True, + ) + if not role_exists: + raise SynapseError(400, "Role doesn't exist") + + # TODO: Check role is part of the summary already + role_exists = self._simple_select_one_onecol_txn( + txn, + table="group_summary_roles", + keyvalues={"group_id": group_id, "role_id": role_id}, + retcol="group_id", + allow_none=True, + ) + if not role_exists: + # If not, add it with an order larger than all others + txn.execute( + """ + INSERT INTO group_summary_roles + (group_id, role_id, role_order) + SELECT ?, ?, COALESCE(MAX(role_order), 0) + 1 + FROM group_summary_roles + WHERE group_id = ? AND role_id = ? + """, + (group_id, role_id, group_id, role_id), + ) + + existing = self._simple_select_one_txn( + txn, + table="group_summary_users", + keyvalues={"group_id": group_id, "user_id": user_id, "role_id": role_id}, + retcols=("user_order", "is_public"), + allow_none=True, + ) + + if order is not None: + # Shuffle other users orders that come after the given order + sql = """ + UPDATE group_summary_users SET user_order = user_order + 1 + WHERE group_id = ? AND role_id = ? AND user_order >= ? + """ + txn.execute(sql, (group_id, role_id, order)) + elif not existing: + sql = """ + SELECT COALESCE(MAX(user_order), 0) + 1 FROM group_summary_users + WHERE group_id = ? AND role_id = ? + """ + txn.execute(sql, (group_id, role_id)) + order, = txn.fetchone() + + if existing: + to_update = {} + if order is not None: + to_update["user_order"] = order + if is_public is not None: + to_update["is_public"] = is_public + self._simple_update_txn( + txn, + table="group_summary_users", + keyvalues={ + "group_id": group_id, + "role_id": role_id, + "user_id": user_id, + }, + values=to_update, + ) + else: + if is_public is None: + is_public = True + + self._simple_insert_txn( + txn, + table="group_summary_users", + values={ + "group_id": group_id, + "role_id": role_id, + "user_id": user_id, + "user_order": order, + "is_public": is_public, + }, + ) + + def remove_user_from_summary(self, group_id, user_id, role_id): + if role_id is None: + role_id = _DEFAULT_ROLE_ID + + return self._simple_delete( + table="group_summary_users", + keyvalues={"group_id": group_id, "role_id": role_id, "user_id": user_id}, + desc="remove_user_from_summary", + ) + + def get_users_for_summary_by_role(self, group_id, include_private=False): + """Get the users and roles that should be included in a summary request + + Returns ([users], [roles]) + """ + + def _get_users_for_summary_txn(txn): + keyvalues = {"group_id": group_id} + if not include_private: + keyvalues["is_public"] = True + + sql = """ + SELECT user_id, is_public, role_id, user_order + FROM group_summary_users + WHERE group_id = ? + """ + + if not include_private: + sql += " AND is_public = ?" + txn.execute(sql, (group_id, True)) + else: + txn.execute(sql, (group_id,)) + + users = [ + { + "user_id": row[0], + "is_public": row[1], + "role_id": row[2] if row[2] != _DEFAULT_ROLE_ID else None, + "order": row[3], + } + for row in txn + ] + + sql = """ + SELECT role_id, is_public, profile, role_order + FROM group_summary_roles + INNER JOIN group_roles USING (group_id, role_id) + WHERE group_id = ? + """ + + if not include_private: + sql += " AND is_public = ?" + txn.execute(sql, (group_id, True)) + else: + txn.execute(sql, (group_id,)) + + roles = { + row[0]: { + "is_public": row[1], + "profile": json.loads(row[2]), + "order": row[3], + } + for row in txn + } + + return users, roles + + return self.runInteraction( + "get_users_for_summary_by_role", _get_users_for_summary_txn + ) + + def is_user_in_group(self, user_id, group_id): + return self._simple_select_one_onecol( + table="group_users", + keyvalues={"group_id": group_id, "user_id": user_id}, + retcol="user_id", + allow_none=True, + desc="is_user_in_group", + ).addCallback(lambda r: bool(r)) + + def is_user_admin_in_group(self, group_id, user_id): + return self._simple_select_one_onecol( + table="group_users", + keyvalues={"group_id": group_id, "user_id": user_id}, + retcol="is_admin", + allow_none=True, + desc="is_user_admin_in_group", + ) + + def add_group_invite(self, group_id, user_id): + """Record that the group server has invited a user + """ + return self._simple_insert( + table="group_invites", + values={"group_id": group_id, "user_id": user_id}, + desc="add_group_invite", + ) + + def is_user_invited_to_local_group(self, group_id, user_id): + """Has the group server invited a user? + """ + return self._simple_select_one_onecol( + table="group_invites", + keyvalues={"group_id": group_id, "user_id": user_id}, + retcol="user_id", + desc="is_user_invited_to_local_group", + allow_none=True, + ) + + def get_users_membership_info_in_group(self, group_id, user_id): + """Get a dict describing the membership of a user in a group. + + Example if joined: + + { + "membership": "join", + "is_public": True, + "is_privileged": False, + } + + Returns an empty dict if the user is not join/invite/etc + """ + + def _get_users_membership_in_group_txn(txn): + row = self._simple_select_one_txn( + txn, + table="group_users", + keyvalues={"group_id": group_id, "user_id": user_id}, + retcols=("is_admin", "is_public"), + allow_none=True, + ) + + if row: + return { + "membership": "join", + "is_public": row["is_public"], + "is_privileged": row["is_admin"], + } + + row = self._simple_select_one_onecol_txn( + txn, + table="group_invites", + keyvalues={"group_id": group_id, "user_id": user_id}, + retcol="user_id", + allow_none=True, + ) + + if row: + return {"membership": "invite"} + + return {} + + return self.runInteraction( + "get_users_membership_info_in_group", _get_users_membership_in_group_txn + ) + + def add_user_to_group( + self, + group_id, + user_id, + is_admin=False, + is_public=True, + local_attestation=None, + remote_attestation=None, + ): + """Add a user to the group server. + + Args: + group_id (str) + user_id (str) + is_admin (bool) + is_public (bool) + local_attestation (dict): The attestation the GS created to give + to the remote server. Optional if the user and group are on the + same server + remote_attestation (dict): The attestation given to GS by remote + server. Optional if the user and group are on the same server + """ + + def _add_user_to_group_txn(txn): + self._simple_insert_txn( + txn, + table="group_users", + values={ + "group_id": group_id, + "user_id": user_id, + "is_admin": is_admin, + "is_public": is_public, + }, + ) + + self._simple_delete_txn( + txn, + table="group_invites", + keyvalues={"group_id": group_id, "user_id": user_id}, + ) + + if local_attestation: + self._simple_insert_txn( + txn, + table="group_attestations_renewals", + values={ + "group_id": group_id, + "user_id": user_id, + "valid_until_ms": local_attestation["valid_until_ms"], + }, + ) + if remote_attestation: + self._simple_insert_txn( + txn, + table="group_attestations_remote", + values={ + "group_id": group_id, + "user_id": user_id, + "valid_until_ms": remote_attestation["valid_until_ms"], + "attestation_json": json.dumps(remote_attestation), + }, + ) + + return self.runInteraction("add_user_to_group", _add_user_to_group_txn) + + def remove_user_from_group(self, group_id, user_id): + def _remove_user_from_group_txn(txn): + self._simple_delete_txn( + txn, + table="group_users", + keyvalues={"group_id": group_id, "user_id": user_id}, + ) + self._simple_delete_txn( + txn, + table="group_invites", + keyvalues={"group_id": group_id, "user_id": user_id}, + ) + self._simple_delete_txn( + txn, + table="group_attestations_renewals", + keyvalues={"group_id": group_id, "user_id": user_id}, + ) + self._simple_delete_txn( + txn, + table="group_attestations_remote", + keyvalues={"group_id": group_id, "user_id": user_id}, + ) + self._simple_delete_txn( + txn, + table="group_summary_users", + keyvalues={"group_id": group_id, "user_id": user_id}, + ) + + return self.runInteraction( + "remove_user_from_group", _remove_user_from_group_txn + ) + + def add_room_to_group(self, group_id, room_id, is_public): + return self._simple_insert( + table="group_rooms", + values={"group_id": group_id, "room_id": room_id, "is_public": is_public}, + desc="add_room_to_group", + ) + + def update_room_in_group_visibility(self, group_id, room_id, is_public): + return self._simple_update( + table="group_rooms", + keyvalues={"group_id": group_id, "room_id": room_id}, + updatevalues={"is_public": is_public}, + desc="update_room_in_group_visibility", + ) + + def remove_room_from_group(self, group_id, room_id): + def _remove_room_from_group_txn(txn): + self._simple_delete_txn( + txn, + table="group_rooms", + keyvalues={"group_id": group_id, "room_id": room_id}, + ) + + self._simple_delete_txn( + txn, + table="group_summary_rooms", + keyvalues={"group_id": group_id, "room_id": room_id}, + ) + + return self.runInteraction( + "remove_room_from_group", _remove_room_from_group_txn + ) + + def get_publicised_groups_for_user(self, user_id): + """Get all groups a user is publicising + """ + return self._simple_select_onecol( + table="local_group_membership", + keyvalues={"user_id": user_id, "membership": "join", "is_publicised": True}, + retcol="group_id", + desc="get_publicised_groups_for_user", + ) + + def update_group_publicity(self, group_id, user_id, publicise): + """Update whether the user is publicising their membership of the group + """ + return self._simple_update_one( + table="local_group_membership", + keyvalues={"group_id": group_id, "user_id": user_id}, + updatevalues={"is_publicised": publicise}, + desc="update_group_publicity", + ) + + @defer.inlineCallbacks + def register_user_group_membership( + self, + group_id, + user_id, + membership, + is_admin=False, + content={}, + local_attestation=None, + remote_attestation=None, + is_publicised=False, + ): + """Registers that a local user is a member of a (local or remote) group. + + Args: + group_id (str) + user_id (str) + membership (str) + is_admin (bool) + content (dict): Content of the membership, e.g. includes the inviter + if the user has been invited. + local_attestation (dict): If remote group then store the fact that we + have given out an attestation, else None. + remote_attestation (dict): If remote group then store the remote + attestation from the group, else None. + """ + + def _register_user_group_membership_txn(txn, next_id): + # TODO: Upsert? + self._simple_delete_txn( + txn, + table="local_group_membership", + keyvalues={"group_id": group_id, "user_id": user_id}, + ) + self._simple_insert_txn( + txn, + table="local_group_membership", + values={ + "group_id": group_id, + "user_id": user_id, + "is_admin": is_admin, + "membership": membership, + "is_publicised": is_publicised, + "content": json.dumps(content), + }, + ) + + self._simple_insert_txn( + txn, + table="local_group_updates", + values={ + "stream_id": next_id, + "group_id": group_id, + "user_id": user_id, + "type": "membership", + "content": json.dumps( + {"membership": membership, "content": content} + ), + }, + ) + self._group_updates_stream_cache.entity_has_changed(user_id, next_id) + + # TODO: Insert profile to ensure it comes down stream if its a join. + + if membership == "join": + if local_attestation: + self._simple_insert_txn( + txn, + table="group_attestations_renewals", + values={ + "group_id": group_id, + "user_id": user_id, + "valid_until_ms": local_attestation["valid_until_ms"], + }, + ) + if remote_attestation: + self._simple_insert_txn( + txn, + table="group_attestations_remote", + values={ + "group_id": group_id, + "user_id": user_id, + "valid_until_ms": remote_attestation["valid_until_ms"], + "attestation_json": json.dumps(remote_attestation), + }, + ) + else: + self._simple_delete_txn( + txn, + table="group_attestations_renewals", + keyvalues={"group_id": group_id, "user_id": user_id}, + ) + self._simple_delete_txn( + txn, + table="group_attestations_remote", + keyvalues={"group_id": group_id, "user_id": user_id}, + ) + + return next_id + + with self._group_updates_id_gen.get_next() as next_id: + res = yield self.runInteraction( + "register_user_group_membership", + _register_user_group_membership_txn, + next_id, + ) + return res + + @defer.inlineCallbacks + def create_group( + self, group_id, user_id, name, avatar_url, short_description, long_description + ): + yield self._simple_insert( + table="groups", + values={ + "group_id": group_id, + "name": name, + "avatar_url": avatar_url, + "short_description": short_description, + "long_description": long_description, + "is_public": True, + }, + desc="create_group", + ) + + @defer.inlineCallbacks + def update_group_profile(self, group_id, profile): + yield self._simple_update_one( + table="groups", + keyvalues={"group_id": group_id}, + updatevalues=profile, + desc="update_group_profile", + ) + + def get_attestations_need_renewals(self, valid_until_ms): + """Get all attestations that need to be renewed until givent time + """ + + def _get_attestations_need_renewals_txn(txn): + sql = """ + SELECT group_id, user_id FROM group_attestations_renewals + WHERE valid_until_ms <= ? + """ + txn.execute(sql, (valid_until_ms,)) + return self.cursor_to_dict(txn) + + return self.runInteraction( + "get_attestations_need_renewals", _get_attestations_need_renewals_txn + ) + + def update_attestation_renewal(self, group_id, user_id, attestation): + """Update an attestation that we have renewed + """ + return self._simple_update_one( + table="group_attestations_renewals", + keyvalues={"group_id": group_id, "user_id": user_id}, + updatevalues={"valid_until_ms": attestation["valid_until_ms"]}, + desc="update_attestation_renewal", + ) + + def update_remote_attestion(self, group_id, user_id, attestation): + """Update an attestation that a remote has renewed + """ + return self._simple_update_one( + table="group_attestations_remote", + keyvalues={"group_id": group_id, "user_id": user_id}, + updatevalues={ + "valid_until_ms": attestation["valid_until_ms"], + "attestation_json": json.dumps(attestation), + }, + desc="update_remote_attestion", + ) + + def remove_attestation_renewal(self, group_id, user_id): + """Remove an attestation that we thought we should renew, but actually + shouldn't. Ideally this would never get called as we would never + incorrectly try and do attestations for local users on local groups. + + Args: + group_id (str) + user_id (str) + """ + return self._simple_delete( + table="group_attestations_renewals", + keyvalues={"group_id": group_id, "user_id": user_id}, + desc="remove_attestation_renewal", + ) + + @defer.inlineCallbacks + def get_remote_attestation(self, group_id, user_id): + """Get the attestation that proves the remote agrees that the user is + in the group. + """ + row = yield self._simple_select_one( + table="group_attestations_remote", + keyvalues={"group_id": group_id, "user_id": user_id}, + retcols=("valid_until_ms", "attestation_json"), + desc="get_remote_attestation", + allow_none=True, + ) + + now = int(self._clock.time_msec()) + if row and now < row["valid_until_ms"]: + return json.loads(row["attestation_json"]) + + return None + + def get_joined_groups(self, user_id): + return self._simple_select_onecol( + table="local_group_membership", + keyvalues={"user_id": user_id, "membership": "join"}, + retcol="group_id", + desc="get_joined_groups", + ) + + def get_all_groups_for_user(self, user_id, now_token): + def _get_all_groups_for_user_txn(txn): + sql = """ + SELECT group_id, type, membership, u.content + FROM local_group_updates AS u + INNER JOIN local_group_membership USING (group_id, user_id) + WHERE user_id = ? AND membership != 'leave' + AND stream_id <= ? + """ + txn.execute(sql, (user_id, now_token)) + return [ + { + "group_id": row[0], + "type": row[1], + "membership": row[2], + "content": json.loads(row[3]), + } + for row in txn + ] + + return self.runInteraction( + "get_all_groups_for_user", _get_all_groups_for_user_txn + ) + + def get_groups_changes_for_user(self, user_id, from_token, to_token): + from_token = int(from_token) + has_changed = self._group_updates_stream_cache.has_entity_changed( + user_id, from_token + ) + if not has_changed: + return [] + + def _get_groups_changes_for_user_txn(txn): + sql = """ + SELECT group_id, membership, type, u.content + FROM local_group_updates AS u + INNER JOIN local_group_membership USING (group_id, user_id) + WHERE user_id = ? AND ? < stream_id AND stream_id <= ? + """ + txn.execute(sql, (user_id, from_token, to_token)) + return [ + { + "group_id": group_id, + "membership": membership, + "type": gtype, + "content": json.loads(content_json), + } + for group_id, membership, gtype, content_json in txn + ] + + return self.runInteraction( + "get_groups_changes_for_user", _get_groups_changes_for_user_txn + ) + + def get_all_groups_changes(self, from_token, to_token, limit): + from_token = int(from_token) + has_changed = self._group_updates_stream_cache.has_any_entity_changed( + from_token + ) + if not has_changed: + return [] + + def _get_all_groups_changes_txn(txn): + sql = """ + SELECT stream_id, group_id, user_id, type, content + FROM local_group_updates + WHERE ? < stream_id AND stream_id <= ? + LIMIT ? + """ + txn.execute(sql, (from_token, to_token, limit)) + return [ + (stream_id, group_id, user_id, gtype, json.loads(content_json)) + for stream_id, group_id, user_id, gtype, content_json in txn + ] + + return self.runInteraction( + "get_all_groups_changes", _get_all_groups_changes_txn + ) + + def get_group_stream_token(self): + return self._group_updates_id_gen.get_current_token() + + def delete_group(self, group_id): + """Deletes a group fully from the database. + + Args: + group_id (str) + + Returns: + Deferred + """ + + def _delete_group_txn(txn): + tables = [ + "groups", + "group_users", + "group_invites", + "group_rooms", + "group_summary_rooms", + "group_summary_room_categories", + "group_room_categories", + "group_summary_users", + "group_summary_roles", + "group_roles", + "group_attestations_renewals", + "group_attestations_remote", + ] + + for table in tables: + self._simple_delete_txn( + txn, table=table, keyvalues={"group_id": group_id} + ) + + return self.runInteraction("delete_group", _delete_group_txn) diff --git a/synapse/storage/data_stores/main/keys.py b/synapse/storage/data_stores/main/keys.py new file mode 100644 index 0000000000..ebc7db3ed6 --- /dev/null +++ b/synapse/storage/data_stores/main/keys.py @@ -0,0 +1,214 @@ +# -*- coding: utf-8 -*- +# Copyright 2014-2016 OpenMarket Ltd +# Copyright 2019 New Vector Ltd. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import itertools +import logging + +import six + +from signedjson.key import decode_verify_key_bytes + +from synapse.storage._base import SQLBaseStore +from synapse.storage.keys import FetchKeyResult +from synapse.util import batch_iter +from synapse.util.caches.descriptors import cached, cachedList + +logger = logging.getLogger(__name__) + +# py2 sqlite has buffer hardcoded as only binary type, so we must use it, +# despite being deprecated and removed in favor of memoryview +if six.PY2: + db_binary_type = six.moves.builtins.buffer +else: + db_binary_type = memoryview + + +class KeyStore(SQLBaseStore): + """Persistence for signature verification keys + """ + + @cached() + def _get_server_verify_key(self, server_name_and_key_id): + raise NotImplementedError() + + @cachedList( + cached_method_name="_get_server_verify_key", list_name="server_name_and_key_ids" + ) + def get_server_verify_keys(self, server_name_and_key_ids): + """ + Args: + server_name_and_key_ids (iterable[Tuple[str, str]]): + iterable of (server_name, key-id) tuples to fetch keys for + + Returns: + Deferred: resolves to dict[Tuple[str, str], FetchKeyResult|None]: + map from (server_name, key_id) -> FetchKeyResult, or None if the key is + unknown + """ + keys = {} + + def _get_keys(txn, batch): + """Processes a batch of keys to fetch, and adds the result to `keys`.""" + + # batch_iter always returns tuples so it's safe to do len(batch) + sql = ( + "SELECT server_name, key_id, verify_key, ts_valid_until_ms " + "FROM server_signature_keys WHERE 1=0" + ) + " OR (server_name=? AND key_id=?)" * len(batch) + + txn.execute(sql, tuple(itertools.chain.from_iterable(batch))) + + for row in txn: + server_name, key_id, key_bytes, ts_valid_until_ms = row + + if ts_valid_until_ms is None: + # Old keys may be stored with a ts_valid_until_ms of null, + # in which case we treat this as if it was set to `0`, i.e. + # it won't match key requests that define a minimum + # `ts_valid_until_ms`. + ts_valid_until_ms = 0 + + res = FetchKeyResult( + verify_key=decode_verify_key_bytes(key_id, bytes(key_bytes)), + valid_until_ts=ts_valid_until_ms, + ) + keys[(server_name, key_id)] = res + + def _txn(txn): + for batch in batch_iter(server_name_and_key_ids, 50): + _get_keys(txn, batch) + return keys + + return self.runInteraction("get_server_verify_keys", _txn) + + def store_server_verify_keys(self, from_server, ts_added_ms, verify_keys): + """Stores NACL verification keys for remote servers. + Args: + from_server (str): Where the verification keys were looked up + ts_added_ms (int): The time to record that the key was added + verify_keys (iterable[tuple[str, str, FetchKeyResult]]): + keys to be stored. Each entry is a triplet of + (server_name, key_id, key). + """ + key_values = [] + value_values = [] + invalidations = [] + for server_name, key_id, fetch_result in verify_keys: + key_values.append((server_name, key_id)) + value_values.append( + ( + from_server, + ts_added_ms, + fetch_result.valid_until_ts, + db_binary_type(fetch_result.verify_key.encode()), + ) + ) + # invalidate takes a tuple corresponding to the params of + # _get_server_verify_key. _get_server_verify_key only takes one + # param, which is itself the 2-tuple (server_name, key_id). + invalidations.append((server_name, key_id)) + + def _invalidate(res): + f = self._get_server_verify_key.invalidate + for i in invalidations: + f((i,)) + return res + + return self.runInteraction( + "store_server_verify_keys", + self._simple_upsert_many_txn, + table="server_signature_keys", + key_names=("server_name", "key_id"), + key_values=key_values, + value_names=( + "from_server", + "ts_added_ms", + "ts_valid_until_ms", + "verify_key", + ), + value_values=value_values, + ).addCallback(_invalidate) + + def store_server_keys_json( + self, server_name, key_id, from_server, ts_now_ms, ts_expires_ms, key_json_bytes + ): + """Stores the JSON bytes for a set of keys from a server + The JSON should be signed by the originating server, the intermediate + server, and by this server. Updates the value for the + (server_name, key_id, from_server) triplet if one already existed. + Args: + server_name (str): The name of the server. + key_id (str): The identifer of the key this JSON is for. + from_server (str): The server this JSON was fetched from. + ts_now_ms (int): The time now in milliseconds. + ts_valid_until_ms (int): The time when this json stops being valid. + key_json (bytes): The encoded JSON. + """ + return self._simple_upsert( + table="server_keys_json", + keyvalues={ + "server_name": server_name, + "key_id": key_id, + "from_server": from_server, + }, + values={ + "server_name": server_name, + "key_id": key_id, + "from_server": from_server, + "ts_added_ms": ts_now_ms, + "ts_valid_until_ms": ts_expires_ms, + "key_json": db_binary_type(key_json_bytes), + }, + desc="store_server_keys_json", + ) + + def get_server_keys_json(self, server_keys): + """Retrive the key json for a list of server_keys and key ids. + If no keys are found for a given server, key_id and source then + that server, key_id, and source triplet entry will be an empty list. + The JSON is returned as a byte array so that it can be efficiently + used in an HTTP response. + Args: + server_keys (list): List of (server_name, key_id, source) triplets. + Returns: + Deferred[dict[Tuple[str, str, str|None], list[dict]]]: + Dict mapping (server_name, key_id, source) triplets to lists of dicts + """ + + def _get_server_keys_json_txn(txn): + results = {} + for server_name, key_id, from_server in server_keys: + keyvalues = {"server_name": server_name} + if key_id is not None: + keyvalues["key_id"] = key_id + if from_server is not None: + keyvalues["from_server"] = from_server + rows = self._simple_select_list_txn( + txn, + "server_keys_json", + keyvalues=keyvalues, + retcols=( + "key_id", + "from_server", + "ts_added_ms", + "ts_valid_until_ms", + "key_json", + ), + ) + results[(server_name, key_id, from_server)] = rows + return results + + return self.runInteraction("get_server_keys_json", _get_server_keys_json_txn) diff --git a/synapse/storage/data_stores/main/media_repository.py b/synapse/storage/data_stores/main/media_repository.py new file mode 100644 index 0000000000..84b5f3ad5e --- /dev/null +++ b/synapse/storage/data_stores/main/media_repository.py @@ -0,0 +1,378 @@ +# -*- coding: utf-8 -*- +# Copyright 2014-2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +from synapse.storage.background_updates import BackgroundUpdateStore + + +class MediaRepositoryBackgroundUpdateStore(BackgroundUpdateStore): + def __init__(self, db_conn, hs): + super(MediaRepositoryBackgroundUpdateStore, self).__init__(db_conn, hs) + + self.register_background_index_update( + update_name="local_media_repository_url_idx", + index_name="local_media_repository_url_idx", + table="local_media_repository", + columns=["created_ts"], + where_clause="url_cache IS NOT NULL", + ) + + +class MediaRepositoryStore(MediaRepositoryBackgroundUpdateStore): + """Persistence for attachments and avatars""" + + def __init__(self, db_conn, hs): + super(MediaRepositoryStore, self).__init__(db_conn, hs) + + def get_local_media(self, media_id): + """Get the metadata for a local piece of media + Returns: + None if the media_id doesn't exist. + """ + return self._simple_select_one( + "local_media_repository", + {"media_id": media_id}, + ( + "media_type", + "media_length", + "upload_name", + "created_ts", + "quarantined_by", + "url_cache", + ), + allow_none=True, + desc="get_local_media", + ) + + def store_local_media( + self, + media_id, + media_type, + time_now_ms, + upload_name, + media_length, + user_id, + url_cache=None, + ): + return self._simple_insert( + "local_media_repository", + { + "media_id": media_id, + "media_type": media_type, + "created_ts": time_now_ms, + "upload_name": upload_name, + "media_length": media_length, + "user_id": user_id.to_string(), + "url_cache": url_cache, + }, + desc="store_local_media", + ) + + def get_url_cache(self, url, ts): + """Get the media_id and ts for a cached URL as of the given timestamp + Returns: + None if the URL isn't cached. + """ + + def get_url_cache_txn(txn): + # get the most recently cached result (relative to the given ts) + sql = ( + "SELECT response_code, etag, expires_ts, og, media_id, download_ts" + " FROM local_media_repository_url_cache" + " WHERE url = ? AND download_ts <= ?" + " ORDER BY download_ts DESC LIMIT 1" + ) + txn.execute(sql, (url, ts)) + row = txn.fetchone() + + if not row: + # ...or if we've requested a timestamp older than the oldest + # copy in the cache, return the oldest copy (if any) + sql = ( + "SELECT response_code, etag, expires_ts, og, media_id, download_ts" + " FROM local_media_repository_url_cache" + " WHERE url = ? AND download_ts > ?" + " ORDER BY download_ts ASC LIMIT 1" + ) + txn.execute(sql, (url, ts)) + row = txn.fetchone() + + if not row: + return None + + return dict( + zip( + ( + "response_code", + "etag", + "expires_ts", + "og", + "media_id", + "download_ts", + ), + row, + ) + ) + + return self.runInteraction("get_url_cache", get_url_cache_txn) + + def store_url_cache( + self, url, response_code, etag, expires_ts, og, media_id, download_ts + ): + return self._simple_insert( + "local_media_repository_url_cache", + { + "url": url, + "response_code": response_code, + "etag": etag, + "expires_ts": expires_ts, + "og": og, + "media_id": media_id, + "download_ts": download_ts, + }, + desc="store_url_cache", + ) + + def get_local_media_thumbnails(self, media_id): + return self._simple_select_list( + "local_media_repository_thumbnails", + {"media_id": media_id}, + ( + "thumbnail_width", + "thumbnail_height", + "thumbnail_method", + "thumbnail_type", + "thumbnail_length", + ), + desc="get_local_media_thumbnails", + ) + + def store_local_thumbnail( + self, + media_id, + thumbnail_width, + thumbnail_height, + thumbnail_type, + thumbnail_method, + thumbnail_length, + ): + return self._simple_insert( + "local_media_repository_thumbnails", + { + "media_id": media_id, + "thumbnail_width": thumbnail_width, + "thumbnail_height": thumbnail_height, + "thumbnail_method": thumbnail_method, + "thumbnail_type": thumbnail_type, + "thumbnail_length": thumbnail_length, + }, + desc="store_local_thumbnail", + ) + + def get_cached_remote_media(self, origin, media_id): + return self._simple_select_one( + "remote_media_cache", + {"media_origin": origin, "media_id": media_id}, + ( + "media_type", + "media_length", + "upload_name", + "created_ts", + "filesystem_id", + "quarantined_by", + ), + allow_none=True, + desc="get_cached_remote_media", + ) + + def store_cached_remote_media( + self, + origin, + media_id, + media_type, + media_length, + time_now_ms, + upload_name, + filesystem_id, + ): + return self._simple_insert( + "remote_media_cache", + { + "media_origin": origin, + "media_id": media_id, + "media_type": media_type, + "media_length": media_length, + "created_ts": time_now_ms, + "upload_name": upload_name, + "filesystem_id": filesystem_id, + "last_access_ts": time_now_ms, + }, + desc="store_cached_remote_media", + ) + + def update_cached_last_access_time(self, local_media, remote_media, time_ms): + """Updates the last access time of the given media + + Args: + local_media (iterable[str]): Set of media_ids + remote_media (iterable[(str, str)]): Set of (server_name, media_id) + time_ms: Current time in milliseconds + """ + + def update_cache_txn(txn): + sql = ( + "UPDATE remote_media_cache SET last_access_ts = ?" + " WHERE media_origin = ? AND media_id = ?" + ) + + txn.executemany( + sql, + ( + (time_ms, media_origin, media_id) + for media_origin, media_id in remote_media + ), + ) + + sql = ( + "UPDATE local_media_repository SET last_access_ts = ?" + " WHERE media_id = ?" + ) + + txn.executemany(sql, ((time_ms, media_id) for media_id in local_media)) + + return self.runInteraction("update_cached_last_access_time", update_cache_txn) + + def get_remote_media_thumbnails(self, origin, media_id): + return self._simple_select_list( + "remote_media_cache_thumbnails", + {"media_origin": origin, "media_id": media_id}, + ( + "thumbnail_width", + "thumbnail_height", + "thumbnail_method", + "thumbnail_type", + "thumbnail_length", + "filesystem_id", + ), + desc="get_remote_media_thumbnails", + ) + + def store_remote_media_thumbnail( + self, + origin, + media_id, + filesystem_id, + thumbnail_width, + thumbnail_height, + thumbnail_type, + thumbnail_method, + thumbnail_length, + ): + return self._simple_insert( + "remote_media_cache_thumbnails", + { + "media_origin": origin, + "media_id": media_id, + "thumbnail_width": thumbnail_width, + "thumbnail_height": thumbnail_height, + "thumbnail_method": thumbnail_method, + "thumbnail_type": thumbnail_type, + "thumbnail_length": thumbnail_length, + "filesystem_id": filesystem_id, + }, + desc="store_remote_media_thumbnail", + ) + + def get_remote_media_before(self, before_ts): + sql = ( + "SELECT media_origin, media_id, filesystem_id" + " FROM remote_media_cache" + " WHERE last_access_ts < ?" + ) + + return self._execute( + "get_remote_media_before", self.cursor_to_dict, sql, before_ts + ) + + def delete_remote_media(self, media_origin, media_id): + def delete_remote_media_txn(txn): + self._simple_delete_txn( + txn, + "remote_media_cache", + keyvalues={"media_origin": media_origin, "media_id": media_id}, + ) + self._simple_delete_txn( + txn, + "remote_media_cache_thumbnails", + keyvalues={"media_origin": media_origin, "media_id": media_id}, + ) + + return self.runInteraction("delete_remote_media", delete_remote_media_txn) + + def get_expired_url_cache(self, now_ts): + sql = ( + "SELECT media_id FROM local_media_repository_url_cache" + " WHERE expires_ts < ?" + " ORDER BY expires_ts ASC" + " LIMIT 500" + ) + + def _get_expired_url_cache_txn(txn): + txn.execute(sql, (now_ts,)) + return [row[0] for row in txn] + + return self.runInteraction("get_expired_url_cache", _get_expired_url_cache_txn) + + def delete_url_cache(self, media_ids): + if len(media_ids) == 0: + return + + sql = "DELETE FROM local_media_repository_url_cache" " WHERE media_id = ?" + + def _delete_url_cache_txn(txn): + txn.executemany(sql, [(media_id,) for media_id in media_ids]) + + return self.runInteraction("delete_url_cache", _delete_url_cache_txn) + + def get_url_cache_media_before(self, before_ts): + sql = ( + "SELECT media_id FROM local_media_repository" + " WHERE created_ts < ? AND url_cache IS NOT NULL" + " ORDER BY created_ts ASC" + " LIMIT 500" + ) + + def _get_url_cache_media_before_txn(txn): + txn.execute(sql, (before_ts,)) + return [row[0] for row in txn] + + return self.runInteraction( + "get_url_cache_media_before", _get_url_cache_media_before_txn + ) + + def delete_url_cache_media(self, media_ids): + if len(media_ids) == 0: + return + + def _delete_url_cache_media_txn(txn): + sql = "DELETE FROM local_media_repository" " WHERE media_id = ?" + + txn.executemany(sql, [(media_id,) for media_id in media_ids]) + + sql = "DELETE FROM local_media_repository_thumbnails" " WHERE media_id = ?" + + txn.executemany(sql, [(media_id,) for media_id in media_ids]) + + return self.runInteraction( + "delete_url_cache_media", _delete_url_cache_media_txn + ) diff --git a/synapse/storage/data_stores/main/monthly_active_users.py b/synapse/storage/data_stores/main/monthly_active_users.py new file mode 100644 index 0000000000..e6ee1e4aaa --- /dev/null +++ b/synapse/storage/data_stores/main/monthly_active_users.py @@ -0,0 +1,328 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import logging + +from twisted.internet import defer + +from synapse.storage._base import SQLBaseStore +from synapse.util.caches.descriptors import cached + +logger = logging.getLogger(__name__) + +# Number of msec of granularity to store the monthly_active_user timestamp +# This means it is not necessary to update the table on every request +LAST_SEEN_GRANULARITY = 60 * 60 * 1000 + + +class MonthlyActiveUsersStore(SQLBaseStore): + def __init__(self, dbconn, hs): + super(MonthlyActiveUsersStore, self).__init__(None, hs) + self._clock = hs.get_clock() + self.hs = hs + # Do not add more reserved users than the total allowable number + self._new_transaction( + dbconn, + "initialise_mau_threepids", + [], + [], + self._initialise_reserved_users, + hs.config.mau_limits_reserved_threepids[: self.hs.config.max_mau_value], + ) + + def _initialise_reserved_users(self, txn, threepids): + """Ensures that reserved threepids are accounted for in the MAU table, should + be called on start up. + + Args: + txn (cursor): + threepids (list[dict]): List of threepid dicts to reserve + """ + + for tp in threepids: + user_id = self.get_user_id_by_threepid_txn(txn, tp["medium"], tp["address"]) + + if user_id: + is_support = self.is_support_user_txn(txn, user_id) + if not is_support: + self.upsert_monthly_active_user_txn(txn, user_id) + else: + logger.warning("mau limit reserved threepid %s not found in db" % tp) + + @defer.inlineCallbacks + def reap_monthly_active_users(self): + """Cleans out monthly active user table to ensure that no stale + entries exist. + + Returns: + Deferred[] + """ + + def _reap_users(txn, reserved_users): + """ + Args: + reserved_users (tuple): reserved users to preserve + """ + + thirty_days_ago = int(self._clock.time_msec()) - (1000 * 60 * 60 * 24 * 30) + query_args = [thirty_days_ago] + base_sql = "DELETE FROM monthly_active_users WHERE timestamp < ?" + + # Need if/else since 'AND user_id NOT IN ({})' fails on Postgres + # when len(reserved_users) == 0. Works fine on sqlite. + if len(reserved_users) > 0: + # questionmarks is a hack to overcome sqlite not supporting + # tuples in 'WHERE IN %s' + question_marks = ",".join("?" * len(reserved_users)) + + query_args.extend(reserved_users) + sql = base_sql + " AND user_id NOT IN ({})".format(question_marks) + else: + sql = base_sql + + txn.execute(sql, query_args) + + max_mau_value = self.hs.config.max_mau_value + if self.hs.config.limit_usage_by_mau: + # If MAU user count still exceeds the MAU threshold, then delete on + # a least recently active basis. + # Note it is not possible to write this query using OFFSET due to + # incompatibilities in how sqlite and postgres support the feature. + # sqlite requires 'LIMIT -1 OFFSET ?', the LIMIT must be present + # While Postgres does not require 'LIMIT', but also does not support + # negative LIMIT values. So there is no way to write it that both can + # support + if len(reserved_users) == 0: + sql = """ + DELETE FROM monthly_active_users + WHERE user_id NOT IN ( + SELECT user_id FROM monthly_active_users + ORDER BY timestamp DESC + LIMIT ? + ) + """ + txn.execute(sql, (max_mau_value,)) + # Need if/else since 'AND user_id NOT IN ({})' fails on Postgres + # when len(reserved_users) == 0. Works fine on sqlite. + else: + # Must be >= 0 for postgres + num_of_non_reserved_users_to_remove = max( + max_mau_value - len(reserved_users), 0 + ) + + # It is important to filter reserved users twice to guard + # against the case where the reserved user is present in the + # SELECT, meaning that a legitmate mau is deleted. + sql = """ + DELETE FROM monthly_active_users + WHERE user_id NOT IN ( + SELECT user_id FROM monthly_active_users + WHERE user_id NOT IN ({}) + ORDER BY timestamp DESC + LIMIT ? + ) + AND user_id NOT IN ({}) + """.format( + question_marks, question_marks + ) + + query_args = [ + *reserved_users, + num_of_non_reserved_users_to_remove, + *reserved_users, + ] + + txn.execute(sql, query_args) + + reserved_users = yield self.get_registered_reserved_users() + yield self.runInteraction( + "reap_monthly_active_users", _reap_users, reserved_users + ) + # It seems poor to invalidate the whole cache, Postgres supports + # 'Returning' which would allow me to invalidate only the + # specific users, but sqlite has no way to do this and instead + # I would need to SELECT and the DELETE which without locking + # is racy. + # Have resolved to invalidate the whole cache for now and do + # something about it if and when the perf becomes significant + self.user_last_seen_monthly_active.invalidate_all() + self.get_monthly_active_count.invalidate_all() + + @cached(num_args=0) + def get_monthly_active_count(self): + """Generates current count of monthly active users + + Returns: + Defered[int]: Number of current monthly active users + """ + + def _count_users(txn): + sql = "SELECT COALESCE(count(*), 0) FROM monthly_active_users" + + txn.execute(sql) + count, = txn.fetchone() + return count + + return self.runInteraction("count_users", _count_users) + + @defer.inlineCallbacks + def get_registered_reserved_users(self): + """Of the reserved threepids defined in config, which are associated + with registered users? + + Returns: + Defered[list]: Real reserved users + """ + users = [] + + for tp in self.hs.config.mau_limits_reserved_threepids[ + : self.hs.config.max_mau_value + ]: + user_id = yield self.hs.get_datastore().get_user_id_by_threepid( + tp["medium"], tp["address"] + ) + if user_id: + users.append(user_id) + + return users + + @defer.inlineCallbacks + def upsert_monthly_active_user(self, user_id): + """Updates or inserts the user into the monthly active user table, which + is used to track the current MAU usage of the server + + Args: + user_id (str): user to add/update + """ + # Support user never to be included in MAU stats. Note I can't easily call this + # from upsert_monthly_active_user_txn because then I need a _txn form of + # is_support_user which is complicated because I want to cache the result. + # Therefore I call it here and ignore the case where + # upsert_monthly_active_user_txn is called directly from + # _initialise_reserved_users reasoning that it would be very strange to + # include a support user in this context. + + is_support = yield self.is_support_user(user_id) + if is_support: + return + + yield self.runInteraction( + "upsert_monthly_active_user", self.upsert_monthly_active_user_txn, user_id + ) + + user_in_mau = self.user_last_seen_monthly_active.cache.get( + (user_id,), None, update_metrics=False + ) + if user_in_mau is None: + self.get_monthly_active_count.invalidate(()) + + self.user_last_seen_monthly_active.invalidate((user_id,)) + + def upsert_monthly_active_user_txn(self, txn, user_id): + """Updates or inserts monthly active user member + + Note that, after calling this method, it will generally be necessary + to invalidate the caches on user_last_seen_monthly_active and + get_monthly_active_count. We can't do that here, because we are running + in a database thread rather than the main thread, and we can't call + txn.call_after because txn may not be a LoggingTransaction. + + We consciously do not call is_support_txn from this method because it + is not possible to cache the response. is_support_txn will be false in + almost all cases, so it seems reasonable to call it only for + upsert_monthly_active_user and to call is_support_txn manually + for cases where upsert_monthly_active_user_txn is called directly, + like _initialise_reserved_users + + In short, don't call this method with support users. (Support users + should not appear in the MAU stats). + + Args: + txn (cursor): + user_id (str): user to add/update + + Returns: + bool: True if a new entry was created, False if an + existing one was updated. + """ + + # Am consciously deciding to lock the table on the basis that is ought + # never be a big table and alternative approaches (batching multiple + # upserts into a single txn) introduced a lot of extra complexity. + # See https://github.com/matrix-org/synapse/issues/3854 for more + is_insert = self._simple_upsert_txn( + txn, + table="monthly_active_users", + keyvalues={"user_id": user_id}, + values={"timestamp": int(self._clock.time_msec())}, + ) + + return is_insert + + @cached(num_args=1) + def user_last_seen_monthly_active(self, user_id): + """ + Checks if a given user is part of the monthly active user group + Arguments: + user_id (str): user to add/update + Return: + Deferred[int] : timestamp since last seen, None if never seen + + """ + + return self._simple_select_one_onecol( + table="monthly_active_users", + keyvalues={"user_id": user_id}, + retcol="timestamp", + allow_none=True, + desc="user_last_seen_monthly_active", + ) + + @defer.inlineCallbacks + def populate_monthly_active_users(self, user_id): + """Checks on the state of monthly active user limits and optionally + add the user to the monthly active tables + + Args: + user_id(str): the user_id to query + """ + if self.hs.config.limit_usage_by_mau or self.hs.config.mau_stats_only: + # Trial users and guests should not be included as part of MAU group + is_guest = yield self.is_guest(user_id) + if is_guest: + return + is_trial = yield self.is_trial_user(user_id) + if is_trial: + return + + last_seen_timestamp = yield self.user_last_seen_monthly_active(user_id) + now = self.hs.get_clock().time_msec() + + # We want to reduce to the total number of db writes, and are happy + # to trade accuracy of timestamp in order to lighten load. This means + # We always insert new users (where MAU threshold has not been reached), + # but only update if we have not previously seen the user for + # LAST_SEEN_GRANULARITY ms + if last_seen_timestamp is None: + # In the case where mau_stats_only is True and limit_usage_by_mau is + # False, there is no point in checking get_monthly_active_count - it + # adds no value and will break the logic if max_mau_value is exceeded. + if not self.hs.config.limit_usage_by_mau: + yield self.upsert_monthly_active_user(user_id) + else: + count = yield self.get_monthly_active_count() + if count < self.hs.config.max_mau_value: + yield self.upsert_monthly_active_user(user_id) + elif now - last_seen_timestamp > LAST_SEEN_GRANULARITY: + yield self.upsert_monthly_active_user(user_id) diff --git a/synapse/storage/data_stores/main/openid.py b/synapse/storage/data_stores/main/openid.py new file mode 100644 index 0000000000..79b40044d9 --- /dev/null +++ b/synapse/storage/data_stores/main/openid.py @@ -0,0 +1,31 @@ +from synapse.storage._base import SQLBaseStore + + +class OpenIdStore(SQLBaseStore): + def insert_open_id_token(self, token, ts_valid_until_ms, user_id): + return self._simple_insert( + table="open_id_tokens", + values={ + "token": token, + "ts_valid_until_ms": ts_valid_until_ms, + "user_id": user_id, + }, + desc="insert_open_id_token", + ) + + def get_user_id_for_open_id_token(self, token, ts_now_ms): + def get_user_id_for_token_txn(txn): + sql = ( + "SELECT user_id FROM open_id_tokens" + " WHERE token = ? AND ? <= ts_valid_until_ms" + ) + + txn.execute(sql, (token, ts_now_ms)) + + rows = txn.fetchall() + if not rows: + return None + else: + return rows[0][0] + + return self.runInteraction("get_user_id_for_token", get_user_id_for_token_txn) diff --git a/synapse/storage/data_stores/main/presence.py b/synapse/storage/data_stores/main/presence.py new file mode 100644 index 0000000000..523ed6575e --- /dev/null +++ b/synapse/storage/data_stores/main/presence.py @@ -0,0 +1,150 @@ +# -*- coding: utf-8 -*- +# Copyright 2014-2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from twisted.internet import defer + +from synapse.storage._base import SQLBaseStore, make_in_list_sql_clause +from synapse.storage.presence import UserPresenceState +from synapse.util import batch_iter +from synapse.util.caches.descriptors import cached, cachedList + + +class PresenceStore(SQLBaseStore): + @defer.inlineCallbacks + def update_presence(self, presence_states): + stream_ordering_manager = self._presence_id_gen.get_next_mult( + len(presence_states) + ) + + with stream_ordering_manager as stream_orderings: + yield self.runInteraction( + "update_presence", + self._update_presence_txn, + stream_orderings, + presence_states, + ) + + return stream_orderings[-1], self._presence_id_gen.get_current_token() + + def _update_presence_txn(self, txn, stream_orderings, presence_states): + for stream_id, state in zip(stream_orderings, presence_states): + txn.call_after( + self.presence_stream_cache.entity_has_changed, state.user_id, stream_id + ) + txn.call_after(self._get_presence_for_user.invalidate, (state.user_id,)) + + # Actually insert new rows + self._simple_insert_many_txn( + txn, + table="presence_stream", + values=[ + { + "stream_id": stream_id, + "user_id": state.user_id, + "state": state.state, + "last_active_ts": state.last_active_ts, + "last_federation_update_ts": state.last_federation_update_ts, + "last_user_sync_ts": state.last_user_sync_ts, + "status_msg": state.status_msg, + "currently_active": state.currently_active, + } + for state in presence_states + ], + ) + + # Delete old rows to stop database from getting really big + sql = "DELETE FROM presence_stream WHERE stream_id < ? AND " + + for states in batch_iter(presence_states, 50): + clause, args = make_in_list_sql_clause( + self.database_engine, "user_id", [s.user_id for s in states] + ) + txn.execute(sql + clause, [stream_id] + list(args)) + + def get_all_presence_updates(self, last_id, current_id): + if last_id == current_id: + return defer.succeed([]) + + def get_all_presence_updates_txn(txn): + sql = ( + "SELECT stream_id, user_id, state, last_active_ts," + " last_federation_update_ts, last_user_sync_ts, status_msg," + " currently_active" + " FROM presence_stream" + " WHERE ? < stream_id AND stream_id <= ?" + ) + txn.execute(sql, (last_id, current_id)) + return txn.fetchall() + + return self.runInteraction( + "get_all_presence_updates", get_all_presence_updates_txn + ) + + @cached() + def _get_presence_for_user(self, user_id): + raise NotImplementedError() + + @cachedList( + cached_method_name="_get_presence_for_user", + list_name="user_ids", + num_args=1, + inlineCallbacks=True, + ) + def get_presence_for_users(self, user_ids): + rows = yield self._simple_select_many_batch( + table="presence_stream", + column="user_id", + iterable=user_ids, + keyvalues={}, + retcols=( + "user_id", + "state", + "last_active_ts", + "last_federation_update_ts", + "last_user_sync_ts", + "status_msg", + "currently_active", + ), + desc="get_presence_for_users", + ) + + for row in rows: + row["currently_active"] = bool(row["currently_active"]) + + return {row["user_id"]: UserPresenceState(**row) for row in rows} + + def get_current_presence_token(self): + return self._presence_id_gen.get_current_token() + + def allow_presence_visible(self, observed_localpart, observer_userid): + return self._simple_insert( + table="presence_allow_inbound", + values={ + "observed_user_id": observed_localpart, + "observer_user_id": observer_userid, + }, + desc="allow_presence_visible", + or_ignore=True, + ) + + def disallow_presence_visible(self, observed_localpart, observer_userid): + return self._simple_delete_one( + table="presence_allow_inbound", + keyvalues={ + "observed_user_id": observed_localpart, + "observer_user_id": observer_userid, + }, + desc="disallow_presence_visible", + ) diff --git a/synapse/storage/data_stores/main/profile.py b/synapse/storage/data_stores/main/profile.py new file mode 100644 index 0000000000..e4e8a1c1d6 --- /dev/null +++ b/synapse/storage/data_stores/main/profile.py @@ -0,0 +1,178 @@ +# -*- coding: utf-8 -*- +# Copyright 2014-2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from twisted.internet import defer + +from synapse.api.errors import StoreError +from synapse.storage._base import SQLBaseStore +from synapse.storage.data_stores.main.roommember import ProfileInfo + + +class ProfileWorkerStore(SQLBaseStore): + @defer.inlineCallbacks + def get_profileinfo(self, user_localpart): + try: + profile = yield self._simple_select_one( + table="profiles", + keyvalues={"user_id": user_localpart}, + retcols=("displayname", "avatar_url"), + desc="get_profileinfo", + ) + except StoreError as e: + if e.code == 404: + # no match + return ProfileInfo(None, None) + else: + raise + + return ProfileInfo( + avatar_url=profile["avatar_url"], display_name=profile["displayname"] + ) + + def get_profile_displayname(self, user_localpart): + return self._simple_select_one_onecol( + table="profiles", + keyvalues={"user_id": user_localpart}, + retcol="displayname", + desc="get_profile_displayname", + ) + + def get_profile_avatar_url(self, user_localpart): + return self._simple_select_one_onecol( + table="profiles", + keyvalues={"user_id": user_localpart}, + retcol="avatar_url", + desc="get_profile_avatar_url", + ) + + def get_from_remote_profile_cache(self, user_id): + return self._simple_select_one( + table="remote_profile_cache", + keyvalues={"user_id": user_id}, + retcols=("displayname", "avatar_url"), + allow_none=True, + desc="get_from_remote_profile_cache", + ) + + def create_profile(self, user_localpart): + return self._simple_insert( + table="profiles", values={"user_id": user_localpart}, desc="create_profile" + ) + + def set_profile_displayname(self, user_localpart, new_displayname): + return self._simple_update_one( + table="profiles", + keyvalues={"user_id": user_localpart}, + updatevalues={"displayname": new_displayname}, + desc="set_profile_displayname", + ) + + def set_profile_avatar_url(self, user_localpart, new_avatar_url): + return self._simple_update_one( + table="profiles", + keyvalues={"user_id": user_localpart}, + updatevalues={"avatar_url": new_avatar_url}, + desc="set_profile_avatar_url", + ) + + +class ProfileStore(ProfileWorkerStore): + def add_remote_profile_cache(self, user_id, displayname, avatar_url): + """Ensure we are caching the remote user's profiles. + + This should only be called when `is_subscribed_remote_profile_for_user` + would return true for the user. + """ + return self._simple_upsert( + table="remote_profile_cache", + keyvalues={"user_id": user_id}, + values={ + "displayname": displayname, + "avatar_url": avatar_url, + "last_check": self._clock.time_msec(), + }, + desc="add_remote_profile_cache", + ) + + def update_remote_profile_cache(self, user_id, displayname, avatar_url): + return self._simple_update( + table="remote_profile_cache", + keyvalues={"user_id": user_id}, + values={ + "displayname": displayname, + "avatar_url": avatar_url, + "last_check": self._clock.time_msec(), + }, + desc="update_remote_profile_cache", + ) + + @defer.inlineCallbacks + def maybe_delete_remote_profile_cache(self, user_id): + """Check if we still care about the remote user's profile, and if we + don't then remove their profile from the cache + """ + subscribed = yield self.is_subscribed_remote_profile_for_user(user_id) + if not subscribed: + yield self._simple_delete( + table="remote_profile_cache", + keyvalues={"user_id": user_id}, + desc="delete_remote_profile_cache", + ) + + def get_remote_profile_cache_entries_that_expire(self, last_checked): + """Get all users who haven't been checked since `last_checked` + """ + + def _get_remote_profile_cache_entries_that_expire_txn(txn): + sql = """ + SELECT user_id, displayname, avatar_url + FROM remote_profile_cache + WHERE last_check < ? + """ + + txn.execute(sql, (last_checked,)) + + return self.cursor_to_dict(txn) + + return self.runInteraction( + "get_remote_profile_cache_entries_that_expire", + _get_remote_profile_cache_entries_that_expire_txn, + ) + + @defer.inlineCallbacks + def is_subscribed_remote_profile_for_user(self, user_id): + """Check whether we are interested in a remote user's profile. + """ + res = yield self._simple_select_one_onecol( + table="group_users", + keyvalues={"user_id": user_id}, + retcol="user_id", + allow_none=True, + desc="should_update_remote_profile_cache_for_user", + ) + + if res: + return True + + res = yield self._simple_select_one_onecol( + table="group_invites", + keyvalues={"user_id": user_id}, + retcol="user_id", + allow_none=True, + desc="should_update_remote_profile_cache_for_user", + ) + + if res: + return True diff --git a/synapse/storage/data_stores/main/push_rule.py b/synapse/storage/data_stores/main/push_rule.py new file mode 100644 index 0000000000..cd95f1ce60 --- /dev/null +++ b/synapse/storage/data_stores/main/push_rule.py @@ -0,0 +1,713 @@ +# -*- coding: utf-8 -*- +# Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import abc +import logging + +from canonicaljson import json + +from twisted.internet import defer + +from synapse.push.baserules import list_with_base_rules +from synapse.storage._base import SQLBaseStore +from synapse.storage.data_stores.main.appservice import ApplicationServiceWorkerStore +from synapse.storage.data_stores.main.pusher import PusherWorkerStore +from synapse.storage.data_stores.main.receipts import ReceiptsWorkerStore +from synapse.storage.data_stores.main.roommember import RoomMemberWorkerStore +from synapse.storage.push_rule import InconsistentRuleException, RuleNotFoundException +from synapse.util.caches.descriptors import cachedInlineCallbacks, cachedList +from synapse.util.caches.stream_change_cache import StreamChangeCache + +logger = logging.getLogger(__name__) + + +def _load_rules(rawrules, enabled_map): + ruleslist = [] + for rawrule in rawrules: + rule = dict(rawrule) + rule["conditions"] = json.loads(rawrule["conditions"]) + rule["actions"] = json.loads(rawrule["actions"]) + ruleslist.append(rule) + + # We're going to be mutating this a lot, so do a deep copy + rules = list(list_with_base_rules(ruleslist)) + + for i, rule in enumerate(rules): + rule_id = rule["rule_id"] + if rule_id in enabled_map: + if rule.get("enabled", True) != bool(enabled_map[rule_id]): + # Rules are cached across users. + rule = dict(rule) + rule["enabled"] = bool(enabled_map[rule_id]) + rules[i] = rule + + return rules + + +class PushRulesWorkerStore( + ApplicationServiceWorkerStore, + ReceiptsWorkerStore, + PusherWorkerStore, + RoomMemberWorkerStore, + SQLBaseStore, +): + """This is an abstract base class where subclasses must implement + `get_max_push_rules_stream_id` which can be called in the initializer. + """ + + # This ABCMeta metaclass ensures that we cannot be instantiated without + # the abstract methods being implemented. + __metaclass__ = abc.ABCMeta + + def __init__(self, db_conn, hs): + super(PushRulesWorkerStore, self).__init__(db_conn, hs) + + push_rules_prefill, push_rules_id = self._get_cache_dict( + db_conn, + "push_rules_stream", + entity_column="user_id", + stream_column="stream_id", + max_value=self.get_max_push_rules_stream_id(), + ) + + self.push_rules_stream_cache = StreamChangeCache( + "PushRulesStreamChangeCache", + push_rules_id, + prefilled_cache=push_rules_prefill, + ) + + @abc.abstractmethod + def get_max_push_rules_stream_id(self): + """Get the position of the push rules stream. + + Returns: + int + """ + raise NotImplementedError() + + @cachedInlineCallbacks(max_entries=5000) + def get_push_rules_for_user(self, user_id): + rows = yield self._simple_select_list( + table="push_rules", + keyvalues={"user_name": user_id}, + retcols=( + "user_name", + "rule_id", + "priority_class", + "priority", + "conditions", + "actions", + ), + desc="get_push_rules_enabled_for_user", + ) + + rows.sort(key=lambda row: (-int(row["priority_class"]), -int(row["priority"]))) + + enabled_map = yield self.get_push_rules_enabled_for_user(user_id) + + rules = _load_rules(rows, enabled_map) + + return rules + + @cachedInlineCallbacks(max_entries=5000) + def get_push_rules_enabled_for_user(self, user_id): + results = yield self._simple_select_list( + table="push_rules_enable", + keyvalues={"user_name": user_id}, + retcols=("user_name", "rule_id", "enabled"), + desc="get_push_rules_enabled_for_user", + ) + return {r["rule_id"]: False if r["enabled"] == 0 else True for r in results} + + def have_push_rules_changed_for_user(self, user_id, last_id): + if not self.push_rules_stream_cache.has_entity_changed(user_id, last_id): + return defer.succeed(False) + else: + + def have_push_rules_changed_txn(txn): + sql = ( + "SELECT COUNT(stream_id) FROM push_rules_stream" + " WHERE user_id = ? AND ? < stream_id" + ) + txn.execute(sql, (user_id, last_id)) + count, = txn.fetchone() + return bool(count) + + return self.runInteraction( + "have_push_rules_changed", have_push_rules_changed_txn + ) + + @cachedList( + cached_method_name="get_push_rules_for_user", + list_name="user_ids", + num_args=1, + inlineCallbacks=True, + ) + def bulk_get_push_rules(self, user_ids): + if not user_ids: + return {} + + results = {user_id: [] for user_id in user_ids} + + rows = yield self._simple_select_many_batch( + table="push_rules", + column="user_name", + iterable=user_ids, + retcols=("*",), + desc="bulk_get_push_rules", + ) + + rows.sort(key=lambda row: (-int(row["priority_class"]), -int(row["priority"]))) + + for row in rows: + results.setdefault(row["user_name"], []).append(row) + + enabled_map_by_user = yield self.bulk_get_push_rules_enabled(user_ids) + + for user_id, rules in results.items(): + results[user_id] = _load_rules(rules, enabled_map_by_user.get(user_id, {})) + + return results + + @defer.inlineCallbacks + def copy_push_rule_from_room_to_room(self, new_room_id, user_id, rule): + """Copy a single push rule from one room to another for a specific user. + + Args: + new_room_id (str): ID of the new room. + user_id (str): ID of user the push rule belongs to. + rule (Dict): A push rule. + """ + # Create new rule id + rule_id_scope = "/".join(rule["rule_id"].split("/")[:-1]) + new_rule_id = rule_id_scope + "/" + new_room_id + + # Change room id in each condition + for condition in rule.get("conditions", []): + if condition.get("key") == "room_id": + condition["pattern"] = new_room_id + + # Add the rule for the new room + yield self.add_push_rule( + user_id=user_id, + rule_id=new_rule_id, + priority_class=rule["priority_class"], + conditions=rule["conditions"], + actions=rule["actions"], + ) + + @defer.inlineCallbacks + def copy_push_rules_from_room_to_room_for_user( + self, old_room_id, new_room_id, user_id + ): + """Copy all of the push rules from one room to another for a specific + user. + + Args: + old_room_id (str): ID of the old room. + new_room_id (str): ID of the new room. + user_id (str): ID of user to copy push rules for. + """ + # Retrieve push rules for this user + user_push_rules = yield self.get_push_rules_for_user(user_id) + + # Get rules relating to the old room and copy them to the new room + for rule in user_push_rules: + conditions = rule.get("conditions", []) + if any( + (c.get("key") == "room_id" and c.get("pattern") == old_room_id) + for c in conditions + ): + yield self.copy_push_rule_from_room_to_room(new_room_id, user_id, rule) + + @defer.inlineCallbacks + def bulk_get_push_rules_for_room(self, event, context): + state_group = context.state_group + if not state_group: + # If state_group is None it means it has yet to be assigned a + # state group, i.e. we need to make sure that calls with a state_group + # of None don't hit previous cached calls with a None state_group. + # To do this we set the state_group to a new object as object() != object() + state_group = object() + + current_state_ids = yield context.get_current_state_ids(self) + result = yield self._bulk_get_push_rules_for_room( + event.room_id, state_group, current_state_ids, event=event + ) + return result + + @cachedInlineCallbacks(num_args=2, cache_context=True) + def _bulk_get_push_rules_for_room( + self, room_id, state_group, current_state_ids, cache_context, event=None + ): + # We don't use `state_group`, its there so that we can cache based + # on it. However, its important that its never None, since two current_state's + # with a state_group of None are likely to be different. + # See bulk_get_push_rules_for_room for how we work around this. + assert state_group is not None + + # We also will want to generate notifs for other people in the room so + # their unread countss are correct in the event stream, but to avoid + # generating them for bot / AS users etc, we only do so for people who've + # sent a read receipt into the room. + + users_in_room = yield self._get_joined_users_from_context( + room_id, + state_group, + current_state_ids, + on_invalidate=cache_context.invalidate, + event=event, + ) + + # We ignore app service users for now. This is so that we don't fill + # up the `get_if_users_have_pushers` cache with AS entries that we + # know don't have pushers, nor even read receipts. + local_users_in_room = set( + u + for u in users_in_room + if self.hs.is_mine_id(u) + and not self.get_if_app_services_interested_in_user(u) + ) + + # users in the room who have pushers need to get push rules run because + # that's how their pushers work + if_users_with_pushers = yield self.get_if_users_have_pushers( + local_users_in_room, on_invalidate=cache_context.invalidate + ) + user_ids = set( + uid for uid, have_pusher in if_users_with_pushers.items() if have_pusher + ) + + users_with_receipts = yield self.get_users_with_read_receipts_in_room( + room_id, on_invalidate=cache_context.invalidate + ) + + # any users with pushers must be ours: they have pushers + for uid in users_with_receipts: + if uid in local_users_in_room: + user_ids.add(uid) + + rules_by_user = yield self.bulk_get_push_rules( + user_ids, on_invalidate=cache_context.invalidate + ) + + rules_by_user = {k: v for k, v in rules_by_user.items() if v is not None} + + return rules_by_user + + @cachedList( + cached_method_name="get_push_rules_enabled_for_user", + list_name="user_ids", + num_args=1, + inlineCallbacks=True, + ) + def bulk_get_push_rules_enabled(self, user_ids): + if not user_ids: + return {} + + results = {user_id: {} for user_id in user_ids} + + rows = yield self._simple_select_many_batch( + table="push_rules_enable", + column="user_name", + iterable=user_ids, + retcols=("user_name", "rule_id", "enabled"), + desc="bulk_get_push_rules_enabled", + ) + for row in rows: + enabled = bool(row["enabled"]) + results.setdefault(row["user_name"], {})[row["rule_id"]] = enabled + return results + + +class PushRuleStore(PushRulesWorkerStore): + @defer.inlineCallbacks + def add_push_rule( + self, + user_id, + rule_id, + priority_class, + conditions, + actions, + before=None, + after=None, + ): + conditions_json = json.dumps(conditions) + actions_json = json.dumps(actions) + with self._push_rules_stream_id_gen.get_next() as ids: + stream_id, event_stream_ordering = ids + if before or after: + yield self.runInteraction( + "_add_push_rule_relative_txn", + self._add_push_rule_relative_txn, + stream_id, + event_stream_ordering, + user_id, + rule_id, + priority_class, + conditions_json, + actions_json, + before, + after, + ) + else: + yield self.runInteraction( + "_add_push_rule_highest_priority_txn", + self._add_push_rule_highest_priority_txn, + stream_id, + event_stream_ordering, + user_id, + rule_id, + priority_class, + conditions_json, + actions_json, + ) + + def _add_push_rule_relative_txn( + self, + txn, + stream_id, + event_stream_ordering, + user_id, + rule_id, + priority_class, + conditions_json, + actions_json, + before, + after, + ): + # Lock the table since otherwise we'll have annoying races between the + # SELECT here and the UPSERT below. + self.database_engine.lock_table(txn, "push_rules") + + relative_to_rule = before or after + + res = self._simple_select_one_txn( + txn, + table="push_rules", + keyvalues={"user_name": user_id, "rule_id": relative_to_rule}, + retcols=["priority_class", "priority"], + allow_none=True, + ) + + if not res: + raise RuleNotFoundException( + "before/after rule not found: %s" % (relative_to_rule,) + ) + + base_priority_class = res["priority_class"] + base_rule_priority = res["priority"] + + if base_priority_class != priority_class: + raise InconsistentRuleException( + "Given priority class does not match class of relative rule" + ) + + if before: + # Higher priority rules are executed first, So adding a rule before + # a rule means giving it a higher priority than that rule. + new_rule_priority = base_rule_priority + 1 + else: + # We increment the priority of the existing rules to make space for + # the new rule. Therefore if we want this rule to appear after + # an existing rule we give it the priority of the existing rule, + # and then increment the priority of the existing rule. + new_rule_priority = base_rule_priority + + sql = ( + "UPDATE push_rules SET priority = priority + 1" + " WHERE user_name = ? AND priority_class = ? AND priority >= ?" + ) + + txn.execute(sql, (user_id, priority_class, new_rule_priority)) + + self._upsert_push_rule_txn( + txn, + stream_id, + event_stream_ordering, + user_id, + rule_id, + priority_class, + new_rule_priority, + conditions_json, + actions_json, + ) + + def _add_push_rule_highest_priority_txn( + self, + txn, + stream_id, + event_stream_ordering, + user_id, + rule_id, + priority_class, + conditions_json, + actions_json, + ): + # Lock the table since otherwise we'll have annoying races between the + # SELECT here and the UPSERT below. + self.database_engine.lock_table(txn, "push_rules") + + # find the highest priority rule in that class + sql = ( + "SELECT COUNT(*), MAX(priority) FROM push_rules" + " WHERE user_name = ? and priority_class = ?" + ) + txn.execute(sql, (user_id, priority_class)) + res = txn.fetchall() + (how_many, highest_prio) = res[0] + + new_prio = 0 + if how_many > 0: + new_prio = highest_prio + 1 + + self._upsert_push_rule_txn( + txn, + stream_id, + event_stream_ordering, + user_id, + rule_id, + priority_class, + new_prio, + conditions_json, + actions_json, + ) + + def _upsert_push_rule_txn( + self, + txn, + stream_id, + event_stream_ordering, + user_id, + rule_id, + priority_class, + priority, + conditions_json, + actions_json, + update_stream=True, + ): + """Specialised version of _simple_upsert_txn that picks a push_rule_id + using the _push_rule_id_gen if it needs to insert the rule. It assumes + that the "push_rules" table is locked""" + + sql = ( + "UPDATE push_rules" + " SET priority_class = ?, priority = ?, conditions = ?, actions = ?" + " WHERE user_name = ? AND rule_id = ?" + ) + + txn.execute( + sql, + (priority_class, priority, conditions_json, actions_json, user_id, rule_id), + ) + + if txn.rowcount == 0: + # We didn't update a row with the given rule_id so insert one + push_rule_id = self._push_rule_id_gen.get_next() + + self._simple_insert_txn( + txn, + table="push_rules", + values={ + "id": push_rule_id, + "user_name": user_id, + "rule_id": rule_id, + "priority_class": priority_class, + "priority": priority, + "conditions": conditions_json, + "actions": actions_json, + }, + ) + + if update_stream: + self._insert_push_rules_update_txn( + txn, + stream_id, + event_stream_ordering, + user_id, + rule_id, + op="ADD", + data={ + "priority_class": priority_class, + "priority": priority, + "conditions": conditions_json, + "actions": actions_json, + }, + ) + + @defer.inlineCallbacks + def delete_push_rule(self, user_id, rule_id): + """ + Delete a push rule. Args specify the row to be deleted and can be + any of the columns in the push_rule table, but below are the + standard ones + + Args: + user_id (str): The matrix ID of the push rule owner + rule_id (str): The rule_id of the rule to be deleted + """ + + def delete_push_rule_txn(txn, stream_id, event_stream_ordering): + self._simple_delete_one_txn( + txn, "push_rules", {"user_name": user_id, "rule_id": rule_id} + ) + + self._insert_push_rules_update_txn( + txn, stream_id, event_stream_ordering, user_id, rule_id, op="DELETE" + ) + + with self._push_rules_stream_id_gen.get_next() as ids: + stream_id, event_stream_ordering = ids + yield self.runInteraction( + "delete_push_rule", + delete_push_rule_txn, + stream_id, + event_stream_ordering, + ) + + @defer.inlineCallbacks + def set_push_rule_enabled(self, user_id, rule_id, enabled): + with self._push_rules_stream_id_gen.get_next() as ids: + stream_id, event_stream_ordering = ids + yield self.runInteraction( + "_set_push_rule_enabled_txn", + self._set_push_rule_enabled_txn, + stream_id, + event_stream_ordering, + user_id, + rule_id, + enabled, + ) + + def _set_push_rule_enabled_txn( + self, txn, stream_id, event_stream_ordering, user_id, rule_id, enabled + ): + new_id = self._push_rules_enable_id_gen.get_next() + self._simple_upsert_txn( + txn, + "push_rules_enable", + {"user_name": user_id, "rule_id": rule_id}, + {"enabled": 1 if enabled else 0}, + {"id": new_id}, + ) + + self._insert_push_rules_update_txn( + txn, + stream_id, + event_stream_ordering, + user_id, + rule_id, + op="ENABLE" if enabled else "DISABLE", + ) + + @defer.inlineCallbacks + def set_push_rule_actions(self, user_id, rule_id, actions, is_default_rule): + actions_json = json.dumps(actions) + + def set_push_rule_actions_txn(txn, stream_id, event_stream_ordering): + if is_default_rule: + # Add a dummy rule to the rules table with the user specified + # actions. + priority_class = -1 + priority = 1 + self._upsert_push_rule_txn( + txn, + stream_id, + event_stream_ordering, + user_id, + rule_id, + priority_class, + priority, + "[]", + actions_json, + update_stream=False, + ) + else: + self._simple_update_one_txn( + txn, + "push_rules", + {"user_name": user_id, "rule_id": rule_id}, + {"actions": actions_json}, + ) + + self._insert_push_rules_update_txn( + txn, + stream_id, + event_stream_ordering, + user_id, + rule_id, + op="ACTIONS", + data={"actions": actions_json}, + ) + + with self._push_rules_stream_id_gen.get_next() as ids: + stream_id, event_stream_ordering = ids + yield self.runInteraction( + "set_push_rule_actions", + set_push_rule_actions_txn, + stream_id, + event_stream_ordering, + ) + + def _insert_push_rules_update_txn( + self, txn, stream_id, event_stream_ordering, user_id, rule_id, op, data=None + ): + values = { + "stream_id": stream_id, + "event_stream_ordering": event_stream_ordering, + "user_id": user_id, + "rule_id": rule_id, + "op": op, + } + if data is not None: + values.update(data) + + self._simple_insert_txn(txn, "push_rules_stream", values=values) + + txn.call_after(self.get_push_rules_for_user.invalidate, (user_id,)) + txn.call_after(self.get_push_rules_enabled_for_user.invalidate, (user_id,)) + txn.call_after( + self.push_rules_stream_cache.entity_has_changed, user_id, stream_id + ) + + def get_all_push_rule_updates(self, last_id, current_id, limit): + """Get all the push rules changes that have happend on the server""" + if last_id == current_id: + return defer.succeed([]) + + def get_all_push_rule_updates_txn(txn): + sql = ( + "SELECT stream_id, event_stream_ordering, user_id, rule_id," + " op, priority_class, priority, conditions, actions" + " FROM push_rules_stream" + " WHERE ? < stream_id AND stream_id <= ?" + " ORDER BY stream_id ASC LIMIT ?" + ) + txn.execute(sql, (last_id, current_id, limit)) + return txn.fetchall() + + return self.runInteraction( + "get_all_push_rule_updates", get_all_push_rule_updates_txn + ) + + def get_push_rules_stream_token(self): + """Get the position of the push rules stream. + Returns a pair of a stream id for the push_rules stream and the + room stream ordering it corresponds to.""" + return self._push_rules_stream_id_gen.get_current_token() + + def get_max_push_rules_stream_id(self): + return self.get_push_rules_stream_token()[0] diff --git a/synapse/storage/data_stores/main/pusher.py b/synapse/storage/data_stores/main/pusher.py new file mode 100644 index 0000000000..f005c1ae0a --- /dev/null +++ b/synapse/storage/data_stores/main/pusher.py @@ -0,0 +1,371 @@ +# -*- coding: utf-8 -*- +# Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging + +import six + +from canonicaljson import encode_canonical_json, json + +from twisted.internet import defer + +from synapse.storage._base import SQLBaseStore +from synapse.util.caches.descriptors import cachedInlineCallbacks, cachedList + +logger = logging.getLogger(__name__) + +if six.PY2: + db_binary_type = six.moves.builtins.buffer +else: + db_binary_type = memoryview + + +class PusherWorkerStore(SQLBaseStore): + def _decode_pushers_rows(self, rows): + for r in rows: + dataJson = r["data"] + r["data"] = None + try: + if isinstance(dataJson, db_binary_type): + dataJson = str(dataJson).decode("UTF8") + + r["data"] = json.loads(dataJson) + except Exception as e: + logger.warn( + "Invalid JSON in data for pusher %d: %s, %s", + r["id"], + dataJson, + e.args[0], + ) + pass + + if isinstance(r["pushkey"], db_binary_type): + r["pushkey"] = str(r["pushkey"]).decode("UTF8") + + return rows + + @defer.inlineCallbacks + def user_has_pusher(self, user_id): + ret = yield self._simple_select_one_onecol( + "pushers", {"user_name": user_id}, "id", allow_none=True + ) + return ret is not None + + def get_pushers_by_app_id_and_pushkey(self, app_id, pushkey): + return self.get_pushers_by({"app_id": app_id, "pushkey": pushkey}) + + def get_pushers_by_user_id(self, user_id): + return self.get_pushers_by({"user_name": user_id}) + + @defer.inlineCallbacks + def get_pushers_by(self, keyvalues): + ret = yield self._simple_select_list( + "pushers", + keyvalues, + [ + "id", + "user_name", + "access_token", + "profile_tag", + "kind", + "app_id", + "app_display_name", + "device_display_name", + "pushkey", + "ts", + "lang", + "data", + "last_stream_ordering", + "last_success", + "failing_since", + ], + desc="get_pushers_by", + ) + return self._decode_pushers_rows(ret) + + @defer.inlineCallbacks + def get_all_pushers(self): + def get_pushers(txn): + txn.execute("SELECT * FROM pushers") + rows = self.cursor_to_dict(txn) + + return self._decode_pushers_rows(rows) + + rows = yield self.runInteraction("get_all_pushers", get_pushers) + return rows + + def get_all_updated_pushers(self, last_id, current_id, limit): + if last_id == current_id: + return defer.succeed(([], [])) + + def get_all_updated_pushers_txn(txn): + sql = ( + "SELECT id, user_name, access_token, profile_tag, kind," + " app_id, app_display_name, device_display_name, pushkey, ts," + " lang, data" + " FROM pushers" + " WHERE ? < id AND id <= ?" + " ORDER BY id ASC LIMIT ?" + ) + txn.execute(sql, (last_id, current_id, limit)) + updated = txn.fetchall() + + sql = ( + "SELECT stream_id, user_id, app_id, pushkey" + " FROM deleted_pushers" + " WHERE ? < stream_id AND stream_id <= ?" + " ORDER BY stream_id ASC LIMIT ?" + ) + txn.execute(sql, (last_id, current_id, limit)) + deleted = txn.fetchall() + + return updated, deleted + + return self.runInteraction( + "get_all_updated_pushers", get_all_updated_pushers_txn + ) + + def get_all_updated_pushers_rows(self, last_id, current_id, limit): + """Get all the pushers that have changed between the given tokens. + + Returns: + Deferred(list(tuple)): each tuple consists of: + stream_id (str) + user_id (str) + app_id (str) + pushkey (str) + was_deleted (bool): whether the pusher was added/updated (False) + or deleted (True) + """ + + if last_id == current_id: + return defer.succeed([]) + + def get_all_updated_pushers_rows_txn(txn): + sql = ( + "SELECT id, user_name, app_id, pushkey" + " FROM pushers" + " WHERE ? < id AND id <= ?" + " ORDER BY id ASC LIMIT ?" + ) + txn.execute(sql, (last_id, current_id, limit)) + results = [list(row) + [False] for row in txn] + + sql = ( + "SELECT stream_id, user_id, app_id, pushkey" + " FROM deleted_pushers" + " WHERE ? < stream_id AND stream_id <= ?" + " ORDER BY stream_id ASC LIMIT ?" + ) + txn.execute(sql, (last_id, current_id, limit)) + + results.extend(list(row) + [True] for row in txn) + results.sort() # Sort so that they're ordered by stream id + + return results + + return self.runInteraction( + "get_all_updated_pushers_rows", get_all_updated_pushers_rows_txn + ) + + @cachedInlineCallbacks(num_args=1, max_entries=15000) + def get_if_user_has_pusher(self, user_id): + # This only exists for the cachedList decorator + raise NotImplementedError() + + @cachedList( + cached_method_name="get_if_user_has_pusher", + list_name="user_ids", + num_args=1, + inlineCallbacks=True, + ) + def get_if_users_have_pushers(self, user_ids): + rows = yield self._simple_select_many_batch( + table="pushers", + column="user_name", + iterable=user_ids, + retcols=["user_name"], + desc="get_if_users_have_pushers", + ) + + result = {user_id: False for user_id in user_ids} + result.update({r["user_name"]: True for r in rows}) + + return result + + +class PusherStore(PusherWorkerStore): + def get_pushers_stream_token(self): + return self._pushers_id_gen.get_current_token() + + @defer.inlineCallbacks + def add_pusher( + self, + user_id, + access_token, + kind, + app_id, + app_display_name, + device_display_name, + pushkey, + pushkey_ts, + lang, + data, + last_stream_ordering, + profile_tag="", + ): + with self._pushers_id_gen.get_next() as stream_id: + # no need to lock because `pushers` has a unique key on + # (app_id, pushkey, user_name) so _simple_upsert will retry + yield self._simple_upsert( + table="pushers", + keyvalues={"app_id": app_id, "pushkey": pushkey, "user_name": user_id}, + values={ + "access_token": access_token, + "kind": kind, + "app_display_name": app_display_name, + "device_display_name": device_display_name, + "ts": pushkey_ts, + "lang": lang, + "data": bytearray(encode_canonical_json(data)), + "last_stream_ordering": last_stream_ordering, + "profile_tag": profile_tag, + "id": stream_id, + }, + desc="add_pusher", + lock=False, + ) + + user_has_pusher = self.get_if_user_has_pusher.cache.get( + (user_id,), None, update_metrics=False + ) + + if user_has_pusher is not True: + # invalidate, since we the user might not have had a pusher before + yield self.runInteraction( + "add_pusher", + self._invalidate_cache_and_stream, + self.get_if_user_has_pusher, + (user_id,), + ) + + @defer.inlineCallbacks + def delete_pusher_by_app_id_pushkey_user_id(self, app_id, pushkey, user_id): + def delete_pusher_txn(txn, stream_id): + self._invalidate_cache_and_stream( + txn, self.get_if_user_has_pusher, (user_id,) + ) + + self._simple_delete_one_txn( + txn, + "pushers", + {"app_id": app_id, "pushkey": pushkey, "user_name": user_id}, + ) + + # it's possible for us to end up with duplicate rows for + # (app_id, pushkey, user_id) at different stream_ids, but that + # doesn't really matter. + self._simple_insert_txn( + txn, + table="deleted_pushers", + values={ + "stream_id": stream_id, + "app_id": app_id, + "pushkey": pushkey, + "user_id": user_id, + }, + ) + + with self._pushers_id_gen.get_next() as stream_id: + yield self.runInteraction("delete_pusher", delete_pusher_txn, stream_id) + + @defer.inlineCallbacks + def update_pusher_last_stream_ordering( + self, app_id, pushkey, user_id, last_stream_ordering + ): + yield self._simple_update_one( + "pushers", + {"app_id": app_id, "pushkey": pushkey, "user_name": user_id}, + {"last_stream_ordering": last_stream_ordering}, + desc="update_pusher_last_stream_ordering", + ) + + @defer.inlineCallbacks + def update_pusher_last_stream_ordering_and_success( + self, app_id, pushkey, user_id, last_stream_ordering, last_success + ): + """Update the last stream ordering position we've processed up to for + the given pusher. + + Args: + app_id (str) + pushkey (str) + last_stream_ordering (int) + last_success (int) + + Returns: + Deferred[bool]: True if the pusher still exists; False if it has been deleted. + """ + updated = yield self._simple_update( + table="pushers", + keyvalues={"app_id": app_id, "pushkey": pushkey, "user_name": user_id}, + updatevalues={ + "last_stream_ordering": last_stream_ordering, + "last_success": last_success, + }, + desc="update_pusher_last_stream_ordering_and_success", + ) + + return bool(updated) + + @defer.inlineCallbacks + def update_pusher_failing_since(self, app_id, pushkey, user_id, failing_since): + yield self._simple_update( + table="pushers", + keyvalues={"app_id": app_id, "pushkey": pushkey, "user_name": user_id}, + updatevalues={"failing_since": failing_since}, + desc="update_pusher_failing_since", + ) + + @defer.inlineCallbacks + def get_throttle_params_by_room(self, pusher_id): + res = yield self._simple_select_list( + "pusher_throttle", + {"pusher": pusher_id}, + ["room_id", "last_sent_ts", "throttle_ms"], + desc="get_throttle_params_by_room", + ) + + params_by_room = {} + for row in res: + params_by_room[row["room_id"]] = { + "last_sent_ts": row["last_sent_ts"], + "throttle_ms": row["throttle_ms"], + } + + return params_by_room + + @defer.inlineCallbacks + def set_throttle_params(self, pusher_id, room_id, params): + # no need to lock because `pusher_throttle` has a primary key on + # (pusher, room_id) so _simple_upsert will retry + yield self._simple_upsert( + "pusher_throttle", + {"pusher": pusher_id, "room_id": room_id}, + params, + desc="set_throttle_params", + lock=False, + ) diff --git a/synapse/storage/data_stores/main/receipts.py b/synapse/storage/data_stores/main/receipts.py new file mode 100644 index 0000000000..0c24430f28 --- /dev/null +++ b/synapse/storage/data_stores/main/receipts.py @@ -0,0 +1,536 @@ +# -*- coding: utf-8 -*- +# Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import abc +import logging + +from canonicaljson import json + +from twisted.internet import defer + +from synapse.storage._base import SQLBaseStore, make_in_list_sql_clause +from synapse.storage.util.id_generators import StreamIdGenerator +from synapse.util.caches.descriptors import cached, cachedInlineCallbacks, cachedList +from synapse.util.caches.stream_change_cache import StreamChangeCache + +logger = logging.getLogger(__name__) + + +class ReceiptsWorkerStore(SQLBaseStore): + """This is an abstract base class where subclasses must implement + `get_max_receipt_stream_id` which can be called in the initializer. + """ + + # This ABCMeta metaclass ensures that we cannot be instantiated without + # the abstract methods being implemented. + __metaclass__ = abc.ABCMeta + + def __init__(self, db_conn, hs): + super(ReceiptsWorkerStore, self).__init__(db_conn, hs) + + self._receipts_stream_cache = StreamChangeCache( + "ReceiptsRoomChangeCache", self.get_max_receipt_stream_id() + ) + + @abc.abstractmethod + def get_max_receipt_stream_id(self): + """Get the current max stream ID for receipts stream + + Returns: + int + """ + raise NotImplementedError() + + @cachedInlineCallbacks() + def get_users_with_read_receipts_in_room(self, room_id): + receipts = yield self.get_receipts_for_room(room_id, "m.read") + return set(r["user_id"] for r in receipts) + + @cached(num_args=2) + def get_receipts_for_room(self, room_id, receipt_type): + return self._simple_select_list( + table="receipts_linearized", + keyvalues={"room_id": room_id, "receipt_type": receipt_type}, + retcols=("user_id", "event_id"), + desc="get_receipts_for_room", + ) + + @cached(num_args=3) + def get_last_receipt_event_id_for_user(self, user_id, room_id, receipt_type): + return self._simple_select_one_onecol( + table="receipts_linearized", + keyvalues={ + "room_id": room_id, + "receipt_type": receipt_type, + "user_id": user_id, + }, + retcol="event_id", + desc="get_own_receipt_for_user", + allow_none=True, + ) + + @cachedInlineCallbacks(num_args=2) + def get_receipts_for_user(self, user_id, receipt_type): + rows = yield self._simple_select_list( + table="receipts_linearized", + keyvalues={"user_id": user_id, "receipt_type": receipt_type}, + retcols=("room_id", "event_id"), + desc="get_receipts_for_user", + ) + + return {row["room_id"]: row["event_id"] for row in rows} + + @defer.inlineCallbacks + def get_receipts_for_user_with_orderings(self, user_id, receipt_type): + def f(txn): + sql = ( + "SELECT rl.room_id, rl.event_id," + " e.topological_ordering, e.stream_ordering" + " FROM receipts_linearized AS rl" + " INNER JOIN events AS e USING (room_id, event_id)" + " WHERE rl.room_id = e.room_id" + " AND rl.event_id = e.event_id" + " AND user_id = ?" + ) + txn.execute(sql, (user_id,)) + return txn.fetchall() + + rows = yield self.runInteraction("get_receipts_for_user_with_orderings", f) + return { + row[0]: { + "event_id": row[1], + "topological_ordering": row[2], + "stream_ordering": row[3], + } + for row in rows + } + + @defer.inlineCallbacks + def get_linearized_receipts_for_rooms(self, room_ids, to_key, from_key=None): + """Get receipts for multiple rooms for sending to clients. + + Args: + room_ids (list): List of room_ids. + to_key (int): Max stream id to fetch receipts upto. + from_key (int): Min stream id to fetch receipts from. None fetches + from the start. + + Returns: + list: A list of receipts. + """ + room_ids = set(room_ids) + + if from_key is not None: + # Only ask the database about rooms where there have been new + # receipts added since `from_key` + room_ids = yield self._receipts_stream_cache.get_entities_changed( + room_ids, from_key + ) + + results = yield self._get_linearized_receipts_for_rooms( + room_ids, to_key, from_key=from_key + ) + + return [ev for res in results.values() for ev in res] + + def get_linearized_receipts_for_room(self, room_id, to_key, from_key=None): + """Get receipts for a single room for sending to clients. + + Args: + room_ids (str): The room id. + to_key (int): Max stream id to fetch receipts upto. + from_key (int): Min stream id to fetch receipts from. None fetches + from the start. + + Returns: + Deferred[list]: A list of receipts. + """ + if from_key is not None: + # Check the cache first to see if any new receipts have been added + # since`from_key`. If not we can no-op. + if not self._receipts_stream_cache.has_entity_changed(room_id, from_key): + defer.succeed([]) + + return self._get_linearized_receipts_for_room(room_id, to_key, from_key) + + @cachedInlineCallbacks(num_args=3, tree=True) + def _get_linearized_receipts_for_room(self, room_id, to_key, from_key=None): + """See get_linearized_receipts_for_room + """ + + def f(txn): + if from_key: + sql = ( + "SELECT * FROM receipts_linearized WHERE" + " room_id = ? AND stream_id > ? AND stream_id <= ?" + ) + + txn.execute(sql, (room_id, from_key, to_key)) + else: + sql = ( + "SELECT * FROM receipts_linearized WHERE" + " room_id = ? AND stream_id <= ?" + ) + + txn.execute(sql, (room_id, to_key)) + + rows = self.cursor_to_dict(txn) + + return rows + + rows = yield self.runInteraction("get_linearized_receipts_for_room", f) + + if not rows: + return [] + + content = {} + for row in rows: + content.setdefault(row["event_id"], {}).setdefault(row["receipt_type"], {})[ + row["user_id"] + ] = json.loads(row["data"]) + + return [{"type": "m.receipt", "room_id": room_id, "content": content}] + + @cachedList( + cached_method_name="_get_linearized_receipts_for_room", + list_name="room_ids", + num_args=3, + inlineCallbacks=True, + ) + def _get_linearized_receipts_for_rooms(self, room_ids, to_key, from_key=None): + if not room_ids: + return {} + + def f(txn): + if from_key: + sql = """ + SELECT * FROM receipts_linearized WHERE + stream_id > ? AND stream_id <= ? AND + """ + clause, args = make_in_list_sql_clause( + self.database_engine, "room_id", room_ids + ) + + txn.execute(sql + clause, [from_key, to_key] + list(args)) + else: + sql = """ + SELECT * FROM receipts_linearized WHERE + stream_id <= ? AND + """ + + clause, args = make_in_list_sql_clause( + self.database_engine, "room_id", room_ids + ) + + txn.execute(sql + clause, [to_key] + list(args)) + + return self.cursor_to_dict(txn) + + txn_results = yield self.runInteraction("_get_linearized_receipts_for_rooms", f) + + results = {} + for row in txn_results: + # We want a single event per room, since we want to batch the + # receipts by room, event and type. + room_event = results.setdefault( + row["room_id"], + {"type": "m.receipt", "room_id": row["room_id"], "content": {}}, + ) + + # The content is of the form: + # {"$foo:bar": { "read": { "@user:host": }, .. }, .. } + event_entry = room_event["content"].setdefault(row["event_id"], {}) + receipt_type = event_entry.setdefault(row["receipt_type"], {}) + + receipt_type[row["user_id"]] = json.loads(row["data"]) + + results = { + room_id: [results[room_id]] if room_id in results else [] + for room_id in room_ids + } + return results + + def get_all_updated_receipts(self, last_id, current_id, limit=None): + if last_id == current_id: + return defer.succeed([]) + + def get_all_updated_receipts_txn(txn): + sql = ( + "SELECT stream_id, room_id, receipt_type, user_id, event_id, data" + " FROM receipts_linearized" + " WHERE ? < stream_id AND stream_id <= ?" + " ORDER BY stream_id ASC" + ) + args = [last_id, current_id] + if limit is not None: + sql += " LIMIT ?" + args.append(limit) + txn.execute(sql, args) + + return (r[0:5] + (json.loads(r[5]),) for r in txn) + + return self.runInteraction( + "get_all_updated_receipts", get_all_updated_receipts_txn + ) + + def _invalidate_get_users_with_receipts_in_room( + self, room_id, receipt_type, user_id + ): + if receipt_type != "m.read": + return + + # Returns either an ObservableDeferred or the raw result + res = self.get_users_with_read_receipts_in_room.cache.get( + room_id, None, update_metrics=False + ) + + # first handle the Deferred case + if isinstance(res, defer.Deferred): + if res.called: + res = res.result + else: + res = None + + if res and user_id in res: + # We'd only be adding to the set, so no point invalidating if the + # user is already there + return + + self.get_users_with_read_receipts_in_room.invalidate((room_id,)) + + +class ReceiptsStore(ReceiptsWorkerStore): + def __init__(self, db_conn, hs): + # We instantiate this first as the ReceiptsWorkerStore constructor + # needs to be able to call get_max_receipt_stream_id + self._receipts_id_gen = StreamIdGenerator( + db_conn, "receipts_linearized", "stream_id" + ) + + super(ReceiptsStore, self).__init__(db_conn, hs) + + def get_max_receipt_stream_id(self): + return self._receipts_id_gen.get_current_token() + + def insert_linearized_receipt_txn( + self, txn, room_id, receipt_type, user_id, event_id, data, stream_id + ): + """Inserts a read-receipt into the database if it's newer than the current RR + + Returns: int|None + None if the RR is older than the current RR + otherwise, the rx timestamp of the event that the RR corresponds to + (or 0 if the event is unknown) + """ + res = self._simple_select_one_txn( + txn, + table="events", + retcols=["stream_ordering", "received_ts"], + keyvalues={"event_id": event_id}, + allow_none=True, + ) + + stream_ordering = int(res["stream_ordering"]) if res else None + rx_ts = res["received_ts"] if res else 0 + + # We don't want to clobber receipts for more recent events, so we + # have to compare orderings of existing receipts + if stream_ordering is not None: + sql = ( + "SELECT stream_ordering, event_id FROM events" + " INNER JOIN receipts_linearized as r USING (event_id, room_id)" + " WHERE r.room_id = ? AND r.receipt_type = ? AND r.user_id = ?" + ) + txn.execute(sql, (room_id, receipt_type, user_id)) + + for so, eid in txn: + if int(so) >= stream_ordering: + logger.debug( + "Ignoring new receipt for %s in favour of existing " + "one for later event %s", + event_id, + eid, + ) + return None + + txn.call_after(self.get_receipts_for_room.invalidate, (room_id, receipt_type)) + txn.call_after( + self._invalidate_get_users_with_receipts_in_room, + room_id, + receipt_type, + user_id, + ) + txn.call_after(self.get_receipts_for_user.invalidate, (user_id, receipt_type)) + # FIXME: This shouldn't invalidate the whole cache + txn.call_after( + self._get_linearized_receipts_for_room.invalidate_many, (room_id,) + ) + + txn.call_after( + self._receipts_stream_cache.entity_has_changed, room_id, stream_id + ) + + txn.call_after( + self.get_last_receipt_event_id_for_user.invalidate, + (user_id, room_id, receipt_type), + ) + + self._simple_delete_txn( + txn, + table="receipts_linearized", + keyvalues={ + "room_id": room_id, + "receipt_type": receipt_type, + "user_id": user_id, + }, + ) + + self._simple_insert_txn( + txn, + table="receipts_linearized", + values={ + "stream_id": stream_id, + "room_id": room_id, + "receipt_type": receipt_type, + "user_id": user_id, + "event_id": event_id, + "data": json.dumps(data), + }, + ) + + if receipt_type == "m.read" and stream_ordering is not None: + self._remove_old_push_actions_before_txn( + txn, room_id=room_id, user_id=user_id, stream_ordering=stream_ordering + ) + + return rx_ts + + @defer.inlineCallbacks + def insert_receipt(self, room_id, receipt_type, user_id, event_ids, data): + """Insert a receipt, either from local client or remote server. + + Automatically does conversion between linearized and graph + representations. + """ + if not event_ids: + return + + if len(event_ids) == 1: + linearized_event_id = event_ids[0] + else: + # we need to points in graph -> linearized form. + # TODO: Make this better. + def graph_to_linear(txn): + clause, args = make_in_list_sql_clause( + self.database_engine, "event_id", event_ids + ) + + sql = """ + SELECT event_id WHERE room_id = ? AND stream_ordering IN ( + SELECT max(stream_ordering) WHERE %s + ) + """ % ( + clause, + ) + + txn.execute(sql, [room_id] + list(args)) + rows = txn.fetchall() + if rows: + return rows[0][0] + else: + raise RuntimeError("Unrecognized event_ids: %r" % (event_ids,)) + + linearized_event_id = yield self.runInteraction( + "insert_receipt_conv", graph_to_linear + ) + + stream_id_manager = self._receipts_id_gen.get_next() + with stream_id_manager as stream_id: + event_ts = yield self.runInteraction( + "insert_linearized_receipt", + self.insert_linearized_receipt_txn, + room_id, + receipt_type, + user_id, + linearized_event_id, + data, + stream_id=stream_id, + ) + + if event_ts is None: + return None + + now = self._clock.time_msec() + logger.debug( + "RR for event %s in %s (%i ms old)", + linearized_event_id, + room_id, + now - event_ts, + ) + + yield self.insert_graph_receipt(room_id, receipt_type, user_id, event_ids, data) + + max_persisted_id = self._receipts_id_gen.get_current_token() + + return stream_id, max_persisted_id + + def insert_graph_receipt(self, room_id, receipt_type, user_id, event_ids, data): + return self.runInteraction( + "insert_graph_receipt", + self.insert_graph_receipt_txn, + room_id, + receipt_type, + user_id, + event_ids, + data, + ) + + def insert_graph_receipt_txn( + self, txn, room_id, receipt_type, user_id, event_ids, data + ): + txn.call_after(self.get_receipts_for_room.invalidate, (room_id, receipt_type)) + txn.call_after( + self._invalidate_get_users_with_receipts_in_room, + room_id, + receipt_type, + user_id, + ) + txn.call_after(self.get_receipts_for_user.invalidate, (user_id, receipt_type)) + # FIXME: This shouldn't invalidate the whole cache + txn.call_after( + self._get_linearized_receipts_for_room.invalidate_many, (room_id,) + ) + + self._simple_delete_txn( + txn, + table="receipts_graph", + keyvalues={ + "room_id": room_id, + "receipt_type": receipt_type, + "user_id": user_id, + }, + ) + self._simple_insert_txn( + txn, + table="receipts_graph", + values={ + "room_id": room_id, + "receipt_type": receipt_type, + "user_id": user_id, + "event_ids": json.dumps(event_ids), + "data": json.dumps(data), + }, + ) diff --git a/synapse/storage/data_stores/main/registration.py b/synapse/storage/data_stores/main/registration.py new file mode 100644 index 0000000000..6c5b29288a --- /dev/null +++ b/synapse/storage/data_stores/main/registration.py @@ -0,0 +1,1499 @@ +# -*- coding: utf-8 -*- +# Copyright 2014-2016 OpenMarket Ltd +# Copyright 2017-2018 New Vector Ltd +# Copyright 2019 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging +import re + +from six import iterkeys +from six.moves import range + +from twisted.internet import defer +from twisted.internet.defer import Deferred + +from synapse.api.constants import UserTypes +from synapse.api.errors import Codes, StoreError, SynapseError, ThreepidValidationError +from synapse.metrics.background_process_metrics import run_as_background_process +from synapse.storage import background_updates +from synapse.storage._base import SQLBaseStore +from synapse.types import UserID +from synapse.util.caches.descriptors import cached, cachedInlineCallbacks + +THIRTY_MINUTES_IN_MS = 30 * 60 * 1000 + +logger = logging.getLogger(__name__) + + +class RegistrationWorkerStore(SQLBaseStore): + def __init__(self, db_conn, hs): + super(RegistrationWorkerStore, self).__init__(db_conn, hs) + + self.config = hs.config + self.clock = hs.get_clock() + + @cached() + def get_user_by_id(self, user_id): + return self._simple_select_one( + table="users", + keyvalues={"name": user_id}, + retcols=[ + "name", + "password_hash", + "is_guest", + "consent_version", + "consent_server_notice_sent", + "appservice_id", + "creation_ts", + "user_type", + ], + allow_none=True, + desc="get_user_by_id", + ) + + @defer.inlineCallbacks + def is_trial_user(self, user_id): + """Checks if user is in the "trial" period, i.e. within the first + N days of registration defined by `mau_trial_days` config + + Args: + user_id (str) + + Returns: + Deferred[bool] + """ + + info = yield self.get_user_by_id(user_id) + if not info: + return False + + now = self.clock.time_msec() + trial_duration_ms = self.config.mau_trial_days * 24 * 60 * 60 * 1000 + is_trial = (now - info["creation_ts"] * 1000) < trial_duration_ms + return is_trial + + @cached() + def get_user_by_access_token(self, token): + """Get a user from the given access token. + + Args: + token (str): The access token of a user. + Returns: + defer.Deferred: None, if the token did not match, otherwise dict + including the keys `name`, `is_guest`, `device_id`, `token_id`, + `valid_until_ms`. + """ + return self.runInteraction( + "get_user_by_access_token", self._query_for_auth, token + ) + + @cachedInlineCallbacks() + def get_expiration_ts_for_user(self, user_id): + """Get the expiration timestamp for the account bearing a given user ID. + + Args: + user_id (str): The ID of the user. + Returns: + defer.Deferred: None, if the account has no expiration timestamp, + otherwise int representation of the timestamp (as a number of + milliseconds since epoch). + """ + res = yield self._simple_select_one_onecol( + table="account_validity", + keyvalues={"user_id": user_id}, + retcol="expiration_ts_ms", + allow_none=True, + desc="get_expiration_ts_for_user", + ) + return res + + @defer.inlineCallbacks + def set_account_validity_for_user( + self, user_id, expiration_ts, email_sent, renewal_token=None + ): + """Updates the account validity properties of the given account, with the + given values. + + Args: + user_id (str): ID of the account to update properties for. + expiration_ts (int): New expiration date, as a timestamp in milliseconds + since epoch. + email_sent (bool): True means a renewal email has been sent for this + account and there's no need to send another one for the current validity + period. + renewal_token (str): Renewal token the user can use to extend the validity + of their account. Defaults to no token. + """ + + def set_account_validity_for_user_txn(txn): + self._simple_update_txn( + txn=txn, + table="account_validity", + keyvalues={"user_id": user_id}, + updatevalues={ + "expiration_ts_ms": expiration_ts, + "email_sent": email_sent, + "renewal_token": renewal_token, + }, + ) + self._invalidate_cache_and_stream( + txn, self.get_expiration_ts_for_user, (user_id,) + ) + + yield self.runInteraction( + "set_account_validity_for_user", set_account_validity_for_user_txn + ) + + @defer.inlineCallbacks + def set_renewal_token_for_user(self, user_id, renewal_token): + """Defines a renewal token for a given user. + + Args: + user_id (str): ID of the user to set the renewal token for. + renewal_token (str): Random unique string that will be used to renew the + user's account. + + Raises: + StoreError: The provided token is already set for another user. + """ + yield self._simple_update_one( + table="account_validity", + keyvalues={"user_id": user_id}, + updatevalues={"renewal_token": renewal_token}, + desc="set_renewal_token_for_user", + ) + + @defer.inlineCallbacks + def get_user_from_renewal_token(self, renewal_token): + """Get a user ID from a renewal token. + + Args: + renewal_token (str): The renewal token to perform the lookup with. + + Returns: + defer.Deferred[str]: The ID of the user to which the token belongs. + """ + res = yield self._simple_select_one_onecol( + table="account_validity", + keyvalues={"renewal_token": renewal_token}, + retcol="user_id", + desc="get_user_from_renewal_token", + ) + + return res + + @defer.inlineCallbacks + def get_renewal_token_for_user(self, user_id): + """Get the renewal token associated with a given user ID. + + Args: + user_id (str): The user ID to lookup a token for. + + Returns: + defer.Deferred[str]: The renewal token associated with this user ID. + """ + res = yield self._simple_select_one_onecol( + table="account_validity", + keyvalues={"user_id": user_id}, + retcol="renewal_token", + desc="get_renewal_token_for_user", + ) + + return res + + @defer.inlineCallbacks + def get_users_expiring_soon(self): + """Selects users whose account will expire in the [now, now + renew_at] time + window (see configuration for account_validity for information on what renew_at + refers to). + + Returns: + Deferred: Resolves to a list[dict[user_id (str), expiration_ts_ms (int)]] + """ + + def select_users_txn(txn, now_ms, renew_at): + sql = ( + "SELECT user_id, expiration_ts_ms FROM account_validity" + " WHERE email_sent = ? AND (expiration_ts_ms - ?) <= ?" + ) + values = [False, now_ms, renew_at] + txn.execute(sql, values) + return self.cursor_to_dict(txn) + + res = yield self.runInteraction( + "get_users_expiring_soon", + select_users_txn, + self.clock.time_msec(), + self.config.account_validity.renew_at, + ) + + return res + + @defer.inlineCallbacks + def set_renewal_mail_status(self, user_id, email_sent): + """Sets or unsets the flag that indicates whether a renewal email has been sent + to the user (and the user hasn't renewed their account yet). + + Args: + user_id (str): ID of the user to set/unset the flag for. + email_sent (bool): Flag which indicates whether a renewal email has been sent + to this user. + """ + yield self._simple_update_one( + table="account_validity", + keyvalues={"user_id": user_id}, + updatevalues={"email_sent": email_sent}, + desc="set_renewal_mail_status", + ) + + @defer.inlineCallbacks + def delete_account_validity_for_user(self, user_id): + """Deletes the entry for the given user in the account validity table, removing + their expiration date and renewal token. + + Args: + user_id (str): ID of the user to remove from the account validity table. + """ + yield self._simple_delete_one( + table="account_validity", + keyvalues={"user_id": user_id}, + desc="delete_account_validity_for_user", + ) + + @defer.inlineCallbacks + def is_server_admin(self, user): + """Determines if a user is an admin of this homeserver. + + Args: + user (UserID): user ID of the user to test + + Returns (bool): + true iff the user is a server admin, false otherwise. + """ + res = yield self._simple_select_one_onecol( + table="users", + keyvalues={"name": user.to_string()}, + retcol="admin", + allow_none=True, + desc="is_server_admin", + ) + + return res if res else False + + def set_server_admin(self, user, admin): + """Sets whether a user is an admin of this homeserver. + + Args: + user (UserID): user ID of the user to test + admin (bool): true iff the user is to be a server admin, + false otherwise. + """ + return self._simple_update_one( + table="users", + keyvalues={"name": user.to_string()}, + updatevalues={"admin": 1 if admin else 0}, + desc="set_server_admin", + ) + + def _query_for_auth(self, txn, token): + sql = ( + "SELECT users.name, users.is_guest, access_tokens.id as token_id," + " access_tokens.device_id, access_tokens.valid_until_ms" + " FROM users" + " INNER JOIN access_tokens on users.name = access_tokens.user_id" + " WHERE token = ?" + ) + + txn.execute(sql, (token,)) + rows = self.cursor_to_dict(txn) + if rows: + return rows[0] + + return None + + @cachedInlineCallbacks() + def is_real_user(self, user_id): + """Determines if the user is a real user, ie does not have a 'user_type'. + + Args: + user_id (str): user id to test + + Returns: + Deferred[bool]: True if user 'user_type' is null or empty string + """ + res = yield self.runInteraction("is_real_user", self.is_real_user_txn, user_id) + return res + + @cachedInlineCallbacks() + def is_support_user(self, user_id): + """Determines if the user is of type UserTypes.SUPPORT + + Args: + user_id (str): user id to test + + Returns: + Deferred[bool]: True if user is of type UserTypes.SUPPORT + """ + res = yield self.runInteraction( + "is_support_user", self.is_support_user_txn, user_id + ) + return res + + def is_real_user_txn(self, txn, user_id): + res = self._simple_select_one_onecol_txn( + txn=txn, + table="users", + keyvalues={"name": user_id}, + retcol="user_type", + allow_none=True, + ) + return res is None + + def is_support_user_txn(self, txn, user_id): + res = self._simple_select_one_onecol_txn( + txn=txn, + table="users", + keyvalues={"name": user_id}, + retcol="user_type", + allow_none=True, + ) + return True if res == UserTypes.SUPPORT else False + + def get_users_by_id_case_insensitive(self, user_id): + """Gets users that match user_id case insensitively. + Returns a mapping of user_id -> password_hash. + """ + + def f(txn): + sql = ( + "SELECT name, password_hash FROM users" " WHERE lower(name) = lower(?)" + ) + txn.execute(sql, (user_id,)) + return dict(txn) + + return self.runInteraction("get_users_by_id_case_insensitive", f) + + async def get_user_by_external_id( + self, auth_provider: str, external_id: str + ) -> str: + """Look up a user by their external auth id + + Args: + auth_provider: identifier for the remote auth provider + external_id: id on that system + + Returns: + str|None: the mxid of the user, or None if they are not known + """ + return await self._simple_select_one_onecol( + table="user_external_ids", + keyvalues={"auth_provider": auth_provider, "external_id": external_id}, + retcol="user_id", + allow_none=True, + desc="get_user_by_external_id", + ) + + @defer.inlineCallbacks + def count_all_users(self): + """Counts all users registered on the homeserver.""" + + def _count_users(txn): + txn.execute("SELECT COUNT(*) AS users FROM users") + rows = self.cursor_to_dict(txn) + if rows: + return rows[0]["users"] + return 0 + + ret = yield self.runInteraction("count_users", _count_users) + return ret + + def count_daily_user_type(self): + """ + Counts 1) native non guest users + 2) native guests users + 3) bridged users + who registered on the homeserver in the past 24 hours + """ + + def _count_daily_user_type(txn): + yesterday = int(self._clock.time()) - (60 * 60 * 24) + + sql = """ + SELECT user_type, COALESCE(count(*), 0) AS count FROM ( + SELECT + CASE + WHEN is_guest=0 AND appservice_id IS NULL THEN 'native' + WHEN is_guest=1 AND appservice_id IS NULL THEN 'guest' + WHEN is_guest=0 AND appservice_id IS NOT NULL THEN 'bridged' + END AS user_type + FROM users + WHERE creation_ts > ? + ) AS t GROUP BY user_type + """ + results = {"native": 0, "guest": 0, "bridged": 0} + txn.execute(sql, (yesterday,)) + for row in txn: + results[row[0]] = row[1] + return results + + return self.runInteraction("count_daily_user_type", _count_daily_user_type) + + @defer.inlineCallbacks + def count_nonbridged_users(self): + def _count_users(txn): + txn.execute( + """ + SELECT COALESCE(COUNT(*), 0) FROM users + WHERE appservice_id IS NULL + """ + ) + count, = txn.fetchone() + return count + + ret = yield self.runInteraction("count_users", _count_users) + return ret + + @defer.inlineCallbacks + def count_real_users(self): + """Counts all users without a special user_type registered on the homeserver.""" + + def _count_users(txn): + txn.execute("SELECT COUNT(*) AS users FROM users where user_type is null") + rows = self.cursor_to_dict(txn) + if rows: + return rows[0]["users"] + return 0 + + ret = yield self.runInteraction("count_real_users", _count_users) + return ret + + @defer.inlineCallbacks + def find_next_generated_user_id_localpart(self): + """ + Gets the localpart of the next generated user ID. + + Generated user IDs are integers, and we aim for them to be as small as + we can. Unfortunately, it's possible some of them are already taken by + existing users, and there may be gaps in the already taken range. This + function returns the start of the first allocatable gap. This is to + avoid the case of ID 10000000 being pre-allocated, so us wasting the + first (and shortest) many generated user IDs. + """ + + def _find_next_generated_user_id(txn): + # We bound between '@1' and '@a' to avoid pulling the entire table + # out. + txn.execute("SELECT name FROM users WHERE '@1' <= name AND name < '@a'") + + regex = re.compile(r"^@(\d+):") + + found = set() + + for (user_id,) in txn: + match = regex.search(user_id) + if match: + found.add(int(match.group(1))) + for i in range(len(found) + 1): + if i not in found: + return i + + return ( + ( + yield self.runInteraction( + "find_next_generated_user_id", _find_next_generated_user_id + ) + ) + ) + + @defer.inlineCallbacks + def get_user_id_by_threepid(self, medium, address): + """Returns user id from threepid + + Args: + medium (str): threepid medium e.g. email + address (str): threepid address e.g. me@example.com + + Returns: + Deferred[str|None]: user id or None if no user id/threepid mapping exists + """ + user_id = yield self.runInteraction( + "get_user_id_by_threepid", self.get_user_id_by_threepid_txn, medium, address + ) + return user_id + + def get_user_id_by_threepid_txn(self, txn, medium, address): + """Returns user id from threepid + + Args: + txn (cursor): + medium (str): threepid medium e.g. email + address (str): threepid address e.g. me@example.com + + Returns: + str|None: user id or None if no user id/threepid mapping exists + """ + ret = self._simple_select_one_txn( + txn, + "user_threepids", + {"medium": medium, "address": address}, + ["user_id"], + True, + ) + if ret: + return ret["user_id"] + return None + + @defer.inlineCallbacks + def user_add_threepid(self, user_id, medium, address, validated_at, added_at): + yield self._simple_upsert( + "user_threepids", + {"medium": medium, "address": address}, + {"user_id": user_id, "validated_at": validated_at, "added_at": added_at}, + ) + + @defer.inlineCallbacks + def user_get_threepids(self, user_id): + ret = yield self._simple_select_list( + "user_threepids", + {"user_id": user_id}, + ["medium", "address", "validated_at", "added_at"], + "user_get_threepids", + ) + return ret + + def user_delete_threepid(self, user_id, medium, address): + return self._simple_delete( + "user_threepids", + keyvalues={"user_id": user_id, "medium": medium, "address": address}, + desc="user_delete_threepids", + ) + + def add_user_bound_threepid(self, user_id, medium, address, id_server): + """The server proxied a bind request to the given identity server on + behalf of the given user. We need to remember this in case the user + asks us to unbind the threepid. + + Args: + user_id (str) + medium (str) + address (str) + id_server (str) + + Returns: + Deferred + """ + # We need to use an upsert, in case they user had already bound the + # threepid + return self._simple_upsert( + table="user_threepid_id_server", + keyvalues={ + "user_id": user_id, + "medium": medium, + "address": address, + "id_server": id_server, + }, + values={}, + insertion_values={}, + desc="add_user_bound_threepid", + ) + + def user_get_bound_threepids(self, user_id): + """Get the threepids that a user has bound to an identity server through the homeserver + The homeserver remembers where binds to an identity server occurred. Using this + method can retrieve those threepids. + + Args: + user_id (str): The ID of the user to retrieve threepids for + + Returns: + Deferred[list[dict]]: List of dictionaries containing the following: + medium (str): The medium of the threepid (e.g "email") + address (str): The address of the threepid (e.g "bob@example.com") + """ + return self._simple_select_list( + table="user_threepid_id_server", + keyvalues={"user_id": user_id}, + retcols=["medium", "address"], + desc="user_get_bound_threepids", + ) + + def remove_user_bound_threepid(self, user_id, medium, address, id_server): + """The server proxied an unbind request to the given identity server on + behalf of the given user, so we remove the mapping of threepid to + identity server. + + Args: + user_id (str) + medium (str) + address (str) + id_server (str) + + Returns: + Deferred + """ + return self._simple_delete( + table="user_threepid_id_server", + keyvalues={ + "user_id": user_id, + "medium": medium, + "address": address, + "id_server": id_server, + }, + desc="remove_user_bound_threepid", + ) + + def get_id_servers_user_bound(self, user_id, medium, address): + """Get the list of identity servers that the server proxied bind + requests to for given user and threepid + + Args: + user_id (str) + medium (str) + address (str) + + Returns: + Deferred[list[str]]: Resolves to a list of identity servers + """ + return self._simple_select_onecol( + table="user_threepid_id_server", + keyvalues={"user_id": user_id, "medium": medium, "address": address}, + retcol="id_server", + desc="get_id_servers_user_bound", + ) + + @cachedInlineCallbacks() + def get_user_deactivated_status(self, user_id): + """Retrieve the value for the `deactivated` property for the provided user. + + Args: + user_id (str): The ID of the user to retrieve the status for. + + Returns: + defer.Deferred(bool): The requested value. + """ + + res = yield self._simple_select_one_onecol( + table="users", + keyvalues={"name": user_id}, + retcol="deactivated", + desc="get_user_deactivated_status", + ) + + # Convert the integer into a boolean. + return res == 1 + + def get_threepid_validation_session( + self, medium, client_secret, address=None, sid=None, validated=True + ): + """Gets a session_id and last_send_attempt (if available) for a + combination of validation metadata + + Args: + medium (str|None): The medium of the 3PID + address (str|None): The address of the 3PID + sid (str|None): The ID of the validation session + client_secret (str): A unique string provided by the client to help identify this + validation attempt + validated (bool|None): Whether sessions should be filtered by + whether they have been validated already or not. None to + perform no filtering + + Returns: + Deferred[dict|None]: A dict containing the following: + * address - address of the 3pid + * medium - medium of the 3pid + * client_secret - a secret provided by the client for this validation session + * session_id - ID of the validation session + * send_attempt - a number serving to dedupe send attempts for this session + * validated_at - timestamp of when this session was validated if so + + Otherwise None if a validation session is not found + """ + if not client_secret: + raise SynapseError( + 400, "Missing parameter: client_secret", errcode=Codes.MISSING_PARAM + ) + + keyvalues = {"client_secret": client_secret} + if medium: + keyvalues["medium"] = medium + if address: + keyvalues["address"] = address + if sid: + keyvalues["session_id"] = sid + + assert address or sid + + def get_threepid_validation_session_txn(txn): + sql = """ + SELECT address, session_id, medium, client_secret, + last_send_attempt, validated_at + FROM threepid_validation_session WHERE %s + """ % ( + " AND ".join("%s = ?" % k for k in iterkeys(keyvalues)), + ) + + if validated is not None: + sql += " AND validated_at IS " + ("NOT NULL" if validated else "NULL") + + sql += " LIMIT 1" + + txn.execute(sql, list(keyvalues.values())) + rows = self.cursor_to_dict(txn) + if not rows: + return None + + return rows[0] + + return self.runInteraction( + "get_threepid_validation_session", get_threepid_validation_session_txn + ) + + def delete_threepid_session(self, session_id): + """Removes a threepid validation session from the database. This can + be done after validation has been performed and whatever action was + waiting on it has been carried out + + Args: + session_id (str): The ID of the session to delete + """ + + def delete_threepid_session_txn(txn): + self._simple_delete_txn( + txn, + table="threepid_validation_token", + keyvalues={"session_id": session_id}, + ) + self._simple_delete_txn( + txn, + table="threepid_validation_session", + keyvalues={"session_id": session_id}, + ) + + return self.runInteraction( + "delete_threepid_session", delete_threepid_session_txn + ) + + +class RegistrationBackgroundUpdateStore( + RegistrationWorkerStore, background_updates.BackgroundUpdateStore +): + def __init__(self, db_conn, hs): + super(RegistrationBackgroundUpdateStore, self).__init__(db_conn, hs) + + self.clock = hs.get_clock() + self.config = hs.config + + self.register_background_index_update( + "access_tokens_device_index", + index_name="access_tokens_device_id", + table="access_tokens", + columns=["user_id", "device_id"], + ) + + self.register_background_index_update( + "users_creation_ts", + index_name="users_creation_ts", + table="users", + columns=["creation_ts"], + ) + + # we no longer use refresh tokens, but it's possible that some people + # might have a background update queued to build this index. Just + # clear the background update. + self.register_noop_background_update("refresh_tokens_device_index") + + self.register_background_update_handler( + "user_threepids_grandfather", self._bg_user_threepids_grandfather + ) + + self.register_background_update_handler( + "users_set_deactivated_flag", self._background_update_set_deactivated_flag + ) + + @defer.inlineCallbacks + def _background_update_set_deactivated_flag(self, progress, batch_size): + """Retrieves a list of all deactivated users and sets the 'deactivated' flag to 1 + for each of them. + """ + + last_user = progress.get("user_id", "") + + def _background_update_set_deactivated_flag_txn(txn): + txn.execute( + """ + SELECT + users.name, + COUNT(access_tokens.token) AS count_tokens, + COUNT(user_threepids.address) AS count_threepids + FROM users + LEFT JOIN access_tokens ON (access_tokens.user_id = users.name) + LEFT JOIN user_threepids ON (user_threepids.user_id = users.name) + WHERE (users.password_hash IS NULL OR users.password_hash = '') + AND (users.appservice_id IS NULL OR users.appservice_id = '') + AND users.is_guest = 0 + AND users.name > ? + GROUP BY users.name + ORDER BY users.name ASC + LIMIT ?; + """, + (last_user, batch_size), + ) + + rows = self.cursor_to_dict(txn) + + if not rows: + return True, 0 + + rows_processed_nb = 0 + + for user in rows: + if not user["count_tokens"] and not user["count_threepids"]: + self.set_user_deactivated_status_txn(txn, user["name"], True) + rows_processed_nb += 1 + + logger.info("Marked %d rows as deactivated", rows_processed_nb) + + self._background_update_progress_txn( + txn, "users_set_deactivated_flag", {"user_id": rows[-1]["name"]} + ) + + if batch_size > len(rows): + return True, len(rows) + else: + return False, len(rows) + + end, nb_processed = yield self.runInteraction( + "users_set_deactivated_flag", _background_update_set_deactivated_flag_txn + ) + + if end: + yield self._end_background_update("users_set_deactivated_flag") + + return nb_processed + + @defer.inlineCallbacks + def _bg_user_threepids_grandfather(self, progress, batch_size): + """We now track which identity servers a user binds their 3PID to, so + we need to handle the case of existing bindings where we didn't track + this. + + We do this by grandfathering in existing user threepids assuming that + they used one of the server configured trusted identity servers. + """ + id_servers = set(self.config.trusted_third_party_id_servers) + + def _bg_user_threepids_grandfather_txn(txn): + sql = """ + INSERT INTO user_threepid_id_server + (user_id, medium, address, id_server) + SELECT user_id, medium, address, ? + FROM user_threepids + """ + + txn.executemany(sql, [(id_server,) for id_server in id_servers]) + + if id_servers: + yield self.runInteraction( + "_bg_user_threepids_grandfather", _bg_user_threepids_grandfather_txn + ) + + yield self._end_background_update("user_threepids_grandfather") + + return 1 + + +class RegistrationStore(RegistrationBackgroundUpdateStore): + def __init__(self, db_conn, hs): + super(RegistrationStore, self).__init__(db_conn, hs) + + self._account_validity = hs.config.account_validity + + # Create a background job for culling expired 3PID validity tokens + def start_cull(): + # run as a background process to make sure that the database transactions + # have a logcontext to report to + return run_as_background_process( + "cull_expired_threepid_validation_tokens", + self.cull_expired_threepid_validation_tokens, + ) + + hs.get_clock().looping_call(start_cull, THIRTY_MINUTES_IN_MS) + + @defer.inlineCallbacks + def add_access_token_to_user(self, user_id, token, device_id, valid_until_ms): + """Adds an access token for the given user. + + Args: + user_id (str): The user ID. + token (str): The new access token to add. + device_id (str): ID of the device to associate with the access + token + valid_until_ms (int|None): when the token is valid until. None for + no expiry. + Raises: + StoreError if there was a problem adding this. + """ + next_id = self._access_tokens_id_gen.get_next() + + yield self._simple_insert( + "access_tokens", + { + "id": next_id, + "user_id": user_id, + "token": token, + "device_id": device_id, + "valid_until_ms": valid_until_ms, + }, + desc="add_access_token_to_user", + ) + + def register_user( + self, + user_id, + password_hash=None, + was_guest=False, + make_guest=False, + appservice_id=None, + create_profile_with_displayname=None, + admin=False, + user_type=None, + ): + """Attempts to register an account. + + Args: + user_id (str): The desired user ID to register. + password_hash (str): Optional. The password hash for this user. + was_guest (bool): Optional. Whether this is a guest account being + upgraded to a non-guest account. + make_guest (boolean): True if the the new user should be guest, + false to add a regular user account. + appservice_id (str): The ID of the appservice registering the user. + create_profile_with_displayname (unicode): Optionally create a profile for + the user, setting their displayname to the given value + admin (boolean): is an admin user? + user_type (str|None): type of user. One of the values from + api.constants.UserTypes, or None for a normal user. + + Raises: + StoreError if the user_id could not be registered. + """ + return self.runInteraction( + "register_user", + self._register_user, + user_id, + password_hash, + was_guest, + make_guest, + appservice_id, + create_profile_with_displayname, + admin, + user_type, + ) + + def _register_user( + self, + txn, + user_id, + password_hash, + was_guest, + make_guest, + appservice_id, + create_profile_with_displayname, + admin, + user_type, + ): + user_id_obj = UserID.from_string(user_id) + + now = int(self.clock.time()) + + try: + if was_guest: + # Ensure that the guest user actually exists + # ``allow_none=False`` makes this raise an exception + # if the row isn't in the database. + self._simple_select_one_txn( + txn, + "users", + keyvalues={"name": user_id, "is_guest": 1}, + retcols=("name",), + allow_none=False, + ) + + self._simple_update_one_txn( + txn, + "users", + keyvalues={"name": user_id, "is_guest": 1}, + updatevalues={ + "password_hash": password_hash, + "upgrade_ts": now, + "is_guest": 1 if make_guest else 0, + "appservice_id": appservice_id, + "admin": 1 if admin else 0, + "user_type": user_type, + }, + ) + else: + self._simple_insert_txn( + txn, + "users", + values={ + "name": user_id, + "password_hash": password_hash, + "creation_ts": now, + "is_guest": 1 if make_guest else 0, + "appservice_id": appservice_id, + "admin": 1 if admin else 0, + "user_type": user_type, + }, + ) + + except self.database_engine.module.IntegrityError: + raise StoreError(400, "User ID already taken.", errcode=Codes.USER_IN_USE) + + if self._account_validity.enabled: + self.set_expiration_date_for_user_txn(txn, user_id) + + if create_profile_with_displayname: + # set a default displayname serverside to avoid ugly race + # between auto-joins and clients trying to set displaynames + # + # *obviously* the 'profiles' table uses localpart for user_id + # while everything else uses the full mxid. + txn.execute( + "INSERT INTO profiles(user_id, displayname) VALUES (?,?)", + (user_id_obj.localpart, create_profile_with_displayname), + ) + + if self.hs.config.stats_enabled: + # we create a new completed user statistics row + + # we don't strictly need current_token since this user really can't + # have any state deltas before now (as it is a new user), but still, + # we include it for completeness. + current_token = self._get_max_stream_id_in_current_state_deltas_txn(txn) + self._update_stats_delta_txn( + txn, now, "user", user_id, {}, complete_with_stream_id=current_token + ) + + self._invalidate_cache_and_stream(txn, self.get_user_by_id, (user_id,)) + txn.call_after(self.is_guest.invalidate, (user_id,)) + + def record_user_external_id( + self, auth_provider: str, external_id: str, user_id: str + ) -> Deferred: + """Record a mapping from an external user id to a mxid + + Args: + auth_provider: identifier for the remote auth provider + external_id: id on that system + user_id: complete mxid that it is mapped to + """ + return self._simple_insert( + table="user_external_ids", + values={ + "auth_provider": auth_provider, + "external_id": external_id, + "user_id": user_id, + }, + desc="record_user_external_id", + ) + + def user_set_password_hash(self, user_id, password_hash): + """ + NB. This does *not* evict any cache because the one use for this + removes most of the entries subsequently anyway so it would be + pointless. Use flush_user separately. + """ + + def user_set_password_hash_txn(txn): + self._simple_update_one_txn( + txn, "users", {"name": user_id}, {"password_hash": password_hash} + ) + self._invalidate_cache_and_stream(txn, self.get_user_by_id, (user_id,)) + + return self.runInteraction("user_set_password_hash", user_set_password_hash_txn) + + def user_set_consent_version(self, user_id, consent_version): + """Updates the user table to record privacy policy consent + + Args: + user_id (str): full mxid of the user to update + consent_version (str): version of the policy the user has consented + to + + Raises: + StoreError(404) if user not found + """ + + def f(txn): + self._simple_update_one_txn( + txn, + table="users", + keyvalues={"name": user_id}, + updatevalues={"consent_version": consent_version}, + ) + self._invalidate_cache_and_stream(txn, self.get_user_by_id, (user_id,)) + + return self.runInteraction("user_set_consent_version", f) + + def user_set_consent_server_notice_sent(self, user_id, consent_version): + """Updates the user table to record that we have sent the user a server + notice about privacy policy consent + + Args: + user_id (str): full mxid of the user to update + consent_version (str): version of the policy we have notified the + user about + + Raises: + StoreError(404) if user not found + """ + + def f(txn): + self._simple_update_one_txn( + txn, + table="users", + keyvalues={"name": user_id}, + updatevalues={"consent_server_notice_sent": consent_version}, + ) + self._invalidate_cache_and_stream(txn, self.get_user_by_id, (user_id,)) + + return self.runInteraction("user_set_consent_server_notice_sent", f) + + def user_delete_access_tokens(self, user_id, except_token_id=None, device_id=None): + """ + Invalidate access tokens belonging to a user + + Args: + user_id (str): ID of user the tokens belong to + except_token_id (str): list of access_tokens IDs which should + *not* be deleted + device_id (str|None): ID of device the tokens are associated with. + If None, tokens associated with any device (or no device) will + be deleted + Returns: + defer.Deferred[list[str, int, str|None, int]]: a list of + (token, token id, device id) for each of the deleted tokens + """ + + def f(txn): + keyvalues = {"user_id": user_id} + if device_id is not None: + keyvalues["device_id"] = device_id + + items = keyvalues.items() + where_clause = " AND ".join(k + " = ?" for k, _ in items) + values = [v for _, v in items] + if except_token_id: + where_clause += " AND id != ?" + values.append(except_token_id) + + txn.execute( + "SELECT token, id, device_id FROM access_tokens WHERE %s" + % where_clause, + values, + ) + tokens_and_devices = [(r[0], r[1], r[2]) for r in txn] + + for token, _, _ in tokens_and_devices: + self._invalidate_cache_and_stream( + txn, self.get_user_by_access_token, (token,) + ) + + txn.execute("DELETE FROM access_tokens WHERE %s" % where_clause, values) + + return tokens_and_devices + + return self.runInteraction("user_delete_access_tokens", f) + + def delete_access_token(self, access_token): + def f(txn): + self._simple_delete_one_txn( + txn, table="access_tokens", keyvalues={"token": access_token} + ) + + self._invalidate_cache_and_stream( + txn, self.get_user_by_access_token, (access_token,) + ) + + return self.runInteraction("delete_access_token", f) + + @cachedInlineCallbacks() + def is_guest(self, user_id): + res = yield self._simple_select_one_onecol( + table="users", + keyvalues={"name": user_id}, + retcol="is_guest", + allow_none=True, + desc="is_guest", + ) + + return res if res else False + + def add_user_pending_deactivation(self, user_id): + """ + Adds a user to the table of users who need to be parted from all the rooms they're + in + """ + return self._simple_insert( + "users_pending_deactivation", + values={"user_id": user_id}, + desc="add_user_pending_deactivation", + ) + + def del_user_pending_deactivation(self, user_id): + """ + Removes the given user to the table of users who need to be parted from all the + rooms they're in, effectively marking that user as fully deactivated. + """ + # XXX: This should be simple_delete_one but we failed to put a unique index on + # the table, so somehow duplicate entries have ended up in it. + return self._simple_delete( + "users_pending_deactivation", + keyvalues={"user_id": user_id}, + desc="del_user_pending_deactivation", + ) + + def get_user_pending_deactivation(self): + """ + Gets one user from the table of users waiting to be parted from all the rooms + they're in. + """ + return self._simple_select_one_onecol( + "users_pending_deactivation", + keyvalues={}, + retcol="user_id", + allow_none=True, + desc="get_users_pending_deactivation", + ) + + def validate_threepid_session(self, session_id, client_secret, token, current_ts): + """Attempt to validate a threepid session using a token + + Args: + session_id (str): The id of a validation session + client_secret (str): A unique string provided by the client to + help identify this validation attempt + token (str): A validation token + current_ts (int): The current unix time in milliseconds. Used for + checking token expiry status + + Raises: + ThreepidValidationError: if a matching validation token was not found or has + expired + + Returns: + deferred str|None: A str representing a link to redirect the user + to if there is one. + """ + + # Insert everything into a transaction in order to run atomically + def validate_threepid_session_txn(txn): + row = self._simple_select_one_txn( + txn, + table="threepid_validation_session", + keyvalues={"session_id": session_id}, + retcols=["client_secret", "validated_at"], + allow_none=True, + ) + + if not row: + raise ThreepidValidationError(400, "Unknown session_id") + retrieved_client_secret = row["client_secret"] + validated_at = row["validated_at"] + + if retrieved_client_secret != client_secret: + raise ThreepidValidationError( + 400, "This client_secret does not match the provided session_id" + ) + + row = self._simple_select_one_txn( + txn, + table="threepid_validation_token", + keyvalues={"session_id": session_id, "token": token}, + retcols=["expires", "next_link"], + allow_none=True, + ) + + if not row: + raise ThreepidValidationError( + 400, "Validation token not found or has expired" + ) + expires = row["expires"] + next_link = row["next_link"] + + # If the session is already validated, no need to revalidate + if validated_at: + return next_link + + if expires <= current_ts: + raise ThreepidValidationError( + 400, "This token has expired. Please request a new one" + ) + + # Looks good. Validate the session + self._simple_update_txn( + txn, + table="threepid_validation_session", + keyvalues={"session_id": session_id}, + updatevalues={"validated_at": self.clock.time_msec()}, + ) + + return next_link + + # Return next_link if it exists + return self.runInteraction( + "validate_threepid_session_txn", validate_threepid_session_txn + ) + + def upsert_threepid_validation_session( + self, + medium, + address, + client_secret, + send_attempt, + session_id, + validated_at=None, + ): + """Upsert a threepid validation session + Args: + medium (str): The medium of the 3PID + address (str): The address of the 3PID + client_secret (str): A unique string provided by the client to + help identify this validation attempt + send_attempt (int): The latest send_attempt on this session + session_id (str): The id of this validation session + validated_at (int|None): The unix timestamp in milliseconds of + when the session was marked as valid + """ + insertion_values = { + "medium": medium, + "address": address, + "client_secret": client_secret, + } + + if validated_at: + insertion_values["validated_at"] = validated_at + + return self._simple_upsert( + table="threepid_validation_session", + keyvalues={"session_id": session_id}, + values={"last_send_attempt": send_attempt}, + insertion_values=insertion_values, + desc="upsert_threepid_validation_session", + ) + + def start_or_continue_validation_session( + self, + medium, + address, + session_id, + client_secret, + send_attempt, + next_link, + token, + token_expires, + ): + """Creates a new threepid validation session if it does not already + exist and associates a new validation token with it + + Args: + medium (str): The medium of the 3PID + address (str): The address of the 3PID + session_id (str): The id of this validation session + client_secret (str): A unique string provided by the client to + help identify this validation attempt + send_attempt (int): The latest send_attempt on this session + next_link (str|None): The link to redirect the user to upon + successful validation + token (str): The validation token + token_expires (int): The timestamp for which after the token + will no longer be valid + """ + + def start_or_continue_validation_session_txn(txn): + # Create or update a validation session + self._simple_upsert_txn( + txn, + table="threepid_validation_session", + keyvalues={"session_id": session_id}, + values={"last_send_attempt": send_attempt}, + insertion_values={ + "medium": medium, + "address": address, + "client_secret": client_secret, + }, + ) + + # Create a new validation token with this session ID + self._simple_insert_txn( + txn, + table="threepid_validation_token", + values={ + "session_id": session_id, + "token": token, + "next_link": next_link, + "expires": token_expires, + }, + ) + + return self.runInteraction( + "start_or_continue_validation_session", + start_or_continue_validation_session_txn, + ) + + def cull_expired_threepid_validation_tokens(self): + """Remove threepid validation tokens with expiry dates that have passed""" + + def cull_expired_threepid_validation_tokens_txn(txn, ts): + sql = """ + DELETE FROM threepid_validation_token WHERE + expires < ? + """ + return txn.execute(sql, (ts,)) + + return self.runInteraction( + "cull_expired_threepid_validation_tokens", + cull_expired_threepid_validation_tokens_txn, + self.clock.time_msec(), + ) + + @defer.inlineCallbacks + def set_user_deactivated_status(self, user_id, deactivated): + """Set the `deactivated` property for the provided user to the provided value. + + Args: + user_id (str): The ID of the user to set the status for. + deactivated (bool): The value to set for `deactivated`. + """ + + yield self.runInteraction( + "set_user_deactivated_status", + self.set_user_deactivated_status_txn, + user_id, + deactivated, + ) + + def set_user_deactivated_status_txn(self, txn, user_id, deactivated): + self._simple_update_one_txn( + txn=txn, + table="users", + keyvalues={"name": user_id}, + updatevalues={"deactivated": 1 if deactivated else 0}, + ) + self._invalidate_cache_and_stream( + txn, self.get_user_deactivated_status, (user_id,) + ) diff --git a/synapse/storage/data_stores/main/rejections.py b/synapse/storage/data_stores/main/rejections.py new file mode 100644 index 0000000000..7d5de0ea2e --- /dev/null +++ b/synapse/storage/data_stores/main/rejections.py @@ -0,0 +1,42 @@ +# -*- coding: utf-8 -*- +# Copyright 2014-2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging + +from synapse.storage._base import SQLBaseStore + +logger = logging.getLogger(__name__) + + +class RejectionsStore(SQLBaseStore): + def _store_rejections_txn(self, txn, event_id, reason): + self._simple_insert_txn( + txn, + table="rejections", + values={ + "event_id": event_id, + "reason": reason, + "last_check": self._clock.time_msec(), + }, + ) + + def get_rejection_reason(self, event_id): + return self._simple_select_one_onecol( + table="rejections", + retcol="reason", + keyvalues={"event_id": event_id}, + allow_none=True, + desc="get_rejection_reason", + ) diff --git a/synapse/storage/data_stores/main/relations.py b/synapse/storage/data_stores/main/relations.py new file mode 100644 index 0000000000..858f65582b --- /dev/null +++ b/synapse/storage/data_stores/main/relations.py @@ -0,0 +1,385 @@ +# -*- coding: utf-8 -*- +# Copyright 2019 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging + +import attr + +from synapse.api.constants import RelationTypes +from synapse.storage._base import SQLBaseStore +from synapse.storage.data_stores.main.stream import generate_pagination_where_clause +from synapse.storage.relations import ( + AggregationPaginationToken, + PaginationChunk, + RelationPaginationToken, +) +from synapse.util.caches.descriptors import cached, cachedInlineCallbacks + +logger = logging.getLogger(__name__) + + +class RelationsWorkerStore(SQLBaseStore): + @cached(tree=True) + def get_relations_for_event( + self, + event_id, + relation_type=None, + event_type=None, + aggregation_key=None, + limit=5, + direction="b", + from_token=None, + to_token=None, + ): + """Get a list of relations for an event, ordered by topological ordering. + + Args: + event_id (str): Fetch events that relate to this event ID. + relation_type (str|None): Only fetch events with this relation + type, if given. + event_type (str|None): Only fetch events with this event type, if + given. + aggregation_key (str|None): Only fetch events with this aggregation + key, if given. + limit (int): Only fetch the most recent `limit` events. + direction (str): Whether to fetch the most recent first (`"b"`) or + the oldest first (`"f"`). + from_token (RelationPaginationToken|None): Fetch rows from the given + token, or from the start if None. + to_token (RelationPaginationToken|None): Fetch rows up to the given + token, or up to the end if None. + + Returns: + Deferred[PaginationChunk]: List of event IDs that match relations + requested. The rows are of the form `{"event_id": "..."}`. + """ + + where_clause = ["relates_to_id = ?"] + where_args = [event_id] + + if relation_type is not None: + where_clause.append("relation_type = ?") + where_args.append(relation_type) + + if event_type is not None: + where_clause.append("type = ?") + where_args.append(event_type) + + if aggregation_key: + where_clause.append("aggregation_key = ?") + where_args.append(aggregation_key) + + pagination_clause = generate_pagination_where_clause( + direction=direction, + column_names=("topological_ordering", "stream_ordering"), + from_token=attr.astuple(from_token) if from_token else None, + to_token=attr.astuple(to_token) if to_token else None, + engine=self.database_engine, + ) + + if pagination_clause: + where_clause.append(pagination_clause) + + if direction == "b": + order = "DESC" + else: + order = "ASC" + + sql = """ + SELECT event_id, topological_ordering, stream_ordering + FROM event_relations + INNER JOIN events USING (event_id) + WHERE %s + ORDER BY topological_ordering %s, stream_ordering %s + LIMIT ? + """ % ( + " AND ".join(where_clause), + order, + order, + ) + + def _get_recent_references_for_event_txn(txn): + txn.execute(sql, where_args + [limit + 1]) + + last_topo_id = None + last_stream_id = None + events = [] + for row in txn: + events.append({"event_id": row[0]}) + last_topo_id = row[1] + last_stream_id = row[2] + + next_batch = None + if len(events) > limit and last_topo_id and last_stream_id: + next_batch = RelationPaginationToken(last_topo_id, last_stream_id) + + return PaginationChunk( + chunk=list(events[:limit]), next_batch=next_batch, prev_batch=from_token + ) + + return self.runInteraction( + "get_recent_references_for_event", _get_recent_references_for_event_txn + ) + + @cached(tree=True) + def get_aggregation_groups_for_event( + self, + event_id, + event_type=None, + limit=5, + direction="b", + from_token=None, + to_token=None, + ): + """Get a list of annotations on the event, grouped by event type and + aggregation key, sorted by count. + + This is used e.g. to get the what and how many reactions have happend + on an event. + + Args: + event_id (str): Fetch events that relate to this event ID. + event_type (str|None): Only fetch events with this event type, if + given. + limit (int): Only fetch the `limit` groups. + direction (str): Whether to fetch the highest count first (`"b"`) or + the lowest count first (`"f"`). + from_token (AggregationPaginationToken|None): Fetch rows from the + given token, or from the start if None. + to_token (AggregationPaginationToken|None): Fetch rows up to the + given token, or up to the end if None. + + + Returns: + Deferred[PaginationChunk]: List of groups of annotations that + match. Each row is a dict with `type`, `key` and `count` fields. + """ + + where_clause = ["relates_to_id = ?", "relation_type = ?"] + where_args = [event_id, RelationTypes.ANNOTATION] + + if event_type: + where_clause.append("type = ?") + where_args.append(event_type) + + having_clause = generate_pagination_where_clause( + direction=direction, + column_names=("COUNT(*)", "MAX(stream_ordering)"), + from_token=attr.astuple(from_token) if from_token else None, + to_token=attr.astuple(to_token) if to_token else None, + engine=self.database_engine, + ) + + if direction == "b": + order = "DESC" + else: + order = "ASC" + + if having_clause: + having_clause = "HAVING " + having_clause + else: + having_clause = "" + + sql = """ + SELECT type, aggregation_key, COUNT(DISTINCT sender), MAX(stream_ordering) + FROM event_relations + INNER JOIN events USING (event_id) + WHERE {where_clause} + GROUP BY relation_type, type, aggregation_key + {having_clause} + ORDER BY COUNT(*) {order}, MAX(stream_ordering) {order} + LIMIT ? + """.format( + where_clause=" AND ".join(where_clause), + order=order, + having_clause=having_clause, + ) + + def _get_aggregation_groups_for_event_txn(txn): + txn.execute(sql, where_args + [limit + 1]) + + next_batch = None + events = [] + for row in txn: + events.append({"type": row[0], "key": row[1], "count": row[2]}) + next_batch = AggregationPaginationToken(row[2], row[3]) + + if len(events) <= limit: + next_batch = None + + return PaginationChunk( + chunk=list(events[:limit]), next_batch=next_batch, prev_batch=from_token + ) + + return self.runInteraction( + "get_aggregation_groups_for_event", _get_aggregation_groups_for_event_txn + ) + + @cachedInlineCallbacks() + def get_applicable_edit(self, event_id): + """Get the most recent edit (if any) that has happened for the given + event. + + Correctly handles checking whether edits were allowed to happen. + + Args: + event_id (str): The original event ID + + Returns: + Deferred[EventBase|None]: Returns the most recent edit, if any. + """ + + # We only allow edits for `m.room.message` events that have the same sender + # and event type. We can't assert these things during regular event auth so + # we have to do the checks post hoc. + + # Fetches latest edit that has the same type and sender as the + # original, and is an `m.room.message`. + sql = """ + SELECT edit.event_id FROM events AS edit + INNER JOIN event_relations USING (event_id) + INNER JOIN events AS original ON + original.event_id = relates_to_id + AND edit.type = original.type + AND edit.sender = original.sender + WHERE + relates_to_id = ? + AND relation_type = ? + AND edit.type = 'm.room.message' + ORDER by edit.origin_server_ts DESC, edit.event_id DESC + LIMIT 1 + """ + + def _get_applicable_edit_txn(txn): + txn.execute(sql, (event_id, RelationTypes.REPLACE)) + row = txn.fetchone() + if row: + return row[0] + + edit_id = yield self.runInteraction( + "get_applicable_edit", _get_applicable_edit_txn + ) + + if not edit_id: + return + + edit_event = yield self.get_event(edit_id, allow_none=True) + return edit_event + + def has_user_annotated_event(self, parent_id, event_type, aggregation_key, sender): + """Check if a user has already annotated an event with the same key + (e.g. already liked an event). + + Args: + parent_id (str): The event being annotated + event_type (str): The event type of the annotation + aggregation_key (str): The aggregation key of the annotation + sender (str): The sender of the annotation + + Returns: + Deferred[bool] + """ + + sql = """ + SELECT 1 FROM event_relations + INNER JOIN events USING (event_id) + WHERE + relates_to_id = ? + AND relation_type = ? + AND type = ? + AND sender = ? + AND aggregation_key = ? + LIMIT 1; + """ + + def _get_if_user_has_annotated_event(txn): + txn.execute( + sql, + ( + parent_id, + RelationTypes.ANNOTATION, + event_type, + sender, + aggregation_key, + ), + ) + + return bool(txn.fetchone()) + + return self.runInteraction( + "get_if_user_has_annotated_event", _get_if_user_has_annotated_event + ) + + +class RelationsStore(RelationsWorkerStore): + def _handle_event_relations(self, txn, event): + """Handles inserting relation data during peristence of events + + Args: + txn + event (EventBase) + """ + relation = event.content.get("m.relates_to") + if not relation: + # No relations + return + + rel_type = relation.get("rel_type") + if rel_type not in ( + RelationTypes.ANNOTATION, + RelationTypes.REFERENCE, + RelationTypes.REPLACE, + ): + # Unknown relation type + return + + parent_id = relation.get("event_id") + if not parent_id: + # Invalid relation + return + + aggregation_key = relation.get("key") + + self._simple_insert_txn( + txn, + table="event_relations", + values={ + "event_id": event.event_id, + "relates_to_id": parent_id, + "relation_type": rel_type, + "aggregation_key": aggregation_key, + }, + ) + + txn.call_after(self.get_relations_for_event.invalidate_many, (parent_id,)) + txn.call_after( + self.get_aggregation_groups_for_event.invalidate_many, (parent_id,) + ) + + if rel_type == RelationTypes.REPLACE: + txn.call_after(self.get_applicable_edit.invalidate, (parent_id,)) + + def _handle_redaction(self, txn, redacted_event_id): + """Handles receiving a redaction and checking whether we need to remove + any redacted relations from the database. + + Args: + txn + redacted_event_id (str): The event that was redacted. + """ + + self._simple_delete_txn( + txn, table="event_relations", keyvalues={"event_id": redacted_event_id} + ) diff --git a/synapse/storage/data_stores/main/room.py b/synapse/storage/data_stores/main/room.py new file mode 100644 index 0000000000..4428e5c55d --- /dev/null +++ b/synapse/storage/data_stores/main/room.py @@ -0,0 +1,681 @@ +# -*- coding: utf-8 -*- +# Copyright 2014-2016 OpenMarket Ltd +# Copyright 2019 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import collections +import logging +import re +from typing import Optional, Tuple + +from canonicaljson import json + +from twisted.internet import defer + +from synapse.api.errors import StoreError +from synapse.storage._base import SQLBaseStore +from synapse.storage.data_stores.main.search import SearchStore +from synapse.types import ThirdPartyInstanceID +from synapse.util.caches.descriptors import cached, cachedInlineCallbacks + +logger = logging.getLogger(__name__) + + +OpsLevel = collections.namedtuple( + "OpsLevel", ("ban_level", "kick_level", "redact_level") +) + +RatelimitOverride = collections.namedtuple( + "RatelimitOverride", ("messages_per_second", "burst_count") +) + + +class RoomWorkerStore(SQLBaseStore): + def get_room(self, room_id): + """Retrieve a room. + + Args: + room_id (str): The ID of the room to retrieve. + Returns: + A dict containing the room information, or None if the room is unknown. + """ + return self._simple_select_one( + table="rooms", + keyvalues={"room_id": room_id}, + retcols=("room_id", "is_public", "creator"), + desc="get_room", + allow_none=True, + ) + + def get_public_room_ids(self): + return self._simple_select_onecol( + table="rooms", + keyvalues={"is_public": True}, + retcol="room_id", + desc="get_public_room_ids", + ) + + def count_public_rooms(self, network_tuple, ignore_non_federatable): + """Counts the number of public rooms as tracked in the room_stats_current + and room_stats_state table. + + Args: + network_tuple (ThirdPartyInstanceID|None) + ignore_non_federatable (bool): If true filters out non-federatable rooms + """ + + def _count_public_rooms_txn(txn): + query_args = [] + + if network_tuple: + if network_tuple.appservice_id: + published_sql = """ + SELECT room_id from appservice_room_list + WHERE appservice_id = ? AND network_id = ? + """ + query_args.append(network_tuple.appservice_id) + query_args.append(network_tuple.network_id) + else: + published_sql = """ + SELECT room_id FROM rooms WHERE is_public + """ + else: + published_sql = """ + SELECT room_id FROM rooms WHERE is_public + UNION SELECT room_id from appservice_room_list + """ + + sql = """ + SELECT + COALESCE(COUNT(*), 0) + FROM ( + %(published_sql)s + ) published + INNER JOIN room_stats_state USING (room_id) + INNER JOIN room_stats_current USING (room_id) + WHERE + ( + join_rules = 'public' OR history_visibility = 'world_readable' + ) + AND joined_members > 0 + """ % { + "published_sql": published_sql + } + + txn.execute(sql, query_args) + return txn.fetchone()[0] + + return self.runInteraction("count_public_rooms", _count_public_rooms_txn) + + @defer.inlineCallbacks + def get_largest_public_rooms( + self, + network_tuple: Optional[ThirdPartyInstanceID], + search_filter: Optional[dict], + limit: Optional[int], + bounds: Optional[Tuple[int, str]], + forwards: bool, + ignore_non_federatable: bool = False, + ): + """Gets the largest public rooms (where largest is in terms of joined + members, as tracked in the statistics table). + + Args: + network_tuple + search_filter + limit: Maxmimum number of rows to return, unlimited otherwise. + bounds: An uppoer or lower bound to apply to result set if given, + consists of a joined member count and room_id (these are + excluded from result set). + forwards: true iff going forwards, going backwards otherwise + ignore_non_federatable: If true filters out non-federatable rooms. + + Returns: + Rooms in order: biggest number of joined users first. + We then arbitrarily use the room_id as a tie breaker. + + """ + + where_clauses = [] + query_args = [] + + if network_tuple: + if network_tuple.appservice_id: + published_sql = """ + SELECT room_id from appservice_room_list + WHERE appservice_id = ? AND network_id = ? + """ + query_args.append(network_tuple.appservice_id) + query_args.append(network_tuple.network_id) + else: + published_sql = """ + SELECT room_id FROM rooms WHERE is_public + """ + else: + published_sql = """ + SELECT room_id FROM rooms WHERE is_public + UNION SELECT room_id from appservice_room_list + """ + + # Work out the bounds if we're given them, these bounds look slightly + # odd, but are designed to help query planner use indices by pulling + # out a common bound. + if bounds: + last_joined_members, last_room_id = bounds + if forwards: + where_clauses.append( + """ + joined_members <= ? AND ( + joined_members < ? OR room_id < ? + ) + """ + ) + else: + where_clauses.append( + """ + joined_members >= ? AND ( + joined_members > ? OR room_id > ? + ) + """ + ) + + query_args += [last_joined_members, last_joined_members, last_room_id] + + if ignore_non_federatable: + where_clauses.append("is_federatable") + + if search_filter and search_filter.get("generic_search_term", None): + search_term = "%" + search_filter["generic_search_term"] + "%" + + where_clauses.append( + """ + ( + name LIKE ? + OR topic LIKE ? + OR canonical_alias LIKE ? + ) + """ + ) + query_args += [search_term, search_term, search_term] + + where_clause = "" + if where_clauses: + where_clause = " AND " + " AND ".join(where_clauses) + + sql = """ + SELECT + room_id, name, topic, canonical_alias, joined_members, + avatar, history_visibility, joined_members, guest_access + FROM ( + %(published_sql)s + ) published + INNER JOIN room_stats_state USING (room_id) + INNER JOIN room_stats_current USING (room_id) + WHERE + ( + join_rules = 'public' OR history_visibility = 'world_readable' + ) + AND joined_members > 0 + %(where_clause)s + ORDER BY joined_members %(dir)s, room_id %(dir)s + """ % { + "published_sql": published_sql, + "where_clause": where_clause, + "dir": "DESC" if forwards else "ASC", + } + + if limit is not None: + query_args.append(limit) + + sql += """ + LIMIT ? + """ + + def _get_largest_public_rooms_txn(txn): + txn.execute(sql, query_args) + + results = self.cursor_to_dict(txn) + + if not forwards: + results.reverse() + + return results + + ret_val = yield self.runInteraction( + "get_largest_public_rooms", _get_largest_public_rooms_txn + ) + defer.returnValue(ret_val) + + @cached(max_entries=10000) + def is_room_blocked(self, room_id): + return self._simple_select_one_onecol( + table="blocked_rooms", + keyvalues={"room_id": room_id}, + retcol="1", + allow_none=True, + desc="is_room_blocked", + ) + + @cachedInlineCallbacks(max_entries=10000) + def get_ratelimit_for_user(self, user_id): + """Check if there are any overrides for ratelimiting for the given + user + + Args: + user_id (str) + + Returns: + RatelimitOverride if there is an override, else None. If the contents + of RatelimitOverride are None or 0 then ratelimitng has been + disabled for that user entirely. + """ + row = yield self._simple_select_one( + table="ratelimit_override", + keyvalues={"user_id": user_id}, + retcols=("messages_per_second", "burst_count"), + allow_none=True, + desc="get_ratelimit_for_user", + ) + + if row: + return RatelimitOverride( + messages_per_second=row["messages_per_second"], + burst_count=row["burst_count"], + ) + else: + return None + + +class RoomStore(RoomWorkerStore, SearchStore): + @defer.inlineCallbacks + def store_room(self, room_id, room_creator_user_id, is_public): + """Stores a room. + + Args: + room_id (str): The desired room ID, can be None. + room_creator_user_id (str): The user ID of the room creator. + is_public (bool): True to indicate that this room should appear in + public room lists. + Raises: + StoreError if the room could not be stored. + """ + try: + + def store_room_txn(txn, next_id): + self._simple_insert_txn( + txn, + "rooms", + { + "room_id": room_id, + "creator": room_creator_user_id, + "is_public": is_public, + }, + ) + if is_public: + self._simple_insert_txn( + txn, + table="public_room_list_stream", + values={ + "stream_id": next_id, + "room_id": room_id, + "visibility": is_public, + }, + ) + + with self._public_room_id_gen.get_next() as next_id: + yield self.runInteraction("store_room_txn", store_room_txn, next_id) + except Exception as e: + logger.error("store_room with room_id=%s failed: %s", room_id, e) + raise StoreError(500, "Problem creating room.") + + @defer.inlineCallbacks + def set_room_is_public(self, room_id, is_public): + def set_room_is_public_txn(txn, next_id): + self._simple_update_one_txn( + txn, + table="rooms", + keyvalues={"room_id": room_id}, + updatevalues={"is_public": is_public}, + ) + + entries = self._simple_select_list_txn( + txn, + table="public_room_list_stream", + keyvalues={ + "room_id": room_id, + "appservice_id": None, + "network_id": None, + }, + retcols=("stream_id", "visibility"), + ) + + entries.sort(key=lambda r: r["stream_id"]) + + add_to_stream = True + if entries: + add_to_stream = bool(entries[-1]["visibility"]) != is_public + + if add_to_stream: + self._simple_insert_txn( + txn, + table="public_room_list_stream", + values={ + "stream_id": next_id, + "room_id": room_id, + "visibility": is_public, + "appservice_id": None, + "network_id": None, + }, + ) + + with self._public_room_id_gen.get_next() as next_id: + yield self.runInteraction( + "set_room_is_public", set_room_is_public_txn, next_id + ) + self.hs.get_notifier().on_new_replication_data() + + @defer.inlineCallbacks + def set_room_is_public_appservice( + self, room_id, appservice_id, network_id, is_public + ): + """Edit the appservice/network specific public room list. + + Each appservice can have a number of published room lists associated + with them, keyed off of an appservice defined `network_id`, which + basically represents a single instance of a bridge to a third party + network. + + Args: + room_id (str) + appservice_id (str) + network_id (str) + is_public (bool): Whether to publish or unpublish the room from the + list. + """ + + def set_room_is_public_appservice_txn(txn, next_id): + if is_public: + try: + self._simple_insert_txn( + txn, + table="appservice_room_list", + values={ + "appservice_id": appservice_id, + "network_id": network_id, + "room_id": room_id, + }, + ) + except self.database_engine.module.IntegrityError: + # We've already inserted, nothing to do. + return + else: + self._simple_delete_txn( + txn, + table="appservice_room_list", + keyvalues={ + "appservice_id": appservice_id, + "network_id": network_id, + "room_id": room_id, + }, + ) + + entries = self._simple_select_list_txn( + txn, + table="public_room_list_stream", + keyvalues={ + "room_id": room_id, + "appservice_id": appservice_id, + "network_id": network_id, + }, + retcols=("stream_id", "visibility"), + ) + + entries.sort(key=lambda r: r["stream_id"]) + + add_to_stream = True + if entries: + add_to_stream = bool(entries[-1]["visibility"]) != is_public + + if add_to_stream: + self._simple_insert_txn( + txn, + table="public_room_list_stream", + values={ + "stream_id": next_id, + "room_id": room_id, + "visibility": is_public, + "appservice_id": appservice_id, + "network_id": network_id, + }, + ) + + with self._public_room_id_gen.get_next() as next_id: + yield self.runInteraction( + "set_room_is_public_appservice", + set_room_is_public_appservice_txn, + next_id, + ) + self.hs.get_notifier().on_new_replication_data() + + def get_room_count(self): + """Retrieve a list of all rooms + """ + + def f(txn): + sql = "SELECT count(*) FROM rooms" + txn.execute(sql) + row = txn.fetchone() + return row[0] or 0 + + return self.runInteraction("get_rooms", f) + + def _store_room_topic_txn(self, txn, event): + if hasattr(event, "content") and "topic" in event.content: + self.store_event_search_txn( + txn, event, "content.topic", event.content["topic"] + ) + + def _store_room_name_txn(self, txn, event): + if hasattr(event, "content") and "name" in event.content: + self.store_event_search_txn( + txn, event, "content.name", event.content["name"] + ) + + def _store_room_message_txn(self, txn, event): + if hasattr(event, "content") and "body" in event.content: + self.store_event_search_txn( + txn, event, "content.body", event.content["body"] + ) + + def add_event_report( + self, room_id, event_id, user_id, reason, content, received_ts + ): + next_id = self._event_reports_id_gen.get_next() + return self._simple_insert( + table="event_reports", + values={ + "id": next_id, + "received_ts": received_ts, + "room_id": room_id, + "event_id": event_id, + "user_id": user_id, + "reason": reason, + "content": json.dumps(content), + }, + desc="add_event_report", + ) + + def get_current_public_room_stream_id(self): + return self._public_room_id_gen.get_current_token() + + def get_all_new_public_rooms(self, prev_id, current_id, limit): + def get_all_new_public_rooms(txn): + sql = """ + SELECT stream_id, room_id, visibility, appservice_id, network_id + FROM public_room_list_stream + WHERE stream_id > ? AND stream_id <= ? + ORDER BY stream_id ASC + LIMIT ? + """ + + txn.execute(sql, (prev_id, current_id, limit)) + return txn.fetchall() + + if prev_id == current_id: + return defer.succeed([]) + + return self.runInteraction("get_all_new_public_rooms", get_all_new_public_rooms) + + @defer.inlineCallbacks + def block_room(self, room_id, user_id): + """Marks the room as blocked. Can be called multiple times. + + Args: + room_id (str): Room to block + user_id (str): Who blocked it + + Returns: + Deferred + """ + yield self._simple_upsert( + table="blocked_rooms", + keyvalues={"room_id": room_id}, + values={}, + insertion_values={"user_id": user_id}, + desc="block_room", + ) + yield self.runInteraction( + "block_room_invalidation", + self._invalidate_cache_and_stream, + self.is_room_blocked, + (room_id,), + ) + + def get_media_mxcs_in_room(self, room_id): + """Retrieves all the local and remote media MXC URIs in a given room + + Args: + room_id (str) + + Returns: + The local and remote media as a lists of tuples where the key is + the hostname and the value is the media ID. + """ + + def _get_media_mxcs_in_room_txn(txn): + local_mxcs, remote_mxcs = self._get_media_mxcs_in_room_txn(txn, room_id) + local_media_mxcs = [] + remote_media_mxcs = [] + + # Convert the IDs to MXC URIs + for media_id in local_mxcs: + local_media_mxcs.append("mxc://%s/%s" % (self.hs.hostname, media_id)) + for hostname, media_id in remote_mxcs: + remote_media_mxcs.append("mxc://%s/%s" % (hostname, media_id)) + + return local_media_mxcs, remote_media_mxcs + + return self.runInteraction("get_media_ids_in_room", _get_media_mxcs_in_room_txn) + + def quarantine_media_ids_in_room(self, room_id, quarantined_by): + """For a room loops through all events with media and quarantines + the associated media + """ + + def _quarantine_media_in_room_txn(txn): + local_mxcs, remote_mxcs = self._get_media_mxcs_in_room_txn(txn, room_id) + total_media_quarantined = 0 + + # Now update all the tables to set the quarantined_by flag + + txn.executemany( + """ + UPDATE local_media_repository + SET quarantined_by = ? + WHERE media_id = ? + """, + ((quarantined_by, media_id) for media_id in local_mxcs), + ) + + txn.executemany( + """ + UPDATE remote_media_cache + SET quarantined_by = ? + WHERE media_origin = ? AND media_id = ? + """, + ( + (quarantined_by, origin, media_id) + for origin, media_id in remote_mxcs + ), + ) + + total_media_quarantined += len(local_mxcs) + total_media_quarantined += len(remote_mxcs) + + return total_media_quarantined + + return self.runInteraction( + "quarantine_media_in_room", _quarantine_media_in_room_txn + ) + + def _get_media_mxcs_in_room_txn(self, txn, room_id): + """Retrieves all the local and remote media MXC URIs in a given room + + Args: + txn (cursor) + room_id (str) + + Returns: + The local and remote media as a lists of tuples where the key is + the hostname and the value is the media ID. + """ + mxc_re = re.compile("^mxc://([^/]+)/([^/#?]+)") + + next_token = self.get_current_events_token() + 1 + local_media_mxcs = [] + remote_media_mxcs = [] + + while next_token: + sql = """ + SELECT stream_ordering, json FROM events + JOIN event_json USING (room_id, event_id) + WHERE room_id = ? + AND stream_ordering < ? + AND contains_url = ? AND outlier = ? + ORDER BY stream_ordering DESC + LIMIT ? + """ + txn.execute(sql, (room_id, next_token, True, False, 100)) + + next_token = None + for stream_ordering, content_json in txn: + next_token = stream_ordering + event_json = json.loads(content_json) + content = event_json["content"] + content_url = content.get("url") + thumbnail_url = content.get("info", {}).get("thumbnail_url") + + for url in (content_url, thumbnail_url): + if not url: + continue + matches = mxc_re.match(url) + if matches: + hostname = matches.group(1) + media_id = matches.group(2) + if hostname == self.hs.hostname: + local_media_mxcs.append(media_id) + else: + remote_media_mxcs.append((hostname, media_id)) + + return local_media_mxcs, remote_media_mxcs diff --git a/synapse/storage/data_stores/main/roommember.py b/synapse/storage/data_stores/main/roommember.py new file mode 100644 index 0000000000..e47ab604dd --- /dev/null +++ b/synapse/storage/data_stores/main/roommember.py @@ -0,0 +1,1145 @@ +# -*- coding: utf-8 -*- +# Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging + +from six import iteritems, itervalues + +from canonicaljson import json + +from twisted.internet import defer + +from synapse.api.constants import EventTypes, Membership +from synapse.metrics import LaterGauge +from synapse.metrics.background_process_metrics import run_as_background_process +from synapse.storage._base import LoggingTransaction, make_in_list_sql_clause +from synapse.storage.background_updates import BackgroundUpdateStore +from synapse.storage.data_stores.main.events_worker import EventsWorkerStore +from synapse.storage.engines import Sqlite3Engine +from synapse.storage.roommember import ( + GetRoomsForUserWithStreamOrdering, + MemberSummary, + ProfileInfo, + RoomsForUser, +) +from synapse.types import get_domain_from_id +from synapse.util.async_helpers import Linearizer +from synapse.util.caches import intern_string +from synapse.util.caches.descriptors import cached, cachedInlineCallbacks, cachedList +from synapse.util.metrics import Measure +from synapse.util.stringutils import to_ascii + +logger = logging.getLogger(__name__) + + +_MEMBERSHIP_PROFILE_UPDATE_NAME = "room_membership_profile_update" +_CURRENT_STATE_MEMBERSHIP_UPDATE_NAME = "current_state_events_membership" + + +class RoomMemberWorkerStore(EventsWorkerStore): + def __init__(self, db_conn, hs): + super(RoomMemberWorkerStore, self).__init__(db_conn, hs) + + # Is the current_state_events.membership up to date? Or is the + # background update still running? + self._current_state_events_membership_up_to_date = False + + txn = LoggingTransaction( + db_conn.cursor(), + name="_check_safe_current_state_events_membership_updated", + database_engine=self.database_engine, + ) + self._check_safe_current_state_events_membership_updated_txn(txn) + txn.close() + + if self.hs.config.metrics_flags.known_servers: + self._known_servers_count = 1 + self.hs.get_clock().looping_call( + run_as_background_process, + 60 * 1000, + "_count_known_servers", + self._count_known_servers, + ) + self.hs.get_clock().call_later( + 1000, + run_as_background_process, + "_count_known_servers", + self._count_known_servers, + ) + LaterGauge( + "synapse_federation_known_servers", + "", + [], + lambda: self._known_servers_count, + ) + + @defer.inlineCallbacks + def _count_known_servers(self): + """ + Count the servers that this server knows about. + + The statistic is stored on the class for the + `synapse_federation_known_servers` LaterGauge to collect. + """ + + def _transact(txn): + if isinstance(self.database_engine, Sqlite3Engine): + query = """ + SELECT COUNT(DISTINCT substr(out.user_id, pos+1)) + FROM ( + SELECT rm.user_id as user_id, instr(rm.user_id, ':') + AS pos FROM room_memberships as rm + INNER JOIN current_state_events as c ON rm.event_id = c.event_id + WHERE c.type = 'm.room.member' + ) as out + """ + else: + query = """ + SELECT COUNT(DISTINCT split_part(state_key, ':', 2)) + FROM current_state_events + WHERE type = 'm.room.member' AND membership = 'join'; + """ + txn.execute(query) + return list(txn)[0][0] + + count = yield self.runInteraction("get_known_servers", _transact) + + # We always know about ourselves, even if we have nothing in + # room_memberships (for example, the server is new). + self._known_servers_count = max([count, 1]) + return self._known_servers_count + + def _check_safe_current_state_events_membership_updated_txn(self, txn): + """Checks if it is safe to assume the new current_state_events + membership column is up to date + """ + + pending_update = self._simple_select_one_txn( + txn, + table="background_updates", + keyvalues={"update_name": _CURRENT_STATE_MEMBERSHIP_UPDATE_NAME}, + retcols=["update_name"], + allow_none=True, + ) + + self._current_state_events_membership_up_to_date = not pending_update + + # If the update is still running, reschedule to run. + if pending_update: + self._clock.call_later( + 15.0, + run_as_background_process, + "_check_safe_current_state_events_membership_updated", + self.runInteraction, + "_check_safe_current_state_events_membership_updated", + self._check_safe_current_state_events_membership_updated_txn, + ) + + @cachedInlineCallbacks(max_entries=100000, iterable=True, cache_context=True) + def get_hosts_in_room(self, room_id, cache_context): + """Returns the set of all hosts currently in the room + """ + user_ids = yield self.get_users_in_room( + room_id, on_invalidate=cache_context.invalidate + ) + hosts = frozenset(get_domain_from_id(user_id) for user_id in user_ids) + return hosts + + @cached(max_entries=100000, iterable=True) + def get_users_in_room(self, room_id): + return self.runInteraction( + "get_users_in_room", self.get_users_in_room_txn, room_id + ) + + def get_users_in_room_txn(self, txn, room_id): + # If we can assume current_state_events.membership is up to date + # then we can avoid a join, which is a Very Good Thing given how + # frequently this function gets called. + if self._current_state_events_membership_up_to_date: + sql = """ + SELECT state_key FROM current_state_events + WHERE type = 'm.room.member' AND room_id = ? AND membership = ? + """ + else: + sql = """ + SELECT state_key FROM room_memberships as m + INNER JOIN current_state_events as c + ON m.event_id = c.event_id + AND m.room_id = c.room_id + AND m.user_id = c.state_key + WHERE c.type = 'm.room.member' AND c.room_id = ? AND m.membership = ? + """ + + txn.execute(sql, (room_id, Membership.JOIN)) + return [to_ascii(r[0]) for r in txn] + + @cached(max_entries=100000) + def get_room_summary(self, room_id): + """ Get the details of a room roughly suitable for use by the room + summary extension to /sync. Useful when lazy loading room members. + Args: + room_id (str): The room ID to query + Returns: + Deferred[dict[str, MemberSummary]: + dict of membership states, pointing to a MemberSummary named tuple. + """ + + def _get_room_summary_txn(txn): + # first get counts. + # We do this all in one transaction to keep the cache small. + # FIXME: get rid of this when we have room_stats + + # If we can assume current_state_events.membership is up to date + # then we can avoid a join, which is a Very Good Thing given how + # frequently this function gets called. + if self._current_state_events_membership_up_to_date: + # Note, rejected events will have a null membership field, so + # we we manually filter them out. + sql = """ + SELECT count(*), membership FROM current_state_events + WHERE type = 'm.room.member' AND room_id = ? + AND membership IS NOT NULL + GROUP BY membership + """ + else: + sql = """ + SELECT count(*), m.membership FROM room_memberships as m + INNER JOIN current_state_events as c + ON m.event_id = c.event_id + AND m.room_id = c.room_id + AND m.user_id = c.state_key + WHERE c.type = 'm.room.member' AND c.room_id = ? + GROUP BY m.membership + """ + + txn.execute(sql, (room_id,)) + res = {} + for count, membership in txn: + summary = res.setdefault(to_ascii(membership), MemberSummary([], count)) + + # we order by membership and then fairly arbitrarily by event_id so + # heroes are consistent + if self._current_state_events_membership_up_to_date: + # Note, rejected events will have a null membership field, so + # we we manually filter them out. + sql = """ + SELECT state_key, membership, event_id + FROM current_state_events + WHERE type = 'm.room.member' AND room_id = ? + AND membership IS NOT NULL + ORDER BY + CASE membership WHEN ? THEN 1 WHEN ? THEN 2 ELSE 3 END ASC, + event_id ASC + LIMIT ? + """ + else: + sql = """ + SELECT c.state_key, m.membership, c.event_id + FROM room_memberships as m + INNER JOIN current_state_events as c USING (room_id, event_id) + WHERE c.type = 'm.room.member' AND c.room_id = ? + ORDER BY + CASE m.membership WHEN ? THEN 1 WHEN ? THEN 2 ELSE 3 END ASC, + c.event_id ASC + LIMIT ? + """ + + # 6 is 5 (number of heroes) plus 1, in case one of them is the calling user. + txn.execute(sql, (room_id, Membership.JOIN, Membership.INVITE, 6)) + for user_id, membership, event_id in txn: + summary = res[to_ascii(membership)] + # we will always have a summary for this membership type at this + # point given the summary currently contains the counts. + members = summary.members + members.append((to_ascii(user_id), to_ascii(event_id))) + + return res + + return self.runInteraction("get_room_summary", _get_room_summary_txn) + + def _get_user_counts_in_room_txn(self, txn, room_id): + """ + Get the user count in a room by membership. + + Args: + room_id (str) + membership (Membership) + + Returns: + Deferred[int] + """ + sql = """ + SELECT m.membership, count(*) FROM room_memberships as m + INNER JOIN current_state_events as c USING(event_id) + WHERE c.type = 'm.room.member' AND c.room_id = ? + GROUP BY m.membership + """ + + txn.execute(sql, (room_id,)) + return {row[0]: row[1] for row in txn} + + @cached() + def get_invited_rooms_for_user(self, user_id): + """ Get all the rooms the user is invited to + Args: + user_id (str): The user ID. + Returns: + A deferred list of RoomsForUser. + """ + + return self.get_rooms_for_user_where_membership_is(user_id, [Membership.INVITE]) + + @defer.inlineCallbacks + def get_invite_for_user_in_room(self, user_id, room_id): + """Gets the invite for the given user and room + + Args: + user_id (str) + room_id (str) + + Returns: + Deferred: Resolves to either a RoomsForUser or None if no invite was + found. + """ + invites = yield self.get_invited_rooms_for_user(user_id) + for invite in invites: + if invite.room_id == room_id: + return invite + return None + + @defer.inlineCallbacks + def get_rooms_for_user_where_membership_is(self, user_id, membership_list): + """ Get all the rooms for this user where the membership for this user + matches one in the membership list. + + Filters out forgotten rooms. + + Args: + user_id (str): The user ID. + membership_list (list): A list of synapse.api.constants.Membership + values which the user must be in. + + Returns: + Deferred[list[RoomsForUser]] + """ + if not membership_list: + return defer.succeed(None) + + rooms = yield self.runInteraction( + "get_rooms_for_user_where_membership_is", + self._get_rooms_for_user_where_membership_is_txn, + user_id, + membership_list, + ) + + # Now we filter out forgotten rooms + forgotten_rooms = yield self.get_forgotten_rooms_for_user(user_id) + return [room for room in rooms if room.room_id not in forgotten_rooms] + + def _get_rooms_for_user_where_membership_is_txn( + self, txn, user_id, membership_list + ): + + do_invite = Membership.INVITE in membership_list + membership_list = [m for m in membership_list if m != Membership.INVITE] + + results = [] + if membership_list: + if self._current_state_events_membership_up_to_date: + clause, args = make_in_list_sql_clause( + self.database_engine, "c.membership", membership_list + ) + sql = """ + SELECT room_id, e.sender, c.membership, event_id, e.stream_ordering + FROM current_state_events AS c + INNER JOIN events AS e USING (room_id, event_id) + WHERE + c.type = 'm.room.member' + AND state_key = ? + AND %s + """ % ( + clause, + ) + else: + clause, args = make_in_list_sql_clause( + self.database_engine, "m.membership", membership_list + ) + sql = """ + SELECT room_id, e.sender, m.membership, event_id, e.stream_ordering + FROM current_state_events AS c + INNER JOIN room_memberships AS m USING (room_id, event_id) + INNER JOIN events AS e USING (room_id, event_id) + WHERE + c.type = 'm.room.member' + AND state_key = ? + AND %s + """ % ( + clause, + ) + + txn.execute(sql, (user_id, *args)) + results = [RoomsForUser(**r) for r in self.cursor_to_dict(txn)] + + if do_invite: + sql = ( + "SELECT i.room_id, inviter, i.event_id, e.stream_ordering" + " FROM local_invites as i" + " INNER JOIN events as e USING (event_id)" + " WHERE invitee = ? AND locally_rejected is NULL" + " AND replaced_by is NULL" + ) + + txn.execute(sql, (user_id,)) + results.extend( + RoomsForUser( + room_id=r["room_id"], + sender=r["inviter"], + event_id=r["event_id"], + stream_ordering=r["stream_ordering"], + membership=Membership.INVITE, + ) + for r in self.cursor_to_dict(txn) + ) + + return results + + @cachedInlineCallbacks(max_entries=500000, iterable=True) + def get_rooms_for_user_with_stream_ordering(self, user_id): + """Returns a set of room_ids the user is currently joined to + + Args: + user_id (str) + + Returns: + Deferred[frozenset[GetRoomsForUserWithStreamOrdering]]: Returns + the rooms the user is in currently, along with the stream ordering + of the most recent join for that user and room. + """ + rooms = yield self.get_rooms_for_user_where_membership_is( + user_id, membership_list=[Membership.JOIN] + ) + return frozenset( + GetRoomsForUserWithStreamOrdering(r.room_id, r.stream_ordering) + for r in rooms + ) + + @defer.inlineCallbacks + def get_rooms_for_user(self, user_id, on_invalidate=None): + """Returns a set of room_ids the user is currently joined to + """ + rooms = yield self.get_rooms_for_user_with_stream_ordering( + user_id, on_invalidate=on_invalidate + ) + return frozenset(r.room_id for r in rooms) + + @cachedInlineCallbacks(max_entries=500000, cache_context=True, iterable=True) + def get_users_who_share_room_with_user(self, user_id, cache_context): + """Returns the set of users who share a room with `user_id` + """ + room_ids = yield self.get_rooms_for_user( + user_id, on_invalidate=cache_context.invalidate + ) + + user_who_share_room = set() + for room_id in room_ids: + user_ids = yield self.get_users_in_room( + room_id, on_invalidate=cache_context.invalidate + ) + user_who_share_room.update(user_ids) + + return user_who_share_room + + @defer.inlineCallbacks + def get_joined_users_from_context(self, event, context): + state_group = context.state_group + if not state_group: + # If state_group is None it means it has yet to be assigned a + # state group, i.e. we need to make sure that calls with a state_group + # of None don't hit previous cached calls with a None state_group. + # To do this we set the state_group to a new object as object() != object() + state_group = object() + + current_state_ids = yield context.get_current_state_ids(self) + result = yield self._get_joined_users_from_context( + event.room_id, state_group, current_state_ids, event=event, context=context + ) + return result + + @defer.inlineCallbacks + def get_joined_users_from_state(self, room_id, state_entry): + state_group = state_entry.state_group + if not state_group: + # If state_group is None it means it has yet to be assigned a + # state group, i.e. we need to make sure that calls with a state_group + # of None don't hit previous cached calls with a None state_group. + # To do this we set the state_group to a new object as object() != object() + state_group = object() + + with Measure(self._clock, "get_joined_users_from_state"): + return ( + yield self._get_joined_users_from_context( + room_id, state_group, state_entry.state, context=state_entry + ) + ) + + @cachedInlineCallbacks( + num_args=2, cache_context=True, iterable=True, max_entries=100000 + ) + def _get_joined_users_from_context( + self, + room_id, + state_group, + current_state_ids, + cache_context, + event=None, + context=None, + ): + # We don't use `state_group`, it's there so that we can cache based + # on it. However, it's important that it's never None, since two current_states + # with a state_group of None are likely to be different. + # See bulk_get_push_rules_for_room for how we work around this. + assert state_group is not None + + users_in_room = {} + member_event_ids = [ + e_id + for key, e_id in iteritems(current_state_ids) + if key[0] == EventTypes.Member + ] + + if context is not None: + # If we have a context with a delta from a previous state group, + # check if we also have the result from the previous group in cache. + # If we do then we can reuse that result and simply update it with + # any membership changes in `delta_ids` + if context.prev_group and context.delta_ids: + prev_res = self._get_joined_users_from_context.cache.get( + (room_id, context.prev_group), None + ) + if prev_res and isinstance(prev_res, dict): + users_in_room = dict(prev_res) + member_event_ids = [ + e_id + for key, e_id in iteritems(context.delta_ids) + if key[0] == EventTypes.Member + ] + for etype, state_key in context.delta_ids: + users_in_room.pop(state_key, None) + + # We check if we have any of the member event ids in the event cache + # before we ask the DB + + # We don't update the event cache hit ratio as it completely throws off + # the hit ratio counts. After all, we don't populate the cache if we + # miss it here + event_map = self._get_events_from_cache( + member_event_ids, allow_rejected=False, update_metrics=False + ) + + missing_member_event_ids = [] + for event_id in member_event_ids: + ev_entry = event_map.get(event_id) + if ev_entry: + if ev_entry.event.membership == Membership.JOIN: + users_in_room[to_ascii(ev_entry.event.state_key)] = ProfileInfo( + display_name=to_ascii( + ev_entry.event.content.get("displayname", None) + ), + avatar_url=to_ascii( + ev_entry.event.content.get("avatar_url", None) + ), + ) + else: + missing_member_event_ids.append(event_id) + + if missing_member_event_ids: + event_to_memberships = yield self._get_joined_profiles_from_event_ids( + missing_member_event_ids + ) + users_in_room.update((row for row in event_to_memberships.values() if row)) + + if event is not None and event.type == EventTypes.Member: + if event.membership == Membership.JOIN: + if event.event_id in member_event_ids: + users_in_room[to_ascii(event.state_key)] = ProfileInfo( + display_name=to_ascii(event.content.get("displayname", None)), + avatar_url=to_ascii(event.content.get("avatar_url", None)), + ) + + return users_in_room + + @cached(max_entries=10000) + def _get_joined_profile_from_event_id(self, event_id): + raise NotImplementedError() + + @cachedList( + cached_method_name="_get_joined_profile_from_event_id", + list_name="event_ids", + inlineCallbacks=True, + ) + def _get_joined_profiles_from_event_ids(self, event_ids): + """For given set of member event_ids check if they point to a join + event and if so return the associated user and profile info. + + Args: + event_ids (Iterable[str]): The member event IDs to lookup + + Returns: + Deferred[dict[str, Tuple[str, ProfileInfo]|None]]: Map from event ID + to `user_id` and ProfileInfo (or None if not join event). + """ + + rows = yield self._simple_select_many_batch( + table="room_memberships", + column="event_id", + iterable=event_ids, + retcols=("user_id", "display_name", "avatar_url", "event_id"), + keyvalues={"membership": Membership.JOIN}, + batch_size=500, + desc="_get_membership_from_event_ids", + ) + + return { + row["event_id"]: ( + row["user_id"], + ProfileInfo( + avatar_url=row["avatar_url"], display_name=row["display_name"] + ), + ) + for row in rows + } + + @cachedInlineCallbacks(max_entries=10000) + def is_host_joined(self, room_id, host): + if "%" in host or "_" in host: + raise Exception("Invalid host name") + + sql = """ + SELECT state_key FROM current_state_events AS c + INNER JOIN room_memberships AS m USING (event_id) + WHERE m.membership = 'join' + AND type = 'm.room.member' + AND c.room_id = ? + AND state_key LIKE ? + LIMIT 1 + """ + + # We do need to be careful to ensure that host doesn't have any wild cards + # in it, but we checked above for known ones and we'll check below that + # the returned user actually has the correct domain. + like_clause = "%:" + host + + rows = yield self._execute("is_host_joined", None, sql, room_id, like_clause) + + if not rows: + return False + + user_id = rows[0][0] + if get_domain_from_id(user_id) != host: + # This can only happen if the host name has something funky in it + raise Exception("Invalid host name") + + return True + + @cachedInlineCallbacks() + def was_host_joined(self, room_id, host): + """Check whether the server is or ever was in the room. + + Args: + room_id (str) + host (str) + + Returns: + Deferred: Resolves to True if the host is/was in the room, otherwise + False. + """ + if "%" in host or "_" in host: + raise Exception("Invalid host name") + + sql = """ + SELECT user_id FROM room_memberships + WHERE room_id = ? + AND user_id LIKE ? + AND membership = 'join' + LIMIT 1 + """ + + # We do need to be careful to ensure that host doesn't have any wild cards + # in it, but we checked above for known ones and we'll check below that + # the returned user actually has the correct domain. + like_clause = "%:" + host + + rows = yield self._execute("was_host_joined", None, sql, room_id, like_clause) + + if not rows: + return False + + user_id = rows[0][0] + if get_domain_from_id(user_id) != host: + # This can only happen if the host name has something funky in it + raise Exception("Invalid host name") + + return True + + @defer.inlineCallbacks + def get_joined_hosts(self, room_id, state_entry): + state_group = state_entry.state_group + if not state_group: + # If state_group is None it means it has yet to be assigned a + # state group, i.e. we need to make sure that calls with a state_group + # of None don't hit previous cached calls with a None state_group. + # To do this we set the state_group to a new object as object() != object() + state_group = object() + + with Measure(self._clock, "get_joined_hosts"): + return ( + yield self._get_joined_hosts( + room_id, state_group, state_entry.state, state_entry=state_entry + ) + ) + + @cachedInlineCallbacks(num_args=2, max_entries=10000, iterable=True) + # @defer.inlineCallbacks + def _get_joined_hosts(self, room_id, state_group, current_state_ids, state_entry): + # We don't use `state_group`, its there so that we can cache based + # on it. However, its important that its never None, since two current_state's + # with a state_group of None are likely to be different. + # See bulk_get_push_rules_for_room for how we work around this. + assert state_group is not None + + cache = self._get_joined_hosts_cache(room_id) + joined_hosts = yield cache.get_destinations(state_entry) + + return joined_hosts + + @cached(max_entries=10000) + def _get_joined_hosts_cache(self, room_id): + return _JoinedHostsCache(self, room_id) + + @cachedInlineCallbacks(num_args=2) + def did_forget(self, user_id, room_id): + """Returns whether user_id has elected to discard history for room_id. + + Returns False if they have since re-joined.""" + + def f(txn): + sql = ( + "SELECT" + " COUNT(*)" + " FROM" + " room_memberships" + " WHERE" + " user_id = ?" + " AND" + " room_id = ?" + " AND" + " forgotten = 0" + ) + txn.execute(sql, (user_id, room_id)) + rows = txn.fetchall() + return rows[0][0] + + count = yield self.runInteraction("did_forget_membership", f) + return count == 0 + + @cached() + def get_forgotten_rooms_for_user(self, user_id): + """Gets all rooms the user has forgotten. + + Args: + user_id (str) + + Returns: + Deferred[set[str]] + """ + + def _get_forgotten_rooms_for_user_txn(txn): + # This is a slightly convoluted query that first looks up all rooms + # that the user has forgotten in the past, then rechecks that list + # to see if any have subsequently been updated. This is done so that + # we can use a partial index on `forgotten = 1` on the assumption + # that few users will actually forget many rooms. + # + # Note that a room is considered "forgotten" if *all* membership + # events for that user and room have the forgotten field set (as + # when a user forgets a room we update all rows for that user and + # room, not just the current one). + sql = """ + SELECT room_id, ( + SELECT count(*) FROM room_memberships + WHERE room_id = m.room_id AND user_id = m.user_id AND forgotten = 0 + ) AS count + FROM room_memberships AS m + WHERE user_id = ? AND forgotten = 1 + GROUP BY room_id, user_id; + """ + txn.execute(sql, (user_id,)) + return set(row[0] for row in txn if row[1] == 0) + + return self.runInteraction( + "get_forgotten_rooms_for_user", _get_forgotten_rooms_for_user_txn + ) + + @defer.inlineCallbacks + def get_rooms_user_has_been_in(self, user_id): + """Get all rooms that the user has ever been in. + + Args: + user_id (str) + + Returns: + Deferred[set[str]]: Set of room IDs. + """ + + room_ids = yield self._simple_select_onecol( + table="room_memberships", + keyvalues={"membership": Membership.JOIN, "user_id": user_id}, + retcol="room_id", + desc="get_rooms_user_has_been_in", + ) + + return set(room_ids) + + +class RoomMemberBackgroundUpdateStore(BackgroundUpdateStore): + def __init__(self, db_conn, hs): + super(RoomMemberBackgroundUpdateStore, self).__init__(db_conn, hs) + self.register_background_update_handler( + _MEMBERSHIP_PROFILE_UPDATE_NAME, self._background_add_membership_profile + ) + self.register_background_update_handler( + _CURRENT_STATE_MEMBERSHIP_UPDATE_NAME, + self._background_current_state_membership, + ) + self.register_background_index_update( + "room_membership_forgotten_idx", + index_name="room_memberships_user_room_forgotten", + table="room_memberships", + columns=["user_id", "room_id"], + where_clause="forgotten = 1", + ) + + @defer.inlineCallbacks + def _background_add_membership_profile(self, progress, batch_size): + target_min_stream_id = progress.get( + "target_min_stream_id_inclusive", self._min_stream_order_on_start + ) + max_stream_id = progress.get( + "max_stream_id_exclusive", self._stream_order_on_start + 1 + ) + + INSERT_CLUMP_SIZE = 1000 + + def add_membership_profile_txn(txn): + sql = """ + SELECT stream_ordering, event_id, events.room_id, event_json.json + FROM events + INNER JOIN event_json USING (event_id) + INNER JOIN room_memberships USING (event_id) + WHERE ? <= stream_ordering AND stream_ordering < ? + AND type = 'm.room.member' + ORDER BY stream_ordering DESC + LIMIT ? + """ + + txn.execute(sql, (target_min_stream_id, max_stream_id, batch_size)) + + rows = self.cursor_to_dict(txn) + if not rows: + return 0 + + min_stream_id = rows[-1]["stream_ordering"] + + to_update = [] + for row in rows: + event_id = row["event_id"] + room_id = row["room_id"] + try: + event_json = json.loads(row["json"]) + content = event_json["content"] + except Exception: + continue + + display_name = content.get("displayname", None) + avatar_url = content.get("avatar_url", None) + + if display_name or avatar_url: + to_update.append((display_name, avatar_url, event_id, room_id)) + + to_update_sql = """ + UPDATE room_memberships SET display_name = ?, avatar_url = ? + WHERE event_id = ? AND room_id = ? + """ + for index in range(0, len(to_update), INSERT_CLUMP_SIZE): + clump = to_update[index : index + INSERT_CLUMP_SIZE] + txn.executemany(to_update_sql, clump) + + progress = { + "target_min_stream_id_inclusive": target_min_stream_id, + "max_stream_id_exclusive": min_stream_id, + } + + self._background_update_progress_txn( + txn, _MEMBERSHIP_PROFILE_UPDATE_NAME, progress + ) + + return len(rows) + + result = yield self.runInteraction( + _MEMBERSHIP_PROFILE_UPDATE_NAME, add_membership_profile_txn + ) + + if not result: + yield self._end_background_update(_MEMBERSHIP_PROFILE_UPDATE_NAME) + + return result + + @defer.inlineCallbacks + def _background_current_state_membership(self, progress, batch_size): + """Update the new membership column on current_state_events. + + This works by iterating over all rooms in alphebetical order. + """ + + def _background_current_state_membership_txn(txn, last_processed_room): + processed = 0 + while processed < batch_size: + txn.execute( + """ + SELECT MIN(room_id) FROM current_state_events WHERE room_id > ? + """, + (last_processed_room,), + ) + row = txn.fetchone() + if not row or not row[0]: + return processed, True + + next_room, = row + + sql = """ + UPDATE current_state_events + SET membership = ( + SELECT membership FROM room_memberships + WHERE event_id = current_state_events.event_id + ) + WHERE room_id = ? + """ + txn.execute(sql, (next_room,)) + processed += txn.rowcount + + last_processed_room = next_room + + self._background_update_progress_txn( + txn, + _CURRENT_STATE_MEMBERSHIP_UPDATE_NAME, + {"last_processed_room": last_processed_room}, + ) + + return processed, False + + # If we haven't got a last processed room then just use the empty + # string, which will compare before all room IDs correctly. + last_processed_room = progress.get("last_processed_room", "") + + row_count, finished = yield self.runInteraction( + "_background_current_state_membership_update", + _background_current_state_membership_txn, + last_processed_room, + ) + + if finished: + yield self._end_background_update(_CURRENT_STATE_MEMBERSHIP_UPDATE_NAME) + + return row_count + + +class RoomMemberStore(RoomMemberWorkerStore, RoomMemberBackgroundUpdateStore): + def __init__(self, db_conn, hs): + super(RoomMemberStore, self).__init__(db_conn, hs) + + def _store_room_members_txn(self, txn, events, backfilled): + """Store a room member in the database. + """ + self._simple_insert_many_txn( + txn, + table="room_memberships", + values=[ + { + "event_id": event.event_id, + "user_id": event.state_key, + "sender": event.user_id, + "room_id": event.room_id, + "membership": event.membership, + "display_name": event.content.get("displayname", None), + "avatar_url": event.content.get("avatar_url", None), + } + for event in events + ], + ) + + for event in events: + txn.call_after( + self._membership_stream_cache.entity_has_changed, + event.state_key, + event.internal_metadata.stream_ordering, + ) + txn.call_after( + self.get_invited_rooms_for_user.invalidate, (event.state_key,) + ) + + # We update the local_invites table only if the event is "current", + # i.e., its something that has just happened. If the event is an + # outlier it is only current if its an "out of band membership", + # like a remote invite or a rejection of a remote invite. + is_new_state = not backfilled and ( + not event.internal_metadata.is_outlier() + or event.internal_metadata.is_out_of_band_membership() + ) + is_mine = self.hs.is_mine_id(event.state_key) + if is_new_state and is_mine: + if event.membership == Membership.INVITE: + self._simple_insert_txn( + txn, + table="local_invites", + values={ + "event_id": event.event_id, + "invitee": event.state_key, + "inviter": event.sender, + "room_id": event.room_id, + "stream_id": event.internal_metadata.stream_ordering, + }, + ) + else: + sql = ( + "UPDATE local_invites SET stream_id = ?, replaced_by = ? WHERE" + " room_id = ? AND invitee = ? AND locally_rejected is NULL" + " AND replaced_by is NULL" + ) + + txn.execute( + sql, + ( + event.internal_metadata.stream_ordering, + event.event_id, + event.room_id, + event.state_key, + ), + ) + + @defer.inlineCallbacks + def locally_reject_invite(self, user_id, room_id): + sql = ( + "UPDATE local_invites SET stream_id = ?, locally_rejected = ? WHERE" + " room_id = ? AND invitee = ? AND locally_rejected is NULL" + " AND replaced_by is NULL" + ) + + def f(txn, stream_ordering): + txn.execute(sql, (stream_ordering, True, room_id, user_id)) + + with self._stream_id_gen.get_next() as stream_ordering: + yield self.runInteraction("locally_reject_invite", f, stream_ordering) + + def forget(self, user_id, room_id): + """Indicate that user_id wishes to discard history for room_id.""" + + def f(txn): + sql = ( + "UPDATE" + " room_memberships" + " SET" + " forgotten = 1" + " WHERE" + " user_id = ?" + " AND" + " room_id = ?" + ) + txn.execute(sql, (user_id, room_id)) + + self._invalidate_cache_and_stream(txn, self.did_forget, (user_id, room_id)) + self._invalidate_cache_and_stream( + txn, self.get_forgotten_rooms_for_user, (user_id,) + ) + + return self.runInteraction("forget_membership", f) + + +class _JoinedHostsCache(object): + """Cache for joined hosts in a room that is optimised to handle updates + via state deltas. + """ + + def __init__(self, store, room_id): + self.store = store + self.room_id = room_id + + self.hosts_to_joined_users = {} + + self.state_group = object() + + self.linearizer = Linearizer("_JoinedHostsCache") + + self._len = 0 + + @defer.inlineCallbacks + def get_destinations(self, state_entry): + """Get set of destinations for a state entry + + Args: + state_entry(synapse.state._StateCacheEntry) + """ + if state_entry.state_group == self.state_group: + return frozenset(self.hosts_to_joined_users) + + with (yield self.linearizer.queue(())): + if state_entry.state_group == self.state_group: + pass + elif state_entry.prev_group == self.state_group: + for (typ, state_key), event_id in iteritems(state_entry.delta_ids): + if typ != EventTypes.Member: + continue + + host = intern_string(get_domain_from_id(state_key)) + user_id = state_key + known_joins = self.hosts_to_joined_users.setdefault(host, set()) + + event = yield self.store.get_event(event_id) + if event.membership == Membership.JOIN: + known_joins.add(user_id) + else: + known_joins.discard(user_id) + + if not known_joins: + self.hosts_to_joined_users.pop(host, None) + else: + joined_users = yield self.store.get_joined_users_from_state( + self.room_id, state_entry + ) + + self.hosts_to_joined_users = {} + for user_id in joined_users: + host = intern_string(get_domain_from_id(user_id)) + self.hosts_to_joined_users.setdefault(host, set()).add(user_id) + + if state_entry.state_group: + self.state_group = state_entry.state_group + else: + self.state_group = object() + self._len = sum(len(v) for v in itervalues(self.hosts_to_joined_users)) + return frozenset(self.hosts_to_joined_users) + + def __len__(self): + return self._len diff --git a/synapse/storage/data_stores/main/schema/delta/12/v12.sql b/synapse/storage/data_stores/main/schema/delta/12/v12.sql new file mode 100644 index 0000000000..5964c5aaac --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/12/v12.sql @@ -0,0 +1,63 @@ +/* Copyright 2015, 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE TABLE IF NOT EXISTS rejections( + event_id TEXT NOT NULL, + reason TEXT NOT NULL, + last_check TEXT NOT NULL, + UNIQUE (event_id) +); + +-- Push notification endpoints that users have configured +CREATE TABLE IF NOT EXISTS pushers ( + id INTEGER PRIMARY KEY AUTOINCREMENT, + user_name TEXT NOT NULL, + profile_tag VARCHAR(32) NOT NULL, + kind VARCHAR(8) NOT NULL, + app_id VARCHAR(64) NOT NULL, + app_display_name VARCHAR(64) NOT NULL, + device_display_name VARCHAR(128) NOT NULL, + pushkey VARBINARY(512) NOT NULL, + ts BIGINT UNSIGNED NOT NULL, + lang VARCHAR(8), + data LONGBLOB, + last_token TEXT, + last_success BIGINT UNSIGNED, + failing_since BIGINT UNSIGNED, + UNIQUE (app_id, pushkey) +); + +CREATE TABLE IF NOT EXISTS push_rules ( + id INTEGER PRIMARY KEY AUTOINCREMENT, + user_name TEXT NOT NULL, + rule_id TEXT NOT NULL, + priority_class TINYINT NOT NULL, + priority INTEGER NOT NULL DEFAULT 0, + conditions TEXT NOT NULL, + actions TEXT NOT NULL, + UNIQUE(user_name, rule_id) +); + +CREATE INDEX IF NOT EXISTS push_rules_user_name on push_rules (user_name); + +CREATE TABLE IF NOT EXISTS user_filters( + user_id TEXT, + filter_id BIGINT UNSIGNED, + filter_json LONGBLOB +); + +CREATE INDEX IF NOT EXISTS user_filters_by_user_id_filter_id ON user_filters( + user_id, filter_id +); diff --git a/synapse/storage/data_stores/main/schema/delta/13/v13.sql b/synapse/storage/data_stores/main/schema/delta/13/v13.sql new file mode 100644 index 0000000000..f8649e5d99 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/13/v13.sql @@ -0,0 +1,19 @@ +/* Copyright 2015, 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* We used to create a tables called application_services and + * application_services_regex, but these are no longer used and are removed in + * delta 54. + */ diff --git a/synapse/storage/data_stores/main/schema/delta/14/v14.sql b/synapse/storage/data_stores/main/schema/delta/14/v14.sql new file mode 100644 index 0000000000..a831920da6 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/14/v14.sql @@ -0,0 +1,23 @@ +/* Copyright 2015, 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ +CREATE TABLE IF NOT EXISTS push_rules_enable ( + id INTEGER PRIMARY KEY AUTOINCREMENT, + user_name TEXT NOT NULL, + rule_id TEXT NOT NULL, + enabled TINYINT, + UNIQUE(user_name, rule_id) +); + +CREATE INDEX IF NOT EXISTS push_rules_enable_user_name on push_rules_enable (user_name); diff --git a/synapse/storage/data_stores/main/schema/delta/15/appservice_txns.sql b/synapse/storage/data_stores/main/schema/delta/15/appservice_txns.sql new file mode 100644 index 0000000000..e4f5e76aec --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/15/appservice_txns.sql @@ -0,0 +1,31 @@ +/* Copyright 2015, 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE TABLE IF NOT EXISTS application_services_state( + as_id TEXT PRIMARY KEY, + state VARCHAR(5), + last_txn INTEGER +); + +CREATE TABLE IF NOT EXISTS application_services_txns( + as_id TEXT NOT NULL, + txn_id INTEGER NOT NULL, + event_ids TEXT NOT NULL, + UNIQUE(as_id, txn_id) +); + +CREATE INDEX IF NOT EXISTS application_services_txns_id ON application_services_txns ( + as_id +); diff --git a/synapse/storage/data_stores/main/schema/delta/15/presence_indices.sql b/synapse/storage/data_stores/main/schema/delta/15/presence_indices.sql new file mode 100644 index 0000000000..6b8d0f1ca7 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/15/presence_indices.sql @@ -0,0 +1,2 @@ + +CREATE INDEX IF NOT EXISTS presence_list_user_id ON presence_list (user_id); diff --git a/synapse/storage/data_stores/main/schema/delta/15/v15.sql b/synapse/storage/data_stores/main/schema/delta/15/v15.sql new file mode 100644 index 0000000000..9523d2bcc3 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/15/v15.sql @@ -0,0 +1,24 @@ +-- Drop, copy & recreate pushers table to change unique key +-- Also add access_token column at the same time +CREATE TABLE IF NOT EXISTS pushers2 ( + id BIGINT PRIMARY KEY, + user_name TEXT NOT NULL, + access_token BIGINT DEFAULT NULL, + profile_tag VARCHAR(32) NOT NULL, + kind VARCHAR(8) NOT NULL, + app_id VARCHAR(64) NOT NULL, + app_display_name VARCHAR(64) NOT NULL, + device_display_name VARCHAR(128) NOT NULL, + pushkey bytea NOT NULL, + ts BIGINT NOT NULL, + lang VARCHAR(8), + data bytea, + last_token TEXT, + last_success BIGINT, + failing_since BIGINT, + UNIQUE (app_id, pushkey) +); +INSERT INTO pushers2 (id, user_name, profile_tag, kind, app_id, app_display_name, device_display_name, pushkey, ts, lang, data, last_token, last_success, failing_since) + SELECT id, user_name, profile_tag, kind, app_id, app_display_name, device_display_name, pushkey, ts, lang, data, last_token, last_success, failing_since FROM pushers; +DROP TABLE pushers; +ALTER TABLE pushers2 RENAME TO pushers; diff --git a/synapse/storage/data_stores/main/schema/delta/16/events_order_index.sql b/synapse/storage/data_stores/main/schema/delta/16/events_order_index.sql new file mode 100644 index 0000000000..a48f215170 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/16/events_order_index.sql @@ -0,0 +1,4 @@ +CREATE INDEX events_order ON events (topological_ordering, stream_ordering); +CREATE INDEX events_order_room ON events ( + room_id, topological_ordering, stream_ordering +); diff --git a/synapse/storage/data_stores/main/schema/delta/16/remote_media_cache_index.sql b/synapse/storage/data_stores/main/schema/delta/16/remote_media_cache_index.sql new file mode 100644 index 0000000000..7a15265cb1 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/16/remote_media_cache_index.sql @@ -0,0 +1,2 @@ +CREATE INDEX IF NOT EXISTS remote_media_cache_thumbnails_media_id + ON remote_media_cache_thumbnails (media_id); \ No newline at end of file diff --git a/synapse/storage/data_stores/main/schema/delta/16/remove_duplicates.sql b/synapse/storage/data_stores/main/schema/delta/16/remove_duplicates.sql new file mode 100644 index 0000000000..65c97b5e2f --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/16/remove_duplicates.sql @@ -0,0 +1,9 @@ + + +DELETE FROM event_to_state_groups WHERE state_group not in ( + SELECT MAX(state_group) FROM event_to_state_groups GROUP BY event_id +); + +DELETE FROM event_to_state_groups WHERE rowid not in ( + SELECT MIN(rowid) FROM event_to_state_groups GROUP BY event_id +); diff --git a/synapse/storage/data_stores/main/schema/delta/16/room_alias_index.sql b/synapse/storage/data_stores/main/schema/delta/16/room_alias_index.sql new file mode 100644 index 0000000000..f82486132b --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/16/room_alias_index.sql @@ -0,0 +1,3 @@ + +CREATE INDEX IF NOT EXISTS room_aliases_id ON room_aliases(room_id); +CREATE INDEX IF NOT EXISTS room_alias_servers_alias ON room_alias_servers(room_alias); diff --git a/synapse/storage/data_stores/main/schema/delta/16/unique_constraints.sql b/synapse/storage/data_stores/main/schema/delta/16/unique_constraints.sql new file mode 100644 index 0000000000..5b8de52c33 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/16/unique_constraints.sql @@ -0,0 +1,72 @@ + +-- We can use SQLite features here, since other db support was only added in v16 + +-- +DELETE FROM current_state_events WHERE rowid not in ( + SELECT MIN(rowid) FROM current_state_events GROUP BY event_id +); + +DROP INDEX IF EXISTS current_state_events_event_id; +CREATE UNIQUE INDEX current_state_events_event_id ON current_state_events(event_id); + +-- +DELETE FROM room_memberships WHERE rowid not in ( + SELECT MIN(rowid) FROM room_memberships GROUP BY event_id +); + +DROP INDEX IF EXISTS room_memberships_event_id; +CREATE UNIQUE INDEX room_memberships_event_id ON room_memberships(event_id); + +-- +DELETE FROM topics WHERE rowid not in ( + SELECT MIN(rowid) FROM topics GROUP BY event_id +); + +DROP INDEX IF EXISTS topics_event_id; +CREATE UNIQUE INDEX topics_event_id ON topics(event_id); + +-- +DELETE FROM room_names WHERE rowid not in ( + SELECT MIN(rowid) FROM room_names GROUP BY event_id +); + +DROP INDEX IF EXISTS room_names_id; +CREATE UNIQUE INDEX room_names_id ON room_names(event_id); + +-- +DELETE FROM presence WHERE rowid not in ( + SELECT MIN(rowid) FROM presence GROUP BY user_id +); + +DROP INDEX IF EXISTS presence_id; +CREATE UNIQUE INDEX presence_id ON presence(user_id); + +-- +DELETE FROM presence_allow_inbound WHERE rowid not in ( + SELECT MIN(rowid) FROM presence_allow_inbound + GROUP BY observed_user_id, observer_user_id +); + +DROP INDEX IF EXISTS presence_allow_inbound_observers; +CREATE UNIQUE INDEX presence_allow_inbound_observers ON presence_allow_inbound( + observed_user_id, observer_user_id +); + +-- +DELETE FROM presence_list WHERE rowid not in ( + SELECT MIN(rowid) FROM presence_list + GROUP BY user_id, observed_user_id +); + +DROP INDEX IF EXISTS presence_list_observers; +CREATE UNIQUE INDEX presence_list_observers ON presence_list( + user_id, observed_user_id +); + +-- +DELETE FROM room_aliases WHERE rowid not in ( + SELECT MIN(rowid) FROM room_aliases GROUP BY room_alias +); + +DROP INDEX IF EXISTS room_aliases_id; +CREATE INDEX room_aliases_id ON room_aliases(room_id); diff --git a/synapse/storage/data_stores/main/schema/delta/16/users.sql b/synapse/storage/data_stores/main/schema/delta/16/users.sql new file mode 100644 index 0000000000..cd0709250d --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/16/users.sql @@ -0,0 +1,56 @@ +-- Convert `access_tokens`.user from rowids to user strings. +-- MUST BE DONE BEFORE REMOVING ID COLUMN FROM USERS TABLE BELOW +CREATE TABLE IF NOT EXISTS new_access_tokens( + id BIGINT UNSIGNED PRIMARY KEY, + user_id TEXT NOT NULL, + device_id TEXT, + token TEXT NOT NULL, + last_used BIGINT UNSIGNED, + UNIQUE(token) +); + +INSERT INTO new_access_tokens + SELECT a.id, u.name, a.device_id, a.token, a.last_used + FROM access_tokens as a + INNER JOIN users as u ON u.id = a.user_id; + +DROP TABLE access_tokens; + +ALTER TABLE new_access_tokens RENAME TO access_tokens; + +-- Remove ID column from `users` table +CREATE TABLE IF NOT EXISTS new_users( + name TEXT, + password_hash TEXT, + creation_ts BIGINT UNSIGNED, + admin BOOL DEFAULT 0 NOT NULL, + UNIQUE(name) +); + +INSERT INTO new_users SELECT name, password_hash, creation_ts, admin FROM users; + +DROP TABLE users; + +ALTER TABLE new_users RENAME TO users; + + +-- Remove UNIQUE constraint from `user_ips` table +CREATE TABLE IF NOT EXISTS new_user_ips ( + user_id TEXT NOT NULL, + access_token TEXT NOT NULL, + device_id TEXT, + ip TEXT NOT NULL, + user_agent TEXT NOT NULL, + last_seen BIGINT UNSIGNED NOT NULL +); + +INSERT INTO new_user_ips + SELECT user, access_token, device_id, ip, user_agent, last_seen FROM user_ips; + +DROP TABLE user_ips; + +ALTER TABLE new_user_ips RENAME TO user_ips; + +CREATE INDEX IF NOT EXISTS user_ips_user ON user_ips(user_id); +CREATE INDEX IF NOT EXISTS user_ips_user_ip ON user_ips(user_id, access_token, ip); + diff --git a/synapse/storage/data_stores/main/schema/delta/17/drop_indexes.sql b/synapse/storage/data_stores/main/schema/delta/17/drop_indexes.sql new file mode 100644 index 0000000000..7c9a90e27f --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/17/drop_indexes.sql @@ -0,0 +1,18 @@ +/* Copyright 2015, 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +DROP INDEX IF EXISTS sent_transaction_dest; +DROP INDEX IF EXISTS sent_transaction_sent; +DROP INDEX IF EXISTS user_ips_user; diff --git a/synapse/storage/data_stores/main/schema/delta/17/server_keys.sql b/synapse/storage/data_stores/main/schema/delta/17/server_keys.sql new file mode 100644 index 0000000000..70b247a06b --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/17/server_keys.sql @@ -0,0 +1,24 @@ +/* Copyright 2015, 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE TABLE IF NOT EXISTS server_keys_json ( + server_name TEXT, -- Server name. + key_id TEXT, -- Requested key id. + from_server TEXT, -- Which server the keys were fetched from. + ts_added_ms INTEGER, -- When the keys were fetched + ts_valid_until_ms INTEGER, -- When this version of the keys exipires. + key_json bytea, -- JSON certificate for the remote server. + CONSTRAINT uniqueness UNIQUE (server_name, key_id, from_server) +); diff --git a/synapse/storage/data_stores/main/schema/delta/17/user_threepids.sql b/synapse/storage/data_stores/main/schema/delta/17/user_threepids.sql new file mode 100644 index 0000000000..c17715ac80 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/17/user_threepids.sql @@ -0,0 +1,9 @@ +CREATE TABLE user_threepids ( + user_id TEXT NOT NULL, + medium TEXT NOT NULL, + address TEXT NOT NULL, + validated_at BIGINT NOT NULL, + added_at BIGINT NOT NULL, + CONSTRAINT user_medium_address UNIQUE (user_id, medium, address) +); +CREATE INDEX user_threepids_user_id ON user_threepids(user_id); diff --git a/synapse/storage/data_stores/main/schema/delta/18/server_keys_bigger_ints.sql b/synapse/storage/data_stores/main/schema/delta/18/server_keys_bigger_ints.sql new file mode 100644 index 0000000000..6e0871c92b --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/18/server_keys_bigger_ints.sql @@ -0,0 +1,32 @@ +/* Copyright 2015, 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + +CREATE TABLE IF NOT EXISTS new_server_keys_json ( + server_name TEXT NOT NULL, -- Server name. + key_id TEXT NOT NULL, -- Requested key id. + from_server TEXT NOT NULL, -- Which server the keys were fetched from. + ts_added_ms BIGINT NOT NULL, -- When the keys were fetched + ts_valid_until_ms BIGINT NOT NULL, -- When this version of the keys exipires. + key_json bytea NOT NULL, -- JSON certificate for the remote server. + CONSTRAINT server_keys_json_uniqueness UNIQUE (server_name, key_id, from_server) +); + +INSERT INTO new_server_keys_json + SELECT server_name, key_id, from_server,ts_added_ms, ts_valid_until_ms, key_json FROM server_keys_json ; + +DROP TABLE server_keys_json; + +ALTER TABLE new_server_keys_json RENAME TO server_keys_json; diff --git a/synapse/storage/data_stores/main/schema/delta/19/event_index.sql b/synapse/storage/data_stores/main/schema/delta/19/event_index.sql new file mode 100644 index 0000000000..18b97b4332 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/19/event_index.sql @@ -0,0 +1,19 @@ +/* Copyright 2015, 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + +CREATE INDEX events_order_topo_stream_room ON events( + topological_ordering, stream_ordering, room_id +); diff --git a/synapse/storage/data_stores/main/schema/delta/20/dummy.sql b/synapse/storage/data_stores/main/schema/delta/20/dummy.sql new file mode 100644 index 0000000000..e0ac49d1ec --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/20/dummy.sql @@ -0,0 +1 @@ +SELECT 1; diff --git a/synapse/storage/data_stores/main/schema/delta/20/pushers.py b/synapse/storage/data_stores/main/schema/delta/20/pushers.py new file mode 100644 index 0000000000..3edfcfd783 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/20/pushers.py @@ -0,0 +1,88 @@ +# Copyright 2015, 2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + + +""" +Main purpose of this upgrade is to change the unique key on the +pushers table again (it was missed when the v16 full schema was +made) but this also changes the pushkey and data columns to text. +When selecting a bytea column into a text column, postgres inserts +the hex encoded data, and there's no portable way of getting the +UTF-8 bytes, so we have to do it in Python. +""" + +import logging + +logger = logging.getLogger(__name__) + + +def run_create(cur, database_engine, *args, **kwargs): + logger.info("Porting pushers table...") + cur.execute( + """ + CREATE TABLE IF NOT EXISTS pushers2 ( + id BIGINT PRIMARY KEY, + user_name TEXT NOT NULL, + access_token BIGINT DEFAULT NULL, + profile_tag VARCHAR(32) NOT NULL, + kind VARCHAR(8) NOT NULL, + app_id VARCHAR(64) NOT NULL, + app_display_name VARCHAR(64) NOT NULL, + device_display_name VARCHAR(128) NOT NULL, + pushkey TEXT NOT NULL, + ts BIGINT NOT NULL, + lang VARCHAR(8), + data TEXT, + last_token TEXT, + last_success BIGINT, + failing_since BIGINT, + UNIQUE (app_id, pushkey, user_name) + ) + """ + ) + cur.execute( + """SELECT + id, user_name, access_token, profile_tag, kind, + app_id, app_display_name, device_display_name, + pushkey, ts, lang, data, last_token, last_success, + failing_since + FROM pushers + """ + ) + count = 0 + for row in cur.fetchall(): + row = list(row) + row[8] = bytes(row[8]).decode("utf-8") + row[11] = bytes(row[11]).decode("utf-8") + cur.execute( + database_engine.convert_param_style( + """ + INSERT into pushers2 ( + id, user_name, access_token, profile_tag, kind, + app_id, app_display_name, device_display_name, + pushkey, ts, lang, data, last_token, last_success, + failing_since + ) values (%s)""" + % (",".join(["?" for _ in range(len(row))])) + ), + row, + ) + count += 1 + cur.execute("DROP TABLE pushers") + cur.execute("ALTER TABLE pushers2 RENAME TO pushers") + logger.info("Moved %d pushers to new table", count) + + +def run_upgrade(*args, **kwargs): + pass diff --git a/synapse/storage/data_stores/main/schema/delta/21/end_to_end_keys.sql b/synapse/storage/data_stores/main/schema/delta/21/end_to_end_keys.sql new file mode 100644 index 0000000000..4c2fb20b77 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/21/end_to_end_keys.sql @@ -0,0 +1,34 @@ +/* Copyright 2015, 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + +CREATE TABLE IF NOT EXISTS e2e_device_keys_json ( + user_id TEXT NOT NULL, -- The user these keys are for. + device_id TEXT NOT NULL, -- Which of the user's devices these keys are for. + ts_added_ms BIGINT NOT NULL, -- When the keys were uploaded. + key_json TEXT NOT NULL, -- The keys for the device as a JSON blob. + CONSTRAINT e2e_device_keys_json_uniqueness UNIQUE (user_id, device_id) +); + + +CREATE TABLE IF NOT EXISTS e2e_one_time_keys_json ( + user_id TEXT NOT NULL, -- The user this one-time key is for. + device_id TEXT NOT NULL, -- The device this one-time key is for. + algorithm TEXT NOT NULL, -- Which algorithm this one-time key is for. + key_id TEXT NOT NULL, -- An id for suppressing duplicate uploads. + ts_added_ms BIGINT NOT NULL, -- When this key was uploaded. + key_json TEXT NOT NULL, -- The key as a JSON blob. + CONSTRAINT e2e_one_time_keys_json_uniqueness UNIQUE (user_id, device_id, algorithm, key_id) +); diff --git a/synapse/storage/data_stores/main/schema/delta/21/receipts.sql b/synapse/storage/data_stores/main/schema/delta/21/receipts.sql new file mode 100644 index 0000000000..d070845477 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/21/receipts.sql @@ -0,0 +1,38 @@ +/* Copyright 2015, 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + +CREATE TABLE IF NOT EXISTS receipts_graph( + room_id TEXT NOT NULL, + receipt_type TEXT NOT NULL, + user_id TEXT NOT NULL, + event_ids TEXT NOT NULL, + data TEXT NOT NULL, + CONSTRAINT receipts_graph_uniqueness UNIQUE (room_id, receipt_type, user_id) +); + +CREATE TABLE IF NOT EXISTS receipts_linearized ( + stream_id BIGINT NOT NULL, + room_id TEXT NOT NULL, + receipt_type TEXT NOT NULL, + user_id TEXT NOT NULL, + event_id TEXT NOT NULL, + data TEXT NOT NULL, + CONSTRAINT receipts_linearized_uniqueness UNIQUE (room_id, receipt_type, user_id) +); + +CREATE INDEX receipts_linearized_id ON receipts_linearized( + stream_id +); diff --git a/synapse/storage/data_stores/main/schema/delta/22/receipts_index.sql b/synapse/storage/data_stores/main/schema/delta/22/receipts_index.sql new file mode 100644 index 0000000000..bfc0b3bcaa --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/22/receipts_index.sql @@ -0,0 +1,22 @@ +/* Copyright 2015, 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/** Using CREATE INDEX directly is deprecated in favour of using background + * update see synapse/storage/schema/delta/33/access_tokens_device_index.sql + * and synapse/storage/registration.py for an example using + * "access_tokens_device_index" **/ +CREATE INDEX receipts_linearized_room_stream ON receipts_linearized( + room_id, stream_id +); diff --git a/synapse/storage/data_stores/main/schema/delta/22/user_threepids_unique.sql b/synapse/storage/data_stores/main/schema/delta/22/user_threepids_unique.sql new file mode 100644 index 0000000000..87edfa454c --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/22/user_threepids_unique.sql @@ -0,0 +1,19 @@ +CREATE TABLE IF NOT EXISTS user_threepids2 ( + user_id TEXT NOT NULL, + medium TEXT NOT NULL, + address TEXT NOT NULL, + validated_at BIGINT NOT NULL, + added_at BIGINT NOT NULL, + CONSTRAINT medium_address UNIQUE (medium, address) +); + +INSERT INTO user_threepids2 + SELECT * FROM user_threepids WHERE added_at IN ( + SELECT max(added_at) FROM user_threepids GROUP BY medium, address + ) +; + +DROP TABLE user_threepids; +ALTER TABLE user_threepids2 RENAME TO user_threepids; + +CREATE INDEX user_threepids_user_id ON user_threepids(user_id); diff --git a/synapse/storage/data_stores/main/schema/delta/23/drop_state_index.sql b/synapse/storage/data_stores/main/schema/delta/23/drop_state_index.sql new file mode 100644 index 0000000000..ae09fa0065 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/23/drop_state_index.sql @@ -0,0 +1,16 @@ +/* Copyright 2015, 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +DROP INDEX IF EXISTS state_groups_state_tuple; diff --git a/synapse/storage/data_stores/main/schema/delta/24/stats_reporting.sql b/synapse/storage/data_stores/main/schema/delta/24/stats_reporting.sql new file mode 100644 index 0000000000..acea7483bd --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/24/stats_reporting.sql @@ -0,0 +1,18 @@ +/* Copyright 2019 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + /* We used to create a table called stats_reporting, but this is no longer + * used and is removed in delta 54. + */ \ No newline at end of file diff --git a/synapse/storage/data_stores/main/schema/delta/25/00background_updates.sql b/synapse/storage/data_stores/main/schema/delta/25/00background_updates.sql new file mode 100644 index 0000000000..2ad9e8fa56 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/25/00background_updates.sql @@ -0,0 +1,21 @@ +/* Copyright 2015, 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + +CREATE TABLE IF NOT EXISTS background_updates( + update_name TEXT NOT NULL, -- The name of the background update. + progress_json TEXT NOT NULL, -- The current progress of the update as JSON. + CONSTRAINT background_updates_uniqueness UNIQUE (update_name) +); diff --git a/synapse/storage/data_stores/main/schema/delta/25/fts.py b/synapse/storage/data_stores/main/schema/delta/25/fts.py new file mode 100644 index 0000000000..4b2ffd35fd --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/25/fts.py @@ -0,0 +1,82 @@ +# Copyright 2015, 2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging + +import simplejson + +from synapse.storage.engines import PostgresEngine, Sqlite3Engine +from synapse.storage.prepare_database import get_statements + +logger = logging.getLogger(__name__) + + +POSTGRES_TABLE = """ +CREATE TABLE IF NOT EXISTS event_search ( + event_id TEXT, + room_id TEXT, + sender TEXT, + key TEXT, + vector tsvector +); + +CREATE INDEX event_search_fts_idx ON event_search USING gin(vector); +CREATE INDEX event_search_ev_idx ON event_search(event_id); +CREATE INDEX event_search_ev_ridx ON event_search(room_id); +""" + + +SQLITE_TABLE = ( + "CREATE VIRTUAL TABLE event_search" + " USING fts4 ( event_id, room_id, sender, key, value )" +) + + +def run_create(cur, database_engine, *args, **kwargs): + if isinstance(database_engine, PostgresEngine): + for statement in get_statements(POSTGRES_TABLE.splitlines()): + cur.execute(statement) + elif isinstance(database_engine, Sqlite3Engine): + cur.execute(SQLITE_TABLE) + else: + raise Exception("Unrecognized database engine") + + cur.execute("SELECT MIN(stream_ordering) FROM events") + rows = cur.fetchall() + min_stream_id = rows[0][0] + + cur.execute("SELECT MAX(stream_ordering) FROM events") + rows = cur.fetchall() + max_stream_id = rows[0][0] + + if min_stream_id is not None and max_stream_id is not None: + progress = { + "target_min_stream_id_inclusive": min_stream_id, + "max_stream_id_exclusive": max_stream_id + 1, + "rows_inserted": 0, + } + progress_json = simplejson.dumps(progress) + + sql = ( + "INSERT into background_updates (update_name, progress_json)" + " VALUES (?, ?)" + ) + + sql = database_engine.convert_param_style(sql) + + cur.execute(sql, ("event_search", progress_json)) + + +def run_upgrade(*args, **kwargs): + pass diff --git a/synapse/storage/data_stores/main/schema/delta/25/guest_access.sql b/synapse/storage/data_stores/main/schema/delta/25/guest_access.sql new file mode 100644 index 0000000000..1ea389b471 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/25/guest_access.sql @@ -0,0 +1,25 @@ +/* Copyright 2015, 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* + * This is a manual index of guest_access content of state events, + * so that we can join on them in SELECT statements. + */ +CREATE TABLE IF NOT EXISTS guest_access( + event_id TEXT NOT NULL, + room_id TEXT NOT NULL, + guest_access TEXT NOT NULL, + UNIQUE (event_id) +); diff --git a/synapse/storage/data_stores/main/schema/delta/25/history_visibility.sql b/synapse/storage/data_stores/main/schema/delta/25/history_visibility.sql new file mode 100644 index 0000000000..f468fc1897 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/25/history_visibility.sql @@ -0,0 +1,25 @@ +/* Copyright 2015, 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* + * This is a manual index of history_visibility content of state events, + * so that we can join on them in SELECT statements. + */ +CREATE TABLE IF NOT EXISTS history_visibility( + event_id TEXT NOT NULL, + room_id TEXT NOT NULL, + history_visibility TEXT NOT NULL, + UNIQUE (event_id) +); diff --git a/synapse/storage/data_stores/main/schema/delta/25/tags.sql b/synapse/storage/data_stores/main/schema/delta/25/tags.sql new file mode 100644 index 0000000000..7a32ce68e4 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/25/tags.sql @@ -0,0 +1,38 @@ +/* Copyright 2015, 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + +CREATE TABLE IF NOT EXISTS room_tags( + user_id TEXT NOT NULL, + room_id TEXT NOT NULL, + tag TEXT NOT NULL, -- The name of the tag. + content TEXT NOT NULL, -- The JSON content of the tag. + CONSTRAINT room_tag_uniqueness UNIQUE (user_id, room_id, tag) +); + +CREATE TABLE IF NOT EXISTS room_tags_revisions ( + user_id TEXT NOT NULL, + room_id TEXT NOT NULL, + stream_id BIGINT NOT NULL, -- The current version of the room tags. + CONSTRAINT room_tag_revisions_uniqueness UNIQUE (user_id, room_id) +); + +CREATE TABLE IF NOT EXISTS private_user_data_max_stream_id( + Lock CHAR(1) NOT NULL DEFAULT 'X' UNIQUE, -- Makes sure this table only has one row. + stream_id BIGINT NOT NULL, + CHECK (Lock='X') +); + +INSERT INTO private_user_data_max_stream_id (stream_id) VALUES (0); diff --git a/synapse/storage/data_stores/main/schema/delta/26/account_data.sql b/synapse/storage/data_stores/main/schema/delta/26/account_data.sql new file mode 100644 index 0000000000..e395de2b5e --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/26/account_data.sql @@ -0,0 +1,17 @@ +/* Copyright 2015, 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + +ALTER TABLE private_user_data_max_stream_id RENAME TO account_data_max_stream_id; diff --git a/synapse/storage/data_stores/main/schema/delta/27/account_data.sql b/synapse/storage/data_stores/main/schema/delta/27/account_data.sql new file mode 100644 index 0000000000..bf0558b5b3 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/27/account_data.sql @@ -0,0 +1,36 @@ +/* Copyright 2015, 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE TABLE IF NOT EXISTS account_data( + user_id TEXT NOT NULL, + account_data_type TEXT NOT NULL, -- The type of the account_data. + stream_id BIGINT NOT NULL, -- The version of the account_data. + content TEXT NOT NULL, -- The JSON content of the account_data + CONSTRAINT account_data_uniqueness UNIQUE (user_id, account_data_type) +); + + +CREATE TABLE IF NOT EXISTS room_account_data( + user_id TEXT NOT NULL, + room_id TEXT NOT NULL, + account_data_type TEXT NOT NULL, -- The type of the account_data. + stream_id BIGINT NOT NULL, -- The version of the account_data. + content TEXT NOT NULL, -- The JSON content of the account_data + CONSTRAINT room_account_data_uniqueness UNIQUE (user_id, room_id, account_data_type) +); + + +CREATE INDEX account_data_stream_id on account_data(user_id, stream_id); +CREATE INDEX room_account_data_stream_id on room_account_data(user_id, stream_id); diff --git a/synapse/storage/data_stores/main/schema/delta/27/forgotten_memberships.sql b/synapse/storage/data_stores/main/schema/delta/27/forgotten_memberships.sql new file mode 100644 index 0000000000..e2094f37fe --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/27/forgotten_memberships.sql @@ -0,0 +1,26 @@ +/* Copyright 2015, 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* + * Keeps track of what rooms users have left and don't want to be able to + * access again. + * + * If all users on this server have left a room, we can delete the room + * entirely. + * + * This column should always contain either 0 or 1. + */ + + ALTER TABLE room_memberships ADD COLUMN forgotten INTEGER DEFAULT 0; diff --git a/synapse/storage/data_stores/main/schema/delta/27/ts.py b/synapse/storage/data_stores/main/schema/delta/27/ts.py new file mode 100644 index 0000000000..414f9f5aa0 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/27/ts.py @@ -0,0 +1,61 @@ +# Copyright 2015, 2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging + +import simplejson + +from synapse.storage.prepare_database import get_statements + +logger = logging.getLogger(__name__) + + +ALTER_TABLE = ( + "ALTER TABLE events ADD COLUMN origin_server_ts BIGINT;" + "CREATE INDEX events_ts ON events(origin_server_ts, stream_ordering);" +) + + +def run_create(cur, database_engine, *args, **kwargs): + for statement in get_statements(ALTER_TABLE.splitlines()): + cur.execute(statement) + + cur.execute("SELECT MIN(stream_ordering) FROM events") + rows = cur.fetchall() + min_stream_id = rows[0][0] + + cur.execute("SELECT MAX(stream_ordering) FROM events") + rows = cur.fetchall() + max_stream_id = rows[0][0] + + if min_stream_id is not None and max_stream_id is not None: + progress = { + "target_min_stream_id_inclusive": min_stream_id, + "max_stream_id_exclusive": max_stream_id + 1, + "rows_inserted": 0, + } + progress_json = simplejson.dumps(progress) + + sql = ( + "INSERT into background_updates (update_name, progress_json)" + " VALUES (?, ?)" + ) + + sql = database_engine.convert_param_style(sql) + + cur.execute(sql, ("event_origin_server_ts", progress_json)) + + +def run_upgrade(*args, **kwargs): + pass diff --git a/synapse/storage/data_stores/main/schema/delta/28/event_push_actions.sql b/synapse/storage/data_stores/main/schema/delta/28/event_push_actions.sql new file mode 100644 index 0000000000..4d519849df --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/28/event_push_actions.sql @@ -0,0 +1,27 @@ +/* Copyright 2015 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE TABLE IF NOT EXISTS event_push_actions( + room_id TEXT NOT NULL, + event_id TEXT NOT NULL, + user_id TEXT NOT NULL, + profile_tag VARCHAR(32), + actions TEXT NOT NULL, + CONSTRAINT event_id_user_id_profile_tag_uniqueness UNIQUE (room_id, event_id, user_id, profile_tag) +); + + +CREATE INDEX event_push_actions_room_id_event_id_user_id_profile_tag on event_push_actions(room_id, event_id, user_id, profile_tag); +CREATE INDEX event_push_actions_room_id_user_id on event_push_actions(room_id, user_id); diff --git a/synapse/storage/data_stores/main/schema/delta/28/events_room_stream.sql b/synapse/storage/data_stores/main/schema/delta/28/events_room_stream.sql new file mode 100644 index 0000000000..36609475f1 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/28/events_room_stream.sql @@ -0,0 +1,20 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. +*/ + +/** Using CREATE INDEX directly is deprecated in favour of using background + * update see synapse/storage/schema/delta/33/access_tokens_device_index.sql + * and synapse/storage/registration.py for an example using + * "access_tokens_device_index" **/ +CREATE INDEX events_room_stream on events(room_id, stream_ordering); diff --git a/synapse/storage/data_stores/main/schema/delta/28/public_roms_index.sql b/synapse/storage/data_stores/main/schema/delta/28/public_roms_index.sql new file mode 100644 index 0000000000..6c1fd68c5b --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/28/public_roms_index.sql @@ -0,0 +1,20 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. +*/ + +/** Using CREATE INDEX directly is deprecated in favour of using background + * update see synapse/storage/schema/delta/33/access_tokens_device_index.sql + * and synapse/storage/registration.py for an example using + * "access_tokens_device_index" **/ +CREATE INDEX public_room_index on rooms(is_public); diff --git a/synapse/storage/data_stores/main/schema/delta/28/receipts_user_id_index.sql b/synapse/storage/data_stores/main/schema/delta/28/receipts_user_id_index.sql new file mode 100644 index 0000000000..cb84c69baa --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/28/receipts_user_id_index.sql @@ -0,0 +1,22 @@ +/* Copyright 2015, 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/** Using CREATE INDEX directly is deprecated in favour of using background + * update see synapse/storage/schema/delta/33/access_tokens_device_index.sql + * and synapse/storage/registration.py for an example using + * "access_tokens_device_index" **/ +CREATE INDEX receipts_linearized_user ON receipts_linearized( + user_id +); diff --git a/synapse/storage/data_stores/main/schema/delta/28/upgrade_times.sql b/synapse/storage/data_stores/main/schema/delta/28/upgrade_times.sql new file mode 100644 index 0000000000..3e4a9ab455 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/28/upgrade_times.sql @@ -0,0 +1,21 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* + * Stores the timestamp when a user upgraded from a guest to a full user, if + * that happened. + */ + +ALTER TABLE users ADD COLUMN upgrade_ts BIGINT; diff --git a/synapse/storage/data_stores/main/schema/delta/28/users_is_guest.sql b/synapse/storage/data_stores/main/schema/delta/28/users_is_guest.sql new file mode 100644 index 0000000000..21d2b420bf --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/28/users_is_guest.sql @@ -0,0 +1,22 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +ALTER TABLE users ADD is_guest SMALLINT DEFAULT 0 NOT NULL; +/* + * NB: any guest users created between 27 and 28 will be incorrectly + * marked as not guests: we don't bother to fill these in correctly + * because guest access is not really complete in 27 anyway so it's + * very unlikley there will be any guest users created. + */ diff --git a/synapse/storage/data_stores/main/schema/delta/29/push_actions.sql b/synapse/storage/data_stores/main/schema/delta/29/push_actions.sql new file mode 100644 index 0000000000..84b21cf813 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/29/push_actions.sql @@ -0,0 +1,35 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +ALTER TABLE event_push_actions ADD COLUMN topological_ordering BIGINT; +ALTER TABLE event_push_actions ADD COLUMN stream_ordering BIGINT; +ALTER TABLE event_push_actions ADD COLUMN notif SMALLINT; +ALTER TABLE event_push_actions ADD COLUMN highlight SMALLINT; + +UPDATE event_push_actions SET stream_ordering = ( + SELECT stream_ordering FROM events WHERE event_id = event_push_actions.event_id +), topological_ordering = ( + SELECT topological_ordering FROM events WHERE event_id = event_push_actions.event_id +); + +UPDATE event_push_actions SET notif = 1, highlight = 0; + +/** Using CREATE INDEX directly is deprecated in favour of using background + * update see synapse/storage/schema/delta/33/access_tokens_device_index.sql + * and synapse/storage/registration.py for an example using + * "access_tokens_device_index" **/ +CREATE INDEX event_push_actions_rm_tokens on event_push_actions( + user_id, room_id, topological_ordering, stream_ordering +); diff --git a/synapse/storage/data_stores/main/schema/delta/30/alias_creator.sql b/synapse/storage/data_stores/main/schema/delta/30/alias_creator.sql new file mode 100644 index 0000000000..c9d0dde638 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/30/alias_creator.sql @@ -0,0 +1,16 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +ALTER TABLE room_aliases ADD COLUMN creator TEXT; diff --git a/synapse/storage/data_stores/main/schema/delta/30/as_users.py b/synapse/storage/data_stores/main/schema/delta/30/as_users.py new file mode 100644 index 0000000000..9b95411fb6 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/30/as_users.py @@ -0,0 +1,69 @@ +# Copyright 2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import logging + +from six.moves import range + +from synapse.config.appservice import load_appservices + +logger = logging.getLogger(__name__) + + +def run_create(cur, database_engine, *args, **kwargs): + # NULL indicates user was not registered by an appservice. + try: + cur.execute("ALTER TABLE users ADD COLUMN appservice_id TEXT") + except Exception: + # Maybe we already added the column? Hope so... + pass + + +def run_upgrade(cur, database_engine, config, *args, **kwargs): + cur.execute("SELECT name FROM users") + rows = cur.fetchall() + + config_files = [] + try: + config_files = config.app_service_config_files + except AttributeError: + logger.warning("Could not get app_service_config_files from config") + pass + + appservices = load_appservices(config.server_name, config_files) + + owned = {} + + for row in rows: + user_id = row[0] + for appservice in appservices: + if appservice.is_exclusive_user(user_id): + if user_id in owned.keys(): + logger.error( + "user_id %s was owned by more than one application" + " service (IDs %s and %s); assigning arbitrarily to %s" + % (user_id, owned[user_id], appservice.id, owned[user_id]) + ) + owned.setdefault(appservice.id, []).append(user_id) + + for as_id, user_ids in owned.items(): + n = 100 + user_chunks = (user_ids[i : i + 100] for i in range(0, len(user_ids), n)) + for chunk in user_chunks: + cur.execute( + database_engine.convert_param_style( + "UPDATE users SET appservice_id = ? WHERE name IN (%s)" + % (",".join("?" for _ in chunk),) + ), + [as_id] + chunk, + ) diff --git a/synapse/storage/data_stores/main/schema/delta/30/deleted_pushers.sql b/synapse/storage/data_stores/main/schema/delta/30/deleted_pushers.sql new file mode 100644 index 0000000000..712c454aa1 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/30/deleted_pushers.sql @@ -0,0 +1,25 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE TABLE IF NOT EXISTS deleted_pushers( + stream_id BIGINT NOT NULL, + app_id TEXT NOT NULL, + pushkey TEXT NOT NULL, + user_id TEXT NOT NULL, + /* We only track the most recent delete for each app_id, pushkey and user_id. */ + UNIQUE (app_id, pushkey, user_id) +); + +CREATE INDEX deleted_pushers_stream_id ON deleted_pushers (stream_id); diff --git a/synapse/storage/data_stores/main/schema/delta/30/presence_stream.sql b/synapse/storage/data_stores/main/schema/delta/30/presence_stream.sql new file mode 100644 index 0000000000..606bbb037d --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/30/presence_stream.sql @@ -0,0 +1,30 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + + CREATE TABLE presence_stream( + stream_id BIGINT, + user_id TEXT, + state TEXT, + last_active_ts BIGINT, + last_federation_update_ts BIGINT, + last_user_sync_ts BIGINT, + status_msg TEXT, + currently_active BOOLEAN + ); + + CREATE INDEX presence_stream_id ON presence_stream(stream_id, user_id); + CREATE INDEX presence_stream_user_id ON presence_stream(user_id); + CREATE INDEX presence_stream_state ON presence_stream(state); diff --git a/synapse/storage/data_stores/main/schema/delta/30/public_rooms.sql b/synapse/storage/data_stores/main/schema/delta/30/public_rooms.sql new file mode 100644 index 0000000000..f09db4faa6 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/30/public_rooms.sql @@ -0,0 +1,23 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + +/* This release removes the restriction that published rooms must have an alias, + * so we go back and ensure the only 'public' rooms are ones with an alias. + * We use (1 = 0) and (1 = 1) so that it works in both postgres and sqlite + */ +UPDATE rooms SET is_public = (1 = 0) WHERE is_public = (1 = 1) AND room_id not in ( + SELECT room_id FROM room_aliases +); diff --git a/synapse/storage/data_stores/main/schema/delta/30/push_rule_stream.sql b/synapse/storage/data_stores/main/schema/delta/30/push_rule_stream.sql new file mode 100644 index 0000000000..735aa8d5f6 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/30/push_rule_stream.sql @@ -0,0 +1,38 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + + +CREATE TABLE push_rules_stream( + stream_id BIGINT NOT NULL, + event_stream_ordering BIGINT NOT NULL, + user_id TEXT NOT NULL, + rule_id TEXT NOT NULL, + op TEXT NOT NULL, -- One of "ENABLE", "DISABLE", "ACTIONS", "ADD", "DELETE" + priority_class SMALLINT, + priority INTEGER, + conditions TEXT, + actions TEXT +); + +-- The extra data for each operation is: +-- * ENABLE, DISABLE, DELETE: [] +-- * ACTIONS: ["actions"] +-- * ADD: ["priority_class", "priority", "actions", "conditions"] + +-- Index for replication queries. +CREATE INDEX push_rules_stream_id ON push_rules_stream(stream_id); +-- Index for /sync queries. +CREATE INDEX push_rules_stream_user_stream_id on push_rules_stream(user_id, stream_id); diff --git a/synapse/storage/data_stores/main/schema/delta/30/state_stream.sql b/synapse/storage/data_stores/main/schema/delta/30/state_stream.sql new file mode 100644 index 0000000000..e85699e82e --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/30/state_stream.sql @@ -0,0 +1,33 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + +/* We used to create a table called current_state_resets, but this is no + * longer used and is removed in delta 54. + */ + +/* The outlier events that have aquired a state group typically through + * backfill. This is tracked separately to the events table, as assigning a + * state group change the position of the existing event in the stream + * ordering. + * However since a stream_ordering is assigned in persist_event for the + * (event, state) pair, we can use that stream_ordering to identify when + * the new state was assigned for the event. + */ +CREATE TABLE IF NOT EXISTS ex_outlier_stream( + event_stream_ordering BIGINT PRIMARY KEY NOT NULL, + event_id TEXT NOT NULL, + state_group BIGINT NOT NULL +); diff --git a/synapse/storage/data_stores/main/schema/delta/30/threepid_guest_access_tokens.sql b/synapse/storage/data_stores/main/schema/delta/30/threepid_guest_access_tokens.sql new file mode 100644 index 0000000000..0dd2f1360c --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/30/threepid_guest_access_tokens.sql @@ -0,0 +1,24 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- Stores guest account access tokens generated for unbound 3pids. +CREATE TABLE threepid_guest_access_tokens( + medium TEXT, -- The medium of the 3pid. Must be "email". + address TEXT, -- The 3pid address. + guest_access_token TEXT, -- The access token for a guest user for this 3pid. + first_inviter TEXT -- User ID of the first user to invite this 3pid to a room. +); + +CREATE UNIQUE INDEX threepid_guest_access_tokens_index ON threepid_guest_access_tokens(medium, address); diff --git a/synapse/storage/data_stores/main/schema/delta/31/invites.sql b/synapse/storage/data_stores/main/schema/delta/31/invites.sql new file mode 100644 index 0000000000..2c57846d5a --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/31/invites.sql @@ -0,0 +1,42 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + +CREATE TABLE local_invites( + stream_id BIGINT NOT NULL, + inviter TEXT NOT NULL, + invitee TEXT NOT NULL, + event_id TEXT NOT NULL, + room_id TEXT NOT NULL, + locally_rejected TEXT, + replaced_by TEXT +); + +-- Insert all invites for local users into new `invites` table +INSERT INTO local_invites SELECT + stream_ordering as stream_id, + sender as inviter, + state_key as invitee, + event_id, + room_id, + NULL as locally_rejected, + NULL as replaced_by + FROM events + NATURAL JOIN current_state_events + NATURAL JOIN room_memberships + WHERE membership = 'invite' AND state_key IN (SELECT name FROM users); + +CREATE INDEX local_invites_id ON local_invites(stream_id); +CREATE INDEX local_invites_for_user_idx ON local_invites(invitee, locally_rejected, replaced_by, room_id); diff --git a/synapse/storage/data_stores/main/schema/delta/31/local_media_repository_url_cache.sql b/synapse/storage/data_stores/main/schema/delta/31/local_media_repository_url_cache.sql new file mode 100644 index 0000000000..9efb4280eb --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/31/local_media_repository_url_cache.sql @@ -0,0 +1,27 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE TABLE local_media_repository_url_cache( + url TEXT, -- the URL being cached + response_code INTEGER, -- the HTTP response code of this download attempt + etag TEXT, -- the etag header of this response + expires INTEGER, -- the number of ms this response was valid for + og TEXT, -- cache of the OG metadata of this URL as JSON + media_id TEXT, -- the media_id, if any, of the URL's content in the repo + download_ts BIGINT -- the timestamp of this download attempt +); + +CREATE INDEX local_media_repository_url_cache_by_url_download_ts + ON local_media_repository_url_cache(url, download_ts); diff --git a/synapse/storage/data_stores/main/schema/delta/31/pushers.py b/synapse/storage/data_stores/main/schema/delta/31/pushers.py new file mode 100644 index 0000000000..9bb504aad5 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/31/pushers.py @@ -0,0 +1,87 @@ +# Copyright 2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + + +# Change the last_token to last_stream_ordering now that pushers no longer +# listen on an event stream but instead select out of the event_push_actions +# table. + + +import logging + +logger = logging.getLogger(__name__) + + +def token_to_stream_ordering(token): + return int(token[1:].split("_")[0]) + + +def run_create(cur, database_engine, *args, **kwargs): + logger.info("Porting pushers table, delta 31...") + cur.execute( + """ + CREATE TABLE IF NOT EXISTS pushers2 ( + id BIGINT PRIMARY KEY, + user_name TEXT NOT NULL, + access_token BIGINT DEFAULT NULL, + profile_tag VARCHAR(32) NOT NULL, + kind VARCHAR(8) NOT NULL, + app_id VARCHAR(64) NOT NULL, + app_display_name VARCHAR(64) NOT NULL, + device_display_name VARCHAR(128) NOT NULL, + pushkey TEXT NOT NULL, + ts BIGINT NOT NULL, + lang VARCHAR(8), + data TEXT, + last_stream_ordering INTEGER, + last_success BIGINT, + failing_since BIGINT, + UNIQUE (app_id, pushkey, user_name) + ) + """ + ) + cur.execute( + """SELECT + id, user_name, access_token, profile_tag, kind, + app_id, app_display_name, device_display_name, + pushkey, ts, lang, data, last_token, last_success, + failing_since + FROM pushers + """ + ) + count = 0 + for row in cur.fetchall(): + row = list(row) + row[12] = token_to_stream_ordering(row[12]) + cur.execute( + database_engine.convert_param_style( + """ + INSERT into pushers2 ( + id, user_name, access_token, profile_tag, kind, + app_id, app_display_name, device_display_name, + pushkey, ts, lang, data, last_stream_ordering, last_success, + failing_since + ) values (%s)""" + % (",".join(["?" for _ in range(len(row))])) + ), + row, + ) + count += 1 + cur.execute("DROP TABLE pushers") + cur.execute("ALTER TABLE pushers2 RENAME TO pushers") + logger.info("Moved %d pushers to new table", count) + + +def run_upgrade(cur, database_engine, *args, **kwargs): + pass diff --git a/synapse/storage/data_stores/main/schema/delta/31/pushers_index.sql b/synapse/storage/data_stores/main/schema/delta/31/pushers_index.sql new file mode 100644 index 0000000000..a82add88fd --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/31/pushers_index.sql @@ -0,0 +1,22 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/** Using CREATE INDEX directly is deprecated in favour of using background + * update see synapse/storage/schema/delta/33/access_tokens_device_index.sql + * and synapse/storage/registration.py for an example using + * "access_tokens_device_index" **/ + CREATE INDEX event_push_actions_stream_ordering on event_push_actions( + stream_ordering, user_id + ); diff --git a/synapse/storage/data_stores/main/schema/delta/31/search_update.py b/synapse/storage/data_stores/main/schema/delta/31/search_update.py new file mode 100644 index 0000000000..7d8ca5f93f --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/31/search_update.py @@ -0,0 +1,66 @@ +# Copyright 2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging + +import simplejson + +from synapse.storage.engines import PostgresEngine +from synapse.storage.prepare_database import get_statements + +logger = logging.getLogger(__name__) + + +ALTER_TABLE = """ +ALTER TABLE event_search ADD COLUMN origin_server_ts BIGINT; +ALTER TABLE event_search ADD COLUMN stream_ordering BIGINT; +""" + + +def run_create(cur, database_engine, *args, **kwargs): + if not isinstance(database_engine, PostgresEngine): + return + + for statement in get_statements(ALTER_TABLE.splitlines()): + cur.execute(statement) + + cur.execute("SELECT MIN(stream_ordering) FROM events") + rows = cur.fetchall() + min_stream_id = rows[0][0] + + cur.execute("SELECT MAX(stream_ordering) FROM events") + rows = cur.fetchall() + max_stream_id = rows[0][0] + + if min_stream_id is not None and max_stream_id is not None: + progress = { + "target_min_stream_id_inclusive": min_stream_id, + "max_stream_id_exclusive": max_stream_id + 1, + "rows_inserted": 0, + "have_added_indexes": False, + } + progress_json = simplejson.dumps(progress) + + sql = ( + "INSERT into background_updates (update_name, progress_json)" + " VALUES (?, ?)" + ) + + sql = database_engine.convert_param_style(sql) + + cur.execute(sql, ("event_search_order", progress_json)) + + +def run_upgrade(cur, database_engine, *args, **kwargs): + pass diff --git a/synapse/storage/data_stores/main/schema/delta/32/events.sql b/synapse/storage/data_stores/main/schema/delta/32/events.sql new file mode 100644 index 0000000000..1dd0f9e170 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/32/events.sql @@ -0,0 +1,16 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +ALTER TABLE events ADD COLUMN received_ts BIGINT; diff --git a/synapse/storage/data_stores/main/schema/delta/32/openid.sql b/synapse/storage/data_stores/main/schema/delta/32/openid.sql new file mode 100644 index 0000000000..36f37b11c8 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/32/openid.sql @@ -0,0 +1,9 @@ + +CREATE TABLE open_id_tokens ( + token TEXT NOT NULL PRIMARY KEY, + ts_valid_until_ms bigint NOT NULL, + user_id TEXT NOT NULL, + UNIQUE (token) +); + +CREATE index open_id_tokens_ts_valid_until_ms ON open_id_tokens(ts_valid_until_ms); diff --git a/synapse/storage/data_stores/main/schema/delta/32/pusher_throttle.sql b/synapse/storage/data_stores/main/schema/delta/32/pusher_throttle.sql new file mode 100644 index 0000000000..d86d30c13c --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/32/pusher_throttle.sql @@ -0,0 +1,23 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + +CREATE TABLE pusher_throttle( + pusher BIGINT NOT NULL, + room_id TEXT NOT NULL, + last_sent_ts BIGINT, + throttle_ms BIGINT, + PRIMARY KEY (pusher, room_id) +); diff --git a/synapse/storage/data_stores/main/schema/delta/32/remove_indices.sql b/synapse/storage/data_stores/main/schema/delta/32/remove_indices.sql new file mode 100644 index 0000000000..4219cdd06a --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/32/remove_indices.sql @@ -0,0 +1,34 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + +-- The following indices are redundant, other indices are equivalent or +-- supersets +DROP INDEX IF EXISTS events_room_id; -- Prefix of events_room_stream +DROP INDEX IF EXISTS events_order; -- Prefix of events_order_topo_stream_room +DROP INDEX IF EXISTS events_topological_ordering; -- Prefix of events_order_topo_stream_room +DROP INDEX IF EXISTS events_stream_ordering; -- Duplicate of PRIMARY KEY +DROP INDEX IF EXISTS state_groups_id; -- Duplicate of PRIMARY KEY +DROP INDEX IF EXISTS event_to_state_groups_id; -- Duplicate of PRIMARY KEY +DROP INDEX IF EXISTS event_push_actions_room_id_event_id_user_id_profile_tag; -- Duplicate of UNIQUE CONSTRAINT + +DROP INDEX IF EXISTS st_extrem_id; -- Prefix of UNIQUE CONSTRAINT +DROP INDEX IF EXISTS event_signatures_id; -- Prefix of UNIQUE CONSTRAINT +DROP INDEX IF EXISTS redactions_event_id; -- Duplicate of UNIQUE CONSTRAINT + +-- The following indices were unused +DROP INDEX IF EXISTS remote_media_cache_thumbnails_media_id; +DROP INDEX IF EXISTS evauth_edges_auth_id; +DROP INDEX IF EXISTS presence_stream_state; diff --git a/synapse/storage/data_stores/main/schema/delta/32/reports.sql b/synapse/storage/data_stores/main/schema/delta/32/reports.sql new file mode 100644 index 0000000000..d13609776f --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/32/reports.sql @@ -0,0 +1,25 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + +CREATE TABLE event_reports( + id BIGINT NOT NULL PRIMARY KEY, + received_ts BIGINT NOT NULL, + room_id TEXT NOT NULL, + event_id TEXT NOT NULL, + user_id TEXT NOT NULL, + reason TEXT, + content TEXT +); diff --git a/synapse/storage/data_stores/main/schema/delta/33/access_tokens_device_index.sql b/synapse/storage/data_stores/main/schema/delta/33/access_tokens_device_index.sql new file mode 100644 index 0000000000..61ad3fe3e8 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/33/access_tokens_device_index.sql @@ -0,0 +1,17 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +INSERT INTO background_updates (update_name, progress_json) VALUES + ('access_tokens_device_index', '{}'); diff --git a/synapse/storage/data_stores/main/schema/delta/33/devices.sql b/synapse/storage/data_stores/main/schema/delta/33/devices.sql new file mode 100644 index 0000000000..eca7268d82 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/33/devices.sql @@ -0,0 +1,21 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE TABLE devices ( + user_id TEXT NOT NULL, + device_id TEXT NOT NULL, + display_name TEXT, + CONSTRAINT device_uniqueness UNIQUE (user_id, device_id) +); diff --git a/synapse/storage/data_stores/main/schema/delta/33/devices_for_e2e_keys.sql b/synapse/storage/data_stores/main/schema/delta/33/devices_for_e2e_keys.sql new file mode 100644 index 0000000000..aa4a3b9f2f --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/33/devices_for_e2e_keys.sql @@ -0,0 +1,19 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- make sure that we have a device record for each set of E2E keys, so that the +-- user can delete them if they like. +INSERT INTO devices + SELECT user_id, device_id, NULL FROM e2e_device_keys_json; diff --git a/synapse/storage/data_stores/main/schema/delta/33/devices_for_e2e_keys_clear_unknown_device.sql b/synapse/storage/data_stores/main/schema/delta/33/devices_for_e2e_keys_clear_unknown_device.sql new file mode 100644 index 0000000000..6671573398 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/33/devices_for_e2e_keys_clear_unknown_device.sql @@ -0,0 +1,20 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- a previous version of the "devices_for_e2e_keys" delta set all the device +-- names to "unknown device". This wasn't terribly helpful +UPDATE devices + SET display_name = NULL + WHERE display_name = 'unknown device'; diff --git a/synapse/storage/data_stores/main/schema/delta/33/event_fields.py b/synapse/storage/data_stores/main/schema/delta/33/event_fields.py new file mode 100644 index 0000000000..bff1256a7b --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/33/event_fields.py @@ -0,0 +1,61 @@ +# Copyright 2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging + +import simplejson + +from synapse.storage.prepare_database import get_statements + +logger = logging.getLogger(__name__) + + +ALTER_TABLE = """ +ALTER TABLE events ADD COLUMN sender TEXT; +ALTER TABLE events ADD COLUMN contains_url BOOLEAN; +""" + + +def run_create(cur, database_engine, *args, **kwargs): + for statement in get_statements(ALTER_TABLE.splitlines()): + cur.execute(statement) + + cur.execute("SELECT MIN(stream_ordering) FROM events") + rows = cur.fetchall() + min_stream_id = rows[0][0] + + cur.execute("SELECT MAX(stream_ordering) FROM events") + rows = cur.fetchall() + max_stream_id = rows[0][0] + + if min_stream_id is not None and max_stream_id is not None: + progress = { + "target_min_stream_id_inclusive": min_stream_id, + "max_stream_id_exclusive": max_stream_id + 1, + "rows_inserted": 0, + } + progress_json = simplejson.dumps(progress) + + sql = ( + "INSERT into background_updates (update_name, progress_json)" + " VALUES (?, ?)" + ) + + sql = database_engine.convert_param_style(sql) + + cur.execute(sql, ("event_fields_sender_url", progress_json)) + + +def run_upgrade(cur, database_engine, *args, **kwargs): + pass diff --git a/synapse/storage/data_stores/main/schema/delta/33/remote_media_ts.py b/synapse/storage/data_stores/main/schema/delta/33/remote_media_ts.py new file mode 100644 index 0000000000..a26057dfb6 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/33/remote_media_ts.py @@ -0,0 +1,30 @@ +# Copyright 2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import time + +ALTER_TABLE = "ALTER TABLE remote_media_cache ADD COLUMN last_access_ts BIGINT" + + +def run_create(cur, database_engine, *args, **kwargs): + cur.execute(ALTER_TABLE) + + +def run_upgrade(cur, database_engine, *args, **kwargs): + cur.execute( + database_engine.convert_param_style( + "UPDATE remote_media_cache SET last_access_ts = ?" + ), + (int(time.time() * 1000),), + ) diff --git a/synapse/storage/data_stores/main/schema/delta/33/user_ips_index.sql b/synapse/storage/data_stores/main/schema/delta/33/user_ips_index.sql new file mode 100644 index 0000000000..473f75a78e --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/33/user_ips_index.sql @@ -0,0 +1,17 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +INSERT INTO background_updates (update_name, progress_json) VALUES + ('user_ips_device_index', '{}'); diff --git a/synapse/storage/data_stores/main/schema/delta/34/appservice_stream.sql b/synapse/storage/data_stores/main/schema/delta/34/appservice_stream.sql new file mode 100644 index 0000000000..69e16eda0f --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/34/appservice_stream.sql @@ -0,0 +1,23 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE TABLE IF NOT EXISTS appservice_stream_position( + Lock CHAR(1) NOT NULL DEFAULT 'X' UNIQUE, -- Makes sure this table only has one row. + stream_ordering BIGINT, + CHECK (Lock='X') +); + +INSERT INTO appservice_stream_position (stream_ordering) + SELECT COALESCE(MAX(stream_ordering), 0) FROM events; diff --git a/synapse/storage/data_stores/main/schema/delta/34/cache_stream.py b/synapse/storage/data_stores/main/schema/delta/34/cache_stream.py new file mode 100644 index 0000000000..cf09e43e2b --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/34/cache_stream.py @@ -0,0 +1,46 @@ +# Copyright 2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging + +from synapse.storage.engines import PostgresEngine +from synapse.storage.prepare_database import get_statements + +logger = logging.getLogger(__name__) + + +# This stream is used to notify replication slaves that some caches have +# been invalidated that they cannot infer from the other streams. +CREATE_TABLE = """ +CREATE TABLE cache_invalidation_stream ( + stream_id BIGINT, + cache_func TEXT, + keys TEXT[], + invalidation_ts BIGINT +); + +CREATE INDEX cache_invalidation_stream_id ON cache_invalidation_stream(stream_id); +""" + + +def run_create(cur, database_engine, *args, **kwargs): + if not isinstance(database_engine, PostgresEngine): + return + + for statement in get_statements(CREATE_TABLE.splitlines()): + cur.execute(statement) + + +def run_upgrade(cur, database_engine, *args, **kwargs): + pass diff --git a/synapse/storage/data_stores/main/schema/delta/34/device_inbox.sql b/synapse/storage/data_stores/main/schema/delta/34/device_inbox.sql new file mode 100644 index 0000000000..e68844c74a --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/34/device_inbox.sql @@ -0,0 +1,24 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE TABLE device_inbox ( + user_id TEXT NOT NULL, + device_id TEXT NOT NULL, + stream_id BIGINT NOT NULL, + message_json TEXT NOT NULL -- {"type":, "sender":, "content",} +); + +CREATE INDEX device_inbox_user_stream_id ON device_inbox(user_id, device_id, stream_id); +CREATE INDEX device_inbox_stream_id ON device_inbox(stream_id); diff --git a/synapse/storage/data_stores/main/schema/delta/34/push_display_name_rename.sql b/synapse/storage/data_stores/main/schema/delta/34/push_display_name_rename.sql new file mode 100644 index 0000000000..0d9fe1a99a --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/34/push_display_name_rename.sql @@ -0,0 +1,20 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +DELETE FROM push_rules WHERE rule_id = 'global/override/.m.rule.contains_display_name'; +UPDATE push_rules SET rule_id = 'global/override/.m.rule.contains_display_name' WHERE rule_id = 'global/underride/.m.rule.contains_display_name'; + +DELETE FROM push_rules_enable WHERE rule_id = 'global/override/.m.rule.contains_display_name'; +UPDATE push_rules_enable SET rule_id = 'global/override/.m.rule.contains_display_name' WHERE rule_id = 'global/underride/.m.rule.contains_display_name'; diff --git a/synapse/storage/data_stores/main/schema/delta/34/received_txn_purge.py b/synapse/storage/data_stores/main/schema/delta/34/received_txn_purge.py new file mode 100644 index 0000000000..67d505e68b --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/34/received_txn_purge.py @@ -0,0 +1,32 @@ +# Copyright 2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging + +from synapse.storage.engines import PostgresEngine + +logger = logging.getLogger(__name__) + + +def run_create(cur, database_engine, *args, **kwargs): + if isinstance(database_engine, PostgresEngine): + cur.execute("TRUNCATE received_transactions") + else: + cur.execute("DELETE FROM received_transactions") + + cur.execute("CREATE INDEX received_transactions_ts ON received_transactions(ts)") + + +def run_upgrade(cur, database_engine, *args, **kwargs): + pass diff --git a/synapse/storage/data_stores/main/schema/delta/35/add_state_index.sql b/synapse/storage/data_stores/main/schema/delta/35/add_state_index.sql new file mode 100644 index 0000000000..33980d02f0 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/35/add_state_index.sql @@ -0,0 +1,17 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +INSERT into background_updates (update_name, progress_json, depends_on) + VALUES ('state_group_state_type_index', '{}', 'state_group_state_deduplication'); diff --git a/synapse/storage/data_stores/main/schema/delta/35/contains_url.sql b/synapse/storage/data_stores/main/schema/delta/35/contains_url.sql new file mode 100644 index 0000000000..6cd123027b --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/35/contains_url.sql @@ -0,0 +1,17 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + INSERT into background_updates (update_name, progress_json) + VALUES ('event_contains_url_index', '{}'); diff --git a/synapse/storage/data_stores/main/schema/delta/35/device_outbox.sql b/synapse/storage/data_stores/main/schema/delta/35/device_outbox.sql new file mode 100644 index 0000000000..17e6c43105 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/35/device_outbox.sql @@ -0,0 +1,39 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +DROP TABLE IF EXISTS device_federation_outbox; +CREATE TABLE device_federation_outbox ( + destination TEXT NOT NULL, + stream_id BIGINT NOT NULL, + queued_ts BIGINT NOT NULL, + messages_json TEXT NOT NULL +); + + +DROP INDEX IF EXISTS device_federation_outbox_destination_id; +CREATE INDEX device_federation_outbox_destination_id + ON device_federation_outbox(destination, stream_id); + + +DROP TABLE IF EXISTS device_federation_inbox; +CREATE TABLE device_federation_inbox ( + origin TEXT NOT NULL, + message_id TEXT NOT NULL, + received_ts BIGINT NOT NULL +); + +DROP INDEX IF EXISTS device_federation_inbox_sender_id; +CREATE INDEX device_federation_inbox_sender_id + ON device_federation_inbox(origin, message_id); diff --git a/synapse/storage/data_stores/main/schema/delta/35/device_stream_id.sql b/synapse/storage/data_stores/main/schema/delta/35/device_stream_id.sql new file mode 100644 index 0000000000..7ab7d942e2 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/35/device_stream_id.sql @@ -0,0 +1,21 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE TABLE device_max_stream_id ( + stream_id BIGINT NOT NULL +); + +INSERT INTO device_max_stream_id (stream_id) + SELECT COALESCE(MAX(stream_id), 0) FROM device_inbox; diff --git a/synapse/storage/data_stores/main/schema/delta/35/event_push_actions_index.sql b/synapse/storage/data_stores/main/schema/delta/35/event_push_actions_index.sql new file mode 100644 index 0000000000..2e836d8e9c --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/35/event_push_actions_index.sql @@ -0,0 +1,17 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + INSERT into background_updates (update_name, progress_json) + VALUES ('epa_highlight_index', '{}'); diff --git a/synapse/storage/data_stores/main/schema/delta/35/public_room_list_change_stream.sql b/synapse/storage/data_stores/main/schema/delta/35/public_room_list_change_stream.sql new file mode 100644 index 0000000000..dd2bf2e28a --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/35/public_room_list_change_stream.sql @@ -0,0 +1,33 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + +CREATE TABLE public_room_list_stream ( + stream_id BIGINT NOT NULL, + room_id TEXT NOT NULL, + visibility BOOLEAN NOT NULL +); + +INSERT INTO public_room_list_stream (stream_id, room_id, visibility) + SELECT 1, room_id, is_public FROM rooms + WHERE is_public = CAST(1 AS BOOLEAN); + +CREATE INDEX public_room_list_stream_idx on public_room_list_stream( + stream_id +); + +CREATE INDEX public_room_list_stream_rm_idx on public_room_list_stream( + room_id, stream_id +); diff --git a/synapse/storage/data_stores/main/schema/delta/35/state.sql b/synapse/storage/data_stores/main/schema/delta/35/state.sql new file mode 100644 index 0000000000..0f1fa68a89 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/35/state.sql @@ -0,0 +1,22 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE TABLE state_group_edges( + state_group BIGINT NOT NULL, + prev_state_group BIGINT NOT NULL +); + +CREATE INDEX state_group_edges_idx ON state_group_edges(state_group); +CREATE INDEX state_group_edges_prev_idx ON state_group_edges(prev_state_group); diff --git a/synapse/storage/data_stores/main/schema/delta/35/state_dedupe.sql b/synapse/storage/data_stores/main/schema/delta/35/state_dedupe.sql new file mode 100644 index 0000000000..97e5067ef4 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/35/state_dedupe.sql @@ -0,0 +1,17 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +INSERT into background_updates (update_name, progress_json) + VALUES ('state_group_state_deduplication', '{}'); diff --git a/synapse/storage/data_stores/main/schema/delta/35/stream_order_to_extrem.sql b/synapse/storage/data_stores/main/schema/delta/35/stream_order_to_extrem.sql new file mode 100644 index 0000000000..2b945d8a57 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/35/stream_order_to_extrem.sql @@ -0,0 +1,37 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + +CREATE TABLE stream_ordering_to_exterm ( + stream_ordering BIGINT NOT NULL, + room_id TEXT NOT NULL, + event_id TEXT NOT NULL +); + +INSERT INTO stream_ordering_to_exterm (stream_ordering, room_id, event_id) + SELECT stream_ordering, room_id, event_id FROM event_forward_extremities + INNER JOIN ( + SELECT room_id, max(stream_ordering) as stream_ordering FROM events + INNER JOIN event_forward_extremities USING (room_id, event_id) + GROUP BY room_id + ) AS rms USING (room_id); + +CREATE INDEX stream_ordering_to_exterm_idx on stream_ordering_to_exterm( + stream_ordering +); + +CREATE INDEX stream_ordering_to_exterm_rm_idx on stream_ordering_to_exterm( + room_id, stream_ordering +); diff --git a/synapse/storage/data_stores/main/schema/delta/36/readd_public_rooms.sql b/synapse/storage/data_stores/main/schema/delta/36/readd_public_rooms.sql new file mode 100644 index 0000000000..90d8fd18f9 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/36/readd_public_rooms.sql @@ -0,0 +1,26 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- Re-add some entries to stream_ordering_to_exterm that were incorrectly deleted +INSERT INTO stream_ordering_to_exterm (stream_ordering, room_id, event_id) + SELECT + (SELECT stream_ordering FROM events where event_id = e.event_id) AS stream_ordering, + room_id, + event_id + FROM event_forward_extremities AS e + WHERE NOT EXISTS ( + SELECT room_id FROM stream_ordering_to_exterm AS s + WHERE s.room_id = e.room_id + ); diff --git a/synapse/storage/data_stores/main/schema/delta/37/remove_auth_idx.py b/synapse/storage/data_stores/main/schema/delta/37/remove_auth_idx.py new file mode 100644 index 0000000000..a377884169 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/37/remove_auth_idx.py @@ -0,0 +1,85 @@ +# Copyright 2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging + +from synapse.storage.engines import PostgresEngine +from synapse.storage.prepare_database import get_statements + +logger = logging.getLogger(__name__) + +DROP_INDICES = """ +-- We only ever query based on event_id +DROP INDEX IF EXISTS state_events_room_id; +DROP INDEX IF EXISTS state_events_type; +DROP INDEX IF EXISTS state_events_state_key; + +-- room_id is indexed elsewhere +DROP INDEX IF EXISTS current_state_events_room_id; +DROP INDEX IF EXISTS current_state_events_state_key; +DROP INDEX IF EXISTS current_state_events_type; + +DROP INDEX IF EXISTS transactions_have_ref; + +-- (topological_ordering, stream_ordering, room_id) seems like a strange index, +-- and is used incredibly rarely. +DROP INDEX IF EXISTS events_order_topo_stream_room; + +-- an equivalent index to this actually gets re-created in delta 41, because it +-- turned out that deleting it wasn't a great plan :/. In any case, let's +-- delete it here, and delta 41 will create a new one with an added UNIQUE +-- constraint +DROP INDEX IF EXISTS event_search_ev_idx; +""" + +POSTGRES_DROP_CONSTRAINT = """ +ALTER TABLE event_auth DROP CONSTRAINT IF EXISTS event_auth_event_id_auth_id_room_id_key; +""" + +SQLITE_DROP_CONSTRAINT = """ +DROP INDEX IF EXISTS evauth_edges_id; + +CREATE TABLE IF NOT EXISTS event_auth_new( + event_id TEXT NOT NULL, + auth_id TEXT NOT NULL, + room_id TEXT NOT NULL +); + +INSERT INTO event_auth_new + SELECT event_id, auth_id, room_id + FROM event_auth; + +DROP TABLE event_auth; + +ALTER TABLE event_auth_new RENAME TO event_auth; + +CREATE INDEX evauth_edges_id ON event_auth(event_id); +""" + + +def run_create(cur, database_engine, *args, **kwargs): + for statement in get_statements(DROP_INDICES.splitlines()): + cur.execute(statement) + + if isinstance(database_engine, PostgresEngine): + drop_constraint = POSTGRES_DROP_CONSTRAINT + else: + drop_constraint = SQLITE_DROP_CONSTRAINT + + for statement in get_statements(drop_constraint.splitlines()): + cur.execute(statement) + + +def run_upgrade(cur, database_engine, *args, **kwargs): + pass diff --git a/synapse/storage/data_stores/main/schema/delta/37/user_threepids.sql b/synapse/storage/data_stores/main/schema/delta/37/user_threepids.sql new file mode 100644 index 0000000000..cf7a90dd10 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/37/user_threepids.sql @@ -0,0 +1,52 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* + * Update any email addresses that were stored with mixed case into all + * lowercase + */ + + -- There may be "duplicate" emails (with different case) already in the table, + -- so we find them and move all but the most recently used account. + UPDATE user_threepids + SET medium = 'email_old' + WHERE medium = 'email' + AND address IN ( + -- We select all the addresses that are linked to the user_id that is NOT + -- the most recently created. + SELECT u.address + FROM + user_threepids AS u, + -- `duplicate_addresses` is a table of all the email addresses that + -- appear multiple times and when the binding was created + ( + SELECT lower(u1.address) AS address, max(u1.added_at) AS max_ts + FROM user_threepids AS u1 + INNER JOIN user_threepids AS u2 ON u1.medium = u2.medium AND lower(u1.address) = lower(u2.address) AND u1.address != u2.address + WHERE u1.medium = 'email' AND u2.medium = 'email' + GROUP BY lower(u1.address) + ) AS duplicate_addresses + WHERE + lower(u.address) = duplicate_addresses.address + AND u.added_at != max_ts -- NOT the most recently created + ); + + +-- This update is now safe since we've removed the duplicate addresses. +UPDATE user_threepids SET address = LOWER(address) WHERE medium = 'email'; + + +/* Add an index for the select we do on passwored reset */ +CREATE INDEX user_threepids_medium_address on user_threepids (medium, address); diff --git a/synapse/storage/data_stores/main/schema/delta/38/postgres_fts_gist.sql b/synapse/storage/data_stores/main/schema/delta/38/postgres_fts_gist.sql new file mode 100644 index 0000000000..515e6b8e84 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/38/postgres_fts_gist.sql @@ -0,0 +1,19 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- We no longer do this given we back it out again in schema 47 + +-- INSERT into background_updates (update_name, progress_json) +-- VALUES ('event_search_postgres_gist', '{}'); diff --git a/synapse/storage/data_stores/main/schema/delta/39/appservice_room_list.sql b/synapse/storage/data_stores/main/schema/delta/39/appservice_room_list.sql new file mode 100644 index 0000000000..74bdc49073 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/39/appservice_room_list.sql @@ -0,0 +1,29 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE TABLE appservice_room_list( + appservice_id TEXT NOT NULL, + network_id TEXT NOT NULL, + room_id TEXT NOT NULL +); + +-- Each appservice can have multiple published room lists associated with them, +-- keyed of a particular network_id +CREATE UNIQUE INDEX appservice_room_list_idx ON appservice_room_list( + appservice_id, network_id, room_id +); + +ALTER TABLE public_room_list_stream ADD COLUMN appservice_id TEXT; +ALTER TABLE public_room_list_stream ADD COLUMN network_id TEXT; diff --git a/synapse/storage/data_stores/main/schema/delta/39/device_federation_stream_idx.sql b/synapse/storage/data_stores/main/schema/delta/39/device_federation_stream_idx.sql new file mode 100644 index 0000000000..00be801e90 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/39/device_federation_stream_idx.sql @@ -0,0 +1,16 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE INDEX device_federation_outbox_id ON device_federation_outbox(stream_id); diff --git a/synapse/storage/data_stores/main/schema/delta/39/event_push_index.sql b/synapse/storage/data_stores/main/schema/delta/39/event_push_index.sql new file mode 100644 index 0000000000..de2ad93e5c --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/39/event_push_index.sql @@ -0,0 +1,17 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +INSERT INTO background_updates (update_name, progress_json) VALUES + ('event_push_actions_highlights_index', '{}'); diff --git a/synapse/storage/data_stores/main/schema/delta/39/federation_out_position.sql b/synapse/storage/data_stores/main/schema/delta/39/federation_out_position.sql new file mode 100644 index 0000000000..5af814290b --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/39/federation_out_position.sql @@ -0,0 +1,22 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + CREATE TABLE federation_stream_position( + type TEXT NOT NULL, + stream_id INTEGER NOT NULL + ); + + INSERT INTO federation_stream_position (type, stream_id) VALUES ('federation', -1); + INSERT INTO federation_stream_position (type, stream_id) SELECT 'events', coalesce(max(stream_ordering), -1) FROM events; diff --git a/synapse/storage/data_stores/main/schema/delta/39/membership_profile.sql b/synapse/storage/data_stores/main/schema/delta/39/membership_profile.sql new file mode 100644 index 0000000000..1bf911c8ab --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/39/membership_profile.sql @@ -0,0 +1,20 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +ALTER TABLE room_memberships ADD COLUMN display_name TEXT; +ALTER TABLE room_memberships ADD COLUMN avatar_url TEXT; + +INSERT into background_updates (update_name, progress_json) + VALUES ('room_membership_profile_update', '{}'); diff --git a/synapse/storage/data_stores/main/schema/delta/40/current_state_idx.sql b/synapse/storage/data_stores/main/schema/delta/40/current_state_idx.sql new file mode 100644 index 0000000000..7ffa189f39 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/40/current_state_idx.sql @@ -0,0 +1,17 @@ +/* Copyright 2017 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +INSERT INTO background_updates (update_name, progress_json) VALUES + ('current_state_members_idx', '{}'); diff --git a/synapse/storage/data_stores/main/schema/delta/40/device_inbox.sql b/synapse/storage/data_stores/main/schema/delta/40/device_inbox.sql new file mode 100644 index 0000000000..b9fe1f0480 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/40/device_inbox.sql @@ -0,0 +1,21 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- turn the pre-fill startup query into a index-only scan on postgresql. +INSERT into background_updates (update_name, progress_json) + VALUES ('device_inbox_stream_index', '{}'); + +INSERT into background_updates (update_name, progress_json, depends_on) + VALUES ('device_inbox_stream_drop', '{}', 'device_inbox_stream_index'); diff --git a/synapse/storage/data_stores/main/schema/delta/40/device_list_streams.sql b/synapse/storage/data_stores/main/schema/delta/40/device_list_streams.sql new file mode 100644 index 0000000000..dd6dcb65f1 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/40/device_list_streams.sql @@ -0,0 +1,60 @@ +/* Copyright 2017 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- Cache of remote devices. +CREATE TABLE device_lists_remote_cache ( + user_id TEXT NOT NULL, + device_id TEXT NOT NULL, + content TEXT NOT NULL +); + +-- The last update we got for a user. Empty if we're not receiving updates for +-- that user. +CREATE TABLE device_lists_remote_extremeties ( + user_id TEXT NOT NULL, + stream_id TEXT NOT NULL +); + +-- we used to create non-unique indexes on these tables, but as of update 52 we create +-- unique indexes concurrently: +-- +-- CREATE INDEX device_lists_remote_cache_id ON device_lists_remote_cache(user_id, device_id); +-- CREATE INDEX device_lists_remote_extremeties_id ON device_lists_remote_extremeties(user_id, stream_id); + + +-- Stream of device lists updates. Includes both local and remotes +CREATE TABLE device_lists_stream ( + stream_id BIGINT NOT NULL, + user_id TEXT NOT NULL, + device_id TEXT NOT NULL +); + +CREATE INDEX device_lists_stream_id ON device_lists_stream(stream_id, user_id); + + +-- The stream of updates to send to other servers. We keep at least one row +-- per user that was sent so that the prev_id for any new updates can be +-- calculated +CREATE TABLE device_lists_outbound_pokes ( + destination TEXT NOT NULL, + stream_id BIGINT NOT NULL, + user_id TEXT NOT NULL, + device_id TEXT NOT NULL, + sent BOOLEAN NOT NULL, + ts BIGINT NOT NULL -- So that in future we can clear out pokes to dead servers +); + +CREATE INDEX device_lists_outbound_pokes_id ON device_lists_outbound_pokes(destination, stream_id); +CREATE INDEX device_lists_outbound_pokes_user ON device_lists_outbound_pokes(destination, user_id); diff --git a/synapse/storage/data_stores/main/schema/delta/40/event_push_summary.sql b/synapse/storage/data_stores/main/schema/delta/40/event_push_summary.sql new file mode 100644 index 0000000000..3918f0b794 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/40/event_push_summary.sql @@ -0,0 +1,37 @@ +/* Copyright 2017 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- Aggregate of old notification counts that have been deleted out of the +-- main event_push_actions table. This count does not include those that were +-- highlights, as they remain in the event_push_actions table. +CREATE TABLE event_push_summary ( + user_id TEXT NOT NULL, + room_id TEXT NOT NULL, + notif_count BIGINT NOT NULL, + stream_ordering BIGINT NOT NULL +); + +CREATE INDEX event_push_summary_user_rm ON event_push_summary(user_id, room_id); + + +-- The stream ordering up to which we have aggregated the event_push_actions +-- table into event_push_summary +CREATE TABLE event_push_summary_stream_ordering ( + Lock CHAR(1) NOT NULL DEFAULT 'X' UNIQUE, -- Makes sure this table only has one row. + stream_ordering BIGINT NOT NULL, + CHECK (Lock='X') +); + +INSERT INTO event_push_summary_stream_ordering (stream_ordering) VALUES (0); diff --git a/synapse/storage/data_stores/main/schema/delta/40/pushers.sql b/synapse/storage/data_stores/main/schema/delta/40/pushers.sql new file mode 100644 index 0000000000..054a223f14 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/40/pushers.sql @@ -0,0 +1,39 @@ +/* Copyright 2017 Vector Creations Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE TABLE IF NOT EXISTS pushers2 ( + id BIGINT PRIMARY KEY, + user_name TEXT NOT NULL, + access_token BIGINT DEFAULT NULL, + profile_tag TEXT NOT NULL, + kind TEXT NOT NULL, + app_id TEXT NOT NULL, + app_display_name TEXT NOT NULL, + device_display_name TEXT NOT NULL, + pushkey TEXT NOT NULL, + ts BIGINT NOT NULL, + lang TEXT, + data TEXT, + last_stream_ordering INTEGER, + last_success BIGINT, + failing_since BIGINT, + UNIQUE (app_id, pushkey, user_name) +); + +INSERT INTO pushers2 SELECT * FROM PUSHERS; + +DROP TABLE PUSHERS; + +ALTER TABLE pushers2 RENAME TO pushers; diff --git a/synapse/storage/data_stores/main/schema/delta/41/device_list_stream_idx.sql b/synapse/storage/data_stores/main/schema/delta/41/device_list_stream_idx.sql new file mode 100644 index 0000000000..b7bee8b692 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/41/device_list_stream_idx.sql @@ -0,0 +1,17 @@ +/* Copyright 2017 Vector Creations Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +INSERT into background_updates (update_name, progress_json) + VALUES ('device_lists_stream_idx', '{}'); diff --git a/synapse/storage/data_stores/main/schema/delta/41/device_outbound_index.sql b/synapse/storage/data_stores/main/schema/delta/41/device_outbound_index.sql new file mode 100644 index 0000000000..62f0b9892b --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/41/device_outbound_index.sql @@ -0,0 +1,16 @@ +/* Copyright 2017 Vector Creations Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE INDEX device_lists_outbound_pokes_stream ON device_lists_outbound_pokes(stream_id); diff --git a/synapse/storage/data_stores/main/schema/delta/41/event_search_event_id_idx.sql b/synapse/storage/data_stores/main/schema/delta/41/event_search_event_id_idx.sql new file mode 100644 index 0000000000..5d9cfecf36 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/41/event_search_event_id_idx.sql @@ -0,0 +1,17 @@ +/* Copyright 2017 Vector Creations Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +INSERT into background_updates (update_name, progress_json) + VALUES ('event_search_event_id_idx', '{}'); diff --git a/synapse/storage/data_stores/main/schema/delta/41/ratelimit.sql b/synapse/storage/data_stores/main/schema/delta/41/ratelimit.sql new file mode 100644 index 0000000000..a194bf0238 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/41/ratelimit.sql @@ -0,0 +1,22 @@ +/* Copyright 2017 Vector Creations Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE TABLE ratelimit_override ( + user_id TEXT NOT NULL, + messages_per_second BIGINT, + burst_count BIGINT +); + +CREATE UNIQUE INDEX ratelimit_override_idx ON ratelimit_override(user_id); diff --git a/synapse/storage/data_stores/main/schema/delta/42/current_state_delta.sql b/synapse/storage/data_stores/main/schema/delta/42/current_state_delta.sql new file mode 100644 index 0000000000..d28851aff8 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/42/current_state_delta.sql @@ -0,0 +1,26 @@ +/* Copyright 2017 Vector Creations Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + +CREATE TABLE current_state_delta_stream ( + stream_id BIGINT NOT NULL, + room_id TEXT NOT NULL, + type TEXT NOT NULL, + state_key TEXT NOT NULL, + event_id TEXT, -- Is null if the key was removed + prev_event_id TEXT -- Is null if the key was added +); + +CREATE INDEX current_state_delta_stream_idx ON current_state_delta_stream(stream_id); diff --git a/synapse/storage/data_stores/main/schema/delta/42/device_list_last_id.sql b/synapse/storage/data_stores/main/schema/delta/42/device_list_last_id.sql new file mode 100644 index 0000000000..9ab8c14fa3 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/42/device_list_last_id.sql @@ -0,0 +1,33 @@ +/* Copyright 2017 Vector Creations Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + +-- Table of last stream_id that we sent to destination for user_id. This is +-- used to fill out the `prev_id` fields of outbound device list updates. +CREATE TABLE device_lists_outbound_last_success ( + destination TEXT NOT NULL, + user_id TEXT NOT NULL, + stream_id BIGINT NOT NULL +); + +INSERT INTO device_lists_outbound_last_success + SELECT destination, user_id, coalesce(max(stream_id), 0) as stream_id + FROM device_lists_outbound_pokes + WHERE sent = (1 = 1) -- sqlite doesn't have inbuilt boolean values + GROUP BY destination, user_id; + +CREATE INDEX device_lists_outbound_last_success_idx ON device_lists_outbound_last_success( + destination, user_id, stream_id +); diff --git a/synapse/storage/data_stores/main/schema/delta/42/event_auth_state_only.sql b/synapse/storage/data_stores/main/schema/delta/42/event_auth_state_only.sql new file mode 100644 index 0000000000..b8821ac759 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/42/event_auth_state_only.sql @@ -0,0 +1,17 @@ +/* Copyright 2017 Vector Creations Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +INSERT INTO background_updates (update_name, progress_json) VALUES + ('event_auth_state_only', '{}'); diff --git a/synapse/storage/data_stores/main/schema/delta/42/user_dir.py b/synapse/storage/data_stores/main/schema/delta/42/user_dir.py new file mode 100644 index 0000000000..506f326f4d --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/42/user_dir.py @@ -0,0 +1,84 @@ +# Copyright 2017 Vector Creations Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging + +from synapse.storage.engines import PostgresEngine, Sqlite3Engine +from synapse.storage.prepare_database import get_statements + +logger = logging.getLogger(__name__) + + +BOTH_TABLES = """ +CREATE TABLE user_directory_stream_pos ( + Lock CHAR(1) NOT NULL DEFAULT 'X' UNIQUE, -- Makes sure this table only has one row. + stream_id BIGINT, + CHECK (Lock='X') +); + +INSERT INTO user_directory_stream_pos (stream_id) VALUES (null); + +CREATE TABLE user_directory ( + user_id TEXT NOT NULL, + room_id TEXT NOT NULL, -- A room_id that we know the user is joined to + display_name TEXT, + avatar_url TEXT +); + +CREATE INDEX user_directory_room_idx ON user_directory(room_id); +CREATE UNIQUE INDEX user_directory_user_idx ON user_directory(user_id); + +CREATE TABLE users_in_pubic_room ( + user_id TEXT NOT NULL, + room_id TEXT NOT NULL -- A room_id that we know is public +); + +CREATE INDEX users_in_pubic_room_room_idx ON users_in_pubic_room(room_id); +CREATE UNIQUE INDEX users_in_pubic_room_user_idx ON users_in_pubic_room(user_id); +""" + + +POSTGRES_TABLE = """ +CREATE TABLE user_directory_search ( + user_id TEXT NOT NULL, + vector tsvector +); + +CREATE INDEX user_directory_search_fts_idx ON user_directory_search USING gin(vector); +CREATE UNIQUE INDEX user_directory_search_user_idx ON user_directory_search(user_id); +""" + + +SQLITE_TABLE = """ +CREATE VIRTUAL TABLE user_directory_search + USING fts4 ( user_id, value ); +""" + + +def run_create(cur, database_engine, *args, **kwargs): + for statement in get_statements(BOTH_TABLES.splitlines()): + cur.execute(statement) + + if isinstance(database_engine, PostgresEngine): + for statement in get_statements(POSTGRES_TABLE.splitlines()): + cur.execute(statement) + elif isinstance(database_engine, Sqlite3Engine): + for statement in get_statements(SQLITE_TABLE.splitlines()): + cur.execute(statement) + else: + raise Exception("Unrecognized database engine") + + +def run_upgrade(*args, **kwargs): + pass diff --git a/synapse/storage/data_stores/main/schema/delta/43/blocked_rooms.sql b/synapse/storage/data_stores/main/schema/delta/43/blocked_rooms.sql new file mode 100644 index 0000000000..0e3cd143ff --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/43/blocked_rooms.sql @@ -0,0 +1,21 @@ +/* Copyright 2017 Vector Creations Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE TABLE blocked_rooms ( + room_id TEXT NOT NULL, + user_id TEXT NOT NULL -- Admin who blocked the room +); + +CREATE UNIQUE INDEX blocked_rooms_idx ON blocked_rooms(room_id); diff --git a/synapse/storage/data_stores/main/schema/delta/43/quarantine_media.sql b/synapse/storage/data_stores/main/schema/delta/43/quarantine_media.sql new file mode 100644 index 0000000000..630907ec4f --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/43/quarantine_media.sql @@ -0,0 +1,17 @@ +/* Copyright 2017 Vector Creations Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +ALTER TABLE local_media_repository ADD COLUMN quarantined_by TEXT; +ALTER TABLE remote_media_cache ADD COLUMN quarantined_by TEXT; diff --git a/synapse/storage/data_stores/main/schema/delta/43/url_cache.sql b/synapse/storage/data_stores/main/schema/delta/43/url_cache.sql new file mode 100644 index 0000000000..45ebe020da --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/43/url_cache.sql @@ -0,0 +1,16 @@ +/* Copyright 2017 Vector Creations Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +ALTER TABLE local_media_repository ADD COLUMN url_cache TEXT; diff --git a/synapse/storage/data_stores/main/schema/delta/43/user_share.sql b/synapse/storage/data_stores/main/schema/delta/43/user_share.sql new file mode 100644 index 0000000000..ee7062abe4 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/43/user_share.sql @@ -0,0 +1,33 @@ +/* Copyright 2017 Vector Creations Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- Table keeping track of who shares a room with who. We only keep track +-- of this for local users, so `user_id` is local users only (but we do keep track +-- of which remote users share a room) +CREATE TABLE users_who_share_rooms ( + user_id TEXT NOT NULL, + other_user_id TEXT NOT NULL, + room_id TEXT NOT NULL, + share_private BOOLEAN NOT NULL -- is the shared room private? i.e. they share a private room +); + + +CREATE UNIQUE INDEX users_who_share_rooms_u_idx ON users_who_share_rooms(user_id, other_user_id); +CREATE INDEX users_who_share_rooms_r_idx ON users_who_share_rooms(room_id); +CREATE INDEX users_who_share_rooms_o_idx ON users_who_share_rooms(other_user_id); + + +-- Make sure that we populate the table initially +UPDATE user_directory_stream_pos SET stream_id = NULL; diff --git a/synapse/storage/data_stores/main/schema/delta/44/expire_url_cache.sql b/synapse/storage/data_stores/main/schema/delta/44/expire_url_cache.sql new file mode 100644 index 0000000000..b12f9b2ebf --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/44/expire_url_cache.sql @@ -0,0 +1,41 @@ +/* Copyright 2017 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- this didn't work on SQLite 3.7 (because of lack of partial indexes), so was +-- removed and replaced with 46/local_media_repository_url_idx.sql. +-- +-- CREATE INDEX local_media_repository_url_idx ON local_media_repository(created_ts) WHERE url_cache IS NOT NULL; + +-- we need to change `expires` to `expires_ts` so that we can index on it. SQLite doesn't support +-- indices on expressions until 3.9. +CREATE TABLE local_media_repository_url_cache_new( + url TEXT, + response_code INTEGER, + etag TEXT, + expires_ts BIGINT, + og TEXT, + media_id TEXT, + download_ts BIGINT +); + +INSERT INTO local_media_repository_url_cache_new + SELECT url, response_code, etag, expires + download_ts, og, media_id, download_ts FROM local_media_repository_url_cache; + +DROP TABLE local_media_repository_url_cache; +ALTER TABLE local_media_repository_url_cache_new RENAME TO local_media_repository_url_cache; + +CREATE INDEX local_media_repository_url_cache_expires_idx ON local_media_repository_url_cache(expires_ts); +CREATE INDEX local_media_repository_url_cache_by_url_download_ts ON local_media_repository_url_cache(url, download_ts); +CREATE INDEX local_media_repository_url_cache_media_idx ON local_media_repository_url_cache(media_id); diff --git a/synapse/storage/data_stores/main/schema/delta/45/group_server.sql b/synapse/storage/data_stores/main/schema/delta/45/group_server.sql new file mode 100644 index 0000000000..b2333848a0 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/45/group_server.sql @@ -0,0 +1,167 @@ +/* Copyright 2017 Vector Creations Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE TABLE groups ( + group_id TEXT NOT NULL, + name TEXT, -- the display name of the room + avatar_url TEXT, + short_description TEXT, + long_description TEXT +); + +CREATE UNIQUE INDEX groups_idx ON groups(group_id); + + +-- list of users the group server thinks are joined +CREATE TABLE group_users ( + group_id TEXT NOT NULL, + user_id TEXT NOT NULL, + is_admin BOOLEAN NOT NULL, + is_public BOOLEAN NOT NULL -- whether the users membership can be seen by everyone +); + + +CREATE INDEX groups_users_g_idx ON group_users(group_id, user_id); +CREATE INDEX groups_users_u_idx ON group_users(user_id); + +-- list of users the group server thinks are invited +CREATE TABLE group_invites ( + group_id TEXT NOT NULL, + user_id TEXT NOT NULL +); + +CREATE INDEX groups_invites_g_idx ON group_invites(group_id, user_id); +CREATE INDEX groups_invites_u_idx ON group_invites(user_id); + + +CREATE TABLE group_rooms ( + group_id TEXT NOT NULL, + room_id TEXT NOT NULL, + is_public BOOLEAN NOT NULL -- whether the room can be seen by everyone +); + +CREATE UNIQUE INDEX groups_rooms_g_idx ON group_rooms(group_id, room_id); +CREATE INDEX groups_rooms_r_idx ON group_rooms(room_id); + + +-- Rooms to include in the summary +CREATE TABLE group_summary_rooms ( + group_id TEXT NOT NULL, + room_id TEXT NOT NULL, + category_id TEXT NOT NULL, + room_order BIGINT NOT NULL, + is_public BOOLEAN NOT NULL, -- whether the room should be show to everyone + UNIQUE (group_id, category_id, room_id, room_order), + CHECK (room_order > 0) +); + +CREATE UNIQUE INDEX group_summary_rooms_g_idx ON group_summary_rooms(group_id, room_id, category_id); + + +-- Categories to include in the summary +CREATE TABLE group_summary_room_categories ( + group_id TEXT NOT NULL, + category_id TEXT NOT NULL, + cat_order BIGINT NOT NULL, + UNIQUE (group_id, category_id, cat_order), + CHECK (cat_order > 0) +); + +-- The categories in the group +CREATE TABLE group_room_categories ( + group_id TEXT NOT NULL, + category_id TEXT NOT NULL, + profile TEXT NOT NULL, + is_public BOOLEAN NOT NULL, -- whether the category should be show to everyone + UNIQUE (group_id, category_id) +); + +-- The users to include in the group summary +CREATE TABLE group_summary_users ( + group_id TEXT NOT NULL, + user_id TEXT NOT NULL, + role_id TEXT NOT NULL, + user_order BIGINT NOT NULL, + is_public BOOLEAN NOT NULL -- whether the user should be show to everyone +); + +CREATE INDEX group_summary_users_g_idx ON group_summary_users(group_id); + +-- The roles to include in the group summary +CREATE TABLE group_summary_roles ( + group_id TEXT NOT NULL, + role_id TEXT NOT NULL, + role_order BIGINT NOT NULL, + UNIQUE (group_id, role_id, role_order), + CHECK (role_order > 0) +); + + +-- The roles in a groups +CREATE TABLE group_roles ( + group_id TEXT NOT NULL, + role_id TEXT NOT NULL, + profile TEXT NOT NULL, + is_public BOOLEAN NOT NULL, -- whether the role should be show to everyone + UNIQUE (group_id, role_id) +); + + +-- List of attestations we've given out and need to renew +CREATE TABLE group_attestations_renewals ( + group_id TEXT NOT NULL, + user_id TEXT NOT NULL, + valid_until_ms BIGINT NOT NULL +); + +CREATE INDEX group_attestations_renewals_g_idx ON group_attestations_renewals(group_id, user_id); +CREATE INDEX group_attestations_renewals_u_idx ON group_attestations_renewals(user_id); +CREATE INDEX group_attestations_renewals_v_idx ON group_attestations_renewals(valid_until_ms); + + +-- List of attestations we've received from remotes and are interested in. +CREATE TABLE group_attestations_remote ( + group_id TEXT NOT NULL, + user_id TEXT NOT NULL, + valid_until_ms BIGINT NOT NULL, + attestation_json TEXT NOT NULL +); + +CREATE INDEX group_attestations_remote_g_idx ON group_attestations_remote(group_id, user_id); +CREATE INDEX group_attestations_remote_u_idx ON group_attestations_remote(user_id); +CREATE INDEX group_attestations_remote_v_idx ON group_attestations_remote(valid_until_ms); + + +-- The group membership for the HS's users +CREATE TABLE local_group_membership ( + group_id TEXT NOT NULL, + user_id TEXT NOT NULL, + is_admin BOOLEAN NOT NULL, + membership TEXT NOT NULL, + is_publicised BOOLEAN NOT NULL, -- if the user is publicising their membership + content TEXT NOT NULL +); + +CREATE INDEX local_group_membership_u_idx ON local_group_membership(user_id, group_id); +CREATE INDEX local_group_membership_g_idx ON local_group_membership(group_id); + + +CREATE TABLE local_group_updates ( + stream_id BIGINT NOT NULL, + group_id TEXT NOT NULL, + user_id TEXT NOT NULL, + type TEXT NOT NULL, + content TEXT NOT NULL +); diff --git a/synapse/storage/data_stores/main/schema/delta/45/profile_cache.sql b/synapse/storage/data_stores/main/schema/delta/45/profile_cache.sql new file mode 100644 index 0000000000..e5ddc84df0 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/45/profile_cache.sql @@ -0,0 +1,28 @@ +/* Copyright 2017 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + +-- A subset of remote users whose profiles we have cached. +-- Whether a user is in this table or not is defined by the storage function +-- `is_subscribed_remote_profile_for_user` +CREATE TABLE remote_profile_cache ( + user_id TEXT NOT NULL, + displayname TEXT, + avatar_url TEXT, + last_check BIGINT NOT NULL +); + +CREATE UNIQUE INDEX remote_profile_cache_user_id ON remote_profile_cache(user_id); +CREATE INDEX remote_profile_cache_time ON remote_profile_cache(last_check); diff --git a/synapse/storage/data_stores/main/schema/delta/46/drop_refresh_tokens.sql b/synapse/storage/data_stores/main/schema/delta/46/drop_refresh_tokens.sql new file mode 100644 index 0000000000..68c48a89a9 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/46/drop_refresh_tokens.sql @@ -0,0 +1,17 @@ +/* Copyright 2017 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* we no longer use (or create) the refresh_tokens table */ +DROP TABLE IF EXISTS refresh_tokens; diff --git a/synapse/storage/data_stores/main/schema/delta/46/drop_unique_deleted_pushers.sql b/synapse/storage/data_stores/main/schema/delta/46/drop_unique_deleted_pushers.sql new file mode 100644 index 0000000000..bb307889c1 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/46/drop_unique_deleted_pushers.sql @@ -0,0 +1,35 @@ +/* Copyright 2017 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- drop the unique constraint on deleted_pushers so that we can just insert +-- into it rather than upserting. + +CREATE TABLE deleted_pushers2 ( + stream_id BIGINT NOT NULL, + app_id TEXT NOT NULL, + pushkey TEXT NOT NULL, + user_id TEXT NOT NULL +); + +INSERT INTO deleted_pushers2 (stream_id, app_id, pushkey, user_id) + SELECT stream_id, app_id, pushkey, user_id from deleted_pushers; + +DROP TABLE deleted_pushers; +ALTER TABLE deleted_pushers2 RENAME TO deleted_pushers; + +-- create the index after doing the inserts because that's more efficient. +-- it also means we can give it the same name as the old one without renaming. +CREATE INDEX deleted_pushers_stream_id ON deleted_pushers (stream_id); + diff --git a/synapse/storage/data_stores/main/schema/delta/46/group_server.sql b/synapse/storage/data_stores/main/schema/delta/46/group_server.sql new file mode 100644 index 0000000000..097679bc9a --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/46/group_server.sql @@ -0,0 +1,32 @@ +/* Copyright 2017 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE TABLE groups_new ( + group_id TEXT NOT NULL, + name TEXT, -- the display name of the room + avatar_url TEXT, + short_description TEXT, + long_description TEXT, + is_public BOOL NOT NULL -- whether non-members can access group APIs +); + +-- NB: awful hack to get the default to be true on postgres and 1 on sqlite +INSERT INTO groups_new + SELECT group_id, name, avatar_url, short_description, long_description, (1=1) FROM groups; + +DROP TABLE groups; +ALTER TABLE groups_new RENAME TO groups; + +CREATE UNIQUE INDEX groups_idx ON groups(group_id); diff --git a/synapse/storage/data_stores/main/schema/delta/46/local_media_repository_url_idx.sql b/synapse/storage/data_stores/main/schema/delta/46/local_media_repository_url_idx.sql new file mode 100644 index 0000000000..bbfc7f5d1a --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/46/local_media_repository_url_idx.sql @@ -0,0 +1,24 @@ +/* Copyright 2017 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- register a background update which will recreate the +-- local_media_repository_url_idx index. +-- +-- We do this as a bg update not because it is a particularly onerous +-- operation, but because we'd like it to be a partial index if possible, and +-- the background_index_update code will understand whether we are on +-- postgres or sqlite and behave accordingly. +INSERT INTO background_updates (update_name, progress_json) VALUES + ('local_media_repository_url_idx', '{}'); diff --git a/synapse/storage/data_stores/main/schema/delta/46/user_dir_null_room_ids.sql b/synapse/storage/data_stores/main/schema/delta/46/user_dir_null_room_ids.sql new file mode 100644 index 0000000000..cb0d5a2576 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/46/user_dir_null_room_ids.sql @@ -0,0 +1,35 @@ +/* Copyright 2017 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- change the user_directory table to also cover global local user profiles +-- rather than just profiles within specific rooms. + +CREATE TABLE user_directory2 ( + user_id TEXT NOT NULL, + room_id TEXT, + display_name TEXT, + avatar_url TEXT +); + +INSERT INTO user_directory2(user_id, room_id, display_name, avatar_url) + SELECT user_id, room_id, display_name, avatar_url from user_directory; + +DROP TABLE user_directory; +ALTER TABLE user_directory2 RENAME TO user_directory; + +-- create indexes after doing the inserts because that's more efficient. +-- it also means we can give it the same name as the old one without renaming. +CREATE INDEX user_directory_room_idx ON user_directory(room_id); +CREATE UNIQUE INDEX user_directory_user_idx ON user_directory(user_id); diff --git a/synapse/storage/data_stores/main/schema/delta/46/user_dir_typos.sql b/synapse/storage/data_stores/main/schema/delta/46/user_dir_typos.sql new file mode 100644 index 0000000000..d9505f8da1 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/46/user_dir_typos.sql @@ -0,0 +1,24 @@ +/* Copyright 2017 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- this is just embarassing :| +ALTER TABLE users_in_pubic_room RENAME TO users_in_public_rooms; + +-- this is only 300K rows on matrix.org and takes ~3s to generate the index, +-- so is hopefully not going to block anyone else for that long... +CREATE INDEX users_in_public_rooms_room_idx ON users_in_public_rooms(room_id); +CREATE UNIQUE INDEX users_in_public_rooms_user_idx ON users_in_public_rooms(user_id); +DROP INDEX users_in_pubic_room_room_idx; +DROP INDEX users_in_pubic_room_user_idx; diff --git a/synapse/storage/data_stores/main/schema/delta/47/last_access_media.sql b/synapse/storage/data_stores/main/schema/delta/47/last_access_media.sql new file mode 100644 index 0000000000..f505fb22b5 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/47/last_access_media.sql @@ -0,0 +1,16 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +ALTER TABLE local_media_repository ADD COLUMN last_access_ts BIGINT; diff --git a/synapse/storage/data_stores/main/schema/delta/47/postgres_fts_gin.sql b/synapse/storage/data_stores/main/schema/delta/47/postgres_fts_gin.sql new file mode 100644 index 0000000000..31d7a817eb --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/47/postgres_fts_gin.sql @@ -0,0 +1,17 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +INSERT into background_updates (update_name, progress_json) + VALUES ('event_search_postgres_gin', '{}'); diff --git a/synapse/storage/data_stores/main/schema/delta/47/push_actions_staging.sql b/synapse/storage/data_stores/main/schema/delta/47/push_actions_staging.sql new file mode 100644 index 0000000000..edccf4a96f --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/47/push_actions_staging.sql @@ -0,0 +1,28 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- Temporary staging area for push actions that have been calculated for an +-- event, but the event hasn't yet been persisted. +-- When the event is persisted the rows are moved over to the +-- event_push_actions table. +CREATE TABLE event_push_actions_staging ( + event_id TEXT NOT NULL, + user_id TEXT NOT NULL, + actions TEXT NOT NULL, + notif SMALLINT NOT NULL, + highlight SMALLINT NOT NULL +); + +CREATE INDEX event_push_actions_staging_id ON event_push_actions_staging(event_id); diff --git a/synapse/storage/data_stores/main/schema/delta/47/state_group_seq.py b/synapse/storage/data_stores/main/schema/delta/47/state_group_seq.py new file mode 100644 index 0000000000..9fd1ccf6f7 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/47/state_group_seq.py @@ -0,0 +1,34 @@ +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from synapse.storage.engines import PostgresEngine + + +def run_create(cur, database_engine, *args, **kwargs): + if isinstance(database_engine, PostgresEngine): + # if we already have some state groups, we want to start making new + # ones with a higher id. + cur.execute("SELECT max(id) FROM state_groups") + row = cur.fetchone() + + if row[0] is None: + start_val = 1 + else: + start_val = row[0] + 1 + + cur.execute("CREATE SEQUENCE state_group_id_seq START WITH %s", (start_val,)) + + +def run_upgrade(*args, **kwargs): + pass diff --git a/synapse/storage/data_stores/main/schema/delta/48/add_user_consent.sql b/synapse/storage/data_stores/main/schema/delta/48/add_user_consent.sql new file mode 100644 index 0000000000..5237491506 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/48/add_user_consent.sql @@ -0,0 +1,18 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* record the version of the privacy policy the user has consented to + */ +ALTER TABLE users ADD COLUMN consent_version TEXT; diff --git a/synapse/storage/data_stores/main/schema/delta/48/add_user_ips_last_seen_index.sql b/synapse/storage/data_stores/main/schema/delta/48/add_user_ips_last_seen_index.sql new file mode 100644 index 0000000000..9248b0b24a --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/48/add_user_ips_last_seen_index.sql @@ -0,0 +1,17 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +INSERT into background_updates (update_name, progress_json) + VALUES ('user_ips_last_seen_index', '{}'); diff --git a/synapse/storage/data_stores/main/schema/delta/48/deactivated_users.sql b/synapse/storage/data_stores/main/schema/delta/48/deactivated_users.sql new file mode 100644 index 0000000000..e9013a6969 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/48/deactivated_users.sql @@ -0,0 +1,25 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* + * Store any accounts that have been requested to be deactivated. + * We part the account from all the rooms its in when its + * deactivated. This can take some time and synapse may be restarted + * before it completes, so store the user IDs here until the process + * is complete. + */ +CREATE TABLE users_pending_deactivation ( + user_id TEXT NOT NULL +); diff --git a/synapse/storage/data_stores/main/schema/delta/48/group_unique_indexes.py b/synapse/storage/data_stores/main/schema/delta/48/group_unique_indexes.py new file mode 100644 index 0000000000..49f5f2c003 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/48/group_unique_indexes.py @@ -0,0 +1,63 @@ +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from synapse.storage.engines import PostgresEngine +from synapse.storage.prepare_database import get_statements + +FIX_INDEXES = """ +-- rebuild indexes as uniques +DROP INDEX groups_invites_g_idx; +CREATE UNIQUE INDEX group_invites_g_idx ON group_invites(group_id, user_id); +DROP INDEX groups_users_g_idx; +CREATE UNIQUE INDEX group_users_g_idx ON group_users(group_id, user_id); + +-- rename other indexes to actually match their table names.. +DROP INDEX groups_users_u_idx; +CREATE INDEX group_users_u_idx ON group_users(user_id); +DROP INDEX groups_invites_u_idx; +CREATE INDEX group_invites_u_idx ON group_invites(user_id); +DROP INDEX groups_rooms_g_idx; +CREATE UNIQUE INDEX group_rooms_g_idx ON group_rooms(group_id, room_id); +DROP INDEX groups_rooms_r_idx; +CREATE INDEX group_rooms_r_idx ON group_rooms(room_id); +""" + + +def run_create(cur, database_engine, *args, **kwargs): + rowid = "ctid" if isinstance(database_engine, PostgresEngine) else "rowid" + + # remove duplicates from group_users & group_invites tables + cur.execute( + """ + DELETE FROM group_users WHERE %s NOT IN ( + SELECT min(%s) FROM group_users GROUP BY group_id, user_id + ); + """ + % (rowid, rowid) + ) + cur.execute( + """ + DELETE FROM group_invites WHERE %s NOT IN ( + SELECT min(%s) FROM group_invites GROUP BY group_id, user_id + ); + """ + % (rowid, rowid) + ) + + for statement in get_statements(FIX_INDEXES.splitlines()): + cur.execute(statement) + + +def run_upgrade(*args, **kwargs): + pass diff --git a/synapse/storage/data_stores/main/schema/delta/48/groups_joinable.sql b/synapse/storage/data_stores/main/schema/delta/48/groups_joinable.sql new file mode 100644 index 0000000000..ce26eaf0c9 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/48/groups_joinable.sql @@ -0,0 +1,22 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* + * This isn't a real ENUM because sqlite doesn't support it + * and we use a default of NULL for inserted rows and interpret + * NULL at the python store level as necessary so that existing + * rows are given the correct default policy. + */ +ALTER TABLE groups ADD COLUMN join_policy TEXT NOT NULL DEFAULT 'invite'; diff --git a/synapse/storage/data_stores/main/schema/delta/49/add_user_consent_server_notice_sent.sql b/synapse/storage/data_stores/main/schema/delta/49/add_user_consent_server_notice_sent.sql new file mode 100644 index 0000000000..14dcf18d73 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/49/add_user_consent_server_notice_sent.sql @@ -0,0 +1,20 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* record whether we have sent a server notice about consenting to the + * privacy policy. Specifically records the version of the policy we sent + * a message about. + */ +ALTER TABLE users ADD COLUMN consent_server_notice_sent TEXT; diff --git a/synapse/storage/data_stores/main/schema/delta/49/add_user_daily_visits.sql b/synapse/storage/data_stores/main/schema/delta/49/add_user_daily_visits.sql new file mode 100644 index 0000000000..3dd478196f --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/49/add_user_daily_visits.sql @@ -0,0 +1,21 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + +CREATE TABLE user_daily_visits ( user_id TEXT NOT NULL, + device_id TEXT, + timestamp BIGINT NOT NULL ); +CREATE INDEX user_daily_visits_uts_idx ON user_daily_visits(user_id, timestamp); +CREATE INDEX user_daily_visits_ts_idx ON user_daily_visits(timestamp); diff --git a/synapse/storage/data_stores/main/schema/delta/49/add_user_ips_last_seen_only_index.sql b/synapse/storage/data_stores/main/schema/delta/49/add_user_ips_last_seen_only_index.sql new file mode 100644 index 0000000000..3a4ed59b5b --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/49/add_user_ips_last_seen_only_index.sql @@ -0,0 +1,17 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +INSERT into background_updates (update_name, progress_json) + VALUES ('user_ips_last_seen_only_index', '{}'); diff --git a/synapse/storage/data_stores/main/schema/delta/50/add_creation_ts_users_index.sql b/synapse/storage/data_stores/main/schema/delta/50/add_creation_ts_users_index.sql new file mode 100644 index 0000000000..c93ae47532 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/50/add_creation_ts_users_index.sql @@ -0,0 +1,19 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + + +INSERT into background_updates (update_name, progress_json) + VALUES ('users_creation_ts', '{}'); diff --git a/synapse/storage/data_stores/main/schema/delta/50/erasure_store.sql b/synapse/storage/data_stores/main/schema/delta/50/erasure_store.sql new file mode 100644 index 0000000000..5d8641a9ab --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/50/erasure_store.sql @@ -0,0 +1,21 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- a table of users who have requested that their details be erased +CREATE TABLE erased_users ( + user_id TEXT NOT NULL +); + +CREATE UNIQUE INDEX erased_users_user ON erased_users(user_id); diff --git a/synapse/storage/data_stores/main/schema/delta/50/make_event_content_nullable.py b/synapse/storage/data_stores/main/schema/delta/50/make_event_content_nullable.py new file mode 100644 index 0000000000..b1684a8441 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/50/make_event_content_nullable.py @@ -0,0 +1,96 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +""" +We want to stop populating 'event.content', so we need to make it nullable. + +If this has to be rolled back, then the following should populate the missing data: + +Postgres: + + UPDATE events SET content=(ej.json::json)->'content' FROM event_json ej + WHERE ej.event_id = events.event_id AND + stream_ordering < ( + SELECT stream_ordering FROM events WHERE content IS NOT NULL + ORDER BY stream_ordering LIMIT 1 + ); + + UPDATE events SET content=(ej.json::json)->'content' FROM event_json ej + WHERE ej.event_id = events.event_id AND + stream_ordering > ( + SELECT stream_ordering FROM events WHERE content IS NOT NULL + ORDER BY stream_ordering DESC LIMIT 1 + ); + +SQLite: + + UPDATE events SET content=( + SELECT json_extract(json,'$.content') FROM event_json ej + WHERE ej.event_id = events.event_id + ) + WHERE + stream_ordering < ( + SELECT stream_ordering FROM events WHERE content IS NOT NULL + ORDER BY stream_ordering LIMIT 1 + ) + OR stream_ordering > ( + SELECT stream_ordering FROM events WHERE content IS NOT NULL + ORDER BY stream_ordering DESC LIMIT 1 + ); + +""" + +import logging + +from synapse.storage.engines import PostgresEngine + +logger = logging.getLogger(__name__) + + +def run_create(cur, database_engine, *args, **kwargs): + pass + + +def run_upgrade(cur, database_engine, *args, **kwargs): + if isinstance(database_engine, PostgresEngine): + cur.execute( + """ + ALTER TABLE events ALTER COLUMN content DROP NOT NULL; + """ + ) + return + + # sqlite is an arse about this. ref: https://www.sqlite.org/lang_altertable.html + + cur.execute( + "SELECT sql FROM sqlite_master WHERE tbl_name='events' AND type='table'" + ) + (oldsql,) = cur.fetchone() + + sql = oldsql.replace("content TEXT NOT NULL", "content TEXT") + if sql == oldsql: + raise Exception("Couldn't find null constraint to drop in %s" % oldsql) + + logger.info("Replacing definition of 'events' with: %s", sql) + + cur.execute("PRAGMA schema_version") + (oldver,) = cur.fetchone() + cur.execute("PRAGMA writable_schema=ON") + cur.execute( + "UPDATE sqlite_master SET sql=? WHERE tbl_name='events' AND type='table'", + (sql,), + ) + cur.execute("PRAGMA schema_version=%i" % (oldver + 1,)) + cur.execute("PRAGMA writable_schema=OFF") diff --git a/synapse/storage/data_stores/main/schema/delta/51/e2e_room_keys.sql b/synapse/storage/data_stores/main/schema/delta/51/e2e_room_keys.sql new file mode 100644 index 0000000000..c0e66a697d --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/51/e2e_room_keys.sql @@ -0,0 +1,39 @@ +/* Copyright 2017 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- users' optionally backed up encrypted e2e sessions +CREATE TABLE e2e_room_keys ( + user_id TEXT NOT NULL, + room_id TEXT NOT NULL, + session_id TEXT NOT NULL, + version TEXT NOT NULL, + first_message_index INT, + forwarded_count INT, + is_verified BOOLEAN, + session_data TEXT NOT NULL +); + +CREATE UNIQUE INDEX e2e_room_keys_idx ON e2e_room_keys(user_id, room_id, session_id); + +-- the metadata for each generation of encrypted e2e session backups +CREATE TABLE e2e_room_keys_versions ( + user_id TEXT NOT NULL, + version TEXT NOT NULL, + algorithm TEXT NOT NULL, + auth_data TEXT NOT NULL, + deleted SMALLINT DEFAULT 0 NOT NULL +); + +CREATE UNIQUE INDEX e2e_room_keys_versions_idx ON e2e_room_keys_versions(user_id, version); diff --git a/synapse/storage/data_stores/main/schema/delta/51/monthly_active_users.sql b/synapse/storage/data_stores/main/schema/delta/51/monthly_active_users.sql new file mode 100644 index 0000000000..c9d537d5a3 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/51/monthly_active_users.sql @@ -0,0 +1,27 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- a table of monthly active users, for use where blocking based on mau limits +CREATE TABLE monthly_active_users ( + user_id TEXT NOT NULL, + -- Last time we saw the user. Not guaranteed to be accurate due to rate limiting + -- on updates, Granularity of updates governed by + -- synapse.storage.monthly_active_users.LAST_SEEN_GRANULARITY + -- Measured in ms since epoch. + timestamp BIGINT NOT NULL +); + +CREATE UNIQUE INDEX monthly_active_users_users ON monthly_active_users(user_id); +CREATE INDEX monthly_active_users_time_stamp ON monthly_active_users(timestamp); diff --git a/synapse/storage/data_stores/main/schema/delta/52/add_event_to_state_group_index.sql b/synapse/storage/data_stores/main/schema/delta/52/add_event_to_state_group_index.sql new file mode 100644 index 0000000000..91e03d13e1 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/52/add_event_to_state_group_index.sql @@ -0,0 +1,19 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- This is needed to efficiently check for unreferenced state groups during +-- purge. Added events_to_state_group(state_group) index +INSERT into background_updates (update_name, progress_json) + VALUES ('event_to_state_groups_sg_index', '{}'); diff --git a/synapse/storage/data_stores/main/schema/delta/52/device_list_streams_unique_idx.sql b/synapse/storage/data_stores/main/schema/delta/52/device_list_streams_unique_idx.sql new file mode 100644 index 0000000000..bfa49e6f92 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/52/device_list_streams_unique_idx.sql @@ -0,0 +1,36 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- register a background update which will create a unique index on +-- device_lists_remote_cache +INSERT into background_updates (update_name, progress_json) + VALUES ('device_lists_remote_cache_unique_idx', '{}'); + +-- and one on device_lists_remote_extremeties +INSERT into background_updates (update_name, progress_json, depends_on) + VALUES ( + 'device_lists_remote_extremeties_unique_idx', '{}', + + -- doesn't really depend on this, but we need to make sure both happen + -- before we drop the old indexes. + 'device_lists_remote_cache_unique_idx' + ); + +-- once they complete, we can drop the old indexes. +INSERT into background_updates (update_name, progress_json, depends_on) + VALUES ( + 'drop_device_list_streams_non_unique_indexes', '{}', + 'device_lists_remote_extremeties_unique_idx' + ); diff --git a/synapse/storage/data_stores/main/schema/delta/52/e2e_room_keys.sql b/synapse/storage/data_stores/main/schema/delta/52/e2e_room_keys.sql new file mode 100644 index 0000000000..db687cccae --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/52/e2e_room_keys.sql @@ -0,0 +1,53 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* Change version column to an integer so we can do MAX() sensibly + */ +CREATE TABLE e2e_room_keys_versions_new ( + user_id TEXT NOT NULL, + version BIGINT NOT NULL, + algorithm TEXT NOT NULL, + auth_data TEXT NOT NULL, + deleted SMALLINT DEFAULT 0 NOT NULL +); + +INSERT INTO e2e_room_keys_versions_new + SELECT user_id, CAST(version as BIGINT), algorithm, auth_data, deleted FROM e2e_room_keys_versions; + +DROP TABLE e2e_room_keys_versions; +ALTER TABLE e2e_room_keys_versions_new RENAME TO e2e_room_keys_versions; + +CREATE UNIQUE INDEX e2e_room_keys_versions_idx ON e2e_room_keys_versions(user_id, version); + +/* Change e2e_rooms_keys to match + */ +CREATE TABLE e2e_room_keys_new ( + user_id TEXT NOT NULL, + room_id TEXT NOT NULL, + session_id TEXT NOT NULL, + version BIGINT NOT NULL, + first_message_index INT, + forwarded_count INT, + is_verified BOOLEAN, + session_data TEXT NOT NULL +); + +INSERT INTO e2e_room_keys_new + SELECT user_id, room_id, session_id, CAST(version as BIGINT), first_message_index, forwarded_count, is_verified, session_data FROM e2e_room_keys; + +DROP TABLE e2e_room_keys; +ALTER TABLE e2e_room_keys_new RENAME TO e2e_room_keys; + +CREATE UNIQUE INDEX e2e_room_keys_idx ON e2e_room_keys(user_id, room_id, session_id); diff --git a/synapse/storage/data_stores/main/schema/delta/53/add_user_type_to_users.sql b/synapse/storage/data_stores/main/schema/delta/53/add_user_type_to_users.sql new file mode 100644 index 0000000000..88ec2f83e5 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/53/add_user_type_to_users.sql @@ -0,0 +1,19 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* The type of the user: NULL for a regular user, or one of the constants in + * synapse.api.constants.UserTypes + */ +ALTER TABLE users ADD COLUMN user_type TEXT DEFAULT NULL; diff --git a/synapse/storage/data_stores/main/schema/delta/53/drop_sent_transactions.sql b/synapse/storage/data_stores/main/schema/delta/53/drop_sent_transactions.sql new file mode 100644 index 0000000000..e372f5a44a --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/53/drop_sent_transactions.sql @@ -0,0 +1,16 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +DROP TABLE IF EXISTS sent_transactions; diff --git a/synapse/storage/data_stores/main/schema/delta/53/event_format_version.sql b/synapse/storage/data_stores/main/schema/delta/53/event_format_version.sql new file mode 100644 index 0000000000..1d977c2834 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/53/event_format_version.sql @@ -0,0 +1,16 @@ +/* Copyright 2019 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +ALTER TABLE event_json ADD COLUMN format_version INTEGER; diff --git a/synapse/storage/data_stores/main/schema/delta/53/user_dir_populate.sql b/synapse/storage/data_stores/main/schema/delta/53/user_dir_populate.sql new file mode 100644 index 0000000000..ffcc896b58 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/53/user_dir_populate.sql @@ -0,0 +1,30 @@ +/* Copyright 2019 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- Set up staging tables +INSERT INTO background_updates (update_name, progress_json) VALUES + ('populate_user_directory_createtables', '{}'); + +-- Run through each room and update the user directory according to who is in it +INSERT INTO background_updates (update_name, progress_json, depends_on) VALUES + ('populate_user_directory_process_rooms', '{}', 'populate_user_directory_createtables'); + +-- Insert all users, if search_all_users is on +INSERT INTO background_updates (update_name, progress_json, depends_on) VALUES + ('populate_user_directory_process_users', '{}', 'populate_user_directory_process_rooms'); + +-- Clean up staging tables +INSERT INTO background_updates (update_name, progress_json, depends_on) VALUES + ('populate_user_directory_cleanup', '{}', 'populate_user_directory_process_users'); diff --git a/synapse/storage/data_stores/main/schema/delta/53/user_ips_index.sql b/synapse/storage/data_stores/main/schema/delta/53/user_ips_index.sql new file mode 100644 index 0000000000..b812c5794f --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/53/user_ips_index.sql @@ -0,0 +1,30 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + -- analyze user_ips, to help ensure the correct indices are used +INSERT INTO background_updates (update_name, progress_json) VALUES + ('user_ips_analyze', '{}'); + +-- delete duplicates +INSERT INTO background_updates (update_name, progress_json, depends_on) VALUES + ('user_ips_remove_dupes', '{}', 'user_ips_analyze'); + +-- add a new unique index to user_ips table +INSERT INTO background_updates (update_name, progress_json, depends_on) VALUES + ('user_ips_device_unique_index', '{}', 'user_ips_remove_dupes'); + +-- drop the old original index +INSERT INTO background_updates (update_name, progress_json, depends_on) VALUES + ('user_ips_drop_nonunique_index', '{}', 'user_ips_device_unique_index'); diff --git a/synapse/storage/data_stores/main/schema/delta/53/user_share.sql b/synapse/storage/data_stores/main/schema/delta/53/user_share.sql new file mode 100644 index 0000000000..5831b1a6f8 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/53/user_share.sql @@ -0,0 +1,44 @@ +/* Copyright 2017 Vector Creations Ltd, 2019 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- Old disused version of the tables below. +DROP TABLE IF EXISTS users_who_share_rooms; + +-- Tables keeping track of what users share rooms. This is a map of local users +-- to local or remote users, per room. Remote users cannot be in the user_id +-- column, only the other_user_id column. There are two tables, one for public +-- rooms and those for private rooms. +CREATE TABLE IF NOT EXISTS users_who_share_public_rooms ( + user_id TEXT NOT NULL, + other_user_id TEXT NOT NULL, + room_id TEXT NOT NULL +); + +CREATE TABLE IF NOT EXISTS users_who_share_private_rooms ( + user_id TEXT NOT NULL, + other_user_id TEXT NOT NULL, + room_id TEXT NOT NULL +); + +CREATE UNIQUE INDEX users_who_share_public_rooms_u_idx ON users_who_share_public_rooms(user_id, other_user_id, room_id); +CREATE INDEX users_who_share_public_rooms_r_idx ON users_who_share_public_rooms(room_id); +CREATE INDEX users_who_share_public_rooms_o_idx ON users_who_share_public_rooms(other_user_id); + +CREATE UNIQUE INDEX users_who_share_private_rooms_u_idx ON users_who_share_private_rooms(user_id, other_user_id, room_id); +CREATE INDEX users_who_share_private_rooms_r_idx ON users_who_share_private_rooms(room_id); +CREATE INDEX users_who_share_private_rooms_o_idx ON users_who_share_private_rooms(other_user_id); + +-- Make sure that we populate the tables initially by resetting the stream ID +UPDATE user_directory_stream_pos SET stream_id = NULL; diff --git a/synapse/storage/data_stores/main/schema/delta/53/user_threepid_id.sql b/synapse/storage/data_stores/main/schema/delta/53/user_threepid_id.sql new file mode 100644 index 0000000000..80c2c573b6 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/53/user_threepid_id.sql @@ -0,0 +1,29 @@ +/* Copyright 2019 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- Tracks which identity server a user bound their threepid via. +CREATE TABLE user_threepid_id_server ( + user_id TEXT NOT NULL, + medium TEXT NOT NULL, + address TEXT NOT NULL, + id_server TEXT NOT NULL +); + +CREATE UNIQUE INDEX user_threepid_id_server_idx ON user_threepid_id_server( + user_id, medium, address, id_server +); + +INSERT INTO background_updates (update_name, progress_json) VALUES + ('user_threepids_grandfather', '{}'); diff --git a/synapse/storage/data_stores/main/schema/delta/53/users_in_public_rooms.sql b/synapse/storage/data_stores/main/schema/delta/53/users_in_public_rooms.sql new file mode 100644 index 0000000000..f7827ca6d2 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/53/users_in_public_rooms.sql @@ -0,0 +1,28 @@ +/* Copyright 2019 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- We don't need the old version of this table. +DROP TABLE IF EXISTS users_in_public_rooms; + +-- Old version of users_in_public_rooms +DROP TABLE IF EXISTS users_who_share_public_rooms; + +-- Track what users are in public rooms. +CREATE TABLE IF NOT EXISTS users_in_public_rooms ( + user_id TEXT NOT NULL, + room_id TEXT NOT NULL +); + +CREATE UNIQUE INDEX users_in_public_rooms_u_idx ON users_in_public_rooms(user_id, room_id); diff --git a/synapse/storage/data_stores/main/schema/delta/54/account_validity_with_renewal.sql b/synapse/storage/data_stores/main/schema/delta/54/account_validity_with_renewal.sql new file mode 100644 index 0000000000..0adb2ad55e --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/54/account_validity_with_renewal.sql @@ -0,0 +1,30 @@ +/* Copyright 2019 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- We previously changed the schema for this table without renaming the file, which means +-- that some databases might still be using the old schema. This ensures Synapse uses the +-- right schema for the table. +DROP TABLE IF EXISTS account_validity; + +-- Track what users are in public rooms. +CREATE TABLE IF NOT EXISTS account_validity ( + user_id TEXT PRIMARY KEY, + expiration_ts_ms BIGINT NOT NULL, + email_sent BOOLEAN NOT NULL, + renewal_token TEXT +); + +CREATE INDEX account_validity_email_sent_idx ON account_validity(email_sent, expiration_ts_ms) +CREATE UNIQUE INDEX account_validity_renewal_string_idx ON account_validity(renewal_token) diff --git a/synapse/storage/data_stores/main/schema/delta/54/add_validity_to_server_keys.sql b/synapse/storage/data_stores/main/schema/delta/54/add_validity_to_server_keys.sql new file mode 100644 index 0000000000..c01aa9d2d9 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/54/add_validity_to_server_keys.sql @@ -0,0 +1,23 @@ +/* Copyright 2019 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* When we can use this key until, before we have to refresh it. */ +ALTER TABLE server_signature_keys ADD COLUMN ts_valid_until_ms BIGINT; + +UPDATE server_signature_keys SET ts_valid_until_ms = ( + SELECT MAX(ts_valid_until_ms) FROM server_keys_json skj WHERE + skj.server_name = server_signature_keys.server_name AND + skj.key_id = server_signature_keys.key_id +); diff --git a/synapse/storage/data_stores/main/schema/delta/54/delete_forward_extremities.sql b/synapse/storage/data_stores/main/schema/delta/54/delete_forward_extremities.sql new file mode 100644 index 0000000000..b062ec840c --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/54/delete_forward_extremities.sql @@ -0,0 +1,23 @@ +/* Copyright 2019 The Matrix.org Foundation C.I.C. + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- Start a background job to cleanup extremities that were incorrectly added +-- by bug #5269. +INSERT INTO background_updates (update_name, progress_json) VALUES + ('delete_soft_failed_extremities', '{}'); + +DROP TABLE IF EXISTS _extremities_to_check; -- To make this delta schema file idempotent. +CREATE TABLE _extremities_to_check AS SELECT event_id FROM event_forward_extremities; +CREATE INDEX _extremities_to_check_id ON _extremities_to_check(event_id); diff --git a/synapse/storage/data_stores/main/schema/delta/54/drop_legacy_tables.sql b/synapse/storage/data_stores/main/schema/delta/54/drop_legacy_tables.sql new file mode 100644 index 0000000000..dbbe682697 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/54/drop_legacy_tables.sql @@ -0,0 +1,30 @@ +/* Copyright 2019 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- we need to do this first due to foreign constraints +DROP TABLE IF EXISTS application_services_regex; + +DROP TABLE IF EXISTS application_services; +DROP TABLE IF EXISTS transaction_id_to_pdu; +DROP TABLE IF EXISTS stats_reporting; +DROP TABLE IF EXISTS current_state_resets; +DROP TABLE IF EXISTS event_content_hashes; +DROP TABLE IF EXISTS event_destinations; +DROP TABLE IF EXISTS event_edge_hashes; +DROP TABLE IF EXISTS event_signatures; +DROP TABLE IF EXISTS feedback; +DROP TABLE IF EXISTS room_hosts; +DROP TABLE IF EXISTS server_tls_certificates; +DROP TABLE IF EXISTS state_forward_extremities; diff --git a/synapse/storage/data_stores/main/schema/delta/54/drop_presence_list.sql b/synapse/storage/data_stores/main/schema/delta/54/drop_presence_list.sql new file mode 100644 index 0000000000..e6ee70c623 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/54/drop_presence_list.sql @@ -0,0 +1,16 @@ +/* Copyright 2019 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +DROP TABLE IF EXISTS presence_list; diff --git a/synapse/storage/data_stores/main/schema/delta/54/relations.sql b/synapse/storage/data_stores/main/schema/delta/54/relations.sql new file mode 100644 index 0000000000..134862b870 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/54/relations.sql @@ -0,0 +1,27 @@ +/* Copyright 2019 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- Tracks related events, like reactions, replies, edits, etc. Note that things +-- in this table are not necessarily "valid", e.g. it may contain edits from +-- people who don't have power to edit other peoples events. +CREATE TABLE IF NOT EXISTS event_relations ( + event_id TEXT NOT NULL, + relates_to_id TEXT NOT NULL, + relation_type TEXT NOT NULL, + aggregation_key TEXT +); + +CREATE UNIQUE INDEX event_relations_id ON event_relations(event_id); +CREATE INDEX event_relations_relates ON event_relations(relates_to_id, relation_type, aggregation_key); diff --git a/synapse/storage/data_stores/main/schema/delta/54/stats.sql b/synapse/storage/data_stores/main/schema/delta/54/stats.sql new file mode 100644 index 0000000000..652e58308e --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/54/stats.sql @@ -0,0 +1,80 @@ +/* Copyright 2018 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE TABLE stats_stream_pos ( + Lock CHAR(1) NOT NULL DEFAULT 'X' UNIQUE, -- Makes sure this table only has one row. + stream_id BIGINT, + CHECK (Lock='X') +); + +INSERT INTO stats_stream_pos (stream_id) VALUES (null); + +CREATE TABLE user_stats ( + user_id TEXT NOT NULL, + ts BIGINT NOT NULL, + bucket_size INT NOT NULL, + public_rooms INT NOT NULL, + private_rooms INT NOT NULL +); + +CREATE UNIQUE INDEX user_stats_user_ts ON user_stats(user_id, ts); + +CREATE TABLE room_stats ( + room_id TEXT NOT NULL, + ts BIGINT NOT NULL, + bucket_size INT NOT NULL, + current_state_events INT NOT NULL, + joined_members INT NOT NULL, + invited_members INT NOT NULL, + left_members INT NOT NULL, + banned_members INT NOT NULL, + state_events INT NOT NULL +); + +CREATE UNIQUE INDEX room_stats_room_ts ON room_stats(room_id, ts); + +-- cache of current room state; useful for the publicRooms list +CREATE TABLE room_state ( + room_id TEXT NOT NULL, + join_rules TEXT, + history_visibility TEXT, + encryption TEXT, + name TEXT, + topic TEXT, + avatar TEXT, + canonical_alias TEXT + -- get aliases straight from the right table +); + +CREATE UNIQUE INDEX room_state_room ON room_state(room_id); + +CREATE TABLE room_stats_earliest_token ( + room_id TEXT NOT NULL, + token BIGINT NOT NULL +); + +CREATE UNIQUE INDEX room_stats_earliest_token_idx ON room_stats_earliest_token(room_id); + +-- Set up staging tables +INSERT INTO background_updates (update_name, progress_json) VALUES + ('populate_stats_createtables', '{}'); + +-- Run through each room and update stats +INSERT INTO background_updates (update_name, progress_json, depends_on) VALUES + ('populate_stats_process_rooms', '{}', 'populate_stats_createtables'); + +-- Clean up staging tables +INSERT INTO background_updates (update_name, progress_json, depends_on) VALUES + ('populate_stats_cleanup', '{}', 'populate_stats_process_rooms'); diff --git a/synapse/storage/data_stores/main/schema/delta/54/stats2.sql b/synapse/storage/data_stores/main/schema/delta/54/stats2.sql new file mode 100644 index 0000000000..3b2d48447f --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/54/stats2.sql @@ -0,0 +1,28 @@ +/* Copyright 2019 The Matrix.org Foundation C.I.C. + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- This delta file gets run after `54/stats.sql` delta. + +-- We want to add some indices to the temporary stats table, so we re-insert +-- 'populate_stats_createtables' if we are still processing the rooms update. +INSERT INTO background_updates (update_name, progress_json) + SELECT 'populate_stats_createtables', '{}' + WHERE + 'populate_stats_process_rooms' IN ( + SELECT update_name FROM background_updates + ) + AND 'populate_stats_createtables' NOT IN ( -- don't insert if already exists + SELECT update_name FROM background_updates + ); diff --git a/synapse/storage/data_stores/main/schema/delta/55/access_token_expiry.sql b/synapse/storage/data_stores/main/schema/delta/55/access_token_expiry.sql new file mode 100644 index 0000000000..4590604bfd --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/55/access_token_expiry.sql @@ -0,0 +1,18 @@ +/* Copyright 2019 The Matrix.org Foundation C.I.C. + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- when this access token can be used until, in ms since the epoch. NULL means the token +-- never expires. +ALTER TABLE access_tokens ADD COLUMN valid_until_ms BIGINT; diff --git a/synapse/storage/data_stores/main/schema/delta/55/track_threepid_validations.sql b/synapse/storage/data_stores/main/schema/delta/55/track_threepid_validations.sql new file mode 100644 index 0000000000..a8eced2e0a --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/55/track_threepid_validations.sql @@ -0,0 +1,31 @@ +/* Copyright 2019 The Matrix.org Foundation C.I.C. + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ +CREATE TABLE IF NOT EXISTS threepid_validation_session ( + session_id TEXT PRIMARY KEY, + medium TEXT NOT NULL, + address TEXT NOT NULL, + client_secret TEXT NOT NULL, + last_send_attempt BIGINT NOT NULL, + validated_at BIGINT +); + +CREATE TABLE IF NOT EXISTS threepid_validation_token ( + token TEXT PRIMARY KEY, + session_id TEXT NOT NULL, + next_link TEXT, + expires BIGINT NOT NULL +); + +CREATE INDEX threepid_validation_token_session_id ON threepid_validation_token(session_id); diff --git a/synapse/storage/data_stores/main/schema/delta/55/users_alter_deactivated.sql b/synapse/storage/data_stores/main/schema/delta/55/users_alter_deactivated.sql new file mode 100644 index 0000000000..dabdde489b --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/55/users_alter_deactivated.sql @@ -0,0 +1,19 @@ +/* Copyright 2019 The Matrix.org Foundation C.I.C. + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +ALTER TABLE users ADD deactivated SMALLINT DEFAULT 0 NOT NULL; + +INSERT INTO background_updates (update_name, progress_json) VALUES + ('users_set_deactivated_flag', '{}'); diff --git a/synapse/storage/data_stores/main/schema/delta/56/add_spans_to_device_lists.sql b/synapse/storage/data_stores/main/schema/delta/56/add_spans_to_device_lists.sql new file mode 100644 index 0000000000..41807eb1e7 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/56/add_spans_to_device_lists.sql @@ -0,0 +1,20 @@ +/* Copyright 2019 The Matrix.org Foundation C.I.C + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* + * Opentracing context data for inclusion in the device_list_update EDUs, as a + * json-encoded dictionary. NULL if opentracing is disabled (or not enabled for this destination). + */ +ALTER TABLE device_lists_outbound_pokes ADD opentracing_context TEXT; diff --git a/synapse/storage/data_stores/main/schema/delta/56/current_state_events_membership.sql b/synapse/storage/data_stores/main/schema/delta/56/current_state_events_membership.sql new file mode 100644 index 0000000000..473018676f --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/56/current_state_events_membership.sql @@ -0,0 +1,22 @@ +/* Copyright 2019 The Matrix.org Foundation C.I.C. + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- We add membership to current state so that we don't need to join against +-- room_memberships, which can be surprisingly costly (we do such queries +-- very frequently). +-- This will be null for non-membership events and the content.membership key +-- for membership events. (Will also be null for membership events until the +-- background update job has finished). +ALTER TABLE current_state_events ADD membership TEXT; diff --git a/synapse/storage/data_stores/main/schema/delta/56/current_state_events_membership_mk2.sql b/synapse/storage/data_stores/main/schema/delta/56/current_state_events_membership_mk2.sql new file mode 100644 index 0000000000..3133d42d4a --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/56/current_state_events_membership_mk2.sql @@ -0,0 +1,24 @@ +/* Copyright 2019 The Matrix.org Foundation C.I.C. + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- We add membership to current state so that we don't need to join against +-- room_memberships, which can be surprisingly costly (we do such queries +-- very frequently). +-- This will be null for non-membership events and the content.membership key +-- for membership events. (Will also be null for membership events until the +-- background update job has finished). + +INSERT INTO background_updates (update_name, progress_json) VALUES + ('current_state_events_membership', '{}'); diff --git a/synapse/storage/data_stores/main/schema/delta/56/destinations_failure_ts.sql b/synapse/storage/data_stores/main/schema/delta/56/destinations_failure_ts.sql new file mode 100644 index 0000000000..f00889290b --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/56/destinations_failure_ts.sql @@ -0,0 +1,25 @@ +/* Copyright 2019 The Matrix.org Foundation C.I.C + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* + * Record the timestamp when a given server started failing + */ +ALTER TABLE destinations ADD failure_ts BIGINT; + +/* as a rough approximation, we assume that the server started failing at + * retry_interval before the last retry + */ +UPDATE destinations SET failure_ts = retry_last_ts - retry_interval + WHERE retry_last_ts > 0; diff --git a/synapse/storage/data_stores/main/schema/delta/56/destinations_retry_interval_type.sql.postgres b/synapse/storage/data_stores/main/schema/delta/56/destinations_retry_interval_type.sql.postgres new file mode 100644 index 0000000000..b9bbb18a91 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/56/destinations_retry_interval_type.sql.postgres @@ -0,0 +1,18 @@ +/* Copyright 2019 The Matrix.org Foundation C.I.C + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- We want to store large retry intervals so we upgrade the column from INT +-- to BIGINT. We don't need to do this on SQLite. +ALTER TABLE destinations ALTER retry_interval SET DATA TYPE BIGINT; diff --git a/synapse/storage/data_stores/main/schema/delta/56/devices_last_seen.sql b/synapse/storage/data_stores/main/schema/delta/56/devices_last_seen.sql new file mode 100644 index 0000000000..dfa902d0ba --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/56/devices_last_seen.sql @@ -0,0 +1,24 @@ +/* Copyright 2019 Matrix.org Foundation CIC + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- Track last seen information for a device in the devices table, rather +-- than relying on it being in the user_ips table (which we want to be able +-- to purge old entries from) +ALTER TABLE devices ADD COLUMN last_seen BIGINT; +ALTER TABLE devices ADD COLUMN ip TEXT; +ALTER TABLE devices ADD COLUMN user_agent TEXT; + +INSERT INTO background_updates (update_name, progress_json) VALUES + ('devices_last_seen', '{}'); diff --git a/synapse/storage/data_stores/main/schema/delta/56/drop_unused_event_tables.sql b/synapse/storage/data_stores/main/schema/delta/56/drop_unused_event_tables.sql new file mode 100644 index 0000000000..9f09922c67 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/56/drop_unused_event_tables.sql @@ -0,0 +1,20 @@ +/* Copyright 2019 The Matrix.org Foundation C.I.C. + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- these tables are never used. +DROP TABLE IF EXISTS room_names; +DROP TABLE IF EXISTS topics; +DROP TABLE IF EXISTS history_visibility; +DROP TABLE IF EXISTS guest_access; diff --git a/synapse/storage/data_stores/main/schema/delta/56/fix_room_keys_index.sql b/synapse/storage/data_stores/main/schema/delta/56/fix_room_keys_index.sql new file mode 100644 index 0000000000..014cb3b538 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/56/fix_room_keys_index.sql @@ -0,0 +1,18 @@ +/* Copyright 2019 Matrix.org Foundation CIC + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- version is supposed to be part of the room keys index +CREATE UNIQUE INDEX e2e_room_keys_with_version_idx ON e2e_room_keys(user_id, version, room_id, session_id); +DROP INDEX IF EXISTS e2e_room_keys_idx; diff --git a/synapse/storage/data_stores/main/schema/delta/56/public_room_list_idx.sql b/synapse/storage/data_stores/main/schema/delta/56/public_room_list_idx.sql new file mode 100644 index 0000000000..7be31ffebb --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/56/public_room_list_idx.sql @@ -0,0 +1,16 @@ +/* Copyright 2019 The Matrix.org Foundation C.I.C. + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE INDEX public_room_list_stream_network ON public_room_list_stream (appservice_id, network_id, room_id); diff --git a/synapse/storage/data_stores/main/schema/delta/56/redaction_censor.sql b/synapse/storage/data_stores/main/schema/delta/56/redaction_censor.sql new file mode 100644 index 0000000000..fe51b02309 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/56/redaction_censor.sql @@ -0,0 +1,17 @@ +/* Copyright 2019 The Matrix.org Foundation C.I.C. + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +ALTER TABLE redactions ADD COLUMN have_censored BOOL NOT NULL DEFAULT false; +CREATE INDEX redactions_have_censored ON redactions(event_id) WHERE not have_censored; diff --git a/synapse/storage/data_stores/main/schema/delta/56/redaction_censor2.sql b/synapse/storage/data_stores/main/schema/delta/56/redaction_censor2.sql new file mode 100644 index 0000000000..77a5eca499 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/56/redaction_censor2.sql @@ -0,0 +1,20 @@ +/* Copyright 2019 The Matrix.org Foundation C.I.C. + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +ALTER TABLE redactions ADD COLUMN received_ts BIGINT; +CREATE INDEX redactions_have_censored_ts ON redactions(received_ts) WHERE not have_censored; + +INSERT INTO background_updates (update_name, progress_json) VALUES + ('redactions_received_ts', '{}'); diff --git a/synapse/storage/data_stores/main/schema/delta/56/redaction_censor3_fix_update.sql.postgres b/synapse/storage/data_stores/main/schema/delta/56/redaction_censor3_fix_update.sql.postgres new file mode 100644 index 0000000000..67471f3ef5 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/56/redaction_censor3_fix_update.sql.postgres @@ -0,0 +1,25 @@ +/* Copyright 2019 The Matrix.org Foundation C.I.C. + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + +-- There was a bug where we may have updated censored redactions as bytes, +-- which can (somehow) cause json to be inserted hex encoded. These updates go +-- and undoes any such hex encoded JSON. + +INSERT into background_updates (update_name, progress_json) + VALUES ('event_fix_redactions_bytes_create_index', '{}'); + +INSERT into background_updates (update_name, progress_json, depends_on) + VALUES ('event_fix_redactions_bytes', '{}', 'event_fix_redactions_bytes_create_index'); diff --git a/synapse/storage/data_stores/main/schema/delta/56/room_membership_idx.sql b/synapse/storage/data_stores/main/schema/delta/56/room_membership_idx.sql new file mode 100644 index 0000000000..92ab1f5e65 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/56/room_membership_idx.sql @@ -0,0 +1,18 @@ +/* Copyright 2019 The Matrix.org Foundation C.I.C. + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- Adds an index on room_memberships for fetching all forgotten rooms for a user +INSERT INTO background_updates (update_name, progress_json) VALUES + ('room_membership_forgotten_idx', '{}'); diff --git a/synapse/storage/data_stores/main/schema/delta/56/stats_separated.sql b/synapse/storage/data_stores/main/schema/delta/56/stats_separated.sql new file mode 100644 index 0000000000..163529c071 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/56/stats_separated.sql @@ -0,0 +1,152 @@ +/* Copyright 2018 New Vector Ltd + * Copyright 2019 The Matrix.org Foundation C.I.C. + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + +----- First clean up from previous versions of room stats. + +-- First remove old stats stuff +DROP TABLE IF EXISTS room_stats; +DROP TABLE IF EXISTS room_state; +DROP TABLE IF EXISTS room_stats_state; +DROP TABLE IF EXISTS user_stats; +DROP TABLE IF EXISTS room_stats_earliest_tokens; +DROP TABLE IF EXISTS _temp_populate_stats_position; +DROP TABLE IF EXISTS _temp_populate_stats_rooms; +DROP TABLE IF EXISTS stats_stream_pos; + +-- Unschedule old background updates if they're still scheduled +DELETE FROM background_updates WHERE update_name IN ( + 'populate_stats_createtables', + 'populate_stats_process_rooms', + 'populate_stats_process_users', + 'populate_stats_cleanup' +); + +INSERT INTO background_updates (update_name, progress_json, depends_on) VALUES + ('populate_stats_process_rooms', '{}', ''); + +INSERT INTO background_updates (update_name, progress_json, depends_on) VALUES + ('populate_stats_process_users', '{}', 'populate_stats_process_rooms'); + +----- Create tables for our version of room stats. + +-- single-row table to track position of incremental updates +DROP TABLE IF EXISTS stats_incremental_position; +CREATE TABLE stats_incremental_position ( + Lock CHAR(1) NOT NULL DEFAULT 'X' UNIQUE, -- Makes sure this table only has one row. + stream_id BIGINT NOT NULL, + CHECK (Lock='X') +); + +-- insert a null row and make sure it is the only one. +INSERT INTO stats_incremental_position ( + stream_id +) SELECT COALESCE(MAX(stream_ordering), 0) from events; + +-- represents PRESENT room statistics for a room +-- only holds absolute fields +DROP TABLE IF EXISTS room_stats_current; +CREATE TABLE room_stats_current ( + room_id TEXT NOT NULL PRIMARY KEY, + + -- These are absolute counts + current_state_events INT NOT NULL, + joined_members INT NOT NULL, + invited_members INT NOT NULL, + left_members INT NOT NULL, + banned_members INT NOT NULL, + + local_users_in_room INT NOT NULL, + + -- The maximum delta stream position that this row takes into account. + completed_delta_stream_id BIGINT NOT NULL +); + + +-- represents HISTORICAL room statistics for a room +DROP TABLE IF EXISTS room_stats_historical; +CREATE TABLE room_stats_historical ( + room_id TEXT NOT NULL, + -- These stats cover the time from (end_ts - bucket_size)...end_ts (in ms). + -- Note that end_ts is quantised. + end_ts BIGINT NOT NULL, + bucket_size BIGINT NOT NULL, + + -- These stats are absolute counts + current_state_events BIGINT NOT NULL, + joined_members BIGINT NOT NULL, + invited_members BIGINT NOT NULL, + left_members BIGINT NOT NULL, + banned_members BIGINT NOT NULL, + local_users_in_room BIGINT NOT NULL, + + -- These stats are per time slice + total_events BIGINT NOT NULL, + total_event_bytes BIGINT NOT NULL, + + PRIMARY KEY (room_id, end_ts) +); + +-- We use this index to speed up deletion of ancient room stats. +CREATE INDEX room_stats_historical_end_ts ON room_stats_historical (end_ts); + +-- represents PRESENT statistics for a user +-- only holds absolute fields +DROP TABLE IF EXISTS user_stats_current; +CREATE TABLE user_stats_current ( + user_id TEXT NOT NULL PRIMARY KEY, + + joined_rooms BIGINT NOT NULL, + + -- The maximum delta stream position that this row takes into account. + completed_delta_stream_id BIGINT NOT NULL +); + +-- represents HISTORICAL statistics for a user +DROP TABLE IF EXISTS user_stats_historical; +CREATE TABLE user_stats_historical ( + user_id TEXT NOT NULL, + end_ts BIGINT NOT NULL, + bucket_size BIGINT NOT NULL, + + joined_rooms BIGINT NOT NULL, + + invites_sent BIGINT NOT NULL, + rooms_created BIGINT NOT NULL, + total_events BIGINT NOT NULL, + total_event_bytes BIGINT NOT NULL, + + PRIMARY KEY (user_id, end_ts) +); + +-- We use this index to speed up deletion of ancient user stats. +CREATE INDEX user_stats_historical_end_ts ON user_stats_historical (end_ts); + + +CREATE TABLE room_stats_state ( + room_id TEXT NOT NULL, + name TEXT, + canonical_alias TEXT, + join_rules TEXT, + history_visibility TEXT, + encryption TEXT, + avatar TEXT, + guest_access TEXT, + is_federatable BOOLEAN, + topic TEXT +); + +CREATE UNIQUE INDEX room_stats_state_room ON room_stats_state(room_id); diff --git a/synapse/storage/data_stores/main/schema/delta/56/unique_user_filter_index.py b/synapse/storage/data_stores/main/schema/delta/56/unique_user_filter_index.py new file mode 100644 index 0000000000..1de8b54961 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/56/unique_user_filter_index.py @@ -0,0 +1,52 @@ +import logging + +from synapse.storage.engines import PostgresEngine + +logger = logging.getLogger(__name__) + + +""" +This migration updates the user_filters table as follows: + + - drops any (user_id, filter_id) duplicates + - makes the columns NON-NULLable + - turns the index into a UNIQUE index +""" + + +def run_upgrade(cur, database_engine, *args, **kwargs): + pass + + +def run_create(cur, database_engine, *args, **kwargs): + if isinstance(database_engine, PostgresEngine): + select_clause = """ + SELECT DISTINCT ON (user_id, filter_id) user_id, filter_id, filter_json + FROM user_filters + """ + else: + select_clause = """ + SELECT * FROM user_filters GROUP BY user_id, filter_id + """ + sql = """ + DROP TABLE IF EXISTS user_filters_migration; + DROP INDEX IF EXISTS user_filters_unique; + CREATE TABLE user_filters_migration ( + user_id TEXT NOT NULL, + filter_id BIGINT NOT NULL, + filter_json BYTEA NOT NULL + ); + INSERT INTO user_filters_migration (user_id, filter_id, filter_json) + %s; + CREATE UNIQUE INDEX user_filters_unique ON user_filters_migration + (user_id, filter_id); + DROP TABLE user_filters; + ALTER TABLE user_filters_migration RENAME TO user_filters; + """ % ( + select_clause, + ) + + if isinstance(database_engine, PostgresEngine): + cur.execute(sql) + else: + cur.executescript(sql) diff --git a/synapse/storage/data_stores/main/schema/delta/56/user_external_ids.sql b/synapse/storage/data_stores/main/schema/delta/56/user_external_ids.sql new file mode 100644 index 0000000000..91390c4527 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/56/user_external_ids.sql @@ -0,0 +1,24 @@ +/* Copyright 2019 The Matrix.org Foundation C.I.C. + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* + * a table which records mappings from external auth providers to mxids + */ +CREATE TABLE IF NOT EXISTS user_external_ids ( + auth_provider TEXT NOT NULL, + external_id TEXT NOT NULL, + user_id TEXT NOT NULL, + UNIQUE (auth_provider, external_id) +); diff --git a/synapse/storage/data_stores/main/schema/delta/56/users_in_public_rooms_idx.sql b/synapse/storage/data_stores/main/schema/delta/56/users_in_public_rooms_idx.sql new file mode 100644 index 0000000000..149f8be8b6 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/56/users_in_public_rooms_idx.sql @@ -0,0 +1,17 @@ +/* Copyright 2019 Matrix.org Foundation CIC + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- this was apparently forgotten when the table was created back in delta 53. +CREATE INDEX users_in_public_rooms_r_idx ON users_in_public_rooms(room_id); diff --git a/synapse/storage/data_stores/main/schema/full_schemas/16/application_services.sql b/synapse/storage/data_stores/main/schema/full_schemas/16/application_services.sql new file mode 100644 index 0000000000..883fcd10b2 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/full_schemas/16/application_services.sql @@ -0,0 +1,37 @@ +/* Copyright 2015, 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* We used to create tables called application_services and + * application_services_regex, but these are no longer used and are removed in + * delta 54. + */ + + +CREATE TABLE IF NOT EXISTS application_services_state( + as_id TEXT PRIMARY KEY, + state VARCHAR(5), + last_txn INTEGER +); + +CREATE TABLE IF NOT EXISTS application_services_txns( + as_id TEXT NOT NULL, + txn_id INTEGER NOT NULL, + event_ids TEXT NOT NULL, + UNIQUE(as_id, txn_id) +); + +CREATE INDEX application_services_txns_id ON application_services_txns ( + as_id +); diff --git a/synapse/storage/data_stores/main/schema/full_schemas/16/event_edges.sql b/synapse/storage/data_stores/main/schema/full_schemas/16/event_edges.sql new file mode 100644 index 0000000000..10ce2aa7a0 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/full_schemas/16/event_edges.sql @@ -0,0 +1,70 @@ +/* Copyright 2014-2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* We used to create tables called event_destinations and + * state_forward_extremities, but these are no longer used and are removed in + * delta 54. + */ + +CREATE TABLE IF NOT EXISTS event_forward_extremities( + event_id TEXT NOT NULL, + room_id TEXT NOT NULL, + UNIQUE (event_id, room_id) +); + +CREATE INDEX ev_extrem_room ON event_forward_extremities(room_id); +CREATE INDEX ev_extrem_id ON event_forward_extremities(event_id); + + +CREATE TABLE IF NOT EXISTS event_backward_extremities( + event_id TEXT NOT NULL, + room_id TEXT NOT NULL, + UNIQUE (event_id, room_id) +); + +CREATE INDEX ev_b_extrem_room ON event_backward_extremities(room_id); +CREATE INDEX ev_b_extrem_id ON event_backward_extremities(event_id); + + +CREATE TABLE IF NOT EXISTS event_edges( + event_id TEXT NOT NULL, + prev_event_id TEXT NOT NULL, + room_id TEXT NOT NULL, + is_state BOOL NOT NULL, -- true if this is a prev_state edge rather than a regular + -- event dag edge. + UNIQUE (event_id, prev_event_id, room_id, is_state) +); + +CREATE INDEX ev_edges_id ON event_edges(event_id); +CREATE INDEX ev_edges_prev_id ON event_edges(prev_event_id); + + +CREATE TABLE IF NOT EXISTS room_depth( + room_id TEXT NOT NULL, + min_depth INTEGER NOT NULL, + UNIQUE (room_id) +); + +CREATE INDEX room_depth_room ON room_depth(room_id); + +CREATE TABLE IF NOT EXISTS event_auth( + event_id TEXT NOT NULL, + auth_id TEXT NOT NULL, + room_id TEXT NOT NULL, + UNIQUE (event_id, auth_id, room_id) +); + +CREATE INDEX evauth_edges_id ON event_auth(event_id); +CREATE INDEX evauth_edges_auth_id ON event_auth(auth_id); diff --git a/synapse/storage/data_stores/main/schema/full_schemas/16/event_signatures.sql b/synapse/storage/data_stores/main/schema/full_schemas/16/event_signatures.sql new file mode 100644 index 0000000000..95826da431 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/full_schemas/16/event_signatures.sql @@ -0,0 +1,38 @@ +/* Copyright 2014-2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + /* We used to create tables called event_content_hashes and event_edge_hashes, + * but these are no longer used and are removed in delta 54. + */ + +CREATE TABLE IF NOT EXISTS event_reference_hashes ( + event_id TEXT, + algorithm TEXT, + hash bytea, + UNIQUE (event_id, algorithm) +); + +CREATE INDEX event_reference_hashes_id ON event_reference_hashes(event_id); + + +CREATE TABLE IF NOT EXISTS event_signatures ( + event_id TEXT, + signature_name TEXT, + key_id TEXT, + signature bytea, + UNIQUE (event_id, signature_name, key_id) +); + +CREATE INDEX event_signatures_id ON event_signatures(event_id); diff --git a/synapse/storage/data_stores/main/schema/full_schemas/16/im.sql b/synapse/storage/data_stores/main/schema/full_schemas/16/im.sql new file mode 100644 index 0000000000..a1a2aa8e5b --- /dev/null +++ b/synapse/storage/data_stores/main/schema/full_schemas/16/im.sql @@ -0,0 +1,120 @@ +/* Copyright 2014-2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* We used to create tables called room_hosts and feedback, + * but these are no longer used and are removed in delta 54. + */ + +CREATE TABLE IF NOT EXISTS events( + stream_ordering INTEGER PRIMARY KEY, + topological_ordering BIGINT NOT NULL, + event_id TEXT NOT NULL, + type TEXT NOT NULL, + room_id TEXT NOT NULL, + + -- 'content' used to be created NULLable, but as of delta 50 we drop that constraint. + -- the hack we use to drop the constraint doesn't work for an in-memory sqlite + -- database, which breaks the sytests. Hence, we no longer make it nullable. + content TEXT, + + unrecognized_keys TEXT, + processed BOOL NOT NULL, + outlier BOOL NOT NULL, + depth BIGINT DEFAULT 0 NOT NULL, + UNIQUE (event_id) +); + +CREATE INDEX events_stream_ordering ON events (stream_ordering); +CREATE INDEX events_topological_ordering ON events (topological_ordering); +CREATE INDEX events_order ON events (topological_ordering, stream_ordering); +CREATE INDEX events_room_id ON events (room_id); +CREATE INDEX events_order_room ON events ( + room_id, topological_ordering, stream_ordering +); + + +CREATE TABLE IF NOT EXISTS event_json( + event_id TEXT NOT NULL, + room_id TEXT NOT NULL, + internal_metadata TEXT NOT NULL, + json TEXT NOT NULL, + UNIQUE (event_id) +); + +CREATE INDEX event_json_room_id ON event_json(room_id); + + +CREATE TABLE IF NOT EXISTS state_events( + event_id TEXT NOT NULL, + room_id TEXT NOT NULL, + type TEXT NOT NULL, + state_key TEXT NOT NULL, + prev_state TEXT, + UNIQUE (event_id) +); + +CREATE INDEX state_events_room_id ON state_events (room_id); +CREATE INDEX state_events_type ON state_events (type); +CREATE INDEX state_events_state_key ON state_events (state_key); + + +CREATE TABLE IF NOT EXISTS current_state_events( + event_id TEXT NOT NULL, + room_id TEXT NOT NULL, + type TEXT NOT NULL, + state_key TEXT NOT NULL, + UNIQUE (event_id), + UNIQUE (room_id, type, state_key) +); + +CREATE INDEX current_state_events_room_id ON current_state_events (room_id); +CREATE INDEX current_state_events_type ON current_state_events (type); +CREATE INDEX current_state_events_state_key ON current_state_events (state_key); + +CREATE TABLE IF NOT EXISTS room_memberships( + event_id TEXT NOT NULL, + user_id TEXT NOT NULL, + sender TEXT NOT NULL, + room_id TEXT NOT NULL, + membership TEXT NOT NULL, + UNIQUE (event_id) +); + +CREATE INDEX room_memberships_room_id ON room_memberships (room_id); +CREATE INDEX room_memberships_user_id ON room_memberships (user_id); + +CREATE TABLE IF NOT EXISTS topics( + event_id TEXT NOT NULL, + room_id TEXT NOT NULL, + topic TEXT NOT NULL, + UNIQUE (event_id) +); + +CREATE INDEX topics_room_id ON topics(room_id); + +CREATE TABLE IF NOT EXISTS room_names( + event_id TEXT NOT NULL, + room_id TEXT NOT NULL, + name TEXT NOT NULL, + UNIQUE (event_id) +); + +CREATE INDEX room_names_room_id ON room_names(room_id); + +CREATE TABLE IF NOT EXISTS rooms( + room_id TEXT PRIMARY KEY NOT NULL, + is_public BOOL, + creator TEXT +); diff --git a/synapse/storage/data_stores/main/schema/full_schemas/16/keys.sql b/synapse/storage/data_stores/main/schema/full_schemas/16/keys.sql new file mode 100644 index 0000000000..11cdffdbb3 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/full_schemas/16/keys.sql @@ -0,0 +1,26 @@ +/* Copyright 2014-2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- we used to create a table called server_tls_certificates, but this is no +-- longer used, and is removed in delta 54. + +CREATE TABLE IF NOT EXISTS server_signature_keys( + server_name TEXT, -- Server name. + key_id TEXT, -- Key version. + from_server TEXT, -- Which key server the key was fetched form. + ts_added_ms BIGINT, -- When the key was added. + verify_key bytea, -- NACL verification key. + UNIQUE (server_name, key_id) +); diff --git a/synapse/storage/data_stores/main/schema/full_schemas/16/media_repository.sql b/synapse/storage/data_stores/main/schema/full_schemas/16/media_repository.sql new file mode 100644 index 0000000000..8f3759bb2a --- /dev/null +++ b/synapse/storage/data_stores/main/schema/full_schemas/16/media_repository.sql @@ -0,0 +1,68 @@ +/* Copyright 2014-2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE TABLE IF NOT EXISTS local_media_repository ( + media_id TEXT, -- The id used to refer to the media. + media_type TEXT, -- The MIME-type of the media. + media_length INTEGER, -- Length of the media in bytes. + created_ts BIGINT, -- When the content was uploaded in ms. + upload_name TEXT, -- The name the media was uploaded with. + user_id TEXT, -- The user who uploaded the file. + UNIQUE (media_id) +); + +CREATE TABLE IF NOT EXISTS local_media_repository_thumbnails ( + media_id TEXT, -- The id used to refer to the media. + thumbnail_width INTEGER, -- The width of the thumbnail in pixels. + thumbnail_height INTEGER, -- The height of the thumbnail in pixels. + thumbnail_type TEXT, -- The MIME-type of the thumbnail. + thumbnail_method TEXT, -- The method used to make the thumbnail. + thumbnail_length INTEGER, -- The length of the thumbnail in bytes. + UNIQUE ( + media_id, thumbnail_width, thumbnail_height, thumbnail_type + ) +); + +CREATE INDEX local_media_repository_thumbnails_media_id + ON local_media_repository_thumbnails (media_id); + +CREATE TABLE IF NOT EXISTS remote_media_cache ( + media_origin TEXT, -- The remote HS the media came from. + media_id TEXT, -- The id used to refer to the media on that server. + media_type TEXT, -- The MIME-type of the media. + created_ts BIGINT, -- When the content was uploaded in ms. + upload_name TEXT, -- The name the media was uploaded with. + media_length INTEGER, -- Length of the media in bytes. + filesystem_id TEXT, -- The name used to store the media on disk. + UNIQUE (media_origin, media_id) +); + +CREATE TABLE IF NOT EXISTS remote_media_cache_thumbnails ( + media_origin TEXT, -- The remote HS the media came from. + media_id TEXT, -- The id used to refer to the media. + thumbnail_width INTEGER, -- The width of the thumbnail in pixels. + thumbnail_height INTEGER, -- The height of the thumbnail in pixels. + thumbnail_method TEXT, -- The method used to make the thumbnail + thumbnail_type TEXT, -- The MIME-type of the thumbnail. + thumbnail_length INTEGER, -- The length of the thumbnail in bytes. + filesystem_id TEXT, -- The name used to store the media on disk. + UNIQUE ( + media_origin, media_id, thumbnail_width, thumbnail_height, + thumbnail_type + ) +); + +CREATE INDEX remote_media_cache_thumbnails_media_id + ON remote_media_cache_thumbnails (media_id); diff --git a/synapse/storage/data_stores/main/schema/full_schemas/16/presence.sql b/synapse/storage/data_stores/main/schema/full_schemas/16/presence.sql new file mode 100644 index 0000000000..01d2d8f833 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/full_schemas/16/presence.sql @@ -0,0 +1,32 @@ +/* Copyright 2014-2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ +CREATE TABLE IF NOT EXISTS presence( + user_id TEXT NOT NULL, + state VARCHAR(20), + status_msg TEXT, + mtime BIGINT, -- miliseconds since last state change + UNIQUE (user_id) +); + +-- For each of /my/ users which possibly-remote users are allowed to see their +-- presence state +CREATE TABLE IF NOT EXISTS presence_allow_inbound( + observed_user_id TEXT NOT NULL, + observer_user_id TEXT NOT NULL, -- a UserID, + UNIQUE (observed_user_id, observer_user_id) +); + +-- We used to create a table called presence_list, but this is no longer used +-- and is removed in delta 54. \ No newline at end of file diff --git a/synapse/storage/data_stores/main/schema/full_schemas/16/profiles.sql b/synapse/storage/data_stores/main/schema/full_schemas/16/profiles.sql new file mode 100644 index 0000000000..c04f4747d9 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/full_schemas/16/profiles.sql @@ -0,0 +1,20 @@ +/* Copyright 2014-2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ +CREATE TABLE IF NOT EXISTS profiles( + user_id TEXT NOT NULL, + displayname TEXT, + avatar_url TEXT, + UNIQUE(user_id) +); diff --git a/synapse/storage/data_stores/main/schema/full_schemas/16/push.sql b/synapse/storage/data_stores/main/schema/full_schemas/16/push.sql new file mode 100644 index 0000000000..e44465cf45 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/full_schemas/16/push.sql @@ -0,0 +1,74 @@ +/* Copyright 2015, 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE TABLE IF NOT EXISTS rejections( + event_id TEXT NOT NULL, + reason TEXT NOT NULL, + last_check TEXT NOT NULL, + UNIQUE (event_id) +); + +-- Push notification endpoints that users have configured +CREATE TABLE IF NOT EXISTS pushers ( + id BIGINT PRIMARY KEY, + user_name TEXT NOT NULL, + access_token BIGINT DEFAULT NULL, + profile_tag VARCHAR(32) NOT NULL, + kind VARCHAR(8) NOT NULL, + app_id VARCHAR(64) NOT NULL, + app_display_name VARCHAR(64) NOT NULL, + device_display_name VARCHAR(128) NOT NULL, + pushkey bytea NOT NULL, + ts BIGINT NOT NULL, + lang VARCHAR(8), + data bytea, + last_token TEXT, + last_success BIGINT, + failing_since BIGINT, + UNIQUE (app_id, pushkey) +); + +CREATE TABLE IF NOT EXISTS push_rules ( + id BIGINT PRIMARY KEY, + user_name TEXT NOT NULL, + rule_id TEXT NOT NULL, + priority_class SMALLINT NOT NULL, + priority INTEGER NOT NULL DEFAULT 0, + conditions TEXT NOT NULL, + actions TEXT NOT NULL, + UNIQUE(user_name, rule_id) +); + +CREATE INDEX push_rules_user_name on push_rules (user_name); + +CREATE TABLE IF NOT EXISTS user_filters( + user_id TEXT, + filter_id BIGINT, + filter_json bytea +); + +CREATE INDEX user_filters_by_user_id_filter_id ON user_filters( + user_id, filter_id +); + +CREATE TABLE IF NOT EXISTS push_rules_enable ( + id BIGINT PRIMARY KEY, + user_name TEXT NOT NULL, + rule_id TEXT NOT NULL, + enabled SMALLINT, + UNIQUE(user_name, rule_id) +); + +CREATE INDEX push_rules_enable_user_name on push_rules_enable (user_name); diff --git a/synapse/storage/data_stores/main/schema/full_schemas/16/redactions.sql b/synapse/storage/data_stores/main/schema/full_schemas/16/redactions.sql new file mode 100644 index 0000000000..318f0d9aa5 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/full_schemas/16/redactions.sql @@ -0,0 +1,22 @@ +/* Copyright 2014-2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ +CREATE TABLE IF NOT EXISTS redactions ( + event_id TEXT NOT NULL, + redacts TEXT NOT NULL, + UNIQUE (event_id) +); + +CREATE INDEX redactions_event_id ON redactions (event_id); +CREATE INDEX redactions_redacts ON redactions (redacts); diff --git a/synapse/storage/data_stores/main/schema/full_schemas/16/room_aliases.sql b/synapse/storage/data_stores/main/schema/full_schemas/16/room_aliases.sql new file mode 100644 index 0000000000..d47da3b12f --- /dev/null +++ b/synapse/storage/data_stores/main/schema/full_schemas/16/room_aliases.sql @@ -0,0 +1,29 @@ +/* Copyright 2014-2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE TABLE IF NOT EXISTS room_aliases( + room_alias TEXT NOT NULL, + room_id TEXT NOT NULL, + UNIQUE (room_alias) +); + +CREATE INDEX room_aliases_id ON room_aliases(room_id); + +CREATE TABLE IF NOT EXISTS room_alias_servers( + room_alias TEXT NOT NULL, + server TEXT NOT NULL +); + +CREATE INDEX room_alias_servers_alias ON room_alias_servers(room_alias); diff --git a/synapse/storage/data_stores/main/schema/full_schemas/16/state.sql b/synapse/storage/data_stores/main/schema/full_schemas/16/state.sql new file mode 100644 index 0000000000..96391a8f0e --- /dev/null +++ b/synapse/storage/data_stores/main/schema/full_schemas/16/state.sql @@ -0,0 +1,40 @@ +/* Copyright 2014-2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE TABLE IF NOT EXISTS state_groups( + id BIGINT PRIMARY KEY, + room_id TEXT NOT NULL, + event_id TEXT NOT NULL +); + +CREATE TABLE IF NOT EXISTS state_groups_state( + state_group BIGINT NOT NULL, + room_id TEXT NOT NULL, + type TEXT NOT NULL, + state_key TEXT NOT NULL, + event_id TEXT NOT NULL +); + +CREATE TABLE IF NOT EXISTS event_to_state_groups( + event_id TEXT NOT NULL, + state_group BIGINT NOT NULL, + UNIQUE (event_id) +); + +CREATE INDEX state_groups_id ON state_groups(id); + +CREATE INDEX state_groups_state_id ON state_groups_state(state_group); +CREATE INDEX state_groups_state_tuple ON state_groups_state(room_id, type, state_key); +CREATE INDEX event_to_state_groups_id ON event_to_state_groups(event_id); diff --git a/synapse/storage/data_stores/main/schema/full_schemas/16/transactions.sql b/synapse/storage/data_stores/main/schema/full_schemas/16/transactions.sql new file mode 100644 index 0000000000..17e67bedac --- /dev/null +++ b/synapse/storage/data_stores/main/schema/full_schemas/16/transactions.sql @@ -0,0 +1,44 @@ +/* Copyright 2014-2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ +-- Stores what transaction ids we have received and what our response was +CREATE TABLE IF NOT EXISTS received_transactions( + transaction_id TEXT, + origin TEXT, + ts BIGINT, + response_code INTEGER, + response_json bytea, + has_been_referenced smallint default 0, -- Whether thishas been referenced by a prev_tx + UNIQUE (transaction_id, origin) +); + +CREATE INDEX transactions_have_ref ON received_transactions(origin, has_been_referenced);-- WHERE has_been_referenced = 0; + +-- For sent transactions only. +CREATE TABLE IF NOT EXISTS transaction_id_to_pdu( + transaction_id INTEGER, + destination TEXT, + pdu_id TEXT, + pdu_origin TEXT, + UNIQUE (transaction_id, destination) +); + +CREATE INDEX transaction_id_to_pdu_dest ON transaction_id_to_pdu(destination); + +-- To track destination health +CREATE TABLE IF NOT EXISTS destinations( + destination TEXT PRIMARY KEY, + retry_last_ts BIGINT, + retry_interval INTEGER +); diff --git a/synapse/storage/data_stores/main/schema/full_schemas/16/users.sql b/synapse/storage/data_stores/main/schema/full_schemas/16/users.sql new file mode 100644 index 0000000000..f013aa8b18 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/full_schemas/16/users.sql @@ -0,0 +1,42 @@ +/* Copyright 2014-2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ +CREATE TABLE IF NOT EXISTS users( + name TEXT, + password_hash TEXT, + creation_ts BIGINT, + admin SMALLINT DEFAULT 0 NOT NULL, + UNIQUE(name) +); + +CREATE TABLE IF NOT EXISTS access_tokens( + id BIGINT PRIMARY KEY, + user_id TEXT NOT NULL, + device_id TEXT, + token TEXT NOT NULL, + last_used BIGINT, + UNIQUE(token) +); + +CREATE TABLE IF NOT EXISTS user_ips ( + user_id TEXT NOT NULL, + access_token TEXT NOT NULL, + device_id TEXT, + ip TEXT NOT NULL, + user_agent TEXT NOT NULL, + last_seen BIGINT NOT NULL +); + +CREATE INDEX user_ips_user ON user_ips(user_id); +CREATE INDEX user_ips_user_ip ON user_ips(user_id, access_token, ip); diff --git a/synapse/storage/data_stores/main/schema/full_schemas/54/full.sql.postgres b/synapse/storage/data_stores/main/schema/full_schemas/54/full.sql.postgres new file mode 100644 index 0000000000..4ad2929f32 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/full_schemas/54/full.sql.postgres @@ -0,0 +1,2035 @@ + + + + + +CREATE TABLE access_tokens ( + id bigint NOT NULL, + user_id text NOT NULL, + device_id text, + token text NOT NULL, + last_used bigint +); + + + +CREATE TABLE account_data ( + user_id text NOT NULL, + account_data_type text NOT NULL, + stream_id bigint NOT NULL, + content text NOT NULL +); + + + +CREATE TABLE account_data_max_stream_id ( + lock character(1) DEFAULT 'X'::bpchar NOT NULL, + stream_id bigint NOT NULL, + CONSTRAINT private_user_data_max_stream_id_lock_check CHECK ((lock = 'X'::bpchar)) +); + + + +CREATE TABLE account_validity ( + user_id text NOT NULL, + expiration_ts_ms bigint NOT NULL, + email_sent boolean NOT NULL, + renewal_token text +); + + + +CREATE TABLE application_services_state ( + as_id text NOT NULL, + state character varying(5), + last_txn integer +); + + + +CREATE TABLE application_services_txns ( + as_id text NOT NULL, + txn_id integer NOT NULL, + event_ids text NOT NULL +); + + + +CREATE TABLE appservice_room_list ( + appservice_id text NOT NULL, + network_id text NOT NULL, + room_id text NOT NULL +); + + + +CREATE TABLE appservice_stream_position ( + lock character(1) DEFAULT 'X'::bpchar NOT NULL, + stream_ordering bigint, + CONSTRAINT appservice_stream_position_lock_check CHECK ((lock = 'X'::bpchar)) +); + + +CREATE TABLE blocked_rooms ( + room_id text NOT NULL, + user_id text NOT NULL +); + + + +CREATE TABLE cache_invalidation_stream ( + stream_id bigint, + cache_func text, + keys text[], + invalidation_ts bigint +); + + + +CREATE TABLE current_state_delta_stream ( + stream_id bigint NOT NULL, + room_id text NOT NULL, + type text NOT NULL, + state_key text NOT NULL, + event_id text, + prev_event_id text +); + + + +CREATE TABLE current_state_events ( + event_id text NOT NULL, + room_id text NOT NULL, + type text NOT NULL, + state_key text NOT NULL +); + + + +CREATE TABLE deleted_pushers ( + stream_id bigint NOT NULL, + app_id text NOT NULL, + pushkey text NOT NULL, + user_id text NOT NULL +); + + + +CREATE TABLE destinations ( + destination text NOT NULL, + retry_last_ts bigint, + retry_interval integer +); + + + +CREATE TABLE device_federation_inbox ( + origin text NOT NULL, + message_id text NOT NULL, + received_ts bigint NOT NULL +); + + + +CREATE TABLE device_federation_outbox ( + destination text NOT NULL, + stream_id bigint NOT NULL, + queued_ts bigint NOT NULL, + messages_json text NOT NULL +); + + + +CREATE TABLE device_inbox ( + user_id text NOT NULL, + device_id text NOT NULL, + stream_id bigint NOT NULL, + message_json text NOT NULL +); + + + +CREATE TABLE device_lists_outbound_last_success ( + destination text NOT NULL, + user_id text NOT NULL, + stream_id bigint NOT NULL +); + + + +CREATE TABLE device_lists_outbound_pokes ( + destination text NOT NULL, + stream_id bigint NOT NULL, + user_id text NOT NULL, + device_id text NOT NULL, + sent boolean NOT NULL, + ts bigint NOT NULL +); + + + +CREATE TABLE device_lists_remote_cache ( + user_id text NOT NULL, + device_id text NOT NULL, + content text NOT NULL +); + + + +CREATE TABLE device_lists_remote_extremeties ( + user_id text NOT NULL, + stream_id text NOT NULL +); + + + +CREATE TABLE device_lists_stream ( + stream_id bigint NOT NULL, + user_id text NOT NULL, + device_id text NOT NULL +); + + + +CREATE TABLE device_max_stream_id ( + stream_id bigint NOT NULL +); + + + +CREATE TABLE devices ( + user_id text NOT NULL, + device_id text NOT NULL, + display_name text +); + + + +CREATE TABLE e2e_device_keys_json ( + user_id text NOT NULL, + device_id text NOT NULL, + ts_added_ms bigint NOT NULL, + key_json text NOT NULL +); + + + +CREATE TABLE e2e_one_time_keys_json ( + user_id text NOT NULL, + device_id text NOT NULL, + algorithm text NOT NULL, + key_id text NOT NULL, + ts_added_ms bigint NOT NULL, + key_json text NOT NULL +); + + + +CREATE TABLE e2e_room_keys ( + user_id text NOT NULL, + room_id text NOT NULL, + session_id text NOT NULL, + version bigint NOT NULL, + first_message_index integer, + forwarded_count integer, + is_verified boolean, + session_data text NOT NULL +); + + + +CREATE TABLE e2e_room_keys_versions ( + user_id text NOT NULL, + version bigint NOT NULL, + algorithm text NOT NULL, + auth_data text NOT NULL, + deleted smallint DEFAULT 0 NOT NULL +); + + + +CREATE TABLE erased_users ( + user_id text NOT NULL +); + + + +CREATE TABLE event_auth ( + event_id text NOT NULL, + auth_id text NOT NULL, + room_id text NOT NULL +); + + + +CREATE TABLE event_backward_extremities ( + event_id text NOT NULL, + room_id text NOT NULL +); + + + +CREATE TABLE event_edges ( + event_id text NOT NULL, + prev_event_id text NOT NULL, + room_id text NOT NULL, + is_state boolean NOT NULL +); + + + +CREATE TABLE event_forward_extremities ( + event_id text NOT NULL, + room_id text NOT NULL +); + + + +CREATE TABLE event_json ( + event_id text NOT NULL, + room_id text NOT NULL, + internal_metadata text NOT NULL, + json text NOT NULL, + format_version integer +); + + + +CREATE TABLE event_push_actions ( + room_id text NOT NULL, + event_id text NOT NULL, + user_id text NOT NULL, + profile_tag character varying(32), + actions text NOT NULL, + topological_ordering bigint, + stream_ordering bigint, + notif smallint, + highlight smallint +); + + + +CREATE TABLE event_push_actions_staging ( + event_id text NOT NULL, + user_id text NOT NULL, + actions text NOT NULL, + notif smallint NOT NULL, + highlight smallint NOT NULL +); + + + +CREATE TABLE event_push_summary ( + user_id text NOT NULL, + room_id text NOT NULL, + notif_count bigint NOT NULL, + stream_ordering bigint NOT NULL +); + + + +CREATE TABLE event_push_summary_stream_ordering ( + lock character(1) DEFAULT 'X'::bpchar NOT NULL, + stream_ordering bigint NOT NULL, + CONSTRAINT event_push_summary_stream_ordering_lock_check CHECK ((lock = 'X'::bpchar)) +); + + + +CREATE TABLE event_reference_hashes ( + event_id text, + algorithm text, + hash bytea +); + + + +CREATE TABLE event_relations ( + event_id text NOT NULL, + relates_to_id text NOT NULL, + relation_type text NOT NULL, + aggregation_key text +); + + + +CREATE TABLE event_reports ( + id bigint NOT NULL, + received_ts bigint NOT NULL, + room_id text NOT NULL, + event_id text NOT NULL, + user_id text NOT NULL, + reason text, + content text +); + + + +CREATE TABLE event_search ( + event_id text, + room_id text, + sender text, + key text, + vector tsvector, + origin_server_ts bigint, + stream_ordering bigint +); + + + +CREATE TABLE event_to_state_groups ( + event_id text NOT NULL, + state_group bigint NOT NULL +); + + + +CREATE TABLE events ( + stream_ordering integer NOT NULL, + topological_ordering bigint NOT NULL, + event_id text NOT NULL, + type text NOT NULL, + room_id text NOT NULL, + content text, + unrecognized_keys text, + processed boolean NOT NULL, + outlier boolean NOT NULL, + depth bigint DEFAULT 0 NOT NULL, + origin_server_ts bigint, + received_ts bigint, + sender text, + contains_url boolean +); + + + +CREATE TABLE ex_outlier_stream ( + event_stream_ordering bigint NOT NULL, + event_id text NOT NULL, + state_group bigint NOT NULL +); + + + +CREATE TABLE federation_stream_position ( + type text NOT NULL, + stream_id integer NOT NULL +); + + + +CREATE TABLE group_attestations_remote ( + group_id text NOT NULL, + user_id text NOT NULL, + valid_until_ms bigint NOT NULL, + attestation_json text NOT NULL +); + + + +CREATE TABLE group_attestations_renewals ( + group_id text NOT NULL, + user_id text NOT NULL, + valid_until_ms bigint NOT NULL +); + + + +CREATE TABLE group_invites ( + group_id text NOT NULL, + user_id text NOT NULL +); + + + +CREATE TABLE group_roles ( + group_id text NOT NULL, + role_id text NOT NULL, + profile text NOT NULL, + is_public boolean NOT NULL +); + + + +CREATE TABLE group_room_categories ( + group_id text NOT NULL, + category_id text NOT NULL, + profile text NOT NULL, + is_public boolean NOT NULL +); + + + +CREATE TABLE group_rooms ( + group_id text NOT NULL, + room_id text NOT NULL, + is_public boolean NOT NULL +); + + + +CREATE TABLE group_summary_roles ( + group_id text NOT NULL, + role_id text NOT NULL, + role_order bigint NOT NULL, + CONSTRAINT group_summary_roles_role_order_check CHECK ((role_order > 0)) +); + + + +CREATE TABLE group_summary_room_categories ( + group_id text NOT NULL, + category_id text NOT NULL, + cat_order bigint NOT NULL, + CONSTRAINT group_summary_room_categories_cat_order_check CHECK ((cat_order > 0)) +); + + + +CREATE TABLE group_summary_rooms ( + group_id text NOT NULL, + room_id text NOT NULL, + category_id text NOT NULL, + room_order bigint NOT NULL, + is_public boolean NOT NULL, + CONSTRAINT group_summary_rooms_room_order_check CHECK ((room_order > 0)) +); + + + +CREATE TABLE group_summary_users ( + group_id text NOT NULL, + user_id text NOT NULL, + role_id text NOT NULL, + user_order bigint NOT NULL, + is_public boolean NOT NULL +); + + + +CREATE TABLE group_users ( + group_id text NOT NULL, + user_id text NOT NULL, + is_admin boolean NOT NULL, + is_public boolean NOT NULL +); + + + +CREATE TABLE groups ( + group_id text NOT NULL, + name text, + avatar_url text, + short_description text, + long_description text, + is_public boolean NOT NULL, + join_policy text DEFAULT 'invite'::text NOT NULL +); + + + +CREATE TABLE guest_access ( + event_id text NOT NULL, + room_id text NOT NULL, + guest_access text NOT NULL +); + + + +CREATE TABLE history_visibility ( + event_id text NOT NULL, + room_id text NOT NULL, + history_visibility text NOT NULL +); + + + +CREATE TABLE local_group_membership ( + group_id text NOT NULL, + user_id text NOT NULL, + is_admin boolean NOT NULL, + membership text NOT NULL, + is_publicised boolean NOT NULL, + content text NOT NULL +); + + + +CREATE TABLE local_group_updates ( + stream_id bigint NOT NULL, + group_id text NOT NULL, + user_id text NOT NULL, + type text NOT NULL, + content text NOT NULL +); + + + +CREATE TABLE local_invites ( + stream_id bigint NOT NULL, + inviter text NOT NULL, + invitee text NOT NULL, + event_id text NOT NULL, + room_id text NOT NULL, + locally_rejected text, + replaced_by text +); + + + +CREATE TABLE local_media_repository ( + media_id text, + media_type text, + media_length integer, + created_ts bigint, + upload_name text, + user_id text, + quarantined_by text, + url_cache text, + last_access_ts bigint +); + + + +CREATE TABLE local_media_repository_thumbnails ( + media_id text, + thumbnail_width integer, + thumbnail_height integer, + thumbnail_type text, + thumbnail_method text, + thumbnail_length integer +); + + + +CREATE TABLE local_media_repository_url_cache ( + url text, + response_code integer, + etag text, + expires_ts bigint, + og text, + media_id text, + download_ts bigint +); + + + +CREATE TABLE monthly_active_users ( + user_id text NOT NULL, + "timestamp" bigint NOT NULL +); + + + +CREATE TABLE open_id_tokens ( + token text NOT NULL, + ts_valid_until_ms bigint NOT NULL, + user_id text NOT NULL +); + + + +CREATE TABLE presence ( + user_id text NOT NULL, + state character varying(20), + status_msg text, + mtime bigint +); + + + +CREATE TABLE presence_allow_inbound ( + observed_user_id text NOT NULL, + observer_user_id text NOT NULL +); + + + +CREATE TABLE presence_stream ( + stream_id bigint, + user_id text, + state text, + last_active_ts bigint, + last_federation_update_ts bigint, + last_user_sync_ts bigint, + status_msg text, + currently_active boolean +); + + + +CREATE TABLE profiles ( + user_id text NOT NULL, + displayname text, + avatar_url text +); + + + +CREATE TABLE public_room_list_stream ( + stream_id bigint NOT NULL, + room_id text NOT NULL, + visibility boolean NOT NULL, + appservice_id text, + network_id text +); + + + +CREATE TABLE push_rules ( + id bigint NOT NULL, + user_name text NOT NULL, + rule_id text NOT NULL, + priority_class smallint NOT NULL, + priority integer DEFAULT 0 NOT NULL, + conditions text NOT NULL, + actions text NOT NULL +); + + + +CREATE TABLE push_rules_enable ( + id bigint NOT NULL, + user_name text NOT NULL, + rule_id text NOT NULL, + enabled smallint +); + + + +CREATE TABLE push_rules_stream ( + stream_id bigint NOT NULL, + event_stream_ordering bigint NOT NULL, + user_id text NOT NULL, + rule_id text NOT NULL, + op text NOT NULL, + priority_class smallint, + priority integer, + conditions text, + actions text +); + + + +CREATE TABLE pusher_throttle ( + pusher bigint NOT NULL, + room_id text NOT NULL, + last_sent_ts bigint, + throttle_ms bigint +); + + + +CREATE TABLE pushers ( + id bigint NOT NULL, + user_name text NOT NULL, + access_token bigint, + profile_tag text NOT NULL, + kind text NOT NULL, + app_id text NOT NULL, + app_display_name text NOT NULL, + device_display_name text NOT NULL, + pushkey text NOT NULL, + ts bigint NOT NULL, + lang text, + data text, + last_stream_ordering integer, + last_success bigint, + failing_since bigint +); + + + +CREATE TABLE ratelimit_override ( + user_id text NOT NULL, + messages_per_second bigint, + burst_count bigint +); + + + +CREATE TABLE receipts_graph ( + room_id text NOT NULL, + receipt_type text NOT NULL, + user_id text NOT NULL, + event_ids text NOT NULL, + data text NOT NULL +); + + + +CREATE TABLE receipts_linearized ( + stream_id bigint NOT NULL, + room_id text NOT NULL, + receipt_type text NOT NULL, + user_id text NOT NULL, + event_id text NOT NULL, + data text NOT NULL +); + + + +CREATE TABLE received_transactions ( + transaction_id text, + origin text, + ts bigint, + response_code integer, + response_json bytea, + has_been_referenced smallint DEFAULT 0 +); + + + +CREATE TABLE redactions ( + event_id text NOT NULL, + redacts text NOT NULL +); + + + +CREATE TABLE rejections ( + event_id text NOT NULL, + reason text NOT NULL, + last_check text NOT NULL +); + + + +CREATE TABLE remote_media_cache ( + media_origin text, + media_id text, + media_type text, + created_ts bigint, + upload_name text, + media_length integer, + filesystem_id text, + last_access_ts bigint, + quarantined_by text +); + + + +CREATE TABLE remote_media_cache_thumbnails ( + media_origin text, + media_id text, + thumbnail_width integer, + thumbnail_height integer, + thumbnail_method text, + thumbnail_type text, + thumbnail_length integer, + filesystem_id text +); + + + +CREATE TABLE remote_profile_cache ( + user_id text NOT NULL, + displayname text, + avatar_url text, + last_check bigint NOT NULL +); + + + +CREATE TABLE room_account_data ( + user_id text NOT NULL, + room_id text NOT NULL, + account_data_type text NOT NULL, + stream_id bigint NOT NULL, + content text NOT NULL +); + + + +CREATE TABLE room_alias_servers ( + room_alias text NOT NULL, + server text NOT NULL +); + + + +CREATE TABLE room_aliases ( + room_alias text NOT NULL, + room_id text NOT NULL, + creator text +); + + + +CREATE TABLE room_depth ( + room_id text NOT NULL, + min_depth integer NOT NULL +); + + + +CREATE TABLE room_memberships ( + event_id text NOT NULL, + user_id text NOT NULL, + sender text NOT NULL, + room_id text NOT NULL, + membership text NOT NULL, + forgotten integer DEFAULT 0, + display_name text, + avatar_url text +); + + + +CREATE TABLE room_names ( + event_id text NOT NULL, + room_id text NOT NULL, + name text NOT NULL +); + + + +CREATE TABLE room_state ( + room_id text NOT NULL, + join_rules text, + history_visibility text, + encryption text, + name text, + topic text, + avatar text, + canonical_alias text +); + + + +CREATE TABLE room_stats ( + room_id text NOT NULL, + ts bigint NOT NULL, + bucket_size integer NOT NULL, + current_state_events integer NOT NULL, + joined_members integer NOT NULL, + invited_members integer NOT NULL, + left_members integer NOT NULL, + banned_members integer NOT NULL, + state_events integer NOT NULL +); + + + +CREATE TABLE room_stats_earliest_token ( + room_id text NOT NULL, + token bigint NOT NULL +); + + + +CREATE TABLE room_tags ( + user_id text NOT NULL, + room_id text NOT NULL, + tag text NOT NULL, + content text NOT NULL +); + + + +CREATE TABLE room_tags_revisions ( + user_id text NOT NULL, + room_id text NOT NULL, + stream_id bigint NOT NULL +); + + + +CREATE TABLE rooms ( + room_id text NOT NULL, + is_public boolean, + creator text +); + + + +CREATE TABLE server_keys_json ( + server_name text NOT NULL, + key_id text NOT NULL, + from_server text NOT NULL, + ts_added_ms bigint NOT NULL, + ts_valid_until_ms bigint NOT NULL, + key_json bytea NOT NULL +); + + + +CREATE TABLE server_signature_keys ( + server_name text, + key_id text, + from_server text, + ts_added_ms bigint, + verify_key bytea, + ts_valid_until_ms bigint +); + + + +CREATE TABLE state_events ( + event_id text NOT NULL, + room_id text NOT NULL, + type text NOT NULL, + state_key text NOT NULL, + prev_state text +); + + + +CREATE TABLE state_group_edges ( + state_group bigint NOT NULL, + prev_state_group bigint NOT NULL +); + + + +CREATE SEQUENCE state_group_id_seq + START WITH 1 + INCREMENT BY 1 + NO MINVALUE + NO MAXVALUE + CACHE 1; + + + +CREATE TABLE state_groups ( + id bigint NOT NULL, + room_id text NOT NULL, + event_id text NOT NULL +); + + + +CREATE TABLE state_groups_state ( + state_group bigint NOT NULL, + room_id text NOT NULL, + type text NOT NULL, + state_key text NOT NULL, + event_id text NOT NULL +); + + + +CREATE TABLE stats_stream_pos ( + lock character(1) DEFAULT 'X'::bpchar NOT NULL, + stream_id bigint, + CONSTRAINT stats_stream_pos_lock_check CHECK ((lock = 'X'::bpchar)) +); + + + +CREATE TABLE stream_ordering_to_exterm ( + stream_ordering bigint NOT NULL, + room_id text NOT NULL, + event_id text NOT NULL +); + + + +CREATE TABLE threepid_guest_access_tokens ( + medium text, + address text, + guest_access_token text, + first_inviter text +); + + + +CREATE TABLE topics ( + event_id text NOT NULL, + room_id text NOT NULL, + topic text NOT NULL +); + + + +CREATE TABLE user_daily_visits ( + user_id text NOT NULL, + device_id text, + "timestamp" bigint NOT NULL +); + + + +CREATE TABLE user_directory ( + user_id text NOT NULL, + room_id text, + display_name text, + avatar_url text +); + + + +CREATE TABLE user_directory_search ( + user_id text NOT NULL, + vector tsvector +); + + + +CREATE TABLE user_directory_stream_pos ( + lock character(1) DEFAULT 'X'::bpchar NOT NULL, + stream_id bigint, + CONSTRAINT user_directory_stream_pos_lock_check CHECK ((lock = 'X'::bpchar)) +); + + + +CREATE TABLE user_filters ( + user_id text, + filter_id bigint, + filter_json bytea +); + + + +CREATE TABLE user_ips ( + user_id text NOT NULL, + access_token text NOT NULL, + device_id text, + ip text NOT NULL, + user_agent text NOT NULL, + last_seen bigint NOT NULL +); + + + +CREATE TABLE user_stats ( + user_id text NOT NULL, + ts bigint NOT NULL, + bucket_size integer NOT NULL, + public_rooms integer NOT NULL, + private_rooms integer NOT NULL +); + + + +CREATE TABLE user_threepid_id_server ( + user_id text NOT NULL, + medium text NOT NULL, + address text NOT NULL, + id_server text NOT NULL +); + + + +CREATE TABLE user_threepids ( + user_id text NOT NULL, + medium text NOT NULL, + address text NOT NULL, + validated_at bigint NOT NULL, + added_at bigint NOT NULL +); + + + +CREATE TABLE users ( + name text, + password_hash text, + creation_ts bigint, + admin smallint DEFAULT 0 NOT NULL, + upgrade_ts bigint, + is_guest smallint DEFAULT 0 NOT NULL, + appservice_id text, + consent_version text, + consent_server_notice_sent text, + user_type text +); + + + +CREATE TABLE users_in_public_rooms ( + user_id text NOT NULL, + room_id text NOT NULL +); + + + +CREATE TABLE users_pending_deactivation ( + user_id text NOT NULL +); + + + +CREATE TABLE users_who_share_private_rooms ( + user_id text NOT NULL, + other_user_id text NOT NULL, + room_id text NOT NULL +); + + + +ALTER TABLE ONLY access_tokens + ADD CONSTRAINT access_tokens_pkey PRIMARY KEY (id); + + + +ALTER TABLE ONLY access_tokens + ADD CONSTRAINT access_tokens_token_key UNIQUE (token); + + + +ALTER TABLE ONLY account_data + ADD CONSTRAINT account_data_uniqueness UNIQUE (user_id, account_data_type); + + + +ALTER TABLE ONLY account_validity + ADD CONSTRAINT account_validity_pkey PRIMARY KEY (user_id); + + + +ALTER TABLE ONLY application_services_state + ADD CONSTRAINT application_services_state_pkey PRIMARY KEY (as_id); + + + +ALTER TABLE ONLY application_services_txns + ADD CONSTRAINT application_services_txns_as_id_txn_id_key UNIQUE (as_id, txn_id); + + + +ALTER TABLE ONLY appservice_stream_position + ADD CONSTRAINT appservice_stream_position_lock_key UNIQUE (lock); + + + +ALTER TABLE ONLY current_state_events + ADD CONSTRAINT current_state_events_event_id_key UNIQUE (event_id); + + + +ALTER TABLE ONLY current_state_events + ADD CONSTRAINT current_state_events_room_id_type_state_key_key UNIQUE (room_id, type, state_key); + + + +ALTER TABLE ONLY destinations + ADD CONSTRAINT destinations_pkey PRIMARY KEY (destination); + + + +ALTER TABLE ONLY devices + ADD CONSTRAINT device_uniqueness UNIQUE (user_id, device_id); + + + +ALTER TABLE ONLY e2e_device_keys_json + ADD CONSTRAINT e2e_device_keys_json_uniqueness UNIQUE (user_id, device_id); + + + +ALTER TABLE ONLY e2e_one_time_keys_json + ADD CONSTRAINT e2e_one_time_keys_json_uniqueness UNIQUE (user_id, device_id, algorithm, key_id); + + + +ALTER TABLE ONLY event_backward_extremities + ADD CONSTRAINT event_backward_extremities_event_id_room_id_key UNIQUE (event_id, room_id); + + + +ALTER TABLE ONLY event_edges + ADD CONSTRAINT event_edges_event_id_prev_event_id_room_id_is_state_key UNIQUE (event_id, prev_event_id, room_id, is_state); + + + +ALTER TABLE ONLY event_forward_extremities + ADD CONSTRAINT event_forward_extremities_event_id_room_id_key UNIQUE (event_id, room_id); + + + +ALTER TABLE ONLY event_push_actions + ADD CONSTRAINT event_id_user_id_profile_tag_uniqueness UNIQUE (room_id, event_id, user_id, profile_tag); + + + +ALTER TABLE ONLY event_json + ADD CONSTRAINT event_json_event_id_key UNIQUE (event_id); + + + +ALTER TABLE ONLY event_push_summary_stream_ordering + ADD CONSTRAINT event_push_summary_stream_ordering_lock_key UNIQUE (lock); + + + +ALTER TABLE ONLY event_reference_hashes + ADD CONSTRAINT event_reference_hashes_event_id_algorithm_key UNIQUE (event_id, algorithm); + + + +ALTER TABLE ONLY event_reports + ADD CONSTRAINT event_reports_pkey PRIMARY KEY (id); + + + +ALTER TABLE ONLY event_to_state_groups + ADD CONSTRAINT event_to_state_groups_event_id_key UNIQUE (event_id); + + + +ALTER TABLE ONLY events + ADD CONSTRAINT events_event_id_key UNIQUE (event_id); + + + +ALTER TABLE ONLY events + ADD CONSTRAINT events_pkey PRIMARY KEY (stream_ordering); + + + +ALTER TABLE ONLY ex_outlier_stream + ADD CONSTRAINT ex_outlier_stream_pkey PRIMARY KEY (event_stream_ordering); + + + +ALTER TABLE ONLY group_roles + ADD CONSTRAINT group_roles_group_id_role_id_key UNIQUE (group_id, role_id); + + + +ALTER TABLE ONLY group_room_categories + ADD CONSTRAINT group_room_categories_group_id_category_id_key UNIQUE (group_id, category_id); + + + +ALTER TABLE ONLY group_summary_roles + ADD CONSTRAINT group_summary_roles_group_id_role_id_role_order_key UNIQUE (group_id, role_id, role_order); + + + +ALTER TABLE ONLY group_summary_room_categories + ADD CONSTRAINT group_summary_room_categories_group_id_category_id_cat_orde_key UNIQUE (group_id, category_id, cat_order); + + + +ALTER TABLE ONLY group_summary_rooms + ADD CONSTRAINT group_summary_rooms_group_id_category_id_room_id_room_order_key UNIQUE (group_id, category_id, room_id, room_order); + + + +ALTER TABLE ONLY guest_access + ADD CONSTRAINT guest_access_event_id_key UNIQUE (event_id); + + + +ALTER TABLE ONLY history_visibility + ADD CONSTRAINT history_visibility_event_id_key UNIQUE (event_id); + + + +ALTER TABLE ONLY local_media_repository + ADD CONSTRAINT local_media_repository_media_id_key UNIQUE (media_id); + + + +ALTER TABLE ONLY local_media_repository_thumbnails + ADD CONSTRAINT local_media_repository_thumbn_media_id_thumbnail_width_thum_key UNIQUE (media_id, thumbnail_width, thumbnail_height, thumbnail_type); + + + +ALTER TABLE ONLY user_threepids + ADD CONSTRAINT medium_address UNIQUE (medium, address); + + + +ALTER TABLE ONLY open_id_tokens + ADD CONSTRAINT open_id_tokens_pkey PRIMARY KEY (token); + + + +ALTER TABLE ONLY presence_allow_inbound + ADD CONSTRAINT presence_allow_inbound_observed_user_id_observer_user_id_key UNIQUE (observed_user_id, observer_user_id); + + + +ALTER TABLE ONLY presence + ADD CONSTRAINT presence_user_id_key UNIQUE (user_id); + + + +ALTER TABLE ONLY account_data_max_stream_id + ADD CONSTRAINT private_user_data_max_stream_id_lock_key UNIQUE (lock); + + + +ALTER TABLE ONLY profiles + ADD CONSTRAINT profiles_user_id_key UNIQUE (user_id); + + + +ALTER TABLE ONLY push_rules_enable + ADD CONSTRAINT push_rules_enable_pkey PRIMARY KEY (id); + + + +ALTER TABLE ONLY push_rules_enable + ADD CONSTRAINT push_rules_enable_user_name_rule_id_key UNIQUE (user_name, rule_id); + + + +ALTER TABLE ONLY push_rules + ADD CONSTRAINT push_rules_pkey PRIMARY KEY (id); + + + +ALTER TABLE ONLY push_rules + ADD CONSTRAINT push_rules_user_name_rule_id_key UNIQUE (user_name, rule_id); + + + +ALTER TABLE ONLY pusher_throttle + ADD CONSTRAINT pusher_throttle_pkey PRIMARY KEY (pusher, room_id); + + + +ALTER TABLE ONLY pushers + ADD CONSTRAINT pushers2_app_id_pushkey_user_name_key UNIQUE (app_id, pushkey, user_name); + + + +ALTER TABLE ONLY pushers + ADD CONSTRAINT pushers2_pkey PRIMARY KEY (id); + + + +ALTER TABLE ONLY receipts_graph + ADD CONSTRAINT receipts_graph_uniqueness UNIQUE (room_id, receipt_type, user_id); + + + +ALTER TABLE ONLY receipts_linearized + ADD CONSTRAINT receipts_linearized_uniqueness UNIQUE (room_id, receipt_type, user_id); + + + +ALTER TABLE ONLY received_transactions + ADD CONSTRAINT received_transactions_transaction_id_origin_key UNIQUE (transaction_id, origin); + + + +ALTER TABLE ONLY redactions + ADD CONSTRAINT redactions_event_id_key UNIQUE (event_id); + + + +ALTER TABLE ONLY rejections + ADD CONSTRAINT rejections_event_id_key UNIQUE (event_id); + + + +ALTER TABLE ONLY remote_media_cache + ADD CONSTRAINT remote_media_cache_media_origin_media_id_key UNIQUE (media_origin, media_id); + + + +ALTER TABLE ONLY remote_media_cache_thumbnails + ADD CONSTRAINT remote_media_cache_thumbnails_media_origin_media_id_thumbna_key UNIQUE (media_origin, media_id, thumbnail_width, thumbnail_height, thumbnail_type); + + + +ALTER TABLE ONLY room_account_data + ADD CONSTRAINT room_account_data_uniqueness UNIQUE (user_id, room_id, account_data_type); + + + +ALTER TABLE ONLY room_aliases + ADD CONSTRAINT room_aliases_room_alias_key UNIQUE (room_alias); + + + +ALTER TABLE ONLY room_depth + ADD CONSTRAINT room_depth_room_id_key UNIQUE (room_id); + + + +ALTER TABLE ONLY room_memberships + ADD CONSTRAINT room_memberships_event_id_key UNIQUE (event_id); + + + +ALTER TABLE ONLY room_names + ADD CONSTRAINT room_names_event_id_key UNIQUE (event_id); + + + +ALTER TABLE ONLY room_tags_revisions + ADD CONSTRAINT room_tag_revisions_uniqueness UNIQUE (user_id, room_id); + + + +ALTER TABLE ONLY room_tags + ADD CONSTRAINT room_tag_uniqueness UNIQUE (user_id, room_id, tag); + + + +ALTER TABLE ONLY rooms + ADD CONSTRAINT rooms_pkey PRIMARY KEY (room_id); + + + +ALTER TABLE ONLY server_keys_json + ADD CONSTRAINT server_keys_json_uniqueness UNIQUE (server_name, key_id, from_server); + + + +ALTER TABLE ONLY server_signature_keys + ADD CONSTRAINT server_signature_keys_server_name_key_id_key UNIQUE (server_name, key_id); + + + +ALTER TABLE ONLY state_events + ADD CONSTRAINT state_events_event_id_key UNIQUE (event_id); + + + +ALTER TABLE ONLY state_groups + ADD CONSTRAINT state_groups_pkey PRIMARY KEY (id); + + + +ALTER TABLE ONLY stats_stream_pos + ADD CONSTRAINT stats_stream_pos_lock_key UNIQUE (lock); + + + +ALTER TABLE ONLY topics + ADD CONSTRAINT topics_event_id_key UNIQUE (event_id); + + + +ALTER TABLE ONLY user_directory_stream_pos + ADD CONSTRAINT user_directory_stream_pos_lock_key UNIQUE (lock); + + + +ALTER TABLE ONLY users + ADD CONSTRAINT users_name_key UNIQUE (name); + + + +CREATE INDEX access_tokens_device_id ON access_tokens USING btree (user_id, device_id); + + + +CREATE INDEX account_data_stream_id ON account_data USING btree (user_id, stream_id); + + + +CREATE INDEX application_services_txns_id ON application_services_txns USING btree (as_id); + + + +CREATE UNIQUE INDEX appservice_room_list_idx ON appservice_room_list USING btree (appservice_id, network_id, room_id); + + + +CREATE UNIQUE INDEX blocked_rooms_idx ON blocked_rooms USING btree (room_id); + + + +CREATE INDEX cache_invalidation_stream_id ON cache_invalidation_stream USING btree (stream_id); + + + +CREATE INDEX current_state_delta_stream_idx ON current_state_delta_stream USING btree (stream_id); + + + +CREATE INDEX current_state_events_member_index ON current_state_events USING btree (state_key) WHERE (type = 'm.room.member'::text); + + + +CREATE INDEX deleted_pushers_stream_id ON deleted_pushers USING btree (stream_id); + + + +CREATE INDEX device_federation_inbox_sender_id ON device_federation_inbox USING btree (origin, message_id); + + + +CREATE INDEX device_federation_outbox_destination_id ON device_federation_outbox USING btree (destination, stream_id); + + + +CREATE INDEX device_federation_outbox_id ON device_federation_outbox USING btree (stream_id); + + + +CREATE INDEX device_inbox_stream_id_user_id ON device_inbox USING btree (stream_id, user_id); + + + +CREATE INDEX device_inbox_user_stream_id ON device_inbox USING btree (user_id, device_id, stream_id); + + + +CREATE INDEX device_lists_outbound_last_success_idx ON device_lists_outbound_last_success USING btree (destination, user_id, stream_id); + + + +CREATE INDEX device_lists_outbound_pokes_id ON device_lists_outbound_pokes USING btree (destination, stream_id); + + + +CREATE INDEX device_lists_outbound_pokes_stream ON device_lists_outbound_pokes USING btree (stream_id); + + + +CREATE INDEX device_lists_outbound_pokes_user ON device_lists_outbound_pokes USING btree (destination, user_id); + + + +CREATE UNIQUE INDEX device_lists_remote_cache_unique_id ON device_lists_remote_cache USING btree (user_id, device_id); + + + +CREATE UNIQUE INDEX device_lists_remote_extremeties_unique_idx ON device_lists_remote_extremeties USING btree (user_id); + + + +CREATE INDEX device_lists_stream_id ON device_lists_stream USING btree (stream_id, user_id); + + + +CREATE INDEX device_lists_stream_user_id ON device_lists_stream USING btree (user_id, device_id); + + + +CREATE UNIQUE INDEX e2e_room_keys_idx ON e2e_room_keys USING btree (user_id, room_id, session_id); + + + +CREATE UNIQUE INDEX e2e_room_keys_versions_idx ON e2e_room_keys_versions USING btree (user_id, version); + + + +CREATE UNIQUE INDEX erased_users_user ON erased_users USING btree (user_id); + + + +CREATE INDEX ev_b_extrem_id ON event_backward_extremities USING btree (event_id); + + + +CREATE INDEX ev_b_extrem_room ON event_backward_extremities USING btree (room_id); + + + +CREATE INDEX ev_edges_id ON event_edges USING btree (event_id); + + + +CREATE INDEX ev_edges_prev_id ON event_edges USING btree (prev_event_id); + + + +CREATE INDEX ev_extrem_id ON event_forward_extremities USING btree (event_id); + + + +CREATE INDEX ev_extrem_room ON event_forward_extremities USING btree (room_id); + + + +CREATE INDEX evauth_edges_id ON event_auth USING btree (event_id); + + + +CREATE INDEX event_contains_url_index ON events USING btree (room_id, topological_ordering, stream_ordering) WHERE ((contains_url = true) AND (outlier = false)); + + + +CREATE INDEX event_json_room_id ON event_json USING btree (room_id); + + + +CREATE INDEX event_push_actions_highlights_index ON event_push_actions USING btree (user_id, room_id, topological_ordering, stream_ordering) WHERE (highlight = 1); + + + +CREATE INDEX event_push_actions_rm_tokens ON event_push_actions USING btree (user_id, room_id, topological_ordering, stream_ordering); + + + +CREATE INDEX event_push_actions_room_id_user_id ON event_push_actions USING btree (room_id, user_id); + + + +CREATE INDEX event_push_actions_staging_id ON event_push_actions_staging USING btree (event_id); + + + +CREATE INDEX event_push_actions_stream_ordering ON event_push_actions USING btree (stream_ordering, user_id); + + + +CREATE INDEX event_push_actions_u_highlight ON event_push_actions USING btree (user_id, stream_ordering); + + + +CREATE INDEX event_push_summary_user_rm ON event_push_summary USING btree (user_id, room_id); + + + +CREATE INDEX event_reference_hashes_id ON event_reference_hashes USING btree (event_id); + + + +CREATE UNIQUE INDEX event_relations_id ON event_relations USING btree (event_id); + + + +CREATE INDEX event_relations_relates ON event_relations USING btree (relates_to_id, relation_type, aggregation_key); + + + +CREATE INDEX event_search_ev_ridx ON event_search USING btree (room_id); + + + +CREATE UNIQUE INDEX event_search_event_id_idx ON event_search USING btree (event_id); + + + +CREATE INDEX event_search_fts_idx ON event_search USING gin (vector); + + + +CREATE INDEX event_to_state_groups_sg_index ON event_to_state_groups USING btree (state_group); + + + +CREATE INDEX events_order_room ON events USING btree (room_id, topological_ordering, stream_ordering); + + + +CREATE INDEX events_room_stream ON events USING btree (room_id, stream_ordering); + + + +CREATE INDEX events_ts ON events USING btree (origin_server_ts, stream_ordering); + + + +CREATE INDEX group_attestations_remote_g_idx ON group_attestations_remote USING btree (group_id, user_id); + + + +CREATE INDEX group_attestations_remote_u_idx ON group_attestations_remote USING btree (user_id); + + + +CREATE INDEX group_attestations_remote_v_idx ON group_attestations_remote USING btree (valid_until_ms); + + + +CREATE INDEX group_attestations_renewals_g_idx ON group_attestations_renewals USING btree (group_id, user_id); + + + +CREATE INDEX group_attestations_renewals_u_idx ON group_attestations_renewals USING btree (user_id); + + + +CREATE INDEX group_attestations_renewals_v_idx ON group_attestations_renewals USING btree (valid_until_ms); + + + +CREATE UNIQUE INDEX group_invites_g_idx ON group_invites USING btree (group_id, user_id); + + + +CREATE INDEX group_invites_u_idx ON group_invites USING btree (user_id); + + + +CREATE UNIQUE INDEX group_rooms_g_idx ON group_rooms USING btree (group_id, room_id); + + + +CREATE INDEX group_rooms_r_idx ON group_rooms USING btree (room_id); + + + +CREATE UNIQUE INDEX group_summary_rooms_g_idx ON group_summary_rooms USING btree (group_id, room_id, category_id); + + + +CREATE INDEX group_summary_users_g_idx ON group_summary_users USING btree (group_id); + + + +CREATE UNIQUE INDEX group_users_g_idx ON group_users USING btree (group_id, user_id); + + + +CREATE INDEX group_users_u_idx ON group_users USING btree (user_id); + + + +CREATE UNIQUE INDEX groups_idx ON groups USING btree (group_id); + + + +CREATE INDEX local_group_membership_g_idx ON local_group_membership USING btree (group_id); + + + +CREATE INDEX local_group_membership_u_idx ON local_group_membership USING btree (user_id, group_id); + + + +CREATE INDEX local_invites_for_user_idx ON local_invites USING btree (invitee, locally_rejected, replaced_by, room_id); + + + +CREATE INDEX local_invites_id ON local_invites USING btree (stream_id); + + + +CREATE INDEX local_media_repository_thumbnails_media_id ON local_media_repository_thumbnails USING btree (media_id); + + + +CREATE INDEX local_media_repository_url_cache_by_url_download_ts ON local_media_repository_url_cache USING btree (url, download_ts); + + + +CREATE INDEX local_media_repository_url_cache_expires_idx ON local_media_repository_url_cache USING btree (expires_ts); + + + +CREATE INDEX local_media_repository_url_cache_media_idx ON local_media_repository_url_cache USING btree (media_id); + + + +CREATE INDEX local_media_repository_url_idx ON local_media_repository USING btree (created_ts) WHERE (url_cache IS NOT NULL); + + + +CREATE INDEX monthly_active_users_time_stamp ON monthly_active_users USING btree ("timestamp"); + + + +CREATE UNIQUE INDEX monthly_active_users_users ON monthly_active_users USING btree (user_id); + + + +CREATE INDEX open_id_tokens_ts_valid_until_ms ON open_id_tokens USING btree (ts_valid_until_ms); + + + +CREATE INDEX presence_stream_id ON presence_stream USING btree (stream_id, user_id); + + + +CREATE INDEX presence_stream_user_id ON presence_stream USING btree (user_id); + + + +CREATE INDEX public_room_index ON rooms USING btree (is_public); + + + +CREATE INDEX public_room_list_stream_idx ON public_room_list_stream USING btree (stream_id); + + + +CREATE INDEX public_room_list_stream_rm_idx ON public_room_list_stream USING btree (room_id, stream_id); + + + +CREATE INDEX push_rules_enable_user_name ON push_rules_enable USING btree (user_name); + + + +CREATE INDEX push_rules_stream_id ON push_rules_stream USING btree (stream_id); + + + +CREATE INDEX push_rules_stream_user_stream_id ON push_rules_stream USING btree (user_id, stream_id); + + + +CREATE INDEX push_rules_user_name ON push_rules USING btree (user_name); + + + +CREATE UNIQUE INDEX ratelimit_override_idx ON ratelimit_override USING btree (user_id); + + + +CREATE INDEX receipts_linearized_id ON receipts_linearized USING btree (stream_id); + + + +CREATE INDEX receipts_linearized_room_stream ON receipts_linearized USING btree (room_id, stream_id); + + + +CREATE INDEX receipts_linearized_user ON receipts_linearized USING btree (user_id); + + + +CREATE INDEX received_transactions_ts ON received_transactions USING btree (ts); + + + +CREATE INDEX redactions_redacts ON redactions USING btree (redacts); + + + +CREATE INDEX remote_profile_cache_time ON remote_profile_cache USING btree (last_check); + + + +CREATE UNIQUE INDEX remote_profile_cache_user_id ON remote_profile_cache USING btree (user_id); + + + +CREATE INDEX room_account_data_stream_id ON room_account_data USING btree (user_id, stream_id); + + + +CREATE INDEX room_alias_servers_alias ON room_alias_servers USING btree (room_alias); + + + +CREATE INDEX room_aliases_id ON room_aliases USING btree (room_id); + + + +CREATE INDEX room_depth_room ON room_depth USING btree (room_id); + + + +CREATE INDEX room_memberships_room_id ON room_memberships USING btree (room_id); + + + +CREATE INDEX room_memberships_user_id ON room_memberships USING btree (user_id); + + + +CREATE INDEX room_names_room_id ON room_names USING btree (room_id); + + + +CREATE UNIQUE INDEX room_state_room ON room_state USING btree (room_id); + + + +CREATE UNIQUE INDEX room_stats_earliest_token_idx ON room_stats_earliest_token USING btree (room_id); + + + +CREATE UNIQUE INDEX room_stats_room_ts ON room_stats USING btree (room_id, ts); + + + +CREATE INDEX state_group_edges_idx ON state_group_edges USING btree (state_group); + + + +CREATE INDEX state_group_edges_prev_idx ON state_group_edges USING btree (prev_state_group); + + + +CREATE INDEX state_groups_state_type_idx ON state_groups_state USING btree (state_group, type, state_key); + + + +CREATE INDEX stream_ordering_to_exterm_idx ON stream_ordering_to_exterm USING btree (stream_ordering); + + + +CREATE INDEX stream_ordering_to_exterm_rm_idx ON stream_ordering_to_exterm USING btree (room_id, stream_ordering); + + + +CREATE UNIQUE INDEX threepid_guest_access_tokens_index ON threepid_guest_access_tokens USING btree (medium, address); + + + +CREATE INDEX topics_room_id ON topics USING btree (room_id); + + + +CREATE INDEX user_daily_visits_ts_idx ON user_daily_visits USING btree ("timestamp"); + + + +CREATE INDEX user_daily_visits_uts_idx ON user_daily_visits USING btree (user_id, "timestamp"); + + + +CREATE INDEX user_directory_room_idx ON user_directory USING btree (room_id); + + + +CREATE INDEX user_directory_search_fts_idx ON user_directory_search USING gin (vector); + + + +CREATE UNIQUE INDEX user_directory_search_user_idx ON user_directory_search USING btree (user_id); + + + +CREATE UNIQUE INDEX user_directory_user_idx ON user_directory USING btree (user_id); + + + +CREATE INDEX user_filters_by_user_id_filter_id ON user_filters USING btree (user_id, filter_id); + + + +CREATE INDEX user_ips_device_id ON user_ips USING btree (user_id, device_id, last_seen); + + + +CREATE INDEX user_ips_last_seen ON user_ips USING btree (user_id, last_seen); + + + +CREATE INDEX user_ips_last_seen_only ON user_ips USING btree (last_seen); + + + +CREATE UNIQUE INDEX user_ips_user_token_ip_unique_index ON user_ips USING btree (user_id, access_token, ip); + + + +CREATE UNIQUE INDEX user_stats_user_ts ON user_stats USING btree (user_id, ts); + + + +CREATE UNIQUE INDEX user_threepid_id_server_idx ON user_threepid_id_server USING btree (user_id, medium, address, id_server); + + + +CREATE INDEX user_threepids_medium_address ON user_threepids USING btree (medium, address); + + + +CREATE INDEX user_threepids_user_id ON user_threepids USING btree (user_id); + + + +CREATE INDEX users_creation_ts ON users USING btree (creation_ts); + + + +CREATE UNIQUE INDEX users_in_public_rooms_u_idx ON users_in_public_rooms USING btree (user_id, room_id); + + + +CREATE INDEX users_who_share_private_rooms_o_idx ON users_who_share_private_rooms USING btree (other_user_id); + + + +CREATE INDEX users_who_share_private_rooms_r_idx ON users_who_share_private_rooms USING btree (room_id); + + + +CREATE UNIQUE INDEX users_who_share_private_rooms_u_idx ON users_who_share_private_rooms USING btree (user_id, other_user_id, room_id); diff --git a/synapse/storage/data_stores/main/schema/full_schemas/54/full.sql.sqlite b/synapse/storage/data_stores/main/schema/full_schemas/54/full.sql.sqlite new file mode 100644 index 0000000000..bad33291e7 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/full_schemas/54/full.sql.sqlite @@ -0,0 +1,259 @@ +CREATE TABLE application_services_state( as_id TEXT PRIMARY KEY, state VARCHAR(5), last_txn INTEGER ); +CREATE TABLE application_services_txns( as_id TEXT NOT NULL, txn_id INTEGER NOT NULL, event_ids TEXT NOT NULL, UNIQUE(as_id, txn_id) ); +CREATE INDEX application_services_txns_id ON application_services_txns ( as_id ); +CREATE TABLE presence( user_id TEXT NOT NULL, state VARCHAR(20), status_msg TEXT, mtime BIGINT, UNIQUE (user_id) ); +CREATE TABLE presence_allow_inbound( observed_user_id TEXT NOT NULL, observer_user_id TEXT NOT NULL, UNIQUE (observed_user_id, observer_user_id) ); +CREATE TABLE users( name TEXT, password_hash TEXT, creation_ts BIGINT, admin SMALLINT DEFAULT 0 NOT NULL, upgrade_ts BIGINT, is_guest SMALLINT DEFAULT 0 NOT NULL, appservice_id TEXT, consent_version TEXT, consent_server_notice_sent TEXT, user_type TEXT DEFAULT NULL, UNIQUE(name) ); +CREATE TABLE access_tokens( id BIGINT PRIMARY KEY, user_id TEXT NOT NULL, device_id TEXT, token TEXT NOT NULL, last_used BIGINT, UNIQUE(token) ); +CREATE TABLE user_ips ( user_id TEXT NOT NULL, access_token TEXT NOT NULL, device_id TEXT, ip TEXT NOT NULL, user_agent TEXT NOT NULL, last_seen BIGINT NOT NULL ); +CREATE TABLE profiles( user_id TEXT NOT NULL, displayname TEXT, avatar_url TEXT, UNIQUE(user_id) ); +CREATE TABLE received_transactions( transaction_id TEXT, origin TEXT, ts BIGINT, response_code INTEGER, response_json bytea, has_been_referenced smallint default 0, UNIQUE (transaction_id, origin) ); +CREATE TABLE destinations( destination TEXT PRIMARY KEY, retry_last_ts BIGINT, retry_interval INTEGER ); +CREATE TABLE events( stream_ordering INTEGER PRIMARY KEY, topological_ordering BIGINT NOT NULL, event_id TEXT NOT NULL, type TEXT NOT NULL, room_id TEXT NOT NULL, content TEXT, unrecognized_keys TEXT, processed BOOL NOT NULL, outlier BOOL NOT NULL, depth BIGINT DEFAULT 0 NOT NULL, origin_server_ts BIGINT, received_ts BIGINT, sender TEXT, contains_url BOOLEAN, UNIQUE (event_id) ); +CREATE INDEX events_order_room ON events ( room_id, topological_ordering, stream_ordering ); +CREATE TABLE event_json( event_id TEXT NOT NULL, room_id TEXT NOT NULL, internal_metadata TEXT NOT NULL, json TEXT NOT NULL, format_version INTEGER, UNIQUE (event_id) ); +CREATE INDEX event_json_room_id ON event_json(room_id); +CREATE TABLE state_events( event_id TEXT NOT NULL, room_id TEXT NOT NULL, type TEXT NOT NULL, state_key TEXT NOT NULL, prev_state TEXT, UNIQUE (event_id) ); +CREATE TABLE current_state_events( event_id TEXT NOT NULL, room_id TEXT NOT NULL, type TEXT NOT NULL, state_key TEXT NOT NULL, UNIQUE (event_id), UNIQUE (room_id, type, state_key) ); +CREATE TABLE room_memberships( event_id TEXT NOT NULL, user_id TEXT NOT NULL, sender TEXT NOT NULL, room_id TEXT NOT NULL, membership TEXT NOT NULL, forgotten INTEGER DEFAULT 0, display_name TEXT, avatar_url TEXT, UNIQUE (event_id) ); +CREATE INDEX room_memberships_room_id ON room_memberships (room_id); +CREATE INDEX room_memberships_user_id ON room_memberships (user_id); +CREATE TABLE topics( event_id TEXT NOT NULL, room_id TEXT NOT NULL, topic TEXT NOT NULL, UNIQUE (event_id) ); +CREATE INDEX topics_room_id ON topics(room_id); +CREATE TABLE room_names( event_id TEXT NOT NULL, room_id TEXT NOT NULL, name TEXT NOT NULL, UNIQUE (event_id) ); +CREATE INDEX room_names_room_id ON room_names(room_id); +CREATE TABLE rooms( room_id TEXT PRIMARY KEY NOT NULL, is_public BOOL, creator TEXT ); +CREATE TABLE server_signature_keys( server_name TEXT, key_id TEXT, from_server TEXT, ts_added_ms BIGINT, verify_key bytea, ts_valid_until_ms BIGINT, UNIQUE (server_name, key_id) ); +CREATE TABLE rejections( event_id TEXT NOT NULL, reason TEXT NOT NULL, last_check TEXT NOT NULL, UNIQUE (event_id) ); +CREATE TABLE push_rules ( id BIGINT PRIMARY KEY, user_name TEXT NOT NULL, rule_id TEXT NOT NULL, priority_class SMALLINT NOT NULL, priority INTEGER NOT NULL DEFAULT 0, conditions TEXT NOT NULL, actions TEXT NOT NULL, UNIQUE(user_name, rule_id) ); +CREATE INDEX push_rules_user_name on push_rules (user_name); +CREATE TABLE user_filters( user_id TEXT, filter_id BIGINT, filter_json bytea ); +CREATE INDEX user_filters_by_user_id_filter_id ON user_filters( user_id, filter_id ); +CREATE TABLE push_rules_enable ( id BIGINT PRIMARY KEY, user_name TEXT NOT NULL, rule_id TEXT NOT NULL, enabled SMALLINT, UNIQUE(user_name, rule_id) ); +CREATE INDEX push_rules_enable_user_name on push_rules_enable (user_name); +CREATE TABLE event_forward_extremities( event_id TEXT NOT NULL, room_id TEXT NOT NULL, UNIQUE (event_id, room_id) ); +CREATE INDEX ev_extrem_room ON event_forward_extremities(room_id); +CREATE INDEX ev_extrem_id ON event_forward_extremities(event_id); +CREATE TABLE event_backward_extremities( event_id TEXT NOT NULL, room_id TEXT NOT NULL, UNIQUE (event_id, room_id) ); +CREATE INDEX ev_b_extrem_room ON event_backward_extremities(room_id); +CREATE INDEX ev_b_extrem_id ON event_backward_extremities(event_id); +CREATE TABLE event_edges( event_id TEXT NOT NULL, prev_event_id TEXT NOT NULL, room_id TEXT NOT NULL, is_state BOOL NOT NULL, UNIQUE (event_id, prev_event_id, room_id, is_state) ); +CREATE INDEX ev_edges_id ON event_edges(event_id); +CREATE INDEX ev_edges_prev_id ON event_edges(prev_event_id); +CREATE TABLE room_depth( room_id TEXT NOT NULL, min_depth INTEGER NOT NULL, UNIQUE (room_id) ); +CREATE INDEX room_depth_room ON room_depth(room_id); +CREATE TABLE state_groups( id BIGINT PRIMARY KEY, room_id TEXT NOT NULL, event_id TEXT NOT NULL ); +CREATE TABLE state_groups_state( state_group BIGINT NOT NULL, room_id TEXT NOT NULL, type TEXT NOT NULL, state_key TEXT NOT NULL, event_id TEXT NOT NULL ); +CREATE TABLE event_to_state_groups( event_id TEXT NOT NULL, state_group BIGINT NOT NULL, UNIQUE (event_id) ); +CREATE TABLE local_media_repository ( media_id TEXT, media_type TEXT, media_length INTEGER, created_ts BIGINT, upload_name TEXT, user_id TEXT, quarantined_by TEXT, url_cache TEXT, last_access_ts BIGINT, UNIQUE (media_id) ); +CREATE TABLE local_media_repository_thumbnails ( media_id TEXT, thumbnail_width INTEGER, thumbnail_height INTEGER, thumbnail_type TEXT, thumbnail_method TEXT, thumbnail_length INTEGER, UNIQUE ( media_id, thumbnail_width, thumbnail_height, thumbnail_type ) ); +CREATE INDEX local_media_repository_thumbnails_media_id ON local_media_repository_thumbnails (media_id); +CREATE TABLE remote_media_cache ( media_origin TEXT, media_id TEXT, media_type TEXT, created_ts BIGINT, upload_name TEXT, media_length INTEGER, filesystem_id TEXT, last_access_ts BIGINT, quarantined_by TEXT, UNIQUE (media_origin, media_id) ); +CREATE TABLE remote_media_cache_thumbnails ( media_origin TEXT, media_id TEXT, thumbnail_width INTEGER, thumbnail_height INTEGER, thumbnail_method TEXT, thumbnail_type TEXT, thumbnail_length INTEGER, filesystem_id TEXT, UNIQUE ( media_origin, media_id, thumbnail_width, thumbnail_height, thumbnail_type ) ); +CREATE TABLE redactions ( event_id TEXT NOT NULL, redacts TEXT NOT NULL, UNIQUE (event_id) ); +CREATE INDEX redactions_redacts ON redactions (redacts); +CREATE TABLE room_aliases( room_alias TEXT NOT NULL, room_id TEXT NOT NULL, creator TEXT, UNIQUE (room_alias) ); +CREATE INDEX room_aliases_id ON room_aliases(room_id); +CREATE TABLE room_alias_servers( room_alias TEXT NOT NULL, server TEXT NOT NULL ); +CREATE INDEX room_alias_servers_alias ON room_alias_servers(room_alias); +CREATE TABLE event_reference_hashes ( event_id TEXT, algorithm TEXT, hash bytea, UNIQUE (event_id, algorithm) ); +CREATE INDEX event_reference_hashes_id ON event_reference_hashes(event_id); +CREATE TABLE IF NOT EXISTS "server_keys_json" ( server_name TEXT NOT NULL, key_id TEXT NOT NULL, from_server TEXT NOT NULL, ts_added_ms BIGINT NOT NULL, ts_valid_until_ms BIGINT NOT NULL, key_json bytea NOT NULL, CONSTRAINT server_keys_json_uniqueness UNIQUE (server_name, key_id, from_server) ); +CREATE TABLE e2e_device_keys_json ( user_id TEXT NOT NULL, device_id TEXT NOT NULL, ts_added_ms BIGINT NOT NULL, key_json TEXT NOT NULL, CONSTRAINT e2e_device_keys_json_uniqueness UNIQUE (user_id, device_id) ); +CREATE TABLE e2e_one_time_keys_json ( user_id TEXT NOT NULL, device_id TEXT NOT NULL, algorithm TEXT NOT NULL, key_id TEXT NOT NULL, ts_added_ms BIGINT NOT NULL, key_json TEXT NOT NULL, CONSTRAINT e2e_one_time_keys_json_uniqueness UNIQUE (user_id, device_id, algorithm, key_id) ); +CREATE TABLE receipts_graph( room_id TEXT NOT NULL, receipt_type TEXT NOT NULL, user_id TEXT NOT NULL, event_ids TEXT NOT NULL, data TEXT NOT NULL, CONSTRAINT receipts_graph_uniqueness UNIQUE (room_id, receipt_type, user_id) ); +CREATE TABLE receipts_linearized ( stream_id BIGINT NOT NULL, room_id TEXT NOT NULL, receipt_type TEXT NOT NULL, user_id TEXT NOT NULL, event_id TEXT NOT NULL, data TEXT NOT NULL, CONSTRAINT receipts_linearized_uniqueness UNIQUE (room_id, receipt_type, user_id) ); +CREATE INDEX receipts_linearized_id ON receipts_linearized( stream_id ); +CREATE INDEX receipts_linearized_room_stream ON receipts_linearized( room_id, stream_id ); +CREATE TABLE IF NOT EXISTS "user_threepids" ( user_id TEXT NOT NULL, medium TEXT NOT NULL, address TEXT NOT NULL, validated_at BIGINT NOT NULL, added_at BIGINT NOT NULL, CONSTRAINT medium_address UNIQUE (medium, address) ); +CREATE INDEX user_threepids_user_id ON user_threepids(user_id); +CREATE VIRTUAL TABLE event_search USING fts4 ( event_id, room_id, sender, key, value ) +/* event_search(event_id,room_id,sender,"key",value) */; +CREATE TABLE IF NOT EXISTS 'event_search_content'(docid INTEGER PRIMARY KEY, 'c0event_id', 'c1room_id', 'c2sender', 'c3key', 'c4value'); +CREATE TABLE IF NOT EXISTS 'event_search_segments'(blockid INTEGER PRIMARY KEY, block BLOB); +CREATE TABLE IF NOT EXISTS 'event_search_segdir'(level INTEGER,idx INTEGER,start_block INTEGER,leaves_end_block INTEGER,end_block INTEGER,root BLOB,PRIMARY KEY(level, idx)); +CREATE TABLE IF NOT EXISTS 'event_search_docsize'(docid INTEGER PRIMARY KEY, size BLOB); +CREATE TABLE IF NOT EXISTS 'event_search_stat'(id INTEGER PRIMARY KEY, value BLOB); +CREATE TABLE guest_access( event_id TEXT NOT NULL, room_id TEXT NOT NULL, guest_access TEXT NOT NULL, UNIQUE (event_id) ); +CREATE TABLE history_visibility( event_id TEXT NOT NULL, room_id TEXT NOT NULL, history_visibility TEXT NOT NULL, UNIQUE (event_id) ); +CREATE TABLE room_tags( user_id TEXT NOT NULL, room_id TEXT NOT NULL, tag TEXT NOT NULL, content TEXT NOT NULL, CONSTRAINT room_tag_uniqueness UNIQUE (user_id, room_id, tag) ); +CREATE TABLE room_tags_revisions ( user_id TEXT NOT NULL, room_id TEXT NOT NULL, stream_id BIGINT NOT NULL, CONSTRAINT room_tag_revisions_uniqueness UNIQUE (user_id, room_id) ); +CREATE TABLE IF NOT EXISTS "account_data_max_stream_id"( Lock CHAR(1) NOT NULL DEFAULT 'X' UNIQUE, stream_id BIGINT NOT NULL, CHECK (Lock='X') ); +CREATE TABLE account_data( user_id TEXT NOT NULL, account_data_type TEXT NOT NULL, stream_id BIGINT NOT NULL, content TEXT NOT NULL, CONSTRAINT account_data_uniqueness UNIQUE (user_id, account_data_type) ); +CREATE TABLE room_account_data( user_id TEXT NOT NULL, room_id TEXT NOT NULL, account_data_type TEXT NOT NULL, stream_id BIGINT NOT NULL, content TEXT NOT NULL, CONSTRAINT room_account_data_uniqueness UNIQUE (user_id, room_id, account_data_type) ); +CREATE INDEX account_data_stream_id on account_data(user_id, stream_id); +CREATE INDEX room_account_data_stream_id on room_account_data(user_id, stream_id); +CREATE INDEX events_ts ON events(origin_server_ts, stream_ordering); +CREATE TABLE event_push_actions( room_id TEXT NOT NULL, event_id TEXT NOT NULL, user_id TEXT NOT NULL, profile_tag VARCHAR(32), actions TEXT NOT NULL, topological_ordering BIGINT, stream_ordering BIGINT, notif SMALLINT, highlight SMALLINT, CONSTRAINT event_id_user_id_profile_tag_uniqueness UNIQUE (room_id, event_id, user_id, profile_tag) ); +CREATE INDEX event_push_actions_room_id_user_id on event_push_actions(room_id, user_id); +CREATE INDEX events_room_stream on events(room_id, stream_ordering); +CREATE INDEX public_room_index on rooms(is_public); +CREATE INDEX receipts_linearized_user ON receipts_linearized( user_id ); +CREATE INDEX event_push_actions_rm_tokens on event_push_actions( user_id, room_id, topological_ordering, stream_ordering ); +CREATE TABLE presence_stream( stream_id BIGINT, user_id TEXT, state TEXT, last_active_ts BIGINT, last_federation_update_ts BIGINT, last_user_sync_ts BIGINT, status_msg TEXT, currently_active BOOLEAN ); +CREATE INDEX presence_stream_id ON presence_stream(stream_id, user_id); +CREATE INDEX presence_stream_user_id ON presence_stream(user_id); +CREATE TABLE push_rules_stream( stream_id BIGINT NOT NULL, event_stream_ordering BIGINT NOT NULL, user_id TEXT NOT NULL, rule_id TEXT NOT NULL, op TEXT NOT NULL, priority_class SMALLINT, priority INTEGER, conditions TEXT, actions TEXT ); +CREATE INDEX push_rules_stream_id ON push_rules_stream(stream_id); +CREATE INDEX push_rules_stream_user_stream_id on push_rules_stream(user_id, stream_id); +CREATE TABLE ex_outlier_stream( event_stream_ordering BIGINT PRIMARY KEY NOT NULL, event_id TEXT NOT NULL, state_group BIGINT NOT NULL ); +CREATE TABLE threepid_guest_access_tokens( medium TEXT, address TEXT, guest_access_token TEXT, first_inviter TEXT ); +CREATE UNIQUE INDEX threepid_guest_access_tokens_index ON threepid_guest_access_tokens(medium, address); +CREATE TABLE local_invites( stream_id BIGINT NOT NULL, inviter TEXT NOT NULL, invitee TEXT NOT NULL, event_id TEXT NOT NULL, room_id TEXT NOT NULL, locally_rejected TEXT, replaced_by TEXT ); +CREATE INDEX local_invites_id ON local_invites(stream_id); +CREATE INDEX local_invites_for_user_idx ON local_invites(invitee, locally_rejected, replaced_by, room_id); +CREATE INDEX event_push_actions_stream_ordering on event_push_actions( stream_ordering, user_id ); +CREATE TABLE open_id_tokens ( token TEXT NOT NULL PRIMARY KEY, ts_valid_until_ms bigint NOT NULL, user_id TEXT NOT NULL, UNIQUE (token) ); +CREATE INDEX open_id_tokens_ts_valid_until_ms ON open_id_tokens(ts_valid_until_ms); +CREATE TABLE pusher_throttle( pusher BIGINT NOT NULL, room_id TEXT NOT NULL, last_sent_ts BIGINT, throttle_ms BIGINT, PRIMARY KEY (pusher, room_id) ); +CREATE TABLE event_reports( id BIGINT NOT NULL PRIMARY KEY, received_ts BIGINT NOT NULL, room_id TEXT NOT NULL, event_id TEXT NOT NULL, user_id TEXT NOT NULL, reason TEXT, content TEXT ); +CREATE TABLE devices ( user_id TEXT NOT NULL, device_id TEXT NOT NULL, display_name TEXT, CONSTRAINT device_uniqueness UNIQUE (user_id, device_id) ); +CREATE TABLE appservice_stream_position( Lock CHAR(1) NOT NULL DEFAULT 'X' UNIQUE, stream_ordering BIGINT, CHECK (Lock='X') ); +CREATE TABLE device_inbox ( user_id TEXT NOT NULL, device_id TEXT NOT NULL, stream_id BIGINT NOT NULL, message_json TEXT NOT NULL ); +CREATE INDEX device_inbox_user_stream_id ON device_inbox(user_id, device_id, stream_id); +CREATE INDEX received_transactions_ts ON received_transactions(ts); +CREATE TABLE device_federation_outbox ( destination TEXT NOT NULL, stream_id BIGINT NOT NULL, queued_ts BIGINT NOT NULL, messages_json TEXT NOT NULL ); +CREATE INDEX device_federation_outbox_destination_id ON device_federation_outbox(destination, stream_id); +CREATE TABLE device_federation_inbox ( origin TEXT NOT NULL, message_id TEXT NOT NULL, received_ts BIGINT NOT NULL ); +CREATE INDEX device_federation_inbox_sender_id ON device_federation_inbox(origin, message_id); +CREATE TABLE device_max_stream_id ( stream_id BIGINT NOT NULL ); +CREATE TABLE public_room_list_stream ( stream_id BIGINT NOT NULL, room_id TEXT NOT NULL, visibility BOOLEAN NOT NULL , appservice_id TEXT, network_id TEXT); +CREATE INDEX public_room_list_stream_idx on public_room_list_stream( stream_id ); +CREATE INDEX public_room_list_stream_rm_idx on public_room_list_stream( room_id, stream_id ); +CREATE TABLE state_group_edges( state_group BIGINT NOT NULL, prev_state_group BIGINT NOT NULL ); +CREATE INDEX state_group_edges_idx ON state_group_edges(state_group); +CREATE INDEX state_group_edges_prev_idx ON state_group_edges(prev_state_group); +CREATE TABLE stream_ordering_to_exterm ( stream_ordering BIGINT NOT NULL, room_id TEXT NOT NULL, event_id TEXT NOT NULL ); +CREATE INDEX stream_ordering_to_exterm_idx on stream_ordering_to_exterm( stream_ordering ); +CREATE INDEX stream_ordering_to_exterm_rm_idx on stream_ordering_to_exterm( room_id, stream_ordering ); +CREATE TABLE IF NOT EXISTS "event_auth"( event_id TEXT NOT NULL, auth_id TEXT NOT NULL, room_id TEXT NOT NULL ); +CREATE INDEX evauth_edges_id ON event_auth(event_id); +CREATE INDEX user_threepids_medium_address on user_threepids (medium, address); +CREATE TABLE appservice_room_list( appservice_id TEXT NOT NULL, network_id TEXT NOT NULL, room_id TEXT NOT NULL ); +CREATE UNIQUE INDEX appservice_room_list_idx ON appservice_room_list( appservice_id, network_id, room_id ); +CREATE INDEX device_federation_outbox_id ON device_federation_outbox(stream_id); +CREATE TABLE federation_stream_position( type TEXT NOT NULL, stream_id INTEGER NOT NULL ); +CREATE TABLE device_lists_remote_cache ( user_id TEXT NOT NULL, device_id TEXT NOT NULL, content TEXT NOT NULL ); +CREATE TABLE device_lists_remote_extremeties ( user_id TEXT NOT NULL, stream_id TEXT NOT NULL ); +CREATE TABLE device_lists_stream ( stream_id BIGINT NOT NULL, user_id TEXT NOT NULL, device_id TEXT NOT NULL ); +CREATE INDEX device_lists_stream_id ON device_lists_stream(stream_id, user_id); +CREATE TABLE device_lists_outbound_pokes ( destination TEXT NOT NULL, stream_id BIGINT NOT NULL, user_id TEXT NOT NULL, device_id TEXT NOT NULL, sent BOOLEAN NOT NULL, ts BIGINT NOT NULL ); +CREATE INDEX device_lists_outbound_pokes_id ON device_lists_outbound_pokes(destination, stream_id); +CREATE INDEX device_lists_outbound_pokes_user ON device_lists_outbound_pokes(destination, user_id); +CREATE TABLE event_push_summary ( user_id TEXT NOT NULL, room_id TEXT NOT NULL, notif_count BIGINT NOT NULL, stream_ordering BIGINT NOT NULL ); +CREATE INDEX event_push_summary_user_rm ON event_push_summary(user_id, room_id); +CREATE TABLE event_push_summary_stream_ordering ( Lock CHAR(1) NOT NULL DEFAULT 'X' UNIQUE, stream_ordering BIGINT NOT NULL, CHECK (Lock='X') ); +CREATE TABLE IF NOT EXISTS "pushers" ( id BIGINT PRIMARY KEY, user_name TEXT NOT NULL, access_token BIGINT DEFAULT NULL, profile_tag TEXT NOT NULL, kind TEXT NOT NULL, app_id TEXT NOT NULL, app_display_name TEXT NOT NULL, device_display_name TEXT NOT NULL, pushkey TEXT NOT NULL, ts BIGINT NOT NULL, lang TEXT, data TEXT, last_stream_ordering INTEGER, last_success BIGINT, failing_since BIGINT, UNIQUE (app_id, pushkey, user_name) ); +CREATE INDEX device_lists_outbound_pokes_stream ON device_lists_outbound_pokes(stream_id); +CREATE TABLE ratelimit_override ( user_id TEXT NOT NULL, messages_per_second BIGINT, burst_count BIGINT ); +CREATE UNIQUE INDEX ratelimit_override_idx ON ratelimit_override(user_id); +CREATE TABLE current_state_delta_stream ( stream_id BIGINT NOT NULL, room_id TEXT NOT NULL, type TEXT NOT NULL, state_key TEXT NOT NULL, event_id TEXT, prev_event_id TEXT ); +CREATE INDEX current_state_delta_stream_idx ON current_state_delta_stream(stream_id); +CREATE TABLE device_lists_outbound_last_success ( destination TEXT NOT NULL, user_id TEXT NOT NULL, stream_id BIGINT NOT NULL ); +CREATE INDEX device_lists_outbound_last_success_idx ON device_lists_outbound_last_success( destination, user_id, stream_id ); +CREATE TABLE user_directory_stream_pos ( Lock CHAR(1) NOT NULL DEFAULT 'X' UNIQUE, stream_id BIGINT, CHECK (Lock='X') ); +CREATE VIRTUAL TABLE user_directory_search USING fts4 ( user_id, value ) +/* user_directory_search(user_id,value) */; +CREATE TABLE IF NOT EXISTS 'user_directory_search_content'(docid INTEGER PRIMARY KEY, 'c0user_id', 'c1value'); +CREATE TABLE IF NOT EXISTS 'user_directory_search_segments'(blockid INTEGER PRIMARY KEY, block BLOB); +CREATE TABLE IF NOT EXISTS 'user_directory_search_segdir'(level INTEGER,idx INTEGER,start_block INTEGER,leaves_end_block INTEGER,end_block INTEGER,root BLOB,PRIMARY KEY(level, idx)); +CREATE TABLE IF NOT EXISTS 'user_directory_search_docsize'(docid INTEGER PRIMARY KEY, size BLOB); +CREATE TABLE IF NOT EXISTS 'user_directory_search_stat'(id INTEGER PRIMARY KEY, value BLOB); +CREATE TABLE blocked_rooms ( room_id TEXT NOT NULL, user_id TEXT NOT NULL ); +CREATE UNIQUE INDEX blocked_rooms_idx ON blocked_rooms(room_id); +CREATE TABLE IF NOT EXISTS "local_media_repository_url_cache"( url TEXT, response_code INTEGER, etag TEXT, expires_ts BIGINT, og TEXT, media_id TEXT, download_ts BIGINT ); +CREATE INDEX local_media_repository_url_cache_expires_idx ON local_media_repository_url_cache(expires_ts); +CREATE INDEX local_media_repository_url_cache_by_url_download_ts ON local_media_repository_url_cache(url, download_ts); +CREATE INDEX local_media_repository_url_cache_media_idx ON local_media_repository_url_cache(media_id); +CREATE TABLE group_users ( group_id TEXT NOT NULL, user_id TEXT NOT NULL, is_admin BOOLEAN NOT NULL, is_public BOOLEAN NOT NULL ); +CREATE TABLE group_invites ( group_id TEXT NOT NULL, user_id TEXT NOT NULL ); +CREATE TABLE group_rooms ( group_id TEXT NOT NULL, room_id TEXT NOT NULL, is_public BOOLEAN NOT NULL ); +CREATE TABLE group_summary_rooms ( group_id TEXT NOT NULL, room_id TEXT NOT NULL, category_id TEXT NOT NULL, room_order BIGINT NOT NULL, is_public BOOLEAN NOT NULL, UNIQUE (group_id, category_id, room_id, room_order), CHECK (room_order > 0) ); +CREATE UNIQUE INDEX group_summary_rooms_g_idx ON group_summary_rooms(group_id, room_id, category_id); +CREATE TABLE group_summary_room_categories ( group_id TEXT NOT NULL, category_id TEXT NOT NULL, cat_order BIGINT NOT NULL, UNIQUE (group_id, category_id, cat_order), CHECK (cat_order > 0) ); +CREATE TABLE group_room_categories ( group_id TEXT NOT NULL, category_id TEXT NOT NULL, profile TEXT NOT NULL, is_public BOOLEAN NOT NULL, UNIQUE (group_id, category_id) ); +CREATE TABLE group_summary_users ( group_id TEXT NOT NULL, user_id TEXT NOT NULL, role_id TEXT NOT NULL, user_order BIGINT NOT NULL, is_public BOOLEAN NOT NULL ); +CREATE INDEX group_summary_users_g_idx ON group_summary_users(group_id); +CREATE TABLE group_summary_roles ( group_id TEXT NOT NULL, role_id TEXT NOT NULL, role_order BIGINT NOT NULL, UNIQUE (group_id, role_id, role_order), CHECK (role_order > 0) ); +CREATE TABLE group_roles ( group_id TEXT NOT NULL, role_id TEXT NOT NULL, profile TEXT NOT NULL, is_public BOOLEAN NOT NULL, UNIQUE (group_id, role_id) ); +CREATE TABLE group_attestations_renewals ( group_id TEXT NOT NULL, user_id TEXT NOT NULL, valid_until_ms BIGINT NOT NULL ); +CREATE INDEX group_attestations_renewals_g_idx ON group_attestations_renewals(group_id, user_id); +CREATE INDEX group_attestations_renewals_u_idx ON group_attestations_renewals(user_id); +CREATE INDEX group_attestations_renewals_v_idx ON group_attestations_renewals(valid_until_ms); +CREATE TABLE group_attestations_remote ( group_id TEXT NOT NULL, user_id TEXT NOT NULL, valid_until_ms BIGINT NOT NULL, attestation_json TEXT NOT NULL ); +CREATE INDEX group_attestations_remote_g_idx ON group_attestations_remote(group_id, user_id); +CREATE INDEX group_attestations_remote_u_idx ON group_attestations_remote(user_id); +CREATE INDEX group_attestations_remote_v_idx ON group_attestations_remote(valid_until_ms); +CREATE TABLE local_group_membership ( group_id TEXT NOT NULL, user_id TEXT NOT NULL, is_admin BOOLEAN NOT NULL, membership TEXT NOT NULL, is_publicised BOOLEAN NOT NULL, content TEXT NOT NULL ); +CREATE INDEX local_group_membership_u_idx ON local_group_membership(user_id, group_id); +CREATE INDEX local_group_membership_g_idx ON local_group_membership(group_id); +CREATE TABLE local_group_updates ( stream_id BIGINT NOT NULL, group_id TEXT NOT NULL, user_id TEXT NOT NULL, type TEXT NOT NULL, content TEXT NOT NULL ); +CREATE TABLE remote_profile_cache ( user_id TEXT NOT NULL, displayname TEXT, avatar_url TEXT, last_check BIGINT NOT NULL ); +CREATE UNIQUE INDEX remote_profile_cache_user_id ON remote_profile_cache(user_id); +CREATE INDEX remote_profile_cache_time ON remote_profile_cache(last_check); +CREATE TABLE IF NOT EXISTS "deleted_pushers" ( stream_id BIGINT NOT NULL, app_id TEXT NOT NULL, pushkey TEXT NOT NULL, user_id TEXT NOT NULL ); +CREATE INDEX deleted_pushers_stream_id ON deleted_pushers (stream_id); +CREATE TABLE IF NOT EXISTS "groups" ( group_id TEXT NOT NULL, name TEXT, avatar_url TEXT, short_description TEXT, long_description TEXT, is_public BOOL NOT NULL , join_policy TEXT NOT NULL DEFAULT 'invite'); +CREATE UNIQUE INDEX groups_idx ON groups(group_id); +CREATE TABLE IF NOT EXISTS "user_directory" ( user_id TEXT NOT NULL, room_id TEXT, display_name TEXT, avatar_url TEXT ); +CREATE INDEX user_directory_room_idx ON user_directory(room_id); +CREATE UNIQUE INDEX user_directory_user_idx ON user_directory(user_id); +CREATE TABLE event_push_actions_staging ( event_id TEXT NOT NULL, user_id TEXT NOT NULL, actions TEXT NOT NULL, notif SMALLINT NOT NULL, highlight SMALLINT NOT NULL ); +CREATE INDEX event_push_actions_staging_id ON event_push_actions_staging(event_id); +CREATE TABLE users_pending_deactivation ( user_id TEXT NOT NULL ); +CREATE UNIQUE INDEX group_invites_g_idx ON group_invites(group_id, user_id); +CREATE UNIQUE INDEX group_users_g_idx ON group_users(group_id, user_id); +CREATE INDEX group_users_u_idx ON group_users(user_id); +CREATE INDEX group_invites_u_idx ON group_invites(user_id); +CREATE UNIQUE INDEX group_rooms_g_idx ON group_rooms(group_id, room_id); +CREATE INDEX group_rooms_r_idx ON group_rooms(room_id); +CREATE TABLE user_daily_visits ( user_id TEXT NOT NULL, device_id TEXT, timestamp BIGINT NOT NULL ); +CREATE INDEX user_daily_visits_uts_idx ON user_daily_visits(user_id, timestamp); +CREATE INDEX user_daily_visits_ts_idx ON user_daily_visits(timestamp); +CREATE TABLE erased_users ( user_id TEXT NOT NULL ); +CREATE UNIQUE INDEX erased_users_user ON erased_users(user_id); +CREATE TABLE monthly_active_users ( user_id TEXT NOT NULL, timestamp BIGINT NOT NULL ); +CREATE UNIQUE INDEX monthly_active_users_users ON monthly_active_users(user_id); +CREATE INDEX monthly_active_users_time_stamp ON monthly_active_users(timestamp); +CREATE TABLE IF NOT EXISTS "e2e_room_keys_versions" ( user_id TEXT NOT NULL, version BIGINT NOT NULL, algorithm TEXT NOT NULL, auth_data TEXT NOT NULL, deleted SMALLINT DEFAULT 0 NOT NULL ); +CREATE UNIQUE INDEX e2e_room_keys_versions_idx ON e2e_room_keys_versions(user_id, version); +CREATE TABLE IF NOT EXISTS "e2e_room_keys" ( user_id TEXT NOT NULL, room_id TEXT NOT NULL, session_id TEXT NOT NULL, version BIGINT NOT NULL, first_message_index INT, forwarded_count INT, is_verified BOOLEAN, session_data TEXT NOT NULL ); +CREATE UNIQUE INDEX e2e_room_keys_idx ON e2e_room_keys(user_id, room_id, session_id); +CREATE TABLE users_who_share_private_rooms ( user_id TEXT NOT NULL, other_user_id TEXT NOT NULL, room_id TEXT NOT NULL ); +CREATE UNIQUE INDEX users_who_share_private_rooms_u_idx ON users_who_share_private_rooms(user_id, other_user_id, room_id); +CREATE INDEX users_who_share_private_rooms_r_idx ON users_who_share_private_rooms(room_id); +CREATE INDEX users_who_share_private_rooms_o_idx ON users_who_share_private_rooms(other_user_id); +CREATE TABLE user_threepid_id_server ( user_id TEXT NOT NULL, medium TEXT NOT NULL, address TEXT NOT NULL, id_server TEXT NOT NULL ); +CREATE UNIQUE INDEX user_threepid_id_server_idx ON user_threepid_id_server( user_id, medium, address, id_server ); +CREATE TABLE users_in_public_rooms ( user_id TEXT NOT NULL, room_id TEXT NOT NULL ); +CREATE UNIQUE INDEX users_in_public_rooms_u_idx ON users_in_public_rooms(user_id, room_id); +CREATE TABLE account_validity ( user_id TEXT PRIMARY KEY, expiration_ts_ms BIGINT NOT NULL, email_sent BOOLEAN NOT NULL, renewal_token TEXT ); +CREATE TABLE event_relations ( event_id TEXT NOT NULL, relates_to_id TEXT NOT NULL, relation_type TEXT NOT NULL, aggregation_key TEXT ); +CREATE UNIQUE INDEX event_relations_id ON event_relations(event_id); +CREATE INDEX event_relations_relates ON event_relations(relates_to_id, relation_type, aggregation_key); +CREATE TABLE stats_stream_pos ( Lock CHAR(1) NOT NULL DEFAULT 'X' UNIQUE, stream_id BIGINT, CHECK (Lock='X') ); +CREATE TABLE user_stats ( user_id TEXT NOT NULL, ts BIGINT NOT NULL, bucket_size INT NOT NULL, public_rooms INT NOT NULL, private_rooms INT NOT NULL ); +CREATE UNIQUE INDEX user_stats_user_ts ON user_stats(user_id, ts); +CREATE TABLE room_stats ( room_id TEXT NOT NULL, ts BIGINT NOT NULL, bucket_size INT NOT NULL, current_state_events INT NOT NULL, joined_members INT NOT NULL, invited_members INT NOT NULL, left_members INT NOT NULL, banned_members INT NOT NULL, state_events INT NOT NULL ); +CREATE UNIQUE INDEX room_stats_room_ts ON room_stats(room_id, ts); +CREATE TABLE room_state ( room_id TEXT NOT NULL, join_rules TEXT, history_visibility TEXT, encryption TEXT, name TEXT, topic TEXT, avatar TEXT, canonical_alias TEXT ); +CREATE UNIQUE INDEX room_state_room ON room_state(room_id); +CREATE TABLE room_stats_earliest_token ( room_id TEXT NOT NULL, token BIGINT NOT NULL ); +CREATE UNIQUE INDEX room_stats_earliest_token_idx ON room_stats_earliest_token(room_id); +CREATE INDEX access_tokens_device_id ON access_tokens (user_id, device_id); +CREATE INDEX user_ips_device_id ON user_ips (user_id, device_id, last_seen); +CREATE INDEX event_contains_url_index ON events (room_id, topological_ordering, stream_ordering); +CREATE INDEX event_push_actions_u_highlight ON event_push_actions (user_id, stream_ordering); +CREATE INDEX event_push_actions_highlights_index ON event_push_actions (user_id, room_id, topological_ordering, stream_ordering); +CREATE INDEX current_state_events_member_index ON current_state_events (state_key); +CREATE INDEX device_inbox_stream_id_user_id ON device_inbox (stream_id, user_id); +CREATE INDEX device_lists_stream_user_id ON device_lists_stream (user_id, device_id); +CREATE INDEX local_media_repository_url_idx ON local_media_repository (created_ts); +CREATE INDEX user_ips_last_seen ON user_ips (user_id, last_seen); +CREATE INDEX user_ips_last_seen_only ON user_ips (last_seen); +CREATE INDEX users_creation_ts ON users (creation_ts); +CREATE INDEX event_to_state_groups_sg_index ON event_to_state_groups (state_group); +CREATE UNIQUE INDEX device_lists_remote_cache_unique_id ON device_lists_remote_cache (user_id, device_id); +CREATE INDEX state_groups_state_type_idx ON state_groups_state(state_group, type, state_key); +CREATE UNIQUE INDEX device_lists_remote_extremeties_unique_idx ON device_lists_remote_extremeties (user_id); +CREATE UNIQUE INDEX user_ips_user_token_ip_unique_index ON user_ips (user_id, access_token, ip); diff --git a/synapse/storage/data_stores/main/schema/full_schemas/54/stream_positions.sql b/synapse/storage/data_stores/main/schema/full_schemas/54/stream_positions.sql new file mode 100644 index 0000000000..c265fd20e2 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/full_schemas/54/stream_positions.sql @@ -0,0 +1,7 @@ + +INSERT INTO appservice_stream_position (stream_ordering) SELECT COALESCE(MAX(stream_ordering), 0) FROM events; +INSERT INTO federation_stream_position (type, stream_id) VALUES ('federation', -1); +INSERT INTO federation_stream_position (type, stream_id) SELECT 'events', coalesce(max(stream_ordering), -1) FROM events; +INSERT INTO user_directory_stream_pos (stream_id) VALUES (0); +INSERT INTO stats_stream_pos (stream_id) VALUES (0); +INSERT INTO event_push_summary_stream_ordering (stream_ordering) VALUES (0); diff --git a/synapse/storage/data_stores/main/schema/full_schemas/README.txt b/synapse/storage/data_stores/main/schema/full_schemas/README.txt new file mode 100644 index 0000000000..d3f6401344 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/full_schemas/README.txt @@ -0,0 +1,19 @@ +Building full schema dumps +========================== + +These schemas need to be made from a database that has had all background updates run. + +Postgres +-------- + +$ pg_dump --format=plain --schema-only --no-tablespaces --no-acl --no-owner $DATABASE_NAME| sed -e '/^--/d' -e 's/public\.//g' -e '/^SET /d' -e '/^SELECT /d' > full.sql.postgres + +SQLite +------ + +$ sqlite3 $DATABASE_FILE ".schema" > full.sql.sqlite + +After +----- + +Delete the CREATE statements for "sqlite_stat1", "schema_version", "applied_schema_deltas", and "applied_module_schemas". \ No newline at end of file diff --git a/synapse/storage/data_stores/main/search.py b/synapse/storage/data_stores/main/search.py new file mode 100644 index 0000000000..0e08497452 --- /dev/null +++ b/synapse/storage/data_stores/main/search.py @@ -0,0 +1,712 @@ +# -*- coding: utf-8 -*- +# Copyright 2015, 2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging +import re +from collections import namedtuple + +from six import string_types + +from canonicaljson import json + +from twisted.internet import defer + +from synapse.api.errors import SynapseError +from synapse.storage._base import make_in_list_sql_clause +from synapse.storage.background_updates import BackgroundUpdateStore +from synapse.storage.engines import PostgresEngine, Sqlite3Engine + +logger = logging.getLogger(__name__) + +SearchEntry = namedtuple( + "SearchEntry", + ["key", "value", "event_id", "room_id", "stream_ordering", "origin_server_ts"], +) + + +class SearchBackgroundUpdateStore(BackgroundUpdateStore): + + EVENT_SEARCH_UPDATE_NAME = "event_search" + EVENT_SEARCH_ORDER_UPDATE_NAME = "event_search_order" + EVENT_SEARCH_USE_GIST_POSTGRES_NAME = "event_search_postgres_gist" + EVENT_SEARCH_USE_GIN_POSTGRES_NAME = "event_search_postgres_gin" + + def __init__(self, db_conn, hs): + super(SearchBackgroundUpdateStore, self).__init__(db_conn, hs) + + if not hs.config.enable_search: + return + + self.register_background_update_handler( + self.EVENT_SEARCH_UPDATE_NAME, self._background_reindex_search + ) + self.register_background_update_handler( + self.EVENT_SEARCH_ORDER_UPDATE_NAME, self._background_reindex_search_order + ) + + # we used to have a background update to turn the GIN index into a + # GIST one; we no longer do that (obviously) because we actually want + # a GIN index. However, it's possible that some people might still have + # the background update queued, so we register a handler to clear the + # background update. + self.register_noop_background_update(self.EVENT_SEARCH_USE_GIST_POSTGRES_NAME) + + self.register_background_update_handler( + self.EVENT_SEARCH_USE_GIN_POSTGRES_NAME, self._background_reindex_gin_search + ) + + @defer.inlineCallbacks + def _background_reindex_search(self, progress, batch_size): + # we work through the events table from highest stream id to lowest + target_min_stream_id = progress["target_min_stream_id_inclusive"] + max_stream_id = progress["max_stream_id_exclusive"] + rows_inserted = progress.get("rows_inserted", 0) + + TYPES = ["m.room.name", "m.room.message", "m.room.topic"] + + def reindex_search_txn(txn): + sql = ( + "SELECT stream_ordering, event_id, room_id, type, json, " + " origin_server_ts FROM events" + " JOIN event_json USING (room_id, event_id)" + " WHERE ? <= stream_ordering AND stream_ordering < ?" + " AND (%s)" + " ORDER BY stream_ordering DESC" + " LIMIT ?" + ) % (" OR ".join("type = '%s'" % (t,) for t in TYPES),) + + txn.execute(sql, (target_min_stream_id, max_stream_id, batch_size)) + + # we could stream straight from the results into + # store_search_entries_txn with a generator function, but that + # would mean having two cursors open on the database at once. + # Instead we just build a list of results. + rows = self.cursor_to_dict(txn) + if not rows: + return 0 + + min_stream_id = rows[-1]["stream_ordering"] + + event_search_rows = [] + for row in rows: + try: + event_id = row["event_id"] + room_id = row["room_id"] + etype = row["type"] + stream_ordering = row["stream_ordering"] + origin_server_ts = row["origin_server_ts"] + try: + event_json = json.loads(row["json"]) + content = event_json["content"] + except Exception: + continue + + if etype == "m.room.message": + key = "content.body" + value = content["body"] + elif etype == "m.room.topic": + key = "content.topic" + value = content["topic"] + elif etype == "m.room.name": + key = "content.name" + value = content["name"] + else: + raise Exception("unexpected event type %s" % etype) + except (KeyError, AttributeError): + # If the event is missing a necessary field then + # skip over it. + continue + + if not isinstance(value, string_types): + # If the event body, name or topic isn't a string + # then skip over it + continue + + event_search_rows.append( + SearchEntry( + key=key, + value=value, + event_id=event_id, + room_id=room_id, + stream_ordering=stream_ordering, + origin_server_ts=origin_server_ts, + ) + ) + + self.store_search_entries_txn(txn, event_search_rows) + + progress = { + "target_min_stream_id_inclusive": target_min_stream_id, + "max_stream_id_exclusive": min_stream_id, + "rows_inserted": rows_inserted + len(event_search_rows), + } + + self._background_update_progress_txn( + txn, self.EVENT_SEARCH_UPDATE_NAME, progress + ) + + return len(event_search_rows) + + result = yield self.runInteraction( + self.EVENT_SEARCH_UPDATE_NAME, reindex_search_txn + ) + + if not result: + yield self._end_background_update(self.EVENT_SEARCH_UPDATE_NAME) + + return result + + @defer.inlineCallbacks + def _background_reindex_gin_search(self, progress, batch_size): + """This handles old synapses which used GIST indexes, if any; + converting them back to be GIN as per the actual schema. + """ + + def create_index(conn): + conn.rollback() + + # we have to set autocommit, because postgres refuses to + # CREATE INDEX CONCURRENTLY without it. + conn.set_session(autocommit=True) + + try: + c = conn.cursor() + + # if we skipped the conversion to GIST, we may already/still + # have an event_search_fts_idx; unfortunately postgres 9.4 + # doesn't support CREATE INDEX IF EXISTS so we just catch the + # exception and ignore it. + import psycopg2 + + try: + c.execute( + "CREATE INDEX CONCURRENTLY event_search_fts_idx" + " ON event_search USING GIN (vector)" + ) + except psycopg2.ProgrammingError as e: + logger.warn( + "Ignoring error %r when trying to switch from GIST to GIN", e + ) + + # we should now be able to delete the GIST index. + c.execute("DROP INDEX IF EXISTS event_search_fts_idx_gist") + finally: + conn.set_session(autocommit=False) + + if isinstance(self.database_engine, PostgresEngine): + yield self.runWithConnection(create_index) + + yield self._end_background_update(self.EVENT_SEARCH_USE_GIN_POSTGRES_NAME) + return 1 + + @defer.inlineCallbacks + def _background_reindex_search_order(self, progress, batch_size): + target_min_stream_id = progress["target_min_stream_id_inclusive"] + max_stream_id = progress["max_stream_id_exclusive"] + rows_inserted = progress.get("rows_inserted", 0) + have_added_index = progress["have_added_indexes"] + + if not have_added_index: + + def create_index(conn): + conn.rollback() + conn.set_session(autocommit=True) + c = conn.cursor() + + # We create with NULLS FIRST so that when we search *backwards* + # we get the ones with non null origin_server_ts *first* + c.execute( + "CREATE INDEX CONCURRENTLY event_search_room_order ON event_search(" + "room_id, origin_server_ts NULLS FIRST, stream_ordering NULLS FIRST)" + ) + c.execute( + "CREATE INDEX CONCURRENTLY event_search_order ON event_search(" + "origin_server_ts NULLS FIRST, stream_ordering NULLS FIRST)" + ) + conn.set_session(autocommit=False) + + yield self.runWithConnection(create_index) + + pg = dict(progress) + pg["have_added_indexes"] = True + + yield self.runInteraction( + self.EVENT_SEARCH_ORDER_UPDATE_NAME, + self._background_update_progress_txn, + self.EVENT_SEARCH_ORDER_UPDATE_NAME, + pg, + ) + + def reindex_search_txn(txn): + sql = ( + "UPDATE event_search AS es SET stream_ordering = e.stream_ordering," + " origin_server_ts = e.origin_server_ts" + " FROM events AS e" + " WHERE e.event_id = es.event_id" + " AND ? <= e.stream_ordering AND e.stream_ordering < ?" + " RETURNING es.stream_ordering" + ) + + min_stream_id = max_stream_id - batch_size + txn.execute(sql, (min_stream_id, max_stream_id)) + rows = txn.fetchall() + + if min_stream_id < target_min_stream_id: + # We've recached the end. + return len(rows), False + + progress = { + "target_min_stream_id_inclusive": target_min_stream_id, + "max_stream_id_exclusive": min_stream_id, + "rows_inserted": rows_inserted + len(rows), + "have_added_indexes": True, + } + + self._background_update_progress_txn( + txn, self.EVENT_SEARCH_ORDER_UPDATE_NAME, progress + ) + + return len(rows), True + + num_rows, finished = yield self.runInteraction( + self.EVENT_SEARCH_ORDER_UPDATE_NAME, reindex_search_txn + ) + + if not finished: + yield self._end_background_update(self.EVENT_SEARCH_ORDER_UPDATE_NAME) + + return num_rows + + def store_search_entries_txn(self, txn, entries): + """Add entries to the search table + + Args: + txn (cursor): + entries (iterable[SearchEntry]): + entries to be added to the table + """ + if not self.hs.config.enable_search: + return + if isinstance(self.database_engine, PostgresEngine): + sql = ( + "INSERT INTO event_search" + " (event_id, room_id, key, vector, stream_ordering, origin_server_ts)" + " VALUES (?,?,?,to_tsvector('english', ?),?,?)" + ) + + args = ( + ( + entry.event_id, + entry.room_id, + entry.key, + entry.value, + entry.stream_ordering, + entry.origin_server_ts, + ) + for entry in entries + ) + + txn.executemany(sql, args) + + elif isinstance(self.database_engine, Sqlite3Engine): + sql = ( + "INSERT INTO event_search (event_id, room_id, key, value)" + " VALUES (?,?,?,?)" + ) + args = ( + (entry.event_id, entry.room_id, entry.key, entry.value) + for entry in entries + ) + + txn.executemany(sql, args) + else: + # This should be unreachable. + raise Exception("Unrecognized database engine") + + +class SearchStore(SearchBackgroundUpdateStore): + def __init__(self, db_conn, hs): + super(SearchStore, self).__init__(db_conn, hs) + + def store_event_search_txn(self, txn, event, key, value): + """Add event to the search table + + Args: + txn (cursor): + event (EventBase): + key (str): + value (str): + """ + self.store_search_entries_txn( + txn, + ( + SearchEntry( + key=key, + value=value, + event_id=event.event_id, + room_id=event.room_id, + stream_ordering=event.internal_metadata.stream_ordering, + origin_server_ts=event.origin_server_ts, + ), + ), + ) + + @defer.inlineCallbacks + def search_msgs(self, room_ids, search_term, keys): + """Performs a full text search over events with given keys. + + Args: + room_ids (list): List of room ids to search in + search_term (str): Search term to search for + keys (list): List of keys to search in, currently supports + "content.body", "content.name", "content.topic" + + Returns: + list of dicts + """ + clauses = [] + + search_query = search_query = _parse_query(self.database_engine, search_term) + + args = [] + + # Make sure we don't explode because the person is in too many rooms. + # We filter the results below regardless. + if len(room_ids) < 500: + clause, args = make_in_list_sql_clause( + self.database_engine, "room_id", room_ids + ) + clauses = [clause] + + local_clauses = [] + for key in keys: + local_clauses.append("key = ?") + args.append(key) + + clauses.append("(%s)" % (" OR ".join(local_clauses),)) + + count_args = args + count_clauses = clauses + + if isinstance(self.database_engine, PostgresEngine): + sql = ( + "SELECT ts_rank_cd(vector, to_tsquery('english', ?)) AS rank," + " room_id, event_id" + " FROM event_search" + " WHERE vector @@ to_tsquery('english', ?)" + ) + args = [search_query, search_query] + args + + count_sql = ( + "SELECT room_id, count(*) as count FROM event_search" + " WHERE vector @@ to_tsquery('english', ?)" + ) + count_args = [search_query] + count_args + elif isinstance(self.database_engine, Sqlite3Engine): + sql = ( + "SELECT rank(matchinfo(event_search)) as rank, room_id, event_id" + " FROM event_search" + " WHERE value MATCH ?" + ) + args = [search_query] + args + + count_sql = ( + "SELECT room_id, count(*) as count FROM event_search" + " WHERE value MATCH ?" + ) + count_args = [search_term] + count_args + else: + # This should be unreachable. + raise Exception("Unrecognized database engine") + + for clause in clauses: + sql += " AND " + clause + + for clause in count_clauses: + count_sql += " AND " + clause + + # We add an arbitrary limit here to ensure we don't try to pull the + # entire table from the database. + sql += " ORDER BY rank DESC LIMIT 500" + + results = yield self._execute("search_msgs", self.cursor_to_dict, sql, *args) + + results = list(filter(lambda row: row["room_id"] in room_ids, results)) + + events = yield self.get_events_as_list([r["event_id"] for r in results]) + + event_map = {ev.event_id: ev for ev in events} + + highlights = None + if isinstance(self.database_engine, PostgresEngine): + highlights = yield self._find_highlights_in_postgres(search_query, events) + + count_sql += " GROUP BY room_id" + + count_results = yield self._execute( + "search_rooms_count", self.cursor_to_dict, count_sql, *count_args + ) + + count = sum(row["count"] for row in count_results if row["room_id"] in room_ids) + + return { + "results": [ + {"event": event_map[r["event_id"]], "rank": r["rank"]} + for r in results + if r["event_id"] in event_map + ], + "highlights": highlights, + "count": count, + } + + @defer.inlineCallbacks + def search_rooms(self, room_ids, search_term, keys, limit, pagination_token=None): + """Performs a full text search over events with given keys. + + Args: + room_id (list): The room_ids to search in + search_term (str): Search term to search for + keys (list): List of keys to search in, currently supports + "content.body", "content.name", "content.topic" + pagination_token (str): A pagination token previously returned + + Returns: + list of dicts + """ + clauses = [] + + search_query = search_query = _parse_query(self.database_engine, search_term) + + args = [] + + # Make sure we don't explode because the person is in too many rooms. + # We filter the results below regardless. + if len(room_ids) < 500: + clause, args = make_in_list_sql_clause( + self.database_engine, "room_id", room_ids + ) + clauses = [clause] + + local_clauses = [] + for key in keys: + local_clauses.append("key = ?") + args.append(key) + + clauses.append("(%s)" % (" OR ".join(local_clauses),)) + + # take copies of the current args and clauses lists, before adding + # pagination clauses to main query. + count_args = list(args) + count_clauses = list(clauses) + + if pagination_token: + try: + origin_server_ts, stream = pagination_token.split(",") + origin_server_ts = int(origin_server_ts) + stream = int(stream) + except Exception: + raise SynapseError(400, "Invalid pagination token") + + clauses.append( + "(origin_server_ts < ?" + " OR (origin_server_ts = ? AND stream_ordering < ?))" + ) + args.extend([origin_server_ts, origin_server_ts, stream]) + + if isinstance(self.database_engine, PostgresEngine): + sql = ( + "SELECT ts_rank_cd(vector, to_tsquery('english', ?)) as rank," + " origin_server_ts, stream_ordering, room_id, event_id" + " FROM event_search" + " WHERE vector @@ to_tsquery('english', ?) AND " + ) + args = [search_query, search_query] + args + + count_sql = ( + "SELECT room_id, count(*) as count FROM event_search" + " WHERE vector @@ to_tsquery('english', ?) AND " + ) + count_args = [search_query] + count_args + elif isinstance(self.database_engine, Sqlite3Engine): + # We use CROSS JOIN here to ensure we use the right indexes. + # https://sqlite.org/optoverview.html#crossjoin + # + # We want to use the full text search index on event_search to + # extract all possible matches first, then lookup those matches + # in the events table to get the topological ordering. We need + # to use the indexes in this order because sqlite refuses to + # MATCH unless it uses the full text search index + sql = ( + "SELECT rank(matchinfo) as rank, room_id, event_id," + " origin_server_ts, stream_ordering" + " FROM (SELECT key, event_id, matchinfo(event_search) as matchinfo" + " FROM event_search" + " WHERE value MATCH ?" + " )" + " CROSS JOIN events USING (event_id)" + " WHERE " + ) + args = [search_query] + args + + count_sql = ( + "SELECT room_id, count(*) as count FROM event_search" + " WHERE value MATCH ? AND " + ) + count_args = [search_term] + count_args + else: + # This should be unreachable. + raise Exception("Unrecognized database engine") + + sql += " AND ".join(clauses) + count_sql += " AND ".join(count_clauses) + + # We add an arbitrary limit here to ensure we don't try to pull the + # entire table from the database. + if isinstance(self.database_engine, PostgresEngine): + sql += ( + " ORDER BY origin_server_ts DESC NULLS LAST," + " stream_ordering DESC NULLS LAST LIMIT ?" + ) + elif isinstance(self.database_engine, Sqlite3Engine): + sql += " ORDER BY origin_server_ts DESC, stream_ordering DESC LIMIT ?" + else: + raise Exception("Unrecognized database engine") + + args.append(limit) + + results = yield self._execute("search_rooms", self.cursor_to_dict, sql, *args) + + results = list(filter(lambda row: row["room_id"] in room_ids, results)) + + events = yield self.get_events_as_list([r["event_id"] for r in results]) + + event_map = {ev.event_id: ev for ev in events} + + highlights = None + if isinstance(self.database_engine, PostgresEngine): + highlights = yield self._find_highlights_in_postgres(search_query, events) + + count_sql += " GROUP BY room_id" + + count_results = yield self._execute( + "search_rooms_count", self.cursor_to_dict, count_sql, *count_args + ) + + count = sum(row["count"] for row in count_results if row["room_id"] in room_ids) + + return { + "results": [ + { + "event": event_map[r["event_id"]], + "rank": r["rank"], + "pagination_token": "%s,%s" + % (r["origin_server_ts"], r["stream_ordering"]), + } + for r in results + if r["event_id"] in event_map + ], + "highlights": highlights, + "count": count, + } + + def _find_highlights_in_postgres(self, search_query, events): + """Given a list of events and a search term, return a list of words + that match from the content of the event. + + This is used to give a list of words that clients can match against to + highlight the matching parts. + + Args: + search_query (str) + events (list): A list of events + + Returns: + deferred : A set of strings. + """ + + def f(txn): + highlight_words = set() + for event in events: + # As a hack we simply join values of all possible keys. This is + # fine since we're only using them to find possible highlights. + values = [] + for key in ("body", "name", "topic"): + v = event.content.get(key, None) + if v: + values.append(v) + + if not values: + continue + + value = " ".join(values) + + # We need to find some values for StartSel and StopSel that + # aren't in the value so that we can pick results out. + start_sel = "<" + stop_sel = ">" + + while start_sel in value: + start_sel += "<" + while stop_sel in value: + stop_sel += ">" + + query = "SELECT ts_headline(?, to_tsquery('english', ?), %s)" % ( + _to_postgres_options( + { + "StartSel": start_sel, + "StopSel": stop_sel, + "MaxFragments": "50", + } + ) + ) + txn.execute(query, (value, search_query)) + headline, = txn.fetchall()[0] + + # Now we need to pick the possible highlights out of the haedline + # result. + matcher_regex = "%s(.*?)%s" % ( + re.escape(start_sel), + re.escape(stop_sel), + ) + + res = re.findall(matcher_regex, headline) + highlight_words.update([r.lower() for r in res]) + + return highlight_words + + return self.runInteraction("_find_highlights", f) + + +def _to_postgres_options(options_dict): + return "'%s'" % (",".join("%s=%s" % (k, v) for k, v in options_dict.items()),) + + +def _parse_query(database_engine, search_term): + """Takes a plain unicode string from the user and converts it into a form + that can be passed to database. + We use this so that we can add prefix matching, which isn't something + that is supported by default. + """ + + # Pull out the individual words, discarding any non-word characters. + results = re.findall(r"([\w\-]+)", search_term, re.UNICODE) + + if isinstance(database_engine, PostgresEngine): + return " & ".join(result + ":*" for result in results) + elif isinstance(database_engine, Sqlite3Engine): + return " & ".join(result + "*" for result in results) + else: + # This should be unreachable. + raise Exception("Unrecognized database engine") diff --git a/synapse/storage/data_stores/main/signatures.py b/synapse/storage/data_stores/main/signatures.py new file mode 100644 index 0000000000..556191b76f --- /dev/null +++ b/synapse/storage/data_stores/main/signatures.py @@ -0,0 +1,101 @@ +# -*- coding: utf-8 -*- +# Copyright 2014-2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import six + +from unpaddedbase64 import encode_base64 + +from twisted.internet import defer + +from synapse.crypto.event_signing import compute_event_reference_hash +from synapse.storage._base import SQLBaseStore +from synapse.util.caches.descriptors import cached, cachedList + +# py2 sqlite has buffer hardcoded as only binary type, so we must use it, +# despite being deprecated and removed in favor of memoryview +if six.PY2: + db_binary_type = six.moves.builtins.buffer +else: + db_binary_type = memoryview + + +class SignatureWorkerStore(SQLBaseStore): + @cached() + def get_event_reference_hash(self, event_id): + # This is a dummy function to allow get_event_reference_hashes + # to use its cache + raise NotImplementedError() + + @cachedList( + cached_method_name="get_event_reference_hash", list_name="event_ids", num_args=1 + ) + def get_event_reference_hashes(self, event_ids): + def f(txn): + return { + event_id: self._get_event_reference_hashes_txn(txn, event_id) + for event_id in event_ids + } + + return self.runInteraction("get_event_reference_hashes", f) + + @defer.inlineCallbacks + def add_event_hashes(self, event_ids): + hashes = yield self.get_event_reference_hashes(event_ids) + hashes = { + e_id: {k: encode_base64(v) for k, v in h.items() if k == "sha256"} + for e_id, h in hashes.items() + } + + return list(hashes.items()) + + def _get_event_reference_hashes_txn(self, txn, event_id): + """Get all the hashes for a given PDU. + Args: + txn (cursor): + event_id (str): Id for the Event. + Returns: + A dict[unicode, bytes] of algorithm -> hash. + """ + query = ( + "SELECT algorithm, hash" + " FROM event_reference_hashes" + " WHERE event_id = ?" + ) + txn.execute(query, (event_id,)) + return {k: v for k, v in txn} + + +class SignatureStore(SignatureWorkerStore): + """Persistence for event signatures and hashes""" + + def _store_event_reference_hashes_txn(self, txn, events): + """Store a hash for a PDU + Args: + txn (cursor): + events (list): list of Events. + """ + + vals = [] + for event in events: + ref_alg, ref_hash_bytes = compute_event_reference_hash(event) + vals.append( + { + "event_id": event.event_id, + "algorithm": ref_alg, + "hash": db_binary_type(ref_hash_bytes), + } + ) + + self._simple_insert_many_txn(txn, table="event_reference_hashes", values=vals) diff --git a/synapse/storage/data_stores/main/state.py b/synapse/storage/data_stores/main/state.py new file mode 100644 index 0000000000..d54442e5fa --- /dev/null +++ b/synapse/storage/data_stores/main/state.py @@ -0,0 +1,1244 @@ +# -*- coding: utf-8 -*- +# Copyright 2014-2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging +from collections import namedtuple + +from six import iteritems, itervalues +from six.moves import range + +from twisted.internet import defer + +from synapse.api.constants import EventTypes +from synapse.api.errors import NotFoundError +from synapse.storage._base import SQLBaseStore +from synapse.storage.background_updates import BackgroundUpdateStore +from synapse.storage.data_stores.main.events_worker import EventsWorkerStore +from synapse.storage.engines import PostgresEngine +from synapse.storage.state import StateFilter +from synapse.util.caches import get_cache_factor_for, intern_string +from synapse.util.caches.descriptors import cached, cachedList +from synapse.util.caches.dictionary_cache import DictionaryCache +from synapse.util.stringutils import to_ascii + +logger = logging.getLogger(__name__) + + +MAX_STATE_DELTA_HOPS = 100 + + +class _GetStateGroupDelta( + namedtuple("_GetStateGroupDelta", ("prev_group", "delta_ids")) +): + """Return type of get_state_group_delta that implements __len__, which lets + us use the itrable flag when caching + """ + + __slots__ = [] + + def __len__(self): + return len(self.delta_ids) if self.delta_ids else 0 + + +class StateGroupBackgroundUpdateStore(SQLBaseStore): + """Defines functions related to state groups needed to run the state backgroud + updates. + """ + + def _count_state_group_hops_txn(self, txn, state_group): + """Given a state group, count how many hops there are in the tree. + + This is used to ensure the delta chains don't get too long. + """ + if isinstance(self.database_engine, PostgresEngine): + sql = """ + WITH RECURSIVE state(state_group) AS ( + VALUES(?::bigint) + UNION ALL + SELECT prev_state_group FROM state_group_edges e, state s + WHERE s.state_group = e.state_group + ) + SELECT count(*) FROM state; + """ + + txn.execute(sql, (state_group,)) + row = txn.fetchone() + if row and row[0]: + return row[0] + else: + return 0 + else: + # We don't use WITH RECURSIVE on sqlite3 as there are distributions + # that ship with an sqlite3 version that doesn't support it (e.g. wheezy) + next_group = state_group + count = 0 + + while next_group: + next_group = self._simple_select_one_onecol_txn( + txn, + table="state_group_edges", + keyvalues={"state_group": next_group}, + retcol="prev_state_group", + allow_none=True, + ) + if next_group: + count += 1 + + return count + + def _get_state_groups_from_groups_txn( + self, txn, groups, state_filter=StateFilter.all() + ): + results = {group: {} for group in groups} + + where_clause, where_args = state_filter.make_sql_filter_clause() + + # Unless the filter clause is empty, we're going to append it after an + # existing where clause + if where_clause: + where_clause = " AND (%s)" % (where_clause,) + + if isinstance(self.database_engine, PostgresEngine): + # Temporarily disable sequential scans in this transaction. This is + # a temporary hack until we can add the right indices in + txn.execute("SET LOCAL enable_seqscan=off") + + # The below query walks the state_group tree so that the "state" + # table includes all state_groups in the tree. It then joins + # against `state_groups_state` to fetch the latest state. + # It assumes that previous state groups are always numerically + # lesser. + # The PARTITION is used to get the event_id in the greatest state + # group for the given type, state_key. + # This may return multiple rows per (type, state_key), but last_value + # should be the same. + sql = """ + WITH RECURSIVE state(state_group) AS ( + VALUES(?::bigint) + UNION ALL + SELECT prev_state_group FROM state_group_edges e, state s + WHERE s.state_group = e.state_group + ) + SELECT DISTINCT type, state_key, last_value(event_id) OVER ( + PARTITION BY type, state_key ORDER BY state_group ASC + ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING + ) AS event_id FROM state_groups_state + WHERE state_group IN ( + SELECT state_group FROM state + ) + """ + + for group in groups: + args = [group] + args.extend(where_args) + + txn.execute(sql + where_clause, args) + for row in txn: + typ, state_key, event_id = row + key = (typ, state_key) + results[group][key] = event_id + else: + max_entries_returned = state_filter.max_entries_returned() + + # We don't use WITH RECURSIVE on sqlite3 as there are distributions + # that ship with an sqlite3 version that doesn't support it (e.g. wheezy) + for group in groups: + next_group = group + + while next_group: + # We did this before by getting the list of group ids, and + # then passing that list to sqlite to get latest event for + # each (type, state_key). However, that was terribly slow + # without the right indices (which we can't add until + # after we finish deduping state, which requires this func) + args = [next_group] + args.extend(where_args) + + txn.execute( + "SELECT type, state_key, event_id FROM state_groups_state" + " WHERE state_group = ? " + where_clause, + args, + ) + results[group].update( + ((typ, state_key), event_id) + for typ, state_key, event_id in txn + if (typ, state_key) not in results[group] + ) + + # If the number of entries in the (type,state_key)->event_id dict + # matches the number of (type,state_keys) types we were searching + # for, then we must have found them all, so no need to go walk + # further down the tree... UNLESS our types filter contained + # wildcards (i.e. Nones) in which case we have to do an exhaustive + # search + if ( + max_entries_returned is not None + and len(results[group]) == max_entries_returned + ): + break + + next_group = self._simple_select_one_onecol_txn( + txn, + table="state_group_edges", + keyvalues={"state_group": next_group}, + retcol="prev_state_group", + allow_none=True, + ) + + return results + + +# this inherits from EventsWorkerStore because it calls self.get_events +class StateGroupWorkerStore( + EventsWorkerStore, StateGroupBackgroundUpdateStore, SQLBaseStore +): + """The parts of StateGroupStore that can be called from workers. + """ + + STATE_GROUP_DEDUPLICATION_UPDATE_NAME = "state_group_state_deduplication" + STATE_GROUP_INDEX_UPDATE_NAME = "state_group_state_type_index" + CURRENT_STATE_INDEX_UPDATE_NAME = "current_state_members_idx" + + def __init__(self, db_conn, hs): + super(StateGroupWorkerStore, self).__init__(db_conn, hs) + + # Originally the state store used a single DictionaryCache to cache the + # event IDs for the state types in a given state group to avoid hammering + # on the state_group* tables. + # + # The point of using a DictionaryCache is that it can cache a subset + # of the state events for a given state group (i.e. a subset of the keys for a + # given dict which is an entry in the cache for a given state group ID). + # + # However, this poses problems when performing complicated queries + # on the store - for instance: "give me all the state for this group, but + # limit members to this subset of users", as DictionaryCache's API isn't + # rich enough to say "please cache any of these fields, apart from this subset". + # This is problematic when lazy loading members, which requires this behaviour, + # as without it the cache has no choice but to speculatively load all + # state events for the group, which negates the efficiency being sought. + # + # Rather than overcomplicating DictionaryCache's API, we instead split the + # state_group_cache into two halves - one for tracking non-member events, + # and the other for tracking member_events. This means that lazy loading + # queries can be made in a cache-friendly manner by querying both caches + # separately and then merging the result. So for the example above, you + # would query the members cache for a specific subset of state keys + # (which DictionaryCache will handle efficiently and fine) and the non-members + # cache for all state (which DictionaryCache will similarly handle fine) + # and then just merge the results together. + # + # We size the non-members cache to be smaller than the members cache as the + # vast majority of state in Matrix (today) is member events. + + self._state_group_cache = DictionaryCache( + "*stateGroupCache*", + # TODO: this hasn't been tuned yet + 50000 * get_cache_factor_for("stateGroupCache"), + ) + self._state_group_members_cache = DictionaryCache( + "*stateGroupMembersCache*", + 500000 * get_cache_factor_for("stateGroupMembersCache"), + ) + + @defer.inlineCallbacks + def get_room_version(self, room_id): + """Get the room_version of a given room + + Args: + room_id (str) + + Returns: + Deferred[str] + + Raises: + NotFoundError if the room is unknown + """ + # for now we do this by looking at the create event. We may want to cache this + # more intelligently in future. + + # Retrieve the room's create event + create_event = yield self.get_create_event_for_room(room_id) + return create_event.content.get("room_version", "1") + + @defer.inlineCallbacks + def get_room_predecessor(self, room_id): + """Get the predecessor room of an upgraded room if one exists. + Otherwise return None. + + Args: + room_id (str) + + Returns: + Deferred[unicode|None]: predecessor room id + + Raises: + NotFoundError if the room is unknown + """ + # Retrieve the room's create event + create_event = yield self.get_create_event_for_room(room_id) + + # Return predecessor if present + return create_event.content.get("predecessor", None) + + @defer.inlineCallbacks + def get_create_event_for_room(self, room_id): + """Get the create state event for a room. + + Args: + room_id (str) + + Returns: + Deferred[EventBase]: The room creation event. + + Raises: + NotFoundError if the room is unknown + """ + state_ids = yield self.get_current_state_ids(room_id) + create_id = state_ids.get((EventTypes.Create, "")) + + # If we can't find the create event, assume we've hit a dead end + if not create_id: + raise NotFoundError("Unknown room %s" % (room_id)) + + # Retrieve the room's create event and return + create_event = yield self.get_event(create_id) + return create_event + + @cached(max_entries=100000, iterable=True) + def get_current_state_ids(self, room_id): + """Get the current state event ids for a room based on the + current_state_events table. + + Args: + room_id (str) + + Returns: + deferred: dict of (type, state_key) -> event_id + """ + + def _get_current_state_ids_txn(txn): + txn.execute( + """SELECT type, state_key, event_id FROM current_state_events + WHERE room_id = ? + """, + (room_id,), + ) + + return { + (intern_string(r[0]), intern_string(r[1])): to_ascii(r[2]) for r in txn + } + + return self.runInteraction("get_current_state_ids", _get_current_state_ids_txn) + + # FIXME: how should this be cached? + def get_filtered_current_state_ids(self, room_id, state_filter=StateFilter.all()): + """Get the current state event of a given type for a room based on the + current_state_events table. This may not be as up-to-date as the result + of doing a fresh state resolution as per state_handler.get_current_state + + Args: + room_id (str) + state_filter (StateFilter): The state filter used to fetch state + from the database. + + Returns: + Deferred[dict[tuple[str, str], str]]: Map from type/state_key to + event ID. + """ + + where_clause, where_args = state_filter.make_sql_filter_clause() + + if not where_clause: + # We delegate to the cached version + return self.get_current_state_ids(room_id) + + def _get_filtered_current_state_ids_txn(txn): + results = {} + sql = """ + SELECT type, state_key, event_id FROM current_state_events + WHERE room_id = ? + """ + + if where_clause: + sql += " AND (%s)" % (where_clause,) + + args = [room_id] + args.extend(where_args) + txn.execute(sql, args) + for row in txn: + typ, state_key, event_id = row + key = (intern_string(typ), intern_string(state_key)) + results[key] = event_id + + return results + + return self.runInteraction( + "get_filtered_current_state_ids", _get_filtered_current_state_ids_txn + ) + + @defer.inlineCallbacks + def get_canonical_alias_for_room(self, room_id): + """Get canonical alias for room, if any + + Args: + room_id (str) + + Returns: + Deferred[str|None]: The canonical alias, if any + """ + + state = yield self.get_filtered_current_state_ids( + room_id, StateFilter.from_types([(EventTypes.CanonicalAlias, "")]) + ) + + event_id = state.get((EventTypes.CanonicalAlias, "")) + if not event_id: + return + + event = yield self.get_event(event_id, allow_none=True) + if not event: + return + + return event.content.get("canonical_alias") + + @cached(max_entries=10000, iterable=True) + def get_state_group_delta(self, state_group): + """Given a state group try to return a previous group and a delta between + the old and the new. + + Returns: + (prev_group, delta_ids), where both may be None. + """ + + def _get_state_group_delta_txn(txn): + prev_group = self._simple_select_one_onecol_txn( + txn, + table="state_group_edges", + keyvalues={"state_group": state_group}, + retcol="prev_state_group", + allow_none=True, + ) + + if not prev_group: + return _GetStateGroupDelta(None, None) + + delta_ids = self._simple_select_list_txn( + txn, + table="state_groups_state", + keyvalues={"state_group": state_group}, + retcols=("type", "state_key", "event_id"), + ) + + return _GetStateGroupDelta( + prev_group, + {(row["type"], row["state_key"]): row["event_id"] for row in delta_ids}, + ) + + return self.runInteraction("get_state_group_delta", _get_state_group_delta_txn) + + @defer.inlineCallbacks + def get_state_groups_ids(self, _room_id, event_ids): + """Get the event IDs of all the state for the state groups for the given events + + Args: + _room_id (str): id of the room for these events + event_ids (iterable[str]): ids of the events + + Returns: + Deferred[dict[int, dict[tuple[str, str], str]]]: + dict of state_group_id -> (dict of (type, state_key) -> event id) + """ + if not event_ids: + return {} + + event_to_groups = yield self._get_state_group_for_events(event_ids) + + groups = set(itervalues(event_to_groups)) + group_to_state = yield self._get_state_for_groups(groups) + + return group_to_state + + @defer.inlineCallbacks + def get_state_ids_for_group(self, state_group): + """Get the event IDs of all the state in the given state group + + Args: + state_group (int) + + Returns: + Deferred[dict]: Resolves to a map of (type, state_key) -> event_id + """ + group_to_state = yield self._get_state_for_groups((state_group,)) + + return group_to_state[state_group] + + @defer.inlineCallbacks + def get_state_groups(self, room_id, event_ids): + """ Get the state groups for the given list of event_ids + + Returns: + Deferred[dict[int, list[EventBase]]]: + dict of state_group_id -> list of state events. + """ + if not event_ids: + return {} + + group_to_ids = yield self.get_state_groups_ids(room_id, event_ids) + + state_event_map = yield self.get_events( + [ + ev_id + for group_ids in itervalues(group_to_ids) + for ev_id in itervalues(group_ids) + ], + get_prev_content=False, + ) + + return { + group: [ + state_event_map[v] + for v in itervalues(event_id_map) + if v in state_event_map + ] + for group, event_id_map in iteritems(group_to_ids) + } + + @defer.inlineCallbacks + def _get_state_groups_from_groups(self, groups, state_filter): + """Returns the state groups for a given set of groups, filtering on + types of state events. + + Args: + groups(list[int]): list of state group IDs to query + state_filter (StateFilter): The state filter used to fetch state + from the database. + Returns: + Deferred[dict[int, dict[tuple[str, str], str]]]: + dict of state_group_id -> (dict of (type, state_key) -> event id) + """ + results = {} + + chunks = [groups[i : i + 100] for i in range(0, len(groups), 100)] + for chunk in chunks: + res = yield self.runInteraction( + "_get_state_groups_from_groups", + self._get_state_groups_from_groups_txn, + chunk, + state_filter, + ) + results.update(res) + + return results + + @defer.inlineCallbacks + def get_state_for_events(self, event_ids, state_filter=StateFilter.all()): + """Given a list of event_ids and type tuples, return a list of state + dicts for each event. + + Args: + event_ids (list[string]) + state_filter (StateFilter): The state filter used to fetch state + from the database. + + Returns: + deferred: A dict of (event_id) -> (type, state_key) -> [state_events] + """ + event_to_groups = yield self._get_state_group_for_events(event_ids) + + groups = set(itervalues(event_to_groups)) + group_to_state = yield self._get_state_for_groups(groups, state_filter) + + state_event_map = yield self.get_events( + [ev_id for sd in itervalues(group_to_state) for ev_id in itervalues(sd)], + get_prev_content=False, + ) + + event_to_state = { + event_id: { + k: state_event_map[v] + for k, v in iteritems(group_to_state[group]) + if v in state_event_map + } + for event_id, group in iteritems(event_to_groups) + } + + return {event: event_to_state[event] for event in event_ids} + + @defer.inlineCallbacks + def get_state_ids_for_events(self, event_ids, state_filter=StateFilter.all()): + """ + Get the state dicts corresponding to a list of events, containing the event_ids + of the state events (as opposed to the events themselves) + + Args: + event_ids(list(str)): events whose state should be returned + state_filter (StateFilter): The state filter used to fetch state + from the database. + + Returns: + A deferred dict from event_id -> (type, state_key) -> event_id + """ + event_to_groups = yield self._get_state_group_for_events(event_ids) + + groups = set(itervalues(event_to_groups)) + group_to_state = yield self._get_state_for_groups(groups, state_filter) + + event_to_state = { + event_id: group_to_state[group] + for event_id, group in iteritems(event_to_groups) + } + + return {event: event_to_state[event] for event in event_ids} + + @defer.inlineCallbacks + def get_state_for_event(self, event_id, state_filter=StateFilter.all()): + """ + Get the state dict corresponding to a particular event + + Args: + event_id(str): event whose state should be returned + state_filter (StateFilter): The state filter used to fetch state + from the database. + + Returns: + A deferred dict from (type, state_key) -> state_event + """ + state_map = yield self.get_state_for_events([event_id], state_filter) + return state_map[event_id] + + @defer.inlineCallbacks + def get_state_ids_for_event(self, event_id, state_filter=StateFilter.all()): + """ + Get the state dict corresponding to a particular event + + Args: + event_id(str): event whose state should be returned + state_filter (StateFilter): The state filter used to fetch state + from the database. + + Returns: + A deferred dict from (type, state_key) -> state_event + """ + state_map = yield self.get_state_ids_for_events([event_id], state_filter) + return state_map[event_id] + + @cached(max_entries=50000) + def _get_state_group_for_event(self, event_id): + return self._simple_select_one_onecol( + table="event_to_state_groups", + keyvalues={"event_id": event_id}, + retcol="state_group", + allow_none=True, + desc="_get_state_group_for_event", + ) + + @cachedList( + cached_method_name="_get_state_group_for_event", + list_name="event_ids", + num_args=1, + inlineCallbacks=True, + ) + def _get_state_group_for_events(self, event_ids): + """Returns mapping event_id -> state_group + """ + rows = yield self._simple_select_many_batch( + table="event_to_state_groups", + column="event_id", + iterable=event_ids, + keyvalues={}, + retcols=("event_id", "state_group"), + desc="_get_state_group_for_events", + ) + + return {row["event_id"]: row["state_group"] for row in rows} + + def _get_state_for_group_using_cache(self, cache, group, state_filter): + """Checks if group is in cache. See `_get_state_for_groups` + + Args: + cache(DictionaryCache): the state group cache to use + group(int): The state group to lookup + state_filter (StateFilter): The state filter used to fetch state + from the database. + + Returns 2-tuple (`state_dict`, `got_all`). + `got_all` is a bool indicating if we successfully retrieved all + requests state from the cache, if False we need to query the DB for the + missing state. + """ + is_all, known_absent, state_dict_ids = cache.get(group) + + if is_all or state_filter.is_full(): + # Either we have everything or want everything, either way + # `is_all` tells us whether we've gotten everything. + return state_filter.filter_state(state_dict_ids), is_all + + # tracks whether any of our requested types are missing from the cache + missing_types = False + + if state_filter.has_wildcards(): + # We don't know if we fetched all the state keys for the types in + # the filter that are wildcards, so we have to assume that we may + # have missed some. + missing_types = True + else: + # There aren't any wild cards, so `concrete_types()` returns the + # complete list of event types we're wanting. + for key in state_filter.concrete_types(): + if key not in state_dict_ids and key not in known_absent: + missing_types = True + break + + return state_filter.filter_state(state_dict_ids), not missing_types + + @defer.inlineCallbacks + def _get_state_for_groups(self, groups, state_filter=StateFilter.all()): + """Gets the state at each of a list of state groups, optionally + filtering by type/state_key + + Args: + groups (iterable[int]): list of state groups for which we want + to get the state. + state_filter (StateFilter): The state filter used to fetch state + from the database. + Returns: + Deferred[dict[int, dict[tuple[str, str], str]]]: + dict of state_group_id -> (dict of (type, state_key) -> event id) + """ + + member_filter, non_member_filter = state_filter.get_member_split() + + # Now we look them up in the member and non-member caches + non_member_state, incomplete_groups_nm, = ( + yield self._get_state_for_groups_using_cache( + groups, self._state_group_cache, state_filter=non_member_filter + ) + ) + + member_state, incomplete_groups_m, = ( + yield self._get_state_for_groups_using_cache( + groups, self._state_group_members_cache, state_filter=member_filter + ) + ) + + state = dict(non_member_state) + for group in groups: + state[group].update(member_state[group]) + + # Now fetch any missing groups from the database + + incomplete_groups = incomplete_groups_m | incomplete_groups_nm + + if not incomplete_groups: + return state + + cache_sequence_nm = self._state_group_cache.sequence + cache_sequence_m = self._state_group_members_cache.sequence + + # Help the cache hit ratio by expanding the filter a bit + db_state_filter = state_filter.return_expanded() + + group_to_state_dict = yield self._get_state_groups_from_groups( + list(incomplete_groups), state_filter=db_state_filter + ) + + # Now lets update the caches + self._insert_into_cache( + group_to_state_dict, + db_state_filter, + cache_seq_num_members=cache_sequence_m, + cache_seq_num_non_members=cache_sequence_nm, + ) + + # And finally update the result dict, by filtering out any extra + # stuff we pulled out of the database. + for group, group_state_dict in iteritems(group_to_state_dict): + # We just replace any existing entries, as we will have loaded + # everything we need from the database anyway. + state[group] = state_filter.filter_state(group_state_dict) + + return state + + def _get_state_for_groups_using_cache(self, groups, cache, state_filter): + """Gets the state at each of a list of state groups, optionally + filtering by type/state_key, querying from a specific cache. + + Args: + groups (iterable[int]): list of state groups for which we want + to get the state. + cache (DictionaryCache): the cache of group ids to state dicts which + we will pass through - either the normal state cache or the specific + members state cache. + state_filter (StateFilter): The state filter used to fetch state + from the database. + + Returns: + tuple[dict[int, dict[tuple[str, str], str]], set[int]]: Tuple of + dict of state_group_id -> (dict of (type, state_key) -> event id) + of entries in the cache, and the state group ids either missing + from the cache or incomplete. + """ + results = {} + incomplete_groups = set() + for group in set(groups): + state_dict_ids, got_all = self._get_state_for_group_using_cache( + cache, group, state_filter + ) + results[group] = state_dict_ids + + if not got_all: + incomplete_groups.add(group) + + return results, incomplete_groups + + def _insert_into_cache( + self, + group_to_state_dict, + state_filter, + cache_seq_num_members, + cache_seq_num_non_members, + ): + """Inserts results from querying the database into the relevant cache. + + Args: + group_to_state_dict (dict): The new entries pulled from database. + Map from state group to state dict + state_filter (StateFilter): The state filter used to fetch state + from the database. + cache_seq_num_members (int): Sequence number of member cache since + last lookup in cache + cache_seq_num_non_members (int): Sequence number of member cache since + last lookup in cache + """ + + # We need to work out which types we've fetched from the DB for the + # member vs non-member caches. This should be as accurate as possible, + # but can be an underestimate (e.g. when we have wild cards) + + member_filter, non_member_filter = state_filter.get_member_split() + if member_filter.is_full(): + # We fetched all member events + member_types = None + else: + # `concrete_types()` will only return a subset when there are wild + # cards in the filter, but that's fine. + member_types = member_filter.concrete_types() + + if non_member_filter.is_full(): + # We fetched all non member events + non_member_types = None + else: + non_member_types = non_member_filter.concrete_types() + + for group, group_state_dict in iteritems(group_to_state_dict): + state_dict_members = {} + state_dict_non_members = {} + + for k, v in iteritems(group_state_dict): + if k[0] == EventTypes.Member: + state_dict_members[k] = v + else: + state_dict_non_members[k] = v + + self._state_group_members_cache.update( + cache_seq_num_members, + key=group, + value=state_dict_members, + fetched_keys=member_types, + ) + + self._state_group_cache.update( + cache_seq_num_non_members, + key=group, + value=state_dict_non_members, + fetched_keys=non_member_types, + ) + + def store_state_group( + self, event_id, room_id, prev_group, delta_ids, current_state_ids + ): + """Store a new set of state, returning a newly assigned state group. + + Args: + event_id (str): The event ID for which the state was calculated + room_id (str) + prev_group (int|None): A previous state group for the room, optional. + delta_ids (dict|None): The delta between state at `prev_group` and + `current_state_ids`, if `prev_group` was given. Same format as + `current_state_ids`. + current_state_ids (dict): The state to store. Map of (type, state_key) + to event_id. + + Returns: + Deferred[int]: The state group ID + """ + + def _store_state_group_txn(txn): + if current_state_ids is None: + # AFAIK, this can never happen + raise Exception("current_state_ids cannot be None") + + state_group = self.database_engine.get_next_state_group_id(txn) + + self._simple_insert_txn( + txn, + table="state_groups", + values={"id": state_group, "room_id": room_id, "event_id": event_id}, + ) + + # We persist as a delta if we can, while also ensuring the chain + # of deltas isn't tooo long, as otherwise read performance degrades. + if prev_group: + is_in_db = self._simple_select_one_onecol_txn( + txn, + table="state_groups", + keyvalues={"id": prev_group}, + retcol="id", + allow_none=True, + ) + if not is_in_db: + raise Exception( + "Trying to persist state with unpersisted prev_group: %r" + % (prev_group,) + ) + + potential_hops = self._count_state_group_hops_txn(txn, prev_group) + if prev_group and potential_hops < MAX_STATE_DELTA_HOPS: + self._simple_insert_txn( + txn, + table="state_group_edges", + values={"state_group": state_group, "prev_state_group": prev_group}, + ) + + self._simple_insert_many_txn( + txn, + table="state_groups_state", + values=[ + { + "state_group": state_group, + "room_id": room_id, + "type": key[0], + "state_key": key[1], + "event_id": state_id, + } + for key, state_id in iteritems(delta_ids) + ], + ) + else: + self._simple_insert_many_txn( + txn, + table="state_groups_state", + values=[ + { + "state_group": state_group, + "room_id": room_id, + "type": key[0], + "state_key": key[1], + "event_id": state_id, + } + for key, state_id in iteritems(current_state_ids) + ], + ) + + # Prefill the state group caches with this group. + # It's fine to use the sequence like this as the state group map + # is immutable. (If the map wasn't immutable then this prefill could + # race with another update) + + current_member_state_ids = { + s: ev + for (s, ev) in iteritems(current_state_ids) + if s[0] == EventTypes.Member + } + txn.call_after( + self._state_group_members_cache.update, + self._state_group_members_cache.sequence, + key=state_group, + value=dict(current_member_state_ids), + ) + + current_non_member_state_ids = { + s: ev + for (s, ev) in iteritems(current_state_ids) + if s[0] != EventTypes.Member + } + txn.call_after( + self._state_group_cache.update, + self._state_group_cache.sequence, + key=state_group, + value=dict(current_non_member_state_ids), + ) + + return state_group + + return self.runInteraction("store_state_group", _store_state_group_txn) + + +class StateBackgroundUpdateStore( + StateGroupBackgroundUpdateStore, BackgroundUpdateStore +): + + STATE_GROUP_DEDUPLICATION_UPDATE_NAME = "state_group_state_deduplication" + STATE_GROUP_INDEX_UPDATE_NAME = "state_group_state_type_index" + CURRENT_STATE_INDEX_UPDATE_NAME = "current_state_members_idx" + EVENT_STATE_GROUP_INDEX_UPDATE_NAME = "event_to_state_groups_sg_index" + + def __init__(self, db_conn, hs): + super(StateBackgroundUpdateStore, self).__init__(db_conn, hs) + self.register_background_update_handler( + self.STATE_GROUP_DEDUPLICATION_UPDATE_NAME, + self._background_deduplicate_state, + ) + self.register_background_update_handler( + self.STATE_GROUP_INDEX_UPDATE_NAME, self._background_index_state + ) + self.register_background_index_update( + self.CURRENT_STATE_INDEX_UPDATE_NAME, + index_name="current_state_events_member_index", + table="current_state_events", + columns=["state_key"], + where_clause="type='m.room.member'", + ) + self.register_background_index_update( + self.EVENT_STATE_GROUP_INDEX_UPDATE_NAME, + index_name="event_to_state_groups_sg_index", + table="event_to_state_groups", + columns=["state_group"], + ) + + @defer.inlineCallbacks + def _background_deduplicate_state(self, progress, batch_size): + """This background update will slowly deduplicate state by reencoding + them as deltas. + """ + last_state_group = progress.get("last_state_group", 0) + rows_inserted = progress.get("rows_inserted", 0) + max_group = progress.get("max_group", None) + + BATCH_SIZE_SCALE_FACTOR = 100 + + batch_size = max(1, int(batch_size / BATCH_SIZE_SCALE_FACTOR)) + + if max_group is None: + rows = yield self._execute( + "_background_deduplicate_state", + None, + "SELECT coalesce(max(id), 0) FROM state_groups", + ) + max_group = rows[0][0] + + def reindex_txn(txn): + new_last_state_group = last_state_group + for count in range(batch_size): + txn.execute( + "SELECT id, room_id FROM state_groups" + " WHERE ? < id AND id <= ?" + " ORDER BY id ASC" + " LIMIT 1", + (new_last_state_group, max_group), + ) + row = txn.fetchone() + if row: + state_group, room_id = row + + if not row or not state_group: + return True, count + + txn.execute( + "SELECT state_group FROM state_group_edges" + " WHERE state_group = ?", + (state_group,), + ) + + # If we reach a point where we've already started inserting + # edges we should stop. + if txn.fetchall(): + return True, count + + txn.execute( + "SELECT coalesce(max(id), 0) FROM state_groups" + " WHERE id < ? AND room_id = ?", + (state_group, room_id), + ) + prev_group, = txn.fetchone() + new_last_state_group = state_group + + if prev_group: + potential_hops = self._count_state_group_hops_txn(txn, prev_group) + if potential_hops >= MAX_STATE_DELTA_HOPS: + # We want to ensure chains are at most this long,# + # otherwise read performance degrades. + continue + + prev_state = self._get_state_groups_from_groups_txn( + txn, [prev_group] + ) + prev_state = prev_state[prev_group] + + curr_state = self._get_state_groups_from_groups_txn( + txn, [state_group] + ) + curr_state = curr_state[state_group] + + if not set(prev_state.keys()) - set(curr_state.keys()): + # We can only do a delta if the current has a strict super set + # of keys + + delta_state = { + key: value + for key, value in iteritems(curr_state) + if prev_state.get(key, None) != value + } + + self._simple_delete_txn( + txn, + table="state_group_edges", + keyvalues={"state_group": state_group}, + ) + + self._simple_insert_txn( + txn, + table="state_group_edges", + values={ + "state_group": state_group, + "prev_state_group": prev_group, + }, + ) + + self._simple_delete_txn( + txn, + table="state_groups_state", + keyvalues={"state_group": state_group}, + ) + + self._simple_insert_many_txn( + txn, + table="state_groups_state", + values=[ + { + "state_group": state_group, + "room_id": room_id, + "type": key[0], + "state_key": key[1], + "event_id": state_id, + } + for key, state_id in iteritems(delta_state) + ], + ) + + progress = { + "last_state_group": state_group, + "rows_inserted": rows_inserted + batch_size, + "max_group": max_group, + } + + self._background_update_progress_txn( + txn, self.STATE_GROUP_DEDUPLICATION_UPDATE_NAME, progress + ) + + return False, batch_size + + finished, result = yield self.runInteraction( + self.STATE_GROUP_DEDUPLICATION_UPDATE_NAME, reindex_txn + ) + + if finished: + yield self._end_background_update( + self.STATE_GROUP_DEDUPLICATION_UPDATE_NAME + ) + + return result * BATCH_SIZE_SCALE_FACTOR + + @defer.inlineCallbacks + def _background_index_state(self, progress, batch_size): + def reindex_txn(conn): + conn.rollback() + if isinstance(self.database_engine, PostgresEngine): + # postgres insists on autocommit for the index + conn.set_session(autocommit=True) + try: + txn = conn.cursor() + txn.execute( + "CREATE INDEX CONCURRENTLY state_groups_state_type_idx" + " ON state_groups_state(state_group, type, state_key)" + ) + txn.execute("DROP INDEX IF EXISTS state_groups_state_id") + finally: + conn.set_session(autocommit=False) + else: + txn = conn.cursor() + txn.execute( + "CREATE INDEX state_groups_state_type_idx" + " ON state_groups_state(state_group, type, state_key)" + ) + txn.execute("DROP INDEX IF EXISTS state_groups_state_id") + + yield self.runWithConnection(reindex_txn) + + yield self._end_background_update(self.STATE_GROUP_INDEX_UPDATE_NAME) + + return 1 + + +class StateStore(StateGroupWorkerStore, StateBackgroundUpdateStore): + """ Keeps track of the state at a given event. + + This is done by the concept of `state groups`. Every event is a assigned + a state group (identified by an arbitrary string), which references a + collection of state events. The current state of an event is then the + collection of state events referenced by the event's state group. + + Hence, every change in the current state causes a new state group to be + generated. However, if no change happens (e.g., if we get a message event + with only one parent it inherits the state group from its parent.) + + There are three tables: + * `state_groups`: Stores group name, first event with in the group and + room id. + * `event_to_state_groups`: Maps events to state groups. + * `state_groups_state`: Maps state group to state events. + """ + + def __init__(self, db_conn, hs): + super(StateStore, self).__init__(db_conn, hs) + + def _store_event_state_mappings_txn(self, txn, events_and_contexts): + state_groups = {} + for event, context in events_and_contexts: + if event.internal_metadata.is_outlier(): + continue + + # if the event was rejected, just give it the same state as its + # predecessor. + if context.rejected: + state_groups[event.event_id] = context.prev_group + continue + + state_groups[event.event_id] = context.state_group + + self._simple_insert_many_txn( + txn, + table="event_to_state_groups", + values=[ + {"state_group": state_group_id, "event_id": event_id} + for event_id, state_group_id in iteritems(state_groups) + ], + ) + + for event_id, state_group_id in iteritems(state_groups): + txn.call_after( + self._get_state_group_for_event.prefill, (event_id,), state_group_id + ) diff --git a/synapse/storage/data_stores/main/state_deltas.py b/synapse/storage/data_stores/main/state_deltas.py new file mode 100644 index 0000000000..28f33ec18f --- /dev/null +++ b/synapse/storage/data_stores/main/state_deltas.py @@ -0,0 +1,119 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 Vector Creations Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging + +from synapse.storage._base import SQLBaseStore + +logger = logging.getLogger(__name__) + + +class StateDeltasStore(SQLBaseStore): + def get_current_state_deltas(self, prev_stream_id: int, max_stream_id: int): + """Fetch a list of room state changes since the given stream id + + Each entry in the result contains the following fields: + - stream_id (int) + - room_id (str) + - type (str): event type + - state_key (str): + - event_id (str|None): new event_id for this state key. None if the + state has been deleted. + - prev_event_id (str|None): previous event_id for this state key. None + if it's new state. + + Args: + prev_stream_id (int): point to get changes since (exclusive) + max_stream_id (int): the point that we know has been correctly persisted + - ie, an upper limit to return changes from. + + Returns: + Deferred[tuple[int, list[dict]]: A tuple consisting of: + - the stream id which these results go up to + - list of current_state_delta_stream rows. If it is empty, we are + up to date. + """ + prev_stream_id = int(prev_stream_id) + + # check we're not going backwards + assert prev_stream_id <= max_stream_id + + if not self._curr_state_delta_stream_cache.has_any_entity_changed( + prev_stream_id + ): + # if the CSDs haven't changed between prev_stream_id and now, we + # know for certain that they haven't changed between prev_stream_id and + # max_stream_id. + return max_stream_id, [] + + def get_current_state_deltas_txn(txn): + # First we calculate the max stream id that will give us less than + # N results. + # We arbitarily limit to 100 stream_id entries to ensure we don't + # select toooo many. + sql = """ + SELECT stream_id, count(*) + FROM current_state_delta_stream + WHERE stream_id > ? AND stream_id <= ? + GROUP BY stream_id + ORDER BY stream_id ASC + LIMIT 100 + """ + txn.execute(sql, (prev_stream_id, max_stream_id)) + + total = 0 + + for stream_id, count in txn: + total += count + if total > 100: + # We arbitarily limit to 100 entries to ensure we don't + # select toooo many. + logger.debug( + "Clipping current_state_delta_stream rows to stream_id %i", + stream_id, + ) + clipped_stream_id = stream_id + break + else: + # if there's no problem, we may as well go right up to the max_stream_id + clipped_stream_id = max_stream_id + + # Now actually get the deltas + sql = """ + SELECT stream_id, room_id, type, state_key, event_id, prev_event_id + FROM current_state_delta_stream + WHERE ? < stream_id AND stream_id <= ? + ORDER BY stream_id ASC + """ + txn.execute(sql, (prev_stream_id, clipped_stream_id)) + return clipped_stream_id, self.cursor_to_dict(txn) + + return self.runInteraction( + "get_current_state_deltas", get_current_state_deltas_txn + ) + + def _get_max_stream_id_in_current_state_deltas_txn(self, txn): + return self._simple_select_one_onecol_txn( + txn, + table="current_state_delta_stream", + keyvalues={}, + retcol="COALESCE(MAX(stream_id), -1)", + ) + + def get_max_stream_id_in_current_state_deltas(self): + return self.runInteraction( + "get_max_stream_id_in_current_state_deltas", + self._get_max_stream_id_in_current_state_deltas_txn, + ) diff --git a/synapse/storage/data_stores/main/stats.py b/synapse/storage/data_stores/main/stats.py new file mode 100644 index 0000000000..5ab639b2ad --- /dev/null +++ b/synapse/storage/data_stores/main/stats.py @@ -0,0 +1,881 @@ +# -*- coding: utf-8 -*- +# Copyright 2018, 2019 New Vector Ltd +# Copyright 2019 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging +from itertools import chain + +from twisted.internet import defer +from twisted.internet.defer import DeferredLock + +from synapse.api.constants import EventTypes, Membership +from synapse.storage.data_stores.main.state_deltas import StateDeltasStore +from synapse.storage.engines import PostgresEngine +from synapse.util.caches.descriptors import cached + +logger = logging.getLogger(__name__) + +# these fields track absolutes (e.g. total number of rooms on the server) +# You can think of these as Prometheus Gauges. +# You can draw these stats on a line graph. +# Example: number of users in a room +ABSOLUTE_STATS_FIELDS = { + "room": ( + "current_state_events", + "joined_members", + "invited_members", + "left_members", + "banned_members", + "local_users_in_room", + ), + "user": ("joined_rooms",), +} + +# these fields are per-timeslice and so should be reset to 0 upon a new slice +# You can draw these stats on a histogram. +# Example: number of events sent locally during a time slice +PER_SLICE_FIELDS = { + "room": ("total_events", "total_event_bytes"), + "user": ("invites_sent", "rooms_created", "total_events", "total_event_bytes"), +} + +TYPE_TO_TABLE = {"room": ("room_stats", "room_id"), "user": ("user_stats", "user_id")} + +# these are the tables (& ID columns) which contain our actual subjects +TYPE_TO_ORIGIN_TABLE = {"room": ("rooms", "room_id"), "user": ("users", "name")} + + +class StatsStore(StateDeltasStore): + def __init__(self, db_conn, hs): + super(StatsStore, self).__init__(db_conn, hs) + + self.server_name = hs.hostname + self.clock = self.hs.get_clock() + self.stats_enabled = hs.config.stats_enabled + self.stats_bucket_size = hs.config.stats_bucket_size + + self.stats_delta_processing_lock = DeferredLock() + + self.register_background_update_handler( + "populate_stats_process_rooms", self._populate_stats_process_rooms + ) + self.register_background_update_handler( + "populate_stats_process_users", self._populate_stats_process_users + ) + # we no longer need to perform clean-up, but we will give ourselves + # the potential to reintroduce it in the future – so documentation + # will still encourage the use of this no-op handler. + self.register_noop_background_update("populate_stats_cleanup") + self.register_noop_background_update("populate_stats_prepare") + + def quantise_stats_time(self, ts): + """ + Quantises a timestamp to be a multiple of the bucket size. + + Args: + ts (int): the timestamp to quantise, in milliseconds since the Unix + Epoch + + Returns: + int: a timestamp which + - is divisible by the bucket size; + - is no later than `ts`; and + - is the largest such timestamp. + """ + return (ts // self.stats_bucket_size) * self.stats_bucket_size + + @defer.inlineCallbacks + def _populate_stats_process_users(self, progress, batch_size): + """ + This is a background update which regenerates statistics for users. + """ + if not self.stats_enabled: + yield self._end_background_update("populate_stats_process_users") + return 1 + + last_user_id = progress.get("last_user_id", "") + + def _get_next_batch(txn): + sql = """ + SELECT DISTINCT name FROM users + WHERE name > ? + ORDER BY name ASC + LIMIT ? + """ + txn.execute(sql, (last_user_id, batch_size)) + return [r for r, in txn] + + users_to_work_on = yield self.runInteraction( + "_populate_stats_process_users", _get_next_batch + ) + + # No more rooms -- complete the transaction. + if not users_to_work_on: + yield self._end_background_update("populate_stats_process_users") + return 1 + + for user_id in users_to_work_on: + yield self._calculate_and_set_initial_state_for_user(user_id) + progress["last_user_id"] = user_id + + yield self.runInteraction( + "populate_stats_process_users", + self._background_update_progress_txn, + "populate_stats_process_users", + progress, + ) + + return len(users_to_work_on) + + @defer.inlineCallbacks + def _populate_stats_process_rooms(self, progress, batch_size): + """ + This is a background update which regenerates statistics for rooms. + """ + if not self.stats_enabled: + yield self._end_background_update("populate_stats_process_rooms") + return 1 + + last_room_id = progress.get("last_room_id", "") + + def _get_next_batch(txn): + sql = """ + SELECT DISTINCT room_id FROM current_state_events + WHERE room_id > ? + ORDER BY room_id ASC + LIMIT ? + """ + txn.execute(sql, (last_room_id, batch_size)) + return [r for r, in txn] + + rooms_to_work_on = yield self.runInteraction( + "populate_stats_rooms_get_batch", _get_next_batch + ) + + # No more rooms -- complete the transaction. + if not rooms_to_work_on: + yield self._end_background_update("populate_stats_process_rooms") + return 1 + + for room_id in rooms_to_work_on: + yield self._calculate_and_set_initial_state_for_room(room_id) + progress["last_room_id"] = room_id + + yield self.runInteraction( + "_populate_stats_process_rooms", + self._background_update_progress_txn, + "populate_stats_process_rooms", + progress, + ) + + return len(rooms_to_work_on) + + def get_stats_positions(self): + """ + Returns the stats processor positions. + """ + return self._simple_select_one_onecol( + table="stats_incremental_position", + keyvalues={}, + retcol="stream_id", + desc="stats_incremental_position", + ) + + def update_room_state(self, room_id, fields): + """ + Args: + room_id (str) + fields (dict[str:Any]) + """ + + # For whatever reason some of the fields may contain null bytes, which + # postgres isn't a fan of, so we replace those fields with null. + for col in ( + "join_rules", + "history_visibility", + "encryption", + "name", + "topic", + "avatar", + "canonical_alias", + ): + field = fields.get(col) + if field and "\0" in field: + fields[col] = None + + return self._simple_upsert( + table="room_stats_state", + keyvalues={"room_id": room_id}, + values=fields, + desc="update_room_state", + ) + + def get_statistics_for_subject(self, stats_type, stats_id, start, size=100): + """ + Get statistics for a given subject. + + Args: + stats_type (str): The type of subject + stats_id (str): The ID of the subject (e.g. room_id or user_id) + start (int): Pagination start. Number of entries, not timestamp. + size (int): How many entries to return. + + Returns: + Deferred[list[dict]], where the dict has the keys of + ABSOLUTE_STATS_FIELDS[stats_type], and "bucket_size" and "end_ts". + """ + return self.runInteraction( + "get_statistics_for_subject", + self._get_statistics_for_subject_txn, + stats_type, + stats_id, + start, + size, + ) + + def _get_statistics_for_subject_txn( + self, txn, stats_type, stats_id, start, size=100 + ): + """ + Transaction-bound version of L{get_statistics_for_subject}. + """ + + table, id_col = TYPE_TO_TABLE[stats_type] + selected_columns = list( + ABSOLUTE_STATS_FIELDS[stats_type] + PER_SLICE_FIELDS[stats_type] + ) + + slice_list = self._simple_select_list_paginate_txn( + txn, + table + "_historical", + {id_col: stats_id}, + "end_ts", + start, + size, + retcols=selected_columns + ["bucket_size", "end_ts"], + order_direction="DESC", + ) + + return slice_list + + def get_room_stats_state(self, room_id): + """ + Returns the current room_stats_state for a room. + + Args: + room_id (str): The ID of the room to return state for. + + Returns (dict): + Dictionary containing these keys: + "name", "topic", "canonical_alias", "avatar", "join_rules", + "history_visibility" + """ + return self._simple_select_one( + "room_stats_state", + {"room_id": room_id}, + retcols=( + "name", + "topic", + "canonical_alias", + "avatar", + "join_rules", + "history_visibility", + ), + ) + + @cached() + def get_earliest_token_for_stats(self, stats_type, id): + """ + Fetch the "earliest token". This is used by the room stats delta + processor to ignore deltas that have been processed between the + start of the background task and any particular room's stats + being calculated. + + Returns: + Deferred[int] + """ + table, id_col = TYPE_TO_TABLE[stats_type] + + return self._simple_select_one_onecol( + "%s_current" % (table,), + keyvalues={id_col: id}, + retcol="completed_delta_stream_id", + allow_none=True, + ) + + def bulk_update_stats_delta(self, ts, updates, stream_id): + """Bulk update stats tables for a given stream_id and updates the stats + incremental position. + + Args: + ts (int): Current timestamp in ms + updates(dict[str, dict[str, dict[str, Counter]]]): The updates to + commit as a mapping stats_type -> stats_id -> field -> delta. + stream_id (int): Current position. + + Returns: + Deferred + """ + + def _bulk_update_stats_delta_txn(txn): + for stats_type, stats_updates in updates.items(): + for stats_id, fields in stats_updates.items(): + logger.info( + "Updating %s stats for %s: %s", stats_type, stats_id, fields + ) + self._update_stats_delta_txn( + txn, + ts=ts, + stats_type=stats_type, + stats_id=stats_id, + fields=fields, + complete_with_stream_id=stream_id, + ) + + self._simple_update_one_txn( + txn, + table="stats_incremental_position", + keyvalues={}, + updatevalues={"stream_id": stream_id}, + ) + + return self.runInteraction( + "bulk_update_stats_delta", _bulk_update_stats_delta_txn + ) + + def update_stats_delta( + self, + ts, + stats_type, + stats_id, + fields, + complete_with_stream_id, + absolute_field_overrides=None, + ): + """ + Updates the statistics for a subject, with a delta (difference/relative + change). + + Args: + ts (int): timestamp of the change + stats_type (str): "room" or "user" – the kind of subject + stats_id (str): the subject's ID (room ID or user ID) + fields (dict[str, int]): Deltas of stats values. + complete_with_stream_id (int, optional): + If supplied, converts an incomplete row into a complete row, + with the supplied stream_id marked as the stream_id where the + row was completed. + absolute_field_overrides (dict[str, int]): Current stats values + (i.e. not deltas) of absolute fields. + Does not work with per-slice fields. + """ + + return self.runInteraction( + "update_stats_delta", + self._update_stats_delta_txn, + ts, + stats_type, + stats_id, + fields, + complete_with_stream_id=complete_with_stream_id, + absolute_field_overrides=absolute_field_overrides, + ) + + def _update_stats_delta_txn( + self, + txn, + ts, + stats_type, + stats_id, + fields, + complete_with_stream_id, + absolute_field_overrides=None, + ): + if absolute_field_overrides is None: + absolute_field_overrides = {} + + table, id_col = TYPE_TO_TABLE[stats_type] + + quantised_ts = self.quantise_stats_time(int(ts)) + end_ts = quantised_ts + self.stats_bucket_size + + # Lets be paranoid and check that all the given field names are known + abs_field_names = ABSOLUTE_STATS_FIELDS[stats_type] + slice_field_names = PER_SLICE_FIELDS[stats_type] + for field in chain(fields.keys(), absolute_field_overrides.keys()): + if field not in abs_field_names and field not in slice_field_names: + # guard against potential SQL injection dodginess + raise ValueError( + "%s is not a recognised field" + " for stats type %s" % (field, stats_type) + ) + + # Per slice fields do not get added to the _current table + + # This calculates the deltas (`field = field + ?` values) + # for absolute fields, + # * defaulting to 0 if not specified + # (required for the INSERT part of upserting to work) + # * omitting overrides specified in `absolute_field_overrides` + deltas_of_absolute_fields = { + key: fields.get(key, 0) + for key in abs_field_names + if key not in absolute_field_overrides + } + + # Keep the delta stream ID field up to date + absolute_field_overrides = absolute_field_overrides.copy() + absolute_field_overrides["completed_delta_stream_id"] = complete_with_stream_id + + # first upsert the `_current` table + self._upsert_with_additive_relatives_txn( + txn=txn, + table=table + "_current", + keyvalues={id_col: stats_id}, + absolutes=absolute_field_overrides, + additive_relatives=deltas_of_absolute_fields, + ) + + per_slice_additive_relatives = { + key: fields.get(key, 0) for key in slice_field_names + } + self._upsert_copy_from_table_with_additive_relatives_txn( + txn=txn, + into_table=table + "_historical", + keyvalues={id_col: stats_id}, + extra_dst_insvalues={"bucket_size": self.stats_bucket_size}, + extra_dst_keyvalues={"end_ts": end_ts}, + additive_relatives=per_slice_additive_relatives, + src_table=table + "_current", + copy_columns=abs_field_names, + ) + + def _upsert_with_additive_relatives_txn( + self, txn, table, keyvalues, absolutes, additive_relatives + ): + """Used to update values in the stats tables. + + This is basically a slightly convoluted upsert that *adds* to any + existing rows. + + Args: + txn + table (str): Table name + keyvalues (dict[str, any]): Row-identifying key values + absolutes (dict[str, any]): Absolute (set) fields + additive_relatives (dict[str, int]): Fields that will be added onto + if existing row present. + """ + if self.database_engine.can_native_upsert: + absolute_updates = [ + "%(field)s = EXCLUDED.%(field)s" % {"field": field} + for field in absolutes.keys() + ] + + relative_updates = [ + "%(field)s = EXCLUDED.%(field)s + %(table)s.%(field)s" + % {"table": table, "field": field} + for field in additive_relatives.keys() + ] + + insert_cols = [] + qargs = [] + + for (key, val) in chain( + keyvalues.items(), absolutes.items(), additive_relatives.items() + ): + insert_cols.append(key) + qargs.append(val) + + sql = """ + INSERT INTO %(table)s (%(insert_cols_cs)s) + VALUES (%(insert_vals_qs)s) + ON CONFLICT (%(key_columns)s) DO UPDATE SET %(updates)s + """ % { + "table": table, + "insert_cols_cs": ", ".join(insert_cols), + "insert_vals_qs": ", ".join( + ["?"] * (len(keyvalues) + len(absolutes) + len(additive_relatives)) + ), + "key_columns": ", ".join(keyvalues), + "updates": ", ".join(chain(absolute_updates, relative_updates)), + } + + txn.execute(sql, qargs) + else: + self.database_engine.lock_table(txn, table) + retcols = list(chain(absolutes.keys(), additive_relatives.keys())) + current_row = self._simple_select_one_txn( + txn, table, keyvalues, retcols, allow_none=True + ) + if current_row is None: + merged_dict = {**keyvalues, **absolutes, **additive_relatives} + self._simple_insert_txn(txn, table, merged_dict) + else: + for (key, val) in additive_relatives.items(): + current_row[key] += val + current_row.update(absolutes) + self._simple_update_one_txn(txn, table, keyvalues, current_row) + + def _upsert_copy_from_table_with_additive_relatives_txn( + self, + txn, + into_table, + keyvalues, + extra_dst_keyvalues, + extra_dst_insvalues, + additive_relatives, + src_table, + copy_columns, + ): + """Updates the historic stats table with latest updates. + + This involves copying "absolute" fields from the `_current` table, and + adding relative fields to any existing values. + + Args: + txn: Transaction + into_table (str): The destination table to UPSERT the row into + keyvalues (dict[str, any]): Row-identifying key values + extra_dst_keyvalues (dict[str, any]): Additional keyvalues + for `into_table`. + extra_dst_insvalues (dict[str, any]): Additional values to insert + on new row creation for `into_table`. + additive_relatives (dict[str, any]): Fields that will be added onto + if existing row present. (Must be disjoint from copy_columns.) + src_table (str): The source table to copy from + copy_columns (iterable[str]): The list of columns to copy + """ + if self.database_engine.can_native_upsert: + ins_columns = chain( + keyvalues, + copy_columns, + additive_relatives, + extra_dst_keyvalues, + extra_dst_insvalues, + ) + sel_exprs = chain( + keyvalues, + copy_columns, + ( + "?" + for _ in chain( + additive_relatives, extra_dst_keyvalues, extra_dst_insvalues + ) + ), + ) + keyvalues_where = ("%s = ?" % f for f in keyvalues) + + sets_cc = ("%s = EXCLUDED.%s" % (f, f) for f in copy_columns) + sets_ar = ( + "%s = EXCLUDED.%s + %s.%s" % (f, f, into_table, f) + for f in additive_relatives + ) + + sql = """ + INSERT INTO %(into_table)s (%(ins_columns)s) + SELECT %(sel_exprs)s + FROM %(src_table)s + WHERE %(keyvalues_where)s + ON CONFLICT (%(keyvalues)s) + DO UPDATE SET %(sets)s + """ % { + "into_table": into_table, + "ins_columns": ", ".join(ins_columns), + "sel_exprs": ", ".join(sel_exprs), + "keyvalues_where": " AND ".join(keyvalues_where), + "src_table": src_table, + "keyvalues": ", ".join( + chain(keyvalues.keys(), extra_dst_keyvalues.keys()) + ), + "sets": ", ".join(chain(sets_cc, sets_ar)), + } + + qargs = list( + chain( + additive_relatives.values(), + extra_dst_keyvalues.values(), + extra_dst_insvalues.values(), + keyvalues.values(), + ) + ) + txn.execute(sql, qargs) + else: + self.database_engine.lock_table(txn, into_table) + src_row = self._simple_select_one_txn( + txn, src_table, keyvalues, copy_columns + ) + all_dest_keyvalues = {**keyvalues, **extra_dst_keyvalues} + dest_current_row = self._simple_select_one_txn( + txn, + into_table, + keyvalues=all_dest_keyvalues, + retcols=list(chain(additive_relatives.keys(), copy_columns)), + allow_none=True, + ) + + if dest_current_row is None: + merged_dict = { + **keyvalues, + **extra_dst_keyvalues, + **extra_dst_insvalues, + **src_row, + **additive_relatives, + } + self._simple_insert_txn(txn, into_table, merged_dict) + else: + for (key, val) in additive_relatives.items(): + src_row[key] = dest_current_row[key] + val + self._simple_update_txn(txn, into_table, all_dest_keyvalues, src_row) + + def get_changes_room_total_events_and_bytes(self, min_pos, max_pos): + """Fetches the counts of events in the given range of stream IDs. + + Args: + min_pos (int) + max_pos (int) + + Returns: + Deferred[dict[str, dict[str, int]]]: Mapping of room ID to field + changes. + """ + + return self.runInteraction( + "stats_incremental_total_events_and_bytes", + self.get_changes_room_total_events_and_bytes_txn, + min_pos, + max_pos, + ) + + def get_changes_room_total_events_and_bytes_txn(self, txn, low_pos, high_pos): + """Gets the total_events and total_event_bytes counts for rooms and + senders, in a range of stream_orderings (including backfilled events). + + Args: + txn + low_pos (int): Low stream ordering + high_pos (int): High stream ordering + + Returns: + tuple[dict[str, dict[str, int]], dict[str, dict[str, int]]]: The + room and user deltas for total_events/total_event_bytes in the + format of `stats_id` -> fields + """ + + if low_pos >= high_pos: + # nothing to do here. + return {}, {} + + if isinstance(self.database_engine, PostgresEngine): + new_bytes_expression = "OCTET_LENGTH(json)" + else: + new_bytes_expression = "LENGTH(CAST(json AS BLOB))" + + sql = """ + SELECT events.room_id, COUNT(*) AS new_events, SUM(%s) AS new_bytes + FROM events INNER JOIN event_json USING (event_id) + WHERE (? < stream_ordering AND stream_ordering <= ?) + OR (? <= stream_ordering AND stream_ordering <= ?) + GROUP BY events.room_id + """ % ( + new_bytes_expression, + ) + + txn.execute(sql, (low_pos, high_pos, -high_pos, -low_pos)) + + room_deltas = { + room_id: {"total_events": new_events, "total_event_bytes": new_bytes} + for room_id, new_events, new_bytes in txn + } + + sql = """ + SELECT events.sender, COUNT(*) AS new_events, SUM(%s) AS new_bytes + FROM events INNER JOIN event_json USING (event_id) + WHERE (? < stream_ordering AND stream_ordering <= ?) + OR (? <= stream_ordering AND stream_ordering <= ?) + GROUP BY events.sender + """ % ( + new_bytes_expression, + ) + + txn.execute(sql, (low_pos, high_pos, -high_pos, -low_pos)) + + user_deltas = { + user_id: {"total_events": new_events, "total_event_bytes": new_bytes} + for user_id, new_events, new_bytes in txn + if self.hs.is_mine_id(user_id) + } + + return room_deltas, user_deltas + + @defer.inlineCallbacks + def _calculate_and_set_initial_state_for_room(self, room_id): + """Calculate and insert an entry into room_stats_current. + + Args: + room_id (str) + + Returns: + Deferred[tuple[dict, dict, int]]: A tuple of room state, membership + counts and stream position. + """ + + def _fetch_current_state_stats(txn): + pos = self.get_room_max_stream_ordering() + + rows = self._simple_select_many_txn( + txn, + table="current_state_events", + column="type", + iterable=[ + EventTypes.Create, + EventTypes.JoinRules, + EventTypes.RoomHistoryVisibility, + EventTypes.Encryption, + EventTypes.Name, + EventTypes.Topic, + EventTypes.RoomAvatar, + EventTypes.CanonicalAlias, + ], + keyvalues={"room_id": room_id, "state_key": ""}, + retcols=["event_id"], + ) + + event_ids = [row["event_id"] for row in rows] + + txn.execute( + """ + SELECT membership, count(*) FROM current_state_events + WHERE room_id = ? AND type = 'm.room.member' + GROUP BY membership + """, + (room_id,), + ) + membership_counts = {membership: cnt for membership, cnt in txn} + + txn.execute( + """ + SELECT COALESCE(count(*), 0) FROM current_state_events + WHERE room_id = ? + """, + (room_id,), + ) + + current_state_events_count, = txn.fetchone() + + users_in_room = self.get_users_in_room_txn(txn, room_id) + + return ( + event_ids, + membership_counts, + current_state_events_count, + users_in_room, + pos, + ) + + ( + event_ids, + membership_counts, + current_state_events_count, + users_in_room, + pos, + ) = yield self.runInteraction( + "get_initial_state_for_room", _fetch_current_state_stats + ) + + state_event_map = yield self.get_events(event_ids, get_prev_content=False) + + room_state = { + "join_rules": None, + "history_visibility": None, + "encryption": None, + "name": None, + "topic": None, + "avatar": None, + "canonical_alias": None, + "is_federatable": True, + } + + for event in state_event_map.values(): + if event.type == EventTypes.JoinRules: + room_state["join_rules"] = event.content.get("join_rule") + elif event.type == EventTypes.RoomHistoryVisibility: + room_state["history_visibility"] = event.content.get( + "history_visibility" + ) + elif event.type == EventTypes.Encryption: + room_state["encryption"] = event.content.get("algorithm") + elif event.type == EventTypes.Name: + room_state["name"] = event.content.get("name") + elif event.type == EventTypes.Topic: + room_state["topic"] = event.content.get("topic") + elif event.type == EventTypes.RoomAvatar: + room_state["avatar"] = event.content.get("url") + elif event.type == EventTypes.CanonicalAlias: + room_state["canonical_alias"] = event.content.get("alias") + elif event.type == EventTypes.Create: + room_state["is_federatable"] = ( + event.content.get("m.federate", True) is True + ) + + yield self.update_room_state(room_id, room_state) + + local_users_in_room = [u for u in users_in_room if self.hs.is_mine_id(u)] + + yield self.update_stats_delta( + ts=self.clock.time_msec(), + stats_type="room", + stats_id=room_id, + fields={}, + complete_with_stream_id=pos, + absolute_field_overrides={ + "current_state_events": current_state_events_count, + "joined_members": membership_counts.get(Membership.JOIN, 0), + "invited_members": membership_counts.get(Membership.INVITE, 0), + "left_members": membership_counts.get(Membership.LEAVE, 0), + "banned_members": membership_counts.get(Membership.BAN, 0), + "local_users_in_room": len(local_users_in_room), + }, + ) + + @defer.inlineCallbacks + def _calculate_and_set_initial_state_for_user(self, user_id): + def _calculate_and_set_initial_state_for_user_txn(txn): + pos = self._get_max_stream_id_in_current_state_deltas_txn(txn) + + txn.execute( + """ + SELECT COUNT(distinct room_id) FROM current_state_events + WHERE type = 'm.room.member' AND state_key = ? + AND membership = 'join' + """, + (user_id,), + ) + count, = txn.fetchone() + return count, pos + + joined_rooms, pos = yield self.runInteraction( + "calculate_and_set_initial_state_for_user", + _calculate_and_set_initial_state_for_user_txn, + ) + + yield self.update_stats_delta( + ts=self.clock.time_msec(), + stats_type="user", + stats_id=user_id, + fields={}, + complete_with_stream_id=pos, + absolute_field_overrides={"joined_rooms": joined_rooms}, + ) diff --git a/synapse/storage/data_stores/main/stream.py b/synapse/storage/data_stores/main/stream.py new file mode 100644 index 0000000000..263999dfca --- /dev/null +++ b/synapse/storage/data_stores/main/stream.py @@ -0,0 +1,948 @@ +# -*- coding: utf-8 -*- +# Copyright 2014-2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +""" This module is responsible for getting events from the DB for pagination +and event streaming. + +The order it returns events in depend on whether we are streaming forwards or +are paginating backwards. We do this because we want to handle out of order +messages nicely, while still returning them in the correct order when we +paginate bacwards. + +This is implemented by keeping two ordering columns: stream_ordering and +topological_ordering. Stream ordering is basically insertion/received order +(except for events from backfill requests). The topological_ordering is a +weak ordering of events based on the pdu graph. + +This means that we have to have two different types of tokens, depending on +what sort order was used: + - stream tokens are of the form: "s%d", which maps directly to the column + - topological tokems: "t%d-%d", where the integers map to the topological + and stream ordering columns respectively. +""" + +import abc +import logging +from collections import namedtuple + +from six.moves import range + +from twisted.internet import defer + +from synapse.logging.context import make_deferred_yieldable, run_in_background +from synapse.storage._base import SQLBaseStore +from synapse.storage.data_stores.main.events_worker import EventsWorkerStore +from synapse.storage.engines import PostgresEngine +from synapse.types import RoomStreamToken +from synapse.util.caches.stream_change_cache import StreamChangeCache + +logger = logging.getLogger(__name__) + + +MAX_STREAM_SIZE = 1000 + + +_STREAM_TOKEN = "stream" +_TOPOLOGICAL_TOKEN = "topological" + + +# Used as return values for pagination APIs +_EventDictReturn = namedtuple( + "_EventDictReturn", ("event_id", "topological_ordering", "stream_ordering") +) + + +def generate_pagination_where_clause( + direction, column_names, from_token, to_token, engine +): + """Creates an SQL expression to bound the columns by the pagination + tokens. + + For example creates an SQL expression like: + + (6, 7) >= (topological_ordering, stream_ordering) + AND (5, 3) < (topological_ordering, stream_ordering) + + would be generated for dir=b, from_token=(6, 7) and to_token=(5, 3). + + Note that tokens are considered to be after the row they are in, e.g. if + a row A has a token T, then we consider A to be before T. This convention + is important when figuring out inequalities for the generated SQL, and + produces the following result: + - If paginating forwards then we exclude any rows matching the from + token, but include those that match the to token. + - If paginating backwards then we include any rows matching the from + token, but include those that match the to token. + + Args: + direction (str): Whether we're paginating backwards("b") or + forwards ("f"). + column_names (tuple[str, str]): The column names to bound. Must *not* + be user defined as these get inserted directly into the SQL + statement without escapes. + from_token (tuple[int, int]|None): The start point for the pagination. + This is an exclusive minimum bound if direction is "f", and an + inclusive maximum bound if direction is "b". + to_token (tuple[int, int]|None): The endpoint point for the pagination. + This is an inclusive maximum bound if direction is "f", and an + exclusive minimum bound if direction is "b". + engine: The database engine to generate the clauses for + + Returns: + str: The sql expression + """ + assert direction in ("b", "f") + + where_clause = [] + if from_token: + where_clause.append( + _make_generic_sql_bound( + bound=">=" if direction == "b" else "<", + column_names=column_names, + values=from_token, + engine=engine, + ) + ) + + if to_token: + where_clause.append( + _make_generic_sql_bound( + bound="<" if direction == "b" else ">=", + column_names=column_names, + values=to_token, + engine=engine, + ) + ) + + return " AND ".join(where_clause) + + +def _make_generic_sql_bound(bound, column_names, values, engine): + """Create an SQL expression that bounds the given column names by the + values, e.g. create the equivalent of `(1, 2) < (col1, col2)`. + + Only works with two columns. + + Older versions of SQLite don't support that syntax so we have to expand it + out manually. + + Args: + bound (str): The comparison operator to use. One of ">", "<", ">=", + "<=", where the values are on the left and columns on the right. + names (tuple[str, str]): The column names. Must *not* be user defined + as these get inserted directly into the SQL statement without + escapes. + values (tuple[int|None, int]): The values to bound the columns by. If + the first value is None then only creates a bound on the second + column. + engine: The database engine to generate the SQL for + + Returns: + str + """ + + assert bound in (">", "<", ">=", "<=") + + name1, name2 = column_names + val1, val2 = values + + if val1 is None: + val2 = int(val2) + return "(%d %s %s)" % (val2, bound, name2) + + val1 = int(val1) + val2 = int(val2) + + if isinstance(engine, PostgresEngine): + # Postgres doesn't optimise ``(x < a) OR (x=a AND y ? AND stream_ordering <= ?" + " ORDER BY stream_ordering %s LIMIT ?" + ) % (order,) + txn.execute(sql, (room_id, from_id, to_id, limit)) + + rows = [_EventDictReturn(row[0], None, row[1]) for row in txn] + return rows + + rows = yield self.runInteraction("get_room_events_stream_for_room", f) + + ret = yield self.get_events_as_list( + [r.event_id for r in rows], get_prev_content=True + ) + + self._set_before_and_after(ret, rows, topo_order=from_id is None) + + if order.lower() == "desc": + ret.reverse() + + if rows: + key = "s%d" % min(r.stream_ordering for r in rows) + else: + # Assume we didn't get anything because there was nothing to + # get. + key = from_key + + return ret, key + + @defer.inlineCallbacks + def get_membership_changes_for_user(self, user_id, from_key, to_key): + from_id = RoomStreamToken.parse_stream_token(from_key).stream + to_id = RoomStreamToken.parse_stream_token(to_key).stream + + if from_key == to_key: + return [] + + if from_id: + has_changed = self._membership_stream_cache.has_entity_changed( + user_id, int(from_id) + ) + if not has_changed: + return [] + + def f(txn): + sql = ( + "SELECT m.event_id, stream_ordering FROM events AS e," + " room_memberships AS m" + " WHERE e.event_id = m.event_id" + " AND m.user_id = ?" + " AND e.stream_ordering > ? AND e.stream_ordering <= ?" + " ORDER BY e.stream_ordering ASC" + ) + txn.execute(sql, (user_id, from_id, to_id)) + + rows = [_EventDictReturn(row[0], None, row[1]) for row in txn] + + return rows + + rows = yield self.runInteraction("get_membership_changes_for_user", f) + + ret = yield self.get_events_as_list( + [r.event_id for r in rows], get_prev_content=True + ) + + self._set_before_and_after(ret, rows, topo_order=False) + + return ret + + @defer.inlineCallbacks + def get_recent_events_for_room(self, room_id, limit, end_token): + """Get the most recent events in the room in topological ordering. + + Args: + room_id (str) + limit (int) + end_token (str): The stream token representing now. + + Returns: + Deferred[tuple[list[FrozenEvent], str]]: Returns a list of + events and a token pointing to the start of the returned + events. + The events returned are in ascending order. + """ + + rows, token = yield self.get_recent_event_ids_for_room( + room_id, limit, end_token + ) + + logger.debug("stream before") + events = yield self.get_events_as_list( + [r.event_id for r in rows], get_prev_content=True + ) + logger.debug("stream after") + + self._set_before_and_after(events, rows) + + return (events, token) + + @defer.inlineCallbacks + def get_recent_event_ids_for_room(self, room_id, limit, end_token): + """Get the most recent events in the room in topological ordering. + + Args: + room_id (str) + limit (int) + end_token (str): The stream token representing now. + + Returns: + Deferred[tuple[list[_EventDictReturn], str]]: Returns a list of + _EventDictReturn and a token pointing to the start of the returned + events. + The events returned are in ascending order. + """ + # Allow a zero limit here, and no-op. + if limit == 0: + return [], end_token + + end_token = RoomStreamToken.parse(end_token) + + rows, token = yield self.runInteraction( + "get_recent_event_ids_for_room", + self._paginate_room_events_txn, + room_id, + from_token=end_token, + limit=limit, + ) + + # We want to return the results in ascending order. + rows.reverse() + + return rows, token + + def get_room_event_after_stream_ordering(self, room_id, stream_ordering): + """Gets details of the first event in a room at or after a stream ordering + + Args: + room_id (str): + stream_ordering (int): + + Returns: + Deferred[(int, int, str)]: + (stream ordering, topological ordering, event_id) + """ + + def _f(txn): + sql = ( + "SELECT stream_ordering, topological_ordering, event_id" + " FROM events" + " WHERE room_id = ? AND stream_ordering >= ?" + " AND NOT outlier" + " ORDER BY stream_ordering" + " LIMIT 1" + ) + txn.execute(sql, (room_id, stream_ordering)) + return txn.fetchone() + + return self.runInteraction("get_room_event_after_stream_ordering", _f) + + @defer.inlineCallbacks + def get_room_events_max_id(self, room_id=None): + """Returns the current token for rooms stream. + + By default, it returns the current global stream token. Specifying a + `room_id` causes it to return the current room specific topological + token. + """ + token = yield self.get_room_max_stream_ordering() + if room_id is None: + return "s%d" % (token,) + else: + topo = yield self.runInteraction( + "_get_max_topological_txn", self._get_max_topological_txn, room_id + ) + return "t%d-%d" % (topo, token) + + def get_stream_token_for_event(self, event_id): + """The stream token for an event + Args: + event_id(str): The id of the event to look up a stream token for. + Raises: + StoreError if the event wasn't in the database. + Returns: + A deferred "s%d" stream token. + """ + return self._simple_select_one_onecol( + table="events", keyvalues={"event_id": event_id}, retcol="stream_ordering" + ).addCallback(lambda row: "s%d" % (row,)) + + def get_topological_token_for_event(self, event_id): + """The stream token for an event + Args: + event_id(str): The id of the event to look up a stream token for. + Raises: + StoreError if the event wasn't in the database. + Returns: + A deferred "t%d-%d" topological token. + """ + return self._simple_select_one( + table="events", + keyvalues={"event_id": event_id}, + retcols=("stream_ordering", "topological_ordering"), + desc="get_topological_token_for_event", + ).addCallback( + lambda row: "t%d-%d" % (row["topological_ordering"], row["stream_ordering"]) + ) + + def get_max_topological_token(self, room_id, stream_key): + """Get the max topological token in a room before the given stream + ordering. + + Args: + room_id (str) + stream_key (int) + + Returns: + Deferred[int] + """ + sql = ( + "SELECT coalesce(max(topological_ordering), 0) FROM events" + " WHERE room_id = ? AND stream_ordering < ?" + ) + return self._execute( + "get_max_topological_token", None, sql, room_id, stream_key + ).addCallback(lambda r: r[0][0] if r else 0) + + def _get_max_topological_txn(self, txn, room_id): + txn.execute( + "SELECT MAX(topological_ordering) FROM events" " WHERE room_id = ?", + (room_id,), + ) + + rows = txn.fetchall() + return rows[0][0] if rows else 0 + + @staticmethod + def _set_before_and_after(events, rows, topo_order=True): + """Inserts ordering information to events' internal metadata from + the DB rows. + + Args: + events (list[FrozenEvent]) + rows (list[_EventDictReturn]) + topo_order (bool): Whether the events were ordered topologically + or by stream ordering. If true then all rows should have a non + null topological_ordering. + """ + for event, row in zip(events, rows): + stream = row.stream_ordering + if topo_order and row.topological_ordering: + topo = row.topological_ordering + else: + topo = None + internal = event.internal_metadata + internal.before = str(RoomStreamToken(topo, stream - 1)) + internal.after = str(RoomStreamToken(topo, stream)) + internal.order = (int(topo) if topo else 0, int(stream)) + + @defer.inlineCallbacks + def get_events_around( + self, room_id, event_id, before_limit, after_limit, event_filter=None + ): + """Retrieve events and pagination tokens around a given event in a + room. + + Args: + room_id (str) + event_id (str) + before_limit (int) + after_limit (int) + event_filter (Filter|None) + + Returns: + dict + """ + + results = yield self.runInteraction( + "get_events_around", + self._get_events_around_txn, + room_id, + event_id, + before_limit, + after_limit, + event_filter, + ) + + events_before = yield self.get_events_as_list( + [e for e in results["before"]["event_ids"]], get_prev_content=True + ) + + events_after = yield self.get_events_as_list( + [e for e in results["after"]["event_ids"]], get_prev_content=True + ) + + return { + "events_before": events_before, + "events_after": events_after, + "start": results["before"]["token"], + "end": results["after"]["token"], + } + + def _get_events_around_txn( + self, txn, room_id, event_id, before_limit, after_limit, event_filter + ): + """Retrieves event_ids and pagination tokens around a given event in a + room. + + Args: + room_id (str) + event_id (str) + before_limit (int) + after_limit (int) + event_filter (Filter|None) + + Returns: + dict + """ + + results = self._simple_select_one_txn( + txn, + "events", + keyvalues={"event_id": event_id, "room_id": room_id}, + retcols=["stream_ordering", "topological_ordering"], + ) + + # Paginating backwards includes the event at the token, but paginating + # forward doesn't. + before_token = RoomStreamToken( + results["topological_ordering"] - 1, results["stream_ordering"] + ) + + after_token = RoomStreamToken( + results["topological_ordering"], results["stream_ordering"] + ) + + rows, start_token = self._paginate_room_events_txn( + txn, + room_id, + before_token, + direction="b", + limit=before_limit, + event_filter=event_filter, + ) + events_before = [r.event_id for r in rows] + + rows, end_token = self._paginate_room_events_txn( + txn, + room_id, + after_token, + direction="f", + limit=after_limit, + event_filter=event_filter, + ) + events_after = [r.event_id for r in rows] + + return { + "before": {"event_ids": events_before, "token": start_token}, + "after": {"event_ids": events_after, "token": end_token}, + } + + @defer.inlineCallbacks + def get_all_new_events_stream(self, from_id, current_id, limit): + """Get all new events + + Returns all events with from_id < stream_ordering <= current_id. + + Args: + from_id (int): the stream_ordering of the last event we processed + current_id (int): the stream_ordering of the most recently processed event + limit (int): the maximum number of events to return + + Returns: + Deferred[Tuple[int, list[FrozenEvent]]]: A tuple of (next_id, events), where + `next_id` is the next value to pass as `from_id` (it will either be the + stream_ordering of the last returned event, or, if fewer than `limit` events + were found, `current_id`. + """ + + def get_all_new_events_stream_txn(txn): + sql = ( + "SELECT e.stream_ordering, e.event_id" + " FROM events AS e" + " WHERE" + " ? < e.stream_ordering AND e.stream_ordering <= ?" + " ORDER BY e.stream_ordering ASC" + " LIMIT ?" + ) + + txn.execute(sql, (from_id, current_id, limit)) + rows = txn.fetchall() + + upper_bound = current_id + if len(rows) == limit: + upper_bound = rows[-1][0] + + return upper_bound, [row[1] for row in rows] + + upper_bound, event_ids = yield self.runInteraction( + "get_all_new_events_stream", get_all_new_events_stream_txn + ) + + events = yield self.get_events_as_list(event_ids) + + return upper_bound, events + + def get_federation_out_pos(self, typ): + return self._simple_select_one_onecol( + table="federation_stream_position", + retcol="stream_id", + keyvalues={"type": typ}, + desc="get_federation_out_pos", + ) + + def update_federation_out_pos(self, typ, stream_id): + return self._simple_update_one( + table="federation_stream_position", + keyvalues={"type": typ}, + updatevalues={"stream_id": stream_id}, + desc="update_federation_out_pos", + ) + + def has_room_changed_since(self, room_id, stream_id): + return self._events_stream_cache.has_entity_changed(room_id, stream_id) + + def _paginate_room_events_txn( + self, + txn, + room_id, + from_token, + to_token=None, + direction="b", + limit=-1, + event_filter=None, + ): + """Returns list of events before or after a given token. + + Args: + txn + room_id (str) + from_token (RoomStreamToken): The token used to stream from + to_token (RoomStreamToken|None): A token which if given limits the + results to only those before + direction(char): Either 'b' or 'f' to indicate whether we are + paginating forwards or backwards from `from_key`. + limit (int): The maximum number of events to return. + event_filter (Filter|None): If provided filters the events to + those that match the filter. + + Returns: + Deferred[tuple[list[_EventDictReturn], str]]: Returns the results + as a list of _EventDictReturn and a token that points to the end + of the result set. If no events are returned then the end of the + stream has been reached (i.e. there are no events between + `from_token` and `to_token`), or `limit` is zero. + """ + + assert int(limit) >= 0 + + # Tokens really represent positions between elements, but we use + # the convention of pointing to the event before the gap. Hence + # we have a bit of asymmetry when it comes to equalities. + args = [False, room_id] + if direction == "b": + order = "DESC" + else: + order = "ASC" + + bounds = generate_pagination_where_clause( + direction=direction, + column_names=("topological_ordering", "stream_ordering"), + from_token=from_token, + to_token=to_token, + engine=self.database_engine, + ) + + filter_clause, filter_args = filter_to_clause(event_filter) + + if filter_clause: + bounds += " AND " + filter_clause + args.extend(filter_args) + + args.append(int(limit)) + + sql = ( + "SELECT event_id, topological_ordering, stream_ordering" + " FROM events" + " WHERE outlier = ? AND room_id = ? AND %(bounds)s" + " ORDER BY topological_ordering %(order)s," + " stream_ordering %(order)s LIMIT ?" + ) % {"bounds": bounds, "order": order} + + txn.execute(sql, args) + + rows = [_EventDictReturn(row[0], row[1], row[2]) for row in txn] + + if rows: + topo = rows[-1].topological_ordering + toke = rows[-1].stream_ordering + if direction == "b": + # Tokens are positions between events. + # This token points *after* the last event in the chunk. + # We need it to point to the event before it in the chunk + # when we are going backwards so we subtract one from the + # stream part. + toke -= 1 + next_token = RoomStreamToken(topo, toke) + else: + # TODO (erikj): We should work out what to do here instead. + next_token = to_token if to_token else from_token + + return rows, str(next_token) + + @defer.inlineCallbacks + def paginate_room_events( + self, room_id, from_key, to_key=None, direction="b", limit=-1, event_filter=None + ): + """Returns list of events before or after a given token. + + Args: + room_id (str) + from_key (str): The token used to stream from + to_key (str|None): A token which if given limits the results to + only those before + direction(char): Either 'b' or 'f' to indicate whether we are + paginating forwards or backwards from `from_key`. + limit (int): The maximum number of events to return. + event_filter (Filter|None): If provided filters the events to + those that match the filter. + + Returns: + tuple[list[FrozenEvent], str]: Returns the results as a list of + events and a token that points to the end of the result set. If no + events are returned then the end of the stream has been reached + (i.e. there are no events between `from_key` and `to_key`). + """ + + from_key = RoomStreamToken.parse(from_key) + if to_key: + to_key = RoomStreamToken.parse(to_key) + + rows, token = yield self.runInteraction( + "paginate_room_events", + self._paginate_room_events_txn, + room_id, + from_key, + to_key, + direction, + limit, + event_filter, + ) + + events = yield self.get_events_as_list( + [r.event_id for r in rows], get_prev_content=True + ) + + self._set_before_and_after(events, rows) + + return (events, token) + + +class StreamStore(StreamWorkerStore): + def get_room_max_stream_ordering(self): + return self._stream_id_gen.get_current_token() + + def get_room_min_stream_ordering(self): + return self._backfill_id_gen.get_current_token() diff --git a/synapse/storage/data_stores/main/tags.py b/synapse/storage/data_stores/main/tags.py new file mode 100644 index 0000000000..10d1887f75 --- /dev/null +++ b/synapse/storage/data_stores/main/tags.py @@ -0,0 +1,265 @@ +# -*- coding: utf-8 -*- +# Copyright 2014-2016 OpenMarket Ltd +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging + +from six.moves import range + +from canonicaljson import json + +from twisted.internet import defer + +from synapse.storage.data_stores.main.account_data import AccountDataWorkerStore +from synapse.util.caches.descriptors import cached + +logger = logging.getLogger(__name__) + + +class TagsWorkerStore(AccountDataWorkerStore): + @cached() + def get_tags_for_user(self, user_id): + """Get all the tags for a user. + + + Args: + user_id(str): The user to get the tags for. + Returns: + A deferred dict mapping from room_id strings to dicts mapping from + tag strings to tag content. + """ + + deferred = self._simple_select_list( + "room_tags", {"user_id": user_id}, ["room_id", "tag", "content"] + ) + + @deferred.addCallback + def tags_by_room(rows): + tags_by_room = {} + for row in rows: + room_tags = tags_by_room.setdefault(row["room_id"], {}) + room_tags[row["tag"]] = json.loads(row["content"]) + return tags_by_room + + return deferred + + @defer.inlineCallbacks + def get_all_updated_tags(self, last_id, current_id, limit): + """Get all the client tags that have changed on the server + Args: + last_id(int): The position to fetch from. + current_id(int): The position to fetch up to. + Returns: + A deferred list of tuples of stream_id int, user_id string, + room_id string, tag string and content string. + """ + if last_id == current_id: + return [] + + def get_all_updated_tags_txn(txn): + sql = ( + "SELECT stream_id, user_id, room_id" + " FROM room_tags_revisions as r" + " WHERE ? < stream_id AND stream_id <= ?" + " ORDER BY stream_id ASC LIMIT ?" + ) + txn.execute(sql, (last_id, current_id, limit)) + return txn.fetchall() + + tag_ids = yield self.runInteraction( + "get_all_updated_tags", get_all_updated_tags_txn + ) + + def get_tag_content(txn, tag_ids): + sql = ( + "SELECT tag, content" " FROM room_tags" " WHERE user_id=? AND room_id=?" + ) + results = [] + for stream_id, user_id, room_id in tag_ids: + txn.execute(sql, (user_id, room_id)) + tags = [] + for tag, content in txn: + tags.append(json.dumps(tag) + ":" + content) + tag_json = "{" + ",".join(tags) + "}" + results.append((stream_id, user_id, room_id, tag_json)) + + return results + + batch_size = 50 + results = [] + for i in range(0, len(tag_ids), batch_size): + tags = yield self.runInteraction( + "get_all_updated_tag_content", + get_tag_content, + tag_ids[i : i + batch_size], + ) + results.extend(tags) + + return results + + @defer.inlineCallbacks + def get_updated_tags(self, user_id, stream_id): + """Get all the tags for the rooms where the tags have changed since the + given version + + Args: + user_id(str): The user to get the tags for. + stream_id(int): The earliest update to get for the user. + Returns: + A deferred dict mapping from room_id strings to lists of tag + strings for all the rooms that changed since the stream_id token. + """ + + def get_updated_tags_txn(txn): + sql = ( + "SELECT room_id from room_tags_revisions" + " WHERE user_id = ? AND stream_id > ?" + ) + txn.execute(sql, (user_id, stream_id)) + room_ids = [row[0] for row in txn] + return room_ids + + changed = self._account_data_stream_cache.has_entity_changed( + user_id, int(stream_id) + ) + if not changed: + return {} + + room_ids = yield self.runInteraction("get_updated_tags", get_updated_tags_txn) + + results = {} + if room_ids: + tags_by_room = yield self.get_tags_for_user(user_id) + for room_id in room_ids: + results[room_id] = tags_by_room.get(room_id, {}) + + return results + + def get_tags_for_room(self, user_id, room_id): + """Get all the tags for the given room + Args: + user_id(str): The user to get tags for + room_id(str): The room to get tags for + Returns: + A deferred list of string tags. + """ + return self._simple_select_list( + table="room_tags", + keyvalues={"user_id": user_id, "room_id": room_id}, + retcols=("tag", "content"), + desc="get_tags_for_room", + ).addCallback( + lambda rows: {row["tag"]: json.loads(row["content"]) for row in rows} + ) + + +class TagsStore(TagsWorkerStore): + @defer.inlineCallbacks + def add_tag_to_room(self, user_id, room_id, tag, content): + """Add a tag to a room for a user. + Args: + user_id(str): The user to add a tag for. + room_id(str): The room to add a tag for. + tag(str): The tag name to add. + content(dict): A json object to associate with the tag. + Returns: + A deferred that completes once the tag has been added. + """ + content_json = json.dumps(content) + + def add_tag_txn(txn, next_id): + self._simple_upsert_txn( + txn, + table="room_tags", + keyvalues={"user_id": user_id, "room_id": room_id, "tag": tag}, + values={"content": content_json}, + ) + self._update_revision_txn(txn, user_id, room_id, next_id) + + with self._account_data_id_gen.get_next() as next_id: + yield self.runInteraction("add_tag", add_tag_txn, next_id) + + self.get_tags_for_user.invalidate((user_id,)) + + result = self._account_data_id_gen.get_current_token() + return result + + @defer.inlineCallbacks + def remove_tag_from_room(self, user_id, room_id, tag): + """Remove a tag from a room for a user. + Returns: + A deferred that completes once the tag has been removed + """ + + def remove_tag_txn(txn, next_id): + sql = ( + "DELETE FROM room_tags " + " WHERE user_id = ? AND room_id = ? AND tag = ?" + ) + txn.execute(sql, (user_id, room_id, tag)) + self._update_revision_txn(txn, user_id, room_id, next_id) + + with self._account_data_id_gen.get_next() as next_id: + yield self.runInteraction("remove_tag", remove_tag_txn, next_id) + + self.get_tags_for_user.invalidate((user_id,)) + + result = self._account_data_id_gen.get_current_token() + return result + + def _update_revision_txn(self, txn, user_id, room_id, next_id): + """Update the latest revision of the tags for the given user and room. + + Args: + txn: The database cursor + user_id(str): The ID of the user. + room_id(str): The ID of the room. + next_id(int): The the revision to advance to. + """ + + txn.call_after( + self._account_data_stream_cache.entity_has_changed, user_id, next_id + ) + + update_max_id_sql = ( + "UPDATE account_data_max_stream_id" + " SET stream_id = ?" + " WHERE stream_id < ?" + ) + txn.execute(update_max_id_sql, (next_id, next_id)) + + update_sql = ( + "UPDATE room_tags_revisions" + " SET stream_id = ?" + " WHERE user_id = ?" + " AND room_id = ?" + ) + txn.execute(update_sql, (next_id, user_id, room_id)) + + if txn.rowcount == 0: + insert_sql = ( + "INSERT INTO room_tags_revisions (user_id, room_id, stream_id)" + " VALUES (?, ?, ?)" + ) + try: + txn.execute(insert_sql, (user_id, room_id, next_id)) + except self.database_engine.module.IntegrityError: + # Ignore insertion errors. It doesn't matter if the row wasn't + # inserted because if two updates happend concurrently the one + # with the higher stream_id will not be reported to a client + # unless the previous update has completed. It doesn't matter + # which stream_id ends up in the table, as long as it is higher + # than the id that the client has. + pass diff --git a/synapse/storage/data_stores/main/transactions.py b/synapse/storage/data_stores/main/transactions.py new file mode 100644 index 0000000000..01b1be5e14 --- /dev/null +++ b/synapse/storage/data_stores/main/transactions.py @@ -0,0 +1,273 @@ +# -*- coding: utf-8 -*- +# Copyright 2014-2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging +from collections import namedtuple + +import six + +from canonicaljson import encode_canonical_json + +from twisted.internet import defer + +from synapse.metrics.background_process_metrics import run_as_background_process +from synapse.storage._base import SQLBaseStore, db_to_json +from synapse.util.caches.expiringcache import ExpiringCache + +# py2 sqlite has buffer hardcoded as only binary type, so we must use it, +# despite being deprecated and removed in favor of memoryview +if six.PY2: + db_binary_type = six.moves.builtins.buffer +else: + db_binary_type = memoryview + +logger = logging.getLogger(__name__) + + +_TransactionRow = namedtuple( + "_TransactionRow", + ("id", "transaction_id", "destination", "ts", "response_code", "response_json"), +) + +_UpdateTransactionRow = namedtuple( + "_TransactionRow", ("response_code", "response_json") +) + +SENTINEL = object() + + +class TransactionStore(SQLBaseStore): + """A collection of queries for handling PDUs. + """ + + def __init__(self, db_conn, hs): + super(TransactionStore, self).__init__(db_conn, hs) + + self._clock.looping_call(self._start_cleanup_transactions, 30 * 60 * 1000) + + self._destination_retry_cache = ExpiringCache( + cache_name="get_destination_retry_timings", + clock=self._clock, + expiry_ms=5 * 60 * 1000, + ) + + def get_received_txn_response(self, transaction_id, origin): + """For an incoming transaction from a given origin, check if we have + already responded to it. If so, return the response code and response + body (as a dict). + + Args: + transaction_id (str) + origin(str) + + Returns: + tuple: None if we have not previously responded to + this transaction or a 2-tuple of (int, dict) + """ + + return self.runInteraction( + "get_received_txn_response", + self._get_received_txn_response, + transaction_id, + origin, + ) + + def _get_received_txn_response(self, txn, transaction_id, origin): + result = self._simple_select_one_txn( + txn, + table="received_transactions", + keyvalues={"transaction_id": transaction_id, "origin": origin}, + retcols=( + "transaction_id", + "origin", + "ts", + "response_code", + "response_json", + "has_been_referenced", + ), + allow_none=True, + ) + + if result and result["response_code"]: + return result["response_code"], db_to_json(result["response_json"]) + + else: + return None + + def set_received_txn_response(self, transaction_id, origin, code, response_dict): + """Persist the response we returened for an incoming transaction, and + should return for subsequent transactions with the same transaction_id + and origin. + + Args: + txn + transaction_id (str) + origin (str) + code (int) + response_json (str) + """ + + return self._simple_insert( + table="received_transactions", + values={ + "transaction_id": transaction_id, + "origin": origin, + "response_code": code, + "response_json": db_binary_type(encode_canonical_json(response_dict)), + "ts": self._clock.time_msec(), + }, + or_ignore=True, + desc="set_received_txn_response", + ) + + @defer.inlineCallbacks + def get_destination_retry_timings(self, destination): + """Gets the current retry timings (if any) for a given destination. + + Args: + destination (str) + + Returns: + None if not retrying + Otherwise a dict for the retry scheme + """ + + result = self._destination_retry_cache.get(destination, SENTINEL) + if result is not SENTINEL: + return result + + result = yield self.runInteraction( + "get_destination_retry_timings", + self._get_destination_retry_timings, + destination, + ) + + # We don't hugely care about race conditions between getting and + # invalidating the cache, since we time out fairly quickly anyway. + self._destination_retry_cache[destination] = result + return result + + def _get_destination_retry_timings(self, txn, destination): + result = self._simple_select_one_txn( + txn, + table="destinations", + keyvalues={"destination": destination}, + retcols=("destination", "failure_ts", "retry_last_ts", "retry_interval"), + allow_none=True, + ) + + if result and result["retry_last_ts"] > 0: + return result + else: + return None + + def set_destination_retry_timings( + self, destination, failure_ts, retry_last_ts, retry_interval + ): + """Sets the current retry timings for a given destination. + Both timings should be zero if retrying is no longer occuring. + + Args: + destination (str) + failure_ts (int|None) - when the server started failing (ms since epoch) + retry_last_ts (int) - time of last retry attempt in unix epoch ms + retry_interval (int) - how long until next retry in ms + """ + + self._destination_retry_cache.pop(destination, None) + return self.runInteraction( + "set_destination_retry_timings", + self._set_destination_retry_timings, + destination, + failure_ts, + retry_last_ts, + retry_interval, + ) + + def _set_destination_retry_timings( + self, txn, destination, failure_ts, retry_last_ts, retry_interval + ): + + if self.database_engine.can_native_upsert: + # Upsert retry time interval if retry_interval is zero (i.e. we're + # resetting it) or greater than the existing retry interval. + + sql = """ + INSERT INTO destinations ( + destination, failure_ts, retry_last_ts, retry_interval + ) + VALUES (?, ?, ?, ?) + ON CONFLICT (destination) DO UPDATE SET + failure_ts = EXCLUDED.failure_ts, + retry_last_ts = EXCLUDED.retry_last_ts, + retry_interval = EXCLUDED.retry_interval + WHERE + EXCLUDED.retry_interval = 0 + OR destinations.retry_interval < EXCLUDED.retry_interval + """ + + txn.execute(sql, (destination, failure_ts, retry_last_ts, retry_interval)) + + return + + self.database_engine.lock_table(txn, "destinations") + + # We need to be careful here as the data may have changed from under us + # due to a worker setting the timings. + + prev_row = self._simple_select_one_txn( + txn, + table="destinations", + keyvalues={"destination": destination}, + retcols=("failure_ts", "retry_last_ts", "retry_interval"), + allow_none=True, + ) + + if not prev_row: + self._simple_insert_txn( + txn, + table="destinations", + values={ + "destination": destination, + "failure_ts": failure_ts, + "retry_last_ts": retry_last_ts, + "retry_interval": retry_interval, + }, + ) + elif retry_interval == 0 or prev_row["retry_interval"] < retry_interval: + self._simple_update_one_txn( + txn, + "destinations", + keyvalues={"destination": destination}, + updatevalues={ + "failure_ts": failure_ts, + "retry_last_ts": retry_last_ts, + "retry_interval": retry_interval, + }, + ) + + def _start_cleanup_transactions(self): + return run_as_background_process( + "cleanup_transactions", self._cleanup_transactions + ) + + def _cleanup_transactions(self): + now = self._clock.time_msec() + month_ago = now - 30 * 24 * 60 * 60 * 1000 + + def _cleanup_transactions_txn(txn): + txn.execute("DELETE FROM received_transactions WHERE ts < ?", (month_ago,)) + + return self.runInteraction("_cleanup_transactions", _cleanup_transactions_txn) diff --git a/synapse/storage/data_stores/main/user_directory.py b/synapse/storage/data_stores/main/user_directory.py new file mode 100644 index 0000000000..652abe0e6a --- /dev/null +++ b/synapse/storage/data_stores/main/user_directory.py @@ -0,0 +1,827 @@ +# -*- coding: utf-8 -*- +# Copyright 2017 Vector Creations Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging +import re + +from twisted.internet import defer + +from synapse.api.constants import EventTypes, JoinRules +from synapse.storage.background_updates import BackgroundUpdateStore +from synapse.storage.data_stores.main.state import StateFilter +from synapse.storage.data_stores.main.state_deltas import StateDeltasStore +from synapse.storage.engines import PostgresEngine, Sqlite3Engine +from synapse.types import get_domain_from_id, get_localpart_from_id +from synapse.util.caches.descriptors import cached + +logger = logging.getLogger(__name__) + + +TEMP_TABLE = "_temp_populate_user_directory" + + +class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore): + + # How many records do we calculate before sending it to + # add_users_who_share_private_rooms? + SHARE_PRIVATE_WORKING_SET = 500 + + def __init__(self, db_conn, hs): + super(UserDirectoryBackgroundUpdateStore, self).__init__(db_conn, hs) + + self.server_name = hs.hostname + + self.register_background_update_handler( + "populate_user_directory_createtables", + self._populate_user_directory_createtables, + ) + self.register_background_update_handler( + "populate_user_directory_process_rooms", + self._populate_user_directory_process_rooms, + ) + self.register_background_update_handler( + "populate_user_directory_process_users", + self._populate_user_directory_process_users, + ) + self.register_background_update_handler( + "populate_user_directory_cleanup", self._populate_user_directory_cleanup + ) + + @defer.inlineCallbacks + def _populate_user_directory_createtables(self, progress, batch_size): + + # Get all the rooms that we want to process. + def _make_staging_area(txn): + sql = ( + "CREATE TABLE IF NOT EXISTS " + + TEMP_TABLE + + "_rooms(room_id TEXT NOT NULL, events BIGINT NOT NULL)" + ) + txn.execute(sql) + + sql = ( + "CREATE TABLE IF NOT EXISTS " + + TEMP_TABLE + + "_position(position TEXT NOT NULL)" + ) + txn.execute(sql) + + # Get rooms we want to process from the database + sql = """ + SELECT room_id, count(*) FROM current_state_events + GROUP BY room_id + """ + txn.execute(sql) + rooms = [{"room_id": x[0], "events": x[1]} for x in txn.fetchall()] + self._simple_insert_many_txn(txn, TEMP_TABLE + "_rooms", rooms) + del rooms + + # If search all users is on, get all the users we want to add. + if self.hs.config.user_directory_search_all_users: + sql = ( + "CREATE TABLE IF NOT EXISTS " + + TEMP_TABLE + + "_users(user_id TEXT NOT NULL)" + ) + txn.execute(sql) + + txn.execute("SELECT name FROM users") + users = [{"user_id": x[0]} for x in txn.fetchall()] + + self._simple_insert_many_txn(txn, TEMP_TABLE + "_users", users) + + new_pos = yield self.get_max_stream_id_in_current_state_deltas() + yield self.runInteraction( + "populate_user_directory_temp_build", _make_staging_area + ) + yield self._simple_insert(TEMP_TABLE + "_position", {"position": new_pos}) + + yield self._end_background_update("populate_user_directory_createtables") + return 1 + + @defer.inlineCallbacks + def _populate_user_directory_cleanup(self, progress, batch_size): + """ + Update the user directory stream position, then clean up the old tables. + """ + position = yield self._simple_select_one_onecol( + TEMP_TABLE + "_position", None, "position" + ) + yield self.update_user_directory_stream_pos(position) + + def _delete_staging_area(txn): + txn.execute("DROP TABLE IF EXISTS " + TEMP_TABLE + "_rooms") + txn.execute("DROP TABLE IF EXISTS " + TEMP_TABLE + "_users") + txn.execute("DROP TABLE IF EXISTS " + TEMP_TABLE + "_position") + + yield self.runInteraction( + "populate_user_directory_cleanup", _delete_staging_area + ) + + yield self._end_background_update("populate_user_directory_cleanup") + return 1 + + @defer.inlineCallbacks + def _populate_user_directory_process_rooms(self, progress, batch_size): + """ + Args: + progress (dict) + batch_size (int): Maximum number of state events to process + per cycle. + """ + state = self.hs.get_state_handler() + + # If we don't have progress filed, delete everything. + if not progress: + yield self.delete_all_from_user_dir() + + def _get_next_batch(txn): + # Only fetch 250 rooms, so we don't fetch too many at once, even + # if those 250 rooms have less than batch_size state events. + sql = """ + SELECT room_id, events FROM %s + ORDER BY events DESC + LIMIT 250 + """ % ( + TEMP_TABLE + "_rooms", + ) + txn.execute(sql) + rooms_to_work_on = txn.fetchall() + + if not rooms_to_work_on: + return None + + # Get how many are left to process, so we can give status on how + # far we are in processing + txn.execute("SELECT COUNT(*) FROM " + TEMP_TABLE + "_rooms") + progress["remaining"] = txn.fetchone()[0] + + return rooms_to_work_on + + rooms_to_work_on = yield self.runInteraction( + "populate_user_directory_temp_read", _get_next_batch + ) + + # No more rooms -- complete the transaction. + if not rooms_to_work_on: + yield self._end_background_update("populate_user_directory_process_rooms") + return 1 + + logger.info( + "Processing the next %d rooms of %d remaining" + % (len(rooms_to_work_on), progress["remaining"]) + ) + + processed_event_count = 0 + + for room_id, event_count in rooms_to_work_on: + is_in_room = yield self.is_host_joined(room_id, self.server_name) + + if is_in_room: + is_public = yield self.is_room_world_readable_or_publicly_joinable( + room_id + ) + + users_with_profile = yield state.get_current_users_in_room(room_id) + user_ids = set(users_with_profile) + + # Update each user in the user directory. + for user_id, profile in users_with_profile.items(): + yield self.update_profile_in_user_dir( + user_id, profile.display_name, profile.avatar_url + ) + + to_insert = set() + + if is_public: + for user_id in user_ids: + if self.get_if_app_services_interested_in_user(user_id): + continue + + to_insert.add(user_id) + + if to_insert: + yield self.add_users_in_public_rooms(room_id, to_insert) + to_insert.clear() + else: + for user_id in user_ids: + if not self.hs.is_mine_id(user_id): + continue + + if self.get_if_app_services_interested_in_user(user_id): + continue + + for other_user_id in user_ids: + if user_id == other_user_id: + continue + + user_set = (user_id, other_user_id) + to_insert.add(user_set) + + # If it gets too big, stop and write to the database + # to prevent storing too much in RAM. + if len(to_insert) >= self.SHARE_PRIVATE_WORKING_SET: + yield self.add_users_who_share_private_room( + room_id, to_insert + ) + to_insert.clear() + + if to_insert: + yield self.add_users_who_share_private_room(room_id, to_insert) + to_insert.clear() + + # We've finished a room. Delete it from the table. + yield self._simple_delete_one(TEMP_TABLE + "_rooms", {"room_id": room_id}) + # Update the remaining counter. + progress["remaining"] -= 1 + yield self.runInteraction( + "populate_user_directory", + self._background_update_progress_txn, + "populate_user_directory_process_rooms", + progress, + ) + + processed_event_count += event_count + + if processed_event_count > batch_size: + # Don't process any more rooms, we've hit our batch size. + return processed_event_count + + return processed_event_count + + @defer.inlineCallbacks + def _populate_user_directory_process_users(self, progress, batch_size): + """ + If search_all_users is enabled, add all of the users to the user directory. + """ + if not self.hs.config.user_directory_search_all_users: + yield self._end_background_update("populate_user_directory_process_users") + return 1 + + def _get_next_batch(txn): + sql = "SELECT user_id FROM %s LIMIT %s" % ( + TEMP_TABLE + "_users", + str(batch_size), + ) + txn.execute(sql) + users_to_work_on = txn.fetchall() + + if not users_to_work_on: + return None + + users_to_work_on = [x[0] for x in users_to_work_on] + + # Get how many are left to process, so we can give status on how + # far we are in processing + sql = "SELECT COUNT(*) FROM " + TEMP_TABLE + "_users" + txn.execute(sql) + progress["remaining"] = txn.fetchone()[0] + + return users_to_work_on + + users_to_work_on = yield self.runInteraction( + "populate_user_directory_temp_read", _get_next_batch + ) + + # No more users -- complete the transaction. + if not users_to_work_on: + yield self._end_background_update("populate_user_directory_process_users") + return 1 + + logger.info( + "Processing the next %d users of %d remaining" + % (len(users_to_work_on), progress["remaining"]) + ) + + for user_id in users_to_work_on: + profile = yield self.get_profileinfo(get_localpart_from_id(user_id)) + yield self.update_profile_in_user_dir( + user_id, profile.display_name, profile.avatar_url + ) + + # We've finished processing a user. Delete it from the table. + yield self._simple_delete_one(TEMP_TABLE + "_users", {"user_id": user_id}) + # Update the remaining counter. + progress["remaining"] -= 1 + yield self.runInteraction( + "populate_user_directory", + self._background_update_progress_txn, + "populate_user_directory_process_users", + progress, + ) + + return len(users_to_work_on) + + @defer.inlineCallbacks + def is_room_world_readable_or_publicly_joinable(self, room_id): + """Check if the room is either world_readable or publically joinable + """ + + # Create a state filter that only queries join and history state event + types_to_filter = ( + (EventTypes.JoinRules, ""), + (EventTypes.RoomHistoryVisibility, ""), + ) + + current_state_ids = yield self.get_filtered_current_state_ids( + room_id, StateFilter.from_types(types_to_filter) + ) + + join_rules_id = current_state_ids.get((EventTypes.JoinRules, "")) + if join_rules_id: + join_rule_ev = yield self.get_event(join_rules_id, allow_none=True) + if join_rule_ev: + if join_rule_ev.content.get("join_rule") == JoinRules.PUBLIC: + return True + + hist_vis_id = current_state_ids.get((EventTypes.RoomHistoryVisibility, "")) + if hist_vis_id: + hist_vis_ev = yield self.get_event(hist_vis_id, allow_none=True) + if hist_vis_ev: + if hist_vis_ev.content.get("history_visibility") == "world_readable": + return True + + return False + + def update_profile_in_user_dir(self, user_id, display_name, avatar_url): + """ + Update or add a user's profile in the user directory. + """ + + def _update_profile_in_user_dir_txn(txn): + new_entry = self._simple_upsert_txn( + txn, + table="user_directory", + keyvalues={"user_id": user_id}, + values={"display_name": display_name, "avatar_url": avatar_url}, + lock=False, # We're only inserter + ) + + if isinstance(self.database_engine, PostgresEngine): + # We weight the localpart most highly, then display name and finally + # server name + if self.database_engine.can_native_upsert: + sql = """ + INSERT INTO user_directory_search(user_id, vector) + VALUES (?, + setweight(to_tsvector('english', ?), 'A') + || setweight(to_tsvector('english', ?), 'D') + || setweight(to_tsvector('english', COALESCE(?, '')), 'B') + ) ON CONFLICT (user_id) DO UPDATE SET vector=EXCLUDED.vector + """ + txn.execute( + sql, + ( + user_id, + get_localpart_from_id(user_id), + get_domain_from_id(user_id), + display_name, + ), + ) + else: + # TODO: Remove this code after we've bumped the minimum version + # of postgres to always support upserts, so we can get rid of + # `new_entry` usage + if new_entry is True: + sql = """ + INSERT INTO user_directory_search(user_id, vector) + VALUES (?, + setweight(to_tsvector('english', ?), 'A') + || setweight(to_tsvector('english', ?), 'D') + || setweight(to_tsvector('english', COALESCE(?, '')), 'B') + ) + """ + txn.execute( + sql, + ( + user_id, + get_localpart_from_id(user_id), + get_domain_from_id(user_id), + display_name, + ), + ) + elif new_entry is False: + sql = """ + UPDATE user_directory_search + SET vector = setweight(to_tsvector('english', ?), 'A') + || setweight(to_tsvector('english', ?), 'D') + || setweight(to_tsvector('english', COALESCE(?, '')), 'B') + WHERE user_id = ? + """ + txn.execute( + sql, + ( + get_localpart_from_id(user_id), + get_domain_from_id(user_id), + display_name, + user_id, + ), + ) + else: + raise RuntimeError( + "upsert returned None when 'can_native_upsert' is False" + ) + elif isinstance(self.database_engine, Sqlite3Engine): + value = "%s %s" % (user_id, display_name) if display_name else user_id + self._simple_upsert_txn( + txn, + table="user_directory_search", + keyvalues={"user_id": user_id}, + values={"value": value}, + lock=False, # We're only inserter + ) + else: + # This should be unreachable. + raise Exception("Unrecognized database engine") + + txn.call_after(self.get_user_in_directory.invalidate, (user_id,)) + + return self.runInteraction( + "update_profile_in_user_dir", _update_profile_in_user_dir_txn + ) + + def add_users_who_share_private_room(self, room_id, user_id_tuples): + """Insert entries into the users_who_share_private_rooms table. The first + user should be a local user. + + Args: + room_id (str) + user_id_tuples([(str, str)]): iterable of 2-tuple of user IDs. + """ + + def _add_users_who_share_room_txn(txn): + self._simple_upsert_many_txn( + txn, + table="users_who_share_private_rooms", + key_names=["user_id", "other_user_id", "room_id"], + key_values=[ + (user_id, other_user_id, room_id) + for user_id, other_user_id in user_id_tuples + ], + value_names=(), + value_values=None, + ) + + return self.runInteraction( + "add_users_who_share_room", _add_users_who_share_room_txn + ) + + def add_users_in_public_rooms(self, room_id, user_ids): + """Insert entries into the users_who_share_private_rooms table. The first + user should be a local user. + + Args: + room_id (str) + user_ids (list[str]) + """ + + def _add_users_in_public_rooms_txn(txn): + + self._simple_upsert_many_txn( + txn, + table="users_in_public_rooms", + key_names=["user_id", "room_id"], + key_values=[(user_id, room_id) for user_id in user_ids], + value_names=(), + value_values=None, + ) + + return self.runInteraction( + "add_users_in_public_rooms", _add_users_in_public_rooms_txn + ) + + def delete_all_from_user_dir(self): + """Delete the entire user directory + """ + + def _delete_all_from_user_dir_txn(txn): + txn.execute("DELETE FROM user_directory") + txn.execute("DELETE FROM user_directory_search") + txn.execute("DELETE FROM users_in_public_rooms") + txn.execute("DELETE FROM users_who_share_private_rooms") + txn.call_after(self.get_user_in_directory.invalidate_all) + + return self.runInteraction( + "delete_all_from_user_dir", _delete_all_from_user_dir_txn + ) + + @cached() + def get_user_in_directory(self, user_id): + return self._simple_select_one( + table="user_directory", + keyvalues={"user_id": user_id}, + retcols=("display_name", "avatar_url"), + allow_none=True, + desc="get_user_in_directory", + ) + + def update_user_directory_stream_pos(self, stream_id): + return self._simple_update_one( + table="user_directory_stream_pos", + keyvalues={}, + updatevalues={"stream_id": stream_id}, + desc="update_user_directory_stream_pos", + ) + + +class UserDirectoryStore(UserDirectoryBackgroundUpdateStore): + + # How many records do we calculate before sending it to + # add_users_who_share_private_rooms? + SHARE_PRIVATE_WORKING_SET = 500 + + def __init__(self, db_conn, hs): + super(UserDirectoryStore, self).__init__(db_conn, hs) + + def remove_from_user_dir(self, user_id): + def _remove_from_user_dir_txn(txn): + self._simple_delete_txn( + txn, table="user_directory", keyvalues={"user_id": user_id} + ) + self._simple_delete_txn( + txn, table="user_directory_search", keyvalues={"user_id": user_id} + ) + self._simple_delete_txn( + txn, table="users_in_public_rooms", keyvalues={"user_id": user_id} + ) + self._simple_delete_txn( + txn, + table="users_who_share_private_rooms", + keyvalues={"user_id": user_id}, + ) + self._simple_delete_txn( + txn, + table="users_who_share_private_rooms", + keyvalues={"other_user_id": user_id}, + ) + txn.call_after(self.get_user_in_directory.invalidate, (user_id,)) + + return self.runInteraction("remove_from_user_dir", _remove_from_user_dir_txn) + + @defer.inlineCallbacks + def get_users_in_dir_due_to_room(self, room_id): + """Get all user_ids that are in the room directory because they're + in the given room_id + """ + user_ids_share_pub = yield self._simple_select_onecol( + table="users_in_public_rooms", + keyvalues={"room_id": room_id}, + retcol="user_id", + desc="get_users_in_dir_due_to_room", + ) + + user_ids_share_priv = yield self._simple_select_onecol( + table="users_who_share_private_rooms", + keyvalues={"room_id": room_id}, + retcol="other_user_id", + desc="get_users_in_dir_due_to_room", + ) + + user_ids = set(user_ids_share_pub) + user_ids.update(user_ids_share_priv) + + return user_ids + + def remove_user_who_share_room(self, user_id, room_id): + """ + Deletes entries in the users_who_share_*_rooms table. The first + user should be a local user. + + Args: + user_id (str) + room_id (str) + """ + + def _remove_user_who_share_room_txn(txn): + self._simple_delete_txn( + txn, + table="users_who_share_private_rooms", + keyvalues={"user_id": user_id, "room_id": room_id}, + ) + self._simple_delete_txn( + txn, + table="users_who_share_private_rooms", + keyvalues={"other_user_id": user_id, "room_id": room_id}, + ) + self._simple_delete_txn( + txn, + table="users_in_public_rooms", + keyvalues={"user_id": user_id, "room_id": room_id}, + ) + + return self.runInteraction( + "remove_user_who_share_room", _remove_user_who_share_room_txn + ) + + @defer.inlineCallbacks + def get_user_dir_rooms_user_is_in(self, user_id): + """ + Returns the rooms that a user is in. + + Args: + user_id(str): Must be a local user + + Returns: + list: user_id + """ + rows = yield self._simple_select_onecol( + table="users_who_share_private_rooms", + keyvalues={"user_id": user_id}, + retcol="room_id", + desc="get_rooms_user_is_in", + ) + + pub_rows = yield self._simple_select_onecol( + table="users_in_public_rooms", + keyvalues={"user_id": user_id}, + retcol="room_id", + desc="get_rooms_user_is_in", + ) + + users = set(pub_rows) + users.update(rows) + return list(users) + + @defer.inlineCallbacks + def get_rooms_in_common_for_users(self, user_id, other_user_id): + """Given two user_ids find out the list of rooms they share. + """ + sql = """ + SELECT room_id FROM ( + SELECT c.room_id FROM current_state_events AS c + INNER JOIN room_memberships AS m USING (event_id) + WHERE type = 'm.room.member' + AND m.membership = 'join' + AND state_key = ? + ) AS f1 INNER JOIN ( + SELECT c.room_id FROM current_state_events AS c + INNER JOIN room_memberships AS m USING (event_id) + WHERE type = 'm.room.member' + AND m.membership = 'join' + AND state_key = ? + ) f2 USING (room_id) + """ + + rows = yield self._execute( + "get_rooms_in_common_for_users", None, sql, user_id, other_user_id + ) + + return [room_id for room_id, in rows] + + def get_user_directory_stream_pos(self): + return self._simple_select_one_onecol( + table="user_directory_stream_pos", + keyvalues={}, + retcol="stream_id", + desc="get_user_directory_stream_pos", + ) + + @defer.inlineCallbacks + def search_user_dir(self, user_id, search_term, limit): + """Searches for users in directory + + Returns: + dict of the form:: + + { + "limited": , # whether there were more results or not + "results": [ # Ordered by best match first + { + "user_id": , + "display_name": , + "avatar_url": + } + ] + } + """ + + if self.hs.config.user_directory_search_all_users: + join_args = (user_id,) + where_clause = "user_id != ?" + else: + join_args = (user_id,) + where_clause = """ + ( + EXISTS (select 1 from users_in_public_rooms WHERE user_id = t.user_id) + OR EXISTS ( + SELECT 1 FROM users_who_share_private_rooms + WHERE user_id = ? AND other_user_id = t.user_id + ) + ) + """ + + if isinstance(self.database_engine, PostgresEngine): + full_query, exact_query, prefix_query = _parse_query_postgres(search_term) + + # We order by rank and then if they have profile info + # The ranking algorithm is hand tweaked for "best" results. Broadly + # the idea is we give a higher weight to exact matches. + # The array of numbers are the weights for the various part of the + # search: (domain, _, display name, localpart) + sql = """ + SELECT d.user_id AS user_id, display_name, avatar_url + FROM user_directory_search as t + INNER JOIN user_directory AS d USING (user_id) + WHERE + %s + AND vector @@ to_tsquery('english', ?) + ORDER BY + (CASE WHEN d.user_id IS NOT NULL THEN 4.0 ELSE 1.0 END) + * (CASE WHEN display_name IS NOT NULL THEN 1.2 ELSE 1.0 END) + * (CASE WHEN avatar_url IS NOT NULL THEN 1.2 ELSE 1.0 END) + * ( + 3 * ts_rank_cd( + '{0.1, 0.1, 0.9, 1.0}', + vector, + to_tsquery('english', ?), + 8 + ) + + ts_rank_cd( + '{0.1, 0.1, 0.9, 1.0}', + vector, + to_tsquery('english', ?), + 8 + ) + ) + DESC, + display_name IS NULL, + avatar_url IS NULL + LIMIT ? + """ % ( + where_clause, + ) + args = join_args + (full_query, exact_query, prefix_query, limit + 1) + elif isinstance(self.database_engine, Sqlite3Engine): + search_query = _parse_query_sqlite(search_term) + + sql = """ + SELECT d.user_id AS user_id, display_name, avatar_url + FROM user_directory_search as t + INNER JOIN user_directory AS d USING (user_id) + WHERE + %s + AND value MATCH ? + ORDER BY + rank(matchinfo(user_directory_search)) DESC, + display_name IS NULL, + avatar_url IS NULL + LIMIT ? + """ % ( + where_clause, + ) + args = join_args + (search_query, limit + 1) + else: + # This should be unreachable. + raise Exception("Unrecognized database engine") + + results = yield self._execute( + "search_user_dir", self.cursor_to_dict, sql, *args + ) + + limited = len(results) > limit + + return {"limited": limited, "results": results} + + +def _parse_query_sqlite(search_term): + """Takes a plain unicode string from the user and converts it into a form + that can be passed to database. + We use this so that we can add prefix matching, which isn't something + that is supported by default. + + We specifically add both a prefix and non prefix matching term so that + exact matches get ranked higher. + """ + + # Pull out the individual words, discarding any non-word characters. + results = re.findall(r"([\w\-]+)", search_term, re.UNICODE) + return " & ".join("(%s* OR %s)" % (result, result) for result in results) + + +def _parse_query_postgres(search_term): + """Takes a plain unicode string from the user and converts it into a form + that can be passed to database. + We use this so that we can add prefix matching, which isn't something + that is supported by default. + """ + + # Pull out the individual words, discarding any non-word characters. + results = re.findall(r"([\w\-]+)", search_term, re.UNICODE) + + both = " & ".join("(%s:* | %s)" % (result, result) for result in results) + exact = " & ".join("%s" % (result,) for result in results) + prefix = " & ".join("%s:*" % (result,) for result in results) + + return both, exact, prefix diff --git a/synapse/storage/data_stores/main/user_erasure_store.py b/synapse/storage/data_stores/main/user_erasure_store.py new file mode 100644 index 0000000000..aa4f0da5f0 --- /dev/null +++ b/synapse/storage/data_stores/main/user_erasure_store.py @@ -0,0 +1,91 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import operator + +from synapse.storage._base import SQLBaseStore +from synapse.util.caches.descriptors import cached, cachedList + + +class UserErasureWorkerStore(SQLBaseStore): + @cached() + def is_user_erased(self, user_id): + """ + Check if the given user id has requested erasure + + Args: + user_id (str): full user id to check + + Returns: + Deferred[bool]: True if the user has requested erasure + """ + return self._simple_select_onecol( + table="erased_users", + keyvalues={"user_id": user_id}, + retcol="1", + desc="is_user_erased", + ).addCallback(operator.truth) + + @cachedList( + cached_method_name="is_user_erased", list_name="user_ids", inlineCallbacks=True + ) + def are_users_erased(self, user_ids): + """ + Checks which users in a list have requested erasure + + Args: + user_ids (iterable[str]): full user id to check + + Returns: + Deferred[dict[str, bool]]: + for each user, whether the user has requested erasure. + """ + # this serves the dual purpose of (a) making sure we can do len and + # iterate it multiple times, and (b) avoiding duplicates. + user_ids = tuple(set(user_ids)) + + rows = yield self._simple_select_many_batch( + table="erased_users", + column="user_id", + iterable=user_ids, + retcols=("user_id",), + desc="are_users_erased", + ) + erased_users = set(row["user_id"] for row in rows) + + res = dict((u, u in erased_users) for u in user_ids) + return res + + +class UserErasureStore(UserErasureWorkerStore): + def mark_user_erased(self, user_id): + """Indicate that user_id wishes their message history to be erased. + + Args: + user_id (str): full user_id to be erased + """ + + def f(txn): + # first check if they are already in the list + txn.execute("SELECT 1 FROM erased_users WHERE user_id = ?", (user_id,)) + if txn.fetchone(): + return + + # they are not already there: do the insert. + txn.execute("INSERT INTO erased_users (user_id) VALUES (?)", (user_id,)) + + self._invalidate_cache_and_stream(txn, self.is_user_erased, (user_id,)) + + return self.runInteraction("mark_user_erased", f) diff --git a/synapse/storage/deviceinbox.py b/synapse/storage/deviceinbox.py deleted file mode 100644 index f04aad0743..0000000000 --- a/synapse/storage/deviceinbox.py +++ /dev/null @@ -1,457 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import logging - -from canonicaljson import json - -from twisted.internet import defer - -from synapse.logging.opentracing import log_kv, set_tag, trace -from synapse.storage._base import SQLBaseStore, make_in_list_sql_clause -from synapse.storage.background_updates import BackgroundUpdateStore -from synapse.util.caches.expiringcache import ExpiringCache - -logger = logging.getLogger(__name__) - - -class DeviceInboxWorkerStore(SQLBaseStore): - def get_to_device_stream_token(self): - return self._device_inbox_id_gen.get_current_token() - - def get_new_messages_for_device( - self, user_id, device_id, last_stream_id, current_stream_id, limit=100 - ): - """ - Args: - user_id(str): The recipient user_id. - device_id(str): The recipient device_id. - current_stream_id(int): The current position of the to device - message stream. - Returns: - Deferred ([dict], int): List of messages for the device and where - in the stream the messages got to. - """ - has_changed = self._device_inbox_stream_cache.has_entity_changed( - user_id, last_stream_id - ) - if not has_changed: - return defer.succeed(([], current_stream_id)) - - def get_new_messages_for_device_txn(txn): - sql = ( - "SELECT stream_id, message_json FROM device_inbox" - " WHERE user_id = ? AND device_id = ?" - " AND ? < stream_id AND stream_id <= ?" - " ORDER BY stream_id ASC" - " LIMIT ?" - ) - txn.execute( - sql, (user_id, device_id, last_stream_id, current_stream_id, limit) - ) - messages = [] - for row in txn: - stream_pos = row[0] - messages.append(json.loads(row[1])) - if len(messages) < limit: - stream_pos = current_stream_id - return messages, stream_pos - - return self.runInteraction( - "get_new_messages_for_device", get_new_messages_for_device_txn - ) - - @trace - @defer.inlineCallbacks - def delete_messages_for_device(self, user_id, device_id, up_to_stream_id): - """ - Args: - user_id(str): The recipient user_id. - device_id(str): The recipient device_id. - up_to_stream_id(int): Where to delete messages up to. - Returns: - A deferred that resolves to the number of messages deleted. - """ - # If we have cached the last stream id we've deleted up to, we can - # check if there is likely to be anything that needs deleting - last_deleted_stream_id = self._last_device_delete_cache.get( - (user_id, device_id), None - ) - - set_tag("last_deleted_stream_id", last_deleted_stream_id) - - if last_deleted_stream_id: - has_changed = self._device_inbox_stream_cache.has_entity_changed( - user_id, last_deleted_stream_id - ) - if not has_changed: - log_kv({"message": "No changes in cache since last check"}) - return 0 - - def delete_messages_for_device_txn(txn): - sql = ( - "DELETE FROM device_inbox" - " WHERE user_id = ? AND device_id = ?" - " AND stream_id <= ?" - ) - txn.execute(sql, (user_id, device_id, up_to_stream_id)) - return txn.rowcount - - count = yield self.runInteraction( - "delete_messages_for_device", delete_messages_for_device_txn - ) - - log_kv( - {"message": "deleted {} messages for device".format(count), "count": count} - ) - - # Update the cache, ensuring that we only ever increase the value - last_deleted_stream_id = self._last_device_delete_cache.get( - (user_id, device_id), 0 - ) - self._last_device_delete_cache[(user_id, device_id)] = max( - last_deleted_stream_id, up_to_stream_id - ) - - return count - - @trace - def get_new_device_msgs_for_remote( - self, destination, last_stream_id, current_stream_id, limit - ): - """ - Args: - destination(str): The name of the remote server. - last_stream_id(int|long): The last position of the device message stream - that the server sent up to. - current_stream_id(int|long): The current position of the device - message stream. - Returns: - Deferred ([dict], int|long): List of messages for the device and where - in the stream the messages got to. - """ - - set_tag("destination", destination) - set_tag("last_stream_id", last_stream_id) - set_tag("current_stream_id", current_stream_id) - set_tag("limit", limit) - - has_changed = self._device_federation_outbox_stream_cache.has_entity_changed( - destination, last_stream_id - ) - if not has_changed or last_stream_id == current_stream_id: - log_kv({"message": "No new messages in stream"}) - return defer.succeed(([], current_stream_id)) - - if limit <= 0: - # This can happen if we run out of room for EDUs in the transaction. - return defer.succeed(([], last_stream_id)) - - @trace - def get_new_messages_for_remote_destination_txn(txn): - sql = ( - "SELECT stream_id, messages_json FROM device_federation_outbox" - " WHERE destination = ?" - " AND ? < stream_id AND stream_id <= ?" - " ORDER BY stream_id ASC" - " LIMIT ?" - ) - txn.execute(sql, (destination, last_stream_id, current_stream_id, limit)) - messages = [] - for row in txn: - stream_pos = row[0] - messages.append(json.loads(row[1])) - if len(messages) < limit: - log_kv({"message": "Set stream position to current position"}) - stream_pos = current_stream_id - return messages, stream_pos - - return self.runInteraction( - "get_new_device_msgs_for_remote", - get_new_messages_for_remote_destination_txn, - ) - - @trace - def delete_device_msgs_for_remote(self, destination, up_to_stream_id): - """Used to delete messages when the remote destination acknowledges - their receipt. - - Args: - destination(str): The destination server_name - up_to_stream_id(int): Where to delete messages up to. - Returns: - A deferred that resolves when the messages have been deleted. - """ - - def delete_messages_for_remote_destination_txn(txn): - sql = ( - "DELETE FROM device_federation_outbox" - " WHERE destination = ?" - " AND stream_id <= ?" - ) - txn.execute(sql, (destination, up_to_stream_id)) - - return self.runInteraction( - "delete_device_msgs_for_remote", delete_messages_for_remote_destination_txn - ) - - -class DeviceInboxBackgroundUpdateStore(BackgroundUpdateStore): - DEVICE_INBOX_STREAM_ID = "device_inbox_stream_drop" - - def __init__(self, db_conn, hs): - super(DeviceInboxBackgroundUpdateStore, self).__init__(db_conn, hs) - - self.register_background_index_update( - "device_inbox_stream_index", - index_name="device_inbox_stream_id_user_id", - table="device_inbox", - columns=["stream_id", "user_id"], - ) - - self.register_background_update_handler( - self.DEVICE_INBOX_STREAM_ID, self._background_drop_index_device_inbox - ) - - @defer.inlineCallbacks - def _background_drop_index_device_inbox(self, progress, batch_size): - def reindex_txn(conn): - txn = conn.cursor() - txn.execute("DROP INDEX IF EXISTS device_inbox_stream_id") - txn.close() - - yield self.runWithConnection(reindex_txn) - - yield self._end_background_update(self.DEVICE_INBOX_STREAM_ID) - - return 1 - - -class DeviceInboxStore(DeviceInboxWorkerStore, DeviceInboxBackgroundUpdateStore): - DEVICE_INBOX_STREAM_ID = "device_inbox_stream_drop" - - def __init__(self, db_conn, hs): - super(DeviceInboxStore, self).__init__(db_conn, hs) - - # Map of (user_id, device_id) to the last stream_id that has been - # deleted up to. This is so that we can no op deletions. - self._last_device_delete_cache = ExpiringCache( - cache_name="last_device_delete_cache", - clock=self._clock, - max_len=10000, - expiry_ms=30 * 60 * 1000, - ) - - @trace - @defer.inlineCallbacks - def add_messages_to_device_inbox( - self, local_messages_by_user_then_device, remote_messages_by_destination - ): - """Used to send messages from this server. - - Args: - sender_user_id(str): The ID of the user sending these messages. - local_messages_by_user_and_device(dict): - Dictionary of user_id to device_id to message. - remote_messages_by_destination(dict): - Dictionary of destination server_name to the EDU JSON to send. - Returns: - A deferred stream_id that resolves when the messages have been - inserted. - """ - - def add_messages_txn(txn, now_ms, stream_id): - # Add the local messages directly to the local inbox. - self._add_messages_to_local_device_inbox_txn( - txn, stream_id, local_messages_by_user_then_device - ) - - # Add the remote messages to the federation outbox. - # We'll send them to a remote server when we next send a - # federation transaction to that destination. - sql = ( - "INSERT INTO device_federation_outbox" - " (destination, stream_id, queued_ts, messages_json)" - " VALUES (?,?,?,?)" - ) - rows = [] - for destination, edu in remote_messages_by_destination.items(): - edu_json = json.dumps(edu) - rows.append((destination, stream_id, now_ms, edu_json)) - txn.executemany(sql, rows) - - with self._device_inbox_id_gen.get_next() as stream_id: - now_ms = self.clock.time_msec() - yield self.runInteraction( - "add_messages_to_device_inbox", add_messages_txn, now_ms, stream_id - ) - for user_id in local_messages_by_user_then_device.keys(): - self._device_inbox_stream_cache.entity_has_changed(user_id, stream_id) - for destination in remote_messages_by_destination.keys(): - self._device_federation_outbox_stream_cache.entity_has_changed( - destination, stream_id - ) - - return self._device_inbox_id_gen.get_current_token() - - @defer.inlineCallbacks - def add_messages_from_remote_to_device_inbox( - self, origin, message_id, local_messages_by_user_then_device - ): - def add_messages_txn(txn, now_ms, stream_id): - # Check if we've already inserted a matching message_id for that - # origin. This can happen if the origin doesn't receive our - # acknowledgement from the first time we received the message. - already_inserted = self._simple_select_one_txn( - txn, - table="device_federation_inbox", - keyvalues={"origin": origin, "message_id": message_id}, - retcols=("message_id",), - allow_none=True, - ) - if already_inserted is not None: - return - - # Add an entry for this message_id so that we know we've processed - # it. - self._simple_insert_txn( - txn, - table="device_federation_inbox", - values={ - "origin": origin, - "message_id": message_id, - "received_ts": now_ms, - }, - ) - - # Add the messages to the approriate local device inboxes so that - # they'll be sent to the devices when they next sync. - self._add_messages_to_local_device_inbox_txn( - txn, stream_id, local_messages_by_user_then_device - ) - - with self._device_inbox_id_gen.get_next() as stream_id: - now_ms = self.clock.time_msec() - yield self.runInteraction( - "add_messages_from_remote_to_device_inbox", - add_messages_txn, - now_ms, - stream_id, - ) - for user_id in local_messages_by_user_then_device.keys(): - self._device_inbox_stream_cache.entity_has_changed(user_id, stream_id) - - return stream_id - - def _add_messages_to_local_device_inbox_txn( - self, txn, stream_id, messages_by_user_then_device - ): - sql = "UPDATE device_max_stream_id" " SET stream_id = ?" " WHERE stream_id < ?" - txn.execute(sql, (stream_id, stream_id)) - - local_by_user_then_device = {} - for user_id, messages_by_device in messages_by_user_then_device.items(): - messages_json_for_user = {} - devices = list(messages_by_device.keys()) - if len(devices) == 1 and devices[0] == "*": - # Handle wildcard device_ids. - sql = "SELECT device_id FROM devices" " WHERE user_id = ?" - txn.execute(sql, (user_id,)) - message_json = json.dumps(messages_by_device["*"]) - for row in txn: - # Add the message for all devices for this user on this - # server. - device = row[0] - messages_json_for_user[device] = message_json - else: - if not devices: - continue - - clause, args = make_in_list_sql_clause( - txn.database_engine, "device_id", devices - ) - sql = "SELECT device_id FROM devices WHERE user_id = ? AND " + clause - - # TODO: Maybe this needs to be done in batches if there are - # too many local devices for a given user. - txn.execute(sql, [user_id] + list(args)) - for row in txn: - # Only insert into the local inbox if the device exists on - # this server - device = row[0] - message_json = json.dumps(messages_by_device[device]) - messages_json_for_user[device] = message_json - - if messages_json_for_user: - local_by_user_then_device[user_id] = messages_json_for_user - - if not local_by_user_then_device: - return - - sql = ( - "INSERT INTO device_inbox" - " (user_id, device_id, stream_id, message_json)" - " VALUES (?,?,?,?)" - ) - rows = [] - for user_id, messages_by_device in local_by_user_then_device.items(): - for device_id, message_json in messages_by_device.items(): - rows.append((user_id, device_id, stream_id, message_json)) - - txn.executemany(sql, rows) - - def get_all_new_device_messages(self, last_pos, current_pos, limit): - """ - Args: - last_pos(int): - current_pos(int): - limit(int): - Returns: - A deferred list of rows from the device inbox - """ - if last_pos == current_pos: - return defer.succeed([]) - - def get_all_new_device_messages_txn(txn): - # We limit like this as we might have multiple rows per stream_id, and - # we want to make sure we always get all entries for any stream_id - # we return. - upper_pos = min(current_pos, last_pos + limit) - sql = ( - "SELECT max(stream_id), user_id" - " FROM device_inbox" - " WHERE ? < stream_id AND stream_id <= ?" - " GROUP BY user_id" - ) - txn.execute(sql, (last_pos, upper_pos)) - rows = txn.fetchall() - - sql = ( - "SELECT max(stream_id), destination" - " FROM device_federation_outbox" - " WHERE ? < stream_id AND stream_id <= ?" - " GROUP BY destination" - ) - txn.execute(sql, (last_pos, upper_pos)) - rows.extend(txn) - - # Order by ascending stream ordering - rows.sort() - - return rows - - return self.runInteraction( - "get_all_new_device_messages", get_all_new_device_messages_txn - ) diff --git a/synapse/storage/devices.py b/synapse/storage/devices.py deleted file mode 100644 index ac5239e50a..0000000000 --- a/synapse/storage/devices.py +++ /dev/null @@ -1,937 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. -import logging - -from six import iteritems - -from canonicaljson import json - -from twisted.internet import defer - -from synapse.api.errors import StoreError -from synapse.logging.opentracing import ( - get_active_span_text_map, - set_tag, - trace, - whitelisted_homeserver, -) -from synapse.metrics.background_process_metrics import run_as_background_process -from synapse.storage._base import ( - Cache, - SQLBaseStore, - db_to_json, - make_in_list_sql_clause, -) -from synapse.storage.background_updates import BackgroundUpdateStore -from synapse.util import batch_iter -from synapse.util.caches.descriptors import cached, cachedInlineCallbacks, cachedList - -logger = logging.getLogger(__name__) - -DROP_DEVICE_LIST_STREAMS_NON_UNIQUE_INDEXES = ( - "drop_device_list_streams_non_unique_indexes" -) - - -class DeviceWorkerStore(SQLBaseStore): - def get_device(self, user_id, device_id): - """Retrieve a device. - - Args: - user_id (str): The ID of the user which owns the device - device_id (str): The ID of the device to retrieve - Returns: - defer.Deferred for a dict containing the device information - Raises: - StoreError: if the device is not found - """ - return self._simple_select_one( - table="devices", - keyvalues={"user_id": user_id, "device_id": device_id}, - retcols=("user_id", "device_id", "display_name"), - desc="get_device", - ) - - @defer.inlineCallbacks - def get_devices_by_user(self, user_id): - """Retrieve all of a user's registered devices. - - Args: - user_id (str): - Returns: - defer.Deferred: resolves to a dict from device_id to a dict - containing "device_id", "user_id" and "display_name" for each - device. - """ - devices = yield self._simple_select_list( - table="devices", - keyvalues={"user_id": user_id}, - retcols=("user_id", "device_id", "display_name"), - desc="get_devices_by_user", - ) - - return {d["device_id"]: d for d in devices} - - @trace - @defer.inlineCallbacks - def get_devices_by_remote(self, destination, from_stream_id, limit): - """Get stream of updates to send to remote servers - - Returns: - Deferred[tuple[int, list[dict]]]: - current stream id (ie, the stream id of the last update included in the - response), and the list of updates - """ - now_stream_id = self._device_list_id_gen.get_current_token() - - has_changed = self._device_list_federation_stream_cache.has_entity_changed( - destination, int(from_stream_id) - ) - if not has_changed: - return now_stream_id, [] - - # We retrieve n+1 devices from the list of outbound pokes where n is - # our outbound device update limit. We then check if the very last - # device has the same stream_id as the second-to-last device. If so, - # then we ignore all devices with that stream_id and only send the - # devices with a lower stream_id. - # - # If when culling the list we end up with no devices afterwards, we - # consider the device update to be too large, and simply skip the - # stream_id; the rationale being that such a large device list update - # is likely an error. - updates = yield self.runInteraction( - "get_devices_by_remote", - self._get_devices_by_remote_txn, - destination, - from_stream_id, - now_stream_id, - limit + 1, - ) - - # Return an empty list if there are no updates - if not updates: - return now_stream_id, [] - - # if we have exceeded the limit, we need to exclude any results with the - # same stream_id as the last row. - if len(updates) > limit: - stream_id_cutoff = updates[-1][2] - now_stream_id = stream_id_cutoff - 1 - else: - stream_id_cutoff = None - - # Perform the equivalent of a GROUP BY - # - # Iterate through the updates list and copy non-duplicate - # (user_id, device_id) entries into a map, with the value being - # the max stream_id across each set of duplicate entries - # - # maps (user_id, device_id) -> (stream_id, opentracing_context) - # as long as their stream_id does not match that of the last row - # - # opentracing_context contains the opentracing metadata for the request - # that created the poke - # - # The most recent request's opentracing_context is used as the - # context which created the Edu. - - query_map = {} - for update in updates: - if stream_id_cutoff is not None and update[2] >= stream_id_cutoff: - # Stop processing updates - break - - key = (update[0], update[1]) - - update_context = update[3] - update_stream_id = update[2] - - previous_update_stream_id, _ = query_map.get(key, (0, None)) - - if update_stream_id > previous_update_stream_id: - query_map[key] = (update_stream_id, update_context) - - # If we didn't find any updates with a stream_id lower than the cutoff, it - # means that there are more than limit updates all of which have the same - # steam_id. - - # That should only happen if a client is spamming the server with new - # devices, in which case E2E isn't going to work well anyway. We'll just - # skip that stream_id and return an empty list, and continue with the next - # stream_id next time. - if not query_map: - return stream_id_cutoff, [] - - results = yield self._get_device_update_edus_by_remote( - destination, from_stream_id, query_map - ) - - return now_stream_id, results - - def _get_devices_by_remote_txn( - self, txn, destination, from_stream_id, now_stream_id, limit - ): - """Return device update information for a given remote destination - - Args: - txn (LoggingTransaction): The transaction to execute - destination (str): The host the device updates are intended for - from_stream_id (int): The minimum stream_id to filter updates by, exclusive - now_stream_id (int): The maximum stream_id to filter updates by, inclusive - limit (int): Maximum number of device updates to return - - Returns: - List: List of device updates - """ - sql = """ - SELECT user_id, device_id, stream_id, opentracing_context FROM device_lists_outbound_pokes - WHERE destination = ? AND ? < stream_id AND stream_id <= ? AND sent = ? - ORDER BY stream_id - LIMIT ? - """ - txn.execute(sql, (destination, from_stream_id, now_stream_id, False, limit)) - - return list(txn) - - @defer.inlineCallbacks - def _get_device_update_edus_by_remote(self, destination, from_stream_id, query_map): - """Returns a list of device update EDUs as well as E2EE keys - - Args: - destination (str): The host the device updates are intended for - from_stream_id (int): The minimum stream_id to filter updates by, exclusive - query_map (Dict[(str, str): (int, str|None)]): Dictionary mapping - user_id/device_id to update stream_id and the relevent json-encoded - opentracing context - - Returns: - List[Dict]: List of objects representing an device update EDU - - """ - devices = yield self.runInteraction( - "_get_e2e_device_keys_txn", - self._get_e2e_device_keys_txn, - query_map.keys(), - include_all_devices=True, - include_deleted_devices=True, - ) - - results = [] - for user_id, user_devices in iteritems(devices): - # The prev_id for the first row is always the last row before - # `from_stream_id` - prev_id = yield self._get_last_device_update_for_remote_user( - destination, user_id, from_stream_id - ) - for device_id, device in iteritems(user_devices): - stream_id, opentracing_context = query_map[(user_id, device_id)] - result = { - "user_id": user_id, - "device_id": device_id, - "prev_id": [prev_id] if prev_id else [], - "stream_id": stream_id, - "org.matrix.opentracing_context": opentracing_context, - } - - prev_id = stream_id - - if device is not None: - key_json = device.get("key_json", None) - if key_json: - result["keys"] = db_to_json(key_json) - device_display_name = device.get("device_display_name", None) - if device_display_name: - result["device_display_name"] = device_display_name - else: - result["deleted"] = True - - results.append(result) - - return results - - def _get_last_device_update_for_remote_user( - self, destination, user_id, from_stream_id - ): - def f(txn): - prev_sent_id_sql = """ - SELECT coalesce(max(stream_id), 0) as stream_id - FROM device_lists_outbound_last_success - WHERE destination = ? AND user_id = ? AND stream_id <= ? - """ - txn.execute(prev_sent_id_sql, (destination, user_id, from_stream_id)) - rows = txn.fetchall() - return rows[0][0] - - return self.runInteraction("get_last_device_update_for_remote_user", f) - - def mark_as_sent_devices_by_remote(self, destination, stream_id): - """Mark that updates have successfully been sent to the destination. - """ - return self.runInteraction( - "mark_as_sent_devices_by_remote", - self._mark_as_sent_devices_by_remote_txn, - destination, - stream_id, - ) - - def _mark_as_sent_devices_by_remote_txn(self, txn, destination, stream_id): - # We update the device_lists_outbound_last_success with the successfully - # poked users. We do the join to see which users need to be inserted and - # which updated. - sql = """ - SELECT user_id, coalesce(max(o.stream_id), 0), (max(s.stream_id) IS NOT NULL) - FROM device_lists_outbound_pokes as o - LEFT JOIN device_lists_outbound_last_success as s - USING (destination, user_id) - WHERE destination = ? AND o.stream_id <= ? - GROUP BY user_id - """ - txn.execute(sql, (destination, stream_id)) - rows = txn.fetchall() - - sql = """ - UPDATE device_lists_outbound_last_success - SET stream_id = ? - WHERE destination = ? AND user_id = ? - """ - txn.executemany(sql, ((row[1], destination, row[0]) for row in rows if row[2])) - - sql = """ - INSERT INTO device_lists_outbound_last_success - (destination, user_id, stream_id) VALUES (?, ?, ?) - """ - txn.executemany( - sql, ((destination, row[0], row[1]) for row in rows if not row[2]) - ) - - # Delete all sent outbound pokes - sql = """ - DELETE FROM device_lists_outbound_pokes - WHERE destination = ? AND stream_id <= ? - """ - txn.execute(sql, (destination, stream_id)) - - def get_device_stream_token(self): - return self._device_list_id_gen.get_current_token() - - @trace - @defer.inlineCallbacks - def get_user_devices_from_cache(self, query_list): - """Get the devices (and keys if any) for remote users from the cache. - - Args: - query_list(list): List of (user_id, device_ids), if device_ids is - falsey then return all device ids for that user. - - Returns: - (user_ids_not_in_cache, results_map), where user_ids_not_in_cache is - a set of user_ids and results_map is a mapping of - user_id -> device_id -> device_info - """ - user_ids = set(user_id for user_id, _ in query_list) - user_map = yield self.get_device_list_last_stream_id_for_remotes(list(user_ids)) - user_ids_in_cache = set( - user_id for user_id, stream_id in user_map.items() if stream_id - ) - user_ids_not_in_cache = user_ids - user_ids_in_cache - - results = {} - for user_id, device_id in query_list: - if user_id not in user_ids_in_cache: - continue - - if device_id: - device = yield self._get_cached_user_device(user_id, device_id) - results.setdefault(user_id, {})[device_id] = device - else: - results[user_id] = yield self._get_cached_devices_for_user(user_id) - - set_tag("in_cache", results) - set_tag("not_in_cache", user_ids_not_in_cache) - - return user_ids_not_in_cache, results - - @cachedInlineCallbacks(num_args=2, tree=True) - def _get_cached_user_device(self, user_id, device_id): - content = yield self._simple_select_one_onecol( - table="device_lists_remote_cache", - keyvalues={"user_id": user_id, "device_id": device_id}, - retcol="content", - desc="_get_cached_user_device", - ) - return db_to_json(content) - - @cachedInlineCallbacks() - def _get_cached_devices_for_user(self, user_id): - devices = yield self._simple_select_list( - table="device_lists_remote_cache", - keyvalues={"user_id": user_id}, - retcols=("device_id", "content"), - desc="_get_cached_devices_for_user", - ) - return { - device["device_id"]: db_to_json(device["content"]) for device in devices - } - - def get_devices_with_keys_by_user(self, user_id): - """Get all devices (with any device keys) for a user - - Returns: - (stream_id, devices) - """ - return self.runInteraction( - "get_devices_with_keys_by_user", - self._get_devices_with_keys_by_user_txn, - user_id, - ) - - def _get_devices_with_keys_by_user_txn(self, txn, user_id): - now_stream_id = self._device_list_id_gen.get_current_token() - - devices = self._get_e2e_device_keys_txn( - txn, [(user_id, None)], include_all_devices=True - ) - - if devices: - user_devices = devices[user_id] - results = [] - for device_id, device in iteritems(user_devices): - result = {"device_id": device_id} - - key_json = device.get("key_json", None) - if key_json: - result["keys"] = db_to_json(key_json) - device_display_name = device.get("device_display_name", None) - if device_display_name: - result["device_display_name"] = device_display_name - - results.append(result) - - return now_stream_id, results - - return now_stream_id, [] - - def get_users_whose_devices_changed(self, from_key, user_ids): - """Get set of users whose devices have changed since `from_key` that - are in the given list of user_ids. - - Args: - from_key (str): The device lists stream token - user_ids (Iterable[str]) - - Returns: - Deferred[set[str]]: The set of user_ids whose devices have changed - since `from_key` - """ - from_key = int(from_key) - - # Get set of users who *may* have changed. Users not in the returned - # list have definitely not changed. - to_check = list( - self._device_list_stream_cache.get_entities_changed(user_ids, from_key) - ) - - if not to_check: - return defer.succeed(set()) - - def _get_users_whose_devices_changed_txn(txn): - changes = set() - - sql = """ - SELECT DISTINCT user_id FROM device_lists_stream - WHERE stream_id > ? - AND - """ - - for chunk in batch_iter(to_check, 100): - clause, args = make_in_list_sql_clause( - txn.database_engine, "user_id", chunk - ) - txn.execute(sql + clause, (from_key,) + tuple(args)) - changes.update(user_id for user_id, in txn) - - return changes - - return self.runInteraction( - "get_users_whose_devices_changed", _get_users_whose_devices_changed_txn - ) - - def get_all_device_list_changes_for_remotes(self, from_key, to_key): - """Return a list of `(stream_id, user_id, destination)` which is the - combined list of changes to devices, and which destinations need to be - poked. `destination` may be None if no destinations need to be poked. - """ - # We do a group by here as there can be a large number of duplicate - # entries, since we throw away device IDs. - sql = """ - SELECT MAX(stream_id) AS stream_id, user_id, destination - FROM device_lists_stream - LEFT JOIN device_lists_outbound_pokes USING (stream_id, user_id, device_id) - WHERE ? < stream_id AND stream_id <= ? - GROUP BY user_id, destination - """ - return self._execute( - "get_all_device_list_changes_for_remotes", None, sql, from_key, to_key - ) - - @cached(max_entries=10000) - def get_device_list_last_stream_id_for_remote(self, user_id): - """Get the last stream_id we got for a user. May be None if we haven't - got any information for them. - """ - return self._simple_select_one_onecol( - table="device_lists_remote_extremeties", - keyvalues={"user_id": user_id}, - retcol="stream_id", - desc="get_device_list_last_stream_id_for_remote", - allow_none=True, - ) - - @cachedList( - cached_method_name="get_device_list_last_stream_id_for_remote", - list_name="user_ids", - inlineCallbacks=True, - ) - def get_device_list_last_stream_id_for_remotes(self, user_ids): - rows = yield self._simple_select_many_batch( - table="device_lists_remote_extremeties", - column="user_id", - iterable=user_ids, - retcols=("user_id", "stream_id"), - desc="get_device_list_last_stream_id_for_remotes", - ) - - results = {user_id: None for user_id in user_ids} - results.update({row["user_id"]: row["stream_id"] for row in rows}) - - return results - - -class DeviceBackgroundUpdateStore(BackgroundUpdateStore): - def __init__(self, db_conn, hs): - super(DeviceBackgroundUpdateStore, self).__init__(db_conn, hs) - - self.register_background_index_update( - "device_lists_stream_idx", - index_name="device_lists_stream_user_id", - table="device_lists_stream", - columns=["user_id", "device_id"], - ) - - # create a unique index on device_lists_remote_cache - self.register_background_index_update( - "device_lists_remote_cache_unique_idx", - index_name="device_lists_remote_cache_unique_id", - table="device_lists_remote_cache", - columns=["user_id", "device_id"], - unique=True, - ) - - # And one on device_lists_remote_extremeties - self.register_background_index_update( - "device_lists_remote_extremeties_unique_idx", - index_name="device_lists_remote_extremeties_unique_idx", - table="device_lists_remote_extremeties", - columns=["user_id"], - unique=True, - ) - - # once they complete, we can remove the old non-unique indexes. - self.register_background_update_handler( - DROP_DEVICE_LIST_STREAMS_NON_UNIQUE_INDEXES, - self._drop_device_list_streams_non_unique_indexes, - ) - - @defer.inlineCallbacks - def _drop_device_list_streams_non_unique_indexes(self, progress, batch_size): - def f(conn): - txn = conn.cursor() - txn.execute("DROP INDEX IF EXISTS device_lists_remote_cache_id") - txn.execute("DROP INDEX IF EXISTS device_lists_remote_extremeties_id") - txn.close() - - yield self.runWithConnection(f) - yield self._end_background_update(DROP_DEVICE_LIST_STREAMS_NON_UNIQUE_INDEXES) - return 1 - - -class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): - def __init__(self, db_conn, hs): - super(DeviceStore, self).__init__(db_conn, hs) - - # Map of (user_id, device_id) -> bool. If there is an entry that implies - # the device exists. - self.device_id_exists_cache = Cache( - name="device_id_exists", keylen=2, max_entries=10000 - ) - - self._clock.looping_call(self._prune_old_outbound_device_pokes, 60 * 60 * 1000) - - @defer.inlineCallbacks - def store_device(self, user_id, device_id, initial_device_display_name): - """Ensure the given device is known; add it to the store if not - - Args: - user_id (str): id of user associated with the device - device_id (str): id of device - initial_device_display_name (str): initial displayname of the - device. Ignored if device exists. - Returns: - defer.Deferred: boolean whether the device was inserted or an - existing device existed with that ID. - """ - key = (user_id, device_id) - if self.device_id_exists_cache.get(key, None): - return False - - try: - inserted = yield self._simple_insert( - "devices", - values={ - "user_id": user_id, - "device_id": device_id, - "display_name": initial_device_display_name, - }, - desc="store_device", - or_ignore=True, - ) - self.device_id_exists_cache.prefill(key, True) - return inserted - except Exception as e: - logger.error( - "store_device with device_id=%s(%r) user_id=%s(%r)" - " display_name=%s(%r) failed: %s", - type(device_id).__name__, - device_id, - type(user_id).__name__, - user_id, - type(initial_device_display_name).__name__, - initial_device_display_name, - e, - ) - raise StoreError(500, "Problem storing device.") - - @defer.inlineCallbacks - def delete_device(self, user_id, device_id): - """Delete a device. - - Args: - user_id (str): The ID of the user which owns the device - device_id (str): The ID of the device to delete - Returns: - defer.Deferred - """ - yield self._simple_delete_one( - table="devices", - keyvalues={"user_id": user_id, "device_id": device_id}, - desc="delete_device", - ) - - self.device_id_exists_cache.invalidate((user_id, device_id)) - - @defer.inlineCallbacks - def delete_devices(self, user_id, device_ids): - """Deletes several devices. - - Args: - user_id (str): The ID of the user which owns the devices - device_ids (list): The IDs of the devices to delete - Returns: - defer.Deferred - """ - yield self._simple_delete_many( - table="devices", - column="device_id", - iterable=device_ids, - keyvalues={"user_id": user_id}, - desc="delete_devices", - ) - for device_id in device_ids: - self.device_id_exists_cache.invalidate((user_id, device_id)) - - def update_device(self, user_id, device_id, new_display_name=None): - """Update a device. - - Args: - user_id (str): The ID of the user which owns the device - device_id (str): The ID of the device to update - new_display_name (str|None): new displayname for device; None - to leave unchanged - Raises: - StoreError: if the device is not found - Returns: - defer.Deferred - """ - updates = {} - if new_display_name is not None: - updates["display_name"] = new_display_name - if not updates: - return defer.succeed(None) - return self._simple_update_one( - table="devices", - keyvalues={"user_id": user_id, "device_id": device_id}, - updatevalues=updates, - desc="update_device", - ) - - @defer.inlineCallbacks - def mark_remote_user_device_list_as_unsubscribed(self, user_id): - """Mark that we no longer track device lists for remote user. - """ - yield self._simple_delete( - table="device_lists_remote_extremeties", - keyvalues={"user_id": user_id}, - desc="mark_remote_user_device_list_as_unsubscribed", - ) - self.get_device_list_last_stream_id_for_remote.invalidate((user_id,)) - - def update_remote_device_list_cache_entry( - self, user_id, device_id, content, stream_id - ): - """Updates a single device in the cache of a remote user's devicelist. - - Note: assumes that we are the only thread that can be updating this user's - device list. - - Args: - user_id (str): User to update device list for - device_id (str): ID of decivice being updated - content (dict): new data on this device - stream_id (int): the version of the device list - - Returns: - Deferred[None] - """ - return self.runInteraction( - "update_remote_device_list_cache_entry", - self._update_remote_device_list_cache_entry_txn, - user_id, - device_id, - content, - stream_id, - ) - - def _update_remote_device_list_cache_entry_txn( - self, txn, user_id, device_id, content, stream_id - ): - if content.get("deleted"): - self._simple_delete_txn( - txn, - table="device_lists_remote_cache", - keyvalues={"user_id": user_id, "device_id": device_id}, - ) - - txn.call_after(self.device_id_exists_cache.invalidate, (user_id, device_id)) - else: - self._simple_upsert_txn( - txn, - table="device_lists_remote_cache", - keyvalues={"user_id": user_id, "device_id": device_id}, - values={"content": json.dumps(content)}, - # we don't need to lock, because we assume we are the only thread - # updating this user's devices. - lock=False, - ) - - txn.call_after(self._get_cached_user_device.invalidate, (user_id, device_id)) - txn.call_after(self._get_cached_devices_for_user.invalidate, (user_id,)) - txn.call_after( - self.get_device_list_last_stream_id_for_remote.invalidate, (user_id,) - ) - - self._simple_upsert_txn( - txn, - table="device_lists_remote_extremeties", - keyvalues={"user_id": user_id}, - values={"stream_id": stream_id}, - # again, we can assume we are the only thread updating this user's - # extremity. - lock=False, - ) - - def update_remote_device_list_cache(self, user_id, devices, stream_id): - """Replace the entire cache of the remote user's devices. - - Note: assumes that we are the only thread that can be updating this user's - device list. - - Args: - user_id (str): User to update device list for - devices (list[dict]): list of device objects supplied over federation - stream_id (int): the version of the device list - - Returns: - Deferred[None] - """ - return self.runInteraction( - "update_remote_device_list_cache", - self._update_remote_device_list_cache_txn, - user_id, - devices, - stream_id, - ) - - def _update_remote_device_list_cache_txn(self, txn, user_id, devices, stream_id): - self._simple_delete_txn( - txn, table="device_lists_remote_cache", keyvalues={"user_id": user_id} - ) - - self._simple_insert_many_txn( - txn, - table="device_lists_remote_cache", - values=[ - { - "user_id": user_id, - "device_id": content["device_id"], - "content": json.dumps(content), - } - for content in devices - ], - ) - - txn.call_after(self._get_cached_devices_for_user.invalidate, (user_id,)) - txn.call_after(self._get_cached_user_device.invalidate_many, (user_id,)) - txn.call_after( - self.get_device_list_last_stream_id_for_remote.invalidate, (user_id,) - ) - - self._simple_upsert_txn( - txn, - table="device_lists_remote_extremeties", - keyvalues={"user_id": user_id}, - values={"stream_id": stream_id}, - # we don't need to lock, because we can assume we are the only thread - # updating this user's extremity. - lock=False, - ) - - @defer.inlineCallbacks - def add_device_change_to_streams(self, user_id, device_ids, hosts): - """Persist that a user's devices have been updated, and which hosts - (if any) should be poked. - """ - with self._device_list_id_gen.get_next() as stream_id: - yield self.runInteraction( - "add_device_change_to_streams", - self._add_device_change_txn, - user_id, - device_ids, - hosts, - stream_id, - ) - return stream_id - - def _add_device_change_txn(self, txn, user_id, device_ids, hosts, stream_id): - now = self._clock.time_msec() - - txn.call_after( - self._device_list_stream_cache.entity_has_changed, user_id, stream_id - ) - for host in hosts: - txn.call_after( - self._device_list_federation_stream_cache.entity_has_changed, - host, - stream_id, - ) - - # Delete older entries in the table, as we really only care about - # when the latest change happened. - txn.executemany( - """ - DELETE FROM device_lists_stream - WHERE user_id = ? AND device_id = ? AND stream_id < ? - """, - [(user_id, device_id, stream_id) for device_id in device_ids], - ) - - self._simple_insert_many_txn( - txn, - table="device_lists_stream", - values=[ - {"stream_id": stream_id, "user_id": user_id, "device_id": device_id} - for device_id in device_ids - ], - ) - - context = get_active_span_text_map() - - self._simple_insert_many_txn( - txn, - table="device_lists_outbound_pokes", - values=[ - { - "destination": destination, - "stream_id": stream_id, - "user_id": user_id, - "device_id": device_id, - "sent": False, - "ts": now, - "opentracing_context": json.dumps(context) - if whitelisted_homeserver(destination) - else "{}", - } - for destination in hosts - for device_id in device_ids - ], - ) - - def _prune_old_outbound_device_pokes(self): - """Delete old entries out of the device_lists_outbound_pokes to ensure - that we don't fill up due to dead servers. We keep one entry per - (destination, user_id) tuple to ensure that the prev_ids remain correct - if the server does come back. - """ - yesterday = self._clock.time_msec() - 24 * 60 * 60 * 1000 - - def _prune_txn(txn): - select_sql = """ - SELECT destination, user_id, max(stream_id) as stream_id - FROM device_lists_outbound_pokes - GROUP BY destination, user_id - HAVING min(ts) < ? AND count(*) > 1 - """ - - txn.execute(select_sql, (yesterday,)) - rows = txn.fetchall() - - if not rows: - return - - delete_sql = """ - DELETE FROM device_lists_outbound_pokes - WHERE ts < ? AND destination = ? AND user_id = ? AND stream_id < ? - """ - - txn.executemany( - delete_sql, ((yesterday, row[0], row[1], row[2]) for row in rows) - ) - - # Since we've deleted unsent deltas, we need to remove the entry - # of last successful sent so that the prev_ids are correctly set. - sql = """ - DELETE FROM device_lists_outbound_last_success - WHERE destination = ? AND user_id = ? - """ - txn.executemany(sql, ((row[0], row[1]) for row in rows)) - - logger.info("Pruned %d device list outbound pokes", txn.rowcount) - - return run_as_background_process( - "prune_old_outbound_device_pokes", - self.runInteraction, - "_prune_old_outbound_device_pokes", - _prune_txn, - ) diff --git a/synapse/storage/directory.py b/synapse/storage/directory.py deleted file mode 100644 index eed7757ed5..0000000000 --- a/synapse/storage/directory.py +++ /dev/null @@ -1,174 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2014-2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -from collections import namedtuple - -from twisted.internet import defer - -from synapse.api.errors import SynapseError -from synapse.util.caches.descriptors import cached - -from ._base import SQLBaseStore - -RoomAliasMapping = namedtuple("RoomAliasMapping", ("room_id", "room_alias", "servers")) - - -class DirectoryWorkerStore(SQLBaseStore): - @defer.inlineCallbacks - def get_association_from_room_alias(self, room_alias): - """ Get's the room_id and server list for a given room_alias - - Args: - room_alias (RoomAlias) - - Returns: - Deferred: results in namedtuple with keys "room_id" and - "servers" or None if no association can be found - """ - room_id = yield self._simple_select_one_onecol( - "room_aliases", - {"room_alias": room_alias.to_string()}, - "room_id", - allow_none=True, - desc="get_association_from_room_alias", - ) - - if not room_id: - return None - - servers = yield self._simple_select_onecol( - "room_alias_servers", - {"room_alias": room_alias.to_string()}, - "server", - desc="get_association_from_room_alias", - ) - - if not servers: - return None - - return RoomAliasMapping(room_id, room_alias.to_string(), servers) - - def get_room_alias_creator(self, room_alias): - return self._simple_select_one_onecol( - table="room_aliases", - keyvalues={"room_alias": room_alias}, - retcol="creator", - desc="get_room_alias_creator", - ) - - @cached(max_entries=5000) - def get_aliases_for_room(self, room_id): - return self._simple_select_onecol( - "room_aliases", - {"room_id": room_id}, - "room_alias", - desc="get_aliases_for_room", - ) - - -class DirectoryStore(DirectoryWorkerStore): - @defer.inlineCallbacks - def create_room_alias_association(self, room_alias, room_id, servers, creator=None): - """ Creates an association between a room alias and room_id/servers - - Args: - room_alias (RoomAlias) - room_id (str) - servers (list) - creator (str): Optional user_id of creator. - - Returns: - Deferred - """ - - def alias_txn(txn): - self._simple_insert_txn( - txn, - "room_aliases", - { - "room_alias": room_alias.to_string(), - "room_id": room_id, - "creator": creator, - }, - ) - - self._simple_insert_many_txn( - txn, - table="room_alias_servers", - values=[ - {"room_alias": room_alias.to_string(), "server": server} - for server in servers - ], - ) - - self._invalidate_cache_and_stream( - txn, self.get_aliases_for_room, (room_id,) - ) - - try: - ret = yield self.runInteraction("create_room_alias_association", alias_txn) - except self.database_engine.module.IntegrityError: - raise SynapseError( - 409, "Room alias %s already exists" % room_alias.to_string() - ) - return ret - - @defer.inlineCallbacks - def delete_room_alias(self, room_alias): - room_id = yield self.runInteraction( - "delete_room_alias", self._delete_room_alias_txn, room_alias - ) - - return room_id - - def _delete_room_alias_txn(self, txn, room_alias): - txn.execute( - "SELECT room_id FROM room_aliases WHERE room_alias = ?", - (room_alias.to_string(),), - ) - - res = txn.fetchone() - if res: - room_id = res[0] - else: - return None - - txn.execute( - "DELETE FROM room_aliases WHERE room_alias = ?", (room_alias.to_string(),) - ) - - txn.execute( - "DELETE FROM room_alias_servers WHERE room_alias = ?", - (room_alias.to_string(),), - ) - - self._invalidate_cache_and_stream(txn, self.get_aliases_for_room, (room_id,)) - - return room_id - - def update_aliases_for_room(self, old_room_id, new_room_id, creator): - def _update_aliases_for_room_txn(txn): - sql = "UPDATE room_aliases SET room_id = ?, creator = ? WHERE room_id = ?" - txn.execute(sql, (new_room_id, creator, old_room_id)) - self._invalidate_cache_and_stream( - txn, self.get_aliases_for_room, (old_room_id,) - ) - self._invalidate_cache_and_stream( - txn, self.get_aliases_for_room, (new_room_id,) - ) - - return self.runInteraction( - "_update_aliases_for_room_txn", _update_aliases_for_room_txn - ) diff --git a/synapse/storage/e2e_room_keys.py b/synapse/storage/e2e_room_keys.py deleted file mode 100644 index be2fe2bab6..0000000000 --- a/synapse/storage/e2e_room_keys.py +++ /dev/null @@ -1,337 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2017 New Vector Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import json - -from twisted.internet import defer - -from synapse.api.errors import StoreError -from synapse.logging.opentracing import log_kv, trace - -from ._base import SQLBaseStore - - -class EndToEndRoomKeyStore(SQLBaseStore): - @defer.inlineCallbacks - def get_e2e_room_key(self, user_id, version, room_id, session_id): - """Get the encrypted E2E room key for a given session from a given - backup version of room_keys. We only store the 'best' room key for a given - session at a given time, as determined by the handler. - - Args: - user_id(str): the user whose backup we're querying - version(str): the version ID of the backup for the set of keys we're querying - room_id(str): the ID of the room whose keys we're querying. - This is a bit redundant as it's implied by the session_id, but - we include for consistency with the rest of the API. - session_id(str): the session whose room_key we're querying. - - Returns: - A deferred dict giving the session_data and message metadata for - this room key. - """ - - row = yield self._simple_select_one( - table="e2e_room_keys", - keyvalues={ - "user_id": user_id, - "version": version, - "room_id": room_id, - "session_id": session_id, - }, - retcols=( - "first_message_index", - "forwarded_count", - "is_verified", - "session_data", - ), - desc="get_e2e_room_key", - ) - - row["session_data"] = json.loads(row["session_data"]) - - return row - - @defer.inlineCallbacks - def set_e2e_room_key(self, user_id, version, room_id, session_id, room_key): - """Replaces or inserts the encrypted E2E room key for a given session in - a given backup - - Args: - user_id(str): the user whose backup we're setting - version(str): the version ID of the backup we're updating - room_id(str): the ID of the room whose keys we're setting - session_id(str): the session whose room_key we're setting - room_key(dict): the room_key being set - Raises: - StoreError - """ - - yield self._simple_upsert( - table="e2e_room_keys", - keyvalues={ - "user_id": user_id, - "version": version, - "room_id": room_id, - "session_id": session_id, - }, - values={ - "first_message_index": room_key["first_message_index"], - "forwarded_count": room_key["forwarded_count"], - "is_verified": room_key["is_verified"], - "session_data": json.dumps(room_key["session_data"]), - }, - lock=False, - ) - log_kv( - { - "message": "Set room key", - "room_id": room_id, - "session_id": session_id, - "room_key": room_key, - } - ) - - @trace - @defer.inlineCallbacks - def get_e2e_room_keys(self, user_id, version, room_id=None, session_id=None): - """Bulk get the E2E room keys for a given backup, optionally filtered to a given - room, or a given session. - - Args: - user_id(str): the user whose backup we're querying - version(str): the version ID of the backup for the set of keys we're querying - room_id(str): Optional. the ID of the room whose keys we're querying, if any. - If not specified, we return the keys for all the rooms in the backup. - session_id(str): Optional. the session whose room_key we're querying, if any. - If specified, we also require the room_id to be specified. - If not specified, we return all the keys in this version of - the backup (or for the specified room) - - Returns: - A deferred list of dicts giving the session_data and message metadata for - these room keys. - """ - - try: - version = int(version) - except ValueError: - return {"rooms": {}} - - keyvalues = {"user_id": user_id, "version": version} - if room_id: - keyvalues["room_id"] = room_id - if session_id: - keyvalues["session_id"] = session_id - - rows = yield self._simple_select_list( - table="e2e_room_keys", - keyvalues=keyvalues, - retcols=( - "user_id", - "room_id", - "session_id", - "first_message_index", - "forwarded_count", - "is_verified", - "session_data", - ), - desc="get_e2e_room_keys", - ) - - sessions = {"rooms": {}} - for row in rows: - room_entry = sessions["rooms"].setdefault(row["room_id"], {"sessions": {}}) - room_entry["sessions"][row["session_id"]] = { - "first_message_index": row["first_message_index"], - "forwarded_count": row["forwarded_count"], - "is_verified": row["is_verified"], - "session_data": json.loads(row["session_data"]), - } - - return sessions - - @trace - @defer.inlineCallbacks - def delete_e2e_room_keys(self, user_id, version, room_id=None, session_id=None): - """Bulk delete the E2E room keys for a given backup, optionally filtered to a given - room or a given session. - - Args: - user_id(str): the user whose backup we're deleting from - version(str): the version ID of the backup for the set of keys we're deleting - room_id(str): Optional. the ID of the room whose keys we're deleting, if any. - If not specified, we delete the keys for all the rooms in the backup. - session_id(str): Optional. the session whose room_key we're querying, if any. - If specified, we also require the room_id to be specified. - If not specified, we delete all the keys in this version of - the backup (or for the specified room) - - Returns: - A deferred of the deletion transaction - """ - - keyvalues = {"user_id": user_id, "version": int(version)} - if room_id: - keyvalues["room_id"] = room_id - if session_id: - keyvalues["session_id"] = session_id - - yield self._simple_delete( - table="e2e_room_keys", keyvalues=keyvalues, desc="delete_e2e_room_keys" - ) - - @staticmethod - def _get_current_version(txn, user_id): - txn.execute( - "SELECT MAX(version) FROM e2e_room_keys_versions " - "WHERE user_id=? AND deleted=0", - (user_id,), - ) - row = txn.fetchone() - if not row: - raise StoreError(404, "No current backup version") - return row[0] - - def get_e2e_room_keys_version_info(self, user_id, version=None): - """Get info metadata about a version of our room_keys backup. - - Args: - user_id(str): the user whose backup we're querying - version(str): Optional. the version ID of the backup we're querying about - If missing, we return the information about the current version. - Raises: - StoreError: with code 404 if there are no e2e_room_keys_versions present - Returns: - A deferred dict giving the info metadata for this backup version, with - fields including: - version(str) - algorithm(str) - auth_data(object): opaque dict supplied by the client - """ - - def _get_e2e_room_keys_version_info_txn(txn): - if version is None: - this_version = self._get_current_version(txn, user_id) - else: - try: - this_version = int(version) - except ValueError: - # Our versions are all ints so if we can't convert it to an integer, - # it isn't there. - raise StoreError(404, "No row found") - - result = self._simple_select_one_txn( - txn, - table="e2e_room_keys_versions", - keyvalues={"user_id": user_id, "version": this_version, "deleted": 0}, - retcols=("version", "algorithm", "auth_data"), - ) - result["auth_data"] = json.loads(result["auth_data"]) - result["version"] = str(result["version"]) - return result - - return self.runInteraction( - "get_e2e_room_keys_version_info", _get_e2e_room_keys_version_info_txn - ) - - @trace - def create_e2e_room_keys_version(self, user_id, info): - """Atomically creates a new version of this user's e2e_room_keys store - with the given version info. - - Args: - user_id(str): the user whose backup we're creating a version - info(dict): the info about the backup version to be created - - Returns: - A deferred string for the newly created version ID - """ - - def _create_e2e_room_keys_version_txn(txn): - txn.execute( - "SELECT MAX(version) FROM e2e_room_keys_versions WHERE user_id=?", - (user_id,), - ) - current_version = txn.fetchone()[0] - if current_version is None: - current_version = "0" - - new_version = str(int(current_version) + 1) - - self._simple_insert_txn( - txn, - table="e2e_room_keys_versions", - values={ - "user_id": user_id, - "version": new_version, - "algorithm": info["algorithm"], - "auth_data": json.dumps(info["auth_data"]), - }, - ) - - return new_version - - return self.runInteraction( - "create_e2e_room_keys_version_txn", _create_e2e_room_keys_version_txn - ) - - @trace - def update_e2e_room_keys_version(self, user_id, version, info): - """Update a given backup version - - Args: - user_id(str): the user whose backup version we're updating - version(str): the version ID of the backup version we're updating - info(dict): the new backup version info to store - """ - - return self._simple_update( - table="e2e_room_keys_versions", - keyvalues={"user_id": user_id, "version": version}, - updatevalues={"auth_data": json.dumps(info["auth_data"])}, - desc="update_e2e_room_keys_version", - ) - - @trace - def delete_e2e_room_keys_version(self, user_id, version=None): - """Delete a given backup version of the user's room keys. - Doesn't delete their actual key data. - - Args: - user_id(str): the user whose backup version we're deleting - version(str): Optional. the version ID of the backup version we're deleting - If missing, we delete the current backup version info. - Raises: - StoreError: with code 404 if there are no e2e_room_keys_versions present, - or if the version requested doesn't exist. - """ - - def _delete_e2e_room_keys_version_txn(txn): - if version is None: - this_version = self._get_current_version(txn, user_id) - else: - this_version = version - - return self._simple_update_one_txn( - txn, - table="e2e_room_keys_versions", - keyvalues={"user_id": user_id, "version": this_version}, - updatevalues={"deleted": 1}, - ) - - return self.runInteraction( - "delete_e2e_room_keys_version", _delete_e2e_room_keys_version_txn - ) diff --git a/synapse/storage/end_to_end_keys.py b/synapse/storage/end_to_end_keys.py deleted file mode 100644 index 872bc75490..0000000000 --- a/synapse/storage/end_to_end_keys.py +++ /dev/null @@ -1,323 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2015, 2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. -from six import iteritems - -from canonicaljson import encode_canonical_json - -from twisted.internet import defer - -from synapse.logging.opentracing import log_kv, set_tag, trace -from synapse.util.caches.descriptors import cached - -from ._base import SQLBaseStore, db_to_json - - -class EndToEndKeyWorkerStore(SQLBaseStore): - @trace - @defer.inlineCallbacks - def get_e2e_device_keys( - self, query_list, include_all_devices=False, include_deleted_devices=False - ): - """Fetch a list of device keys. - Args: - query_list(list): List of pairs of user_ids and device_ids. - include_all_devices (bool): whether to include entries for devices - that don't have device keys - include_deleted_devices (bool): whether to include null entries for - devices which no longer exist (but were in the query_list). - This option only takes effect if include_all_devices is true. - Returns: - Dict mapping from user-id to dict mapping from device_id to - key data. The key data will be a dict in the same format as the - DeviceKeys type returned by POST /_matrix/client/r0/keys/query. - """ - set_tag("query_list", query_list) - if not query_list: - return {} - - results = yield self.runInteraction( - "get_e2e_device_keys", - self._get_e2e_device_keys_txn, - query_list, - include_all_devices, - include_deleted_devices, - ) - - # Build the result structure, un-jsonify the results, and add the - # "unsigned" section - rv = {} - for user_id, device_keys in iteritems(results): - rv[user_id] = {} - for device_id, device_info in iteritems(device_keys): - r = db_to_json(device_info.pop("key_json")) - r["unsigned"] = {} - display_name = device_info["device_display_name"] - if display_name is not None: - r["unsigned"]["device_display_name"] = display_name - rv[user_id][device_id] = r - - return rv - - @trace - def _get_e2e_device_keys_txn( - self, txn, query_list, include_all_devices=False, include_deleted_devices=False - ): - set_tag("include_all_devices", include_all_devices) - set_tag("include_deleted_devices", include_deleted_devices) - - query_clauses = [] - query_params = [] - - if include_all_devices is False: - include_deleted_devices = False - - if include_deleted_devices: - deleted_devices = set(query_list) - - for (user_id, device_id) in query_list: - query_clause = "user_id = ?" - query_params.append(user_id) - - if device_id is not None: - query_clause += " AND device_id = ?" - query_params.append(device_id) - - query_clauses.append(query_clause) - - sql = ( - "SELECT user_id, device_id, " - " d.display_name AS device_display_name, " - " k.key_json" - " FROM devices d" - " %s JOIN e2e_device_keys_json k USING (user_id, device_id)" - " WHERE %s" - ) % ( - "LEFT" if include_all_devices else "INNER", - " OR ".join("(" + q + ")" for q in query_clauses), - ) - - txn.execute(sql, query_params) - rows = self.cursor_to_dict(txn) - - result = {} - for row in rows: - if include_deleted_devices: - deleted_devices.remove((row["user_id"], row["device_id"])) - result.setdefault(row["user_id"], {})[row["device_id"]] = row - - if include_deleted_devices: - for user_id, device_id in deleted_devices: - result.setdefault(user_id, {})[device_id] = None - - log_kv(result) - return result - - @defer.inlineCallbacks - def get_e2e_one_time_keys(self, user_id, device_id, key_ids): - """Retrieve a number of one-time keys for a user - - Args: - user_id(str): id of user to get keys for - device_id(str): id of device to get keys for - key_ids(list[str]): list of key ids (excluding algorithm) to - retrieve - - Returns: - deferred resolving to Dict[(str, str), str]: map from (algorithm, - key_id) to json string for key - """ - - rows = yield self._simple_select_many_batch( - table="e2e_one_time_keys_json", - column="key_id", - iterable=key_ids, - retcols=("algorithm", "key_id", "key_json"), - keyvalues={"user_id": user_id, "device_id": device_id}, - desc="add_e2e_one_time_keys_check", - ) - result = {(row["algorithm"], row["key_id"]): row["key_json"] for row in rows} - log_kv({"message": "Fetched one time keys for user", "one_time_keys": result}) - return result - - @defer.inlineCallbacks - def add_e2e_one_time_keys(self, user_id, device_id, time_now, new_keys): - """Insert some new one time keys for a device. Errors if any of the - keys already exist. - - Args: - user_id(str): id of user to get keys for - device_id(str): id of device to get keys for - time_now(long): insertion time to record (ms since epoch) - new_keys(iterable[(str, str, str)]: keys to add - each a tuple of - (algorithm, key_id, key json) - """ - - def _add_e2e_one_time_keys(txn): - set_tag("user_id", user_id) - set_tag("device_id", device_id) - set_tag("new_keys", new_keys) - # We are protected from race between lookup and insertion due to - # a unique constraint. If there is a race of two calls to - # `add_e2e_one_time_keys` then they'll conflict and we will only - # insert one set. - self._simple_insert_many_txn( - txn, - table="e2e_one_time_keys_json", - values=[ - { - "user_id": user_id, - "device_id": device_id, - "algorithm": algorithm, - "key_id": key_id, - "ts_added_ms": time_now, - "key_json": json_bytes, - } - for algorithm, key_id, json_bytes in new_keys - ], - ) - self._invalidate_cache_and_stream( - txn, self.count_e2e_one_time_keys, (user_id, device_id) - ) - - yield self.runInteraction( - "add_e2e_one_time_keys_insert", _add_e2e_one_time_keys - ) - - @cached(max_entries=10000) - def count_e2e_one_time_keys(self, user_id, device_id): - """ Count the number of one time keys the server has for a device - Returns: - Dict mapping from algorithm to number of keys for that algorithm. - """ - - def _count_e2e_one_time_keys(txn): - sql = ( - "SELECT algorithm, COUNT(key_id) FROM e2e_one_time_keys_json" - " WHERE user_id = ? AND device_id = ?" - " GROUP BY algorithm" - ) - txn.execute(sql, (user_id, device_id)) - result = {} - for algorithm, key_count in txn: - result[algorithm] = key_count - return result - - return self.runInteraction("count_e2e_one_time_keys", _count_e2e_one_time_keys) - - -class EndToEndKeyStore(EndToEndKeyWorkerStore, SQLBaseStore): - def set_e2e_device_keys(self, user_id, device_id, time_now, device_keys): - """Stores device keys for a device. Returns whether there was a change - or the keys were already in the database. - """ - - def _set_e2e_device_keys_txn(txn): - set_tag("user_id", user_id) - set_tag("device_id", device_id) - set_tag("time_now", time_now) - set_tag("device_keys", device_keys) - - old_key_json = self._simple_select_one_onecol_txn( - txn, - table="e2e_device_keys_json", - keyvalues={"user_id": user_id, "device_id": device_id}, - retcol="key_json", - allow_none=True, - ) - - # In py3 we need old_key_json to match new_key_json type. The DB - # returns unicode while encode_canonical_json returns bytes. - new_key_json = encode_canonical_json(device_keys).decode("utf-8") - - if old_key_json == new_key_json: - log_kv({"Message": "Device key already stored."}) - return False - - self._simple_upsert_txn( - txn, - table="e2e_device_keys_json", - keyvalues={"user_id": user_id, "device_id": device_id}, - values={"ts_added_ms": time_now, "key_json": new_key_json}, - ) - log_kv({"message": "Device keys stored."}) - return True - - return self.runInteraction("set_e2e_device_keys", _set_e2e_device_keys_txn) - - def claim_e2e_one_time_keys(self, query_list): - """Take a list of one time keys out of the database""" - - @trace - def _claim_e2e_one_time_keys(txn): - sql = ( - "SELECT key_id, key_json FROM e2e_one_time_keys_json" - " WHERE user_id = ? AND device_id = ? AND algorithm = ?" - " LIMIT 1" - ) - result = {} - delete = [] - for user_id, device_id, algorithm in query_list: - user_result = result.setdefault(user_id, {}) - device_result = user_result.setdefault(device_id, {}) - txn.execute(sql, (user_id, device_id, algorithm)) - for key_id, key_json in txn: - device_result[algorithm + ":" + key_id] = key_json - delete.append((user_id, device_id, algorithm, key_id)) - sql = ( - "DELETE FROM e2e_one_time_keys_json" - " WHERE user_id = ? AND device_id = ? AND algorithm = ?" - " AND key_id = ?" - ) - for user_id, device_id, algorithm, key_id in delete: - log_kv( - { - "message": "Executing claim e2e_one_time_keys transaction on database." - } - ) - txn.execute(sql, (user_id, device_id, algorithm, key_id)) - log_kv({"message": "finished executing and invalidating cache"}) - self._invalidate_cache_and_stream( - txn, self.count_e2e_one_time_keys, (user_id, device_id) - ) - return result - - return self.runInteraction("claim_e2e_one_time_keys", _claim_e2e_one_time_keys) - - def delete_e2e_keys_by_device(self, user_id, device_id): - def delete_e2e_keys_by_device_txn(txn): - log_kv( - { - "message": "Deleting keys for device", - "device_id": device_id, - "user_id": user_id, - } - ) - self._simple_delete_txn( - txn, - table="e2e_device_keys_json", - keyvalues={"user_id": user_id, "device_id": device_id}, - ) - self._simple_delete_txn( - txn, - table="e2e_one_time_keys_json", - keyvalues={"user_id": user_id, "device_id": device_id}, - ) - self._invalidate_cache_and_stream( - txn, self.count_e2e_one_time_keys, (user_id, device_id) - ) - - return self.runInteraction( - "delete_e2e_keys_by_device", delete_e2e_keys_by_device_txn - ) diff --git a/synapse/storage/event_federation.py b/synapse/storage/event_federation.py deleted file mode 100644 index 47cc10d32a..0000000000 --- a/synapse/storage/event_federation.py +++ /dev/null @@ -1,672 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2014-2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. -import itertools -import logging -import random - -from six.moves import range -from six.moves.queue import Empty, PriorityQueue - -from unpaddedbase64 import encode_base64 - -from twisted.internet import defer - -from synapse.api.errors import StoreError -from synapse.metrics.background_process_metrics import run_as_background_process -from synapse.storage._base import SQLBaseStore, make_in_list_sql_clause -from synapse.storage.events_worker import EventsWorkerStore -from synapse.storage.signatures import SignatureWorkerStore -from synapse.util.caches.descriptors import cached - -logger = logging.getLogger(__name__) - - -class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBaseStore): - def get_auth_chain(self, event_ids, include_given=False): - """Get auth events for given event_ids. The events *must* be state events. - - Args: - event_ids (list): state events - include_given (bool): include the given events in result - - Returns: - list of events - """ - return self.get_auth_chain_ids( - event_ids, include_given=include_given - ).addCallback(self.get_events_as_list) - - def get_auth_chain_ids(self, event_ids, include_given=False): - """Get auth events for given event_ids. The events *must* be state events. - - Args: - event_ids (list): state events - include_given (bool): include the given events in result - - Returns: - list of event_ids - """ - return self.runInteraction( - "get_auth_chain_ids", self._get_auth_chain_ids_txn, event_ids, include_given - ) - - def _get_auth_chain_ids_txn(self, txn, event_ids, include_given): - if include_given: - results = set(event_ids) - else: - results = set() - - base_sql = "SELECT auth_id FROM event_auth WHERE " - - front = set(event_ids) - while front: - new_front = set() - front_list = list(front) - chunks = [front_list[x : x + 100] for x in range(0, len(front), 100)] - for chunk in chunks: - clause, args = make_in_list_sql_clause( - txn.database_engine, "event_id", chunk - ) - txn.execute(base_sql + clause, list(args)) - new_front.update([r[0] for r in txn]) - - new_front -= results - - front = new_front - results.update(front) - - return list(results) - - def get_oldest_events_in_room(self, room_id): - return self.runInteraction( - "get_oldest_events_in_room", self._get_oldest_events_in_room_txn, room_id - ) - - def get_oldest_events_with_depth_in_room(self, room_id): - return self.runInteraction( - "get_oldest_events_with_depth_in_room", - self.get_oldest_events_with_depth_in_room_txn, - room_id, - ) - - def get_oldest_events_with_depth_in_room_txn(self, txn, room_id): - sql = ( - "SELECT b.event_id, MAX(e.depth) FROM events as e" - " INNER JOIN event_edges as g" - " ON g.event_id = e.event_id" - " INNER JOIN event_backward_extremities as b" - " ON g.prev_event_id = b.event_id" - " WHERE b.room_id = ? AND g.is_state is ?" - " GROUP BY b.event_id" - ) - - txn.execute(sql, (room_id, False)) - - return dict(txn) - - @defer.inlineCallbacks - def get_max_depth_of(self, event_ids): - """Returns the max depth of a set of event IDs - - Args: - event_ids (list[str]) - - Returns - Deferred[int] - """ - rows = yield self._simple_select_many_batch( - table="events", - column="event_id", - iterable=event_ids, - retcols=("depth",), - desc="get_max_depth_of", - ) - - if not rows: - return 0 - else: - return max(row["depth"] for row in rows) - - def _get_oldest_events_in_room_txn(self, txn, room_id): - return self._simple_select_onecol_txn( - txn, - table="event_backward_extremities", - keyvalues={"room_id": room_id}, - retcol="event_id", - ) - - @defer.inlineCallbacks - def get_prev_events_for_room(self, room_id): - """ - Gets a subset of the current forward extremities in the given room. - - Limits the result to 10 extremities, so that we can avoid creating - events which refer to hundreds of prev_events. - - Args: - room_id (str): room_id - - Returns: - Deferred[list[(str, dict[str, str], int)]] - for each event, a tuple of (event_id, hashes, depth) - where *hashes* is a map from algorithm to hash. - """ - res = yield self.get_latest_event_ids_and_hashes_in_room(room_id) - if len(res) > 10: - # Sort by reverse depth, so we point to the most recent. - res.sort(key=lambda a: -a[2]) - - # we use half of the limit for the actual most recent events, and - # the other half to randomly point to some of the older events, to - # make sure that we don't completely ignore the older events. - res = res[0:5] + random.sample(res[5:], 5) - - return res - - def get_latest_event_ids_and_hashes_in_room(self, room_id): - """ - Gets the current forward extremities in the given room - - Args: - room_id (str): room_id - - Returns: - Deferred[list[(str, dict[str, str], int)]] - for each event, a tuple of (event_id, hashes, depth) - where *hashes* is a map from algorithm to hash. - """ - - return self.runInteraction( - "get_latest_event_ids_and_hashes_in_room", - self._get_latest_event_ids_and_hashes_in_room, - room_id, - ) - - def get_rooms_with_many_extremities(self, min_count, limit, room_id_filter): - """Get the top rooms with at least N extremities. - - Args: - min_count (int): The minimum number of extremities - limit (int): The maximum number of rooms to return. - room_id_filter (iterable[str]): room_ids to exclude from the results - - Returns: - Deferred[list]: At most `limit` room IDs that have at least - `min_count` extremities, sorted by extremity count. - """ - - def _get_rooms_with_many_extremities_txn(txn): - where_clause = "1=1" - if room_id_filter: - where_clause = "room_id NOT IN (%s)" % ( - ",".join("?" for _ in room_id_filter), - ) - - sql = """ - SELECT room_id FROM event_forward_extremities - WHERE %s - GROUP BY room_id - HAVING count(*) > ? - ORDER BY count(*) DESC - LIMIT ? - """ % ( - where_clause, - ) - - query_args = list(itertools.chain(room_id_filter, [min_count, limit])) - txn.execute(sql, query_args) - return [room_id for room_id, in txn] - - return self.runInteraction( - "get_rooms_with_many_extremities", _get_rooms_with_many_extremities_txn - ) - - @cached(max_entries=5000, iterable=True) - def get_latest_event_ids_in_room(self, room_id): - return self._simple_select_onecol( - table="event_forward_extremities", - keyvalues={"room_id": room_id}, - retcol="event_id", - desc="get_latest_event_ids_in_room", - ) - - def _get_latest_event_ids_and_hashes_in_room(self, txn, room_id): - sql = ( - "SELECT e.event_id, e.depth FROM events as e " - "INNER JOIN event_forward_extremities as f " - "ON e.event_id = f.event_id " - "AND e.room_id = f.room_id " - "WHERE f.room_id = ?" - ) - - txn.execute(sql, (room_id,)) - - results = [] - for event_id, depth in txn.fetchall(): - hashes = self._get_event_reference_hashes_txn(txn, event_id) - prev_hashes = { - k: encode_base64(v) for k, v in hashes.items() if k == "sha256" - } - results.append((event_id, prev_hashes, depth)) - - return results - - def get_min_depth(self, room_id): - """ For hte given room, get the minimum depth we have seen for it. - """ - return self.runInteraction( - "get_min_depth", self._get_min_depth_interaction, room_id - ) - - def _get_min_depth_interaction(self, txn, room_id): - min_depth = self._simple_select_one_onecol_txn( - txn, - table="room_depth", - keyvalues={"room_id": room_id}, - retcol="min_depth", - allow_none=True, - ) - - return int(min_depth) if min_depth is not None else None - - def get_forward_extremeties_for_room(self, room_id, stream_ordering): - """For a given room_id and stream_ordering, return the forward - extremeties of the room at that point in "time". - - Throws a StoreError if we have since purged the index for - stream_orderings from that point. - - Args: - room_id (str): - stream_ordering (int): - - Returns: - deferred, which resolves to a list of event_ids - """ - # We want to make the cache more effective, so we clamp to the last - # change before the given ordering. - last_change = self._events_stream_cache.get_max_pos_of_last_change(room_id) - - # We don't always have a full stream_to_exterm_id table, e.g. after - # the upgrade that introduced it, so we make sure we never ask for a - # stream_ordering from before a restart - last_change = max(self._stream_order_on_start, last_change) - - # provided the last_change is recent enough, we now clamp the requested - # stream_ordering to it. - if last_change > self.stream_ordering_month_ago: - stream_ordering = min(last_change, stream_ordering) - - return self._get_forward_extremeties_for_room(room_id, stream_ordering) - - @cached(max_entries=5000, num_args=2) - def _get_forward_extremeties_for_room(self, room_id, stream_ordering): - """For a given room_id and stream_ordering, return the forward - extremeties of the room at that point in "time". - - Throws a StoreError if we have since purged the index for - stream_orderings from that point. - """ - - if stream_ordering <= self.stream_ordering_month_ago: - raise StoreError(400, "stream_ordering too old") - - sql = """ - SELECT event_id FROM stream_ordering_to_exterm - INNER JOIN ( - SELECT room_id, MAX(stream_ordering) AS stream_ordering - FROM stream_ordering_to_exterm - WHERE stream_ordering <= ? GROUP BY room_id - ) AS rms USING (room_id, stream_ordering) - WHERE room_id = ? - """ - - def get_forward_extremeties_for_room_txn(txn): - txn.execute(sql, (stream_ordering, room_id)) - return [event_id for event_id, in txn] - - return self.runInteraction( - "get_forward_extremeties_for_room", get_forward_extremeties_for_room_txn - ) - - def get_backfill_events(self, room_id, event_list, limit): - """Get a list of Events for a given topic that occurred before (and - including) the events in event_list. Return a list of max size `limit` - - Args: - txn - room_id (str) - event_list (list) - limit (int) - """ - return ( - self.runInteraction( - "get_backfill_events", - self._get_backfill_events, - room_id, - event_list, - limit, - ) - .addCallback(self.get_events_as_list) - .addCallback(lambda l: sorted(l, key=lambda e: -e.depth)) - ) - - def _get_backfill_events(self, txn, room_id, event_list, limit): - logger.debug( - "_get_backfill_events: %s, %s, %s", room_id, repr(event_list), limit - ) - - event_results = set() - - # We want to make sure that we do a breadth-first, "depth" ordered - # search. - - query = ( - "SELECT depth, prev_event_id FROM event_edges" - " INNER JOIN events" - " ON prev_event_id = events.event_id" - " WHERE event_edges.event_id = ?" - " AND event_edges.is_state = ?" - " LIMIT ?" - ) - - queue = PriorityQueue() - - for event_id in event_list: - depth = self._simple_select_one_onecol_txn( - txn, - table="events", - keyvalues={"event_id": event_id, "room_id": room_id}, - retcol="depth", - allow_none=True, - ) - - if depth: - queue.put((-depth, event_id)) - - while not queue.empty() and len(event_results) < limit: - try: - _, event_id = queue.get_nowait() - except Empty: - break - - if event_id in event_results: - continue - - event_results.add(event_id) - - txn.execute(query, (event_id, False, limit - len(event_results))) - - for row in txn: - if row[1] not in event_results: - queue.put((-row[0], row[1])) - - return event_results - - @defer.inlineCallbacks - def get_missing_events(self, room_id, earliest_events, latest_events, limit): - ids = yield self.runInteraction( - "get_missing_events", - self._get_missing_events, - room_id, - earliest_events, - latest_events, - limit, - ) - events = yield self.get_events_as_list(ids) - return events - - def _get_missing_events(self, txn, room_id, earliest_events, latest_events, limit): - - seen_events = set(earliest_events) - front = set(latest_events) - seen_events - event_results = [] - - query = ( - "SELECT prev_event_id FROM event_edges " - "WHERE room_id = ? AND event_id = ? AND is_state = ? " - "LIMIT ?" - ) - - while front and len(event_results) < limit: - new_front = set() - for event_id in front: - txn.execute( - query, (room_id, event_id, False, limit - len(event_results)) - ) - - new_results = set(t[0] for t in txn) - seen_events - - new_front |= new_results - seen_events |= new_results - event_results.extend(new_results) - - front = new_front - - # we built the list working backwards from latest_events; we now need to - # reverse it so that the events are approximately chronological. - event_results.reverse() - return event_results - - @defer.inlineCallbacks - def get_successor_events(self, event_ids): - """Fetch all events that have the given events as a prev event - - Args: - event_ids (iterable[str]) - - Returns: - Deferred[list[str]] - """ - rows = yield self._simple_select_many_batch( - table="event_edges", - column="prev_event_id", - iterable=event_ids, - retcols=("event_id",), - desc="get_successor_events", - ) - - return [row["event_id"] for row in rows] - - -class EventFederationStore(EventFederationWorkerStore): - """ Responsible for storing and serving up the various graphs associated - with an event. Including the main event graph and the auth chains for an - event. - - Also has methods for getting the front (latest) and back (oldest) edges - of the event graphs. These are used to generate the parents for new events - and backfilling from another server respectively. - """ - - EVENT_AUTH_STATE_ONLY = "event_auth_state_only" - - def __init__(self, db_conn, hs): - super(EventFederationStore, self).__init__(db_conn, hs) - - self.register_background_update_handler( - self.EVENT_AUTH_STATE_ONLY, self._background_delete_non_state_event_auth - ) - - hs.get_clock().looping_call( - self._delete_old_forward_extrem_cache, 60 * 60 * 1000 - ) - - def _update_min_depth_for_room_txn(self, txn, room_id, depth): - min_depth = self._get_min_depth_interaction(txn, room_id) - - if min_depth and depth >= min_depth: - return - - self._simple_upsert_txn( - txn, - table="room_depth", - keyvalues={"room_id": room_id}, - values={"min_depth": depth}, - ) - - def _handle_mult_prev_events(self, txn, events): - """ - For the given event, update the event edges table and forward and - backward extremities tables. - """ - self._simple_insert_many_txn( - txn, - table="event_edges", - values=[ - { - "event_id": ev.event_id, - "prev_event_id": e_id, - "room_id": ev.room_id, - "is_state": False, - } - for ev in events - for e_id in ev.prev_event_ids() - ], - ) - - self._update_backward_extremeties(txn, events) - - def _update_backward_extremeties(self, txn, events): - """Updates the event_backward_extremities tables based on the new/updated - events being persisted. - - This is called for new events *and* for events that were outliers, but - are now being persisted as non-outliers. - - Forward extremities are handled when we first start persisting the events. - """ - events_by_room = {} - for ev in events: - events_by_room.setdefault(ev.room_id, []).append(ev) - - query = ( - "INSERT INTO event_backward_extremities (event_id, room_id)" - " SELECT ?, ? WHERE NOT EXISTS (" - " SELECT 1 FROM event_backward_extremities" - " WHERE event_id = ? AND room_id = ?" - " )" - " AND NOT EXISTS (" - " SELECT 1 FROM events WHERE event_id = ? AND room_id = ? " - " AND outlier = ?" - " )" - ) - - txn.executemany( - query, - [ - (e_id, ev.room_id, e_id, ev.room_id, e_id, ev.room_id, False) - for ev in events - for e_id in ev.prev_event_ids() - if not ev.internal_metadata.is_outlier() - ], - ) - - query = ( - "DELETE FROM event_backward_extremities" - " WHERE event_id = ? AND room_id = ?" - ) - txn.executemany( - query, - [ - (ev.event_id, ev.room_id) - for ev in events - if not ev.internal_metadata.is_outlier() - ], - ) - - def _delete_old_forward_extrem_cache(self): - def _delete_old_forward_extrem_cache_txn(txn): - # Delete entries older than a month, while making sure we don't delete - # the only entries for a room. - sql = """ - DELETE FROM stream_ordering_to_exterm - WHERE - room_id IN ( - SELECT room_id - FROM stream_ordering_to_exterm - WHERE stream_ordering > ? - ) AND stream_ordering < ? - """ - txn.execute( - sql, (self.stream_ordering_month_ago, self.stream_ordering_month_ago) - ) - - return run_as_background_process( - "delete_old_forward_extrem_cache", - self.runInteraction, - "_delete_old_forward_extrem_cache", - _delete_old_forward_extrem_cache_txn, - ) - - def clean_room_for_join(self, room_id): - return self.runInteraction( - "clean_room_for_join", self._clean_room_for_join_txn, room_id - ) - - def _clean_room_for_join_txn(self, txn, room_id): - query = "DELETE FROM event_forward_extremities WHERE room_id = ?" - - txn.execute(query, (room_id,)) - txn.call_after(self.get_latest_event_ids_in_room.invalidate, (room_id,)) - - @defer.inlineCallbacks - def _background_delete_non_state_event_auth(self, progress, batch_size): - def delete_event_auth(txn): - target_min_stream_id = progress.get("target_min_stream_id_inclusive") - max_stream_id = progress.get("max_stream_id_exclusive") - - if not target_min_stream_id or not max_stream_id: - txn.execute("SELECT COALESCE(MIN(stream_ordering), 0) FROM events") - rows = txn.fetchall() - target_min_stream_id = rows[0][0] - - txn.execute("SELECT COALESCE(MAX(stream_ordering), 0) FROM events") - rows = txn.fetchall() - max_stream_id = rows[0][0] - - min_stream_id = max_stream_id - batch_size - - sql = """ - DELETE FROM event_auth - WHERE event_id IN ( - SELECT event_id FROM events - LEFT JOIN state_events USING (room_id, event_id) - WHERE ? <= stream_ordering AND stream_ordering < ? - AND state_key IS null - ) - """ - - txn.execute(sql, (min_stream_id, max_stream_id)) - - new_progress = { - "target_min_stream_id_inclusive": target_min_stream_id, - "max_stream_id_exclusive": min_stream_id, - } - - self._background_update_progress_txn( - txn, self.EVENT_AUTH_STATE_ONLY, new_progress - ) - - return min_stream_id >= target_min_stream_id - - result = yield self.runInteraction( - self.EVENT_AUTH_STATE_ONLY, delete_event_auth - ) - - if not result: - yield self._end_background_update(self.EVENT_AUTH_STATE_ONLY) - - return batch_size diff --git a/synapse/storage/event_push_actions.py b/synapse/storage/event_push_actions.py deleted file mode 100644 index 22025effbc..0000000000 --- a/synapse/storage/event_push_actions.py +++ /dev/null @@ -1,960 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2015 OpenMarket Ltd -# Copyright 2018 New Vector Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import logging - -from six import iteritems - -from canonicaljson import json - -from twisted.internet import defer - -from synapse.metrics.background_process_metrics import run_as_background_process -from synapse.storage._base import LoggingTransaction, SQLBaseStore -from synapse.util.caches.descriptors import cachedInlineCallbacks - -logger = logging.getLogger(__name__) - - -DEFAULT_NOTIF_ACTION = ["notify", {"set_tweak": "highlight", "value": False}] -DEFAULT_HIGHLIGHT_ACTION = [ - "notify", - {"set_tweak": "sound", "value": "default"}, - {"set_tweak": "highlight"}, -] - - -def _serialize_action(actions, is_highlight): - """Custom serializer for actions. This allows us to "compress" common actions. - - We use the fact that most users have the same actions for notifs (and for - highlights). - We store these default actions as the empty string rather than the full JSON. - Since the empty string isn't valid JSON there is no risk of this clashing with - any real JSON actions - """ - if is_highlight: - if actions == DEFAULT_HIGHLIGHT_ACTION: - return "" # We use empty string as the column is non-NULL - else: - if actions == DEFAULT_NOTIF_ACTION: - return "" - return json.dumps(actions) - - -def _deserialize_action(actions, is_highlight): - """Custom deserializer for actions. This allows us to "compress" common actions - """ - if actions: - return json.loads(actions) - - if is_highlight: - return DEFAULT_HIGHLIGHT_ACTION - else: - return DEFAULT_NOTIF_ACTION - - -class EventPushActionsWorkerStore(SQLBaseStore): - def __init__(self, db_conn, hs): - super(EventPushActionsWorkerStore, self).__init__(db_conn, hs) - - # These get correctly set by _find_stream_orderings_for_times_txn - self.stream_ordering_month_ago = None - self.stream_ordering_day_ago = None - - cur = LoggingTransaction( - db_conn.cursor(), - name="_find_stream_orderings_for_times_txn", - database_engine=self.database_engine, - ) - self._find_stream_orderings_for_times_txn(cur) - cur.close() - - self.find_stream_orderings_looping_call = self._clock.looping_call( - self._find_stream_orderings_for_times, 10 * 60 * 1000 - ) - self._rotate_delay = 3 - self._rotate_count = 10000 - - @cachedInlineCallbacks(num_args=3, tree=True, max_entries=5000) - def get_unread_event_push_actions_by_room_for_user( - self, room_id, user_id, last_read_event_id - ): - ret = yield self.runInteraction( - "get_unread_event_push_actions_by_room", - self._get_unread_counts_by_receipt_txn, - room_id, - user_id, - last_read_event_id, - ) - return ret - - def _get_unread_counts_by_receipt_txn( - self, txn, room_id, user_id, last_read_event_id - ): - sql = ( - "SELECT stream_ordering" - " FROM events" - " WHERE room_id = ? AND event_id = ?" - ) - txn.execute(sql, (room_id, last_read_event_id)) - results = txn.fetchall() - if len(results) == 0: - return {"notify_count": 0, "highlight_count": 0} - - stream_ordering = results[0][0] - - return self._get_unread_counts_by_pos_txn( - txn, room_id, user_id, stream_ordering - ) - - def _get_unread_counts_by_pos_txn(self, txn, room_id, user_id, stream_ordering): - - # First get number of notifications. - # We don't need to put a notif=1 clause as all rows always have - # notif=1 - sql = ( - "SELECT count(*)" - " FROM event_push_actions ea" - " WHERE" - " user_id = ?" - " AND room_id = ?" - " AND stream_ordering > ?" - ) - - txn.execute(sql, (user_id, room_id, stream_ordering)) - row = txn.fetchone() - notify_count = row[0] if row else 0 - - txn.execute( - """ - SELECT notif_count FROM event_push_summary - WHERE room_id = ? AND user_id = ? AND stream_ordering > ? - """, - (room_id, user_id, stream_ordering), - ) - rows = txn.fetchall() - if rows: - notify_count += rows[0][0] - - # Now get the number of highlights - sql = ( - "SELECT count(*)" - " FROM event_push_actions ea" - " WHERE" - " highlight = 1" - " AND user_id = ?" - " AND room_id = ?" - " AND stream_ordering > ?" - ) - - txn.execute(sql, (user_id, room_id, stream_ordering)) - row = txn.fetchone() - highlight_count = row[0] if row else 0 - - return {"notify_count": notify_count, "highlight_count": highlight_count} - - @defer.inlineCallbacks - def get_push_action_users_in_range(self, min_stream_ordering, max_stream_ordering): - def f(txn): - sql = ( - "SELECT DISTINCT(user_id) FROM event_push_actions WHERE" - " stream_ordering >= ? AND stream_ordering <= ?" - ) - txn.execute(sql, (min_stream_ordering, max_stream_ordering)) - return [r[0] for r in txn] - - ret = yield self.runInteraction("get_push_action_users_in_range", f) - return ret - - @defer.inlineCallbacks - def get_unread_push_actions_for_user_in_range_for_http( - self, user_id, min_stream_ordering, max_stream_ordering, limit=20 - ): - """Get a list of the most recent unread push actions for a given user, - within the given stream ordering range. Called by the httppusher. - - Args: - user_id (str): The user to fetch push actions for. - min_stream_ordering(int): The exclusive lower bound on the - stream ordering of event push actions to fetch. - max_stream_ordering(int): The inclusive upper bound on the - stream ordering of event push actions to fetch. - limit (int): The maximum number of rows to return. - Returns: - A promise which resolves to a list of dicts with the keys "event_id", - "room_id", "stream_ordering", "actions". - The list will be ordered by ascending stream_ordering. - The list will have between 0~limit entries. - """ - # find rooms that have a read receipt in them and return the next - # push actions - def get_after_receipt(txn): - # find rooms that have a read receipt in them and return the next - # push actions - sql = ( - "SELECT ep.event_id, ep.room_id, ep.stream_ordering, ep.actions," - " ep.highlight " - " FROM (" - " SELECT room_id," - " MAX(stream_ordering) as stream_ordering" - " FROM events" - " INNER JOIN receipts_linearized USING (room_id, event_id)" - " WHERE receipt_type = 'm.read' AND user_id = ?" - " GROUP BY room_id" - ") AS rl," - " event_push_actions AS ep" - " WHERE" - " ep.room_id = rl.room_id" - " AND ep.stream_ordering > rl.stream_ordering" - " AND ep.user_id = ?" - " AND ep.stream_ordering > ?" - " AND ep.stream_ordering <= ?" - " ORDER BY ep.stream_ordering ASC LIMIT ?" - ) - args = [user_id, user_id, min_stream_ordering, max_stream_ordering, limit] - txn.execute(sql, args) - return txn.fetchall() - - after_read_receipt = yield self.runInteraction( - "get_unread_push_actions_for_user_in_range_http_arr", get_after_receipt - ) - - # There are rooms with push actions in them but you don't have a read receipt in - # them e.g. rooms you've been invited to, so get push actions for rooms which do - # not have read receipts in them too. - def get_no_receipt(txn): - sql = ( - "SELECT ep.event_id, ep.room_id, ep.stream_ordering, ep.actions," - " ep.highlight " - " FROM event_push_actions AS ep" - " INNER JOIN events AS e USING (room_id, event_id)" - " WHERE" - " ep.room_id NOT IN (" - " SELECT room_id FROM receipts_linearized" - " WHERE receipt_type = 'm.read' AND user_id = ?" - " GROUP BY room_id" - " )" - " AND ep.user_id = ?" - " AND ep.stream_ordering > ?" - " AND ep.stream_ordering <= ?" - " ORDER BY ep.stream_ordering ASC LIMIT ?" - ) - args = [user_id, user_id, min_stream_ordering, max_stream_ordering, limit] - txn.execute(sql, args) - return txn.fetchall() - - no_read_receipt = yield self.runInteraction( - "get_unread_push_actions_for_user_in_range_http_nrr", get_no_receipt - ) - - notifs = [ - { - "event_id": row[0], - "room_id": row[1], - "stream_ordering": row[2], - "actions": _deserialize_action(row[3], row[4]), - } - for row in after_read_receipt + no_read_receipt - ] - - # Now sort it so it's ordered correctly, since currently it will - # contain results from the first query, correctly ordered, followed - # by results from the second query, but we want them all ordered - # by stream_ordering, oldest first. - notifs.sort(key=lambda r: r["stream_ordering"]) - - # Take only up to the limit. We have to stop at the limit because - # one of the subqueries may have hit the limit. - return notifs[:limit] - - @defer.inlineCallbacks - def get_unread_push_actions_for_user_in_range_for_email( - self, user_id, min_stream_ordering, max_stream_ordering, limit=20 - ): - """Get a list of the most recent unread push actions for a given user, - within the given stream ordering range. Called by the emailpusher - - Args: - user_id (str): The user to fetch push actions for. - min_stream_ordering(int): The exclusive lower bound on the - stream ordering of event push actions to fetch. - max_stream_ordering(int): The inclusive upper bound on the - stream ordering of event push actions to fetch. - limit (int): The maximum number of rows to return. - Returns: - A promise which resolves to a list of dicts with the keys "event_id", - "room_id", "stream_ordering", "actions", "received_ts". - The list will be ordered by descending received_ts. - The list will have between 0~limit entries. - """ - # find rooms that have a read receipt in them and return the most recent - # push actions - def get_after_receipt(txn): - sql = ( - "SELECT ep.event_id, ep.room_id, ep.stream_ordering, ep.actions," - " ep.highlight, e.received_ts" - " FROM (" - " SELECT room_id," - " MAX(stream_ordering) as stream_ordering" - " FROM events" - " INNER JOIN receipts_linearized USING (room_id, event_id)" - " WHERE receipt_type = 'm.read' AND user_id = ?" - " GROUP BY room_id" - ") AS rl," - " event_push_actions AS ep" - " INNER JOIN events AS e USING (room_id, event_id)" - " WHERE" - " ep.room_id = rl.room_id" - " AND ep.stream_ordering > rl.stream_ordering" - " AND ep.user_id = ?" - " AND ep.stream_ordering > ?" - " AND ep.stream_ordering <= ?" - " ORDER BY ep.stream_ordering DESC LIMIT ?" - ) - args = [user_id, user_id, min_stream_ordering, max_stream_ordering, limit] - txn.execute(sql, args) - return txn.fetchall() - - after_read_receipt = yield self.runInteraction( - "get_unread_push_actions_for_user_in_range_email_arr", get_after_receipt - ) - - # There are rooms with push actions in them but you don't have a read receipt in - # them e.g. rooms you've been invited to, so get push actions for rooms which do - # not have read receipts in them too. - def get_no_receipt(txn): - sql = ( - "SELECT ep.event_id, ep.room_id, ep.stream_ordering, ep.actions," - " ep.highlight, e.received_ts" - " FROM event_push_actions AS ep" - " INNER JOIN events AS e USING (room_id, event_id)" - " WHERE" - " ep.room_id NOT IN (" - " SELECT room_id FROM receipts_linearized" - " WHERE receipt_type = 'm.read' AND user_id = ?" - " GROUP BY room_id" - " )" - " AND ep.user_id = ?" - " AND ep.stream_ordering > ?" - " AND ep.stream_ordering <= ?" - " ORDER BY ep.stream_ordering DESC LIMIT ?" - ) - args = [user_id, user_id, min_stream_ordering, max_stream_ordering, limit] - txn.execute(sql, args) - return txn.fetchall() - - no_read_receipt = yield self.runInteraction( - "get_unread_push_actions_for_user_in_range_email_nrr", get_no_receipt - ) - - # Make a list of dicts from the two sets of results. - notifs = [ - { - "event_id": row[0], - "room_id": row[1], - "stream_ordering": row[2], - "actions": _deserialize_action(row[3], row[4]), - "received_ts": row[5], - } - for row in after_read_receipt + no_read_receipt - ] - - # Now sort it so it's ordered correctly, since currently it will - # contain results from the first query, correctly ordered, followed - # by results from the second query, but we want them all ordered - # by received_ts (most recent first) - notifs.sort(key=lambda r: -(r["received_ts"] or 0)) - - # Now return the first `limit` - return notifs[:limit] - - def get_if_maybe_push_in_range_for_user(self, user_id, min_stream_ordering): - """A fast check to see if there might be something to push for the - user since the given stream ordering. May return false positives. - - Useful to know whether to bother starting a pusher on start up or not. - - Args: - user_id (str) - min_stream_ordering (int) - - Returns: - Deferred[bool]: True if there may be push to process, False if - there definitely isn't. - """ - - def _get_if_maybe_push_in_range_for_user_txn(txn): - sql = """ - SELECT 1 FROM event_push_actions - WHERE user_id = ? AND stream_ordering > ? - LIMIT 1 - """ - - txn.execute(sql, (user_id, min_stream_ordering)) - return bool(txn.fetchone()) - - return self.runInteraction( - "get_if_maybe_push_in_range_for_user", - _get_if_maybe_push_in_range_for_user_txn, - ) - - def add_push_actions_to_staging(self, event_id, user_id_actions): - """Add the push actions for the event to the push action staging area. - - Args: - event_id (str) - user_id_actions (dict[str, list[dict|str])]): A dictionary mapping - user_id to list of push actions, where an action can either be - a string or dict. - - Returns: - Deferred - """ - - if not user_id_actions: - return - - # This is a helper function for generating the necessary tuple that - # can be used to inert into the `event_push_actions_staging` table. - def _gen_entry(user_id, actions): - is_highlight = 1 if _action_has_highlight(actions) else 0 - return ( - event_id, # event_id column - user_id, # user_id column - _serialize_action(actions, is_highlight), # actions column - 1, # notif column - is_highlight, # highlight column - ) - - def _add_push_actions_to_staging_txn(txn): - # We don't use _simple_insert_many here to avoid the overhead - # of generating lists of dicts. - - sql = """ - INSERT INTO event_push_actions_staging - (event_id, user_id, actions, notif, highlight) - VALUES (?, ?, ?, ?, ?) - """ - - txn.executemany( - sql, - ( - _gen_entry(user_id, actions) - for user_id, actions in iteritems(user_id_actions) - ), - ) - - return self.runInteraction( - "add_push_actions_to_staging", _add_push_actions_to_staging_txn - ) - - @defer.inlineCallbacks - def remove_push_actions_from_staging(self, event_id): - """Called if we failed to persist the event to ensure that stale push - actions don't build up in the DB - - Args: - event_id (str) - """ - - try: - res = yield self._simple_delete( - table="event_push_actions_staging", - keyvalues={"event_id": event_id}, - desc="remove_push_actions_from_staging", - ) - return res - except Exception: - # this method is called from an exception handler, so propagating - # another exception here really isn't helpful - there's nothing - # the caller can do about it. Just log the exception and move on. - logger.exception( - "Error removing push actions after event persistence failure" - ) - - def _find_stream_orderings_for_times(self): - return run_as_background_process( - "event_push_action_stream_orderings", - self.runInteraction, - "_find_stream_orderings_for_times", - self._find_stream_orderings_for_times_txn, - ) - - def _find_stream_orderings_for_times_txn(self, txn): - logger.info("Searching for stream ordering 1 month ago") - self.stream_ordering_month_ago = self._find_first_stream_ordering_after_ts_txn( - txn, self._clock.time_msec() - 30 * 24 * 60 * 60 * 1000 - ) - logger.info( - "Found stream ordering 1 month ago: it's %d", self.stream_ordering_month_ago - ) - logger.info("Searching for stream ordering 1 day ago") - self.stream_ordering_day_ago = self._find_first_stream_ordering_after_ts_txn( - txn, self._clock.time_msec() - 24 * 60 * 60 * 1000 - ) - logger.info( - "Found stream ordering 1 day ago: it's %d", self.stream_ordering_day_ago - ) - - def find_first_stream_ordering_after_ts(self, ts): - """Gets the stream ordering corresponding to a given timestamp. - - Specifically, finds the stream_ordering of the first event that was - received on or after the timestamp. This is done by a binary search on - the events table, since there is no index on received_ts, so is - relatively slow. - - Args: - ts (int): timestamp in millis - - Returns: - Deferred[int]: stream ordering of the first event received on/after - the timestamp - """ - return self.runInteraction( - "_find_first_stream_ordering_after_ts_txn", - self._find_first_stream_ordering_after_ts_txn, - ts, - ) - - @staticmethod - def _find_first_stream_ordering_after_ts_txn(txn, ts): - """ - Find the stream_ordering of the first event that was received on or - after a given timestamp. This is relatively slow as there is no index - on received_ts but we can then use this to delete push actions before - this. - - received_ts must necessarily be in the same order as stream_ordering - and stream_ordering is indexed, so we manually binary search using - stream_ordering - - Args: - txn (twisted.enterprise.adbapi.Transaction): - ts (int): timestamp to search for - - Returns: - int: stream ordering - """ - txn.execute("SELECT MAX(stream_ordering) FROM events") - max_stream_ordering = txn.fetchone()[0] - - if max_stream_ordering is None: - return 0 - - # We want the first stream_ordering in which received_ts is greater - # than or equal to ts. Call this point X. - # - # We maintain the invariants: - # - # range_start <= X <= range_end - # - range_start = 0 - range_end = max_stream_ordering + 1 - - # Given a stream_ordering, look up the timestamp at that - # stream_ordering. - # - # The array may be sparse (we may be missing some stream_orderings). - # We treat the gaps as the same as having the same value as the - # preceding entry, because we will pick the lowest stream_ordering - # which satisfies our requirement of received_ts >= ts. - # - # For example, if our array of events indexed by stream_ordering is - # [10, , 20], we should treat this as being equivalent to - # [10, 10, 20]. - # - sql = ( - "SELECT received_ts FROM events" - " WHERE stream_ordering <= ?" - " ORDER BY stream_ordering DESC" - " LIMIT 1" - ) - - while range_end - range_start > 0: - middle = (range_end + range_start) // 2 - txn.execute(sql, (middle,)) - row = txn.fetchone() - if row is None: - # no rows with stream_ordering<=middle - range_start = middle + 1 - continue - - middle_ts = row[0] - if ts > middle_ts: - # we got a timestamp lower than the one we were looking for. - # definitely need to look higher: X > middle. - range_start = middle + 1 - else: - # we got a timestamp higher than (or the same as) the one we - # were looking for. We aren't yet sure about the point we - # looked up, but we can be sure that X <= middle. - range_end = middle - - return range_end - - -class EventPushActionsStore(EventPushActionsWorkerStore): - EPA_HIGHLIGHT_INDEX = "epa_highlight_index" - - def __init__(self, db_conn, hs): - super(EventPushActionsStore, self).__init__(db_conn, hs) - - self.register_background_index_update( - self.EPA_HIGHLIGHT_INDEX, - index_name="event_push_actions_u_highlight", - table="event_push_actions", - columns=["user_id", "stream_ordering"], - ) - - self.register_background_index_update( - "event_push_actions_highlights_index", - index_name="event_push_actions_highlights_index", - table="event_push_actions", - columns=["user_id", "room_id", "topological_ordering", "stream_ordering"], - where_clause="highlight=1", - ) - - self._doing_notif_rotation = False - self._rotate_notif_loop = self._clock.looping_call( - self._start_rotate_notifs, 30 * 60 * 1000 - ) - - def _set_push_actions_for_event_and_users_txn( - self, txn, events_and_contexts, all_events_and_contexts - ): - """Handles moving push actions from staging table to main - event_push_actions table for all events in `events_and_contexts`. - - Also ensures that all events in `all_events_and_contexts` are removed - from the push action staging area. - - Args: - events_and_contexts (list[(EventBase, EventContext)]): events - we are persisting - all_events_and_contexts (list[(EventBase, EventContext)]): all - events that we were going to persist. This includes events - we've already persisted, etc, that wouldn't appear in - events_and_context. - """ - - sql = """ - INSERT INTO event_push_actions ( - room_id, event_id, user_id, actions, stream_ordering, - topological_ordering, notif, highlight - ) - SELECT ?, event_id, user_id, actions, ?, ?, notif, highlight - FROM event_push_actions_staging - WHERE event_id = ? - """ - - if events_and_contexts: - txn.executemany( - sql, - ( - ( - event.room_id, - event.internal_metadata.stream_ordering, - event.depth, - event.event_id, - ) - for event, _ in events_and_contexts - ), - ) - - for event, _ in events_and_contexts: - user_ids = self._simple_select_onecol_txn( - txn, - table="event_push_actions_staging", - keyvalues={"event_id": event.event_id}, - retcol="user_id", - ) - - for uid in user_ids: - txn.call_after( - self.get_unread_event_push_actions_by_room_for_user.invalidate_many, - (event.room_id, uid), - ) - - # Now we delete the staging area for *all* events that were being - # persisted. - txn.executemany( - "DELETE FROM event_push_actions_staging WHERE event_id = ?", - ((event.event_id,) for event, _ in all_events_and_contexts), - ) - - @defer.inlineCallbacks - def get_push_actions_for_user( - self, user_id, before=None, limit=50, only_highlight=False - ): - def f(txn): - before_clause = "" - if before: - before_clause = "AND epa.stream_ordering < ?" - args = [user_id, before, limit] - else: - args = [user_id, limit] - - if only_highlight: - if len(before_clause) > 0: - before_clause += " " - before_clause += "AND epa.highlight = 1" - - # NB. This assumes event_ids are globally unique since - # it makes the query easier to index - sql = ( - "SELECT epa.event_id, epa.room_id," - " epa.stream_ordering, epa.topological_ordering," - " epa.actions, epa.highlight, epa.profile_tag, e.received_ts" - " FROM event_push_actions epa, events e" - " WHERE epa.event_id = e.event_id" - " AND epa.user_id = ? %s" - " ORDER BY epa.stream_ordering DESC" - " LIMIT ?" % (before_clause,) - ) - txn.execute(sql, args) - return self.cursor_to_dict(txn) - - push_actions = yield self.runInteraction("get_push_actions_for_user", f) - for pa in push_actions: - pa["actions"] = _deserialize_action(pa["actions"], pa["highlight"]) - return push_actions - - @defer.inlineCallbacks - def get_time_of_last_push_action_before(self, stream_ordering): - def f(txn): - sql = ( - "SELECT e.received_ts" - " FROM event_push_actions AS ep" - " JOIN events e ON ep.room_id = e.room_id AND ep.event_id = e.event_id" - " WHERE ep.stream_ordering > ?" - " ORDER BY ep.stream_ordering ASC" - " LIMIT 1" - ) - txn.execute(sql, (stream_ordering,)) - return txn.fetchone() - - result = yield self.runInteraction("get_time_of_last_push_action_before", f) - return result[0] if result else None - - @defer.inlineCallbacks - def get_latest_push_action_stream_ordering(self): - def f(txn): - txn.execute("SELECT MAX(stream_ordering) FROM event_push_actions") - return txn.fetchone() - - result = yield self.runInteraction("get_latest_push_action_stream_ordering", f) - return result[0] or 0 - - def _remove_push_actions_for_event_id_txn(self, txn, room_id, event_id): - # Sad that we have to blow away the cache for the whole room here - txn.call_after( - self.get_unread_event_push_actions_by_room_for_user.invalidate_many, - (room_id,), - ) - txn.execute( - "DELETE FROM event_push_actions WHERE room_id = ? AND event_id = ?", - (room_id, event_id), - ) - - def _remove_old_push_actions_before_txn( - self, txn, room_id, user_id, stream_ordering - ): - """ - Purges old push actions for a user and room before a given - stream_ordering. - - We however keep a months worth of highlighted notifications, so that - users can still get a list of recent highlights. - - Args: - txn: The transcation - room_id: Room ID to delete from - user_id: user ID to delete for - stream_ordering: The lowest stream ordering which will - not be deleted. - """ - txn.call_after( - self.get_unread_event_push_actions_by_room_for_user.invalidate_many, - (room_id, user_id), - ) - - # We need to join on the events table to get the received_ts for - # event_push_actions and sqlite won't let us use a join in a delete so - # we can't just delete where received_ts < x. Furthermore we can - # only identify event_push_actions by a tuple of room_id, event_id - # we we can't use a subquery. - # Instead, we look up the stream ordering for the last event in that - # room received before the threshold time and delete event_push_actions - # in the room with a stream_odering before that. - txn.execute( - "DELETE FROM event_push_actions " - " WHERE user_id = ? AND room_id = ? AND " - " stream_ordering <= ?" - " AND ((stream_ordering < ? AND highlight = 1) or highlight = 0)", - (user_id, room_id, stream_ordering, self.stream_ordering_month_ago), - ) - - txn.execute( - """ - DELETE FROM event_push_summary - WHERE room_id = ? AND user_id = ? AND stream_ordering <= ? - """, - (room_id, user_id, stream_ordering), - ) - - def _start_rotate_notifs(self): - return run_as_background_process("rotate_notifs", self._rotate_notifs) - - @defer.inlineCallbacks - def _rotate_notifs(self): - if self._doing_notif_rotation or self.stream_ordering_day_ago is None: - return - self._doing_notif_rotation = True - - try: - while True: - logger.info("Rotating notifications") - - caught_up = yield self.runInteraction( - "_rotate_notifs", self._rotate_notifs_txn - ) - if caught_up: - break - yield self.hs.get_clock().sleep(self._rotate_delay) - finally: - self._doing_notif_rotation = False - - def _rotate_notifs_txn(self, txn): - """Archives older notifications into event_push_summary. Returns whether - the archiving process has caught up or not. - """ - - old_rotate_stream_ordering = self._simple_select_one_onecol_txn( - txn, - table="event_push_summary_stream_ordering", - keyvalues={}, - retcol="stream_ordering", - ) - - # We don't to try and rotate millions of rows at once, so we cap the - # maximum stream ordering we'll rotate before. - txn.execute( - """ - SELECT stream_ordering FROM event_push_actions - WHERE stream_ordering > ? - ORDER BY stream_ordering ASC LIMIT 1 OFFSET ? - """, - (old_rotate_stream_ordering, self._rotate_count), - ) - stream_row = txn.fetchone() - if stream_row: - offset_stream_ordering, = stream_row - rotate_to_stream_ordering = min( - self.stream_ordering_day_ago, offset_stream_ordering - ) - caught_up = offset_stream_ordering >= self.stream_ordering_day_ago - else: - rotate_to_stream_ordering = self.stream_ordering_day_ago - caught_up = True - - logger.info("Rotating notifications up to: %s", rotate_to_stream_ordering) - - self._rotate_notifs_before_txn(txn, rotate_to_stream_ordering) - - # We have caught up iff we were limited by `stream_ordering_day_ago` - return caught_up - - def _rotate_notifs_before_txn(self, txn, rotate_to_stream_ordering): - old_rotate_stream_ordering = self._simple_select_one_onecol_txn( - txn, - table="event_push_summary_stream_ordering", - keyvalues={}, - retcol="stream_ordering", - ) - - # Calculate the new counts that should be upserted into event_push_summary - sql = """ - SELECT user_id, room_id, - coalesce(old.notif_count, 0) + upd.notif_count, - upd.stream_ordering, - old.user_id - FROM ( - SELECT user_id, room_id, count(*) as notif_count, - max(stream_ordering) as stream_ordering - FROM event_push_actions - WHERE ? <= stream_ordering AND stream_ordering < ? - AND highlight = 0 - GROUP BY user_id, room_id - ) AS upd - LEFT JOIN event_push_summary AS old USING (user_id, room_id) - """ - - txn.execute(sql, (old_rotate_stream_ordering, rotate_to_stream_ordering)) - rows = txn.fetchall() - - logger.info("Rotating notifications, handling %d rows", len(rows)) - - # If the `old.user_id` above is NULL then we know there isn't already an - # entry in the table, so we simply insert it. Otherwise we update the - # existing table. - self._simple_insert_many_txn( - txn, - table="event_push_summary", - values=[ - { - "user_id": row[0], - "room_id": row[1], - "notif_count": row[2], - "stream_ordering": row[3], - } - for row in rows - if row[4] is None - ], - ) - - txn.executemany( - """ - UPDATE event_push_summary SET notif_count = ?, stream_ordering = ? - WHERE user_id = ? AND room_id = ? - """, - ((row[2], row[3], row[0], row[1]) for row in rows if row[4] is not None), - ) - - txn.execute( - "DELETE FROM event_push_actions" - " WHERE ? <= stream_ordering AND stream_ordering < ? AND highlight = 0", - (old_rotate_stream_ordering, rotate_to_stream_ordering), - ) - - logger.info("Rotating notifications, deleted %s push actions", txn.rowcount) - - txn.execute( - "UPDATE event_push_summary_stream_ordering SET stream_ordering = ?", - (rotate_to_stream_ordering,), - ) - - -def _action_has_highlight(actions): - for action in actions: - try: - if action.get("set_tweak", None) == "highlight": - return action.get("value", True) - except AttributeError: - pass - - return False diff --git a/synapse/storage/events.py b/synapse/storage/events.py deleted file mode 100644 index ee49ef235d..0000000000 --- a/synapse/storage/events.py +++ /dev/null @@ -1,2489 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2014-2016 OpenMarket Ltd -# Copyright 2018-2019 New Vector Ltd -# Copyright 2019 The Matrix.org Foundation C.I.C. -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import itertools -import logging -from collections import Counter as c_counter, OrderedDict, deque, namedtuple -from functools import wraps - -from six import iteritems, text_type -from six.moves import range - -from canonicaljson import json -from prometheus_client import Counter, Histogram - -from twisted.internet import defer - -import synapse.metrics -from synapse.api.constants import EventTypes -from synapse.api.errors import SynapseError -from synapse.events import EventBase # noqa: F401 -from synapse.events.snapshot import EventContext # noqa: F401 -from synapse.events.utils import prune_event_dict -from synapse.logging.context import PreserveLoggingContext, make_deferred_yieldable -from synapse.logging.utils import log_function -from synapse.metrics import BucketCollector -from synapse.metrics.background_process_metrics import run_as_background_process -from synapse.state import StateResolutionStore -from synapse.storage._base import make_in_list_sql_clause -from synapse.storage.background_updates import BackgroundUpdateStore -from synapse.storage.event_federation import EventFederationStore -from synapse.storage.events_worker import EventsWorkerStore -from synapse.storage.state import StateGroupWorkerStore -from synapse.types import RoomStreamToken, get_domain_from_id -from synapse.util import batch_iter -from synapse.util.async_helpers import ObservableDeferred -from synapse.util.caches.descriptors import cached, cachedInlineCallbacks -from synapse.util.frozenutils import frozendict_json_encoder -from synapse.util.metrics import Measure - -logger = logging.getLogger(__name__) - -persist_event_counter = Counter("synapse_storage_events_persisted_events", "") -event_counter = Counter( - "synapse_storage_events_persisted_events_sep", - "", - ["type", "origin_type", "origin_entity"], -) - -# The number of times we are recalculating the current state -state_delta_counter = Counter("synapse_storage_events_state_delta", "") - -# The number of times we are recalculating state when there is only a -# single forward extremity -state_delta_single_event_counter = Counter( - "synapse_storage_events_state_delta_single_event", "" -) - -# The number of times we are reculating state when we could have resonably -# calculated the delta when we calculated the state for an event we were -# persisting. -state_delta_reuse_delta_counter = Counter( - "synapse_storage_events_state_delta_reuse_delta", "" -) - -# The number of forward extremities for each new event. -forward_extremities_counter = Histogram( - "synapse_storage_events_forward_extremities_persisted", - "Number of forward extremities for each new event", - buckets=(1, 2, 3, 5, 7, 10, 15, 20, 50, 100, 200, 500, "+Inf"), -) - -# The number of stale forward extremities for each new event. Stale extremities -# are those that were in the previous set of extremities as well as the new. -stale_forward_extremities_counter = Histogram( - "synapse_storage_events_stale_forward_extremities_persisted", - "Number of unchanged forward extremities for each new event", - buckets=(0, 1, 2, 3, 5, 7, 10, 15, 20, 50, 100, 200, 500, "+Inf"), -) - - -def encode_json(json_object): - """ - Encode a Python object as JSON and return it in a Unicode string. - """ - out = frozendict_json_encoder.encode(json_object) - if isinstance(out, bytes): - out = out.decode("utf8") - return out - - -class _EventPeristenceQueue(object): - """Queues up events so that they can be persisted in bulk with only one - concurrent transaction per room. - """ - - _EventPersistQueueItem = namedtuple( - "_EventPersistQueueItem", ("events_and_contexts", "backfilled", "deferred") - ) - - def __init__(self): - self._event_persist_queues = {} - self._currently_persisting_rooms = set() - - def add_to_queue(self, room_id, events_and_contexts, backfilled): - """Add events to the queue, with the given persist_event options. - - NB: due to the normal usage pattern of this method, it does *not* - follow the synapse logcontext rules, and leaves the logcontext in - place whether or not the returned deferred is ready. - - Args: - room_id (str): - events_and_contexts (list[(EventBase, EventContext)]): - backfilled (bool): - - Returns: - defer.Deferred: a deferred which will resolve once the events are - persisted. Runs its callbacks *without* a logcontext. - """ - queue = self._event_persist_queues.setdefault(room_id, deque()) - if queue: - # if the last item in the queue has the same `backfilled` setting, - # we can just add these new events to that item. - end_item = queue[-1] - if end_item.backfilled == backfilled: - end_item.events_and_contexts.extend(events_and_contexts) - return end_item.deferred.observe() - - deferred = ObservableDeferred(defer.Deferred(), consumeErrors=True) - - queue.append( - self._EventPersistQueueItem( - events_and_contexts=events_and_contexts, - backfilled=backfilled, - deferred=deferred, - ) - ) - - return deferred.observe() - - def handle_queue(self, room_id, per_item_callback): - """Attempts to handle the queue for a room if not already being handled. - - The given callback will be invoked with for each item in the queue, - of type _EventPersistQueueItem. The per_item_callback will continuously - be called with new items, unless the queue becomnes empty. The return - value of the function will be given to the deferreds waiting on the item, - exceptions will be passed to the deferreds as well. - - This function should therefore be called whenever anything is added - to the queue. - - If another callback is currently handling the queue then it will not be - invoked. - """ - - if room_id in self._currently_persisting_rooms: - return - - self._currently_persisting_rooms.add(room_id) - - @defer.inlineCallbacks - def handle_queue_loop(): - try: - queue = self._get_drainining_queue(room_id) - for item in queue: - try: - ret = yield per_item_callback(item) - except Exception: - with PreserveLoggingContext(): - item.deferred.errback() - else: - with PreserveLoggingContext(): - item.deferred.callback(ret) - finally: - queue = self._event_persist_queues.pop(room_id, None) - if queue: - self._event_persist_queues[room_id] = queue - self._currently_persisting_rooms.discard(room_id) - - # set handle_queue_loop off in the background - run_as_background_process("persist_events", handle_queue_loop) - - def _get_drainining_queue(self, room_id): - queue = self._event_persist_queues.setdefault(room_id, deque()) - - try: - while True: - yield queue.popleft() - except IndexError: - # Queue has been drained. - pass - - -_EventCacheEntry = namedtuple("_EventCacheEntry", ("event", "redacted_event")) - - -def _retry_on_integrity_error(func): - """Wraps a database function so that it gets retried on IntegrityError, - with `delete_existing=True` passed in. - - Args: - func: function that returns a Deferred and accepts a `delete_existing` arg - """ - - @wraps(func) - @defer.inlineCallbacks - def f(self, *args, **kwargs): - try: - res = yield func(self, *args, **kwargs) - except self.database_engine.module.IntegrityError: - logger.exception("IntegrityError, retrying.") - res = yield func(self, *args, delete_existing=True, **kwargs) - return res - - return f - - -# inherits from EventFederationStore so that we can call _update_backward_extremities -# and _handle_mult_prev_events (though arguably those could both be moved in here) -class EventsStore( - StateGroupWorkerStore, - EventFederationStore, - EventsWorkerStore, - BackgroundUpdateStore, -): - def __init__(self, db_conn, hs): - super(EventsStore, self).__init__(db_conn, hs) - - self._event_persist_queue = _EventPeristenceQueue() - self._state_resolution_handler = hs.get_state_resolution_handler() - - # Collect metrics on the number of forward extremities that exist. - # Counter of number of extremities to count - self._current_forward_extremities_amount = c_counter() - - BucketCollector( - "synapse_forward_extremities", - lambda: self._current_forward_extremities_amount, - buckets=[1, 2, 3, 5, 7, 10, 15, 20, 50, 100, 200, 500, "+Inf"], - ) - - # Read the extrems every 60 minutes - def read_forward_extremities(): - # run as a background process to make sure that the database transactions - # have a logcontext to report to - return run_as_background_process( - "read_forward_extremities", self._read_forward_extremities - ) - - hs.get_clock().looping_call(read_forward_extremities, 60 * 60 * 1000) - - def _censor_redactions(): - return run_as_background_process( - "_censor_redactions", self._censor_redactions - ) - - if self.hs.config.redaction_retention_period is not None: - hs.get_clock().looping_call(_censor_redactions, 5 * 60 * 1000) - - @defer.inlineCallbacks - def _read_forward_extremities(self): - def fetch(txn): - txn.execute( - """ - select count(*) c from event_forward_extremities - group by room_id - """ - ) - return txn.fetchall() - - res = yield self.runInteraction("read_forward_extremities", fetch) - self._current_forward_extremities_amount = c_counter(list(x[0] for x in res)) - - @defer.inlineCallbacks - def persist_events(self, events_and_contexts, backfilled=False): - """ - Write events to the database - Args: - events_and_contexts: list of tuples of (event, context) - backfilled (bool): Whether the results are retrieved from federation - via backfill or not. Used to determine if they're "new" events - which might update the current state etc. - - Returns: - Deferred[int]: the stream ordering of the latest persisted event - """ - partitioned = {} - for event, ctx in events_and_contexts: - partitioned.setdefault(event.room_id, []).append((event, ctx)) - - deferreds = [] - for room_id, evs_ctxs in iteritems(partitioned): - d = self._event_persist_queue.add_to_queue( - room_id, evs_ctxs, backfilled=backfilled - ) - deferreds.append(d) - - for room_id in partitioned: - self._maybe_start_persisting(room_id) - - yield make_deferred_yieldable( - defer.gatherResults(deferreds, consumeErrors=True) - ) - - max_persisted_id = yield self._stream_id_gen.get_current_token() - - return max_persisted_id - - @defer.inlineCallbacks - @log_function - def persist_event(self, event, context, backfilled=False): - """ - - Args: - event (EventBase): - context (EventContext): - backfilled (bool): - - Returns: - Deferred: resolves to (int, int): the stream ordering of ``event``, - and the stream ordering of the latest persisted event - """ - deferred = self._event_persist_queue.add_to_queue( - event.room_id, [(event, context)], backfilled=backfilled - ) - - self._maybe_start_persisting(event.room_id) - - yield make_deferred_yieldable(deferred) - - max_persisted_id = yield self._stream_id_gen.get_current_token() - return (event.internal_metadata.stream_ordering, max_persisted_id) - - def _maybe_start_persisting(self, room_id): - @defer.inlineCallbacks - def persisting_queue(item): - with Measure(self._clock, "persist_events"): - yield self._persist_events( - item.events_and_contexts, backfilled=item.backfilled - ) - - self._event_persist_queue.handle_queue(room_id, persisting_queue) - - @_retry_on_integrity_error - @defer.inlineCallbacks - def _persist_events( - self, events_and_contexts, backfilled=False, delete_existing=False - ): - """Persist events to db - - Args: - events_and_contexts (list[(EventBase, EventContext)]): - backfilled (bool): - delete_existing (bool): - - Returns: - Deferred: resolves when the events have been persisted - """ - if not events_and_contexts: - return - - chunks = [ - events_and_contexts[x : x + 100] - for x in range(0, len(events_and_contexts), 100) - ] - - for chunk in chunks: - # We can't easily parallelize these since different chunks - # might contain the same event. :( - - # NB: Assumes that we are only persisting events for one room - # at a time. - - # map room_id->list[event_ids] giving the new forward - # extremities in each room - new_forward_extremeties = {} - - # map room_id->(type,state_key)->event_id tracking the full - # state in each room after adding these events. - # This is simply used to prefill the get_current_state_ids - # cache - current_state_for_room = {} - - # map room_id->(to_delete, to_insert) where to_delete is a list - # of type/state keys to remove from current state, and to_insert - # is a map (type,key)->event_id giving the state delta in each - # room - state_delta_for_room = {} - - if not backfilled: - with Measure(self._clock, "_calculate_state_and_extrem"): - # Work out the new "current state" for each room. - # We do this by working out what the new extremities are and then - # calculating the state from that. - events_by_room = {} - for event, context in chunk: - events_by_room.setdefault(event.room_id, []).append( - (event, context) - ) - - for room_id, ev_ctx_rm in iteritems(events_by_room): - latest_event_ids = yield self.get_latest_event_ids_in_room( - room_id - ) - new_latest_event_ids = yield self._calculate_new_extremities( - room_id, ev_ctx_rm, latest_event_ids - ) - - latest_event_ids = set(latest_event_ids) - if new_latest_event_ids == latest_event_ids: - # No change in extremities, so no change in state - continue - - # there should always be at least one forward extremity. - # (except during the initial persistence of the send_join - # results, in which case there will be no existing - # extremities, so we'll `continue` above and skip this bit.) - assert new_latest_event_ids, "No forward extremities left!" - - new_forward_extremeties[room_id] = new_latest_event_ids - - len_1 = ( - len(latest_event_ids) == 1 - and len(new_latest_event_ids) == 1 - ) - if len_1: - all_single_prev_not_state = all( - len(event.prev_event_ids()) == 1 - and not event.is_state() - for event, ctx in ev_ctx_rm - ) - # Don't bother calculating state if they're just - # a long chain of single ancestor non-state events. - if all_single_prev_not_state: - continue - - state_delta_counter.inc() - if len(new_latest_event_ids) == 1: - state_delta_single_event_counter.inc() - - # This is a fairly handwavey check to see if we could - # have guessed what the delta would have been when - # processing one of these events. - # What we're interested in is if the latest extremities - # were the same when we created the event as they are - # now. When this server creates a new event (as opposed - # to receiving it over federation) it will use the - # forward extremities as the prev_events, so we can - # guess this by looking at the prev_events and checking - # if they match the current forward extremities. - for ev, _ in ev_ctx_rm: - prev_event_ids = set(ev.prev_event_ids()) - if latest_event_ids == prev_event_ids: - state_delta_reuse_delta_counter.inc() - break - - logger.info("Calculating state delta for room %s", room_id) - with Measure( - self._clock, "persist_events.get_new_state_after_events" - ): - res = yield self._get_new_state_after_events( - room_id, - ev_ctx_rm, - latest_event_ids, - new_latest_event_ids, - ) - current_state, delta_ids = res - - # If either are not None then there has been a change, - # and we need to work out the delta (or use that - # given) - if delta_ids is not None: - # If there is a delta we know that we've - # only added or replaced state, never - # removed keys entirely. - state_delta_for_room[room_id] = ([], delta_ids) - elif current_state is not None: - with Measure( - self._clock, "persist_events.calculate_state_delta" - ): - delta = yield self._calculate_state_delta( - room_id, current_state - ) - state_delta_for_room[room_id] = delta - - # If we have the current_state then lets prefill - # the cache with it. - if current_state is not None: - current_state_for_room[room_id] = current_state - - # We want to calculate the stream orderings as late as possible, as - # we only notify after all events with a lesser stream ordering have - # been persisted. I.e. if we spend 10s inside the with block then - # that will delay all subsequent events from being notified about. - # Hence why we do it down here rather than wrapping the entire - # function. - # - # Its safe to do this after calculating the state deltas etc as we - # only need to protect the *persistence* of the events. This is to - # ensure that queries of the form "fetch events since X" don't - # return events and stream positions after events that are still in - # flight, as otherwise subsequent requests "fetch event since Y" - # will not return those events. - # - # Note: Multiple instances of this function cannot be in flight at - # the same time for the same room. - if backfilled: - stream_ordering_manager = self._backfill_id_gen.get_next_mult( - len(chunk) - ) - else: - stream_ordering_manager = self._stream_id_gen.get_next_mult(len(chunk)) - - with stream_ordering_manager as stream_orderings: - for (event, context), stream in zip(chunk, stream_orderings): - event.internal_metadata.stream_ordering = stream - - yield self.runInteraction( - "persist_events", - self._persist_events_txn, - events_and_contexts=chunk, - backfilled=backfilled, - delete_existing=delete_existing, - state_delta_for_room=state_delta_for_room, - new_forward_extremeties=new_forward_extremeties, - ) - persist_event_counter.inc(len(chunk)) - - if not backfilled: - # backfilled events have negative stream orderings, so we don't - # want to set the event_persisted_position to that. - synapse.metrics.event_persisted_position.set( - chunk[-1][0].internal_metadata.stream_ordering - ) - - for event, context in chunk: - if context.app_service: - origin_type = "local" - origin_entity = context.app_service.id - elif self.hs.is_mine_id(event.sender): - origin_type = "local" - origin_entity = "*client*" - else: - origin_type = "remote" - origin_entity = get_domain_from_id(event.sender) - - event_counter.labels(event.type, origin_type, origin_entity).inc() - - for room_id, new_state in iteritems(current_state_for_room): - self.get_current_state_ids.prefill((room_id,), new_state) - - for room_id, latest_event_ids in iteritems(new_forward_extremeties): - self.get_latest_event_ids_in_room.prefill( - (room_id,), list(latest_event_ids) - ) - - @defer.inlineCallbacks - def _calculate_new_extremities(self, room_id, event_contexts, latest_event_ids): - """Calculates the new forward extremities for a room given events to - persist. - - Assumes that we are only persisting events for one room at a time. - """ - - # we're only interested in new events which aren't outliers and which aren't - # being rejected. - new_events = [ - event - for event, ctx in event_contexts - if not event.internal_metadata.is_outlier() - and not ctx.rejected - and not event.internal_metadata.is_soft_failed() - ] - - latest_event_ids = set(latest_event_ids) - - # start with the existing forward extremities - result = set(latest_event_ids) - - # add all the new events to the list - result.update(event.event_id for event in new_events) - - # Now remove all events which are prev_events of any of the new events - result.difference_update( - e_id for event in new_events for e_id in event.prev_event_ids() - ) - - # Remove any events which are prev_events of any existing events. - existing_prevs = yield self._get_events_which_are_prevs(result) - result.difference_update(existing_prevs) - - # Finally handle the case where the new events have soft-failed prev - # events. If they do we need to remove them and their prev events, - # otherwise we end up with dangling extremities. - existing_prevs = yield self._get_prevs_before_rejected( - e_id for event in new_events for e_id in event.prev_event_ids() - ) - result.difference_update(existing_prevs) - - # We only update metrics for events that change forward extremities - # (e.g. we ignore backfill/outliers/etc) - if result != latest_event_ids: - forward_extremities_counter.observe(len(result)) - stale = latest_event_ids & result - stale_forward_extremities_counter.observe(len(stale)) - - return result - - @defer.inlineCallbacks - def _get_events_which_are_prevs(self, event_ids): - """Filter the supplied list of event_ids to get those which are prev_events of - existing (non-outlier/rejected) events. - - Args: - event_ids (Iterable[str]): event ids to filter - - Returns: - Deferred[List[str]]: filtered event ids - """ - results = [] - - def _get_events_which_are_prevs_txn(txn, batch): - sql = """ - SELECT prev_event_id, internal_metadata - FROM event_edges - INNER JOIN events USING (event_id) - LEFT JOIN rejections USING (event_id) - LEFT JOIN event_json USING (event_id) - WHERE - NOT events.outlier - AND rejections.event_id IS NULL - AND - """ - - clause, args = make_in_list_sql_clause( - self.database_engine, "prev_event_id", batch - ) - - txn.execute(sql + clause, args) - results.extend(r[0] for r in txn if not json.loads(r[1]).get("soft_failed")) - - for chunk in batch_iter(event_ids, 100): - yield self.runInteraction( - "_get_events_which_are_prevs", _get_events_which_are_prevs_txn, chunk - ) - - return results - - @defer.inlineCallbacks - def _get_prevs_before_rejected(self, event_ids): - """Get soft-failed ancestors to remove from the extremities. - - Given a set of events, find all those that have been soft-failed or - rejected. Returns those soft failed/rejected events and their prev - events (whether soft-failed/rejected or not), and recurses up the - prev-event graph until it finds no more soft-failed/rejected events. - - This is used to find extremities that are ancestors of new events, but - are separated by soft failed events. - - Args: - event_ids (Iterable[str]): Events to find prev events for. Note - that these must have already been persisted. - - Returns: - Deferred[set[str]] - """ - - # The set of event_ids to return. This includes all soft-failed events - # and their prev events. - existing_prevs = set() - - def _get_prevs_before_rejected_txn(txn, batch): - to_recursively_check = batch - - while to_recursively_check: - sql = """ - SELECT - event_id, prev_event_id, internal_metadata, - rejections.event_id IS NOT NULL - FROM event_edges - INNER JOIN events USING (event_id) - LEFT JOIN rejections USING (event_id) - LEFT JOIN event_json USING (event_id) - WHERE - NOT events.outlier - AND - """ - - clause, args = make_in_list_sql_clause( - self.database_engine, "event_id", to_recursively_check - ) - - txn.execute(sql + clause, args) - to_recursively_check = [] - - for event_id, prev_event_id, metadata, rejected in txn: - if prev_event_id in existing_prevs: - continue - - soft_failed = json.loads(metadata).get("soft_failed") - if soft_failed or rejected: - to_recursively_check.append(prev_event_id) - existing_prevs.add(prev_event_id) - - for chunk in batch_iter(event_ids, 100): - yield self.runInteraction( - "_get_prevs_before_rejected", _get_prevs_before_rejected_txn, chunk - ) - - return existing_prevs - - @defer.inlineCallbacks - def _get_new_state_after_events( - self, room_id, events_context, old_latest_event_ids, new_latest_event_ids - ): - """Calculate the current state dict after adding some new events to - a room - - Args: - room_id (str): - room to which the events are being added. Used for logging etc - - events_context (list[(EventBase, EventContext)]): - events and contexts which are being added to the room - - old_latest_event_ids (iterable[str]): - the old forward extremities for the room. - - new_latest_event_ids (iterable[str]): - the new forward extremities for the room. - - Returns: - Deferred[tuple[dict[(str,str), str]|None, dict[(str,str), str]|None]]: - Returns a tuple of two state maps, the first being the full new current - state and the second being the delta to the existing current state. - If both are None then there has been no change. - - If there has been a change then we only return the delta if its - already been calculated. Conversely if we do know the delta then - the new current state is only returned if we've already calculated - it. - """ - # map from state_group to ((type, key) -> event_id) state map - state_groups_map = {} - - # Map from (prev state group, new state group) -> delta state dict - state_group_deltas = {} - - for ev, ctx in events_context: - if ctx.state_group is None: - # This should only happen for outlier events. - if not ev.internal_metadata.is_outlier(): - raise Exception( - "Context for new event %s has no state " - "group" % (ev.event_id,) - ) - continue - - if ctx.state_group in state_groups_map: - continue - - # We're only interested in pulling out state that has already - # been cached in the context. We'll pull stuff out of the DB later - # if necessary. - current_state_ids = ctx.get_cached_current_state_ids() - if current_state_ids is not None: - state_groups_map[ctx.state_group] = current_state_ids - - if ctx.prev_group: - state_group_deltas[(ctx.prev_group, ctx.state_group)] = ctx.delta_ids - - # We need to map the event_ids to their state groups. First, let's - # check if the event is one we're persisting, in which case we can - # pull the state group from its context. - # Otherwise we need to pull the state group from the database. - - # Set of events we need to fetch groups for. (We know none of the old - # extremities are going to be in events_context). - missing_event_ids = set(old_latest_event_ids) - - event_id_to_state_group = {} - for event_id in new_latest_event_ids: - # First search in the list of new events we're adding. - for ev, ctx in events_context: - if event_id == ev.event_id and ctx.state_group is not None: - event_id_to_state_group[event_id] = ctx.state_group - break - else: - # If we couldn't find it, then we'll need to pull - # the state from the database - missing_event_ids.add(event_id) - - if missing_event_ids: - # Now pull out the state groups for any missing events from DB - event_to_groups = yield self._get_state_group_for_events(missing_event_ids) - event_id_to_state_group.update(event_to_groups) - - # State groups of old_latest_event_ids - old_state_groups = set( - event_id_to_state_group[evid] for evid in old_latest_event_ids - ) - - # State groups of new_latest_event_ids - new_state_groups = set( - event_id_to_state_group[evid] for evid in new_latest_event_ids - ) - - # If they old and new groups are the same then we don't need to do - # anything. - if old_state_groups == new_state_groups: - return None, None - - if len(new_state_groups) == 1 and len(old_state_groups) == 1: - # If we're going from one state group to another, lets check if - # we have a delta for that transition. If we do then we can just - # return that. - - new_state_group = next(iter(new_state_groups)) - old_state_group = next(iter(old_state_groups)) - - delta_ids = state_group_deltas.get((old_state_group, new_state_group), None) - if delta_ids is not None: - # We have a delta from the existing to new current state, - # so lets just return that. If we happen to already have - # the current state in memory then lets also return that, - # but it doesn't matter if we don't. - new_state = state_groups_map.get(new_state_group) - return new_state, delta_ids - - # Now that we have calculated new_state_groups we need to get - # their state IDs so we can resolve to a single state set. - missing_state = new_state_groups - set(state_groups_map) - if missing_state: - group_to_state = yield self._get_state_for_groups(missing_state) - state_groups_map.update(group_to_state) - - if len(new_state_groups) == 1: - # If there is only one state group, then we know what the current - # state is. - return state_groups_map[new_state_groups.pop()], None - - # Ok, we need to defer to the state handler to resolve our state sets. - - state_groups = {sg: state_groups_map[sg] for sg in new_state_groups} - - events_map = {ev.event_id: ev for ev, _ in events_context} - - # We need to get the room version, which is in the create event. - # Normally that'd be in the database, but its also possible that we're - # currently trying to persist it. - room_version = None - for ev, _ in events_context: - if ev.type == EventTypes.Create and ev.state_key == "": - room_version = ev.content.get("room_version", "1") - break - - if not room_version: - room_version = yield self.get_room_version(room_id) - - logger.debug("calling resolve_state_groups from preserve_events") - res = yield self._state_resolution_handler.resolve_state_groups( - room_id, - room_version, - state_groups, - events_map, - state_res_store=StateResolutionStore(self), - ) - - return res.state, None - - @defer.inlineCallbacks - def _calculate_state_delta(self, room_id, current_state): - """Calculate the new state deltas for a room. - - Assumes that we are only persisting events for one room at a time. - - Returns: - tuple[list, dict] (to_delete, to_insert): where to_delete are the - type/state_keys to remove from current_state_events and `to_insert` - are the updates to current_state_events. - """ - existing_state = yield self.get_current_state_ids(room_id) - - to_delete = [key for key in existing_state if key not in current_state] - - to_insert = { - key: ev_id - for key, ev_id in iteritems(current_state) - if ev_id != existing_state.get(key) - } - - return to_delete, to_insert - - @log_function - def _persist_events_txn( - self, - txn, - events_and_contexts, - backfilled, - delete_existing=False, - state_delta_for_room={}, - new_forward_extremeties={}, - ): - """Insert some number of room events into the necessary database tables. - - Rejected events are only inserted into the events table, the events_json table, - and the rejections table. Things reading from those table will need to check - whether the event was rejected. - - Args: - txn (twisted.enterprise.adbapi.Connection): db connection - events_and_contexts (list[(EventBase, EventContext)]): - events to persist - backfilled (bool): True if the events were backfilled - delete_existing (bool): True to purge existing table rows for the - events from the database. This is useful when retrying due to - IntegrityError. - state_delta_for_room (dict[str, (list, dict)]): - The current-state delta for each room. For each room, a tuple - (to_delete, to_insert), being a list of type/state keys to be - removed from the current state, and a state set to be added to - the current state. - new_forward_extremeties (dict[str, list[str]]): - The new forward extremities for each room. For each room, a - list of the event ids which are the forward extremities. - - """ - all_events_and_contexts = events_and_contexts - - min_stream_order = events_and_contexts[0][0].internal_metadata.stream_ordering - max_stream_order = events_and_contexts[-1][0].internal_metadata.stream_ordering - - self._update_forward_extremities_txn( - txn, - new_forward_extremities=new_forward_extremeties, - max_stream_order=max_stream_order, - ) - - # Ensure that we don't have the same event twice. - events_and_contexts = self._filter_events_and_contexts_for_duplicates( - events_and_contexts - ) - - self._update_room_depths_txn( - txn, events_and_contexts=events_and_contexts, backfilled=backfilled - ) - - # _update_outliers_txn filters out any events which have already been - # persisted, and returns the filtered list. - events_and_contexts = self._update_outliers_txn( - txn, events_and_contexts=events_and_contexts - ) - - # From this point onwards the events are only events that we haven't - # seen before. - - if delete_existing: - # For paranoia reasons, we go and delete all the existing entries - # for these events so we can reinsert them. - # This gets around any problems with some tables already having - # entries. - self._delete_existing_rows_txn(txn, events_and_contexts=events_and_contexts) - - self._store_event_txn(txn, events_and_contexts=events_and_contexts) - - # Insert into event_to_state_groups. - self._store_event_state_mappings_txn(txn, events_and_contexts) - - # We want to store event_auth mappings for rejected events, as they're - # used in state res v2. - # This is only necessary if the rejected event appears in an accepted - # event's auth chain, but its easier for now just to store them (and - # it doesn't take much storage compared to storing the entire event - # anyway). - self._simple_insert_many_txn( - txn, - table="event_auth", - values=[ - { - "event_id": event.event_id, - "room_id": event.room_id, - "auth_id": auth_id, - } - for event, _ in events_and_contexts - for auth_id in event.auth_event_ids() - if event.is_state() - ], - ) - - # _store_rejected_events_txn filters out any events which were - # rejected, and returns the filtered list. - events_and_contexts = self._store_rejected_events_txn( - txn, events_and_contexts=events_and_contexts - ) - - # From this point onwards the events are only ones that weren't - # rejected. - - self._update_metadata_tables_txn( - txn, - events_and_contexts=events_and_contexts, - all_events_and_contexts=all_events_and_contexts, - backfilled=backfilled, - ) - - # We call this last as it assumes we've inserted the events into - # room_memberships, where applicable. - self._update_current_state_txn(txn, state_delta_for_room, min_stream_order) - - def _update_current_state_txn(self, txn, state_delta_by_room, stream_id): - for room_id, current_state_tuple in iteritems(state_delta_by_room): - to_delete, to_insert = current_state_tuple - - # First we add entries to the current_state_delta_stream. We - # do this before updating the current_state_events table so - # that we can use it to calculate the `prev_event_id`. (This - # allows us to not have to pull out the existing state - # unnecessarily). - # - # The stream_id for the update is chosen to be the minimum of the stream_ids - # for the batch of the events that we are persisting; that means we do not - # end up in a situation where workers see events before the - # current_state_delta updates. - # - sql = """ - INSERT INTO current_state_delta_stream - (stream_id, room_id, type, state_key, event_id, prev_event_id) - SELECT ?, ?, ?, ?, ?, ( - SELECT event_id FROM current_state_events - WHERE room_id = ? AND type = ? AND state_key = ? - ) - """ - txn.executemany( - sql, - ( - ( - stream_id, - room_id, - etype, - state_key, - None, - room_id, - etype, - state_key, - ) - for etype, state_key in to_delete - # We sanity check that we're deleting rather than updating - if (etype, state_key) not in to_insert - ), - ) - txn.executemany( - sql, - ( - ( - stream_id, - room_id, - etype, - state_key, - ev_id, - room_id, - etype, - state_key, - ) - for (etype, state_key), ev_id in iteritems(to_insert) - ), - ) - - # Now we actually update the current_state_events table - - txn.executemany( - "DELETE FROM current_state_events" - " WHERE room_id = ? AND type = ? AND state_key = ?", - ( - (room_id, etype, state_key) - for etype, state_key in itertools.chain(to_delete, to_insert) - ), - ) - - # We include the membership in the current state table, hence we do - # a lookup when we insert. This assumes that all events have already - # been inserted into room_memberships. - txn.executemany( - """INSERT INTO current_state_events - (room_id, type, state_key, event_id, membership) - VALUES (?, ?, ?, ?, (SELECT membership FROM room_memberships WHERE event_id = ?)) - """, - [ - (room_id, key[0], key[1], ev_id, ev_id) - for key, ev_id in iteritems(to_insert) - ], - ) - - txn.call_after( - self._curr_state_delta_stream_cache.entity_has_changed, - room_id, - stream_id, - ) - - # Invalidate the various caches - - # Figure out the changes of membership to invalidate the - # `get_rooms_for_user` cache. - # We find out which membership events we may have deleted - # and which we have added, then we invlidate the caches for all - # those users. - members_changed = set( - state_key - for ev_type, state_key in itertools.chain(to_delete, to_insert) - if ev_type == EventTypes.Member - ) - - for member in members_changed: - txn.call_after( - self.get_rooms_for_user_with_stream_ordering.invalidate, (member,) - ) - - self._invalidate_state_caches_and_stream(txn, room_id, members_changed) - - def _update_forward_extremities_txn( - self, txn, new_forward_extremities, max_stream_order - ): - for room_id, new_extrem in iteritems(new_forward_extremities): - self._simple_delete_txn( - txn, table="event_forward_extremities", keyvalues={"room_id": room_id} - ) - txn.call_after(self.get_latest_event_ids_in_room.invalidate, (room_id,)) - - self._simple_insert_many_txn( - txn, - table="event_forward_extremities", - values=[ - {"event_id": ev_id, "room_id": room_id} - for room_id, new_extrem in iteritems(new_forward_extremities) - for ev_id in new_extrem - ], - ) - # We now insert into stream_ordering_to_exterm a mapping from room_id, - # new stream_ordering to new forward extremeties in the room. - # This allows us to later efficiently look up the forward extremeties - # for a room before a given stream_ordering - self._simple_insert_many_txn( - txn, - table="stream_ordering_to_exterm", - values=[ - { - "room_id": room_id, - "event_id": event_id, - "stream_ordering": max_stream_order, - } - for room_id, new_extrem in iteritems(new_forward_extremities) - for event_id in new_extrem - ], - ) - - @classmethod - def _filter_events_and_contexts_for_duplicates(cls, events_and_contexts): - """Ensure that we don't have the same event twice. - - Pick the earliest non-outlier if there is one, else the earliest one. - - Args: - events_and_contexts (list[(EventBase, EventContext)]): - Returns: - list[(EventBase, EventContext)]: filtered list - """ - new_events_and_contexts = OrderedDict() - for event, context in events_and_contexts: - prev_event_context = new_events_and_contexts.get(event.event_id) - if prev_event_context: - if not event.internal_metadata.is_outlier(): - if prev_event_context[0].internal_metadata.is_outlier(): - # To ensure correct ordering we pop, as OrderedDict is - # ordered by first insertion. - new_events_and_contexts.pop(event.event_id, None) - new_events_and_contexts[event.event_id] = (event, context) - else: - new_events_and_contexts[event.event_id] = (event, context) - return list(new_events_and_contexts.values()) - - def _update_room_depths_txn(self, txn, events_and_contexts, backfilled): - """Update min_depth for each room - - Args: - txn (twisted.enterprise.adbapi.Connection): db connection - events_and_contexts (list[(EventBase, EventContext)]): events - we are persisting - backfilled (bool): True if the events were backfilled - """ - depth_updates = {} - for event, context in events_and_contexts: - # Remove the any existing cache entries for the event_ids - txn.call_after(self._invalidate_get_event_cache, event.event_id) - if not backfilled: - txn.call_after( - self._events_stream_cache.entity_has_changed, - event.room_id, - event.internal_metadata.stream_ordering, - ) - - if not event.internal_metadata.is_outlier() and not context.rejected: - depth_updates[event.room_id] = max( - event.depth, depth_updates.get(event.room_id, event.depth) - ) - - for room_id, depth in iteritems(depth_updates): - self._update_min_depth_for_room_txn(txn, room_id, depth) - - def _update_outliers_txn(self, txn, events_and_contexts): - """Update any outliers with new event info. - - This turns outliers into ex-outliers (unless the new event was - rejected). - - Args: - txn (twisted.enterprise.adbapi.Connection): db connection - events_and_contexts (list[(EventBase, EventContext)]): events - we are persisting - - Returns: - list[(EventBase, EventContext)] new list, without events which - are already in the events table. - """ - txn.execute( - "SELECT event_id, outlier FROM events WHERE event_id in (%s)" - % (",".join(["?"] * len(events_and_contexts)),), - [event.event_id for event, _ in events_and_contexts], - ) - - have_persisted = {event_id: outlier for event_id, outlier in txn} - - to_remove = set() - for event, context in events_and_contexts: - if event.event_id not in have_persisted: - continue - - to_remove.add(event) - - if context.rejected: - # If the event is rejected then we don't care if the event - # was an outlier or not. - continue - - outlier_persisted = have_persisted[event.event_id] - if not event.internal_metadata.is_outlier() and outlier_persisted: - # We received a copy of an event that we had already stored as - # an outlier in the database. We now have some state at that - # so we need to update the state_groups table with that state. - - # insert into event_to_state_groups. - try: - self._store_event_state_mappings_txn(txn, ((event, context),)) - except Exception: - logger.exception("") - raise - - metadata_json = encode_json(event.internal_metadata.get_dict()) - - sql = ( - "UPDATE event_json SET internal_metadata = ?" " WHERE event_id = ?" - ) - txn.execute(sql, (metadata_json, event.event_id)) - - # Add an entry to the ex_outlier_stream table to replicate the - # change in outlier status to our workers. - stream_order = event.internal_metadata.stream_ordering - state_group_id = context.state_group - self._simple_insert_txn( - txn, - table="ex_outlier_stream", - values={ - "event_stream_ordering": stream_order, - "event_id": event.event_id, - "state_group": state_group_id, - }, - ) - - sql = "UPDATE events SET outlier = ?" " WHERE event_id = ?" - txn.execute(sql, (False, event.event_id)) - - # Update the event_backward_extremities table now that this - # event isn't an outlier any more. - self._update_backward_extremeties(txn, [event]) - - return [ec for ec in events_and_contexts if ec[0] not in to_remove] - - @classmethod - def _delete_existing_rows_txn(cls, txn, events_and_contexts): - if not events_and_contexts: - # nothing to do here - return - - logger.info("Deleting existing") - - for table in ( - "events", - "event_auth", - "event_json", - "event_edges", - "event_forward_extremities", - "event_reference_hashes", - "event_search", - "event_to_state_groups", - "local_invites", - "state_events", - "rejections", - "redactions", - "room_memberships", - ): - txn.executemany( - "DELETE FROM %s WHERE event_id = ?" % (table,), - [(ev.event_id,) for ev, _ in events_and_contexts], - ) - - for table in ("event_push_actions",): - txn.executemany( - "DELETE FROM %s WHERE room_id = ? AND event_id = ?" % (table,), - [(ev.room_id, ev.event_id) for ev, _ in events_and_contexts], - ) - - def _store_event_txn(self, txn, events_and_contexts): - """Insert new events into the event and event_json tables - - Args: - txn (twisted.enterprise.adbapi.Connection): db connection - events_and_contexts (list[(EventBase, EventContext)]): events - we are persisting - """ - - if not events_and_contexts: - # nothing to do here - return - - def event_dict(event): - d = event.get_dict() - d.pop("redacted", None) - d.pop("redacted_because", None) - return d - - self._simple_insert_many_txn( - txn, - table="event_json", - values=[ - { - "event_id": event.event_id, - "room_id": event.room_id, - "internal_metadata": encode_json( - event.internal_metadata.get_dict() - ), - "json": encode_json(event_dict(event)), - "format_version": event.format_version, - } - for event, _ in events_and_contexts - ], - ) - - self._simple_insert_many_txn( - txn, - table="events", - values=[ - { - "stream_ordering": event.internal_metadata.stream_ordering, - "topological_ordering": event.depth, - "depth": event.depth, - "event_id": event.event_id, - "room_id": event.room_id, - "type": event.type, - "processed": True, - "outlier": event.internal_metadata.is_outlier(), - "origin_server_ts": int(event.origin_server_ts), - "received_ts": self._clock.time_msec(), - "sender": event.sender, - "contains_url": ( - "url" in event.content - and isinstance(event.content["url"], text_type) - ), - } - for event, _ in events_and_contexts - ], - ) - - for event, _ in events_and_contexts: - if not event.internal_metadata.is_redacted(): - # If we're persisting an unredacted event we go and ensure - # that we mark any redactions that reference this event as - # requiring censoring. - self._simple_update_txn( - txn, - table="redactions", - keyvalues={"redacts": event.event_id}, - updatevalues={"have_censored": False}, - ) - - def _store_rejected_events_txn(self, txn, events_and_contexts): - """Add rows to the 'rejections' table for received events which were - rejected - - Args: - txn (twisted.enterprise.adbapi.Connection): db connection - events_and_contexts (list[(EventBase, EventContext)]): events - we are persisting - - Returns: - list[(EventBase, EventContext)] new list, without the rejected - events. - """ - # Remove the rejected events from the list now that we've added them - # to the events table and the events_json table. - to_remove = set() - for event, context in events_and_contexts: - if context.rejected: - # Insert the event_id into the rejections table - self._store_rejections_txn(txn, event.event_id, context.rejected) - to_remove.add(event) - - return [ec for ec in events_and_contexts if ec[0] not in to_remove] - - def _update_metadata_tables_txn( - self, txn, events_and_contexts, all_events_and_contexts, backfilled - ): - """Update all the miscellaneous tables for new events - - Args: - txn (twisted.enterprise.adbapi.Connection): db connection - events_and_contexts (list[(EventBase, EventContext)]): events - we are persisting - all_events_and_contexts (list[(EventBase, EventContext)]): all - events that we were going to persist. This includes events - we've already persisted, etc, that wouldn't appear in - events_and_context. - backfilled (bool): True if the events were backfilled - """ - - # Insert all the push actions into the event_push_actions table. - self._set_push_actions_for_event_and_users_txn( - txn, - events_and_contexts=events_and_contexts, - all_events_and_contexts=all_events_and_contexts, - ) - - if not events_and_contexts: - # nothing to do here - return - - for event, context in events_and_contexts: - if event.type == EventTypes.Redaction and event.redacts is not None: - # Remove the entries in the event_push_actions table for the - # redacted event. - self._remove_push_actions_for_event_id_txn( - txn, event.room_id, event.redacts - ) - - # Remove from relations table. - self._handle_redaction(txn, event.redacts) - - # Update the event_forward_extremities, event_backward_extremities and - # event_edges tables. - self._handle_mult_prev_events( - txn, events=[event for event, _ in events_and_contexts] - ) - - for event, _ in events_and_contexts: - if event.type == EventTypes.Name: - # Insert into the event_search table. - self._store_room_name_txn(txn, event) - elif event.type == EventTypes.Topic: - # Insert into the event_search table. - self._store_room_topic_txn(txn, event) - elif event.type == EventTypes.Message: - # Insert into the event_search table. - self._store_room_message_txn(txn, event) - elif event.type == EventTypes.Redaction: - # Insert into the redactions table. - self._store_redaction(txn, event) - - self._handle_event_relations(txn, event) - - # Insert into the room_memberships table. - self._store_room_members_txn( - txn, - [ - event - for event, _ in events_and_contexts - if event.type == EventTypes.Member - ], - backfilled=backfilled, - ) - - # Insert event_reference_hashes table. - self._store_event_reference_hashes_txn( - txn, [event for event, _ in events_and_contexts] - ) - - state_events_and_contexts = [ - ec for ec in events_and_contexts if ec[0].is_state() - ] - - state_values = [] - for event, context in state_events_and_contexts: - vals = { - "event_id": event.event_id, - "room_id": event.room_id, - "type": event.type, - "state_key": event.state_key, - } - - # TODO: How does this work with backfilling? - if hasattr(event, "replaces_state"): - vals["prev_state"] = event.replaces_state - - state_values.append(vals) - - self._simple_insert_many_txn(txn, table="state_events", values=state_values) - - # Prefill the event cache - self._add_to_cache(txn, events_and_contexts) - - def _add_to_cache(self, txn, events_and_contexts): - to_prefill = [] - - rows = [] - N = 200 - for i in range(0, len(events_and_contexts), N): - ev_map = {e[0].event_id: e[0] for e in events_and_contexts[i : i + N]} - if not ev_map: - break - - sql = ( - "SELECT " - " e.event_id as event_id, " - " r.redacts as redacts," - " rej.event_id as rejects " - " FROM events as e" - " LEFT JOIN rejections as rej USING (event_id)" - " LEFT JOIN redactions as r ON e.event_id = r.redacts" - " WHERE " - ) - - clause, args = make_in_list_sql_clause( - self.database_engine, "e.event_id", list(ev_map) - ) - - txn.execute(sql + clause, args) - rows = self.cursor_to_dict(txn) - for row in rows: - event = ev_map[row["event_id"]] - if not row["rejects"] and not row["redacts"]: - to_prefill.append( - _EventCacheEntry(event=event, redacted_event=None) - ) - - def prefill(): - for cache_entry in to_prefill: - self._get_event_cache.prefill((cache_entry[0].event_id,), cache_entry) - - txn.call_after(prefill) - - def _store_redaction(self, txn, event): - # invalidate the cache for the redacted event - txn.call_after(self._invalidate_get_event_cache, event.redacts) - - self._simple_insert_txn( - txn, - table="redactions", - values={ - "event_id": event.event_id, - "redacts": event.redacts, - "received_ts": self._clock.time_msec(), - }, - ) - - @defer.inlineCallbacks - def _censor_redactions(self): - """Censors all redactions older than the configured period that haven't - been censored yet. - - By censor we mean update the event_json table with the redacted event. - - Returns: - Deferred - """ - - if self.hs.config.redaction_retention_period is None: - return - - before_ts = self._clock.time_msec() - self.hs.config.redaction_retention_period - - # We fetch all redactions that: - # 1. point to an event we have, - # 2. has a received_ts from before the cut off, and - # 3. we haven't yet censored. - # - # This is limited to 100 events to ensure that we don't try and do too - # much at once. We'll get called again so this should eventually catch - # up. - sql = """ - SELECT redactions.event_id, redacts FROM redactions - LEFT JOIN events AS original_event ON ( - redacts = original_event.event_id - ) - WHERE NOT have_censored - AND redactions.received_ts <= ? - ORDER BY redactions.received_ts ASC - LIMIT ? - """ - - rows = yield self._execute( - "_censor_redactions_fetch", None, sql, before_ts, 100 - ) - - updates = [] - - for redaction_id, event_id in rows: - redaction_event = yield self.get_event(redaction_id, allow_none=True) - original_event = yield self.get_event( - event_id, allow_rejected=True, allow_none=True - ) - - # The SQL above ensures that we have both the redaction and - # original event, so if the `get_event` calls return None it - # means that the redaction wasn't allowed. Either way we know that - # the result won't change so we mark the fact that we've checked. - if ( - redaction_event - and original_event - and original_event.internal_metadata.is_redacted() - ): - # Redaction was allowed - pruned_json = encode_json(prune_event_dict(original_event.get_dict())) - else: - # Redaction wasn't allowed - pruned_json = None - - updates.append((redaction_id, event_id, pruned_json)) - - def _update_censor_txn(txn): - for redaction_id, event_id, pruned_json in updates: - if pruned_json: - self._simple_update_one_txn( - txn, - table="event_json", - keyvalues={"event_id": event_id}, - updatevalues={"json": pruned_json}, - ) - - self._simple_update_one_txn( - txn, - table="redactions", - keyvalues={"event_id": redaction_id}, - updatevalues={"have_censored": True}, - ) - - yield self.runInteraction("_update_censor_txn", _update_censor_txn) - - @defer.inlineCallbacks - def count_daily_messages(self): - """ - Returns an estimate of the number of messages sent in the last day. - - If it has been significantly less or more than one day since the last - call to this function, it will return None. - """ - - def _count_messages(txn): - sql = """ - SELECT COALESCE(COUNT(*), 0) FROM events - WHERE type = 'm.room.message' - AND stream_ordering > ? - """ - txn.execute(sql, (self.stream_ordering_day_ago,)) - count, = txn.fetchone() - return count - - ret = yield self.runInteraction("count_messages", _count_messages) - return ret - - @defer.inlineCallbacks - def count_daily_sent_messages(self): - def _count_messages(txn): - # This is good enough as if you have silly characters in your own - # hostname then thats your own fault. - like_clause = "%:" + self.hs.hostname - - sql = """ - SELECT COALESCE(COUNT(*), 0) FROM events - WHERE type = 'm.room.message' - AND sender LIKE ? - AND stream_ordering > ? - """ - - txn.execute(sql, (like_clause, self.stream_ordering_day_ago)) - count, = txn.fetchone() - return count - - ret = yield self.runInteraction("count_daily_sent_messages", _count_messages) - return ret - - @defer.inlineCallbacks - def count_daily_active_rooms(self): - def _count(txn): - sql = """ - SELECT COALESCE(COUNT(DISTINCT room_id), 0) FROM events - WHERE type = 'm.room.message' - AND stream_ordering > ? - """ - txn.execute(sql, (self.stream_ordering_day_ago,)) - count, = txn.fetchone() - return count - - ret = yield self.runInteraction("count_daily_active_rooms", _count) - return ret - - def get_current_backfill_token(self): - """The current minimum token that backfilled events have reached""" - return -self._backfill_id_gen.get_current_token() - - def get_current_events_token(self): - """The current maximum token that events have reached""" - return self._stream_id_gen.get_current_token() - - def get_all_new_forward_event_rows(self, last_id, current_id, limit): - if last_id == current_id: - return defer.succeed([]) - - def get_all_new_forward_event_rows(txn): - sql = ( - "SELECT e.stream_ordering, e.event_id, e.room_id, e.type," - " state_key, redacts, relates_to_id" - " FROM events AS e" - " LEFT JOIN redactions USING (event_id)" - " LEFT JOIN state_events USING (event_id)" - " LEFT JOIN event_relations USING (event_id)" - " WHERE ? < stream_ordering AND stream_ordering <= ?" - " ORDER BY stream_ordering ASC" - " LIMIT ?" - ) - txn.execute(sql, (last_id, current_id, limit)) - new_event_updates = txn.fetchall() - - if len(new_event_updates) == limit: - upper_bound = new_event_updates[-1][0] - else: - upper_bound = current_id - - sql = ( - "SELECT event_stream_ordering, e.event_id, e.room_id, e.type," - " state_key, redacts, relates_to_id" - " FROM events AS e" - " INNER JOIN ex_outlier_stream USING (event_id)" - " LEFT JOIN redactions USING (event_id)" - " LEFT JOIN state_events USING (event_id)" - " LEFT JOIN event_relations USING (event_id)" - " WHERE ? < event_stream_ordering" - " AND event_stream_ordering <= ?" - " ORDER BY event_stream_ordering DESC" - ) - txn.execute(sql, (last_id, upper_bound)) - new_event_updates.extend(txn) - - return new_event_updates - - return self.runInteraction( - "get_all_new_forward_event_rows", get_all_new_forward_event_rows - ) - - def get_all_new_backfill_event_rows(self, last_id, current_id, limit): - if last_id == current_id: - return defer.succeed([]) - - def get_all_new_backfill_event_rows(txn): - sql = ( - "SELECT -e.stream_ordering, e.event_id, e.room_id, e.type," - " state_key, redacts, relates_to_id" - " FROM events AS e" - " LEFT JOIN redactions USING (event_id)" - " LEFT JOIN state_events USING (event_id)" - " LEFT JOIN event_relations USING (event_id)" - " WHERE ? > stream_ordering AND stream_ordering >= ?" - " ORDER BY stream_ordering ASC" - " LIMIT ?" - ) - txn.execute(sql, (-last_id, -current_id, limit)) - new_event_updates = txn.fetchall() - - if len(new_event_updates) == limit: - upper_bound = new_event_updates[-1][0] - else: - upper_bound = current_id - - sql = ( - "SELECT -event_stream_ordering, e.event_id, e.room_id, e.type," - " state_key, redacts, relates_to_id" - " FROM events AS e" - " INNER JOIN ex_outlier_stream USING (event_id)" - " LEFT JOIN redactions USING (event_id)" - " LEFT JOIN state_events USING (event_id)" - " LEFT JOIN event_relations USING (event_id)" - " WHERE ? > event_stream_ordering" - " AND event_stream_ordering >= ?" - " ORDER BY event_stream_ordering DESC" - ) - txn.execute(sql, (-last_id, -upper_bound)) - new_event_updates.extend(txn.fetchall()) - - return new_event_updates - - return self.runInteraction( - "get_all_new_backfill_event_rows", get_all_new_backfill_event_rows - ) - - @cached(num_args=5, max_entries=10) - def get_all_new_events( - self, - last_backfill_id, - last_forward_id, - current_backfill_id, - current_forward_id, - limit, - ): - """Get all the new events that have arrived at the server either as - new events or as backfilled events""" - have_backfill_events = last_backfill_id != current_backfill_id - have_forward_events = last_forward_id != current_forward_id - - if not have_backfill_events and not have_forward_events: - return defer.succeed(AllNewEventsResult([], [], [], [], [])) - - def get_all_new_events_txn(txn): - sql = ( - "SELECT e.stream_ordering, e.event_id, e.room_id, e.type," - " state_key, redacts" - " FROM events AS e" - " LEFT JOIN redactions USING (event_id)" - " LEFT JOIN state_events USING (event_id)" - " WHERE ? < stream_ordering AND stream_ordering <= ?" - " ORDER BY stream_ordering ASC" - " LIMIT ?" - ) - if have_forward_events: - txn.execute(sql, (last_forward_id, current_forward_id, limit)) - new_forward_events = txn.fetchall() - - if len(new_forward_events) == limit: - upper_bound = new_forward_events[-1][0] - else: - upper_bound = current_forward_id - - sql = ( - "SELECT event_stream_ordering, event_id, state_group" - " FROM ex_outlier_stream" - " WHERE ? > event_stream_ordering" - " AND event_stream_ordering >= ?" - " ORDER BY event_stream_ordering DESC" - ) - txn.execute(sql, (last_forward_id, upper_bound)) - forward_ex_outliers = txn.fetchall() - else: - new_forward_events = [] - forward_ex_outliers = [] - - sql = ( - "SELECT -e.stream_ordering, e.event_id, e.room_id, e.type," - " state_key, redacts" - " FROM events AS e" - " LEFT JOIN redactions USING (event_id)" - " LEFT JOIN state_events USING (event_id)" - " WHERE ? > stream_ordering AND stream_ordering >= ?" - " ORDER BY stream_ordering DESC" - " LIMIT ?" - ) - if have_backfill_events: - txn.execute(sql, (-last_backfill_id, -current_backfill_id, limit)) - new_backfill_events = txn.fetchall() - - if len(new_backfill_events) == limit: - upper_bound = new_backfill_events[-1][0] - else: - upper_bound = current_backfill_id - - sql = ( - "SELECT -event_stream_ordering, event_id, state_group" - " FROM ex_outlier_stream" - " WHERE ? > event_stream_ordering" - " AND event_stream_ordering >= ?" - " ORDER BY event_stream_ordering DESC" - ) - txn.execute(sql, (-last_backfill_id, -upper_bound)) - backward_ex_outliers = txn.fetchall() - else: - new_backfill_events = [] - backward_ex_outliers = [] - - return AllNewEventsResult( - new_forward_events, - new_backfill_events, - forward_ex_outliers, - backward_ex_outliers, - ) - - return self.runInteraction("get_all_new_events", get_all_new_events_txn) - - def purge_history(self, room_id, token, delete_local_events): - """Deletes room history before a certain point - - Args: - room_id (str): - - token (str): A topological token to delete events before - - delete_local_events (bool): - if True, we will delete local events as well as remote ones - (instead of just marking them as outliers and deleting their - state groups). - """ - - return self.runInteraction( - "purge_history", - self._purge_history_txn, - room_id, - token, - delete_local_events, - ) - - def _purge_history_txn(self, txn, room_id, token_str, delete_local_events): - token = RoomStreamToken.parse(token_str) - - # Tables that should be pruned: - # event_auth - # event_backward_extremities - # event_edges - # event_forward_extremities - # event_json - # event_push_actions - # event_reference_hashes - # event_search - # event_to_state_groups - # events - # rejections - # room_depth - # state_groups - # state_groups_state - - # we will build a temporary table listing the events so that we don't - # have to keep shovelling the list back and forth across the - # connection. Annoyingly the python sqlite driver commits the - # transaction on CREATE, so let's do this first. - # - # furthermore, we might already have the table from a previous (failed) - # purge attempt, so let's drop the table first. - - txn.execute("DROP TABLE IF EXISTS events_to_purge") - - txn.execute( - "CREATE TEMPORARY TABLE events_to_purge (" - " event_id TEXT NOT NULL," - " should_delete BOOLEAN NOT NULL" - ")" - ) - - # First ensure that we're not about to delete all the forward extremeties - txn.execute( - "SELECT e.event_id, e.depth FROM events as e " - "INNER JOIN event_forward_extremities as f " - "ON e.event_id = f.event_id " - "AND e.room_id = f.room_id " - "WHERE f.room_id = ?", - (room_id,), - ) - rows = txn.fetchall() - max_depth = max(row[1] for row in rows) - - if max_depth < token.topological: - # We need to ensure we don't delete all the events from the database - # otherwise we wouldn't be able to send any events (due to not - # having any backwards extremeties) - raise SynapseError( - 400, "topological_ordering is greater than forward extremeties" - ) - - logger.info("[purge] looking for events to delete") - - should_delete_expr = "state_key IS NULL" - should_delete_params = () - if not delete_local_events: - should_delete_expr += " AND event_id NOT LIKE ?" - - # We include the parameter twice since we use the expression twice - should_delete_params += ("%:" + self.hs.hostname, "%:" + self.hs.hostname) - - should_delete_params += (room_id, token.topological) - - # Note that we insert events that are outliers and aren't going to be - # deleted, as nothing will happen to them. - txn.execute( - "INSERT INTO events_to_purge" - " SELECT event_id, %s" - " FROM events AS e LEFT JOIN state_events USING (event_id)" - " WHERE (NOT outlier OR (%s)) AND e.room_id = ? AND topological_ordering < ?" - % (should_delete_expr, should_delete_expr), - should_delete_params, - ) - - # We create the indices *after* insertion as that's a lot faster. - - # create an index on should_delete because later we'll be looking for - # the should_delete / shouldn't_delete subsets - txn.execute( - "CREATE INDEX events_to_purge_should_delete" - " ON events_to_purge(should_delete)" - ) - - # We do joins against events_to_purge for e.g. calculating state - # groups to purge, etc., so lets make an index. - txn.execute("CREATE INDEX events_to_purge_id" " ON events_to_purge(event_id)") - - txn.execute("SELECT event_id, should_delete FROM events_to_purge") - event_rows = txn.fetchall() - logger.info( - "[purge] found %i events before cutoff, of which %i can be deleted", - len(event_rows), - sum(1 for e in event_rows if e[1]), - ) - - logger.info("[purge] Finding new backward extremities") - - # We calculate the new entries for the backward extremeties by finding - # events to be purged that are pointed to by events we're not going to - # purge. - txn.execute( - "SELECT DISTINCT e.event_id FROM events_to_purge AS e" - " INNER JOIN event_edges AS ed ON e.event_id = ed.prev_event_id" - " LEFT JOIN events_to_purge AS ep2 ON ed.event_id = ep2.event_id" - " WHERE ep2.event_id IS NULL" - ) - new_backwards_extrems = txn.fetchall() - - logger.info("[purge] replacing backward extremities: %r", new_backwards_extrems) - - txn.execute( - "DELETE FROM event_backward_extremities WHERE room_id = ?", (room_id,) - ) - - # Update backward extremeties - txn.executemany( - "INSERT INTO event_backward_extremities (room_id, event_id)" - " VALUES (?, ?)", - [(room_id, event_id) for event_id, in new_backwards_extrems], - ) - - logger.info("[purge] finding redundant state groups") - - # Get all state groups that are referenced by events that are to be - # deleted. We then go and check if they are referenced by other events - # or state groups, and if not we delete them. - txn.execute( - """ - SELECT DISTINCT state_group FROM events_to_purge - INNER JOIN event_to_state_groups USING (event_id) - """ - ) - - referenced_state_groups = set(sg for sg, in txn) - logger.info( - "[purge] found %i referenced state groups", len(referenced_state_groups) - ) - - logger.info("[purge] finding state groups that can be deleted") - - _ = self._find_unreferenced_groups_during_purge(txn, referenced_state_groups) - state_groups_to_delete, remaining_state_groups = _ - - logger.info( - "[purge] found %i state groups to delete", len(state_groups_to_delete) - ) - - logger.info( - "[purge] de-delta-ing %i remaining state groups", - len(remaining_state_groups), - ) - - # Now we turn the state groups that reference to-be-deleted state - # groups to non delta versions. - for sg in remaining_state_groups: - logger.info("[purge] de-delta-ing remaining state group %s", sg) - curr_state = self._get_state_groups_from_groups_txn(txn, [sg]) - curr_state = curr_state[sg] - - self._simple_delete_txn( - txn, table="state_groups_state", keyvalues={"state_group": sg} - ) - - self._simple_delete_txn( - txn, table="state_group_edges", keyvalues={"state_group": sg} - ) - - self._simple_insert_many_txn( - txn, - table="state_groups_state", - values=[ - { - "state_group": sg, - "room_id": room_id, - "type": key[0], - "state_key": key[1], - "event_id": state_id, - } - for key, state_id in iteritems(curr_state) - ], - ) - - logger.info("[purge] removing redundant state groups") - txn.executemany( - "DELETE FROM state_groups_state WHERE state_group = ?", - ((sg,) for sg in state_groups_to_delete), - ) - txn.executemany( - "DELETE FROM state_groups WHERE id = ?", - ((sg,) for sg in state_groups_to_delete), - ) - - logger.info("[purge] removing events from event_to_state_groups") - txn.execute( - "DELETE FROM event_to_state_groups " - "WHERE event_id IN (SELECT event_id from events_to_purge)" - ) - for event_id, _ in event_rows: - txn.call_after(self._get_state_group_for_event.invalidate, (event_id,)) - - # Delete all remote non-state events - for table in ( - "events", - "event_json", - "event_auth", - "event_edges", - "event_forward_extremities", - "event_reference_hashes", - "event_search", - "rejections", - ): - logger.info("[purge] removing events from %s", table) - - txn.execute( - "DELETE FROM %s WHERE event_id IN (" - " SELECT event_id FROM events_to_purge WHERE should_delete" - ")" % (table,) - ) - - # event_push_actions lacks an index on event_id, and has one on - # (room_id, event_id) instead. - for table in ("event_push_actions",): - logger.info("[purge] removing events from %s", table) - - txn.execute( - "DELETE FROM %s WHERE room_id = ? AND event_id IN (" - " SELECT event_id FROM events_to_purge WHERE should_delete" - ")" % (table,), - (room_id,), - ) - - # Mark all state and own events as outliers - logger.info("[purge] marking remaining events as outliers") - txn.execute( - "UPDATE events SET outlier = ?" - " WHERE event_id IN (" - " SELECT event_id FROM events_to_purge " - " WHERE NOT should_delete" - ")", - (True,), - ) - - # synapse tries to take out an exclusive lock on room_depth whenever it - # persists events (because upsert), and once we run this update, we - # will block that for the rest of our transaction. - # - # So, let's stick it at the end so that we don't block event - # persistence. - # - # We do this by calculating the minimum depth of the backwards - # extremities. However, the events in event_backward_extremities - # are ones we don't have yet so we need to look at the events that - # point to it via event_edges table. - txn.execute( - """ - SELECT COALESCE(MIN(depth), 0) - FROM event_backward_extremities AS eb - INNER JOIN event_edges AS eg ON eg.prev_event_id = eb.event_id - INNER JOIN events AS e ON e.event_id = eg.event_id - WHERE eb.room_id = ? - """, - (room_id,), - ) - min_depth, = txn.fetchone() - - logger.info("[purge] updating room_depth to %d", min_depth) - - txn.execute( - "UPDATE room_depth SET min_depth = ? WHERE room_id = ?", - (min_depth, room_id), - ) - - # finally, drop the temp table. this will commit the txn in sqlite, - # so make sure to keep this actually last. - txn.execute("DROP TABLE events_to_purge") - - logger.info("[purge] done") - - def _find_unreferenced_groups_during_purge(self, txn, state_groups): - """Used when purging history to figure out which state groups can be - deleted and which need to be de-delta'ed (due to one of its prev groups - being scheduled for deletion). - - Args: - txn - state_groups (set[int]): Set of state groups referenced by events - that are going to be deleted. - - Returns: - tuple[set[int], set[int]]: The set of state groups that can be - deleted and the set of state groups that need to be de-delta'ed - """ - # Graph of state group -> previous group - graph = {} - - # Set of events that we have found to be referenced by events - referenced_groups = set() - - # Set of state groups we've already seen - state_groups_seen = set(state_groups) - - # Set of state groups to handle next. - next_to_search = set(state_groups) - while next_to_search: - # We bound size of groups we're looking up at once, to stop the - # SQL query getting too big - if len(next_to_search) < 100: - current_search = next_to_search - next_to_search = set() - else: - current_search = set(itertools.islice(next_to_search, 100)) - next_to_search -= current_search - - # Check if state groups are referenced - sql = """ - SELECT DISTINCT state_group FROM event_to_state_groups - LEFT JOIN events_to_purge AS ep USING (event_id) - WHERE ep.event_id IS NULL AND - """ - clause, args = make_in_list_sql_clause( - txn.database_engine, "state_group", current_search - ) - txn.execute(sql + clause, list(args)) - - referenced = set(sg for sg, in txn) - referenced_groups |= referenced - - # We don't continue iterating up the state group graphs for state - # groups that are referenced. - current_search -= referenced - - rows = self._simple_select_many_txn( - txn, - table="state_group_edges", - column="prev_state_group", - iterable=current_search, - keyvalues={}, - retcols=("prev_state_group", "state_group"), - ) - - prevs = set(row["state_group"] for row in rows) - # We don't bother re-handling groups we've already seen - prevs -= state_groups_seen - next_to_search |= prevs - state_groups_seen |= prevs - - for row in rows: - # Note: Each state group can have at most one prev group - graph[row["state_group"]] = row["prev_state_group"] - - to_delete = state_groups_seen - referenced_groups - - to_dedelta = set() - for sg in referenced_groups: - prev_sg = graph.get(sg) - if prev_sg and prev_sg in to_delete: - to_dedelta.add(sg) - - return to_delete, to_dedelta - - def purge_room(self, room_id): - """Deletes all record of a room - - Args: - room_id (str): - """ - - return self.runInteraction("purge_room", self._purge_room_txn, room_id) - - def _purge_room_txn(self, txn, room_id): - # first we have to delete the state groups states - logger.info("[purge] removing %s from state_groups_state", room_id) - - txn.execute( - """ - DELETE FROM state_groups_state WHERE state_group IN ( - SELECT state_group FROM events JOIN event_to_state_groups USING(event_id) - WHERE events.room_id=? - ) - """, - (room_id,), - ) - - # ... and the state group edges - logger.info("[purge] removing %s from state_group_edges", room_id) - - txn.execute( - """ - DELETE FROM state_group_edges WHERE state_group IN ( - SELECT state_group FROM events JOIN event_to_state_groups USING(event_id) - WHERE events.room_id=? - ) - """, - (room_id,), - ) - - # ... and the state groups - logger.info("[purge] removing %s from state_groups", room_id) - - txn.execute( - """ - DELETE FROM state_groups WHERE id IN ( - SELECT state_group FROM events JOIN event_to_state_groups USING(event_id) - WHERE events.room_id=? - ) - """, - (room_id,), - ) - - # and then tables which lack an index on room_id but have one on event_id - for table in ( - "event_auth", - "event_edges", - "event_push_actions_staging", - "event_reference_hashes", - "event_relations", - "event_to_state_groups", - "redactions", - "rejections", - "state_events", - ): - logger.info("[purge] removing %s from %s", room_id, table) - - txn.execute( - """ - DELETE FROM %s WHERE event_id IN ( - SELECT event_id FROM events WHERE room_id=? - ) - """ - % (table,), - (room_id,), - ) - - # and finally, the tables with an index on room_id (or no useful index) - for table in ( - "current_state_events", - "event_backward_extremities", - "event_forward_extremities", - "event_json", - "event_push_actions", - "event_search", - "events", - "group_rooms", - "public_room_list_stream", - "receipts_graph", - "receipts_linearized", - "room_aliases", - "room_depth", - "room_memberships", - "room_stats_state", - "room_stats_current", - "room_stats_historical", - "room_stats_earliest_token", - "rooms", - "stream_ordering_to_exterm", - "topics", - "users_in_public_rooms", - "users_who_share_private_rooms", - # no useful index, but let's clear them anyway - "appservice_room_list", - "e2e_room_keys", - "event_push_summary", - "pusher_throttle", - "group_summary_rooms", - "local_invites", - "room_account_data", - "room_tags", - ): - logger.info("[purge] removing %s from %s", room_id, table) - txn.execute("DELETE FROM %s WHERE room_id=?" % (table,), (room_id,)) - - # Other tables we do NOT need to clear out: - # - # - blocked_rooms - # This is important, to make sure that we don't accidentally rejoin a blocked - # room after it was purged - # - # - user_directory - # This has a room_id column, but it is unused - # - - # Other tables that we might want to consider clearing out include: - # - # - event_reports - # Given that these are intended for abuse management my initial - # inclination is to leave them in place. - # - # - current_state_delta_stream - # - ex_outlier_stream - # - room_tags_revisions - # The problem with these is that they are largeish and there is no room_id - # index on them. In any case we should be clearing out 'stream' tables - # periodically anyway (#5888) - - # TODO: we could probably usefully do a bunch of cache invalidation here - - logger.info("[purge] done") - - @defer.inlineCallbacks - def is_event_after(self, event_id1, event_id2): - """Returns True if event_id1 is after event_id2 in the stream - """ - to_1, so_1 = yield self._get_event_ordering(event_id1) - to_2, so_2 = yield self._get_event_ordering(event_id2) - return (to_1, so_1) > (to_2, so_2) - - @cachedInlineCallbacks(max_entries=5000) - def _get_event_ordering(self, event_id): - res = yield self._simple_select_one( - table="events", - retcols=["topological_ordering", "stream_ordering"], - keyvalues={"event_id": event_id}, - allow_none=True, - ) - - if not res: - raise SynapseError(404, "Could not find event %s" % (event_id,)) - - return (int(res["topological_ordering"]), int(res["stream_ordering"])) - - def get_all_updated_current_state_deltas(self, from_token, to_token, limit): - def get_all_updated_current_state_deltas_txn(txn): - sql = """ - SELECT stream_id, room_id, type, state_key, event_id - FROM current_state_delta_stream - WHERE ? < stream_id AND stream_id <= ? - ORDER BY stream_id ASC LIMIT ? - """ - txn.execute(sql, (from_token, to_token, limit)) - return txn.fetchall() - - return self.runInteraction( - "get_all_updated_current_state_deltas", - get_all_updated_current_state_deltas_txn, - ) - - -AllNewEventsResult = namedtuple( - "AllNewEventsResult", - [ - "new_forward_events", - "new_backfill_events", - "forward_ex_outliers", - "backward_ex_outliers", - ], -) diff --git a/synapse/storage/events_bg_updates.py b/synapse/storage/events_bg_updates.py deleted file mode 100644 index 31ea6f917f..0000000000 --- a/synapse/storage/events_bg_updates.py +++ /dev/null @@ -1,505 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2019 The Matrix.org Foundation C.I.C. -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import logging - -from six import text_type - -from canonicaljson import json - -from twisted.internet import defer - -from synapse.storage._base import make_in_list_sql_clause -from synapse.storage.background_updates import BackgroundUpdateStore - -logger = logging.getLogger(__name__) - - -class EventsBackgroundUpdatesStore(BackgroundUpdateStore): - - EVENT_ORIGIN_SERVER_TS_NAME = "event_origin_server_ts" - EVENT_FIELDS_SENDER_URL_UPDATE_NAME = "event_fields_sender_url" - DELETE_SOFT_FAILED_EXTREMITIES = "delete_soft_failed_extremities" - - def __init__(self, db_conn, hs): - super(EventsBackgroundUpdatesStore, self).__init__(db_conn, hs) - - self.register_background_update_handler( - self.EVENT_ORIGIN_SERVER_TS_NAME, self._background_reindex_origin_server_ts - ) - self.register_background_update_handler( - self.EVENT_FIELDS_SENDER_URL_UPDATE_NAME, - self._background_reindex_fields_sender, - ) - - self.register_background_index_update( - "event_contains_url_index", - index_name="event_contains_url_index", - table="events", - columns=["room_id", "topological_ordering", "stream_ordering"], - where_clause="contains_url = true AND outlier = false", - ) - - # an event_id index on event_search is useful for the purge_history - # api. Plus it means we get to enforce some integrity with a UNIQUE - # clause - self.register_background_index_update( - "event_search_event_id_idx", - index_name="event_search_event_id_idx", - table="event_search", - columns=["event_id"], - unique=True, - psql_only=True, - ) - - self.register_background_update_handler( - self.DELETE_SOFT_FAILED_EXTREMITIES, self._cleanup_extremities_bg_update - ) - - self.register_background_update_handler( - "redactions_received_ts", self._redactions_received_ts - ) - - # This index gets deleted in `event_fix_redactions_bytes` update - self.register_background_index_update( - "event_fix_redactions_bytes_create_index", - index_name="redactions_censored_redacts", - table="redactions", - columns=["redacts"], - where_clause="have_censored", - ) - - self.register_background_update_handler( - "event_fix_redactions_bytes", self._event_fix_redactions_bytes - ) - - @defer.inlineCallbacks - def _background_reindex_fields_sender(self, progress, batch_size): - target_min_stream_id = progress["target_min_stream_id_inclusive"] - max_stream_id = progress["max_stream_id_exclusive"] - rows_inserted = progress.get("rows_inserted", 0) - - INSERT_CLUMP_SIZE = 1000 - - def reindex_txn(txn): - sql = ( - "SELECT stream_ordering, event_id, json FROM events" - " INNER JOIN event_json USING (event_id)" - " WHERE ? <= stream_ordering AND stream_ordering < ?" - " ORDER BY stream_ordering DESC" - " LIMIT ?" - ) - - txn.execute(sql, (target_min_stream_id, max_stream_id, batch_size)) - - rows = txn.fetchall() - if not rows: - return 0 - - min_stream_id = rows[-1][0] - - update_rows = [] - for row in rows: - try: - event_id = row[1] - event_json = json.loads(row[2]) - sender = event_json["sender"] - content = event_json["content"] - - contains_url = "url" in content - if contains_url: - contains_url &= isinstance(content["url"], text_type) - except (KeyError, AttributeError): - # If the event is missing a necessary field then - # skip over it. - continue - - update_rows.append((sender, contains_url, event_id)) - - sql = "UPDATE events SET sender = ?, contains_url = ? WHERE event_id = ?" - - for index in range(0, len(update_rows), INSERT_CLUMP_SIZE): - clump = update_rows[index : index + INSERT_CLUMP_SIZE] - txn.executemany(sql, clump) - - progress = { - "target_min_stream_id_inclusive": target_min_stream_id, - "max_stream_id_exclusive": min_stream_id, - "rows_inserted": rows_inserted + len(rows), - } - - self._background_update_progress_txn( - txn, self.EVENT_FIELDS_SENDER_URL_UPDATE_NAME, progress - ) - - return len(rows) - - result = yield self.runInteraction( - self.EVENT_FIELDS_SENDER_URL_UPDATE_NAME, reindex_txn - ) - - if not result: - yield self._end_background_update(self.EVENT_FIELDS_SENDER_URL_UPDATE_NAME) - - return result - - @defer.inlineCallbacks - def _background_reindex_origin_server_ts(self, progress, batch_size): - target_min_stream_id = progress["target_min_stream_id_inclusive"] - max_stream_id = progress["max_stream_id_exclusive"] - rows_inserted = progress.get("rows_inserted", 0) - - INSERT_CLUMP_SIZE = 1000 - - def reindex_search_txn(txn): - sql = ( - "SELECT stream_ordering, event_id FROM events" - " WHERE ? <= stream_ordering AND stream_ordering < ?" - " ORDER BY stream_ordering DESC" - " LIMIT ?" - ) - - txn.execute(sql, (target_min_stream_id, max_stream_id, batch_size)) - - rows = txn.fetchall() - if not rows: - return 0 - - min_stream_id = rows[-1][0] - event_ids = [row[1] for row in rows] - - rows_to_update = [] - - chunks = [event_ids[i : i + 100] for i in range(0, len(event_ids), 100)] - for chunk in chunks: - ev_rows = self._simple_select_many_txn( - txn, - table="event_json", - column="event_id", - iterable=chunk, - retcols=["event_id", "json"], - keyvalues={}, - ) - - for row in ev_rows: - event_id = row["event_id"] - event_json = json.loads(row["json"]) - try: - origin_server_ts = event_json["origin_server_ts"] - except (KeyError, AttributeError): - # If the event is missing a necessary field then - # skip over it. - continue - - rows_to_update.append((origin_server_ts, event_id)) - - sql = "UPDATE events SET origin_server_ts = ? WHERE event_id = ?" - - for index in range(0, len(rows_to_update), INSERT_CLUMP_SIZE): - clump = rows_to_update[index : index + INSERT_CLUMP_SIZE] - txn.executemany(sql, clump) - - progress = { - "target_min_stream_id_inclusive": target_min_stream_id, - "max_stream_id_exclusive": min_stream_id, - "rows_inserted": rows_inserted + len(rows_to_update), - } - - self._background_update_progress_txn( - txn, self.EVENT_ORIGIN_SERVER_TS_NAME, progress - ) - - return len(rows_to_update) - - result = yield self.runInteraction( - self.EVENT_ORIGIN_SERVER_TS_NAME, reindex_search_txn - ) - - if not result: - yield self._end_background_update(self.EVENT_ORIGIN_SERVER_TS_NAME) - - return result - - @defer.inlineCallbacks - def _cleanup_extremities_bg_update(self, progress, batch_size): - """Background update to clean out extremities that should have been - deleted previously. - - Mainly used to deal with the aftermath of #5269. - """ - - # This works by first copying all existing forward extremities into the - # `_extremities_to_check` table at start up, and then checking each - # event in that table whether we have any descendants that are not - # soft-failed/rejected. If that is the case then we delete that event - # from the forward extremities table. - # - # For efficiency, we do this in batches by recursively pulling out all - # descendants of a batch until we find the non soft-failed/rejected - # events, i.e. the set of descendants whose chain of prev events back - # to the batch of extremities are all soft-failed or rejected. - # Typically, we won't find any such events as extremities will rarely - # have any descendants, but if they do then we should delete those - # extremities. - - def _cleanup_extremities_bg_update_txn(txn): - # The set of extremity event IDs that we're checking this round - original_set = set() - - # A dict[str, set[str]] of event ID to their prev events. - graph = {} - - # The set of descendants of the original set that are not rejected - # nor soft-failed. Ancestors of these events should be removed - # from the forward extremities table. - non_rejected_leaves = set() - - # Set of event IDs that have been soft failed, and for which we - # should check if they have descendants which haven't been soft - # failed. - soft_failed_events_to_lookup = set() - - # First, we get `batch_size` events from the table, pulling out - # their successor events, if any, and the successor events' - # rejection status. - txn.execute( - """SELECT prev_event_id, event_id, internal_metadata, - rejections.event_id IS NOT NULL, events.outlier - FROM ( - SELECT event_id AS prev_event_id - FROM _extremities_to_check - LIMIT ? - ) AS f - LEFT JOIN event_edges USING (prev_event_id) - LEFT JOIN events USING (event_id) - LEFT JOIN event_json USING (event_id) - LEFT JOIN rejections USING (event_id) - """, - (batch_size,), - ) - - for prev_event_id, event_id, metadata, rejected, outlier in txn: - original_set.add(prev_event_id) - - if not event_id or outlier: - # Common case where the forward extremity doesn't have any - # descendants. - continue - - graph.setdefault(event_id, set()).add(prev_event_id) - - soft_failed = False - if metadata: - soft_failed = json.loads(metadata).get("soft_failed") - - if soft_failed or rejected: - soft_failed_events_to_lookup.add(event_id) - else: - non_rejected_leaves.add(event_id) - - # Now we recursively check all the soft-failed descendants we - # found above in the same way, until we have nothing left to - # check. - while soft_failed_events_to_lookup: - # We only want to do 100 at a time, so we split given list - # into two. - batch = list(soft_failed_events_to_lookup) - to_check, to_defer = batch[:100], batch[100:] - soft_failed_events_to_lookup = set(to_defer) - - sql = """SELECT prev_event_id, event_id, internal_metadata, - rejections.event_id IS NOT NULL - FROM event_edges - INNER JOIN events USING (event_id) - INNER JOIN event_json USING (event_id) - LEFT JOIN rejections USING (event_id) - WHERE - NOT events.outlier - AND - """ - clause, args = make_in_list_sql_clause( - self.database_engine, "prev_event_id", to_check - ) - txn.execute(sql + clause, list(args)) - - for prev_event_id, event_id, metadata, rejected in txn: - if event_id in graph: - # Already handled this event previously, but we still - # want to record the edge. - graph[event_id].add(prev_event_id) - continue - - graph[event_id] = {prev_event_id} - - soft_failed = json.loads(metadata).get("soft_failed") - if soft_failed or rejected: - soft_failed_events_to_lookup.add(event_id) - else: - non_rejected_leaves.add(event_id) - - # We have a set of non-soft-failed descendants, so we recurse up - # the graph to find all ancestors and add them to the set of event - # IDs that we can delete from forward extremities table. - to_delete = set() - while non_rejected_leaves: - event_id = non_rejected_leaves.pop() - prev_event_ids = graph.get(event_id, set()) - non_rejected_leaves.update(prev_event_ids) - to_delete.update(prev_event_ids) - - to_delete.intersection_update(original_set) - - deleted = self._simple_delete_many_txn( - txn=txn, - table="event_forward_extremities", - column="event_id", - iterable=to_delete, - keyvalues={}, - ) - - logger.info( - "Deleted %d forward extremities of %d checked, to clean up #5269", - deleted, - len(original_set), - ) - - if deleted: - # We now need to invalidate the caches of these rooms - rows = self._simple_select_many_txn( - txn, - table="events", - column="event_id", - iterable=to_delete, - keyvalues={}, - retcols=("room_id",), - ) - room_ids = set(row["room_id"] for row in rows) - for room_id in room_ids: - txn.call_after( - self.get_latest_event_ids_in_room.invalidate, (room_id,) - ) - - self._simple_delete_many_txn( - txn=txn, - table="_extremities_to_check", - column="event_id", - iterable=original_set, - keyvalues={}, - ) - - return len(original_set) - - num_handled = yield self.runInteraction( - "_cleanup_extremities_bg_update", _cleanup_extremities_bg_update_txn - ) - - if not num_handled: - yield self._end_background_update(self.DELETE_SOFT_FAILED_EXTREMITIES) - - def _drop_table_txn(txn): - txn.execute("DROP TABLE _extremities_to_check") - - yield self.runInteraction( - "_cleanup_extremities_bg_update_drop_table", _drop_table_txn - ) - - return num_handled - - @defer.inlineCallbacks - def _redactions_received_ts(self, progress, batch_size): - """Handles filling out the `received_ts` column in redactions. - """ - last_event_id = progress.get("last_event_id", "") - - def _redactions_received_ts_txn(txn): - # Fetch the set of event IDs that we want to update - sql = """ - SELECT event_id FROM redactions - WHERE event_id > ? - ORDER BY event_id ASC - LIMIT ? - """ - - txn.execute(sql, (last_event_id, batch_size)) - - rows = txn.fetchall() - if not rows: - return 0 - - upper_event_id, = rows[-1] - - # Update the redactions with the received_ts. - # - # Note: Not all events have an associated received_ts, so we - # fallback to using origin_server_ts. If we for some reason don't - # have an origin_server_ts, lets just use the current timestamp. - # - # We don't want to leave it null, as then we'll never try and - # censor those redactions. - sql = """ - UPDATE redactions - SET received_ts = ( - SELECT COALESCE(received_ts, origin_server_ts, ?) FROM events - WHERE events.event_id = redactions.event_id - ) - WHERE ? <= event_id AND event_id <= ? - """ - - txn.execute(sql, (self._clock.time_msec(), last_event_id, upper_event_id)) - - self._background_update_progress_txn( - txn, "redactions_received_ts", {"last_event_id": upper_event_id} - ) - - return len(rows) - - count = yield self.runInteraction( - "_redactions_received_ts", _redactions_received_ts_txn - ) - - if not count: - yield self._end_background_update("redactions_received_ts") - - return count - - @defer.inlineCallbacks - def _event_fix_redactions_bytes(self, progress, batch_size): - """Undoes hex encoded censored redacted event JSON. - """ - - def _event_fix_redactions_bytes_txn(txn): - # This update is quite fast due to new index. - txn.execute( - """ - UPDATE event_json - SET - json = convert_from(json::bytea, 'utf8') - FROM redactions - WHERE - redactions.have_censored - AND event_json.event_id = redactions.redacts - AND json NOT LIKE '{%'; - """ - ) - - txn.execute("DROP INDEX redactions_censored_redacts") - - yield self.runInteraction( - "_event_fix_redactions_bytes", _event_fix_redactions_bytes_txn - ) - - yield self._end_background_update("event_fix_redactions_bytes") - - return 1 diff --git a/synapse/storage/events_worker.py b/synapse/storage/events_worker.py deleted file mode 100644 index 4c4b76bd93..0000000000 --- a/synapse/storage/events_worker.py +++ /dev/null @@ -1,882 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2018 New Vector Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -from __future__ import division - -import itertools -import logging -from collections import namedtuple - -from canonicaljson import json - -from twisted.internet import defer - -from synapse.api.constants import EventTypes -from synapse.api.errors import NotFoundError -from synapse.api.room_versions import EventFormatVersions -from synapse.events import FrozenEvent, event_type_from_format_version # noqa: F401 -from synapse.events.snapshot import EventContext # noqa: F401 -from synapse.events.utils import prune_event -from synapse.logging.context import LoggingContext, PreserveLoggingContext -from synapse.metrics.background_process_metrics import run_as_background_process -from synapse.storage._base import SQLBaseStore, make_in_list_sql_clause -from synapse.types import get_domain_from_id -from synapse.util import batch_iter -from synapse.util.metrics import Measure - -logger = logging.getLogger(__name__) - - -# These values are used in the `enqueus_event` and `_do_fetch` methods to -# control how we batch/bulk fetch events from the database. -# The values are plucked out of thing air to make initial sync run faster -# on jki.re -# TODO: Make these configurable. -EVENT_QUEUE_THREADS = 3 # Max number of threads that will fetch events -EVENT_QUEUE_ITERATIONS = 3 # No. times we block waiting for requests for events -EVENT_QUEUE_TIMEOUT_S = 0.1 # Timeout when waiting for requests for events - - -_EventCacheEntry = namedtuple("_EventCacheEntry", ("event", "redacted_event")) - - -class EventsWorkerStore(SQLBaseStore): - def get_received_ts(self, event_id): - """Get received_ts (when it was persisted) for the event. - - Raises an exception for unknown events. - - Args: - event_id (str) - - Returns: - Deferred[int|None]: Timestamp in milliseconds, or None for events - that were persisted before received_ts was implemented. - """ - return self._simple_select_one_onecol( - table="events", - keyvalues={"event_id": event_id}, - retcol="received_ts", - desc="get_received_ts", - ) - - def get_received_ts_by_stream_pos(self, stream_ordering): - """Given a stream ordering get an approximate timestamp of when it - happened. - - This is done by simply taking the received ts of the first event that - has a stream ordering greater than or equal to the given stream pos. - If none exists returns the current time, on the assumption that it must - have happened recently. - - Args: - stream_ordering (int) - - Returns: - Deferred[int] - """ - - def _get_approximate_received_ts_txn(txn): - sql = """ - SELECT received_ts FROM events - WHERE stream_ordering >= ? - LIMIT 1 - """ - - txn.execute(sql, (stream_ordering,)) - row = txn.fetchone() - if row and row[0]: - ts = row[0] - else: - ts = self.clock.time_msec() - - return ts - - return self.runInteraction( - "get_approximate_received_ts", _get_approximate_received_ts_txn - ) - - @defer.inlineCallbacks - def get_event( - self, - event_id, - check_redacted=True, - get_prev_content=False, - allow_rejected=False, - allow_none=False, - check_room_id=None, - ): - """Get an event from the database by event_id. - - Args: - event_id (str): The event_id of the event to fetch - check_redacted (bool): If True, check if event has been redacted - and redact it. - get_prev_content (bool): If True and event is a state event, - include the previous states content in the unsigned field. - allow_rejected (bool): If True return rejected events. - allow_none (bool): If True, return None if no event found, if - False throw a NotFoundError - check_room_id (str|None): if not None, check the room of the found event. - If there is a mismatch, behave as per allow_none. - - Returns: - Deferred[EventBase|None] - """ - if not isinstance(event_id, str): - raise TypeError("Invalid event event_id %r" % (event_id,)) - - events = yield self.get_events_as_list( - [event_id], - check_redacted=check_redacted, - get_prev_content=get_prev_content, - allow_rejected=allow_rejected, - ) - - event = events[0] if events else None - - if event is not None and check_room_id is not None: - if event.room_id != check_room_id: - event = None - - if event is None and not allow_none: - raise NotFoundError("Could not find event %s" % (event_id,)) - - return event - - @defer.inlineCallbacks - def get_events( - self, - event_ids, - check_redacted=True, - get_prev_content=False, - allow_rejected=False, - ): - """Get events from the database - - Args: - event_ids (list): The event_ids of the events to fetch - check_redacted (bool): If True, check if event has been redacted - and redact it. - get_prev_content (bool): If True and event is a state event, - include the previous states content in the unsigned field. - allow_rejected (bool): If True return rejected events. - - Returns: - Deferred : Dict from event_id to event. - """ - events = yield self.get_events_as_list( - event_ids, - check_redacted=check_redacted, - get_prev_content=get_prev_content, - allow_rejected=allow_rejected, - ) - - return {e.event_id: e for e in events} - - @defer.inlineCallbacks - def get_events_as_list( - self, - event_ids, - check_redacted=True, - get_prev_content=False, - allow_rejected=False, - ): - """Get events from the database and return in a list in the same order - as given by `event_ids` arg. - - Args: - event_ids (list): The event_ids of the events to fetch - check_redacted (bool): If True, check if event has been redacted - and redact it. - get_prev_content (bool): If True and event is a state event, - include the previous states content in the unsigned field. - allow_rejected (bool): If True return rejected events. - - Returns: - Deferred[list[EventBase]]: List of events fetched from the database. The - events are in the same order as `event_ids` arg. - - Note that the returned list may be smaller than the list of event - IDs if not all events could be fetched. - """ - - if not event_ids: - return [] - - # there may be duplicates so we cast the list to a set - event_entry_map = yield self._get_events_from_cache_or_db( - set(event_ids), allow_rejected=allow_rejected - ) - - events = [] - for event_id in event_ids: - entry = event_entry_map.get(event_id, None) - if not entry: - continue - - if not allow_rejected: - assert not entry.event.rejected_reason, ( - "rejected event returned from _get_events_from_cache_or_db despite " - "allow_rejected=False" - ) - - # We may not have had the original event when we received a redaction, so - # we have to recheck auth now. - - if not allow_rejected and entry.event.type == EventTypes.Redaction: - if not hasattr(entry.event, "redacts"): - # A redacted redaction doesn't have a `redacts` key, in - # which case lets just withhold the event. - # - # Note: Most of the time if the redactions has been - # redacted we still have the un-redacted event in the DB - # and so we'll still see the `redacts` key. However, this - # isn't always true e.g. if we have censored the event. - logger.debug( - "Withholding redaction event %s as we don't have redacts key", - event_id, - ) - continue - - redacted_event_id = entry.event.redacts - event_map = yield self._get_events_from_cache_or_db([redacted_event_id]) - original_event_entry = event_map.get(redacted_event_id) - if not original_event_entry: - # we don't have the redacted event (or it was rejected). - # - # We assume that the redaction isn't authorized for now; if the - # redacted event later turns up, the redaction will be re-checked, - # and if it is found valid, the original will get redacted before it - # is served to the client. - logger.debug( - "Withholding redaction event %s since we don't (yet) have the " - "original %s", - event_id, - redacted_event_id, - ) - continue - - original_event = original_event_entry.event - if original_event.type == EventTypes.Create: - # we never serve redactions of Creates to clients. - logger.info( - "Withholding redaction %s of create event %s", - event_id, - redacted_event_id, - ) - continue - - if original_event.room_id != entry.event.room_id: - logger.info( - "Withholding redaction %s of event %s from a different room", - event_id, - redacted_event_id, - ) - continue - - if entry.event.internal_metadata.need_to_check_redaction(): - original_domain = get_domain_from_id(original_event.sender) - redaction_domain = get_domain_from_id(entry.event.sender) - if original_domain != redaction_domain: - # the senders don't match, so this is forbidden - logger.info( - "Withholding redaction %s whose sender domain %s doesn't " - "match that of redacted event %s %s", - event_id, - redaction_domain, - redacted_event_id, - original_domain, - ) - continue - - # Update the cache to save doing the checks again. - entry.event.internal_metadata.recheck_redaction = False - - if check_redacted and entry.redacted_event: - event = entry.redacted_event - else: - event = entry.event - - events.append(event) - - if get_prev_content: - if "replaces_state" in event.unsigned: - prev = yield self.get_event( - event.unsigned["replaces_state"], - get_prev_content=False, - allow_none=True, - ) - if prev: - event.unsigned = dict(event.unsigned) - event.unsigned["prev_content"] = prev.content - event.unsigned["prev_sender"] = prev.sender - - return events - - @defer.inlineCallbacks - def _get_events_from_cache_or_db(self, event_ids, allow_rejected=False): - """Fetch a bunch of events from the cache or the database. - - If events are pulled from the database, they will be cached for future lookups. - - Args: - event_ids (Iterable[str]): The event_ids of the events to fetch - allow_rejected (bool): Whether to include rejected events - - Returns: - Deferred[Dict[str, _EventCacheEntry]]: - map from event id to result - """ - event_entry_map = self._get_events_from_cache( - event_ids, allow_rejected=allow_rejected - ) - - missing_events_ids = [e for e in event_ids if e not in event_entry_map] - - if missing_events_ids: - log_ctx = LoggingContext.current_context() - log_ctx.record_event_fetch(len(missing_events_ids)) - - # Note that _get_events_from_db is also responsible for turning db rows - # into FrozenEvents (via _get_event_from_row), which involves seeing if - # the events have been redacted, and if so pulling the redaction event out - # of the database to check it. - # - missing_events = yield self._get_events_from_db( - missing_events_ids, allow_rejected=allow_rejected - ) - - event_entry_map.update(missing_events) - - return event_entry_map - - def _invalidate_get_event_cache(self, event_id): - self._get_event_cache.invalidate((event_id,)) - - def _get_events_from_cache(self, events, allow_rejected, update_metrics=True): - """Fetch events from the caches - - Args: - events (Iterable[str]): list of event_ids to fetch - allow_rejected (bool): Whether to return events that were rejected - update_metrics (bool): Whether to update the cache hit ratio metrics - - Returns: - dict of event_id -> _EventCacheEntry for each event_id in cache. If - allow_rejected is `False` then there will still be an entry but it - will be `None` - """ - event_map = {} - - for event_id in events: - ret = self._get_event_cache.get( - (event_id,), None, update_metrics=update_metrics - ) - if not ret: - continue - - if allow_rejected or not ret.event.rejected_reason: - event_map[event_id] = ret - else: - event_map[event_id] = None - - return event_map - - def _do_fetch(self, conn): - """Takes a database connection and waits for requests for events from - the _event_fetch_list queue. - """ - i = 0 - while True: - with self._event_fetch_lock: - event_list = self._event_fetch_list - self._event_fetch_list = [] - - if not event_list: - single_threaded = self.database_engine.single_threaded - if single_threaded or i > EVENT_QUEUE_ITERATIONS: - self._event_fetch_ongoing -= 1 - return - else: - self._event_fetch_lock.wait(EVENT_QUEUE_TIMEOUT_S) - i += 1 - continue - i = 0 - - self._fetch_event_list(conn, event_list) - - def _fetch_event_list(self, conn, event_list): - """Handle a load of requests from the _event_fetch_list queue - - Args: - conn (twisted.enterprise.adbapi.Connection): database connection - - event_list (list[Tuple[list[str], Deferred]]): - The fetch requests. Each entry consists of a list of event - ids to be fetched, and a deferred to be completed once the - events have been fetched. - - The deferreds are callbacked with a dictionary mapping from event id - to event row. Note that it may well contain additional events that - were not part of this request. - """ - with Measure(self._clock, "_fetch_event_list"): - try: - events_to_fetch = set( - event_id for events, _ in event_list for event_id in events - ) - - row_dict = self._new_transaction( - conn, "do_fetch", [], [], self._fetch_event_rows, events_to_fetch - ) - - # We only want to resolve deferreds from the main thread - def fire(): - for _, d in event_list: - d.callback(row_dict) - - with PreserveLoggingContext(): - self.hs.get_reactor().callFromThread(fire) - except Exception as e: - logger.exception("do_fetch") - - # We only want to resolve deferreds from the main thread - def fire(evs, exc): - for _, d in evs: - if not d.called: - with PreserveLoggingContext(): - d.errback(exc) - - with PreserveLoggingContext(): - self.hs.get_reactor().callFromThread(fire, event_list, e) - - @defer.inlineCallbacks - def _get_events_from_db(self, event_ids, allow_rejected=False): - """Fetch a bunch of events from the database. - - Returned events will be added to the cache for future lookups. - - Args: - event_ids (Iterable[str]): The event_ids of the events to fetch - allow_rejected (bool): Whether to include rejected events - - Returns: - Deferred[Dict[str, _EventCacheEntry]]: - map from event id to result. May return extra events which - weren't asked for. - """ - fetched_events = {} - events_to_fetch = event_ids - - while events_to_fetch: - row_map = yield self._enqueue_events(events_to_fetch) - - # we need to recursively fetch any redactions of those events - redaction_ids = set() - for event_id in events_to_fetch: - row = row_map.get(event_id) - fetched_events[event_id] = row - if row: - redaction_ids.update(row["redactions"]) - - events_to_fetch = redaction_ids.difference(fetched_events.keys()) - if events_to_fetch: - logger.debug("Also fetching redaction events %s", events_to_fetch) - - # build a map from event_id to EventBase - event_map = {} - for event_id, row in fetched_events.items(): - if not row: - continue - assert row["event_id"] == event_id - - rejected_reason = row["rejected_reason"] - - if not allow_rejected and rejected_reason: - continue - - d = json.loads(row["json"]) - internal_metadata = json.loads(row["internal_metadata"]) - - format_version = row["format_version"] - if format_version is None: - # This means that we stored the event before we had the concept - # of a event format version, so it must be a V1 event. - format_version = EventFormatVersions.V1 - - original_ev = event_type_from_format_version(format_version)( - event_dict=d, - internal_metadata_dict=internal_metadata, - rejected_reason=rejected_reason, - ) - - event_map[event_id] = original_ev - - # finally, we can decide whether each one nededs redacting, and build - # the cache entries. - result_map = {} - for event_id, original_ev in event_map.items(): - redactions = fetched_events[event_id]["redactions"] - redacted_event = self._maybe_redact_event_row( - original_ev, redactions, event_map - ) - - cache_entry = _EventCacheEntry( - event=original_ev, redacted_event=redacted_event - ) - - self._get_event_cache.prefill((event_id,), cache_entry) - result_map[event_id] = cache_entry - - return result_map - - @defer.inlineCallbacks - def _enqueue_events(self, events): - """Fetches events from the database using the _event_fetch_list. This - allows batch and bulk fetching of events - it allows us to fetch events - without having to create a new transaction for each request for events. - - Args: - events (Iterable[str]): events to be fetched. - - Returns: - Deferred[Dict[str, Dict]]: map from event id to row data from the database. - May contain events that weren't requested. - """ - - events_d = defer.Deferred() - with self._event_fetch_lock: - self._event_fetch_list.append((events, events_d)) - - self._event_fetch_lock.notify() - - if self._event_fetch_ongoing < EVENT_QUEUE_THREADS: - self._event_fetch_ongoing += 1 - should_start = True - else: - should_start = False - - if should_start: - run_as_background_process( - "fetch_events", self.runWithConnection, self._do_fetch - ) - - logger.debug("Loading %d events: %s", len(events), events) - with PreserveLoggingContext(): - row_map = yield events_d - logger.debug("Loaded %d events (%d rows)", len(events), len(row_map)) - - return row_map - - def _fetch_event_rows(self, txn, event_ids): - """Fetch event rows from the database - - Events which are not found are omitted from the result. - - The returned per-event dicts contain the following keys: - - * event_id (str) - - * json (str): json-encoded event structure - - * internal_metadata (str): json-encoded internal metadata dict - - * format_version (int|None): The format of the event. Hopefully one - of EventFormatVersions. 'None' means the event predates - EventFormatVersions (so the event is format V1). - - * rejected_reason (str|None): if the event was rejected, the reason - why. - - * redactions (List[str]): a list of event-ids which (claim to) redact - this event. - - Args: - txn (twisted.enterprise.adbapi.Connection): - event_ids (Iterable[str]): event IDs to fetch - - Returns: - Dict[str, Dict]: a map from event id to event info. - """ - event_dict = {} - for evs in batch_iter(event_ids, 200): - sql = ( - "SELECT " - " e.event_id, " - " e.internal_metadata," - " e.json," - " e.format_version, " - " rej.reason " - " FROM event_json as e" - " LEFT JOIN rejections as rej USING (event_id)" - " WHERE " - ) - - clause, args = make_in_list_sql_clause( - txn.database_engine, "e.event_id", evs - ) - - txn.execute(sql + clause, args) - - for row in txn: - event_id = row[0] - event_dict[event_id] = { - "event_id": event_id, - "internal_metadata": row[1], - "json": row[2], - "format_version": row[3], - "rejected_reason": row[4], - "redactions": [], - } - - # check for redactions - redactions_sql = "SELECT event_id, redacts FROM redactions WHERE " - - clause, args = make_in_list_sql_clause(txn.database_engine, "redacts", evs) - - txn.execute(redactions_sql + clause, args) - - for (redacter, redacted) in txn: - d = event_dict.get(redacted) - if d: - d["redactions"].append(redacter) - - return event_dict - - def _maybe_redact_event_row(self, original_ev, redactions, event_map): - """Given an event object and a list of possible redacting event ids, - determine whether to honour any of those redactions and if so return a redacted - event. - - Args: - original_ev (EventBase): - redactions (iterable[str]): list of event ids of potential redaction events - event_map (dict[str, EventBase]): other events which have been fetched, in - which we can look up the redaaction events. Map from event id to event. - - Returns: - Deferred[EventBase|None]: if the event should be redacted, a pruned - event object. Otherwise, None. - """ - if original_ev.type == "m.room.create": - # we choose to ignore redactions of m.room.create events. - return None - - for redaction_id in redactions: - redaction_event = event_map.get(redaction_id) - if not redaction_event or redaction_event.rejected_reason: - # we don't have the redaction event, or the redaction event was not - # authorized. - logger.debug( - "%s was redacted by %s but redaction not found/authed", - original_ev.event_id, - redaction_id, - ) - continue - - if redaction_event.room_id != original_ev.room_id: - logger.debug( - "%s was redacted by %s but redaction was in a different room!", - original_ev.event_id, - redaction_id, - ) - continue - - # Starting in room version v3, some redactions need to be - # rechecked if we didn't have the redacted event at the - # time, so we recheck on read instead. - if redaction_event.internal_metadata.need_to_check_redaction(): - expected_domain = get_domain_from_id(original_ev.sender) - if get_domain_from_id(redaction_event.sender) == expected_domain: - # This redaction event is allowed. Mark as not needing a recheck. - redaction_event.internal_metadata.recheck_redaction = False - else: - # Senders don't match, so the event isn't actually redacted - logger.debug( - "%s was redacted by %s but the senders don't match", - original_ev.event_id, - redaction_id, - ) - continue - - logger.debug("Redacting %s due to %s", original_ev.event_id, redaction_id) - - # we found a good redaction event. Redact! - redacted_event = prune_event(original_ev) - redacted_event.unsigned["redacted_by"] = redaction_id - - # It's fine to add the event directly, since get_pdu_json - # will serialise this field correctly - redacted_event.unsigned["redacted_because"] = redaction_event - - return redacted_event - - # no valid redaction found for this event - return None - - @defer.inlineCallbacks - def have_events_in_timeline(self, event_ids): - """Given a list of event ids, check if we have already processed and - stored them as non outliers. - """ - rows = yield self._simple_select_many_batch( - table="events", - retcols=("event_id",), - column="event_id", - iterable=list(event_ids), - keyvalues={"outlier": False}, - desc="have_events_in_timeline", - ) - - return set(r["event_id"] for r in rows) - - @defer.inlineCallbacks - def have_seen_events(self, event_ids): - """Given a list of event ids, check if we have already processed them. - - Args: - event_ids (iterable[str]): - - Returns: - Deferred[set[str]]: The events we have already seen. - """ - results = set() - - def have_seen_events_txn(txn, chunk): - sql = "SELECT event_id FROM events as e WHERE " - clause, args = make_in_list_sql_clause( - txn.database_engine, "e.event_id", chunk - ) - txn.execute(sql + clause, args) - for (event_id,) in txn: - results.add(event_id) - - # break the input up into chunks of 100 - input_iterator = iter(event_ids) - for chunk in iter(lambda: list(itertools.islice(input_iterator, 100)), []): - yield self.runInteraction("have_seen_events", have_seen_events_txn, chunk) - return results - - def get_seen_events_with_rejections(self, event_ids): - """Given a list of event ids, check if we rejected them. - - Args: - event_ids (list[str]) - - Returns: - Deferred[dict[str, str|None): - Has an entry for each event id we already have seen. Maps to - the rejected reason string if we rejected the event, else maps - to None. - """ - if not event_ids: - return defer.succeed({}) - - def f(txn): - sql = ( - "SELECT e.event_id, reason FROM events as e " - "LEFT JOIN rejections as r ON e.event_id = r.event_id " - "WHERE e.event_id = ?" - ) - - res = {} - for event_id in event_ids: - txn.execute(sql, (event_id,)) - row = txn.fetchone() - if row: - _, rejected = row - res[event_id] = rejected - - return res - - return self.runInteraction("get_seen_events_with_rejections", f) - - def _get_total_state_event_counts_txn(self, txn, room_id): - """ - See get_total_state_event_counts. - """ - # We join against the events table as that has an index on room_id - sql = """ - SELECT COUNT(*) FROM state_events - INNER JOIN events USING (room_id, event_id) - WHERE room_id=? - """ - txn.execute(sql, (room_id,)) - row = txn.fetchone() - return row[0] if row else 0 - - def get_total_state_event_counts(self, room_id): - """ - Gets the total number of state events in a room. - - Args: - room_id (str) - - Returns: - Deferred[int] - """ - return self.runInteraction( - "get_total_state_event_counts", - self._get_total_state_event_counts_txn, - room_id, - ) - - def _get_current_state_event_counts_txn(self, txn, room_id): - """ - See get_current_state_event_counts. - """ - sql = "SELECT COUNT(*) FROM current_state_events WHERE room_id=?" - txn.execute(sql, (room_id,)) - row = txn.fetchone() - return row[0] if row else 0 - - def get_current_state_event_counts(self, room_id): - """ - Gets the current number of state events in a room. - - Args: - room_id (str) - - Returns: - Deferred[int] - """ - return self.runInteraction( - "get_current_state_event_counts", - self._get_current_state_event_counts_txn, - room_id, - ) - - @defer.inlineCallbacks - def get_room_complexity(self, room_id): - """ - Get a rough approximation of the complexity of the room. This is used by - remote servers to decide whether they wish to join the room or not. - Higher complexity value indicates that being in the room will consume - more resources. - - Args: - room_id (str) - - Returns: - Deferred[dict[str:int]] of complexity version to complexity. - """ - state_events = yield self.get_current_state_event_counts(room_id) - - # Call this one "v1", so we can introduce new ones as we want to develop - # it. - complexity_v1 = round(state_events / 500, 2) - - return {"v1": complexity_v1} diff --git a/synapse/storage/filtering.py b/synapse/storage/filtering.py deleted file mode 100644 index 7c2a7da836..0000000000 --- a/synapse/storage/filtering.py +++ /dev/null @@ -1,75 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2015, 2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -from canonicaljson import encode_canonical_json - -from synapse.api.errors import Codes, SynapseError -from synapse.util.caches.descriptors import cachedInlineCallbacks - -from ._base import SQLBaseStore, db_to_json - - -class FilteringStore(SQLBaseStore): - @cachedInlineCallbacks(num_args=2) - def get_user_filter(self, user_localpart, filter_id): - # filter_id is BIGINT UNSIGNED, so if it isn't a number, fail - # with a coherent error message rather than 500 M_UNKNOWN. - try: - int(filter_id) - except ValueError: - raise SynapseError(400, "Invalid filter ID", Codes.INVALID_PARAM) - - def_json = yield self._simple_select_one_onecol( - table="user_filters", - keyvalues={"user_id": user_localpart, "filter_id": filter_id}, - retcol="filter_json", - allow_none=False, - desc="get_user_filter", - ) - - return db_to_json(def_json) - - def add_user_filter(self, user_localpart, user_filter): - def_json = encode_canonical_json(user_filter) - - # Need an atomic transaction to SELECT the maximal ID so far then - # INSERT a new one - def _do_txn(txn): - sql = ( - "SELECT filter_id FROM user_filters " - "WHERE user_id = ? AND filter_json = ?" - ) - txn.execute(sql, (user_localpart, bytearray(def_json))) - filter_id_response = txn.fetchone() - if filter_id_response is not None: - return filter_id_response[0] - - sql = "SELECT MAX(filter_id) FROM user_filters " "WHERE user_id = ?" - txn.execute(sql, (user_localpart,)) - max_id = txn.fetchone()[0] - if max_id is None: - filter_id = 0 - else: - filter_id = max_id + 1 - - sql = ( - "INSERT INTO user_filters (user_id, filter_id, filter_json)" - "VALUES(?, ?, ?)" - ) - txn.execute(sql, (user_localpart, filter_id, bytearray(def_json))) - - return filter_id - - return self.runInteraction("add_user_filter", _do_txn) diff --git a/synapse/storage/group_server.py b/synapse/storage/group_server.py deleted file mode 100644 index 15b01c6958..0000000000 --- a/synapse/storage/group_server.py +++ /dev/null @@ -1,1181 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2017 Vector Creations Ltd -# Copyright 2018 New Vector Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -from canonicaljson import json - -from twisted.internet import defer - -from synapse.api.errors import SynapseError - -from ._base import SQLBaseStore - -# The category ID for the "default" category. We don't store as null in the -# database to avoid the fun of null != null -_DEFAULT_CATEGORY_ID = "" -_DEFAULT_ROLE_ID = "" - - -class GroupServerStore(SQLBaseStore): - def set_group_join_policy(self, group_id, join_policy): - """Set the join policy of a group. - - join_policy can be one of: - * "invite" - * "open" - """ - return self._simple_update_one( - table="groups", - keyvalues={"group_id": group_id}, - updatevalues={"join_policy": join_policy}, - desc="set_group_join_policy", - ) - - def get_group(self, group_id): - return self._simple_select_one( - table="groups", - keyvalues={"group_id": group_id}, - retcols=( - "name", - "short_description", - "long_description", - "avatar_url", - "is_public", - "join_policy", - ), - allow_none=True, - desc="get_group", - ) - - def get_users_in_group(self, group_id, include_private=False): - # TODO: Pagination - - keyvalues = {"group_id": group_id} - if not include_private: - keyvalues["is_public"] = True - - return self._simple_select_list( - table="group_users", - keyvalues=keyvalues, - retcols=("user_id", "is_public", "is_admin"), - desc="get_users_in_group", - ) - - def get_invited_users_in_group(self, group_id): - # TODO: Pagination - - return self._simple_select_onecol( - table="group_invites", - keyvalues={"group_id": group_id}, - retcol="user_id", - desc="get_invited_users_in_group", - ) - - def get_rooms_in_group(self, group_id, include_private=False): - # TODO: Pagination - - keyvalues = {"group_id": group_id} - if not include_private: - keyvalues["is_public"] = True - - return self._simple_select_list( - table="group_rooms", - keyvalues=keyvalues, - retcols=("room_id", "is_public"), - desc="get_rooms_in_group", - ) - - def get_rooms_for_summary_by_category(self, group_id, include_private=False): - """Get the rooms and categories that should be included in a summary request - - Returns ([rooms], [categories]) - """ - - def _get_rooms_for_summary_txn(txn): - keyvalues = {"group_id": group_id} - if not include_private: - keyvalues["is_public"] = True - - sql = """ - SELECT room_id, is_public, category_id, room_order - FROM group_summary_rooms - WHERE group_id = ? - """ - - if not include_private: - sql += " AND is_public = ?" - txn.execute(sql, (group_id, True)) - else: - txn.execute(sql, (group_id,)) - - rooms = [ - { - "room_id": row[0], - "is_public": row[1], - "category_id": row[2] if row[2] != _DEFAULT_CATEGORY_ID else None, - "order": row[3], - } - for row in txn - ] - - sql = """ - SELECT category_id, is_public, profile, cat_order - FROM group_summary_room_categories - INNER JOIN group_room_categories USING (group_id, category_id) - WHERE group_id = ? - """ - - if not include_private: - sql += " AND is_public = ?" - txn.execute(sql, (group_id, True)) - else: - txn.execute(sql, (group_id,)) - - categories = { - row[0]: { - "is_public": row[1], - "profile": json.loads(row[2]), - "order": row[3], - } - for row in txn - } - - return rooms, categories - - return self.runInteraction("get_rooms_for_summary", _get_rooms_for_summary_txn) - - def add_room_to_summary(self, group_id, room_id, category_id, order, is_public): - return self.runInteraction( - "add_room_to_summary", - self._add_room_to_summary_txn, - group_id, - room_id, - category_id, - order, - is_public, - ) - - def _add_room_to_summary_txn( - self, txn, group_id, room_id, category_id, order, is_public - ): - """Add (or update) room's entry in summary. - - Args: - group_id (str) - room_id (str) - category_id (str): If not None then adds the category to the end of - the summary if its not already there. [Optional] - order (int): If not None inserts the room at that position, e.g. - an order of 1 will put the room first. Otherwise, the room gets - added to the end. - """ - room_in_group = self._simple_select_one_onecol_txn( - txn, - table="group_rooms", - keyvalues={"group_id": group_id, "room_id": room_id}, - retcol="room_id", - allow_none=True, - ) - if not room_in_group: - raise SynapseError(400, "room not in group") - - if category_id is None: - category_id = _DEFAULT_CATEGORY_ID - else: - cat_exists = self._simple_select_one_onecol_txn( - txn, - table="group_room_categories", - keyvalues={"group_id": group_id, "category_id": category_id}, - retcol="group_id", - allow_none=True, - ) - if not cat_exists: - raise SynapseError(400, "Category doesn't exist") - - # TODO: Check category is part of summary already - cat_exists = self._simple_select_one_onecol_txn( - txn, - table="group_summary_room_categories", - keyvalues={"group_id": group_id, "category_id": category_id}, - retcol="group_id", - allow_none=True, - ) - if not cat_exists: - # If not, add it with an order larger than all others - txn.execute( - """ - INSERT INTO group_summary_room_categories - (group_id, category_id, cat_order) - SELECT ?, ?, COALESCE(MAX(cat_order), 0) + 1 - FROM group_summary_room_categories - WHERE group_id = ? AND category_id = ? - """, - (group_id, category_id, group_id, category_id), - ) - - existing = self._simple_select_one_txn( - txn, - table="group_summary_rooms", - keyvalues={ - "group_id": group_id, - "room_id": room_id, - "category_id": category_id, - }, - retcols=("room_order", "is_public"), - allow_none=True, - ) - - if order is not None: - # Shuffle other room orders that come after the given order - sql = """ - UPDATE group_summary_rooms SET room_order = room_order + 1 - WHERE group_id = ? AND category_id = ? AND room_order >= ? - """ - txn.execute(sql, (group_id, category_id, order)) - elif not existing: - sql = """ - SELECT COALESCE(MAX(room_order), 0) + 1 FROM group_summary_rooms - WHERE group_id = ? AND category_id = ? - """ - txn.execute(sql, (group_id, category_id)) - order, = txn.fetchone() - - if existing: - to_update = {} - if order is not None: - to_update["room_order"] = order - if is_public is not None: - to_update["is_public"] = is_public - self._simple_update_txn( - txn, - table="group_summary_rooms", - keyvalues={ - "group_id": group_id, - "category_id": category_id, - "room_id": room_id, - }, - values=to_update, - ) - else: - if is_public is None: - is_public = True - - self._simple_insert_txn( - txn, - table="group_summary_rooms", - values={ - "group_id": group_id, - "category_id": category_id, - "room_id": room_id, - "room_order": order, - "is_public": is_public, - }, - ) - - def remove_room_from_summary(self, group_id, room_id, category_id): - if category_id is None: - category_id = _DEFAULT_CATEGORY_ID - - return self._simple_delete( - table="group_summary_rooms", - keyvalues={ - "group_id": group_id, - "category_id": category_id, - "room_id": room_id, - }, - desc="remove_room_from_summary", - ) - - @defer.inlineCallbacks - def get_group_categories(self, group_id): - rows = yield self._simple_select_list( - table="group_room_categories", - keyvalues={"group_id": group_id}, - retcols=("category_id", "is_public", "profile"), - desc="get_group_categories", - ) - - return { - row["category_id"]: { - "is_public": row["is_public"], - "profile": json.loads(row["profile"]), - } - for row in rows - } - - @defer.inlineCallbacks - def get_group_category(self, group_id, category_id): - category = yield self._simple_select_one( - table="group_room_categories", - keyvalues={"group_id": group_id, "category_id": category_id}, - retcols=("is_public", "profile"), - desc="get_group_category", - ) - - category["profile"] = json.loads(category["profile"]) - - return category - - def upsert_group_category(self, group_id, category_id, profile, is_public): - """Add/update room category for group - """ - insertion_values = {} - update_values = {"category_id": category_id} # This cannot be empty - - if profile is None: - insertion_values["profile"] = "{}" - else: - update_values["profile"] = json.dumps(profile) - - if is_public is None: - insertion_values["is_public"] = True - else: - update_values["is_public"] = is_public - - return self._simple_upsert( - table="group_room_categories", - keyvalues={"group_id": group_id, "category_id": category_id}, - values=update_values, - insertion_values=insertion_values, - desc="upsert_group_category", - ) - - def remove_group_category(self, group_id, category_id): - return self._simple_delete( - table="group_room_categories", - keyvalues={"group_id": group_id, "category_id": category_id}, - desc="remove_group_category", - ) - - @defer.inlineCallbacks - def get_group_roles(self, group_id): - rows = yield self._simple_select_list( - table="group_roles", - keyvalues={"group_id": group_id}, - retcols=("role_id", "is_public", "profile"), - desc="get_group_roles", - ) - - return { - row["role_id"]: { - "is_public": row["is_public"], - "profile": json.loads(row["profile"]), - } - for row in rows - } - - @defer.inlineCallbacks - def get_group_role(self, group_id, role_id): - role = yield self._simple_select_one( - table="group_roles", - keyvalues={"group_id": group_id, "role_id": role_id}, - retcols=("is_public", "profile"), - desc="get_group_role", - ) - - role["profile"] = json.loads(role["profile"]) - - return role - - def upsert_group_role(self, group_id, role_id, profile, is_public): - """Add/remove user role - """ - insertion_values = {} - update_values = {"role_id": role_id} # This cannot be empty - - if profile is None: - insertion_values["profile"] = "{}" - else: - update_values["profile"] = json.dumps(profile) - - if is_public is None: - insertion_values["is_public"] = True - else: - update_values["is_public"] = is_public - - return self._simple_upsert( - table="group_roles", - keyvalues={"group_id": group_id, "role_id": role_id}, - values=update_values, - insertion_values=insertion_values, - desc="upsert_group_role", - ) - - def remove_group_role(self, group_id, role_id): - return self._simple_delete( - table="group_roles", - keyvalues={"group_id": group_id, "role_id": role_id}, - desc="remove_group_role", - ) - - def add_user_to_summary(self, group_id, user_id, role_id, order, is_public): - return self.runInteraction( - "add_user_to_summary", - self._add_user_to_summary_txn, - group_id, - user_id, - role_id, - order, - is_public, - ) - - def _add_user_to_summary_txn( - self, txn, group_id, user_id, role_id, order, is_public - ): - """Add (or update) user's entry in summary. - - Args: - group_id (str) - user_id (str) - role_id (str): If not None then adds the role to the end of - the summary if its not already there. [Optional] - order (int): If not None inserts the user at that position, e.g. - an order of 1 will put the user first. Otherwise, the user gets - added to the end. - """ - user_in_group = self._simple_select_one_onecol_txn( - txn, - table="group_users", - keyvalues={"group_id": group_id, "user_id": user_id}, - retcol="user_id", - allow_none=True, - ) - if not user_in_group: - raise SynapseError(400, "user not in group") - - if role_id is None: - role_id = _DEFAULT_ROLE_ID - else: - role_exists = self._simple_select_one_onecol_txn( - txn, - table="group_roles", - keyvalues={"group_id": group_id, "role_id": role_id}, - retcol="group_id", - allow_none=True, - ) - if not role_exists: - raise SynapseError(400, "Role doesn't exist") - - # TODO: Check role is part of the summary already - role_exists = self._simple_select_one_onecol_txn( - txn, - table="group_summary_roles", - keyvalues={"group_id": group_id, "role_id": role_id}, - retcol="group_id", - allow_none=True, - ) - if not role_exists: - # If not, add it with an order larger than all others - txn.execute( - """ - INSERT INTO group_summary_roles - (group_id, role_id, role_order) - SELECT ?, ?, COALESCE(MAX(role_order), 0) + 1 - FROM group_summary_roles - WHERE group_id = ? AND role_id = ? - """, - (group_id, role_id, group_id, role_id), - ) - - existing = self._simple_select_one_txn( - txn, - table="group_summary_users", - keyvalues={"group_id": group_id, "user_id": user_id, "role_id": role_id}, - retcols=("user_order", "is_public"), - allow_none=True, - ) - - if order is not None: - # Shuffle other users orders that come after the given order - sql = """ - UPDATE group_summary_users SET user_order = user_order + 1 - WHERE group_id = ? AND role_id = ? AND user_order >= ? - """ - txn.execute(sql, (group_id, role_id, order)) - elif not existing: - sql = """ - SELECT COALESCE(MAX(user_order), 0) + 1 FROM group_summary_users - WHERE group_id = ? AND role_id = ? - """ - txn.execute(sql, (group_id, role_id)) - order, = txn.fetchone() - - if existing: - to_update = {} - if order is not None: - to_update["user_order"] = order - if is_public is not None: - to_update["is_public"] = is_public - self._simple_update_txn( - txn, - table="group_summary_users", - keyvalues={ - "group_id": group_id, - "role_id": role_id, - "user_id": user_id, - }, - values=to_update, - ) - else: - if is_public is None: - is_public = True - - self._simple_insert_txn( - txn, - table="group_summary_users", - values={ - "group_id": group_id, - "role_id": role_id, - "user_id": user_id, - "user_order": order, - "is_public": is_public, - }, - ) - - def remove_user_from_summary(self, group_id, user_id, role_id): - if role_id is None: - role_id = _DEFAULT_ROLE_ID - - return self._simple_delete( - table="group_summary_users", - keyvalues={"group_id": group_id, "role_id": role_id, "user_id": user_id}, - desc="remove_user_from_summary", - ) - - def get_users_for_summary_by_role(self, group_id, include_private=False): - """Get the users and roles that should be included in a summary request - - Returns ([users], [roles]) - """ - - def _get_users_for_summary_txn(txn): - keyvalues = {"group_id": group_id} - if not include_private: - keyvalues["is_public"] = True - - sql = """ - SELECT user_id, is_public, role_id, user_order - FROM group_summary_users - WHERE group_id = ? - """ - - if not include_private: - sql += " AND is_public = ?" - txn.execute(sql, (group_id, True)) - else: - txn.execute(sql, (group_id,)) - - users = [ - { - "user_id": row[0], - "is_public": row[1], - "role_id": row[2] if row[2] != _DEFAULT_ROLE_ID else None, - "order": row[3], - } - for row in txn - ] - - sql = """ - SELECT role_id, is_public, profile, role_order - FROM group_summary_roles - INNER JOIN group_roles USING (group_id, role_id) - WHERE group_id = ? - """ - - if not include_private: - sql += " AND is_public = ?" - txn.execute(sql, (group_id, True)) - else: - txn.execute(sql, (group_id,)) - - roles = { - row[0]: { - "is_public": row[1], - "profile": json.loads(row[2]), - "order": row[3], - } - for row in txn - } - - return users, roles - - return self.runInteraction( - "get_users_for_summary_by_role", _get_users_for_summary_txn - ) - - def is_user_in_group(self, user_id, group_id): - return self._simple_select_one_onecol( - table="group_users", - keyvalues={"group_id": group_id, "user_id": user_id}, - retcol="user_id", - allow_none=True, - desc="is_user_in_group", - ).addCallback(lambda r: bool(r)) - - def is_user_admin_in_group(self, group_id, user_id): - return self._simple_select_one_onecol( - table="group_users", - keyvalues={"group_id": group_id, "user_id": user_id}, - retcol="is_admin", - allow_none=True, - desc="is_user_admin_in_group", - ) - - def add_group_invite(self, group_id, user_id): - """Record that the group server has invited a user - """ - return self._simple_insert( - table="group_invites", - values={"group_id": group_id, "user_id": user_id}, - desc="add_group_invite", - ) - - def is_user_invited_to_local_group(self, group_id, user_id): - """Has the group server invited a user? - """ - return self._simple_select_one_onecol( - table="group_invites", - keyvalues={"group_id": group_id, "user_id": user_id}, - retcol="user_id", - desc="is_user_invited_to_local_group", - allow_none=True, - ) - - def get_users_membership_info_in_group(self, group_id, user_id): - """Get a dict describing the membership of a user in a group. - - Example if joined: - - { - "membership": "join", - "is_public": True, - "is_privileged": False, - } - - Returns an empty dict if the user is not join/invite/etc - """ - - def _get_users_membership_in_group_txn(txn): - row = self._simple_select_one_txn( - txn, - table="group_users", - keyvalues={"group_id": group_id, "user_id": user_id}, - retcols=("is_admin", "is_public"), - allow_none=True, - ) - - if row: - return { - "membership": "join", - "is_public": row["is_public"], - "is_privileged": row["is_admin"], - } - - row = self._simple_select_one_onecol_txn( - txn, - table="group_invites", - keyvalues={"group_id": group_id, "user_id": user_id}, - retcol="user_id", - allow_none=True, - ) - - if row: - return {"membership": "invite"} - - return {} - - return self.runInteraction( - "get_users_membership_info_in_group", _get_users_membership_in_group_txn - ) - - def add_user_to_group( - self, - group_id, - user_id, - is_admin=False, - is_public=True, - local_attestation=None, - remote_attestation=None, - ): - """Add a user to the group server. - - Args: - group_id (str) - user_id (str) - is_admin (bool) - is_public (bool) - local_attestation (dict): The attestation the GS created to give - to the remote server. Optional if the user and group are on the - same server - remote_attestation (dict): The attestation given to GS by remote - server. Optional if the user and group are on the same server - """ - - def _add_user_to_group_txn(txn): - self._simple_insert_txn( - txn, - table="group_users", - values={ - "group_id": group_id, - "user_id": user_id, - "is_admin": is_admin, - "is_public": is_public, - }, - ) - - self._simple_delete_txn( - txn, - table="group_invites", - keyvalues={"group_id": group_id, "user_id": user_id}, - ) - - if local_attestation: - self._simple_insert_txn( - txn, - table="group_attestations_renewals", - values={ - "group_id": group_id, - "user_id": user_id, - "valid_until_ms": local_attestation["valid_until_ms"], - }, - ) - if remote_attestation: - self._simple_insert_txn( - txn, - table="group_attestations_remote", - values={ - "group_id": group_id, - "user_id": user_id, - "valid_until_ms": remote_attestation["valid_until_ms"], - "attestation_json": json.dumps(remote_attestation), - }, - ) - - return self.runInteraction("add_user_to_group", _add_user_to_group_txn) - - def remove_user_from_group(self, group_id, user_id): - def _remove_user_from_group_txn(txn): - self._simple_delete_txn( - txn, - table="group_users", - keyvalues={"group_id": group_id, "user_id": user_id}, - ) - self._simple_delete_txn( - txn, - table="group_invites", - keyvalues={"group_id": group_id, "user_id": user_id}, - ) - self._simple_delete_txn( - txn, - table="group_attestations_renewals", - keyvalues={"group_id": group_id, "user_id": user_id}, - ) - self._simple_delete_txn( - txn, - table="group_attestations_remote", - keyvalues={"group_id": group_id, "user_id": user_id}, - ) - self._simple_delete_txn( - txn, - table="group_summary_users", - keyvalues={"group_id": group_id, "user_id": user_id}, - ) - - return self.runInteraction( - "remove_user_from_group", _remove_user_from_group_txn - ) - - def add_room_to_group(self, group_id, room_id, is_public): - return self._simple_insert( - table="group_rooms", - values={"group_id": group_id, "room_id": room_id, "is_public": is_public}, - desc="add_room_to_group", - ) - - def update_room_in_group_visibility(self, group_id, room_id, is_public): - return self._simple_update( - table="group_rooms", - keyvalues={"group_id": group_id, "room_id": room_id}, - updatevalues={"is_public": is_public}, - desc="update_room_in_group_visibility", - ) - - def remove_room_from_group(self, group_id, room_id): - def _remove_room_from_group_txn(txn): - self._simple_delete_txn( - txn, - table="group_rooms", - keyvalues={"group_id": group_id, "room_id": room_id}, - ) - - self._simple_delete_txn( - txn, - table="group_summary_rooms", - keyvalues={"group_id": group_id, "room_id": room_id}, - ) - - return self.runInteraction( - "remove_room_from_group", _remove_room_from_group_txn - ) - - def get_publicised_groups_for_user(self, user_id): - """Get all groups a user is publicising - """ - return self._simple_select_onecol( - table="local_group_membership", - keyvalues={"user_id": user_id, "membership": "join", "is_publicised": True}, - retcol="group_id", - desc="get_publicised_groups_for_user", - ) - - def update_group_publicity(self, group_id, user_id, publicise): - """Update whether the user is publicising their membership of the group - """ - return self._simple_update_one( - table="local_group_membership", - keyvalues={"group_id": group_id, "user_id": user_id}, - updatevalues={"is_publicised": publicise}, - desc="update_group_publicity", - ) - - @defer.inlineCallbacks - def register_user_group_membership( - self, - group_id, - user_id, - membership, - is_admin=False, - content={}, - local_attestation=None, - remote_attestation=None, - is_publicised=False, - ): - """Registers that a local user is a member of a (local or remote) group. - - Args: - group_id (str) - user_id (str) - membership (str) - is_admin (bool) - content (dict): Content of the membership, e.g. includes the inviter - if the user has been invited. - local_attestation (dict): If remote group then store the fact that we - have given out an attestation, else None. - remote_attestation (dict): If remote group then store the remote - attestation from the group, else None. - """ - - def _register_user_group_membership_txn(txn, next_id): - # TODO: Upsert? - self._simple_delete_txn( - txn, - table="local_group_membership", - keyvalues={"group_id": group_id, "user_id": user_id}, - ) - self._simple_insert_txn( - txn, - table="local_group_membership", - values={ - "group_id": group_id, - "user_id": user_id, - "is_admin": is_admin, - "membership": membership, - "is_publicised": is_publicised, - "content": json.dumps(content), - }, - ) - - self._simple_insert_txn( - txn, - table="local_group_updates", - values={ - "stream_id": next_id, - "group_id": group_id, - "user_id": user_id, - "type": "membership", - "content": json.dumps( - {"membership": membership, "content": content} - ), - }, - ) - self._group_updates_stream_cache.entity_has_changed(user_id, next_id) - - # TODO: Insert profile to ensure it comes down stream if its a join. - - if membership == "join": - if local_attestation: - self._simple_insert_txn( - txn, - table="group_attestations_renewals", - values={ - "group_id": group_id, - "user_id": user_id, - "valid_until_ms": local_attestation["valid_until_ms"], - }, - ) - if remote_attestation: - self._simple_insert_txn( - txn, - table="group_attestations_remote", - values={ - "group_id": group_id, - "user_id": user_id, - "valid_until_ms": remote_attestation["valid_until_ms"], - "attestation_json": json.dumps(remote_attestation), - }, - ) - else: - self._simple_delete_txn( - txn, - table="group_attestations_renewals", - keyvalues={"group_id": group_id, "user_id": user_id}, - ) - self._simple_delete_txn( - txn, - table="group_attestations_remote", - keyvalues={"group_id": group_id, "user_id": user_id}, - ) - - return next_id - - with self._group_updates_id_gen.get_next() as next_id: - res = yield self.runInteraction( - "register_user_group_membership", - _register_user_group_membership_txn, - next_id, - ) - return res - - @defer.inlineCallbacks - def create_group( - self, group_id, user_id, name, avatar_url, short_description, long_description - ): - yield self._simple_insert( - table="groups", - values={ - "group_id": group_id, - "name": name, - "avatar_url": avatar_url, - "short_description": short_description, - "long_description": long_description, - "is_public": True, - }, - desc="create_group", - ) - - @defer.inlineCallbacks - def update_group_profile(self, group_id, profile): - yield self._simple_update_one( - table="groups", - keyvalues={"group_id": group_id}, - updatevalues=profile, - desc="update_group_profile", - ) - - def get_attestations_need_renewals(self, valid_until_ms): - """Get all attestations that need to be renewed until givent time - """ - - def _get_attestations_need_renewals_txn(txn): - sql = """ - SELECT group_id, user_id FROM group_attestations_renewals - WHERE valid_until_ms <= ? - """ - txn.execute(sql, (valid_until_ms,)) - return self.cursor_to_dict(txn) - - return self.runInteraction( - "get_attestations_need_renewals", _get_attestations_need_renewals_txn - ) - - def update_attestation_renewal(self, group_id, user_id, attestation): - """Update an attestation that we have renewed - """ - return self._simple_update_one( - table="group_attestations_renewals", - keyvalues={"group_id": group_id, "user_id": user_id}, - updatevalues={"valid_until_ms": attestation["valid_until_ms"]}, - desc="update_attestation_renewal", - ) - - def update_remote_attestion(self, group_id, user_id, attestation): - """Update an attestation that a remote has renewed - """ - return self._simple_update_one( - table="group_attestations_remote", - keyvalues={"group_id": group_id, "user_id": user_id}, - updatevalues={ - "valid_until_ms": attestation["valid_until_ms"], - "attestation_json": json.dumps(attestation), - }, - desc="update_remote_attestion", - ) - - def remove_attestation_renewal(self, group_id, user_id): - """Remove an attestation that we thought we should renew, but actually - shouldn't. Ideally this would never get called as we would never - incorrectly try and do attestations for local users on local groups. - - Args: - group_id (str) - user_id (str) - """ - return self._simple_delete( - table="group_attestations_renewals", - keyvalues={"group_id": group_id, "user_id": user_id}, - desc="remove_attestation_renewal", - ) - - @defer.inlineCallbacks - def get_remote_attestation(self, group_id, user_id): - """Get the attestation that proves the remote agrees that the user is - in the group. - """ - row = yield self._simple_select_one( - table="group_attestations_remote", - keyvalues={"group_id": group_id, "user_id": user_id}, - retcols=("valid_until_ms", "attestation_json"), - desc="get_remote_attestation", - allow_none=True, - ) - - now = int(self._clock.time_msec()) - if row and now < row["valid_until_ms"]: - return json.loads(row["attestation_json"]) - - return None - - def get_joined_groups(self, user_id): - return self._simple_select_onecol( - table="local_group_membership", - keyvalues={"user_id": user_id, "membership": "join"}, - retcol="group_id", - desc="get_joined_groups", - ) - - def get_all_groups_for_user(self, user_id, now_token): - def _get_all_groups_for_user_txn(txn): - sql = """ - SELECT group_id, type, membership, u.content - FROM local_group_updates AS u - INNER JOIN local_group_membership USING (group_id, user_id) - WHERE user_id = ? AND membership != 'leave' - AND stream_id <= ? - """ - txn.execute(sql, (user_id, now_token)) - return [ - { - "group_id": row[0], - "type": row[1], - "membership": row[2], - "content": json.loads(row[3]), - } - for row in txn - ] - - return self.runInteraction( - "get_all_groups_for_user", _get_all_groups_for_user_txn - ) - - def get_groups_changes_for_user(self, user_id, from_token, to_token): - from_token = int(from_token) - has_changed = self._group_updates_stream_cache.has_entity_changed( - user_id, from_token - ) - if not has_changed: - return [] - - def _get_groups_changes_for_user_txn(txn): - sql = """ - SELECT group_id, membership, type, u.content - FROM local_group_updates AS u - INNER JOIN local_group_membership USING (group_id, user_id) - WHERE user_id = ? AND ? < stream_id AND stream_id <= ? - """ - txn.execute(sql, (user_id, from_token, to_token)) - return [ - { - "group_id": group_id, - "membership": membership, - "type": gtype, - "content": json.loads(content_json), - } - for group_id, membership, gtype, content_json in txn - ] - - return self.runInteraction( - "get_groups_changes_for_user", _get_groups_changes_for_user_txn - ) - - def get_all_groups_changes(self, from_token, to_token, limit): - from_token = int(from_token) - has_changed = self._group_updates_stream_cache.has_any_entity_changed( - from_token - ) - if not has_changed: - return [] - - def _get_all_groups_changes_txn(txn): - sql = """ - SELECT stream_id, group_id, user_id, type, content - FROM local_group_updates - WHERE ? < stream_id AND stream_id <= ? - LIMIT ? - """ - txn.execute(sql, (from_token, to_token, limit)) - return [ - (stream_id, group_id, user_id, gtype, json.loads(content_json)) - for stream_id, group_id, user_id, gtype, content_json in txn - ] - - return self.runInteraction( - "get_all_groups_changes", _get_all_groups_changes_txn - ) - - def get_group_stream_token(self): - return self._group_updates_id_gen.get_current_token() - - def delete_group(self, group_id): - """Deletes a group fully from the database. - - Args: - group_id (str) - - Returns: - Deferred - """ - - def _delete_group_txn(txn): - tables = [ - "groups", - "group_users", - "group_invites", - "group_rooms", - "group_summary_rooms", - "group_summary_room_categories", - "group_room_categories", - "group_summary_users", - "group_summary_roles", - "group_roles", - "group_attestations_renewals", - "group_attestations_remote", - ] - - for table in tables: - self._simple_delete_txn( - txn, table=table, keyvalues={"group_id": group_id} - ) - - return self.runInteraction("delete_group", _delete_group_txn) diff --git a/synapse/storage/keys.py b/synapse/storage/keys.py index e72f89e446..4769b21529 100644 --- a/synapse/storage/keys.py +++ b/synapse/storage/keys.py @@ -14,208 +14,14 @@ # See the License for the specific language governing permissions and # limitations under the License. -import itertools import logging -import six - import attr -from signedjson.key import decode_verify_key_bytes - -from synapse.util import batch_iter -from synapse.util.caches.descriptors import cached, cachedList - -from ._base import SQLBaseStore logger = logging.getLogger(__name__) -# py2 sqlite has buffer hardcoded as only binary type, so we must use it, -# despite being deprecated and removed in favor of memoryview -if six.PY2: - db_binary_type = six.moves.builtins.buffer -else: - db_binary_type = memoryview - @attr.s(slots=True, frozen=True) class FetchKeyResult(object): verify_key = attr.ib() # VerifyKey: the key itself valid_until_ts = attr.ib() # int: how long we can use this key for - - -class KeyStore(SQLBaseStore): - """Persistence for signature verification keys - """ - - @cached() - def _get_server_verify_key(self, server_name_and_key_id): - raise NotImplementedError() - - @cachedList( - cached_method_name="_get_server_verify_key", list_name="server_name_and_key_ids" - ) - def get_server_verify_keys(self, server_name_and_key_ids): - """ - Args: - server_name_and_key_ids (iterable[Tuple[str, str]]): - iterable of (server_name, key-id) tuples to fetch keys for - - Returns: - Deferred: resolves to dict[Tuple[str, str], FetchKeyResult|None]: - map from (server_name, key_id) -> FetchKeyResult, or None if the key is - unknown - """ - keys = {} - - def _get_keys(txn, batch): - """Processes a batch of keys to fetch, and adds the result to `keys`.""" - - # batch_iter always returns tuples so it's safe to do len(batch) - sql = ( - "SELECT server_name, key_id, verify_key, ts_valid_until_ms " - "FROM server_signature_keys WHERE 1=0" - ) + " OR (server_name=? AND key_id=?)" * len(batch) - - txn.execute(sql, tuple(itertools.chain.from_iterable(batch))) - - for row in txn: - server_name, key_id, key_bytes, ts_valid_until_ms = row - - if ts_valid_until_ms is None: - # Old keys may be stored with a ts_valid_until_ms of null, - # in which case we treat this as if it was set to `0`, i.e. - # it won't match key requests that define a minimum - # `ts_valid_until_ms`. - ts_valid_until_ms = 0 - - res = FetchKeyResult( - verify_key=decode_verify_key_bytes(key_id, bytes(key_bytes)), - valid_until_ts=ts_valid_until_ms, - ) - keys[(server_name, key_id)] = res - - def _txn(txn): - for batch in batch_iter(server_name_and_key_ids, 50): - _get_keys(txn, batch) - return keys - - return self.runInteraction("get_server_verify_keys", _txn) - - def store_server_verify_keys(self, from_server, ts_added_ms, verify_keys): - """Stores NACL verification keys for remote servers. - Args: - from_server (str): Where the verification keys were looked up - ts_added_ms (int): The time to record that the key was added - verify_keys (iterable[tuple[str, str, FetchKeyResult]]): - keys to be stored. Each entry is a triplet of - (server_name, key_id, key). - """ - key_values = [] - value_values = [] - invalidations = [] - for server_name, key_id, fetch_result in verify_keys: - key_values.append((server_name, key_id)) - value_values.append( - ( - from_server, - ts_added_ms, - fetch_result.valid_until_ts, - db_binary_type(fetch_result.verify_key.encode()), - ) - ) - # invalidate takes a tuple corresponding to the params of - # _get_server_verify_key. _get_server_verify_key only takes one - # param, which is itself the 2-tuple (server_name, key_id). - invalidations.append((server_name, key_id)) - - def _invalidate(res): - f = self._get_server_verify_key.invalidate - for i in invalidations: - f((i,)) - return res - - return self.runInteraction( - "store_server_verify_keys", - self._simple_upsert_many_txn, - table="server_signature_keys", - key_names=("server_name", "key_id"), - key_values=key_values, - value_names=( - "from_server", - "ts_added_ms", - "ts_valid_until_ms", - "verify_key", - ), - value_values=value_values, - ).addCallback(_invalidate) - - def store_server_keys_json( - self, server_name, key_id, from_server, ts_now_ms, ts_expires_ms, key_json_bytes - ): - """Stores the JSON bytes for a set of keys from a server - The JSON should be signed by the originating server, the intermediate - server, and by this server. Updates the value for the - (server_name, key_id, from_server) triplet if one already existed. - Args: - server_name (str): The name of the server. - key_id (str): The identifer of the key this JSON is for. - from_server (str): The server this JSON was fetched from. - ts_now_ms (int): The time now in milliseconds. - ts_valid_until_ms (int): The time when this json stops being valid. - key_json (bytes): The encoded JSON. - """ - return self._simple_upsert( - table="server_keys_json", - keyvalues={ - "server_name": server_name, - "key_id": key_id, - "from_server": from_server, - }, - values={ - "server_name": server_name, - "key_id": key_id, - "from_server": from_server, - "ts_added_ms": ts_now_ms, - "ts_valid_until_ms": ts_expires_ms, - "key_json": db_binary_type(key_json_bytes), - }, - desc="store_server_keys_json", - ) - - def get_server_keys_json(self, server_keys): - """Retrive the key json for a list of server_keys and key ids. - If no keys are found for a given server, key_id and source then - that server, key_id, and source triplet entry will be an empty list. - The JSON is returned as a byte array so that it can be efficiently - used in an HTTP response. - Args: - server_keys (list): List of (server_name, key_id, source) triplets. - Returns: - Deferred[dict[Tuple[str, str, str|None], list[dict]]]: - Dict mapping (server_name, key_id, source) triplets to lists of dicts - """ - - def _get_server_keys_json_txn(txn): - results = {} - for server_name, key_id, from_server in server_keys: - keyvalues = {"server_name": server_name} - if key_id is not None: - keyvalues["key_id"] = key_id - if from_server is not None: - keyvalues["from_server"] = from_server - rows = self._simple_select_list_txn( - txn, - "server_keys_json", - keyvalues=keyvalues, - retcols=( - "key_id", - "from_server", - "ts_added_ms", - "ts_valid_until_ms", - "key_json", - ), - ) - results[(server_name, key_id, from_server)] = rows - return results - - return self.runInteraction("get_server_keys_json", _get_server_keys_json_txn) diff --git a/synapse/storage/media_repository.py b/synapse/storage/media_repository.py deleted file mode 100644 index 84b5f3ad5e..0000000000 --- a/synapse/storage/media_repository.py +++ /dev/null @@ -1,378 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2014-2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. -from synapse.storage.background_updates import BackgroundUpdateStore - - -class MediaRepositoryBackgroundUpdateStore(BackgroundUpdateStore): - def __init__(self, db_conn, hs): - super(MediaRepositoryBackgroundUpdateStore, self).__init__(db_conn, hs) - - self.register_background_index_update( - update_name="local_media_repository_url_idx", - index_name="local_media_repository_url_idx", - table="local_media_repository", - columns=["created_ts"], - where_clause="url_cache IS NOT NULL", - ) - - -class MediaRepositoryStore(MediaRepositoryBackgroundUpdateStore): - """Persistence for attachments and avatars""" - - def __init__(self, db_conn, hs): - super(MediaRepositoryStore, self).__init__(db_conn, hs) - - def get_local_media(self, media_id): - """Get the metadata for a local piece of media - Returns: - None if the media_id doesn't exist. - """ - return self._simple_select_one( - "local_media_repository", - {"media_id": media_id}, - ( - "media_type", - "media_length", - "upload_name", - "created_ts", - "quarantined_by", - "url_cache", - ), - allow_none=True, - desc="get_local_media", - ) - - def store_local_media( - self, - media_id, - media_type, - time_now_ms, - upload_name, - media_length, - user_id, - url_cache=None, - ): - return self._simple_insert( - "local_media_repository", - { - "media_id": media_id, - "media_type": media_type, - "created_ts": time_now_ms, - "upload_name": upload_name, - "media_length": media_length, - "user_id": user_id.to_string(), - "url_cache": url_cache, - }, - desc="store_local_media", - ) - - def get_url_cache(self, url, ts): - """Get the media_id and ts for a cached URL as of the given timestamp - Returns: - None if the URL isn't cached. - """ - - def get_url_cache_txn(txn): - # get the most recently cached result (relative to the given ts) - sql = ( - "SELECT response_code, etag, expires_ts, og, media_id, download_ts" - " FROM local_media_repository_url_cache" - " WHERE url = ? AND download_ts <= ?" - " ORDER BY download_ts DESC LIMIT 1" - ) - txn.execute(sql, (url, ts)) - row = txn.fetchone() - - if not row: - # ...or if we've requested a timestamp older than the oldest - # copy in the cache, return the oldest copy (if any) - sql = ( - "SELECT response_code, etag, expires_ts, og, media_id, download_ts" - " FROM local_media_repository_url_cache" - " WHERE url = ? AND download_ts > ?" - " ORDER BY download_ts ASC LIMIT 1" - ) - txn.execute(sql, (url, ts)) - row = txn.fetchone() - - if not row: - return None - - return dict( - zip( - ( - "response_code", - "etag", - "expires_ts", - "og", - "media_id", - "download_ts", - ), - row, - ) - ) - - return self.runInteraction("get_url_cache", get_url_cache_txn) - - def store_url_cache( - self, url, response_code, etag, expires_ts, og, media_id, download_ts - ): - return self._simple_insert( - "local_media_repository_url_cache", - { - "url": url, - "response_code": response_code, - "etag": etag, - "expires_ts": expires_ts, - "og": og, - "media_id": media_id, - "download_ts": download_ts, - }, - desc="store_url_cache", - ) - - def get_local_media_thumbnails(self, media_id): - return self._simple_select_list( - "local_media_repository_thumbnails", - {"media_id": media_id}, - ( - "thumbnail_width", - "thumbnail_height", - "thumbnail_method", - "thumbnail_type", - "thumbnail_length", - ), - desc="get_local_media_thumbnails", - ) - - def store_local_thumbnail( - self, - media_id, - thumbnail_width, - thumbnail_height, - thumbnail_type, - thumbnail_method, - thumbnail_length, - ): - return self._simple_insert( - "local_media_repository_thumbnails", - { - "media_id": media_id, - "thumbnail_width": thumbnail_width, - "thumbnail_height": thumbnail_height, - "thumbnail_method": thumbnail_method, - "thumbnail_type": thumbnail_type, - "thumbnail_length": thumbnail_length, - }, - desc="store_local_thumbnail", - ) - - def get_cached_remote_media(self, origin, media_id): - return self._simple_select_one( - "remote_media_cache", - {"media_origin": origin, "media_id": media_id}, - ( - "media_type", - "media_length", - "upload_name", - "created_ts", - "filesystem_id", - "quarantined_by", - ), - allow_none=True, - desc="get_cached_remote_media", - ) - - def store_cached_remote_media( - self, - origin, - media_id, - media_type, - media_length, - time_now_ms, - upload_name, - filesystem_id, - ): - return self._simple_insert( - "remote_media_cache", - { - "media_origin": origin, - "media_id": media_id, - "media_type": media_type, - "media_length": media_length, - "created_ts": time_now_ms, - "upload_name": upload_name, - "filesystem_id": filesystem_id, - "last_access_ts": time_now_ms, - }, - desc="store_cached_remote_media", - ) - - def update_cached_last_access_time(self, local_media, remote_media, time_ms): - """Updates the last access time of the given media - - Args: - local_media (iterable[str]): Set of media_ids - remote_media (iterable[(str, str)]): Set of (server_name, media_id) - time_ms: Current time in milliseconds - """ - - def update_cache_txn(txn): - sql = ( - "UPDATE remote_media_cache SET last_access_ts = ?" - " WHERE media_origin = ? AND media_id = ?" - ) - - txn.executemany( - sql, - ( - (time_ms, media_origin, media_id) - for media_origin, media_id in remote_media - ), - ) - - sql = ( - "UPDATE local_media_repository SET last_access_ts = ?" - " WHERE media_id = ?" - ) - - txn.executemany(sql, ((time_ms, media_id) for media_id in local_media)) - - return self.runInteraction("update_cached_last_access_time", update_cache_txn) - - def get_remote_media_thumbnails(self, origin, media_id): - return self._simple_select_list( - "remote_media_cache_thumbnails", - {"media_origin": origin, "media_id": media_id}, - ( - "thumbnail_width", - "thumbnail_height", - "thumbnail_method", - "thumbnail_type", - "thumbnail_length", - "filesystem_id", - ), - desc="get_remote_media_thumbnails", - ) - - def store_remote_media_thumbnail( - self, - origin, - media_id, - filesystem_id, - thumbnail_width, - thumbnail_height, - thumbnail_type, - thumbnail_method, - thumbnail_length, - ): - return self._simple_insert( - "remote_media_cache_thumbnails", - { - "media_origin": origin, - "media_id": media_id, - "thumbnail_width": thumbnail_width, - "thumbnail_height": thumbnail_height, - "thumbnail_method": thumbnail_method, - "thumbnail_type": thumbnail_type, - "thumbnail_length": thumbnail_length, - "filesystem_id": filesystem_id, - }, - desc="store_remote_media_thumbnail", - ) - - def get_remote_media_before(self, before_ts): - sql = ( - "SELECT media_origin, media_id, filesystem_id" - " FROM remote_media_cache" - " WHERE last_access_ts < ?" - ) - - return self._execute( - "get_remote_media_before", self.cursor_to_dict, sql, before_ts - ) - - def delete_remote_media(self, media_origin, media_id): - def delete_remote_media_txn(txn): - self._simple_delete_txn( - txn, - "remote_media_cache", - keyvalues={"media_origin": media_origin, "media_id": media_id}, - ) - self._simple_delete_txn( - txn, - "remote_media_cache_thumbnails", - keyvalues={"media_origin": media_origin, "media_id": media_id}, - ) - - return self.runInteraction("delete_remote_media", delete_remote_media_txn) - - def get_expired_url_cache(self, now_ts): - sql = ( - "SELECT media_id FROM local_media_repository_url_cache" - " WHERE expires_ts < ?" - " ORDER BY expires_ts ASC" - " LIMIT 500" - ) - - def _get_expired_url_cache_txn(txn): - txn.execute(sql, (now_ts,)) - return [row[0] for row in txn] - - return self.runInteraction("get_expired_url_cache", _get_expired_url_cache_txn) - - def delete_url_cache(self, media_ids): - if len(media_ids) == 0: - return - - sql = "DELETE FROM local_media_repository_url_cache" " WHERE media_id = ?" - - def _delete_url_cache_txn(txn): - txn.executemany(sql, [(media_id,) for media_id in media_ids]) - - return self.runInteraction("delete_url_cache", _delete_url_cache_txn) - - def get_url_cache_media_before(self, before_ts): - sql = ( - "SELECT media_id FROM local_media_repository" - " WHERE created_ts < ? AND url_cache IS NOT NULL" - " ORDER BY created_ts ASC" - " LIMIT 500" - ) - - def _get_url_cache_media_before_txn(txn): - txn.execute(sql, (before_ts,)) - return [row[0] for row in txn] - - return self.runInteraction( - "get_url_cache_media_before", _get_url_cache_media_before_txn - ) - - def delete_url_cache_media(self, media_ids): - if len(media_ids) == 0: - return - - def _delete_url_cache_media_txn(txn): - sql = "DELETE FROM local_media_repository" " WHERE media_id = ?" - - txn.executemany(sql, [(media_id,) for media_id in media_ids]) - - sql = "DELETE FROM local_media_repository_thumbnails" " WHERE media_id = ?" - - txn.executemany(sql, [(media_id,) for media_id in media_ids]) - - return self.runInteraction( - "delete_url_cache_media", _delete_url_cache_media_txn - ) diff --git a/synapse/storage/monthly_active_users.py b/synapse/storage/monthly_active_users.py deleted file mode 100644 index 3803604be7..0000000000 --- a/synapse/storage/monthly_active_users.py +++ /dev/null @@ -1,329 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2018 New Vector -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. -import logging - -from twisted.internet import defer - -from synapse.util.caches.descriptors import cached - -from ._base import SQLBaseStore - -logger = logging.getLogger(__name__) - -# Number of msec of granularity to store the monthly_active_user timestamp -# This means it is not necessary to update the table on every request -LAST_SEEN_GRANULARITY = 60 * 60 * 1000 - - -class MonthlyActiveUsersStore(SQLBaseStore): - def __init__(self, dbconn, hs): - super(MonthlyActiveUsersStore, self).__init__(None, hs) - self._clock = hs.get_clock() - self.hs = hs - # Do not add more reserved users than the total allowable number - self._new_transaction( - dbconn, - "initialise_mau_threepids", - [], - [], - self._initialise_reserved_users, - hs.config.mau_limits_reserved_threepids[: self.hs.config.max_mau_value], - ) - - def _initialise_reserved_users(self, txn, threepids): - """Ensures that reserved threepids are accounted for in the MAU table, should - be called on start up. - - Args: - txn (cursor): - threepids (list[dict]): List of threepid dicts to reserve - """ - - for tp in threepids: - user_id = self.get_user_id_by_threepid_txn(txn, tp["medium"], tp["address"]) - - if user_id: - is_support = self.is_support_user_txn(txn, user_id) - if not is_support: - self.upsert_monthly_active_user_txn(txn, user_id) - else: - logger.warning("mau limit reserved threepid %s not found in db" % tp) - - @defer.inlineCallbacks - def reap_monthly_active_users(self): - """Cleans out monthly active user table to ensure that no stale - entries exist. - - Returns: - Deferred[] - """ - - def _reap_users(txn, reserved_users): - """ - Args: - reserved_users (tuple): reserved users to preserve - """ - - thirty_days_ago = int(self._clock.time_msec()) - (1000 * 60 * 60 * 24 * 30) - query_args = [thirty_days_ago] - base_sql = "DELETE FROM monthly_active_users WHERE timestamp < ?" - - # Need if/else since 'AND user_id NOT IN ({})' fails on Postgres - # when len(reserved_users) == 0. Works fine on sqlite. - if len(reserved_users) > 0: - # questionmarks is a hack to overcome sqlite not supporting - # tuples in 'WHERE IN %s' - question_marks = ",".join("?" * len(reserved_users)) - - query_args.extend(reserved_users) - sql = base_sql + " AND user_id NOT IN ({})".format(question_marks) - else: - sql = base_sql - - txn.execute(sql, query_args) - - max_mau_value = self.hs.config.max_mau_value - if self.hs.config.limit_usage_by_mau: - # If MAU user count still exceeds the MAU threshold, then delete on - # a least recently active basis. - # Note it is not possible to write this query using OFFSET due to - # incompatibilities in how sqlite and postgres support the feature. - # sqlite requires 'LIMIT -1 OFFSET ?', the LIMIT must be present - # While Postgres does not require 'LIMIT', but also does not support - # negative LIMIT values. So there is no way to write it that both can - # support - if len(reserved_users) == 0: - sql = """ - DELETE FROM monthly_active_users - WHERE user_id NOT IN ( - SELECT user_id FROM monthly_active_users - ORDER BY timestamp DESC - LIMIT ? - ) - """ - txn.execute(sql, (max_mau_value,)) - # Need if/else since 'AND user_id NOT IN ({})' fails on Postgres - # when len(reserved_users) == 0. Works fine on sqlite. - else: - # Must be >= 0 for postgres - num_of_non_reserved_users_to_remove = max( - max_mau_value - len(reserved_users), 0 - ) - - # It is important to filter reserved users twice to guard - # against the case where the reserved user is present in the - # SELECT, meaning that a legitmate mau is deleted. - sql = """ - DELETE FROM monthly_active_users - WHERE user_id NOT IN ( - SELECT user_id FROM monthly_active_users - WHERE user_id NOT IN ({}) - ORDER BY timestamp DESC - LIMIT ? - ) - AND user_id NOT IN ({}) - """.format( - question_marks, question_marks - ) - - query_args = [ - *reserved_users, - num_of_non_reserved_users_to_remove, - *reserved_users, - ] - - txn.execute(sql, query_args) - - reserved_users = yield self.get_registered_reserved_users() - yield self.runInteraction( - "reap_monthly_active_users", _reap_users, reserved_users - ) - # It seems poor to invalidate the whole cache, Postgres supports - # 'Returning' which would allow me to invalidate only the - # specific users, but sqlite has no way to do this and instead - # I would need to SELECT and the DELETE which without locking - # is racy. - # Have resolved to invalidate the whole cache for now and do - # something about it if and when the perf becomes significant - self.user_last_seen_monthly_active.invalidate_all() - self.get_monthly_active_count.invalidate_all() - - @cached(num_args=0) - def get_monthly_active_count(self): - """Generates current count of monthly active users - - Returns: - Defered[int]: Number of current monthly active users - """ - - def _count_users(txn): - sql = "SELECT COALESCE(count(*), 0) FROM monthly_active_users" - - txn.execute(sql) - count, = txn.fetchone() - return count - - return self.runInteraction("count_users", _count_users) - - @defer.inlineCallbacks - def get_registered_reserved_users(self): - """Of the reserved threepids defined in config, which are associated - with registered users? - - Returns: - Defered[list]: Real reserved users - """ - users = [] - - for tp in self.hs.config.mau_limits_reserved_threepids[ - : self.hs.config.max_mau_value - ]: - user_id = yield self.hs.get_datastore().get_user_id_by_threepid( - tp["medium"], tp["address"] - ) - if user_id: - users.append(user_id) - - return users - - @defer.inlineCallbacks - def upsert_monthly_active_user(self, user_id): - """Updates or inserts the user into the monthly active user table, which - is used to track the current MAU usage of the server - - Args: - user_id (str): user to add/update - """ - # Support user never to be included in MAU stats. Note I can't easily call this - # from upsert_monthly_active_user_txn because then I need a _txn form of - # is_support_user which is complicated because I want to cache the result. - # Therefore I call it here and ignore the case where - # upsert_monthly_active_user_txn is called directly from - # _initialise_reserved_users reasoning that it would be very strange to - # include a support user in this context. - - is_support = yield self.is_support_user(user_id) - if is_support: - return - - yield self.runInteraction( - "upsert_monthly_active_user", self.upsert_monthly_active_user_txn, user_id - ) - - user_in_mau = self.user_last_seen_monthly_active.cache.get( - (user_id,), None, update_metrics=False - ) - if user_in_mau is None: - self.get_monthly_active_count.invalidate(()) - - self.user_last_seen_monthly_active.invalidate((user_id,)) - - def upsert_monthly_active_user_txn(self, txn, user_id): - """Updates or inserts monthly active user member - - Note that, after calling this method, it will generally be necessary - to invalidate the caches on user_last_seen_monthly_active and - get_monthly_active_count. We can't do that here, because we are running - in a database thread rather than the main thread, and we can't call - txn.call_after because txn may not be a LoggingTransaction. - - We consciously do not call is_support_txn from this method because it - is not possible to cache the response. is_support_txn will be false in - almost all cases, so it seems reasonable to call it only for - upsert_monthly_active_user and to call is_support_txn manually - for cases where upsert_monthly_active_user_txn is called directly, - like _initialise_reserved_users - - In short, don't call this method with support users. (Support users - should not appear in the MAU stats). - - Args: - txn (cursor): - user_id (str): user to add/update - - Returns: - bool: True if a new entry was created, False if an - existing one was updated. - """ - - # Am consciously deciding to lock the table on the basis that is ought - # never be a big table and alternative approaches (batching multiple - # upserts into a single txn) introduced a lot of extra complexity. - # See https://github.com/matrix-org/synapse/issues/3854 for more - is_insert = self._simple_upsert_txn( - txn, - table="monthly_active_users", - keyvalues={"user_id": user_id}, - values={"timestamp": int(self._clock.time_msec())}, - ) - - return is_insert - - @cached(num_args=1) - def user_last_seen_monthly_active(self, user_id): - """ - Checks if a given user is part of the monthly active user group - Arguments: - user_id (str): user to add/update - Return: - Deferred[int] : timestamp since last seen, None if never seen - - """ - - return self._simple_select_one_onecol( - table="monthly_active_users", - keyvalues={"user_id": user_id}, - retcol="timestamp", - allow_none=True, - desc="user_last_seen_monthly_active", - ) - - @defer.inlineCallbacks - def populate_monthly_active_users(self, user_id): - """Checks on the state of monthly active user limits and optionally - add the user to the monthly active tables - - Args: - user_id(str): the user_id to query - """ - if self.hs.config.limit_usage_by_mau or self.hs.config.mau_stats_only: - # Trial users and guests should not be included as part of MAU group - is_guest = yield self.is_guest(user_id) - if is_guest: - return - is_trial = yield self.is_trial_user(user_id) - if is_trial: - return - - last_seen_timestamp = yield self.user_last_seen_monthly_active(user_id) - now = self.hs.get_clock().time_msec() - - # We want to reduce to the total number of db writes, and are happy - # to trade accuracy of timestamp in order to lighten load. This means - # We always insert new users (where MAU threshold has not been reached), - # but only update if we have not previously seen the user for - # LAST_SEEN_GRANULARITY ms - if last_seen_timestamp is None: - # In the case where mau_stats_only is True and limit_usage_by_mau is - # False, there is no point in checking get_monthly_active_count - it - # adds no value and will break the logic if max_mau_value is exceeded. - if not self.hs.config.limit_usage_by_mau: - yield self.upsert_monthly_active_user(user_id) - else: - count = yield self.get_monthly_active_count() - if count < self.hs.config.max_mau_value: - yield self.upsert_monthly_active_user(user_id) - elif now - last_seen_timestamp > LAST_SEEN_GRANULARITY: - yield self.upsert_monthly_active_user(user_id) diff --git a/synapse/storage/openid.py b/synapse/storage/openid.py deleted file mode 100644 index b3318045ee..0000000000 --- a/synapse/storage/openid.py +++ /dev/null @@ -1,31 +0,0 @@ -from ._base import SQLBaseStore - - -class OpenIdStore(SQLBaseStore): - def insert_open_id_token(self, token, ts_valid_until_ms, user_id): - return self._simple_insert( - table="open_id_tokens", - values={ - "token": token, - "ts_valid_until_ms": ts_valid_until_ms, - "user_id": user_id, - }, - desc="insert_open_id_token", - ) - - def get_user_id_for_open_id_token(self, token, ts_now_ms): - def get_user_id_for_token_txn(txn): - sql = ( - "SELECT user_id FROM open_id_tokens" - " WHERE token = ? AND ? <= ts_valid_until_ms" - ) - - txn.execute(sql, (token, ts_now_ms)) - - rows = txn.fetchall() - if not rows: - return None - else: - return rows[0][0] - - return self.runInteraction("get_user_id_for_token", get_user_id_for_token_txn) diff --git a/synapse/storage/presence.py b/synapse/storage/presence.py index 3a641f538b..18a462f0ee 100644 --- a/synapse/storage/presence.py +++ b/synapse/storage/presence.py @@ -15,12 +15,7 @@ from collections import namedtuple -from twisted.internet import defer - from synapse.api.constants import PresenceState -from synapse.storage._base import SQLBaseStore, make_in_list_sql_clause -from synapse.util import batch_iter -from synapse.util.caches.descriptors import cached, cachedList class UserPresenceState( @@ -72,132 +67,3 @@ class UserPresenceState( status_msg=None, currently_active=False, ) - - -class PresenceStore(SQLBaseStore): - @defer.inlineCallbacks - def update_presence(self, presence_states): - stream_ordering_manager = self._presence_id_gen.get_next_mult( - len(presence_states) - ) - - with stream_ordering_manager as stream_orderings: - yield self.runInteraction( - "update_presence", - self._update_presence_txn, - stream_orderings, - presence_states, - ) - - return stream_orderings[-1], self._presence_id_gen.get_current_token() - - def _update_presence_txn(self, txn, stream_orderings, presence_states): - for stream_id, state in zip(stream_orderings, presence_states): - txn.call_after( - self.presence_stream_cache.entity_has_changed, state.user_id, stream_id - ) - txn.call_after(self._get_presence_for_user.invalidate, (state.user_id,)) - - # Actually insert new rows - self._simple_insert_many_txn( - txn, - table="presence_stream", - values=[ - { - "stream_id": stream_id, - "user_id": state.user_id, - "state": state.state, - "last_active_ts": state.last_active_ts, - "last_federation_update_ts": state.last_federation_update_ts, - "last_user_sync_ts": state.last_user_sync_ts, - "status_msg": state.status_msg, - "currently_active": state.currently_active, - } - for state in presence_states - ], - ) - - # Delete old rows to stop database from getting really big - sql = "DELETE FROM presence_stream WHERE stream_id < ? AND " - - for states in batch_iter(presence_states, 50): - clause, args = make_in_list_sql_clause( - self.database_engine, "user_id", [s.user_id for s in states] - ) - txn.execute(sql + clause, [stream_id] + list(args)) - - def get_all_presence_updates(self, last_id, current_id): - if last_id == current_id: - return defer.succeed([]) - - def get_all_presence_updates_txn(txn): - sql = ( - "SELECT stream_id, user_id, state, last_active_ts," - " last_federation_update_ts, last_user_sync_ts, status_msg," - " currently_active" - " FROM presence_stream" - " WHERE ? < stream_id AND stream_id <= ?" - ) - txn.execute(sql, (last_id, current_id)) - return txn.fetchall() - - return self.runInteraction( - "get_all_presence_updates", get_all_presence_updates_txn - ) - - @cached() - def _get_presence_for_user(self, user_id): - raise NotImplementedError() - - @cachedList( - cached_method_name="_get_presence_for_user", - list_name="user_ids", - num_args=1, - inlineCallbacks=True, - ) - def get_presence_for_users(self, user_ids): - rows = yield self._simple_select_many_batch( - table="presence_stream", - column="user_id", - iterable=user_ids, - keyvalues={}, - retcols=( - "user_id", - "state", - "last_active_ts", - "last_federation_update_ts", - "last_user_sync_ts", - "status_msg", - "currently_active", - ), - desc="get_presence_for_users", - ) - - for row in rows: - row["currently_active"] = bool(row["currently_active"]) - - return {row["user_id"]: UserPresenceState(**row) for row in rows} - - def get_current_presence_token(self): - return self._presence_id_gen.get_current_token() - - def allow_presence_visible(self, observed_localpart, observer_userid): - return self._simple_insert( - table="presence_allow_inbound", - values={ - "observed_user_id": observed_localpart, - "observer_user_id": observer_userid, - }, - desc="allow_presence_visible", - or_ignore=True, - ) - - def disallow_presence_visible(self, observed_localpart, observer_userid): - return self._simple_delete_one( - table="presence_allow_inbound", - keyvalues={ - "observed_user_id": observed_localpart, - "observer_user_id": observer_userid, - }, - desc="disallow_presence_visible", - ) diff --git a/synapse/storage/profile.py b/synapse/storage/profile.py deleted file mode 100644 index 912c1df6be..0000000000 --- a/synapse/storage/profile.py +++ /dev/null @@ -1,179 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2014-2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -from twisted.internet import defer - -from synapse.api.errors import StoreError -from synapse.storage.roommember import ProfileInfo - -from ._base import SQLBaseStore - - -class ProfileWorkerStore(SQLBaseStore): - @defer.inlineCallbacks - def get_profileinfo(self, user_localpart): - try: - profile = yield self._simple_select_one( - table="profiles", - keyvalues={"user_id": user_localpart}, - retcols=("displayname", "avatar_url"), - desc="get_profileinfo", - ) - except StoreError as e: - if e.code == 404: - # no match - return ProfileInfo(None, None) - else: - raise - - return ProfileInfo( - avatar_url=profile["avatar_url"], display_name=profile["displayname"] - ) - - def get_profile_displayname(self, user_localpart): - return self._simple_select_one_onecol( - table="profiles", - keyvalues={"user_id": user_localpart}, - retcol="displayname", - desc="get_profile_displayname", - ) - - def get_profile_avatar_url(self, user_localpart): - return self._simple_select_one_onecol( - table="profiles", - keyvalues={"user_id": user_localpart}, - retcol="avatar_url", - desc="get_profile_avatar_url", - ) - - def get_from_remote_profile_cache(self, user_id): - return self._simple_select_one( - table="remote_profile_cache", - keyvalues={"user_id": user_id}, - retcols=("displayname", "avatar_url"), - allow_none=True, - desc="get_from_remote_profile_cache", - ) - - def create_profile(self, user_localpart): - return self._simple_insert( - table="profiles", values={"user_id": user_localpart}, desc="create_profile" - ) - - def set_profile_displayname(self, user_localpart, new_displayname): - return self._simple_update_one( - table="profiles", - keyvalues={"user_id": user_localpart}, - updatevalues={"displayname": new_displayname}, - desc="set_profile_displayname", - ) - - def set_profile_avatar_url(self, user_localpart, new_avatar_url): - return self._simple_update_one( - table="profiles", - keyvalues={"user_id": user_localpart}, - updatevalues={"avatar_url": new_avatar_url}, - desc="set_profile_avatar_url", - ) - - -class ProfileStore(ProfileWorkerStore): - def add_remote_profile_cache(self, user_id, displayname, avatar_url): - """Ensure we are caching the remote user's profiles. - - This should only be called when `is_subscribed_remote_profile_for_user` - would return true for the user. - """ - return self._simple_upsert( - table="remote_profile_cache", - keyvalues={"user_id": user_id}, - values={ - "displayname": displayname, - "avatar_url": avatar_url, - "last_check": self._clock.time_msec(), - }, - desc="add_remote_profile_cache", - ) - - def update_remote_profile_cache(self, user_id, displayname, avatar_url): - return self._simple_update( - table="remote_profile_cache", - keyvalues={"user_id": user_id}, - values={ - "displayname": displayname, - "avatar_url": avatar_url, - "last_check": self._clock.time_msec(), - }, - desc="update_remote_profile_cache", - ) - - @defer.inlineCallbacks - def maybe_delete_remote_profile_cache(self, user_id): - """Check if we still care about the remote user's profile, and if we - don't then remove their profile from the cache - """ - subscribed = yield self.is_subscribed_remote_profile_for_user(user_id) - if not subscribed: - yield self._simple_delete( - table="remote_profile_cache", - keyvalues={"user_id": user_id}, - desc="delete_remote_profile_cache", - ) - - def get_remote_profile_cache_entries_that_expire(self, last_checked): - """Get all users who haven't been checked since `last_checked` - """ - - def _get_remote_profile_cache_entries_that_expire_txn(txn): - sql = """ - SELECT user_id, displayname, avatar_url - FROM remote_profile_cache - WHERE last_check < ? - """ - - txn.execute(sql, (last_checked,)) - - return self.cursor_to_dict(txn) - - return self.runInteraction( - "get_remote_profile_cache_entries_that_expire", - _get_remote_profile_cache_entries_that_expire_txn, - ) - - @defer.inlineCallbacks - def is_subscribed_remote_profile_for_user(self, user_id): - """Check whether we are interested in a remote user's profile. - """ - res = yield self._simple_select_one_onecol( - table="group_users", - keyvalues={"user_id": user_id}, - retcol="user_id", - allow_none=True, - desc="should_update_remote_profile_cache_for_user", - ) - - if res: - return True - - res = yield self._simple_select_one_onecol( - table="group_invites", - keyvalues={"user_id": user_id}, - retcol="user_id", - allow_none=True, - desc="should_update_remote_profile_cache_for_user", - ) - - if res: - return True diff --git a/synapse/storage/push_rule.py b/synapse/storage/push_rule.py index c4e24edff2..f47cec0d86 100644 --- a/synapse/storage/push_rule.py +++ b/synapse/storage/push_rule.py @@ -14,704 +14,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import abc -import logging - -from canonicaljson import json - -from twisted.internet import defer - -from synapse.push.baserules import list_with_base_rules -from synapse.storage.appservice import ApplicationServiceWorkerStore -from synapse.storage.pusher import PusherWorkerStore -from synapse.storage.receipts import ReceiptsWorkerStore -from synapse.storage.roommember import RoomMemberWorkerStore -from synapse.util.caches.descriptors import cachedInlineCallbacks, cachedList -from synapse.util.caches.stream_change_cache import StreamChangeCache - -from ._base import SQLBaseStore - -logger = logging.getLogger(__name__) - - -def _load_rules(rawrules, enabled_map): - ruleslist = [] - for rawrule in rawrules: - rule = dict(rawrule) - rule["conditions"] = json.loads(rawrule["conditions"]) - rule["actions"] = json.loads(rawrule["actions"]) - ruleslist.append(rule) - - # We're going to be mutating this a lot, so do a deep copy - rules = list(list_with_base_rules(ruleslist)) - - for i, rule in enumerate(rules): - rule_id = rule["rule_id"] - if rule_id in enabled_map: - if rule.get("enabled", True) != bool(enabled_map[rule_id]): - # Rules are cached across users. - rule = dict(rule) - rule["enabled"] = bool(enabled_map[rule_id]) - rules[i] = rule - - return rules - - -class PushRulesWorkerStore( - ApplicationServiceWorkerStore, - ReceiptsWorkerStore, - PusherWorkerStore, - RoomMemberWorkerStore, - SQLBaseStore, -): - """This is an abstract base class where subclasses must implement - `get_max_push_rules_stream_id` which can be called in the initializer. - """ - - # This ABCMeta metaclass ensures that we cannot be instantiated without - # the abstract methods being implemented. - __metaclass__ = abc.ABCMeta - - def __init__(self, db_conn, hs): - super(PushRulesWorkerStore, self).__init__(db_conn, hs) - - push_rules_prefill, push_rules_id = self._get_cache_dict( - db_conn, - "push_rules_stream", - entity_column="user_id", - stream_column="stream_id", - max_value=self.get_max_push_rules_stream_id(), - ) - - self.push_rules_stream_cache = StreamChangeCache( - "PushRulesStreamChangeCache", - push_rules_id, - prefilled_cache=push_rules_prefill, - ) - - @abc.abstractmethod - def get_max_push_rules_stream_id(self): - """Get the position of the push rules stream. - - Returns: - int - """ - raise NotImplementedError() - - @cachedInlineCallbacks(max_entries=5000) - def get_push_rules_for_user(self, user_id): - rows = yield self._simple_select_list( - table="push_rules", - keyvalues={"user_name": user_id}, - retcols=( - "user_name", - "rule_id", - "priority_class", - "priority", - "conditions", - "actions", - ), - desc="get_push_rules_enabled_for_user", - ) - - rows.sort(key=lambda row: (-int(row["priority_class"]), -int(row["priority"]))) - - enabled_map = yield self.get_push_rules_enabled_for_user(user_id) - - rules = _load_rules(rows, enabled_map) - - return rules - - @cachedInlineCallbacks(max_entries=5000) - def get_push_rules_enabled_for_user(self, user_id): - results = yield self._simple_select_list( - table="push_rules_enable", - keyvalues={"user_name": user_id}, - retcols=("user_name", "rule_id", "enabled"), - desc="get_push_rules_enabled_for_user", - ) - return {r["rule_id"]: False if r["enabled"] == 0 else True for r in results} - - def have_push_rules_changed_for_user(self, user_id, last_id): - if not self.push_rules_stream_cache.has_entity_changed(user_id, last_id): - return defer.succeed(False) - else: - - def have_push_rules_changed_txn(txn): - sql = ( - "SELECT COUNT(stream_id) FROM push_rules_stream" - " WHERE user_id = ? AND ? < stream_id" - ) - txn.execute(sql, (user_id, last_id)) - count, = txn.fetchone() - return bool(count) - - return self.runInteraction( - "have_push_rules_changed", have_push_rules_changed_txn - ) - - @cachedList( - cached_method_name="get_push_rules_for_user", - list_name="user_ids", - num_args=1, - inlineCallbacks=True, - ) - def bulk_get_push_rules(self, user_ids): - if not user_ids: - return {} - - results = {user_id: [] for user_id in user_ids} - - rows = yield self._simple_select_many_batch( - table="push_rules", - column="user_name", - iterable=user_ids, - retcols=("*",), - desc="bulk_get_push_rules", - ) - - rows.sort(key=lambda row: (-int(row["priority_class"]), -int(row["priority"]))) - - for row in rows: - results.setdefault(row["user_name"], []).append(row) - - enabled_map_by_user = yield self.bulk_get_push_rules_enabled(user_ids) - - for user_id, rules in results.items(): - results[user_id] = _load_rules(rules, enabled_map_by_user.get(user_id, {})) - - return results - - @defer.inlineCallbacks - def copy_push_rule_from_room_to_room(self, new_room_id, user_id, rule): - """Copy a single push rule from one room to another for a specific user. - - Args: - new_room_id (str): ID of the new room. - user_id (str): ID of user the push rule belongs to. - rule (Dict): A push rule. - """ - # Create new rule id - rule_id_scope = "/".join(rule["rule_id"].split("/")[:-1]) - new_rule_id = rule_id_scope + "/" + new_room_id - - # Change room id in each condition - for condition in rule.get("conditions", []): - if condition.get("key") == "room_id": - condition["pattern"] = new_room_id - - # Add the rule for the new room - yield self.add_push_rule( - user_id=user_id, - rule_id=new_rule_id, - priority_class=rule["priority_class"], - conditions=rule["conditions"], - actions=rule["actions"], - ) - - @defer.inlineCallbacks - def copy_push_rules_from_room_to_room_for_user( - self, old_room_id, new_room_id, user_id - ): - """Copy all of the push rules from one room to another for a specific - user. - - Args: - old_room_id (str): ID of the old room. - new_room_id (str): ID of the new room. - user_id (str): ID of user to copy push rules for. - """ - # Retrieve push rules for this user - user_push_rules = yield self.get_push_rules_for_user(user_id) - - # Get rules relating to the old room and copy them to the new room - for rule in user_push_rules: - conditions = rule.get("conditions", []) - if any( - (c.get("key") == "room_id" and c.get("pattern") == old_room_id) - for c in conditions - ): - yield self.copy_push_rule_from_room_to_room(new_room_id, user_id, rule) - - @defer.inlineCallbacks - def bulk_get_push_rules_for_room(self, event, context): - state_group = context.state_group - if not state_group: - # If state_group is None it means it has yet to be assigned a - # state group, i.e. we need to make sure that calls with a state_group - # of None don't hit previous cached calls with a None state_group. - # To do this we set the state_group to a new object as object() != object() - state_group = object() - - current_state_ids = yield context.get_current_state_ids(self) - result = yield self._bulk_get_push_rules_for_room( - event.room_id, state_group, current_state_ids, event=event - ) - return result - - @cachedInlineCallbacks(num_args=2, cache_context=True) - def _bulk_get_push_rules_for_room( - self, room_id, state_group, current_state_ids, cache_context, event=None - ): - # We don't use `state_group`, its there so that we can cache based - # on it. However, its important that its never None, since two current_state's - # with a state_group of None are likely to be different. - # See bulk_get_push_rules_for_room for how we work around this. - assert state_group is not None - - # We also will want to generate notifs for other people in the room so - # their unread countss are correct in the event stream, but to avoid - # generating them for bot / AS users etc, we only do so for people who've - # sent a read receipt into the room. - - users_in_room = yield self._get_joined_users_from_context( - room_id, - state_group, - current_state_ids, - on_invalidate=cache_context.invalidate, - event=event, - ) - - # We ignore app service users for now. This is so that we don't fill - # up the `get_if_users_have_pushers` cache with AS entries that we - # know don't have pushers, nor even read receipts. - local_users_in_room = set( - u - for u in users_in_room - if self.hs.is_mine_id(u) - and not self.get_if_app_services_interested_in_user(u) - ) - - # users in the room who have pushers need to get push rules run because - # that's how their pushers work - if_users_with_pushers = yield self.get_if_users_have_pushers( - local_users_in_room, on_invalidate=cache_context.invalidate - ) - user_ids = set( - uid for uid, have_pusher in if_users_with_pushers.items() if have_pusher - ) - - users_with_receipts = yield self.get_users_with_read_receipts_in_room( - room_id, on_invalidate=cache_context.invalidate - ) - - # any users with pushers must be ours: they have pushers - for uid in users_with_receipts: - if uid in local_users_in_room: - user_ids.add(uid) - - rules_by_user = yield self.bulk_get_push_rules( - user_ids, on_invalidate=cache_context.invalidate - ) - - rules_by_user = {k: v for k, v in rules_by_user.items() if v is not None} - - return rules_by_user - - @cachedList( - cached_method_name="get_push_rules_enabled_for_user", - list_name="user_ids", - num_args=1, - inlineCallbacks=True, - ) - def bulk_get_push_rules_enabled(self, user_ids): - if not user_ids: - return {} - - results = {user_id: {} for user_id in user_ids} - - rows = yield self._simple_select_many_batch( - table="push_rules_enable", - column="user_name", - iterable=user_ids, - retcols=("user_name", "rule_id", "enabled"), - desc="bulk_get_push_rules_enabled", - ) - for row in rows: - enabled = bool(row["enabled"]) - results.setdefault(row["user_name"], {})[row["rule_id"]] = enabled - return results - - -class PushRuleStore(PushRulesWorkerStore): - @defer.inlineCallbacks - def add_push_rule( - self, - user_id, - rule_id, - priority_class, - conditions, - actions, - before=None, - after=None, - ): - conditions_json = json.dumps(conditions) - actions_json = json.dumps(actions) - with self._push_rules_stream_id_gen.get_next() as ids: - stream_id, event_stream_ordering = ids - if before or after: - yield self.runInteraction( - "_add_push_rule_relative_txn", - self._add_push_rule_relative_txn, - stream_id, - event_stream_ordering, - user_id, - rule_id, - priority_class, - conditions_json, - actions_json, - before, - after, - ) - else: - yield self.runInteraction( - "_add_push_rule_highest_priority_txn", - self._add_push_rule_highest_priority_txn, - stream_id, - event_stream_ordering, - user_id, - rule_id, - priority_class, - conditions_json, - actions_json, - ) - - def _add_push_rule_relative_txn( - self, - txn, - stream_id, - event_stream_ordering, - user_id, - rule_id, - priority_class, - conditions_json, - actions_json, - before, - after, - ): - # Lock the table since otherwise we'll have annoying races between the - # SELECT here and the UPSERT below. - self.database_engine.lock_table(txn, "push_rules") - - relative_to_rule = before or after - - res = self._simple_select_one_txn( - txn, - table="push_rules", - keyvalues={"user_name": user_id, "rule_id": relative_to_rule}, - retcols=["priority_class", "priority"], - allow_none=True, - ) - - if not res: - raise RuleNotFoundException( - "before/after rule not found: %s" % (relative_to_rule,) - ) - - base_priority_class = res["priority_class"] - base_rule_priority = res["priority"] - - if base_priority_class != priority_class: - raise InconsistentRuleException( - "Given priority class does not match class of relative rule" - ) - - if before: - # Higher priority rules are executed first, So adding a rule before - # a rule means giving it a higher priority than that rule. - new_rule_priority = base_rule_priority + 1 - else: - # We increment the priority of the existing rules to make space for - # the new rule. Therefore if we want this rule to appear after - # an existing rule we give it the priority of the existing rule, - # and then increment the priority of the existing rule. - new_rule_priority = base_rule_priority - - sql = ( - "UPDATE push_rules SET priority = priority + 1" - " WHERE user_name = ? AND priority_class = ? AND priority >= ?" - ) - - txn.execute(sql, (user_id, priority_class, new_rule_priority)) - - self._upsert_push_rule_txn( - txn, - stream_id, - event_stream_ordering, - user_id, - rule_id, - priority_class, - new_rule_priority, - conditions_json, - actions_json, - ) - - def _add_push_rule_highest_priority_txn( - self, - txn, - stream_id, - event_stream_ordering, - user_id, - rule_id, - priority_class, - conditions_json, - actions_json, - ): - # Lock the table since otherwise we'll have annoying races between the - # SELECT here and the UPSERT below. - self.database_engine.lock_table(txn, "push_rules") - - # find the highest priority rule in that class - sql = ( - "SELECT COUNT(*), MAX(priority) FROM push_rules" - " WHERE user_name = ? and priority_class = ?" - ) - txn.execute(sql, (user_id, priority_class)) - res = txn.fetchall() - (how_many, highest_prio) = res[0] - - new_prio = 0 - if how_many > 0: - new_prio = highest_prio + 1 - - self._upsert_push_rule_txn( - txn, - stream_id, - event_stream_ordering, - user_id, - rule_id, - priority_class, - new_prio, - conditions_json, - actions_json, - ) - - def _upsert_push_rule_txn( - self, - txn, - stream_id, - event_stream_ordering, - user_id, - rule_id, - priority_class, - priority, - conditions_json, - actions_json, - update_stream=True, - ): - """Specialised version of _simple_upsert_txn that picks a push_rule_id - using the _push_rule_id_gen if it needs to insert the rule. It assumes - that the "push_rules" table is locked""" - - sql = ( - "UPDATE push_rules" - " SET priority_class = ?, priority = ?, conditions = ?, actions = ?" - " WHERE user_name = ? AND rule_id = ?" - ) - - txn.execute( - sql, - (priority_class, priority, conditions_json, actions_json, user_id, rule_id), - ) - - if txn.rowcount == 0: - # We didn't update a row with the given rule_id so insert one - push_rule_id = self._push_rule_id_gen.get_next() - - self._simple_insert_txn( - txn, - table="push_rules", - values={ - "id": push_rule_id, - "user_name": user_id, - "rule_id": rule_id, - "priority_class": priority_class, - "priority": priority, - "conditions": conditions_json, - "actions": actions_json, - }, - ) - - if update_stream: - self._insert_push_rules_update_txn( - txn, - stream_id, - event_stream_ordering, - user_id, - rule_id, - op="ADD", - data={ - "priority_class": priority_class, - "priority": priority, - "conditions": conditions_json, - "actions": actions_json, - }, - ) - - @defer.inlineCallbacks - def delete_push_rule(self, user_id, rule_id): - """ - Delete a push rule. Args specify the row to be deleted and can be - any of the columns in the push_rule table, but below are the - standard ones - - Args: - user_id (str): The matrix ID of the push rule owner - rule_id (str): The rule_id of the rule to be deleted - """ - - def delete_push_rule_txn(txn, stream_id, event_stream_ordering): - self._simple_delete_one_txn( - txn, "push_rules", {"user_name": user_id, "rule_id": rule_id} - ) - - self._insert_push_rules_update_txn( - txn, stream_id, event_stream_ordering, user_id, rule_id, op="DELETE" - ) - - with self._push_rules_stream_id_gen.get_next() as ids: - stream_id, event_stream_ordering = ids - yield self.runInteraction( - "delete_push_rule", - delete_push_rule_txn, - stream_id, - event_stream_ordering, - ) - - @defer.inlineCallbacks - def set_push_rule_enabled(self, user_id, rule_id, enabled): - with self._push_rules_stream_id_gen.get_next() as ids: - stream_id, event_stream_ordering = ids - yield self.runInteraction( - "_set_push_rule_enabled_txn", - self._set_push_rule_enabled_txn, - stream_id, - event_stream_ordering, - user_id, - rule_id, - enabled, - ) - - def _set_push_rule_enabled_txn( - self, txn, stream_id, event_stream_ordering, user_id, rule_id, enabled - ): - new_id = self._push_rules_enable_id_gen.get_next() - self._simple_upsert_txn( - txn, - "push_rules_enable", - {"user_name": user_id, "rule_id": rule_id}, - {"enabled": 1 if enabled else 0}, - {"id": new_id}, - ) - - self._insert_push_rules_update_txn( - txn, - stream_id, - event_stream_ordering, - user_id, - rule_id, - op="ENABLE" if enabled else "DISABLE", - ) - - @defer.inlineCallbacks - def set_push_rule_actions(self, user_id, rule_id, actions, is_default_rule): - actions_json = json.dumps(actions) - - def set_push_rule_actions_txn(txn, stream_id, event_stream_ordering): - if is_default_rule: - # Add a dummy rule to the rules table with the user specified - # actions. - priority_class = -1 - priority = 1 - self._upsert_push_rule_txn( - txn, - stream_id, - event_stream_ordering, - user_id, - rule_id, - priority_class, - priority, - "[]", - actions_json, - update_stream=False, - ) - else: - self._simple_update_one_txn( - txn, - "push_rules", - {"user_name": user_id, "rule_id": rule_id}, - {"actions": actions_json}, - ) - - self._insert_push_rules_update_txn( - txn, - stream_id, - event_stream_ordering, - user_id, - rule_id, - op="ACTIONS", - data={"actions": actions_json}, - ) - - with self._push_rules_stream_id_gen.get_next() as ids: - stream_id, event_stream_ordering = ids - yield self.runInteraction( - "set_push_rule_actions", - set_push_rule_actions_txn, - stream_id, - event_stream_ordering, - ) - - def _insert_push_rules_update_txn( - self, txn, stream_id, event_stream_ordering, user_id, rule_id, op, data=None - ): - values = { - "stream_id": stream_id, - "event_stream_ordering": event_stream_ordering, - "user_id": user_id, - "rule_id": rule_id, - "op": op, - } - if data is not None: - values.update(data) - - self._simple_insert_txn(txn, "push_rules_stream", values=values) - - txn.call_after(self.get_push_rules_for_user.invalidate, (user_id,)) - txn.call_after(self.get_push_rules_enabled_for_user.invalidate, (user_id,)) - txn.call_after( - self.push_rules_stream_cache.entity_has_changed, user_id, stream_id - ) - - def get_all_push_rule_updates(self, last_id, current_id, limit): - """Get all the push rules changes that have happend on the server""" - if last_id == current_id: - return defer.succeed([]) - - def get_all_push_rule_updates_txn(txn): - sql = ( - "SELECT stream_id, event_stream_ordering, user_id, rule_id," - " op, priority_class, priority, conditions, actions" - " FROM push_rules_stream" - " WHERE ? < stream_id AND stream_id <= ?" - " ORDER BY stream_id ASC LIMIT ?" - ) - txn.execute(sql, (last_id, current_id, limit)) - return txn.fetchall() - - return self.runInteraction( - "get_all_push_rule_updates", get_all_push_rule_updates_txn - ) - - def get_push_rules_stream_token(self): - """Get the position of the push rules stream. - Returns a pair of a stream id for the push_rules stream and the - room stream ordering it corresponds to.""" - return self._push_rules_stream_id_gen.get_current_token() - - def get_max_push_rules_stream_id(self): - return self.get_push_rules_stream_token()[0] - class RuleNotFoundException(Exception): pass diff --git a/synapse/storage/pusher.py b/synapse/storage/pusher.py deleted file mode 100644 index b12e80440a..0000000000 --- a/synapse/storage/pusher.py +++ /dev/null @@ -1,372 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2014-2016 OpenMarket Ltd -# Copyright 2018 New Vector Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import logging - -import six - -from canonicaljson import encode_canonical_json, json - -from twisted.internet import defer - -from synapse.util.caches.descriptors import cachedInlineCallbacks, cachedList - -from ._base import SQLBaseStore - -logger = logging.getLogger(__name__) - -if six.PY2: - db_binary_type = six.moves.builtins.buffer -else: - db_binary_type = memoryview - - -class PusherWorkerStore(SQLBaseStore): - def _decode_pushers_rows(self, rows): - for r in rows: - dataJson = r["data"] - r["data"] = None - try: - if isinstance(dataJson, db_binary_type): - dataJson = str(dataJson).decode("UTF8") - - r["data"] = json.loads(dataJson) - except Exception as e: - logger.warn( - "Invalid JSON in data for pusher %d: %s, %s", - r["id"], - dataJson, - e.args[0], - ) - pass - - if isinstance(r["pushkey"], db_binary_type): - r["pushkey"] = str(r["pushkey"]).decode("UTF8") - - return rows - - @defer.inlineCallbacks - def user_has_pusher(self, user_id): - ret = yield self._simple_select_one_onecol( - "pushers", {"user_name": user_id}, "id", allow_none=True - ) - return ret is not None - - def get_pushers_by_app_id_and_pushkey(self, app_id, pushkey): - return self.get_pushers_by({"app_id": app_id, "pushkey": pushkey}) - - def get_pushers_by_user_id(self, user_id): - return self.get_pushers_by({"user_name": user_id}) - - @defer.inlineCallbacks - def get_pushers_by(self, keyvalues): - ret = yield self._simple_select_list( - "pushers", - keyvalues, - [ - "id", - "user_name", - "access_token", - "profile_tag", - "kind", - "app_id", - "app_display_name", - "device_display_name", - "pushkey", - "ts", - "lang", - "data", - "last_stream_ordering", - "last_success", - "failing_since", - ], - desc="get_pushers_by", - ) - return self._decode_pushers_rows(ret) - - @defer.inlineCallbacks - def get_all_pushers(self): - def get_pushers(txn): - txn.execute("SELECT * FROM pushers") - rows = self.cursor_to_dict(txn) - - return self._decode_pushers_rows(rows) - - rows = yield self.runInteraction("get_all_pushers", get_pushers) - return rows - - def get_all_updated_pushers(self, last_id, current_id, limit): - if last_id == current_id: - return defer.succeed(([], [])) - - def get_all_updated_pushers_txn(txn): - sql = ( - "SELECT id, user_name, access_token, profile_tag, kind," - " app_id, app_display_name, device_display_name, pushkey, ts," - " lang, data" - " FROM pushers" - " WHERE ? < id AND id <= ?" - " ORDER BY id ASC LIMIT ?" - ) - txn.execute(sql, (last_id, current_id, limit)) - updated = txn.fetchall() - - sql = ( - "SELECT stream_id, user_id, app_id, pushkey" - " FROM deleted_pushers" - " WHERE ? < stream_id AND stream_id <= ?" - " ORDER BY stream_id ASC LIMIT ?" - ) - txn.execute(sql, (last_id, current_id, limit)) - deleted = txn.fetchall() - - return updated, deleted - - return self.runInteraction( - "get_all_updated_pushers", get_all_updated_pushers_txn - ) - - def get_all_updated_pushers_rows(self, last_id, current_id, limit): - """Get all the pushers that have changed between the given tokens. - - Returns: - Deferred(list(tuple)): each tuple consists of: - stream_id (str) - user_id (str) - app_id (str) - pushkey (str) - was_deleted (bool): whether the pusher was added/updated (False) - or deleted (True) - """ - - if last_id == current_id: - return defer.succeed([]) - - def get_all_updated_pushers_rows_txn(txn): - sql = ( - "SELECT id, user_name, app_id, pushkey" - " FROM pushers" - " WHERE ? < id AND id <= ?" - " ORDER BY id ASC LIMIT ?" - ) - txn.execute(sql, (last_id, current_id, limit)) - results = [list(row) + [False] for row in txn] - - sql = ( - "SELECT stream_id, user_id, app_id, pushkey" - " FROM deleted_pushers" - " WHERE ? < stream_id AND stream_id <= ?" - " ORDER BY stream_id ASC LIMIT ?" - ) - txn.execute(sql, (last_id, current_id, limit)) - - results.extend(list(row) + [True] for row in txn) - results.sort() # Sort so that they're ordered by stream id - - return results - - return self.runInteraction( - "get_all_updated_pushers_rows", get_all_updated_pushers_rows_txn - ) - - @cachedInlineCallbacks(num_args=1, max_entries=15000) - def get_if_user_has_pusher(self, user_id): - # This only exists for the cachedList decorator - raise NotImplementedError() - - @cachedList( - cached_method_name="get_if_user_has_pusher", - list_name="user_ids", - num_args=1, - inlineCallbacks=True, - ) - def get_if_users_have_pushers(self, user_ids): - rows = yield self._simple_select_many_batch( - table="pushers", - column="user_name", - iterable=user_ids, - retcols=["user_name"], - desc="get_if_users_have_pushers", - ) - - result = {user_id: False for user_id in user_ids} - result.update({r["user_name"]: True for r in rows}) - - return result - - -class PusherStore(PusherWorkerStore): - def get_pushers_stream_token(self): - return self._pushers_id_gen.get_current_token() - - @defer.inlineCallbacks - def add_pusher( - self, - user_id, - access_token, - kind, - app_id, - app_display_name, - device_display_name, - pushkey, - pushkey_ts, - lang, - data, - last_stream_ordering, - profile_tag="", - ): - with self._pushers_id_gen.get_next() as stream_id: - # no need to lock because `pushers` has a unique key on - # (app_id, pushkey, user_name) so _simple_upsert will retry - yield self._simple_upsert( - table="pushers", - keyvalues={"app_id": app_id, "pushkey": pushkey, "user_name": user_id}, - values={ - "access_token": access_token, - "kind": kind, - "app_display_name": app_display_name, - "device_display_name": device_display_name, - "ts": pushkey_ts, - "lang": lang, - "data": bytearray(encode_canonical_json(data)), - "last_stream_ordering": last_stream_ordering, - "profile_tag": profile_tag, - "id": stream_id, - }, - desc="add_pusher", - lock=False, - ) - - user_has_pusher = self.get_if_user_has_pusher.cache.get( - (user_id,), None, update_metrics=False - ) - - if user_has_pusher is not True: - # invalidate, since we the user might not have had a pusher before - yield self.runInteraction( - "add_pusher", - self._invalidate_cache_and_stream, - self.get_if_user_has_pusher, - (user_id,), - ) - - @defer.inlineCallbacks - def delete_pusher_by_app_id_pushkey_user_id(self, app_id, pushkey, user_id): - def delete_pusher_txn(txn, stream_id): - self._invalidate_cache_and_stream( - txn, self.get_if_user_has_pusher, (user_id,) - ) - - self._simple_delete_one_txn( - txn, - "pushers", - {"app_id": app_id, "pushkey": pushkey, "user_name": user_id}, - ) - - # it's possible for us to end up with duplicate rows for - # (app_id, pushkey, user_id) at different stream_ids, but that - # doesn't really matter. - self._simple_insert_txn( - txn, - table="deleted_pushers", - values={ - "stream_id": stream_id, - "app_id": app_id, - "pushkey": pushkey, - "user_id": user_id, - }, - ) - - with self._pushers_id_gen.get_next() as stream_id: - yield self.runInteraction("delete_pusher", delete_pusher_txn, stream_id) - - @defer.inlineCallbacks - def update_pusher_last_stream_ordering( - self, app_id, pushkey, user_id, last_stream_ordering - ): - yield self._simple_update_one( - "pushers", - {"app_id": app_id, "pushkey": pushkey, "user_name": user_id}, - {"last_stream_ordering": last_stream_ordering}, - desc="update_pusher_last_stream_ordering", - ) - - @defer.inlineCallbacks - def update_pusher_last_stream_ordering_and_success( - self, app_id, pushkey, user_id, last_stream_ordering, last_success - ): - """Update the last stream ordering position we've processed up to for - the given pusher. - - Args: - app_id (str) - pushkey (str) - last_stream_ordering (int) - last_success (int) - - Returns: - Deferred[bool]: True if the pusher still exists; False if it has been deleted. - """ - updated = yield self._simple_update( - table="pushers", - keyvalues={"app_id": app_id, "pushkey": pushkey, "user_name": user_id}, - updatevalues={ - "last_stream_ordering": last_stream_ordering, - "last_success": last_success, - }, - desc="update_pusher_last_stream_ordering_and_success", - ) - - return bool(updated) - - @defer.inlineCallbacks - def update_pusher_failing_since(self, app_id, pushkey, user_id, failing_since): - yield self._simple_update( - table="pushers", - keyvalues={"app_id": app_id, "pushkey": pushkey, "user_name": user_id}, - updatevalues={"failing_since": failing_since}, - desc="update_pusher_failing_since", - ) - - @defer.inlineCallbacks - def get_throttle_params_by_room(self, pusher_id): - res = yield self._simple_select_list( - "pusher_throttle", - {"pusher": pusher_id}, - ["room_id", "last_sent_ts", "throttle_ms"], - desc="get_throttle_params_by_room", - ) - - params_by_room = {} - for row in res: - params_by_room[row["room_id"]] = { - "last_sent_ts": row["last_sent_ts"], - "throttle_ms": row["throttle_ms"], - } - - return params_by_room - - @defer.inlineCallbacks - def set_throttle_params(self, pusher_id, room_id, params): - # no need to lock because `pusher_throttle` has a primary key on - # (pusher, room_id) so _simple_upsert will retry - yield self._simple_upsert( - "pusher_throttle", - {"pusher": pusher_id, "room_id": room_id}, - params, - desc="set_throttle_params", - lock=False, - ) diff --git a/synapse/storage/receipts.py b/synapse/storage/receipts.py deleted file mode 100644 index 0c24430f28..0000000000 --- a/synapse/storage/receipts.py +++ /dev/null @@ -1,536 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2014-2016 OpenMarket Ltd -# Copyright 2018 New Vector Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import abc -import logging - -from canonicaljson import json - -from twisted.internet import defer - -from synapse.storage._base import SQLBaseStore, make_in_list_sql_clause -from synapse.storage.util.id_generators import StreamIdGenerator -from synapse.util.caches.descriptors import cached, cachedInlineCallbacks, cachedList -from synapse.util.caches.stream_change_cache import StreamChangeCache - -logger = logging.getLogger(__name__) - - -class ReceiptsWorkerStore(SQLBaseStore): - """This is an abstract base class where subclasses must implement - `get_max_receipt_stream_id` which can be called in the initializer. - """ - - # This ABCMeta metaclass ensures that we cannot be instantiated without - # the abstract methods being implemented. - __metaclass__ = abc.ABCMeta - - def __init__(self, db_conn, hs): - super(ReceiptsWorkerStore, self).__init__(db_conn, hs) - - self._receipts_stream_cache = StreamChangeCache( - "ReceiptsRoomChangeCache", self.get_max_receipt_stream_id() - ) - - @abc.abstractmethod - def get_max_receipt_stream_id(self): - """Get the current max stream ID for receipts stream - - Returns: - int - """ - raise NotImplementedError() - - @cachedInlineCallbacks() - def get_users_with_read_receipts_in_room(self, room_id): - receipts = yield self.get_receipts_for_room(room_id, "m.read") - return set(r["user_id"] for r in receipts) - - @cached(num_args=2) - def get_receipts_for_room(self, room_id, receipt_type): - return self._simple_select_list( - table="receipts_linearized", - keyvalues={"room_id": room_id, "receipt_type": receipt_type}, - retcols=("user_id", "event_id"), - desc="get_receipts_for_room", - ) - - @cached(num_args=3) - def get_last_receipt_event_id_for_user(self, user_id, room_id, receipt_type): - return self._simple_select_one_onecol( - table="receipts_linearized", - keyvalues={ - "room_id": room_id, - "receipt_type": receipt_type, - "user_id": user_id, - }, - retcol="event_id", - desc="get_own_receipt_for_user", - allow_none=True, - ) - - @cachedInlineCallbacks(num_args=2) - def get_receipts_for_user(self, user_id, receipt_type): - rows = yield self._simple_select_list( - table="receipts_linearized", - keyvalues={"user_id": user_id, "receipt_type": receipt_type}, - retcols=("room_id", "event_id"), - desc="get_receipts_for_user", - ) - - return {row["room_id"]: row["event_id"] for row in rows} - - @defer.inlineCallbacks - def get_receipts_for_user_with_orderings(self, user_id, receipt_type): - def f(txn): - sql = ( - "SELECT rl.room_id, rl.event_id," - " e.topological_ordering, e.stream_ordering" - " FROM receipts_linearized AS rl" - " INNER JOIN events AS e USING (room_id, event_id)" - " WHERE rl.room_id = e.room_id" - " AND rl.event_id = e.event_id" - " AND user_id = ?" - ) - txn.execute(sql, (user_id,)) - return txn.fetchall() - - rows = yield self.runInteraction("get_receipts_for_user_with_orderings", f) - return { - row[0]: { - "event_id": row[1], - "topological_ordering": row[2], - "stream_ordering": row[3], - } - for row in rows - } - - @defer.inlineCallbacks - def get_linearized_receipts_for_rooms(self, room_ids, to_key, from_key=None): - """Get receipts for multiple rooms for sending to clients. - - Args: - room_ids (list): List of room_ids. - to_key (int): Max stream id to fetch receipts upto. - from_key (int): Min stream id to fetch receipts from. None fetches - from the start. - - Returns: - list: A list of receipts. - """ - room_ids = set(room_ids) - - if from_key is not None: - # Only ask the database about rooms where there have been new - # receipts added since `from_key` - room_ids = yield self._receipts_stream_cache.get_entities_changed( - room_ids, from_key - ) - - results = yield self._get_linearized_receipts_for_rooms( - room_ids, to_key, from_key=from_key - ) - - return [ev for res in results.values() for ev in res] - - def get_linearized_receipts_for_room(self, room_id, to_key, from_key=None): - """Get receipts for a single room for sending to clients. - - Args: - room_ids (str): The room id. - to_key (int): Max stream id to fetch receipts upto. - from_key (int): Min stream id to fetch receipts from. None fetches - from the start. - - Returns: - Deferred[list]: A list of receipts. - """ - if from_key is not None: - # Check the cache first to see if any new receipts have been added - # since`from_key`. If not we can no-op. - if not self._receipts_stream_cache.has_entity_changed(room_id, from_key): - defer.succeed([]) - - return self._get_linearized_receipts_for_room(room_id, to_key, from_key) - - @cachedInlineCallbacks(num_args=3, tree=True) - def _get_linearized_receipts_for_room(self, room_id, to_key, from_key=None): - """See get_linearized_receipts_for_room - """ - - def f(txn): - if from_key: - sql = ( - "SELECT * FROM receipts_linearized WHERE" - " room_id = ? AND stream_id > ? AND stream_id <= ?" - ) - - txn.execute(sql, (room_id, from_key, to_key)) - else: - sql = ( - "SELECT * FROM receipts_linearized WHERE" - " room_id = ? AND stream_id <= ?" - ) - - txn.execute(sql, (room_id, to_key)) - - rows = self.cursor_to_dict(txn) - - return rows - - rows = yield self.runInteraction("get_linearized_receipts_for_room", f) - - if not rows: - return [] - - content = {} - for row in rows: - content.setdefault(row["event_id"], {}).setdefault(row["receipt_type"], {})[ - row["user_id"] - ] = json.loads(row["data"]) - - return [{"type": "m.receipt", "room_id": room_id, "content": content}] - - @cachedList( - cached_method_name="_get_linearized_receipts_for_room", - list_name="room_ids", - num_args=3, - inlineCallbacks=True, - ) - def _get_linearized_receipts_for_rooms(self, room_ids, to_key, from_key=None): - if not room_ids: - return {} - - def f(txn): - if from_key: - sql = """ - SELECT * FROM receipts_linearized WHERE - stream_id > ? AND stream_id <= ? AND - """ - clause, args = make_in_list_sql_clause( - self.database_engine, "room_id", room_ids - ) - - txn.execute(sql + clause, [from_key, to_key] + list(args)) - else: - sql = """ - SELECT * FROM receipts_linearized WHERE - stream_id <= ? AND - """ - - clause, args = make_in_list_sql_clause( - self.database_engine, "room_id", room_ids - ) - - txn.execute(sql + clause, [to_key] + list(args)) - - return self.cursor_to_dict(txn) - - txn_results = yield self.runInteraction("_get_linearized_receipts_for_rooms", f) - - results = {} - for row in txn_results: - # We want a single event per room, since we want to batch the - # receipts by room, event and type. - room_event = results.setdefault( - row["room_id"], - {"type": "m.receipt", "room_id": row["room_id"], "content": {}}, - ) - - # The content is of the form: - # {"$foo:bar": { "read": { "@user:host": }, .. }, .. } - event_entry = room_event["content"].setdefault(row["event_id"], {}) - receipt_type = event_entry.setdefault(row["receipt_type"], {}) - - receipt_type[row["user_id"]] = json.loads(row["data"]) - - results = { - room_id: [results[room_id]] if room_id in results else [] - for room_id in room_ids - } - return results - - def get_all_updated_receipts(self, last_id, current_id, limit=None): - if last_id == current_id: - return defer.succeed([]) - - def get_all_updated_receipts_txn(txn): - sql = ( - "SELECT stream_id, room_id, receipt_type, user_id, event_id, data" - " FROM receipts_linearized" - " WHERE ? < stream_id AND stream_id <= ?" - " ORDER BY stream_id ASC" - ) - args = [last_id, current_id] - if limit is not None: - sql += " LIMIT ?" - args.append(limit) - txn.execute(sql, args) - - return (r[0:5] + (json.loads(r[5]),) for r in txn) - - return self.runInteraction( - "get_all_updated_receipts", get_all_updated_receipts_txn - ) - - def _invalidate_get_users_with_receipts_in_room( - self, room_id, receipt_type, user_id - ): - if receipt_type != "m.read": - return - - # Returns either an ObservableDeferred or the raw result - res = self.get_users_with_read_receipts_in_room.cache.get( - room_id, None, update_metrics=False - ) - - # first handle the Deferred case - if isinstance(res, defer.Deferred): - if res.called: - res = res.result - else: - res = None - - if res and user_id in res: - # We'd only be adding to the set, so no point invalidating if the - # user is already there - return - - self.get_users_with_read_receipts_in_room.invalidate((room_id,)) - - -class ReceiptsStore(ReceiptsWorkerStore): - def __init__(self, db_conn, hs): - # We instantiate this first as the ReceiptsWorkerStore constructor - # needs to be able to call get_max_receipt_stream_id - self._receipts_id_gen = StreamIdGenerator( - db_conn, "receipts_linearized", "stream_id" - ) - - super(ReceiptsStore, self).__init__(db_conn, hs) - - def get_max_receipt_stream_id(self): - return self._receipts_id_gen.get_current_token() - - def insert_linearized_receipt_txn( - self, txn, room_id, receipt_type, user_id, event_id, data, stream_id - ): - """Inserts a read-receipt into the database if it's newer than the current RR - - Returns: int|None - None if the RR is older than the current RR - otherwise, the rx timestamp of the event that the RR corresponds to - (or 0 if the event is unknown) - """ - res = self._simple_select_one_txn( - txn, - table="events", - retcols=["stream_ordering", "received_ts"], - keyvalues={"event_id": event_id}, - allow_none=True, - ) - - stream_ordering = int(res["stream_ordering"]) if res else None - rx_ts = res["received_ts"] if res else 0 - - # We don't want to clobber receipts for more recent events, so we - # have to compare orderings of existing receipts - if stream_ordering is not None: - sql = ( - "SELECT stream_ordering, event_id FROM events" - " INNER JOIN receipts_linearized as r USING (event_id, room_id)" - " WHERE r.room_id = ? AND r.receipt_type = ? AND r.user_id = ?" - ) - txn.execute(sql, (room_id, receipt_type, user_id)) - - for so, eid in txn: - if int(so) >= stream_ordering: - logger.debug( - "Ignoring new receipt for %s in favour of existing " - "one for later event %s", - event_id, - eid, - ) - return None - - txn.call_after(self.get_receipts_for_room.invalidate, (room_id, receipt_type)) - txn.call_after( - self._invalidate_get_users_with_receipts_in_room, - room_id, - receipt_type, - user_id, - ) - txn.call_after(self.get_receipts_for_user.invalidate, (user_id, receipt_type)) - # FIXME: This shouldn't invalidate the whole cache - txn.call_after( - self._get_linearized_receipts_for_room.invalidate_many, (room_id,) - ) - - txn.call_after( - self._receipts_stream_cache.entity_has_changed, room_id, stream_id - ) - - txn.call_after( - self.get_last_receipt_event_id_for_user.invalidate, - (user_id, room_id, receipt_type), - ) - - self._simple_delete_txn( - txn, - table="receipts_linearized", - keyvalues={ - "room_id": room_id, - "receipt_type": receipt_type, - "user_id": user_id, - }, - ) - - self._simple_insert_txn( - txn, - table="receipts_linearized", - values={ - "stream_id": stream_id, - "room_id": room_id, - "receipt_type": receipt_type, - "user_id": user_id, - "event_id": event_id, - "data": json.dumps(data), - }, - ) - - if receipt_type == "m.read" and stream_ordering is not None: - self._remove_old_push_actions_before_txn( - txn, room_id=room_id, user_id=user_id, stream_ordering=stream_ordering - ) - - return rx_ts - - @defer.inlineCallbacks - def insert_receipt(self, room_id, receipt_type, user_id, event_ids, data): - """Insert a receipt, either from local client or remote server. - - Automatically does conversion between linearized and graph - representations. - """ - if not event_ids: - return - - if len(event_ids) == 1: - linearized_event_id = event_ids[0] - else: - # we need to points in graph -> linearized form. - # TODO: Make this better. - def graph_to_linear(txn): - clause, args = make_in_list_sql_clause( - self.database_engine, "event_id", event_ids - ) - - sql = """ - SELECT event_id WHERE room_id = ? AND stream_ordering IN ( - SELECT max(stream_ordering) WHERE %s - ) - """ % ( - clause, - ) - - txn.execute(sql, [room_id] + list(args)) - rows = txn.fetchall() - if rows: - return rows[0][0] - else: - raise RuntimeError("Unrecognized event_ids: %r" % (event_ids,)) - - linearized_event_id = yield self.runInteraction( - "insert_receipt_conv", graph_to_linear - ) - - stream_id_manager = self._receipts_id_gen.get_next() - with stream_id_manager as stream_id: - event_ts = yield self.runInteraction( - "insert_linearized_receipt", - self.insert_linearized_receipt_txn, - room_id, - receipt_type, - user_id, - linearized_event_id, - data, - stream_id=stream_id, - ) - - if event_ts is None: - return None - - now = self._clock.time_msec() - logger.debug( - "RR for event %s in %s (%i ms old)", - linearized_event_id, - room_id, - now - event_ts, - ) - - yield self.insert_graph_receipt(room_id, receipt_type, user_id, event_ids, data) - - max_persisted_id = self._receipts_id_gen.get_current_token() - - return stream_id, max_persisted_id - - def insert_graph_receipt(self, room_id, receipt_type, user_id, event_ids, data): - return self.runInteraction( - "insert_graph_receipt", - self.insert_graph_receipt_txn, - room_id, - receipt_type, - user_id, - event_ids, - data, - ) - - def insert_graph_receipt_txn( - self, txn, room_id, receipt_type, user_id, event_ids, data - ): - txn.call_after(self.get_receipts_for_room.invalidate, (room_id, receipt_type)) - txn.call_after( - self._invalidate_get_users_with_receipts_in_room, - room_id, - receipt_type, - user_id, - ) - txn.call_after(self.get_receipts_for_user.invalidate, (user_id, receipt_type)) - # FIXME: This shouldn't invalidate the whole cache - txn.call_after( - self._get_linearized_receipts_for_room.invalidate_many, (room_id,) - ) - - self._simple_delete_txn( - txn, - table="receipts_graph", - keyvalues={ - "room_id": room_id, - "receipt_type": receipt_type, - "user_id": user_id, - }, - ) - self._simple_insert_txn( - txn, - table="receipts_graph", - values={ - "room_id": room_id, - "receipt_type": receipt_type, - "user_id": user_id, - "event_ids": json.dumps(event_ids), - "data": json.dumps(data), - }, - ) diff --git a/synapse/storage/registration.py b/synapse/storage/registration.py deleted file mode 100644 index 6c5b29288a..0000000000 --- a/synapse/storage/registration.py +++ /dev/null @@ -1,1499 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2014-2016 OpenMarket Ltd -# Copyright 2017-2018 New Vector Ltd -# Copyright 2019 The Matrix.org Foundation C.I.C. -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import logging -import re - -from six import iterkeys -from six.moves import range - -from twisted.internet import defer -from twisted.internet.defer import Deferred - -from synapse.api.constants import UserTypes -from synapse.api.errors import Codes, StoreError, SynapseError, ThreepidValidationError -from synapse.metrics.background_process_metrics import run_as_background_process -from synapse.storage import background_updates -from synapse.storage._base import SQLBaseStore -from synapse.types import UserID -from synapse.util.caches.descriptors import cached, cachedInlineCallbacks - -THIRTY_MINUTES_IN_MS = 30 * 60 * 1000 - -logger = logging.getLogger(__name__) - - -class RegistrationWorkerStore(SQLBaseStore): - def __init__(self, db_conn, hs): - super(RegistrationWorkerStore, self).__init__(db_conn, hs) - - self.config = hs.config - self.clock = hs.get_clock() - - @cached() - def get_user_by_id(self, user_id): - return self._simple_select_one( - table="users", - keyvalues={"name": user_id}, - retcols=[ - "name", - "password_hash", - "is_guest", - "consent_version", - "consent_server_notice_sent", - "appservice_id", - "creation_ts", - "user_type", - ], - allow_none=True, - desc="get_user_by_id", - ) - - @defer.inlineCallbacks - def is_trial_user(self, user_id): - """Checks if user is in the "trial" period, i.e. within the first - N days of registration defined by `mau_trial_days` config - - Args: - user_id (str) - - Returns: - Deferred[bool] - """ - - info = yield self.get_user_by_id(user_id) - if not info: - return False - - now = self.clock.time_msec() - trial_duration_ms = self.config.mau_trial_days * 24 * 60 * 60 * 1000 - is_trial = (now - info["creation_ts"] * 1000) < trial_duration_ms - return is_trial - - @cached() - def get_user_by_access_token(self, token): - """Get a user from the given access token. - - Args: - token (str): The access token of a user. - Returns: - defer.Deferred: None, if the token did not match, otherwise dict - including the keys `name`, `is_guest`, `device_id`, `token_id`, - `valid_until_ms`. - """ - return self.runInteraction( - "get_user_by_access_token", self._query_for_auth, token - ) - - @cachedInlineCallbacks() - def get_expiration_ts_for_user(self, user_id): - """Get the expiration timestamp for the account bearing a given user ID. - - Args: - user_id (str): The ID of the user. - Returns: - defer.Deferred: None, if the account has no expiration timestamp, - otherwise int representation of the timestamp (as a number of - milliseconds since epoch). - """ - res = yield self._simple_select_one_onecol( - table="account_validity", - keyvalues={"user_id": user_id}, - retcol="expiration_ts_ms", - allow_none=True, - desc="get_expiration_ts_for_user", - ) - return res - - @defer.inlineCallbacks - def set_account_validity_for_user( - self, user_id, expiration_ts, email_sent, renewal_token=None - ): - """Updates the account validity properties of the given account, with the - given values. - - Args: - user_id (str): ID of the account to update properties for. - expiration_ts (int): New expiration date, as a timestamp in milliseconds - since epoch. - email_sent (bool): True means a renewal email has been sent for this - account and there's no need to send another one for the current validity - period. - renewal_token (str): Renewal token the user can use to extend the validity - of their account. Defaults to no token. - """ - - def set_account_validity_for_user_txn(txn): - self._simple_update_txn( - txn=txn, - table="account_validity", - keyvalues={"user_id": user_id}, - updatevalues={ - "expiration_ts_ms": expiration_ts, - "email_sent": email_sent, - "renewal_token": renewal_token, - }, - ) - self._invalidate_cache_and_stream( - txn, self.get_expiration_ts_for_user, (user_id,) - ) - - yield self.runInteraction( - "set_account_validity_for_user", set_account_validity_for_user_txn - ) - - @defer.inlineCallbacks - def set_renewal_token_for_user(self, user_id, renewal_token): - """Defines a renewal token for a given user. - - Args: - user_id (str): ID of the user to set the renewal token for. - renewal_token (str): Random unique string that will be used to renew the - user's account. - - Raises: - StoreError: The provided token is already set for another user. - """ - yield self._simple_update_one( - table="account_validity", - keyvalues={"user_id": user_id}, - updatevalues={"renewal_token": renewal_token}, - desc="set_renewal_token_for_user", - ) - - @defer.inlineCallbacks - def get_user_from_renewal_token(self, renewal_token): - """Get a user ID from a renewal token. - - Args: - renewal_token (str): The renewal token to perform the lookup with. - - Returns: - defer.Deferred[str]: The ID of the user to which the token belongs. - """ - res = yield self._simple_select_one_onecol( - table="account_validity", - keyvalues={"renewal_token": renewal_token}, - retcol="user_id", - desc="get_user_from_renewal_token", - ) - - return res - - @defer.inlineCallbacks - def get_renewal_token_for_user(self, user_id): - """Get the renewal token associated with a given user ID. - - Args: - user_id (str): The user ID to lookup a token for. - - Returns: - defer.Deferred[str]: The renewal token associated with this user ID. - """ - res = yield self._simple_select_one_onecol( - table="account_validity", - keyvalues={"user_id": user_id}, - retcol="renewal_token", - desc="get_renewal_token_for_user", - ) - - return res - - @defer.inlineCallbacks - def get_users_expiring_soon(self): - """Selects users whose account will expire in the [now, now + renew_at] time - window (see configuration for account_validity for information on what renew_at - refers to). - - Returns: - Deferred: Resolves to a list[dict[user_id (str), expiration_ts_ms (int)]] - """ - - def select_users_txn(txn, now_ms, renew_at): - sql = ( - "SELECT user_id, expiration_ts_ms FROM account_validity" - " WHERE email_sent = ? AND (expiration_ts_ms - ?) <= ?" - ) - values = [False, now_ms, renew_at] - txn.execute(sql, values) - return self.cursor_to_dict(txn) - - res = yield self.runInteraction( - "get_users_expiring_soon", - select_users_txn, - self.clock.time_msec(), - self.config.account_validity.renew_at, - ) - - return res - - @defer.inlineCallbacks - def set_renewal_mail_status(self, user_id, email_sent): - """Sets or unsets the flag that indicates whether a renewal email has been sent - to the user (and the user hasn't renewed their account yet). - - Args: - user_id (str): ID of the user to set/unset the flag for. - email_sent (bool): Flag which indicates whether a renewal email has been sent - to this user. - """ - yield self._simple_update_one( - table="account_validity", - keyvalues={"user_id": user_id}, - updatevalues={"email_sent": email_sent}, - desc="set_renewal_mail_status", - ) - - @defer.inlineCallbacks - def delete_account_validity_for_user(self, user_id): - """Deletes the entry for the given user in the account validity table, removing - their expiration date and renewal token. - - Args: - user_id (str): ID of the user to remove from the account validity table. - """ - yield self._simple_delete_one( - table="account_validity", - keyvalues={"user_id": user_id}, - desc="delete_account_validity_for_user", - ) - - @defer.inlineCallbacks - def is_server_admin(self, user): - """Determines if a user is an admin of this homeserver. - - Args: - user (UserID): user ID of the user to test - - Returns (bool): - true iff the user is a server admin, false otherwise. - """ - res = yield self._simple_select_one_onecol( - table="users", - keyvalues={"name": user.to_string()}, - retcol="admin", - allow_none=True, - desc="is_server_admin", - ) - - return res if res else False - - def set_server_admin(self, user, admin): - """Sets whether a user is an admin of this homeserver. - - Args: - user (UserID): user ID of the user to test - admin (bool): true iff the user is to be a server admin, - false otherwise. - """ - return self._simple_update_one( - table="users", - keyvalues={"name": user.to_string()}, - updatevalues={"admin": 1 if admin else 0}, - desc="set_server_admin", - ) - - def _query_for_auth(self, txn, token): - sql = ( - "SELECT users.name, users.is_guest, access_tokens.id as token_id," - " access_tokens.device_id, access_tokens.valid_until_ms" - " FROM users" - " INNER JOIN access_tokens on users.name = access_tokens.user_id" - " WHERE token = ?" - ) - - txn.execute(sql, (token,)) - rows = self.cursor_to_dict(txn) - if rows: - return rows[0] - - return None - - @cachedInlineCallbacks() - def is_real_user(self, user_id): - """Determines if the user is a real user, ie does not have a 'user_type'. - - Args: - user_id (str): user id to test - - Returns: - Deferred[bool]: True if user 'user_type' is null or empty string - """ - res = yield self.runInteraction("is_real_user", self.is_real_user_txn, user_id) - return res - - @cachedInlineCallbacks() - def is_support_user(self, user_id): - """Determines if the user is of type UserTypes.SUPPORT - - Args: - user_id (str): user id to test - - Returns: - Deferred[bool]: True if user is of type UserTypes.SUPPORT - """ - res = yield self.runInteraction( - "is_support_user", self.is_support_user_txn, user_id - ) - return res - - def is_real_user_txn(self, txn, user_id): - res = self._simple_select_one_onecol_txn( - txn=txn, - table="users", - keyvalues={"name": user_id}, - retcol="user_type", - allow_none=True, - ) - return res is None - - def is_support_user_txn(self, txn, user_id): - res = self._simple_select_one_onecol_txn( - txn=txn, - table="users", - keyvalues={"name": user_id}, - retcol="user_type", - allow_none=True, - ) - return True if res == UserTypes.SUPPORT else False - - def get_users_by_id_case_insensitive(self, user_id): - """Gets users that match user_id case insensitively. - Returns a mapping of user_id -> password_hash. - """ - - def f(txn): - sql = ( - "SELECT name, password_hash FROM users" " WHERE lower(name) = lower(?)" - ) - txn.execute(sql, (user_id,)) - return dict(txn) - - return self.runInteraction("get_users_by_id_case_insensitive", f) - - async def get_user_by_external_id( - self, auth_provider: str, external_id: str - ) -> str: - """Look up a user by their external auth id - - Args: - auth_provider: identifier for the remote auth provider - external_id: id on that system - - Returns: - str|None: the mxid of the user, or None if they are not known - """ - return await self._simple_select_one_onecol( - table="user_external_ids", - keyvalues={"auth_provider": auth_provider, "external_id": external_id}, - retcol="user_id", - allow_none=True, - desc="get_user_by_external_id", - ) - - @defer.inlineCallbacks - def count_all_users(self): - """Counts all users registered on the homeserver.""" - - def _count_users(txn): - txn.execute("SELECT COUNT(*) AS users FROM users") - rows = self.cursor_to_dict(txn) - if rows: - return rows[0]["users"] - return 0 - - ret = yield self.runInteraction("count_users", _count_users) - return ret - - def count_daily_user_type(self): - """ - Counts 1) native non guest users - 2) native guests users - 3) bridged users - who registered on the homeserver in the past 24 hours - """ - - def _count_daily_user_type(txn): - yesterday = int(self._clock.time()) - (60 * 60 * 24) - - sql = """ - SELECT user_type, COALESCE(count(*), 0) AS count FROM ( - SELECT - CASE - WHEN is_guest=0 AND appservice_id IS NULL THEN 'native' - WHEN is_guest=1 AND appservice_id IS NULL THEN 'guest' - WHEN is_guest=0 AND appservice_id IS NOT NULL THEN 'bridged' - END AS user_type - FROM users - WHERE creation_ts > ? - ) AS t GROUP BY user_type - """ - results = {"native": 0, "guest": 0, "bridged": 0} - txn.execute(sql, (yesterday,)) - for row in txn: - results[row[0]] = row[1] - return results - - return self.runInteraction("count_daily_user_type", _count_daily_user_type) - - @defer.inlineCallbacks - def count_nonbridged_users(self): - def _count_users(txn): - txn.execute( - """ - SELECT COALESCE(COUNT(*), 0) FROM users - WHERE appservice_id IS NULL - """ - ) - count, = txn.fetchone() - return count - - ret = yield self.runInteraction("count_users", _count_users) - return ret - - @defer.inlineCallbacks - def count_real_users(self): - """Counts all users without a special user_type registered on the homeserver.""" - - def _count_users(txn): - txn.execute("SELECT COUNT(*) AS users FROM users where user_type is null") - rows = self.cursor_to_dict(txn) - if rows: - return rows[0]["users"] - return 0 - - ret = yield self.runInteraction("count_real_users", _count_users) - return ret - - @defer.inlineCallbacks - def find_next_generated_user_id_localpart(self): - """ - Gets the localpart of the next generated user ID. - - Generated user IDs are integers, and we aim for them to be as small as - we can. Unfortunately, it's possible some of them are already taken by - existing users, and there may be gaps in the already taken range. This - function returns the start of the first allocatable gap. This is to - avoid the case of ID 10000000 being pre-allocated, so us wasting the - first (and shortest) many generated user IDs. - """ - - def _find_next_generated_user_id(txn): - # We bound between '@1' and '@a' to avoid pulling the entire table - # out. - txn.execute("SELECT name FROM users WHERE '@1' <= name AND name < '@a'") - - regex = re.compile(r"^@(\d+):") - - found = set() - - for (user_id,) in txn: - match = regex.search(user_id) - if match: - found.add(int(match.group(1))) - for i in range(len(found) + 1): - if i not in found: - return i - - return ( - ( - yield self.runInteraction( - "find_next_generated_user_id", _find_next_generated_user_id - ) - ) - ) - - @defer.inlineCallbacks - def get_user_id_by_threepid(self, medium, address): - """Returns user id from threepid - - Args: - medium (str): threepid medium e.g. email - address (str): threepid address e.g. me@example.com - - Returns: - Deferred[str|None]: user id or None if no user id/threepid mapping exists - """ - user_id = yield self.runInteraction( - "get_user_id_by_threepid", self.get_user_id_by_threepid_txn, medium, address - ) - return user_id - - def get_user_id_by_threepid_txn(self, txn, medium, address): - """Returns user id from threepid - - Args: - txn (cursor): - medium (str): threepid medium e.g. email - address (str): threepid address e.g. me@example.com - - Returns: - str|None: user id or None if no user id/threepid mapping exists - """ - ret = self._simple_select_one_txn( - txn, - "user_threepids", - {"medium": medium, "address": address}, - ["user_id"], - True, - ) - if ret: - return ret["user_id"] - return None - - @defer.inlineCallbacks - def user_add_threepid(self, user_id, medium, address, validated_at, added_at): - yield self._simple_upsert( - "user_threepids", - {"medium": medium, "address": address}, - {"user_id": user_id, "validated_at": validated_at, "added_at": added_at}, - ) - - @defer.inlineCallbacks - def user_get_threepids(self, user_id): - ret = yield self._simple_select_list( - "user_threepids", - {"user_id": user_id}, - ["medium", "address", "validated_at", "added_at"], - "user_get_threepids", - ) - return ret - - def user_delete_threepid(self, user_id, medium, address): - return self._simple_delete( - "user_threepids", - keyvalues={"user_id": user_id, "medium": medium, "address": address}, - desc="user_delete_threepids", - ) - - def add_user_bound_threepid(self, user_id, medium, address, id_server): - """The server proxied a bind request to the given identity server on - behalf of the given user. We need to remember this in case the user - asks us to unbind the threepid. - - Args: - user_id (str) - medium (str) - address (str) - id_server (str) - - Returns: - Deferred - """ - # We need to use an upsert, in case they user had already bound the - # threepid - return self._simple_upsert( - table="user_threepid_id_server", - keyvalues={ - "user_id": user_id, - "medium": medium, - "address": address, - "id_server": id_server, - }, - values={}, - insertion_values={}, - desc="add_user_bound_threepid", - ) - - def user_get_bound_threepids(self, user_id): - """Get the threepids that a user has bound to an identity server through the homeserver - The homeserver remembers where binds to an identity server occurred. Using this - method can retrieve those threepids. - - Args: - user_id (str): The ID of the user to retrieve threepids for - - Returns: - Deferred[list[dict]]: List of dictionaries containing the following: - medium (str): The medium of the threepid (e.g "email") - address (str): The address of the threepid (e.g "bob@example.com") - """ - return self._simple_select_list( - table="user_threepid_id_server", - keyvalues={"user_id": user_id}, - retcols=["medium", "address"], - desc="user_get_bound_threepids", - ) - - def remove_user_bound_threepid(self, user_id, medium, address, id_server): - """The server proxied an unbind request to the given identity server on - behalf of the given user, so we remove the mapping of threepid to - identity server. - - Args: - user_id (str) - medium (str) - address (str) - id_server (str) - - Returns: - Deferred - """ - return self._simple_delete( - table="user_threepid_id_server", - keyvalues={ - "user_id": user_id, - "medium": medium, - "address": address, - "id_server": id_server, - }, - desc="remove_user_bound_threepid", - ) - - def get_id_servers_user_bound(self, user_id, medium, address): - """Get the list of identity servers that the server proxied bind - requests to for given user and threepid - - Args: - user_id (str) - medium (str) - address (str) - - Returns: - Deferred[list[str]]: Resolves to a list of identity servers - """ - return self._simple_select_onecol( - table="user_threepid_id_server", - keyvalues={"user_id": user_id, "medium": medium, "address": address}, - retcol="id_server", - desc="get_id_servers_user_bound", - ) - - @cachedInlineCallbacks() - def get_user_deactivated_status(self, user_id): - """Retrieve the value for the `deactivated` property for the provided user. - - Args: - user_id (str): The ID of the user to retrieve the status for. - - Returns: - defer.Deferred(bool): The requested value. - """ - - res = yield self._simple_select_one_onecol( - table="users", - keyvalues={"name": user_id}, - retcol="deactivated", - desc="get_user_deactivated_status", - ) - - # Convert the integer into a boolean. - return res == 1 - - def get_threepid_validation_session( - self, medium, client_secret, address=None, sid=None, validated=True - ): - """Gets a session_id and last_send_attempt (if available) for a - combination of validation metadata - - Args: - medium (str|None): The medium of the 3PID - address (str|None): The address of the 3PID - sid (str|None): The ID of the validation session - client_secret (str): A unique string provided by the client to help identify this - validation attempt - validated (bool|None): Whether sessions should be filtered by - whether they have been validated already or not. None to - perform no filtering - - Returns: - Deferred[dict|None]: A dict containing the following: - * address - address of the 3pid - * medium - medium of the 3pid - * client_secret - a secret provided by the client for this validation session - * session_id - ID of the validation session - * send_attempt - a number serving to dedupe send attempts for this session - * validated_at - timestamp of when this session was validated if so - - Otherwise None if a validation session is not found - """ - if not client_secret: - raise SynapseError( - 400, "Missing parameter: client_secret", errcode=Codes.MISSING_PARAM - ) - - keyvalues = {"client_secret": client_secret} - if medium: - keyvalues["medium"] = medium - if address: - keyvalues["address"] = address - if sid: - keyvalues["session_id"] = sid - - assert address or sid - - def get_threepid_validation_session_txn(txn): - sql = """ - SELECT address, session_id, medium, client_secret, - last_send_attempt, validated_at - FROM threepid_validation_session WHERE %s - """ % ( - " AND ".join("%s = ?" % k for k in iterkeys(keyvalues)), - ) - - if validated is not None: - sql += " AND validated_at IS " + ("NOT NULL" if validated else "NULL") - - sql += " LIMIT 1" - - txn.execute(sql, list(keyvalues.values())) - rows = self.cursor_to_dict(txn) - if not rows: - return None - - return rows[0] - - return self.runInteraction( - "get_threepid_validation_session", get_threepid_validation_session_txn - ) - - def delete_threepid_session(self, session_id): - """Removes a threepid validation session from the database. This can - be done after validation has been performed and whatever action was - waiting on it has been carried out - - Args: - session_id (str): The ID of the session to delete - """ - - def delete_threepid_session_txn(txn): - self._simple_delete_txn( - txn, - table="threepid_validation_token", - keyvalues={"session_id": session_id}, - ) - self._simple_delete_txn( - txn, - table="threepid_validation_session", - keyvalues={"session_id": session_id}, - ) - - return self.runInteraction( - "delete_threepid_session", delete_threepid_session_txn - ) - - -class RegistrationBackgroundUpdateStore( - RegistrationWorkerStore, background_updates.BackgroundUpdateStore -): - def __init__(self, db_conn, hs): - super(RegistrationBackgroundUpdateStore, self).__init__(db_conn, hs) - - self.clock = hs.get_clock() - self.config = hs.config - - self.register_background_index_update( - "access_tokens_device_index", - index_name="access_tokens_device_id", - table="access_tokens", - columns=["user_id", "device_id"], - ) - - self.register_background_index_update( - "users_creation_ts", - index_name="users_creation_ts", - table="users", - columns=["creation_ts"], - ) - - # we no longer use refresh tokens, but it's possible that some people - # might have a background update queued to build this index. Just - # clear the background update. - self.register_noop_background_update("refresh_tokens_device_index") - - self.register_background_update_handler( - "user_threepids_grandfather", self._bg_user_threepids_grandfather - ) - - self.register_background_update_handler( - "users_set_deactivated_flag", self._background_update_set_deactivated_flag - ) - - @defer.inlineCallbacks - def _background_update_set_deactivated_flag(self, progress, batch_size): - """Retrieves a list of all deactivated users and sets the 'deactivated' flag to 1 - for each of them. - """ - - last_user = progress.get("user_id", "") - - def _background_update_set_deactivated_flag_txn(txn): - txn.execute( - """ - SELECT - users.name, - COUNT(access_tokens.token) AS count_tokens, - COUNT(user_threepids.address) AS count_threepids - FROM users - LEFT JOIN access_tokens ON (access_tokens.user_id = users.name) - LEFT JOIN user_threepids ON (user_threepids.user_id = users.name) - WHERE (users.password_hash IS NULL OR users.password_hash = '') - AND (users.appservice_id IS NULL OR users.appservice_id = '') - AND users.is_guest = 0 - AND users.name > ? - GROUP BY users.name - ORDER BY users.name ASC - LIMIT ?; - """, - (last_user, batch_size), - ) - - rows = self.cursor_to_dict(txn) - - if not rows: - return True, 0 - - rows_processed_nb = 0 - - for user in rows: - if not user["count_tokens"] and not user["count_threepids"]: - self.set_user_deactivated_status_txn(txn, user["name"], True) - rows_processed_nb += 1 - - logger.info("Marked %d rows as deactivated", rows_processed_nb) - - self._background_update_progress_txn( - txn, "users_set_deactivated_flag", {"user_id": rows[-1]["name"]} - ) - - if batch_size > len(rows): - return True, len(rows) - else: - return False, len(rows) - - end, nb_processed = yield self.runInteraction( - "users_set_deactivated_flag", _background_update_set_deactivated_flag_txn - ) - - if end: - yield self._end_background_update("users_set_deactivated_flag") - - return nb_processed - - @defer.inlineCallbacks - def _bg_user_threepids_grandfather(self, progress, batch_size): - """We now track which identity servers a user binds their 3PID to, so - we need to handle the case of existing bindings where we didn't track - this. - - We do this by grandfathering in existing user threepids assuming that - they used one of the server configured trusted identity servers. - """ - id_servers = set(self.config.trusted_third_party_id_servers) - - def _bg_user_threepids_grandfather_txn(txn): - sql = """ - INSERT INTO user_threepid_id_server - (user_id, medium, address, id_server) - SELECT user_id, medium, address, ? - FROM user_threepids - """ - - txn.executemany(sql, [(id_server,) for id_server in id_servers]) - - if id_servers: - yield self.runInteraction( - "_bg_user_threepids_grandfather", _bg_user_threepids_grandfather_txn - ) - - yield self._end_background_update("user_threepids_grandfather") - - return 1 - - -class RegistrationStore(RegistrationBackgroundUpdateStore): - def __init__(self, db_conn, hs): - super(RegistrationStore, self).__init__(db_conn, hs) - - self._account_validity = hs.config.account_validity - - # Create a background job for culling expired 3PID validity tokens - def start_cull(): - # run as a background process to make sure that the database transactions - # have a logcontext to report to - return run_as_background_process( - "cull_expired_threepid_validation_tokens", - self.cull_expired_threepid_validation_tokens, - ) - - hs.get_clock().looping_call(start_cull, THIRTY_MINUTES_IN_MS) - - @defer.inlineCallbacks - def add_access_token_to_user(self, user_id, token, device_id, valid_until_ms): - """Adds an access token for the given user. - - Args: - user_id (str): The user ID. - token (str): The new access token to add. - device_id (str): ID of the device to associate with the access - token - valid_until_ms (int|None): when the token is valid until. None for - no expiry. - Raises: - StoreError if there was a problem adding this. - """ - next_id = self._access_tokens_id_gen.get_next() - - yield self._simple_insert( - "access_tokens", - { - "id": next_id, - "user_id": user_id, - "token": token, - "device_id": device_id, - "valid_until_ms": valid_until_ms, - }, - desc="add_access_token_to_user", - ) - - def register_user( - self, - user_id, - password_hash=None, - was_guest=False, - make_guest=False, - appservice_id=None, - create_profile_with_displayname=None, - admin=False, - user_type=None, - ): - """Attempts to register an account. - - Args: - user_id (str): The desired user ID to register. - password_hash (str): Optional. The password hash for this user. - was_guest (bool): Optional. Whether this is a guest account being - upgraded to a non-guest account. - make_guest (boolean): True if the the new user should be guest, - false to add a regular user account. - appservice_id (str): The ID of the appservice registering the user. - create_profile_with_displayname (unicode): Optionally create a profile for - the user, setting their displayname to the given value - admin (boolean): is an admin user? - user_type (str|None): type of user. One of the values from - api.constants.UserTypes, or None for a normal user. - - Raises: - StoreError if the user_id could not be registered. - """ - return self.runInteraction( - "register_user", - self._register_user, - user_id, - password_hash, - was_guest, - make_guest, - appservice_id, - create_profile_with_displayname, - admin, - user_type, - ) - - def _register_user( - self, - txn, - user_id, - password_hash, - was_guest, - make_guest, - appservice_id, - create_profile_with_displayname, - admin, - user_type, - ): - user_id_obj = UserID.from_string(user_id) - - now = int(self.clock.time()) - - try: - if was_guest: - # Ensure that the guest user actually exists - # ``allow_none=False`` makes this raise an exception - # if the row isn't in the database. - self._simple_select_one_txn( - txn, - "users", - keyvalues={"name": user_id, "is_guest": 1}, - retcols=("name",), - allow_none=False, - ) - - self._simple_update_one_txn( - txn, - "users", - keyvalues={"name": user_id, "is_guest": 1}, - updatevalues={ - "password_hash": password_hash, - "upgrade_ts": now, - "is_guest": 1 if make_guest else 0, - "appservice_id": appservice_id, - "admin": 1 if admin else 0, - "user_type": user_type, - }, - ) - else: - self._simple_insert_txn( - txn, - "users", - values={ - "name": user_id, - "password_hash": password_hash, - "creation_ts": now, - "is_guest": 1 if make_guest else 0, - "appservice_id": appservice_id, - "admin": 1 if admin else 0, - "user_type": user_type, - }, - ) - - except self.database_engine.module.IntegrityError: - raise StoreError(400, "User ID already taken.", errcode=Codes.USER_IN_USE) - - if self._account_validity.enabled: - self.set_expiration_date_for_user_txn(txn, user_id) - - if create_profile_with_displayname: - # set a default displayname serverside to avoid ugly race - # between auto-joins and clients trying to set displaynames - # - # *obviously* the 'profiles' table uses localpart for user_id - # while everything else uses the full mxid. - txn.execute( - "INSERT INTO profiles(user_id, displayname) VALUES (?,?)", - (user_id_obj.localpart, create_profile_with_displayname), - ) - - if self.hs.config.stats_enabled: - # we create a new completed user statistics row - - # we don't strictly need current_token since this user really can't - # have any state deltas before now (as it is a new user), but still, - # we include it for completeness. - current_token = self._get_max_stream_id_in_current_state_deltas_txn(txn) - self._update_stats_delta_txn( - txn, now, "user", user_id, {}, complete_with_stream_id=current_token - ) - - self._invalidate_cache_and_stream(txn, self.get_user_by_id, (user_id,)) - txn.call_after(self.is_guest.invalidate, (user_id,)) - - def record_user_external_id( - self, auth_provider: str, external_id: str, user_id: str - ) -> Deferred: - """Record a mapping from an external user id to a mxid - - Args: - auth_provider: identifier for the remote auth provider - external_id: id on that system - user_id: complete mxid that it is mapped to - """ - return self._simple_insert( - table="user_external_ids", - values={ - "auth_provider": auth_provider, - "external_id": external_id, - "user_id": user_id, - }, - desc="record_user_external_id", - ) - - def user_set_password_hash(self, user_id, password_hash): - """ - NB. This does *not* evict any cache because the one use for this - removes most of the entries subsequently anyway so it would be - pointless. Use flush_user separately. - """ - - def user_set_password_hash_txn(txn): - self._simple_update_one_txn( - txn, "users", {"name": user_id}, {"password_hash": password_hash} - ) - self._invalidate_cache_and_stream(txn, self.get_user_by_id, (user_id,)) - - return self.runInteraction("user_set_password_hash", user_set_password_hash_txn) - - def user_set_consent_version(self, user_id, consent_version): - """Updates the user table to record privacy policy consent - - Args: - user_id (str): full mxid of the user to update - consent_version (str): version of the policy the user has consented - to - - Raises: - StoreError(404) if user not found - """ - - def f(txn): - self._simple_update_one_txn( - txn, - table="users", - keyvalues={"name": user_id}, - updatevalues={"consent_version": consent_version}, - ) - self._invalidate_cache_and_stream(txn, self.get_user_by_id, (user_id,)) - - return self.runInteraction("user_set_consent_version", f) - - def user_set_consent_server_notice_sent(self, user_id, consent_version): - """Updates the user table to record that we have sent the user a server - notice about privacy policy consent - - Args: - user_id (str): full mxid of the user to update - consent_version (str): version of the policy we have notified the - user about - - Raises: - StoreError(404) if user not found - """ - - def f(txn): - self._simple_update_one_txn( - txn, - table="users", - keyvalues={"name": user_id}, - updatevalues={"consent_server_notice_sent": consent_version}, - ) - self._invalidate_cache_and_stream(txn, self.get_user_by_id, (user_id,)) - - return self.runInteraction("user_set_consent_server_notice_sent", f) - - def user_delete_access_tokens(self, user_id, except_token_id=None, device_id=None): - """ - Invalidate access tokens belonging to a user - - Args: - user_id (str): ID of user the tokens belong to - except_token_id (str): list of access_tokens IDs which should - *not* be deleted - device_id (str|None): ID of device the tokens are associated with. - If None, tokens associated with any device (or no device) will - be deleted - Returns: - defer.Deferred[list[str, int, str|None, int]]: a list of - (token, token id, device id) for each of the deleted tokens - """ - - def f(txn): - keyvalues = {"user_id": user_id} - if device_id is not None: - keyvalues["device_id"] = device_id - - items = keyvalues.items() - where_clause = " AND ".join(k + " = ?" for k, _ in items) - values = [v for _, v in items] - if except_token_id: - where_clause += " AND id != ?" - values.append(except_token_id) - - txn.execute( - "SELECT token, id, device_id FROM access_tokens WHERE %s" - % where_clause, - values, - ) - tokens_and_devices = [(r[0], r[1], r[2]) for r in txn] - - for token, _, _ in tokens_and_devices: - self._invalidate_cache_and_stream( - txn, self.get_user_by_access_token, (token,) - ) - - txn.execute("DELETE FROM access_tokens WHERE %s" % where_clause, values) - - return tokens_and_devices - - return self.runInteraction("user_delete_access_tokens", f) - - def delete_access_token(self, access_token): - def f(txn): - self._simple_delete_one_txn( - txn, table="access_tokens", keyvalues={"token": access_token} - ) - - self._invalidate_cache_and_stream( - txn, self.get_user_by_access_token, (access_token,) - ) - - return self.runInteraction("delete_access_token", f) - - @cachedInlineCallbacks() - def is_guest(self, user_id): - res = yield self._simple_select_one_onecol( - table="users", - keyvalues={"name": user_id}, - retcol="is_guest", - allow_none=True, - desc="is_guest", - ) - - return res if res else False - - def add_user_pending_deactivation(self, user_id): - """ - Adds a user to the table of users who need to be parted from all the rooms they're - in - """ - return self._simple_insert( - "users_pending_deactivation", - values={"user_id": user_id}, - desc="add_user_pending_deactivation", - ) - - def del_user_pending_deactivation(self, user_id): - """ - Removes the given user to the table of users who need to be parted from all the - rooms they're in, effectively marking that user as fully deactivated. - """ - # XXX: This should be simple_delete_one but we failed to put a unique index on - # the table, so somehow duplicate entries have ended up in it. - return self._simple_delete( - "users_pending_deactivation", - keyvalues={"user_id": user_id}, - desc="del_user_pending_deactivation", - ) - - def get_user_pending_deactivation(self): - """ - Gets one user from the table of users waiting to be parted from all the rooms - they're in. - """ - return self._simple_select_one_onecol( - "users_pending_deactivation", - keyvalues={}, - retcol="user_id", - allow_none=True, - desc="get_users_pending_deactivation", - ) - - def validate_threepid_session(self, session_id, client_secret, token, current_ts): - """Attempt to validate a threepid session using a token - - Args: - session_id (str): The id of a validation session - client_secret (str): A unique string provided by the client to - help identify this validation attempt - token (str): A validation token - current_ts (int): The current unix time in milliseconds. Used for - checking token expiry status - - Raises: - ThreepidValidationError: if a matching validation token was not found or has - expired - - Returns: - deferred str|None: A str representing a link to redirect the user - to if there is one. - """ - - # Insert everything into a transaction in order to run atomically - def validate_threepid_session_txn(txn): - row = self._simple_select_one_txn( - txn, - table="threepid_validation_session", - keyvalues={"session_id": session_id}, - retcols=["client_secret", "validated_at"], - allow_none=True, - ) - - if not row: - raise ThreepidValidationError(400, "Unknown session_id") - retrieved_client_secret = row["client_secret"] - validated_at = row["validated_at"] - - if retrieved_client_secret != client_secret: - raise ThreepidValidationError( - 400, "This client_secret does not match the provided session_id" - ) - - row = self._simple_select_one_txn( - txn, - table="threepid_validation_token", - keyvalues={"session_id": session_id, "token": token}, - retcols=["expires", "next_link"], - allow_none=True, - ) - - if not row: - raise ThreepidValidationError( - 400, "Validation token not found or has expired" - ) - expires = row["expires"] - next_link = row["next_link"] - - # If the session is already validated, no need to revalidate - if validated_at: - return next_link - - if expires <= current_ts: - raise ThreepidValidationError( - 400, "This token has expired. Please request a new one" - ) - - # Looks good. Validate the session - self._simple_update_txn( - txn, - table="threepid_validation_session", - keyvalues={"session_id": session_id}, - updatevalues={"validated_at": self.clock.time_msec()}, - ) - - return next_link - - # Return next_link if it exists - return self.runInteraction( - "validate_threepid_session_txn", validate_threepid_session_txn - ) - - def upsert_threepid_validation_session( - self, - medium, - address, - client_secret, - send_attempt, - session_id, - validated_at=None, - ): - """Upsert a threepid validation session - Args: - medium (str): The medium of the 3PID - address (str): The address of the 3PID - client_secret (str): A unique string provided by the client to - help identify this validation attempt - send_attempt (int): The latest send_attempt on this session - session_id (str): The id of this validation session - validated_at (int|None): The unix timestamp in milliseconds of - when the session was marked as valid - """ - insertion_values = { - "medium": medium, - "address": address, - "client_secret": client_secret, - } - - if validated_at: - insertion_values["validated_at"] = validated_at - - return self._simple_upsert( - table="threepid_validation_session", - keyvalues={"session_id": session_id}, - values={"last_send_attempt": send_attempt}, - insertion_values=insertion_values, - desc="upsert_threepid_validation_session", - ) - - def start_or_continue_validation_session( - self, - medium, - address, - session_id, - client_secret, - send_attempt, - next_link, - token, - token_expires, - ): - """Creates a new threepid validation session if it does not already - exist and associates a new validation token with it - - Args: - medium (str): The medium of the 3PID - address (str): The address of the 3PID - session_id (str): The id of this validation session - client_secret (str): A unique string provided by the client to - help identify this validation attempt - send_attempt (int): The latest send_attempt on this session - next_link (str|None): The link to redirect the user to upon - successful validation - token (str): The validation token - token_expires (int): The timestamp for which after the token - will no longer be valid - """ - - def start_or_continue_validation_session_txn(txn): - # Create or update a validation session - self._simple_upsert_txn( - txn, - table="threepid_validation_session", - keyvalues={"session_id": session_id}, - values={"last_send_attempt": send_attempt}, - insertion_values={ - "medium": medium, - "address": address, - "client_secret": client_secret, - }, - ) - - # Create a new validation token with this session ID - self._simple_insert_txn( - txn, - table="threepid_validation_token", - values={ - "session_id": session_id, - "token": token, - "next_link": next_link, - "expires": token_expires, - }, - ) - - return self.runInteraction( - "start_or_continue_validation_session", - start_or_continue_validation_session_txn, - ) - - def cull_expired_threepid_validation_tokens(self): - """Remove threepid validation tokens with expiry dates that have passed""" - - def cull_expired_threepid_validation_tokens_txn(txn, ts): - sql = """ - DELETE FROM threepid_validation_token WHERE - expires < ? - """ - return txn.execute(sql, (ts,)) - - return self.runInteraction( - "cull_expired_threepid_validation_tokens", - cull_expired_threepid_validation_tokens_txn, - self.clock.time_msec(), - ) - - @defer.inlineCallbacks - def set_user_deactivated_status(self, user_id, deactivated): - """Set the `deactivated` property for the provided user to the provided value. - - Args: - user_id (str): The ID of the user to set the status for. - deactivated (bool): The value to set for `deactivated`. - """ - - yield self.runInteraction( - "set_user_deactivated_status", - self.set_user_deactivated_status_txn, - user_id, - deactivated, - ) - - def set_user_deactivated_status_txn(self, txn, user_id, deactivated): - self._simple_update_one_txn( - txn=txn, - table="users", - keyvalues={"name": user_id}, - updatevalues={"deactivated": 1 if deactivated else 0}, - ) - self._invalidate_cache_and_stream( - txn, self.get_user_deactivated_status, (user_id,) - ) diff --git a/synapse/storage/rejections.py b/synapse/storage/rejections.py deleted file mode 100644 index f4c1c2a457..0000000000 --- a/synapse/storage/rejections.py +++ /dev/null @@ -1,42 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2014-2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import logging - -from ._base import SQLBaseStore - -logger = logging.getLogger(__name__) - - -class RejectionsStore(SQLBaseStore): - def _store_rejections_txn(self, txn, event_id, reason): - self._simple_insert_txn( - txn, - table="rejections", - values={ - "event_id": event_id, - "reason": reason, - "last_check": self._clock.time_msec(), - }, - ) - - def get_rejection_reason(self, event_id): - return self._simple_select_one_onecol( - table="rejections", - retcol="reason", - keyvalues={"event_id": event_id}, - allow_none=True, - desc="get_rejection_reason", - ) diff --git a/synapse/storage/relations.py b/synapse/storage/relations.py index fcb5f2f23a..d471ec9860 100644 --- a/synapse/storage/relations.py +++ b/synapse/storage/relations.py @@ -17,11 +17,7 @@ import logging import attr -from synapse.api.constants import RelationTypes from synapse.api.errors import SynapseError -from synapse.storage._base import SQLBaseStore -from synapse.storage.stream import generate_pagination_where_clause -from synapse.util.caches.descriptors import cached, cachedInlineCallbacks logger = logging.getLogger(__name__) @@ -113,358 +109,3 @@ class AggregationPaginationToken(object): def as_tuple(self): return attr.astuple(self) - - -class RelationsWorkerStore(SQLBaseStore): - @cached(tree=True) - def get_relations_for_event( - self, - event_id, - relation_type=None, - event_type=None, - aggregation_key=None, - limit=5, - direction="b", - from_token=None, - to_token=None, - ): - """Get a list of relations for an event, ordered by topological ordering. - - Args: - event_id (str): Fetch events that relate to this event ID. - relation_type (str|None): Only fetch events with this relation - type, if given. - event_type (str|None): Only fetch events with this event type, if - given. - aggregation_key (str|None): Only fetch events with this aggregation - key, if given. - limit (int): Only fetch the most recent `limit` events. - direction (str): Whether to fetch the most recent first (`"b"`) or - the oldest first (`"f"`). - from_token (RelationPaginationToken|None): Fetch rows from the given - token, or from the start if None. - to_token (RelationPaginationToken|None): Fetch rows up to the given - token, or up to the end if None. - - Returns: - Deferred[PaginationChunk]: List of event IDs that match relations - requested. The rows are of the form `{"event_id": "..."}`. - """ - - where_clause = ["relates_to_id = ?"] - where_args = [event_id] - - if relation_type is not None: - where_clause.append("relation_type = ?") - where_args.append(relation_type) - - if event_type is not None: - where_clause.append("type = ?") - where_args.append(event_type) - - if aggregation_key: - where_clause.append("aggregation_key = ?") - where_args.append(aggregation_key) - - pagination_clause = generate_pagination_where_clause( - direction=direction, - column_names=("topological_ordering", "stream_ordering"), - from_token=attr.astuple(from_token) if from_token else None, - to_token=attr.astuple(to_token) if to_token else None, - engine=self.database_engine, - ) - - if pagination_clause: - where_clause.append(pagination_clause) - - if direction == "b": - order = "DESC" - else: - order = "ASC" - - sql = """ - SELECT event_id, topological_ordering, stream_ordering - FROM event_relations - INNER JOIN events USING (event_id) - WHERE %s - ORDER BY topological_ordering %s, stream_ordering %s - LIMIT ? - """ % ( - " AND ".join(where_clause), - order, - order, - ) - - def _get_recent_references_for_event_txn(txn): - txn.execute(sql, where_args + [limit + 1]) - - last_topo_id = None - last_stream_id = None - events = [] - for row in txn: - events.append({"event_id": row[0]}) - last_topo_id = row[1] - last_stream_id = row[2] - - next_batch = None - if len(events) > limit and last_topo_id and last_stream_id: - next_batch = RelationPaginationToken(last_topo_id, last_stream_id) - - return PaginationChunk( - chunk=list(events[:limit]), next_batch=next_batch, prev_batch=from_token - ) - - return self.runInteraction( - "get_recent_references_for_event", _get_recent_references_for_event_txn - ) - - @cached(tree=True) - def get_aggregation_groups_for_event( - self, - event_id, - event_type=None, - limit=5, - direction="b", - from_token=None, - to_token=None, - ): - """Get a list of annotations on the event, grouped by event type and - aggregation key, sorted by count. - - This is used e.g. to get the what and how many reactions have happend - on an event. - - Args: - event_id (str): Fetch events that relate to this event ID. - event_type (str|None): Only fetch events with this event type, if - given. - limit (int): Only fetch the `limit` groups. - direction (str): Whether to fetch the highest count first (`"b"`) or - the lowest count first (`"f"`). - from_token (AggregationPaginationToken|None): Fetch rows from the - given token, or from the start if None. - to_token (AggregationPaginationToken|None): Fetch rows up to the - given token, or up to the end if None. - - - Returns: - Deferred[PaginationChunk]: List of groups of annotations that - match. Each row is a dict with `type`, `key` and `count` fields. - """ - - where_clause = ["relates_to_id = ?", "relation_type = ?"] - where_args = [event_id, RelationTypes.ANNOTATION] - - if event_type: - where_clause.append("type = ?") - where_args.append(event_type) - - having_clause = generate_pagination_where_clause( - direction=direction, - column_names=("COUNT(*)", "MAX(stream_ordering)"), - from_token=attr.astuple(from_token) if from_token else None, - to_token=attr.astuple(to_token) if to_token else None, - engine=self.database_engine, - ) - - if direction == "b": - order = "DESC" - else: - order = "ASC" - - if having_clause: - having_clause = "HAVING " + having_clause - else: - having_clause = "" - - sql = """ - SELECT type, aggregation_key, COUNT(DISTINCT sender), MAX(stream_ordering) - FROM event_relations - INNER JOIN events USING (event_id) - WHERE {where_clause} - GROUP BY relation_type, type, aggregation_key - {having_clause} - ORDER BY COUNT(*) {order}, MAX(stream_ordering) {order} - LIMIT ? - """.format( - where_clause=" AND ".join(where_clause), - order=order, - having_clause=having_clause, - ) - - def _get_aggregation_groups_for_event_txn(txn): - txn.execute(sql, where_args + [limit + 1]) - - next_batch = None - events = [] - for row in txn: - events.append({"type": row[0], "key": row[1], "count": row[2]}) - next_batch = AggregationPaginationToken(row[2], row[3]) - - if len(events) <= limit: - next_batch = None - - return PaginationChunk( - chunk=list(events[:limit]), next_batch=next_batch, prev_batch=from_token - ) - - return self.runInteraction( - "get_aggregation_groups_for_event", _get_aggregation_groups_for_event_txn - ) - - @cachedInlineCallbacks() - def get_applicable_edit(self, event_id): - """Get the most recent edit (if any) that has happened for the given - event. - - Correctly handles checking whether edits were allowed to happen. - - Args: - event_id (str): The original event ID - - Returns: - Deferred[EventBase|None]: Returns the most recent edit, if any. - """ - - # We only allow edits for `m.room.message` events that have the same sender - # and event type. We can't assert these things during regular event auth so - # we have to do the checks post hoc. - - # Fetches latest edit that has the same type and sender as the - # original, and is an `m.room.message`. - sql = """ - SELECT edit.event_id FROM events AS edit - INNER JOIN event_relations USING (event_id) - INNER JOIN events AS original ON - original.event_id = relates_to_id - AND edit.type = original.type - AND edit.sender = original.sender - WHERE - relates_to_id = ? - AND relation_type = ? - AND edit.type = 'm.room.message' - ORDER by edit.origin_server_ts DESC, edit.event_id DESC - LIMIT 1 - """ - - def _get_applicable_edit_txn(txn): - txn.execute(sql, (event_id, RelationTypes.REPLACE)) - row = txn.fetchone() - if row: - return row[0] - - edit_id = yield self.runInteraction( - "get_applicable_edit", _get_applicable_edit_txn - ) - - if not edit_id: - return - - edit_event = yield self.get_event(edit_id, allow_none=True) - return edit_event - - def has_user_annotated_event(self, parent_id, event_type, aggregation_key, sender): - """Check if a user has already annotated an event with the same key - (e.g. already liked an event). - - Args: - parent_id (str): The event being annotated - event_type (str): The event type of the annotation - aggregation_key (str): The aggregation key of the annotation - sender (str): The sender of the annotation - - Returns: - Deferred[bool] - """ - - sql = """ - SELECT 1 FROM event_relations - INNER JOIN events USING (event_id) - WHERE - relates_to_id = ? - AND relation_type = ? - AND type = ? - AND sender = ? - AND aggregation_key = ? - LIMIT 1; - """ - - def _get_if_user_has_annotated_event(txn): - txn.execute( - sql, - ( - parent_id, - RelationTypes.ANNOTATION, - event_type, - sender, - aggregation_key, - ), - ) - - return bool(txn.fetchone()) - - return self.runInteraction( - "get_if_user_has_annotated_event", _get_if_user_has_annotated_event - ) - - -class RelationsStore(RelationsWorkerStore): - def _handle_event_relations(self, txn, event): - """Handles inserting relation data during peristence of events - - Args: - txn - event (EventBase) - """ - relation = event.content.get("m.relates_to") - if not relation: - # No relations - return - - rel_type = relation.get("rel_type") - if rel_type not in ( - RelationTypes.ANNOTATION, - RelationTypes.REFERENCE, - RelationTypes.REPLACE, - ): - # Unknown relation type - return - - parent_id = relation.get("event_id") - if not parent_id: - # Invalid relation - return - - aggregation_key = relation.get("key") - - self._simple_insert_txn( - txn, - table="event_relations", - values={ - "event_id": event.event_id, - "relates_to_id": parent_id, - "relation_type": rel_type, - "aggregation_key": aggregation_key, - }, - ) - - txn.call_after(self.get_relations_for_event.invalidate_many, (parent_id,)) - txn.call_after( - self.get_aggregation_groups_for_event.invalidate_many, (parent_id,) - ) - - if rel_type == RelationTypes.REPLACE: - txn.call_after(self.get_applicable_edit.invalidate, (parent_id,)) - - def _handle_redaction(self, txn, redacted_event_id): - """Handles receiving a redaction and checking whether we need to remove - any redacted relations from the database. - - Args: - txn - redacted_event_id (str): The event that was redacted. - """ - - self._simple_delete_txn( - txn, table="event_relations", keyvalues={"event_id": redacted_event_id} - ) diff --git a/synapse/storage/room.py b/synapse/storage/room.py deleted file mode 100644 index 43cc56fa6f..0000000000 --- a/synapse/storage/room.py +++ /dev/null @@ -1,681 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2014-2016 OpenMarket Ltd -# Copyright 2019 The Matrix.org Foundation C.I.C. -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import collections -import logging -import re -from typing import Optional, Tuple - -from canonicaljson import json - -from twisted.internet import defer - -from synapse.api.errors import StoreError -from synapse.storage._base import SQLBaseStore -from synapse.storage.search import SearchStore -from synapse.types import ThirdPartyInstanceID -from synapse.util.caches.descriptors import cached, cachedInlineCallbacks - -logger = logging.getLogger(__name__) - - -OpsLevel = collections.namedtuple( - "OpsLevel", ("ban_level", "kick_level", "redact_level") -) - -RatelimitOverride = collections.namedtuple( - "RatelimitOverride", ("messages_per_second", "burst_count") -) - - -class RoomWorkerStore(SQLBaseStore): - def get_room(self, room_id): - """Retrieve a room. - - Args: - room_id (str): The ID of the room to retrieve. - Returns: - A dict containing the room information, or None if the room is unknown. - """ - return self._simple_select_one( - table="rooms", - keyvalues={"room_id": room_id}, - retcols=("room_id", "is_public", "creator"), - desc="get_room", - allow_none=True, - ) - - def get_public_room_ids(self): - return self._simple_select_onecol( - table="rooms", - keyvalues={"is_public": True}, - retcol="room_id", - desc="get_public_room_ids", - ) - - def count_public_rooms(self, network_tuple, ignore_non_federatable): - """Counts the number of public rooms as tracked in the room_stats_current - and room_stats_state table. - - Args: - network_tuple (ThirdPartyInstanceID|None) - ignore_non_federatable (bool): If true filters out non-federatable rooms - """ - - def _count_public_rooms_txn(txn): - query_args = [] - - if network_tuple: - if network_tuple.appservice_id: - published_sql = """ - SELECT room_id from appservice_room_list - WHERE appservice_id = ? AND network_id = ? - """ - query_args.append(network_tuple.appservice_id) - query_args.append(network_tuple.network_id) - else: - published_sql = """ - SELECT room_id FROM rooms WHERE is_public - """ - else: - published_sql = """ - SELECT room_id FROM rooms WHERE is_public - UNION SELECT room_id from appservice_room_list - """ - - sql = """ - SELECT - COALESCE(COUNT(*), 0) - FROM ( - %(published_sql)s - ) published - INNER JOIN room_stats_state USING (room_id) - INNER JOIN room_stats_current USING (room_id) - WHERE - ( - join_rules = 'public' OR history_visibility = 'world_readable' - ) - AND joined_members > 0 - """ % { - "published_sql": published_sql - } - - txn.execute(sql, query_args) - return txn.fetchone()[0] - - return self.runInteraction("count_public_rooms", _count_public_rooms_txn) - - @defer.inlineCallbacks - def get_largest_public_rooms( - self, - network_tuple: Optional[ThirdPartyInstanceID], - search_filter: Optional[dict], - limit: Optional[int], - bounds: Optional[Tuple[int, str]], - forwards: bool, - ignore_non_federatable: bool = False, - ): - """Gets the largest public rooms (where largest is in terms of joined - members, as tracked in the statistics table). - - Args: - network_tuple - search_filter - limit: Maxmimum number of rows to return, unlimited otherwise. - bounds: An uppoer or lower bound to apply to result set if given, - consists of a joined member count and room_id (these are - excluded from result set). - forwards: true iff going forwards, going backwards otherwise - ignore_non_federatable: If true filters out non-federatable rooms. - - Returns: - Rooms in order: biggest number of joined users first. - We then arbitrarily use the room_id as a tie breaker. - - """ - - where_clauses = [] - query_args = [] - - if network_tuple: - if network_tuple.appservice_id: - published_sql = """ - SELECT room_id from appservice_room_list - WHERE appservice_id = ? AND network_id = ? - """ - query_args.append(network_tuple.appservice_id) - query_args.append(network_tuple.network_id) - else: - published_sql = """ - SELECT room_id FROM rooms WHERE is_public - """ - else: - published_sql = """ - SELECT room_id FROM rooms WHERE is_public - UNION SELECT room_id from appservice_room_list - """ - - # Work out the bounds if we're given them, these bounds look slightly - # odd, but are designed to help query planner use indices by pulling - # out a common bound. - if bounds: - last_joined_members, last_room_id = bounds - if forwards: - where_clauses.append( - """ - joined_members <= ? AND ( - joined_members < ? OR room_id < ? - ) - """ - ) - else: - where_clauses.append( - """ - joined_members >= ? AND ( - joined_members > ? OR room_id > ? - ) - """ - ) - - query_args += [last_joined_members, last_joined_members, last_room_id] - - if ignore_non_federatable: - where_clauses.append("is_federatable") - - if search_filter and search_filter.get("generic_search_term", None): - search_term = "%" + search_filter["generic_search_term"] + "%" - - where_clauses.append( - """ - ( - name LIKE ? - OR topic LIKE ? - OR canonical_alias LIKE ? - ) - """ - ) - query_args += [search_term, search_term, search_term] - - where_clause = "" - if where_clauses: - where_clause = " AND " + " AND ".join(where_clauses) - - sql = """ - SELECT - room_id, name, topic, canonical_alias, joined_members, - avatar, history_visibility, joined_members, guest_access - FROM ( - %(published_sql)s - ) published - INNER JOIN room_stats_state USING (room_id) - INNER JOIN room_stats_current USING (room_id) - WHERE - ( - join_rules = 'public' OR history_visibility = 'world_readable' - ) - AND joined_members > 0 - %(where_clause)s - ORDER BY joined_members %(dir)s, room_id %(dir)s - """ % { - "published_sql": published_sql, - "where_clause": where_clause, - "dir": "DESC" if forwards else "ASC", - } - - if limit is not None: - query_args.append(limit) - - sql += """ - LIMIT ? - """ - - def _get_largest_public_rooms_txn(txn): - txn.execute(sql, query_args) - - results = self.cursor_to_dict(txn) - - if not forwards: - results.reverse() - - return results - - ret_val = yield self.runInteraction( - "get_largest_public_rooms", _get_largest_public_rooms_txn - ) - defer.returnValue(ret_val) - - @cached(max_entries=10000) - def is_room_blocked(self, room_id): - return self._simple_select_one_onecol( - table="blocked_rooms", - keyvalues={"room_id": room_id}, - retcol="1", - allow_none=True, - desc="is_room_blocked", - ) - - @cachedInlineCallbacks(max_entries=10000) - def get_ratelimit_for_user(self, user_id): - """Check if there are any overrides for ratelimiting for the given - user - - Args: - user_id (str) - - Returns: - RatelimitOverride if there is an override, else None. If the contents - of RatelimitOverride are None or 0 then ratelimitng has been - disabled for that user entirely. - """ - row = yield self._simple_select_one( - table="ratelimit_override", - keyvalues={"user_id": user_id}, - retcols=("messages_per_second", "burst_count"), - allow_none=True, - desc="get_ratelimit_for_user", - ) - - if row: - return RatelimitOverride( - messages_per_second=row["messages_per_second"], - burst_count=row["burst_count"], - ) - else: - return None - - -class RoomStore(RoomWorkerStore, SearchStore): - @defer.inlineCallbacks - def store_room(self, room_id, room_creator_user_id, is_public): - """Stores a room. - - Args: - room_id (str): The desired room ID, can be None. - room_creator_user_id (str): The user ID of the room creator. - is_public (bool): True to indicate that this room should appear in - public room lists. - Raises: - StoreError if the room could not be stored. - """ - try: - - def store_room_txn(txn, next_id): - self._simple_insert_txn( - txn, - "rooms", - { - "room_id": room_id, - "creator": room_creator_user_id, - "is_public": is_public, - }, - ) - if is_public: - self._simple_insert_txn( - txn, - table="public_room_list_stream", - values={ - "stream_id": next_id, - "room_id": room_id, - "visibility": is_public, - }, - ) - - with self._public_room_id_gen.get_next() as next_id: - yield self.runInteraction("store_room_txn", store_room_txn, next_id) - except Exception as e: - logger.error("store_room with room_id=%s failed: %s", room_id, e) - raise StoreError(500, "Problem creating room.") - - @defer.inlineCallbacks - def set_room_is_public(self, room_id, is_public): - def set_room_is_public_txn(txn, next_id): - self._simple_update_one_txn( - txn, - table="rooms", - keyvalues={"room_id": room_id}, - updatevalues={"is_public": is_public}, - ) - - entries = self._simple_select_list_txn( - txn, - table="public_room_list_stream", - keyvalues={ - "room_id": room_id, - "appservice_id": None, - "network_id": None, - }, - retcols=("stream_id", "visibility"), - ) - - entries.sort(key=lambda r: r["stream_id"]) - - add_to_stream = True - if entries: - add_to_stream = bool(entries[-1]["visibility"]) != is_public - - if add_to_stream: - self._simple_insert_txn( - txn, - table="public_room_list_stream", - values={ - "stream_id": next_id, - "room_id": room_id, - "visibility": is_public, - "appservice_id": None, - "network_id": None, - }, - ) - - with self._public_room_id_gen.get_next() as next_id: - yield self.runInteraction( - "set_room_is_public", set_room_is_public_txn, next_id - ) - self.hs.get_notifier().on_new_replication_data() - - @defer.inlineCallbacks - def set_room_is_public_appservice( - self, room_id, appservice_id, network_id, is_public - ): - """Edit the appservice/network specific public room list. - - Each appservice can have a number of published room lists associated - with them, keyed off of an appservice defined `network_id`, which - basically represents a single instance of a bridge to a third party - network. - - Args: - room_id (str) - appservice_id (str) - network_id (str) - is_public (bool): Whether to publish or unpublish the room from the - list. - """ - - def set_room_is_public_appservice_txn(txn, next_id): - if is_public: - try: - self._simple_insert_txn( - txn, - table="appservice_room_list", - values={ - "appservice_id": appservice_id, - "network_id": network_id, - "room_id": room_id, - }, - ) - except self.database_engine.module.IntegrityError: - # We've already inserted, nothing to do. - return - else: - self._simple_delete_txn( - txn, - table="appservice_room_list", - keyvalues={ - "appservice_id": appservice_id, - "network_id": network_id, - "room_id": room_id, - }, - ) - - entries = self._simple_select_list_txn( - txn, - table="public_room_list_stream", - keyvalues={ - "room_id": room_id, - "appservice_id": appservice_id, - "network_id": network_id, - }, - retcols=("stream_id", "visibility"), - ) - - entries.sort(key=lambda r: r["stream_id"]) - - add_to_stream = True - if entries: - add_to_stream = bool(entries[-1]["visibility"]) != is_public - - if add_to_stream: - self._simple_insert_txn( - txn, - table="public_room_list_stream", - values={ - "stream_id": next_id, - "room_id": room_id, - "visibility": is_public, - "appservice_id": appservice_id, - "network_id": network_id, - }, - ) - - with self._public_room_id_gen.get_next() as next_id: - yield self.runInteraction( - "set_room_is_public_appservice", - set_room_is_public_appservice_txn, - next_id, - ) - self.hs.get_notifier().on_new_replication_data() - - def get_room_count(self): - """Retrieve a list of all rooms - """ - - def f(txn): - sql = "SELECT count(*) FROM rooms" - txn.execute(sql) - row = txn.fetchone() - return row[0] or 0 - - return self.runInteraction("get_rooms", f) - - def _store_room_topic_txn(self, txn, event): - if hasattr(event, "content") and "topic" in event.content: - self.store_event_search_txn( - txn, event, "content.topic", event.content["topic"] - ) - - def _store_room_name_txn(self, txn, event): - if hasattr(event, "content") and "name" in event.content: - self.store_event_search_txn( - txn, event, "content.name", event.content["name"] - ) - - def _store_room_message_txn(self, txn, event): - if hasattr(event, "content") and "body" in event.content: - self.store_event_search_txn( - txn, event, "content.body", event.content["body"] - ) - - def add_event_report( - self, room_id, event_id, user_id, reason, content, received_ts - ): - next_id = self._event_reports_id_gen.get_next() - return self._simple_insert( - table="event_reports", - values={ - "id": next_id, - "received_ts": received_ts, - "room_id": room_id, - "event_id": event_id, - "user_id": user_id, - "reason": reason, - "content": json.dumps(content), - }, - desc="add_event_report", - ) - - def get_current_public_room_stream_id(self): - return self._public_room_id_gen.get_current_token() - - def get_all_new_public_rooms(self, prev_id, current_id, limit): - def get_all_new_public_rooms(txn): - sql = """ - SELECT stream_id, room_id, visibility, appservice_id, network_id - FROM public_room_list_stream - WHERE stream_id > ? AND stream_id <= ? - ORDER BY stream_id ASC - LIMIT ? - """ - - txn.execute(sql, (prev_id, current_id, limit)) - return txn.fetchall() - - if prev_id == current_id: - return defer.succeed([]) - - return self.runInteraction("get_all_new_public_rooms", get_all_new_public_rooms) - - @defer.inlineCallbacks - def block_room(self, room_id, user_id): - """Marks the room as blocked. Can be called multiple times. - - Args: - room_id (str): Room to block - user_id (str): Who blocked it - - Returns: - Deferred - """ - yield self._simple_upsert( - table="blocked_rooms", - keyvalues={"room_id": room_id}, - values={}, - insertion_values={"user_id": user_id}, - desc="block_room", - ) - yield self.runInteraction( - "block_room_invalidation", - self._invalidate_cache_and_stream, - self.is_room_blocked, - (room_id,), - ) - - def get_media_mxcs_in_room(self, room_id): - """Retrieves all the local and remote media MXC URIs in a given room - - Args: - room_id (str) - - Returns: - The local and remote media as a lists of tuples where the key is - the hostname and the value is the media ID. - """ - - def _get_media_mxcs_in_room_txn(txn): - local_mxcs, remote_mxcs = self._get_media_mxcs_in_room_txn(txn, room_id) - local_media_mxcs = [] - remote_media_mxcs = [] - - # Convert the IDs to MXC URIs - for media_id in local_mxcs: - local_media_mxcs.append("mxc://%s/%s" % (self.hs.hostname, media_id)) - for hostname, media_id in remote_mxcs: - remote_media_mxcs.append("mxc://%s/%s" % (hostname, media_id)) - - return local_media_mxcs, remote_media_mxcs - - return self.runInteraction("get_media_ids_in_room", _get_media_mxcs_in_room_txn) - - def quarantine_media_ids_in_room(self, room_id, quarantined_by): - """For a room loops through all events with media and quarantines - the associated media - """ - - def _quarantine_media_in_room_txn(txn): - local_mxcs, remote_mxcs = self._get_media_mxcs_in_room_txn(txn, room_id) - total_media_quarantined = 0 - - # Now update all the tables to set the quarantined_by flag - - txn.executemany( - """ - UPDATE local_media_repository - SET quarantined_by = ? - WHERE media_id = ? - """, - ((quarantined_by, media_id) for media_id in local_mxcs), - ) - - txn.executemany( - """ - UPDATE remote_media_cache - SET quarantined_by = ? - WHERE media_origin = ? AND media_id = ? - """, - ( - (quarantined_by, origin, media_id) - for origin, media_id in remote_mxcs - ), - ) - - total_media_quarantined += len(local_mxcs) - total_media_quarantined += len(remote_mxcs) - - return total_media_quarantined - - return self.runInteraction( - "quarantine_media_in_room", _quarantine_media_in_room_txn - ) - - def _get_media_mxcs_in_room_txn(self, txn, room_id): - """Retrieves all the local and remote media MXC URIs in a given room - - Args: - txn (cursor) - room_id (str) - - Returns: - The local and remote media as a lists of tuples where the key is - the hostname and the value is the media ID. - """ - mxc_re = re.compile("^mxc://([^/]+)/([^/#?]+)") - - next_token = self.get_current_events_token() + 1 - local_media_mxcs = [] - remote_media_mxcs = [] - - while next_token: - sql = """ - SELECT stream_ordering, json FROM events - JOIN event_json USING (room_id, event_id) - WHERE room_id = ? - AND stream_ordering < ? - AND contains_url = ? AND outlier = ? - ORDER BY stream_ordering DESC - LIMIT ? - """ - txn.execute(sql, (room_id, next_token, True, False, 100)) - - next_token = None - for stream_ordering, content_json in txn: - next_token = stream_ordering - event_json = json.loads(content_json) - content = event_json["content"] - content_url = content.get("url") - thumbnail_url = content.get("info", {}).get("thumbnail_url") - - for url in (content_url, thumbnail_url): - if not url: - continue - matches = mxc_re.match(url) - if matches: - hostname = matches.group(1) - media_id = matches.group(2) - if hostname == self.hs.hostname: - local_media_mxcs.append(media_id) - else: - remote_media_mxcs.append((hostname, media_id)) - - return local_media_mxcs, remote_media_mxcs diff --git a/synapse/storage/roommember.py b/synapse/storage/roommember.py index ff63487823..8c4a83a840 100644 --- a/synapse/storage/roommember.py +++ b/synapse/storage/roommember.py @@ -17,26 +17,6 @@ import logging from collections import namedtuple -from six import iteritems, itervalues - -from canonicaljson import json - -from twisted.internet import defer - -from synapse.api.constants import EventTypes, Membership -from synapse.metrics import LaterGauge -from synapse.metrics.background_process_metrics import run_as_background_process -from synapse.storage._base import LoggingTransaction, make_in_list_sql_clause -from synapse.storage.background_updates import BackgroundUpdateStore -from synapse.storage.engines import Sqlite3Engine -from synapse.storage.events_worker import EventsWorkerStore -from synapse.types import get_domain_from_id -from synapse.util.async_helpers import Linearizer -from synapse.util.caches import intern_string -from synapse.util.caches.descriptors import cached, cachedInlineCallbacks, cachedList -from synapse.util.metrics import Measure -from synapse.util.stringutils import to_ascii - logger = logging.getLogger(__name__) @@ -57,1102 +37,3 @@ ProfileInfo = namedtuple("ProfileInfo", ("avatar_url", "display_name")) # a given membership type, suitable for use in calculating heroes for a room. # "count" points to the total numberr of users of a given membership type. MemberSummary = namedtuple("MemberSummary", ("members", "count")) - -_MEMBERSHIP_PROFILE_UPDATE_NAME = "room_membership_profile_update" -_CURRENT_STATE_MEMBERSHIP_UPDATE_NAME = "current_state_events_membership" - - -class RoomMemberWorkerStore(EventsWorkerStore): - def __init__(self, db_conn, hs): - super(RoomMemberWorkerStore, self).__init__(db_conn, hs) - - # Is the current_state_events.membership up to date? Or is the - # background update still running? - self._current_state_events_membership_up_to_date = False - - txn = LoggingTransaction( - db_conn.cursor(), - name="_check_safe_current_state_events_membership_updated", - database_engine=self.database_engine, - ) - self._check_safe_current_state_events_membership_updated_txn(txn) - txn.close() - - if self.hs.config.metrics_flags.known_servers: - self._known_servers_count = 1 - self.hs.get_clock().looping_call( - run_as_background_process, - 60 * 1000, - "_count_known_servers", - self._count_known_servers, - ) - self.hs.get_clock().call_later( - 1000, - run_as_background_process, - "_count_known_servers", - self._count_known_servers, - ) - LaterGauge( - "synapse_federation_known_servers", - "", - [], - lambda: self._known_servers_count, - ) - - @defer.inlineCallbacks - def _count_known_servers(self): - """ - Count the servers that this server knows about. - - The statistic is stored on the class for the - `synapse_federation_known_servers` LaterGauge to collect. - """ - - def _transact(txn): - if isinstance(self.database_engine, Sqlite3Engine): - query = """ - SELECT COUNT(DISTINCT substr(out.user_id, pos+1)) - FROM ( - SELECT rm.user_id as user_id, instr(rm.user_id, ':') - AS pos FROM room_memberships as rm - INNER JOIN current_state_events as c ON rm.event_id = c.event_id - WHERE c.type = 'm.room.member' - ) as out - """ - else: - query = """ - SELECT COUNT(DISTINCT split_part(state_key, ':', 2)) - FROM current_state_events - WHERE type = 'm.room.member' AND membership = 'join'; - """ - txn.execute(query) - return list(txn)[0][0] - - count = yield self.runInteraction("get_known_servers", _transact) - - # We always know about ourselves, even if we have nothing in - # room_memberships (for example, the server is new). - self._known_servers_count = max([count, 1]) - return self._known_servers_count - - def _check_safe_current_state_events_membership_updated_txn(self, txn): - """Checks if it is safe to assume the new current_state_events - membership column is up to date - """ - - pending_update = self._simple_select_one_txn( - txn, - table="background_updates", - keyvalues={"update_name": _CURRENT_STATE_MEMBERSHIP_UPDATE_NAME}, - retcols=["update_name"], - allow_none=True, - ) - - self._current_state_events_membership_up_to_date = not pending_update - - # If the update is still running, reschedule to run. - if pending_update: - self._clock.call_later( - 15.0, - run_as_background_process, - "_check_safe_current_state_events_membership_updated", - self.runInteraction, - "_check_safe_current_state_events_membership_updated", - self._check_safe_current_state_events_membership_updated_txn, - ) - - @cachedInlineCallbacks(max_entries=100000, iterable=True, cache_context=True) - def get_hosts_in_room(self, room_id, cache_context): - """Returns the set of all hosts currently in the room - """ - user_ids = yield self.get_users_in_room( - room_id, on_invalidate=cache_context.invalidate - ) - hosts = frozenset(get_domain_from_id(user_id) for user_id in user_ids) - return hosts - - @cached(max_entries=100000, iterable=True) - def get_users_in_room(self, room_id): - return self.runInteraction( - "get_users_in_room", self.get_users_in_room_txn, room_id - ) - - def get_users_in_room_txn(self, txn, room_id): - # If we can assume current_state_events.membership is up to date - # then we can avoid a join, which is a Very Good Thing given how - # frequently this function gets called. - if self._current_state_events_membership_up_to_date: - sql = """ - SELECT state_key FROM current_state_events - WHERE type = 'm.room.member' AND room_id = ? AND membership = ? - """ - else: - sql = """ - SELECT state_key FROM room_memberships as m - INNER JOIN current_state_events as c - ON m.event_id = c.event_id - AND m.room_id = c.room_id - AND m.user_id = c.state_key - WHERE c.type = 'm.room.member' AND c.room_id = ? AND m.membership = ? - """ - - txn.execute(sql, (room_id, Membership.JOIN)) - return [to_ascii(r[0]) for r in txn] - - @cached(max_entries=100000) - def get_room_summary(self, room_id): - """ Get the details of a room roughly suitable for use by the room - summary extension to /sync. Useful when lazy loading room members. - Args: - room_id (str): The room ID to query - Returns: - Deferred[dict[str, MemberSummary]: - dict of membership states, pointing to a MemberSummary named tuple. - """ - - def _get_room_summary_txn(txn): - # first get counts. - # We do this all in one transaction to keep the cache small. - # FIXME: get rid of this when we have room_stats - - # If we can assume current_state_events.membership is up to date - # then we can avoid a join, which is a Very Good Thing given how - # frequently this function gets called. - if self._current_state_events_membership_up_to_date: - # Note, rejected events will have a null membership field, so - # we we manually filter them out. - sql = """ - SELECT count(*), membership FROM current_state_events - WHERE type = 'm.room.member' AND room_id = ? - AND membership IS NOT NULL - GROUP BY membership - """ - else: - sql = """ - SELECT count(*), m.membership FROM room_memberships as m - INNER JOIN current_state_events as c - ON m.event_id = c.event_id - AND m.room_id = c.room_id - AND m.user_id = c.state_key - WHERE c.type = 'm.room.member' AND c.room_id = ? - GROUP BY m.membership - """ - - txn.execute(sql, (room_id,)) - res = {} - for count, membership in txn: - summary = res.setdefault(to_ascii(membership), MemberSummary([], count)) - - # we order by membership and then fairly arbitrarily by event_id so - # heroes are consistent - if self._current_state_events_membership_up_to_date: - # Note, rejected events will have a null membership field, so - # we we manually filter them out. - sql = """ - SELECT state_key, membership, event_id - FROM current_state_events - WHERE type = 'm.room.member' AND room_id = ? - AND membership IS NOT NULL - ORDER BY - CASE membership WHEN ? THEN 1 WHEN ? THEN 2 ELSE 3 END ASC, - event_id ASC - LIMIT ? - """ - else: - sql = """ - SELECT c.state_key, m.membership, c.event_id - FROM room_memberships as m - INNER JOIN current_state_events as c USING (room_id, event_id) - WHERE c.type = 'm.room.member' AND c.room_id = ? - ORDER BY - CASE m.membership WHEN ? THEN 1 WHEN ? THEN 2 ELSE 3 END ASC, - c.event_id ASC - LIMIT ? - """ - - # 6 is 5 (number of heroes) plus 1, in case one of them is the calling user. - txn.execute(sql, (room_id, Membership.JOIN, Membership.INVITE, 6)) - for user_id, membership, event_id in txn: - summary = res[to_ascii(membership)] - # we will always have a summary for this membership type at this - # point given the summary currently contains the counts. - members = summary.members - members.append((to_ascii(user_id), to_ascii(event_id))) - - return res - - return self.runInteraction("get_room_summary", _get_room_summary_txn) - - def _get_user_counts_in_room_txn(self, txn, room_id): - """ - Get the user count in a room by membership. - - Args: - room_id (str) - membership (Membership) - - Returns: - Deferred[int] - """ - sql = """ - SELECT m.membership, count(*) FROM room_memberships as m - INNER JOIN current_state_events as c USING(event_id) - WHERE c.type = 'm.room.member' AND c.room_id = ? - GROUP BY m.membership - """ - - txn.execute(sql, (room_id,)) - return {row[0]: row[1] for row in txn} - - @cached() - def get_invited_rooms_for_user(self, user_id): - """ Get all the rooms the user is invited to - Args: - user_id (str): The user ID. - Returns: - A deferred list of RoomsForUser. - """ - - return self.get_rooms_for_user_where_membership_is(user_id, [Membership.INVITE]) - - @defer.inlineCallbacks - def get_invite_for_user_in_room(self, user_id, room_id): - """Gets the invite for the given user and room - - Args: - user_id (str) - room_id (str) - - Returns: - Deferred: Resolves to either a RoomsForUser or None if no invite was - found. - """ - invites = yield self.get_invited_rooms_for_user(user_id) - for invite in invites: - if invite.room_id == room_id: - return invite - return None - - @defer.inlineCallbacks - def get_rooms_for_user_where_membership_is(self, user_id, membership_list): - """ Get all the rooms for this user where the membership for this user - matches one in the membership list. - - Filters out forgotten rooms. - - Args: - user_id (str): The user ID. - membership_list (list): A list of synapse.api.constants.Membership - values which the user must be in. - - Returns: - Deferred[list[RoomsForUser]] - """ - if not membership_list: - return defer.succeed(None) - - rooms = yield self.runInteraction( - "get_rooms_for_user_where_membership_is", - self._get_rooms_for_user_where_membership_is_txn, - user_id, - membership_list, - ) - - # Now we filter out forgotten rooms - forgotten_rooms = yield self.get_forgotten_rooms_for_user(user_id) - return [room for room in rooms if room.room_id not in forgotten_rooms] - - def _get_rooms_for_user_where_membership_is_txn( - self, txn, user_id, membership_list - ): - - do_invite = Membership.INVITE in membership_list - membership_list = [m for m in membership_list if m != Membership.INVITE] - - results = [] - if membership_list: - if self._current_state_events_membership_up_to_date: - clause, args = make_in_list_sql_clause( - self.database_engine, "c.membership", membership_list - ) - sql = """ - SELECT room_id, e.sender, c.membership, event_id, e.stream_ordering - FROM current_state_events AS c - INNER JOIN events AS e USING (room_id, event_id) - WHERE - c.type = 'm.room.member' - AND state_key = ? - AND %s - """ % ( - clause, - ) - else: - clause, args = make_in_list_sql_clause( - self.database_engine, "m.membership", membership_list - ) - sql = """ - SELECT room_id, e.sender, m.membership, event_id, e.stream_ordering - FROM current_state_events AS c - INNER JOIN room_memberships AS m USING (room_id, event_id) - INNER JOIN events AS e USING (room_id, event_id) - WHERE - c.type = 'm.room.member' - AND state_key = ? - AND %s - """ % ( - clause, - ) - - txn.execute(sql, (user_id, *args)) - results = [RoomsForUser(**r) for r in self.cursor_to_dict(txn)] - - if do_invite: - sql = ( - "SELECT i.room_id, inviter, i.event_id, e.stream_ordering" - " FROM local_invites as i" - " INNER JOIN events as e USING (event_id)" - " WHERE invitee = ? AND locally_rejected is NULL" - " AND replaced_by is NULL" - ) - - txn.execute(sql, (user_id,)) - results.extend( - RoomsForUser( - room_id=r["room_id"], - sender=r["inviter"], - event_id=r["event_id"], - stream_ordering=r["stream_ordering"], - membership=Membership.INVITE, - ) - for r in self.cursor_to_dict(txn) - ) - - return results - - @cachedInlineCallbacks(max_entries=500000, iterable=True) - def get_rooms_for_user_with_stream_ordering(self, user_id): - """Returns a set of room_ids the user is currently joined to - - Args: - user_id (str) - - Returns: - Deferred[frozenset[GetRoomsForUserWithStreamOrdering]]: Returns - the rooms the user is in currently, along with the stream ordering - of the most recent join for that user and room. - """ - rooms = yield self.get_rooms_for_user_where_membership_is( - user_id, membership_list=[Membership.JOIN] - ) - return frozenset( - GetRoomsForUserWithStreamOrdering(r.room_id, r.stream_ordering) - for r in rooms - ) - - @defer.inlineCallbacks - def get_rooms_for_user(self, user_id, on_invalidate=None): - """Returns a set of room_ids the user is currently joined to - """ - rooms = yield self.get_rooms_for_user_with_stream_ordering( - user_id, on_invalidate=on_invalidate - ) - return frozenset(r.room_id for r in rooms) - - @cachedInlineCallbacks(max_entries=500000, cache_context=True, iterable=True) - def get_users_who_share_room_with_user(self, user_id, cache_context): - """Returns the set of users who share a room with `user_id` - """ - room_ids = yield self.get_rooms_for_user( - user_id, on_invalidate=cache_context.invalidate - ) - - user_who_share_room = set() - for room_id in room_ids: - user_ids = yield self.get_users_in_room( - room_id, on_invalidate=cache_context.invalidate - ) - user_who_share_room.update(user_ids) - - return user_who_share_room - - @defer.inlineCallbacks - def get_joined_users_from_context(self, event, context): - state_group = context.state_group - if not state_group: - # If state_group is None it means it has yet to be assigned a - # state group, i.e. we need to make sure that calls with a state_group - # of None don't hit previous cached calls with a None state_group. - # To do this we set the state_group to a new object as object() != object() - state_group = object() - - current_state_ids = yield context.get_current_state_ids(self) - result = yield self._get_joined_users_from_context( - event.room_id, state_group, current_state_ids, event=event, context=context - ) - return result - - @defer.inlineCallbacks - def get_joined_users_from_state(self, room_id, state_entry): - state_group = state_entry.state_group - if not state_group: - # If state_group is None it means it has yet to be assigned a - # state group, i.e. we need to make sure that calls with a state_group - # of None don't hit previous cached calls with a None state_group. - # To do this we set the state_group to a new object as object() != object() - state_group = object() - - with Measure(self._clock, "get_joined_users_from_state"): - return ( - yield self._get_joined_users_from_context( - room_id, state_group, state_entry.state, context=state_entry - ) - ) - - @cachedInlineCallbacks( - num_args=2, cache_context=True, iterable=True, max_entries=100000 - ) - def _get_joined_users_from_context( - self, - room_id, - state_group, - current_state_ids, - cache_context, - event=None, - context=None, - ): - # We don't use `state_group`, it's there so that we can cache based - # on it. However, it's important that it's never None, since two current_states - # with a state_group of None are likely to be different. - # See bulk_get_push_rules_for_room for how we work around this. - assert state_group is not None - - users_in_room = {} - member_event_ids = [ - e_id - for key, e_id in iteritems(current_state_ids) - if key[0] == EventTypes.Member - ] - - if context is not None: - # If we have a context with a delta from a previous state group, - # check if we also have the result from the previous group in cache. - # If we do then we can reuse that result and simply update it with - # any membership changes in `delta_ids` - if context.prev_group and context.delta_ids: - prev_res = self._get_joined_users_from_context.cache.get( - (room_id, context.prev_group), None - ) - if prev_res and isinstance(prev_res, dict): - users_in_room = dict(prev_res) - member_event_ids = [ - e_id - for key, e_id in iteritems(context.delta_ids) - if key[0] == EventTypes.Member - ] - for etype, state_key in context.delta_ids: - users_in_room.pop(state_key, None) - - # We check if we have any of the member event ids in the event cache - # before we ask the DB - - # We don't update the event cache hit ratio as it completely throws off - # the hit ratio counts. After all, we don't populate the cache if we - # miss it here - event_map = self._get_events_from_cache( - member_event_ids, allow_rejected=False, update_metrics=False - ) - - missing_member_event_ids = [] - for event_id in member_event_ids: - ev_entry = event_map.get(event_id) - if ev_entry: - if ev_entry.event.membership == Membership.JOIN: - users_in_room[to_ascii(ev_entry.event.state_key)] = ProfileInfo( - display_name=to_ascii( - ev_entry.event.content.get("displayname", None) - ), - avatar_url=to_ascii( - ev_entry.event.content.get("avatar_url", None) - ), - ) - else: - missing_member_event_ids.append(event_id) - - if missing_member_event_ids: - event_to_memberships = yield self._get_joined_profiles_from_event_ids( - missing_member_event_ids - ) - users_in_room.update((row for row in event_to_memberships.values() if row)) - - if event is not None and event.type == EventTypes.Member: - if event.membership == Membership.JOIN: - if event.event_id in member_event_ids: - users_in_room[to_ascii(event.state_key)] = ProfileInfo( - display_name=to_ascii(event.content.get("displayname", None)), - avatar_url=to_ascii(event.content.get("avatar_url", None)), - ) - - return users_in_room - - @cached(max_entries=10000) - def _get_joined_profile_from_event_id(self, event_id): - raise NotImplementedError() - - @cachedList( - cached_method_name="_get_joined_profile_from_event_id", - list_name="event_ids", - inlineCallbacks=True, - ) - def _get_joined_profiles_from_event_ids(self, event_ids): - """For given set of member event_ids check if they point to a join - event and if so return the associated user and profile info. - - Args: - event_ids (Iterable[str]): The member event IDs to lookup - - Returns: - Deferred[dict[str, Tuple[str, ProfileInfo]|None]]: Map from event ID - to `user_id` and ProfileInfo (or None if not join event). - """ - - rows = yield self._simple_select_many_batch( - table="room_memberships", - column="event_id", - iterable=event_ids, - retcols=("user_id", "display_name", "avatar_url", "event_id"), - keyvalues={"membership": Membership.JOIN}, - batch_size=500, - desc="_get_membership_from_event_ids", - ) - - return { - row["event_id"]: ( - row["user_id"], - ProfileInfo( - avatar_url=row["avatar_url"], display_name=row["display_name"] - ), - ) - for row in rows - } - - @cachedInlineCallbacks(max_entries=10000) - def is_host_joined(self, room_id, host): - if "%" in host or "_" in host: - raise Exception("Invalid host name") - - sql = """ - SELECT state_key FROM current_state_events AS c - INNER JOIN room_memberships AS m USING (event_id) - WHERE m.membership = 'join' - AND type = 'm.room.member' - AND c.room_id = ? - AND state_key LIKE ? - LIMIT 1 - """ - - # We do need to be careful to ensure that host doesn't have any wild cards - # in it, but we checked above for known ones and we'll check below that - # the returned user actually has the correct domain. - like_clause = "%:" + host - - rows = yield self._execute("is_host_joined", None, sql, room_id, like_clause) - - if not rows: - return False - - user_id = rows[0][0] - if get_domain_from_id(user_id) != host: - # This can only happen if the host name has something funky in it - raise Exception("Invalid host name") - - return True - - @cachedInlineCallbacks() - def was_host_joined(self, room_id, host): - """Check whether the server is or ever was in the room. - - Args: - room_id (str) - host (str) - - Returns: - Deferred: Resolves to True if the host is/was in the room, otherwise - False. - """ - if "%" in host or "_" in host: - raise Exception("Invalid host name") - - sql = """ - SELECT user_id FROM room_memberships - WHERE room_id = ? - AND user_id LIKE ? - AND membership = 'join' - LIMIT 1 - """ - - # We do need to be careful to ensure that host doesn't have any wild cards - # in it, but we checked above for known ones and we'll check below that - # the returned user actually has the correct domain. - like_clause = "%:" + host - - rows = yield self._execute("was_host_joined", None, sql, room_id, like_clause) - - if not rows: - return False - - user_id = rows[0][0] - if get_domain_from_id(user_id) != host: - # This can only happen if the host name has something funky in it - raise Exception("Invalid host name") - - return True - - @defer.inlineCallbacks - def get_joined_hosts(self, room_id, state_entry): - state_group = state_entry.state_group - if not state_group: - # If state_group is None it means it has yet to be assigned a - # state group, i.e. we need to make sure that calls with a state_group - # of None don't hit previous cached calls with a None state_group. - # To do this we set the state_group to a new object as object() != object() - state_group = object() - - with Measure(self._clock, "get_joined_hosts"): - return ( - yield self._get_joined_hosts( - room_id, state_group, state_entry.state, state_entry=state_entry - ) - ) - - @cachedInlineCallbacks(num_args=2, max_entries=10000, iterable=True) - # @defer.inlineCallbacks - def _get_joined_hosts(self, room_id, state_group, current_state_ids, state_entry): - # We don't use `state_group`, its there so that we can cache based - # on it. However, its important that its never None, since two current_state's - # with a state_group of None are likely to be different. - # See bulk_get_push_rules_for_room for how we work around this. - assert state_group is not None - - cache = self._get_joined_hosts_cache(room_id) - joined_hosts = yield cache.get_destinations(state_entry) - - return joined_hosts - - @cached(max_entries=10000) - def _get_joined_hosts_cache(self, room_id): - return _JoinedHostsCache(self, room_id) - - @cachedInlineCallbacks(num_args=2) - def did_forget(self, user_id, room_id): - """Returns whether user_id has elected to discard history for room_id. - - Returns False if they have since re-joined.""" - - def f(txn): - sql = ( - "SELECT" - " COUNT(*)" - " FROM" - " room_memberships" - " WHERE" - " user_id = ?" - " AND" - " room_id = ?" - " AND" - " forgotten = 0" - ) - txn.execute(sql, (user_id, room_id)) - rows = txn.fetchall() - return rows[0][0] - - count = yield self.runInteraction("did_forget_membership", f) - return count == 0 - - @cached() - def get_forgotten_rooms_for_user(self, user_id): - """Gets all rooms the user has forgotten. - - Args: - user_id (str) - - Returns: - Deferred[set[str]] - """ - - def _get_forgotten_rooms_for_user_txn(txn): - # This is a slightly convoluted query that first looks up all rooms - # that the user has forgotten in the past, then rechecks that list - # to see if any have subsequently been updated. This is done so that - # we can use a partial index on `forgotten = 1` on the assumption - # that few users will actually forget many rooms. - # - # Note that a room is considered "forgotten" if *all* membership - # events for that user and room have the forgotten field set (as - # when a user forgets a room we update all rows for that user and - # room, not just the current one). - sql = """ - SELECT room_id, ( - SELECT count(*) FROM room_memberships - WHERE room_id = m.room_id AND user_id = m.user_id AND forgotten = 0 - ) AS count - FROM room_memberships AS m - WHERE user_id = ? AND forgotten = 1 - GROUP BY room_id, user_id; - """ - txn.execute(sql, (user_id,)) - return set(row[0] for row in txn if row[1] == 0) - - return self.runInteraction( - "get_forgotten_rooms_for_user", _get_forgotten_rooms_for_user_txn - ) - - @defer.inlineCallbacks - def get_rooms_user_has_been_in(self, user_id): - """Get all rooms that the user has ever been in. - - Args: - user_id (str) - - Returns: - Deferred[set[str]]: Set of room IDs. - """ - - room_ids = yield self._simple_select_onecol( - table="room_memberships", - keyvalues={"membership": Membership.JOIN, "user_id": user_id}, - retcol="room_id", - desc="get_rooms_user_has_been_in", - ) - - return set(room_ids) - - -class RoomMemberBackgroundUpdateStore(BackgroundUpdateStore): - def __init__(self, db_conn, hs): - super(RoomMemberBackgroundUpdateStore, self).__init__(db_conn, hs) - self.register_background_update_handler( - _MEMBERSHIP_PROFILE_UPDATE_NAME, self._background_add_membership_profile - ) - self.register_background_update_handler( - _CURRENT_STATE_MEMBERSHIP_UPDATE_NAME, - self._background_current_state_membership, - ) - self.register_background_index_update( - "room_membership_forgotten_idx", - index_name="room_memberships_user_room_forgotten", - table="room_memberships", - columns=["user_id", "room_id"], - where_clause="forgotten = 1", - ) - - @defer.inlineCallbacks - def _background_add_membership_profile(self, progress, batch_size): - target_min_stream_id = progress.get( - "target_min_stream_id_inclusive", self._min_stream_order_on_start - ) - max_stream_id = progress.get( - "max_stream_id_exclusive", self._stream_order_on_start + 1 - ) - - INSERT_CLUMP_SIZE = 1000 - - def add_membership_profile_txn(txn): - sql = """ - SELECT stream_ordering, event_id, events.room_id, event_json.json - FROM events - INNER JOIN event_json USING (event_id) - INNER JOIN room_memberships USING (event_id) - WHERE ? <= stream_ordering AND stream_ordering < ? - AND type = 'm.room.member' - ORDER BY stream_ordering DESC - LIMIT ? - """ - - txn.execute(sql, (target_min_stream_id, max_stream_id, batch_size)) - - rows = self.cursor_to_dict(txn) - if not rows: - return 0 - - min_stream_id = rows[-1]["stream_ordering"] - - to_update = [] - for row in rows: - event_id = row["event_id"] - room_id = row["room_id"] - try: - event_json = json.loads(row["json"]) - content = event_json["content"] - except Exception: - continue - - display_name = content.get("displayname", None) - avatar_url = content.get("avatar_url", None) - - if display_name or avatar_url: - to_update.append((display_name, avatar_url, event_id, room_id)) - - to_update_sql = """ - UPDATE room_memberships SET display_name = ?, avatar_url = ? - WHERE event_id = ? AND room_id = ? - """ - for index in range(0, len(to_update), INSERT_CLUMP_SIZE): - clump = to_update[index : index + INSERT_CLUMP_SIZE] - txn.executemany(to_update_sql, clump) - - progress = { - "target_min_stream_id_inclusive": target_min_stream_id, - "max_stream_id_exclusive": min_stream_id, - } - - self._background_update_progress_txn( - txn, _MEMBERSHIP_PROFILE_UPDATE_NAME, progress - ) - - return len(rows) - - result = yield self.runInteraction( - _MEMBERSHIP_PROFILE_UPDATE_NAME, add_membership_profile_txn - ) - - if not result: - yield self._end_background_update(_MEMBERSHIP_PROFILE_UPDATE_NAME) - - return result - - @defer.inlineCallbacks - def _background_current_state_membership(self, progress, batch_size): - """Update the new membership column on current_state_events. - - This works by iterating over all rooms in alphebetical order. - """ - - def _background_current_state_membership_txn(txn, last_processed_room): - processed = 0 - while processed < batch_size: - txn.execute( - """ - SELECT MIN(room_id) FROM current_state_events WHERE room_id > ? - """, - (last_processed_room,), - ) - row = txn.fetchone() - if not row or not row[0]: - return processed, True - - next_room, = row - - sql = """ - UPDATE current_state_events - SET membership = ( - SELECT membership FROM room_memberships - WHERE event_id = current_state_events.event_id - ) - WHERE room_id = ? - """ - txn.execute(sql, (next_room,)) - processed += txn.rowcount - - last_processed_room = next_room - - self._background_update_progress_txn( - txn, - _CURRENT_STATE_MEMBERSHIP_UPDATE_NAME, - {"last_processed_room": last_processed_room}, - ) - - return processed, False - - # If we haven't got a last processed room then just use the empty - # string, which will compare before all room IDs correctly. - last_processed_room = progress.get("last_processed_room", "") - - row_count, finished = yield self.runInteraction( - "_background_current_state_membership_update", - _background_current_state_membership_txn, - last_processed_room, - ) - - if finished: - yield self._end_background_update(_CURRENT_STATE_MEMBERSHIP_UPDATE_NAME) - - return row_count - - -class RoomMemberStore(RoomMemberWorkerStore, RoomMemberBackgroundUpdateStore): - def __init__(self, db_conn, hs): - super(RoomMemberStore, self).__init__(db_conn, hs) - - def _store_room_members_txn(self, txn, events, backfilled): - """Store a room member in the database. - """ - self._simple_insert_many_txn( - txn, - table="room_memberships", - values=[ - { - "event_id": event.event_id, - "user_id": event.state_key, - "sender": event.user_id, - "room_id": event.room_id, - "membership": event.membership, - "display_name": event.content.get("displayname", None), - "avatar_url": event.content.get("avatar_url", None), - } - for event in events - ], - ) - - for event in events: - txn.call_after( - self._membership_stream_cache.entity_has_changed, - event.state_key, - event.internal_metadata.stream_ordering, - ) - txn.call_after( - self.get_invited_rooms_for_user.invalidate, (event.state_key,) - ) - - # We update the local_invites table only if the event is "current", - # i.e., its something that has just happened. If the event is an - # outlier it is only current if its an "out of band membership", - # like a remote invite or a rejection of a remote invite. - is_new_state = not backfilled and ( - not event.internal_metadata.is_outlier() - or event.internal_metadata.is_out_of_band_membership() - ) - is_mine = self.hs.is_mine_id(event.state_key) - if is_new_state and is_mine: - if event.membership == Membership.INVITE: - self._simple_insert_txn( - txn, - table="local_invites", - values={ - "event_id": event.event_id, - "invitee": event.state_key, - "inviter": event.sender, - "room_id": event.room_id, - "stream_id": event.internal_metadata.stream_ordering, - }, - ) - else: - sql = ( - "UPDATE local_invites SET stream_id = ?, replaced_by = ? WHERE" - " room_id = ? AND invitee = ? AND locally_rejected is NULL" - " AND replaced_by is NULL" - ) - - txn.execute( - sql, - ( - event.internal_metadata.stream_ordering, - event.event_id, - event.room_id, - event.state_key, - ), - ) - - @defer.inlineCallbacks - def locally_reject_invite(self, user_id, room_id): - sql = ( - "UPDATE local_invites SET stream_id = ?, locally_rejected = ? WHERE" - " room_id = ? AND invitee = ? AND locally_rejected is NULL" - " AND replaced_by is NULL" - ) - - def f(txn, stream_ordering): - txn.execute(sql, (stream_ordering, True, room_id, user_id)) - - with self._stream_id_gen.get_next() as stream_ordering: - yield self.runInteraction("locally_reject_invite", f, stream_ordering) - - def forget(self, user_id, room_id): - """Indicate that user_id wishes to discard history for room_id.""" - - def f(txn): - sql = ( - "UPDATE" - " room_memberships" - " SET" - " forgotten = 1" - " WHERE" - " user_id = ?" - " AND" - " room_id = ?" - ) - txn.execute(sql, (user_id, room_id)) - - self._invalidate_cache_and_stream(txn, self.did_forget, (user_id, room_id)) - self._invalidate_cache_and_stream( - txn, self.get_forgotten_rooms_for_user, (user_id,) - ) - - return self.runInteraction("forget_membership", f) - - -class _JoinedHostsCache(object): - """Cache for joined hosts in a room that is optimised to handle updates - via state deltas. - """ - - def __init__(self, store, room_id): - self.store = store - self.room_id = room_id - - self.hosts_to_joined_users = {} - - self.state_group = object() - - self.linearizer = Linearizer("_JoinedHostsCache") - - self._len = 0 - - @defer.inlineCallbacks - def get_destinations(self, state_entry): - """Get set of destinations for a state entry - - Args: - state_entry(synapse.state._StateCacheEntry) - """ - if state_entry.state_group == self.state_group: - return frozenset(self.hosts_to_joined_users) - - with (yield self.linearizer.queue(())): - if state_entry.state_group == self.state_group: - pass - elif state_entry.prev_group == self.state_group: - for (typ, state_key), event_id in iteritems(state_entry.delta_ids): - if typ != EventTypes.Member: - continue - - host = intern_string(get_domain_from_id(state_key)) - user_id = state_key - known_joins = self.hosts_to_joined_users.setdefault(host, set()) - - event = yield self.store.get_event(event_id) - if event.membership == Membership.JOIN: - known_joins.add(user_id) - else: - known_joins.discard(user_id) - - if not known_joins: - self.hosts_to_joined_users.pop(host, None) - else: - joined_users = yield self.store.get_joined_users_from_state( - self.room_id, state_entry - ) - - self.hosts_to_joined_users = {} - for user_id in joined_users: - host = intern_string(get_domain_from_id(user_id)) - self.hosts_to_joined_users.setdefault(host, set()).add(user_id) - - if state_entry.state_group: - self.state_group = state_entry.state_group - else: - self.state_group = object() - self._len = sum(len(v) for v in itervalues(self.hosts_to_joined_users)) - return frozenset(self.hosts_to_joined_users) - - def __len__(self): - return self._len diff --git a/synapse/storage/schema/delta/12/v12.sql b/synapse/storage/schema/delta/12/v12.sql deleted file mode 100644 index 5964c5aaac..0000000000 --- a/synapse/storage/schema/delta/12/v12.sql +++ /dev/null @@ -1,63 +0,0 @@ -/* Copyright 2015, 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -CREATE TABLE IF NOT EXISTS rejections( - event_id TEXT NOT NULL, - reason TEXT NOT NULL, - last_check TEXT NOT NULL, - UNIQUE (event_id) -); - --- Push notification endpoints that users have configured -CREATE TABLE IF NOT EXISTS pushers ( - id INTEGER PRIMARY KEY AUTOINCREMENT, - user_name TEXT NOT NULL, - profile_tag VARCHAR(32) NOT NULL, - kind VARCHAR(8) NOT NULL, - app_id VARCHAR(64) NOT NULL, - app_display_name VARCHAR(64) NOT NULL, - device_display_name VARCHAR(128) NOT NULL, - pushkey VARBINARY(512) NOT NULL, - ts BIGINT UNSIGNED NOT NULL, - lang VARCHAR(8), - data LONGBLOB, - last_token TEXT, - last_success BIGINT UNSIGNED, - failing_since BIGINT UNSIGNED, - UNIQUE (app_id, pushkey) -); - -CREATE TABLE IF NOT EXISTS push_rules ( - id INTEGER PRIMARY KEY AUTOINCREMENT, - user_name TEXT NOT NULL, - rule_id TEXT NOT NULL, - priority_class TINYINT NOT NULL, - priority INTEGER NOT NULL DEFAULT 0, - conditions TEXT NOT NULL, - actions TEXT NOT NULL, - UNIQUE(user_name, rule_id) -); - -CREATE INDEX IF NOT EXISTS push_rules_user_name on push_rules (user_name); - -CREATE TABLE IF NOT EXISTS user_filters( - user_id TEXT, - filter_id BIGINT UNSIGNED, - filter_json LONGBLOB -); - -CREATE INDEX IF NOT EXISTS user_filters_by_user_id_filter_id ON user_filters( - user_id, filter_id -); diff --git a/synapse/storage/schema/delta/13/v13.sql b/synapse/storage/schema/delta/13/v13.sql deleted file mode 100644 index f8649e5d99..0000000000 --- a/synapse/storage/schema/delta/13/v13.sql +++ /dev/null @@ -1,19 +0,0 @@ -/* Copyright 2015, 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -/* We used to create a tables called application_services and - * application_services_regex, but these are no longer used and are removed in - * delta 54. - */ diff --git a/synapse/storage/schema/delta/14/v14.sql b/synapse/storage/schema/delta/14/v14.sql deleted file mode 100644 index a831920da6..0000000000 --- a/synapse/storage/schema/delta/14/v14.sql +++ /dev/null @@ -1,23 +0,0 @@ -/* Copyright 2015, 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ -CREATE TABLE IF NOT EXISTS push_rules_enable ( - id INTEGER PRIMARY KEY AUTOINCREMENT, - user_name TEXT NOT NULL, - rule_id TEXT NOT NULL, - enabled TINYINT, - UNIQUE(user_name, rule_id) -); - -CREATE INDEX IF NOT EXISTS push_rules_enable_user_name on push_rules_enable (user_name); diff --git a/synapse/storage/schema/delta/15/appservice_txns.sql b/synapse/storage/schema/delta/15/appservice_txns.sql deleted file mode 100644 index e4f5e76aec..0000000000 --- a/synapse/storage/schema/delta/15/appservice_txns.sql +++ /dev/null @@ -1,31 +0,0 @@ -/* Copyright 2015, 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -CREATE TABLE IF NOT EXISTS application_services_state( - as_id TEXT PRIMARY KEY, - state VARCHAR(5), - last_txn INTEGER -); - -CREATE TABLE IF NOT EXISTS application_services_txns( - as_id TEXT NOT NULL, - txn_id INTEGER NOT NULL, - event_ids TEXT NOT NULL, - UNIQUE(as_id, txn_id) -); - -CREATE INDEX IF NOT EXISTS application_services_txns_id ON application_services_txns ( - as_id -); diff --git a/synapse/storage/schema/delta/15/presence_indices.sql b/synapse/storage/schema/delta/15/presence_indices.sql deleted file mode 100644 index 6b8d0f1ca7..0000000000 --- a/synapse/storage/schema/delta/15/presence_indices.sql +++ /dev/null @@ -1,2 +0,0 @@ - -CREATE INDEX IF NOT EXISTS presence_list_user_id ON presence_list (user_id); diff --git a/synapse/storage/schema/delta/15/v15.sql b/synapse/storage/schema/delta/15/v15.sql deleted file mode 100644 index 9523d2bcc3..0000000000 --- a/synapse/storage/schema/delta/15/v15.sql +++ /dev/null @@ -1,24 +0,0 @@ --- Drop, copy & recreate pushers table to change unique key --- Also add access_token column at the same time -CREATE TABLE IF NOT EXISTS pushers2 ( - id BIGINT PRIMARY KEY, - user_name TEXT NOT NULL, - access_token BIGINT DEFAULT NULL, - profile_tag VARCHAR(32) NOT NULL, - kind VARCHAR(8) NOT NULL, - app_id VARCHAR(64) NOT NULL, - app_display_name VARCHAR(64) NOT NULL, - device_display_name VARCHAR(128) NOT NULL, - pushkey bytea NOT NULL, - ts BIGINT NOT NULL, - lang VARCHAR(8), - data bytea, - last_token TEXT, - last_success BIGINT, - failing_since BIGINT, - UNIQUE (app_id, pushkey) -); -INSERT INTO pushers2 (id, user_name, profile_tag, kind, app_id, app_display_name, device_display_name, pushkey, ts, lang, data, last_token, last_success, failing_since) - SELECT id, user_name, profile_tag, kind, app_id, app_display_name, device_display_name, pushkey, ts, lang, data, last_token, last_success, failing_since FROM pushers; -DROP TABLE pushers; -ALTER TABLE pushers2 RENAME TO pushers; diff --git a/synapse/storage/schema/delta/16/events_order_index.sql b/synapse/storage/schema/delta/16/events_order_index.sql deleted file mode 100644 index a48f215170..0000000000 --- a/synapse/storage/schema/delta/16/events_order_index.sql +++ /dev/null @@ -1,4 +0,0 @@ -CREATE INDEX events_order ON events (topological_ordering, stream_ordering); -CREATE INDEX events_order_room ON events ( - room_id, topological_ordering, stream_ordering -); diff --git a/synapse/storage/schema/delta/16/remote_media_cache_index.sql b/synapse/storage/schema/delta/16/remote_media_cache_index.sql deleted file mode 100644 index 7a15265cb1..0000000000 --- a/synapse/storage/schema/delta/16/remote_media_cache_index.sql +++ /dev/null @@ -1,2 +0,0 @@ -CREATE INDEX IF NOT EXISTS remote_media_cache_thumbnails_media_id - ON remote_media_cache_thumbnails (media_id); \ No newline at end of file diff --git a/synapse/storage/schema/delta/16/remove_duplicates.sql b/synapse/storage/schema/delta/16/remove_duplicates.sql deleted file mode 100644 index 65c97b5e2f..0000000000 --- a/synapse/storage/schema/delta/16/remove_duplicates.sql +++ /dev/null @@ -1,9 +0,0 @@ - - -DELETE FROM event_to_state_groups WHERE state_group not in ( - SELECT MAX(state_group) FROM event_to_state_groups GROUP BY event_id -); - -DELETE FROM event_to_state_groups WHERE rowid not in ( - SELECT MIN(rowid) FROM event_to_state_groups GROUP BY event_id -); diff --git a/synapse/storage/schema/delta/16/room_alias_index.sql b/synapse/storage/schema/delta/16/room_alias_index.sql deleted file mode 100644 index f82486132b..0000000000 --- a/synapse/storage/schema/delta/16/room_alias_index.sql +++ /dev/null @@ -1,3 +0,0 @@ - -CREATE INDEX IF NOT EXISTS room_aliases_id ON room_aliases(room_id); -CREATE INDEX IF NOT EXISTS room_alias_servers_alias ON room_alias_servers(room_alias); diff --git a/synapse/storage/schema/delta/16/unique_constraints.sql b/synapse/storage/schema/delta/16/unique_constraints.sql deleted file mode 100644 index 5b8de52c33..0000000000 --- a/synapse/storage/schema/delta/16/unique_constraints.sql +++ /dev/null @@ -1,72 +0,0 @@ - --- We can use SQLite features here, since other db support was only added in v16 - --- -DELETE FROM current_state_events WHERE rowid not in ( - SELECT MIN(rowid) FROM current_state_events GROUP BY event_id -); - -DROP INDEX IF EXISTS current_state_events_event_id; -CREATE UNIQUE INDEX current_state_events_event_id ON current_state_events(event_id); - --- -DELETE FROM room_memberships WHERE rowid not in ( - SELECT MIN(rowid) FROM room_memberships GROUP BY event_id -); - -DROP INDEX IF EXISTS room_memberships_event_id; -CREATE UNIQUE INDEX room_memberships_event_id ON room_memberships(event_id); - --- -DELETE FROM topics WHERE rowid not in ( - SELECT MIN(rowid) FROM topics GROUP BY event_id -); - -DROP INDEX IF EXISTS topics_event_id; -CREATE UNIQUE INDEX topics_event_id ON topics(event_id); - --- -DELETE FROM room_names WHERE rowid not in ( - SELECT MIN(rowid) FROM room_names GROUP BY event_id -); - -DROP INDEX IF EXISTS room_names_id; -CREATE UNIQUE INDEX room_names_id ON room_names(event_id); - --- -DELETE FROM presence WHERE rowid not in ( - SELECT MIN(rowid) FROM presence GROUP BY user_id -); - -DROP INDEX IF EXISTS presence_id; -CREATE UNIQUE INDEX presence_id ON presence(user_id); - --- -DELETE FROM presence_allow_inbound WHERE rowid not in ( - SELECT MIN(rowid) FROM presence_allow_inbound - GROUP BY observed_user_id, observer_user_id -); - -DROP INDEX IF EXISTS presence_allow_inbound_observers; -CREATE UNIQUE INDEX presence_allow_inbound_observers ON presence_allow_inbound( - observed_user_id, observer_user_id -); - --- -DELETE FROM presence_list WHERE rowid not in ( - SELECT MIN(rowid) FROM presence_list - GROUP BY user_id, observed_user_id -); - -DROP INDEX IF EXISTS presence_list_observers; -CREATE UNIQUE INDEX presence_list_observers ON presence_list( - user_id, observed_user_id -); - --- -DELETE FROM room_aliases WHERE rowid not in ( - SELECT MIN(rowid) FROM room_aliases GROUP BY room_alias -); - -DROP INDEX IF EXISTS room_aliases_id; -CREATE INDEX room_aliases_id ON room_aliases(room_id); diff --git a/synapse/storage/schema/delta/16/users.sql b/synapse/storage/schema/delta/16/users.sql deleted file mode 100644 index cd0709250d..0000000000 --- a/synapse/storage/schema/delta/16/users.sql +++ /dev/null @@ -1,56 +0,0 @@ --- Convert `access_tokens`.user from rowids to user strings. --- MUST BE DONE BEFORE REMOVING ID COLUMN FROM USERS TABLE BELOW -CREATE TABLE IF NOT EXISTS new_access_tokens( - id BIGINT UNSIGNED PRIMARY KEY, - user_id TEXT NOT NULL, - device_id TEXT, - token TEXT NOT NULL, - last_used BIGINT UNSIGNED, - UNIQUE(token) -); - -INSERT INTO new_access_tokens - SELECT a.id, u.name, a.device_id, a.token, a.last_used - FROM access_tokens as a - INNER JOIN users as u ON u.id = a.user_id; - -DROP TABLE access_tokens; - -ALTER TABLE new_access_tokens RENAME TO access_tokens; - --- Remove ID column from `users` table -CREATE TABLE IF NOT EXISTS new_users( - name TEXT, - password_hash TEXT, - creation_ts BIGINT UNSIGNED, - admin BOOL DEFAULT 0 NOT NULL, - UNIQUE(name) -); - -INSERT INTO new_users SELECT name, password_hash, creation_ts, admin FROM users; - -DROP TABLE users; - -ALTER TABLE new_users RENAME TO users; - - --- Remove UNIQUE constraint from `user_ips` table -CREATE TABLE IF NOT EXISTS new_user_ips ( - user_id TEXT NOT NULL, - access_token TEXT NOT NULL, - device_id TEXT, - ip TEXT NOT NULL, - user_agent TEXT NOT NULL, - last_seen BIGINT UNSIGNED NOT NULL -); - -INSERT INTO new_user_ips - SELECT user, access_token, device_id, ip, user_agent, last_seen FROM user_ips; - -DROP TABLE user_ips; - -ALTER TABLE new_user_ips RENAME TO user_ips; - -CREATE INDEX IF NOT EXISTS user_ips_user ON user_ips(user_id); -CREATE INDEX IF NOT EXISTS user_ips_user_ip ON user_ips(user_id, access_token, ip); - diff --git a/synapse/storage/schema/delta/17/drop_indexes.sql b/synapse/storage/schema/delta/17/drop_indexes.sql deleted file mode 100644 index 7c9a90e27f..0000000000 --- a/synapse/storage/schema/delta/17/drop_indexes.sql +++ /dev/null @@ -1,18 +0,0 @@ -/* Copyright 2015, 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -DROP INDEX IF EXISTS sent_transaction_dest; -DROP INDEX IF EXISTS sent_transaction_sent; -DROP INDEX IF EXISTS user_ips_user; diff --git a/synapse/storage/schema/delta/17/server_keys.sql b/synapse/storage/schema/delta/17/server_keys.sql deleted file mode 100644 index 70b247a06b..0000000000 --- a/synapse/storage/schema/delta/17/server_keys.sql +++ /dev/null @@ -1,24 +0,0 @@ -/* Copyright 2015, 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -CREATE TABLE IF NOT EXISTS server_keys_json ( - server_name TEXT, -- Server name. - key_id TEXT, -- Requested key id. - from_server TEXT, -- Which server the keys were fetched from. - ts_added_ms INTEGER, -- When the keys were fetched - ts_valid_until_ms INTEGER, -- When this version of the keys exipires. - key_json bytea, -- JSON certificate for the remote server. - CONSTRAINT uniqueness UNIQUE (server_name, key_id, from_server) -); diff --git a/synapse/storage/schema/delta/17/user_threepids.sql b/synapse/storage/schema/delta/17/user_threepids.sql deleted file mode 100644 index c17715ac80..0000000000 --- a/synapse/storage/schema/delta/17/user_threepids.sql +++ /dev/null @@ -1,9 +0,0 @@ -CREATE TABLE user_threepids ( - user_id TEXT NOT NULL, - medium TEXT NOT NULL, - address TEXT NOT NULL, - validated_at BIGINT NOT NULL, - added_at BIGINT NOT NULL, - CONSTRAINT user_medium_address UNIQUE (user_id, medium, address) -); -CREATE INDEX user_threepids_user_id ON user_threepids(user_id); diff --git a/synapse/storage/schema/delta/18/server_keys_bigger_ints.sql b/synapse/storage/schema/delta/18/server_keys_bigger_ints.sql deleted file mode 100644 index 6e0871c92b..0000000000 --- a/synapse/storage/schema/delta/18/server_keys_bigger_ints.sql +++ /dev/null @@ -1,32 +0,0 @@ -/* Copyright 2015, 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - -CREATE TABLE IF NOT EXISTS new_server_keys_json ( - server_name TEXT NOT NULL, -- Server name. - key_id TEXT NOT NULL, -- Requested key id. - from_server TEXT NOT NULL, -- Which server the keys were fetched from. - ts_added_ms BIGINT NOT NULL, -- When the keys were fetched - ts_valid_until_ms BIGINT NOT NULL, -- When this version of the keys exipires. - key_json bytea NOT NULL, -- JSON certificate for the remote server. - CONSTRAINT server_keys_json_uniqueness UNIQUE (server_name, key_id, from_server) -); - -INSERT INTO new_server_keys_json - SELECT server_name, key_id, from_server,ts_added_ms, ts_valid_until_ms, key_json FROM server_keys_json ; - -DROP TABLE server_keys_json; - -ALTER TABLE new_server_keys_json RENAME TO server_keys_json; diff --git a/synapse/storage/schema/delta/19/event_index.sql b/synapse/storage/schema/delta/19/event_index.sql deleted file mode 100644 index 18b97b4332..0000000000 --- a/synapse/storage/schema/delta/19/event_index.sql +++ /dev/null @@ -1,19 +0,0 @@ -/* Copyright 2015, 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - -CREATE INDEX events_order_topo_stream_room ON events( - topological_ordering, stream_ordering, room_id -); diff --git a/synapse/storage/schema/delta/20/dummy.sql b/synapse/storage/schema/delta/20/dummy.sql deleted file mode 100644 index e0ac49d1ec..0000000000 --- a/synapse/storage/schema/delta/20/dummy.sql +++ /dev/null @@ -1 +0,0 @@ -SELECT 1; diff --git a/synapse/storage/schema/delta/20/pushers.py b/synapse/storage/schema/delta/20/pushers.py deleted file mode 100644 index 3edfcfd783..0000000000 --- a/synapse/storage/schema/delta/20/pushers.py +++ /dev/null @@ -1,88 +0,0 @@ -# Copyright 2015, 2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - - -""" -Main purpose of this upgrade is to change the unique key on the -pushers table again (it was missed when the v16 full schema was -made) but this also changes the pushkey and data columns to text. -When selecting a bytea column into a text column, postgres inserts -the hex encoded data, and there's no portable way of getting the -UTF-8 bytes, so we have to do it in Python. -""" - -import logging - -logger = logging.getLogger(__name__) - - -def run_create(cur, database_engine, *args, **kwargs): - logger.info("Porting pushers table...") - cur.execute( - """ - CREATE TABLE IF NOT EXISTS pushers2 ( - id BIGINT PRIMARY KEY, - user_name TEXT NOT NULL, - access_token BIGINT DEFAULT NULL, - profile_tag VARCHAR(32) NOT NULL, - kind VARCHAR(8) NOT NULL, - app_id VARCHAR(64) NOT NULL, - app_display_name VARCHAR(64) NOT NULL, - device_display_name VARCHAR(128) NOT NULL, - pushkey TEXT NOT NULL, - ts BIGINT NOT NULL, - lang VARCHAR(8), - data TEXT, - last_token TEXT, - last_success BIGINT, - failing_since BIGINT, - UNIQUE (app_id, pushkey, user_name) - ) - """ - ) - cur.execute( - """SELECT - id, user_name, access_token, profile_tag, kind, - app_id, app_display_name, device_display_name, - pushkey, ts, lang, data, last_token, last_success, - failing_since - FROM pushers - """ - ) - count = 0 - for row in cur.fetchall(): - row = list(row) - row[8] = bytes(row[8]).decode("utf-8") - row[11] = bytes(row[11]).decode("utf-8") - cur.execute( - database_engine.convert_param_style( - """ - INSERT into pushers2 ( - id, user_name, access_token, profile_tag, kind, - app_id, app_display_name, device_display_name, - pushkey, ts, lang, data, last_token, last_success, - failing_since - ) values (%s)""" - % (",".join(["?" for _ in range(len(row))])) - ), - row, - ) - count += 1 - cur.execute("DROP TABLE pushers") - cur.execute("ALTER TABLE pushers2 RENAME TO pushers") - logger.info("Moved %d pushers to new table", count) - - -def run_upgrade(*args, **kwargs): - pass diff --git a/synapse/storage/schema/delta/21/end_to_end_keys.sql b/synapse/storage/schema/delta/21/end_to_end_keys.sql deleted file mode 100644 index 4c2fb20b77..0000000000 --- a/synapse/storage/schema/delta/21/end_to_end_keys.sql +++ /dev/null @@ -1,34 +0,0 @@ -/* Copyright 2015, 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - -CREATE TABLE IF NOT EXISTS e2e_device_keys_json ( - user_id TEXT NOT NULL, -- The user these keys are for. - device_id TEXT NOT NULL, -- Which of the user's devices these keys are for. - ts_added_ms BIGINT NOT NULL, -- When the keys were uploaded. - key_json TEXT NOT NULL, -- The keys for the device as a JSON blob. - CONSTRAINT e2e_device_keys_json_uniqueness UNIQUE (user_id, device_id) -); - - -CREATE TABLE IF NOT EXISTS e2e_one_time_keys_json ( - user_id TEXT NOT NULL, -- The user this one-time key is for. - device_id TEXT NOT NULL, -- The device this one-time key is for. - algorithm TEXT NOT NULL, -- Which algorithm this one-time key is for. - key_id TEXT NOT NULL, -- An id for suppressing duplicate uploads. - ts_added_ms BIGINT NOT NULL, -- When this key was uploaded. - key_json TEXT NOT NULL, -- The key as a JSON blob. - CONSTRAINT e2e_one_time_keys_json_uniqueness UNIQUE (user_id, device_id, algorithm, key_id) -); diff --git a/synapse/storage/schema/delta/21/receipts.sql b/synapse/storage/schema/delta/21/receipts.sql deleted file mode 100644 index d070845477..0000000000 --- a/synapse/storage/schema/delta/21/receipts.sql +++ /dev/null @@ -1,38 +0,0 @@ -/* Copyright 2015, 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - -CREATE TABLE IF NOT EXISTS receipts_graph( - room_id TEXT NOT NULL, - receipt_type TEXT NOT NULL, - user_id TEXT NOT NULL, - event_ids TEXT NOT NULL, - data TEXT NOT NULL, - CONSTRAINT receipts_graph_uniqueness UNIQUE (room_id, receipt_type, user_id) -); - -CREATE TABLE IF NOT EXISTS receipts_linearized ( - stream_id BIGINT NOT NULL, - room_id TEXT NOT NULL, - receipt_type TEXT NOT NULL, - user_id TEXT NOT NULL, - event_id TEXT NOT NULL, - data TEXT NOT NULL, - CONSTRAINT receipts_linearized_uniqueness UNIQUE (room_id, receipt_type, user_id) -); - -CREATE INDEX receipts_linearized_id ON receipts_linearized( - stream_id -); diff --git a/synapse/storage/schema/delta/22/receipts_index.sql b/synapse/storage/schema/delta/22/receipts_index.sql deleted file mode 100644 index bfc0b3bcaa..0000000000 --- a/synapse/storage/schema/delta/22/receipts_index.sql +++ /dev/null @@ -1,22 +0,0 @@ -/* Copyright 2015, 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -/** Using CREATE INDEX directly is deprecated in favour of using background - * update see synapse/storage/schema/delta/33/access_tokens_device_index.sql - * and synapse/storage/registration.py for an example using - * "access_tokens_device_index" **/ -CREATE INDEX receipts_linearized_room_stream ON receipts_linearized( - room_id, stream_id -); diff --git a/synapse/storage/schema/delta/22/user_threepids_unique.sql b/synapse/storage/schema/delta/22/user_threepids_unique.sql deleted file mode 100644 index 87edfa454c..0000000000 --- a/synapse/storage/schema/delta/22/user_threepids_unique.sql +++ /dev/null @@ -1,19 +0,0 @@ -CREATE TABLE IF NOT EXISTS user_threepids2 ( - user_id TEXT NOT NULL, - medium TEXT NOT NULL, - address TEXT NOT NULL, - validated_at BIGINT NOT NULL, - added_at BIGINT NOT NULL, - CONSTRAINT medium_address UNIQUE (medium, address) -); - -INSERT INTO user_threepids2 - SELECT * FROM user_threepids WHERE added_at IN ( - SELECT max(added_at) FROM user_threepids GROUP BY medium, address - ) -; - -DROP TABLE user_threepids; -ALTER TABLE user_threepids2 RENAME TO user_threepids; - -CREATE INDEX user_threepids_user_id ON user_threepids(user_id); diff --git a/synapse/storage/schema/delta/23/drop_state_index.sql b/synapse/storage/schema/delta/23/drop_state_index.sql deleted file mode 100644 index ae09fa0065..0000000000 --- a/synapse/storage/schema/delta/23/drop_state_index.sql +++ /dev/null @@ -1,16 +0,0 @@ -/* Copyright 2015, 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -DROP INDEX IF EXISTS state_groups_state_tuple; diff --git a/synapse/storage/schema/delta/24/stats_reporting.sql b/synapse/storage/schema/delta/24/stats_reporting.sql deleted file mode 100644 index acea7483bd..0000000000 --- a/synapse/storage/schema/delta/24/stats_reporting.sql +++ /dev/null @@ -1,18 +0,0 @@ -/* Copyright 2019 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - /* We used to create a table called stats_reporting, but this is no longer - * used and is removed in delta 54. - */ \ No newline at end of file diff --git a/synapse/storage/schema/delta/25/fts.py b/synapse/storage/schema/delta/25/fts.py deleted file mode 100644 index 4b2ffd35fd..0000000000 --- a/synapse/storage/schema/delta/25/fts.py +++ /dev/null @@ -1,82 +0,0 @@ -# Copyright 2015, 2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import logging - -import simplejson - -from synapse.storage.engines import PostgresEngine, Sqlite3Engine -from synapse.storage.prepare_database import get_statements - -logger = logging.getLogger(__name__) - - -POSTGRES_TABLE = """ -CREATE TABLE IF NOT EXISTS event_search ( - event_id TEXT, - room_id TEXT, - sender TEXT, - key TEXT, - vector tsvector -); - -CREATE INDEX event_search_fts_idx ON event_search USING gin(vector); -CREATE INDEX event_search_ev_idx ON event_search(event_id); -CREATE INDEX event_search_ev_ridx ON event_search(room_id); -""" - - -SQLITE_TABLE = ( - "CREATE VIRTUAL TABLE event_search" - " USING fts4 ( event_id, room_id, sender, key, value )" -) - - -def run_create(cur, database_engine, *args, **kwargs): - if isinstance(database_engine, PostgresEngine): - for statement in get_statements(POSTGRES_TABLE.splitlines()): - cur.execute(statement) - elif isinstance(database_engine, Sqlite3Engine): - cur.execute(SQLITE_TABLE) - else: - raise Exception("Unrecognized database engine") - - cur.execute("SELECT MIN(stream_ordering) FROM events") - rows = cur.fetchall() - min_stream_id = rows[0][0] - - cur.execute("SELECT MAX(stream_ordering) FROM events") - rows = cur.fetchall() - max_stream_id = rows[0][0] - - if min_stream_id is not None and max_stream_id is not None: - progress = { - "target_min_stream_id_inclusive": min_stream_id, - "max_stream_id_exclusive": max_stream_id + 1, - "rows_inserted": 0, - } - progress_json = simplejson.dumps(progress) - - sql = ( - "INSERT into background_updates (update_name, progress_json)" - " VALUES (?, ?)" - ) - - sql = database_engine.convert_param_style(sql) - - cur.execute(sql, ("event_search", progress_json)) - - -def run_upgrade(*args, **kwargs): - pass diff --git a/synapse/storage/schema/delta/25/guest_access.sql b/synapse/storage/schema/delta/25/guest_access.sql deleted file mode 100644 index 1ea389b471..0000000000 --- a/synapse/storage/schema/delta/25/guest_access.sql +++ /dev/null @@ -1,25 +0,0 @@ -/* Copyright 2015, 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -/* - * This is a manual index of guest_access content of state events, - * so that we can join on them in SELECT statements. - */ -CREATE TABLE IF NOT EXISTS guest_access( - event_id TEXT NOT NULL, - room_id TEXT NOT NULL, - guest_access TEXT NOT NULL, - UNIQUE (event_id) -); diff --git a/synapse/storage/schema/delta/25/history_visibility.sql b/synapse/storage/schema/delta/25/history_visibility.sql deleted file mode 100644 index f468fc1897..0000000000 --- a/synapse/storage/schema/delta/25/history_visibility.sql +++ /dev/null @@ -1,25 +0,0 @@ -/* Copyright 2015, 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -/* - * This is a manual index of history_visibility content of state events, - * so that we can join on them in SELECT statements. - */ -CREATE TABLE IF NOT EXISTS history_visibility( - event_id TEXT NOT NULL, - room_id TEXT NOT NULL, - history_visibility TEXT NOT NULL, - UNIQUE (event_id) -); diff --git a/synapse/storage/schema/delta/25/tags.sql b/synapse/storage/schema/delta/25/tags.sql deleted file mode 100644 index 7a32ce68e4..0000000000 --- a/synapse/storage/schema/delta/25/tags.sql +++ /dev/null @@ -1,38 +0,0 @@ -/* Copyright 2015, 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - -CREATE TABLE IF NOT EXISTS room_tags( - user_id TEXT NOT NULL, - room_id TEXT NOT NULL, - tag TEXT NOT NULL, -- The name of the tag. - content TEXT NOT NULL, -- The JSON content of the tag. - CONSTRAINT room_tag_uniqueness UNIQUE (user_id, room_id, tag) -); - -CREATE TABLE IF NOT EXISTS room_tags_revisions ( - user_id TEXT NOT NULL, - room_id TEXT NOT NULL, - stream_id BIGINT NOT NULL, -- The current version of the room tags. - CONSTRAINT room_tag_revisions_uniqueness UNIQUE (user_id, room_id) -); - -CREATE TABLE IF NOT EXISTS private_user_data_max_stream_id( - Lock CHAR(1) NOT NULL DEFAULT 'X' UNIQUE, -- Makes sure this table only has one row. - stream_id BIGINT NOT NULL, - CHECK (Lock='X') -); - -INSERT INTO private_user_data_max_stream_id (stream_id) VALUES (0); diff --git a/synapse/storage/schema/delta/26/account_data.sql b/synapse/storage/schema/delta/26/account_data.sql deleted file mode 100644 index e395de2b5e..0000000000 --- a/synapse/storage/schema/delta/26/account_data.sql +++ /dev/null @@ -1,17 +0,0 @@ -/* Copyright 2015, 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - -ALTER TABLE private_user_data_max_stream_id RENAME TO account_data_max_stream_id; diff --git a/synapse/storage/schema/delta/27/account_data.sql b/synapse/storage/schema/delta/27/account_data.sql deleted file mode 100644 index bf0558b5b3..0000000000 --- a/synapse/storage/schema/delta/27/account_data.sql +++ /dev/null @@ -1,36 +0,0 @@ -/* Copyright 2015, 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -CREATE TABLE IF NOT EXISTS account_data( - user_id TEXT NOT NULL, - account_data_type TEXT NOT NULL, -- The type of the account_data. - stream_id BIGINT NOT NULL, -- The version of the account_data. - content TEXT NOT NULL, -- The JSON content of the account_data - CONSTRAINT account_data_uniqueness UNIQUE (user_id, account_data_type) -); - - -CREATE TABLE IF NOT EXISTS room_account_data( - user_id TEXT NOT NULL, - room_id TEXT NOT NULL, - account_data_type TEXT NOT NULL, -- The type of the account_data. - stream_id BIGINT NOT NULL, -- The version of the account_data. - content TEXT NOT NULL, -- The JSON content of the account_data - CONSTRAINT room_account_data_uniqueness UNIQUE (user_id, room_id, account_data_type) -); - - -CREATE INDEX account_data_stream_id on account_data(user_id, stream_id); -CREATE INDEX room_account_data_stream_id on room_account_data(user_id, stream_id); diff --git a/synapse/storage/schema/delta/27/forgotten_memberships.sql b/synapse/storage/schema/delta/27/forgotten_memberships.sql deleted file mode 100644 index e2094f37fe..0000000000 --- a/synapse/storage/schema/delta/27/forgotten_memberships.sql +++ /dev/null @@ -1,26 +0,0 @@ -/* Copyright 2015, 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -/* - * Keeps track of what rooms users have left and don't want to be able to - * access again. - * - * If all users on this server have left a room, we can delete the room - * entirely. - * - * This column should always contain either 0 or 1. - */ - - ALTER TABLE room_memberships ADD COLUMN forgotten INTEGER DEFAULT 0; diff --git a/synapse/storage/schema/delta/27/ts.py b/synapse/storage/schema/delta/27/ts.py deleted file mode 100644 index 414f9f5aa0..0000000000 --- a/synapse/storage/schema/delta/27/ts.py +++ /dev/null @@ -1,61 +0,0 @@ -# Copyright 2015, 2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import logging - -import simplejson - -from synapse.storage.prepare_database import get_statements - -logger = logging.getLogger(__name__) - - -ALTER_TABLE = ( - "ALTER TABLE events ADD COLUMN origin_server_ts BIGINT;" - "CREATE INDEX events_ts ON events(origin_server_ts, stream_ordering);" -) - - -def run_create(cur, database_engine, *args, **kwargs): - for statement in get_statements(ALTER_TABLE.splitlines()): - cur.execute(statement) - - cur.execute("SELECT MIN(stream_ordering) FROM events") - rows = cur.fetchall() - min_stream_id = rows[0][0] - - cur.execute("SELECT MAX(stream_ordering) FROM events") - rows = cur.fetchall() - max_stream_id = rows[0][0] - - if min_stream_id is not None and max_stream_id is not None: - progress = { - "target_min_stream_id_inclusive": min_stream_id, - "max_stream_id_exclusive": max_stream_id + 1, - "rows_inserted": 0, - } - progress_json = simplejson.dumps(progress) - - sql = ( - "INSERT into background_updates (update_name, progress_json)" - " VALUES (?, ?)" - ) - - sql = database_engine.convert_param_style(sql) - - cur.execute(sql, ("event_origin_server_ts", progress_json)) - - -def run_upgrade(*args, **kwargs): - pass diff --git a/synapse/storage/schema/delta/28/event_push_actions.sql b/synapse/storage/schema/delta/28/event_push_actions.sql deleted file mode 100644 index 4d519849df..0000000000 --- a/synapse/storage/schema/delta/28/event_push_actions.sql +++ /dev/null @@ -1,27 +0,0 @@ -/* Copyright 2015 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -CREATE TABLE IF NOT EXISTS event_push_actions( - room_id TEXT NOT NULL, - event_id TEXT NOT NULL, - user_id TEXT NOT NULL, - profile_tag VARCHAR(32), - actions TEXT NOT NULL, - CONSTRAINT event_id_user_id_profile_tag_uniqueness UNIQUE (room_id, event_id, user_id, profile_tag) -); - - -CREATE INDEX event_push_actions_room_id_event_id_user_id_profile_tag on event_push_actions(room_id, event_id, user_id, profile_tag); -CREATE INDEX event_push_actions_room_id_user_id on event_push_actions(room_id, user_id); diff --git a/synapse/storage/schema/delta/28/events_room_stream.sql b/synapse/storage/schema/delta/28/events_room_stream.sql deleted file mode 100644 index 36609475f1..0000000000 --- a/synapse/storage/schema/delta/28/events_room_stream.sql +++ /dev/null @@ -1,20 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. -*/ - -/** Using CREATE INDEX directly is deprecated in favour of using background - * update see synapse/storage/schema/delta/33/access_tokens_device_index.sql - * and synapse/storage/registration.py for an example using - * "access_tokens_device_index" **/ -CREATE INDEX events_room_stream on events(room_id, stream_ordering); diff --git a/synapse/storage/schema/delta/28/public_roms_index.sql b/synapse/storage/schema/delta/28/public_roms_index.sql deleted file mode 100644 index 6c1fd68c5b..0000000000 --- a/synapse/storage/schema/delta/28/public_roms_index.sql +++ /dev/null @@ -1,20 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. -*/ - -/** Using CREATE INDEX directly is deprecated in favour of using background - * update see synapse/storage/schema/delta/33/access_tokens_device_index.sql - * and synapse/storage/registration.py for an example using - * "access_tokens_device_index" **/ -CREATE INDEX public_room_index on rooms(is_public); diff --git a/synapse/storage/schema/delta/28/receipts_user_id_index.sql b/synapse/storage/schema/delta/28/receipts_user_id_index.sql deleted file mode 100644 index cb84c69baa..0000000000 --- a/synapse/storage/schema/delta/28/receipts_user_id_index.sql +++ /dev/null @@ -1,22 +0,0 @@ -/* Copyright 2015, 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -/** Using CREATE INDEX directly is deprecated in favour of using background - * update see synapse/storage/schema/delta/33/access_tokens_device_index.sql - * and synapse/storage/registration.py for an example using - * "access_tokens_device_index" **/ -CREATE INDEX receipts_linearized_user ON receipts_linearized( - user_id -); diff --git a/synapse/storage/schema/delta/28/upgrade_times.sql b/synapse/storage/schema/delta/28/upgrade_times.sql deleted file mode 100644 index 3e4a9ab455..0000000000 --- a/synapse/storage/schema/delta/28/upgrade_times.sql +++ /dev/null @@ -1,21 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -/* - * Stores the timestamp when a user upgraded from a guest to a full user, if - * that happened. - */ - -ALTER TABLE users ADD COLUMN upgrade_ts BIGINT; diff --git a/synapse/storage/schema/delta/28/users_is_guest.sql b/synapse/storage/schema/delta/28/users_is_guest.sql deleted file mode 100644 index 21d2b420bf..0000000000 --- a/synapse/storage/schema/delta/28/users_is_guest.sql +++ /dev/null @@ -1,22 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -ALTER TABLE users ADD is_guest SMALLINT DEFAULT 0 NOT NULL; -/* - * NB: any guest users created between 27 and 28 will be incorrectly - * marked as not guests: we don't bother to fill these in correctly - * because guest access is not really complete in 27 anyway so it's - * very unlikley there will be any guest users created. - */ diff --git a/synapse/storage/schema/delta/29/push_actions.sql b/synapse/storage/schema/delta/29/push_actions.sql deleted file mode 100644 index 84b21cf813..0000000000 --- a/synapse/storage/schema/delta/29/push_actions.sql +++ /dev/null @@ -1,35 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -ALTER TABLE event_push_actions ADD COLUMN topological_ordering BIGINT; -ALTER TABLE event_push_actions ADD COLUMN stream_ordering BIGINT; -ALTER TABLE event_push_actions ADD COLUMN notif SMALLINT; -ALTER TABLE event_push_actions ADD COLUMN highlight SMALLINT; - -UPDATE event_push_actions SET stream_ordering = ( - SELECT stream_ordering FROM events WHERE event_id = event_push_actions.event_id -), topological_ordering = ( - SELECT topological_ordering FROM events WHERE event_id = event_push_actions.event_id -); - -UPDATE event_push_actions SET notif = 1, highlight = 0; - -/** Using CREATE INDEX directly is deprecated in favour of using background - * update see synapse/storage/schema/delta/33/access_tokens_device_index.sql - * and synapse/storage/registration.py for an example using - * "access_tokens_device_index" **/ -CREATE INDEX event_push_actions_rm_tokens on event_push_actions( - user_id, room_id, topological_ordering, stream_ordering -); diff --git a/synapse/storage/schema/delta/30/alias_creator.sql b/synapse/storage/schema/delta/30/alias_creator.sql deleted file mode 100644 index c9d0dde638..0000000000 --- a/synapse/storage/schema/delta/30/alias_creator.sql +++ /dev/null @@ -1,16 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -ALTER TABLE room_aliases ADD COLUMN creator TEXT; diff --git a/synapse/storage/schema/delta/30/as_users.py b/synapse/storage/schema/delta/30/as_users.py deleted file mode 100644 index 9b95411fb6..0000000000 --- a/synapse/storage/schema/delta/30/as_users.py +++ /dev/null @@ -1,69 +0,0 @@ -# Copyright 2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. -import logging - -from six.moves import range - -from synapse.config.appservice import load_appservices - -logger = logging.getLogger(__name__) - - -def run_create(cur, database_engine, *args, **kwargs): - # NULL indicates user was not registered by an appservice. - try: - cur.execute("ALTER TABLE users ADD COLUMN appservice_id TEXT") - except Exception: - # Maybe we already added the column? Hope so... - pass - - -def run_upgrade(cur, database_engine, config, *args, **kwargs): - cur.execute("SELECT name FROM users") - rows = cur.fetchall() - - config_files = [] - try: - config_files = config.app_service_config_files - except AttributeError: - logger.warning("Could not get app_service_config_files from config") - pass - - appservices = load_appservices(config.server_name, config_files) - - owned = {} - - for row in rows: - user_id = row[0] - for appservice in appservices: - if appservice.is_exclusive_user(user_id): - if user_id in owned.keys(): - logger.error( - "user_id %s was owned by more than one application" - " service (IDs %s and %s); assigning arbitrarily to %s" - % (user_id, owned[user_id], appservice.id, owned[user_id]) - ) - owned.setdefault(appservice.id, []).append(user_id) - - for as_id, user_ids in owned.items(): - n = 100 - user_chunks = (user_ids[i : i + 100] for i in range(0, len(user_ids), n)) - for chunk in user_chunks: - cur.execute( - database_engine.convert_param_style( - "UPDATE users SET appservice_id = ? WHERE name IN (%s)" - % (",".join("?" for _ in chunk),) - ), - [as_id] + chunk, - ) diff --git a/synapse/storage/schema/delta/30/deleted_pushers.sql b/synapse/storage/schema/delta/30/deleted_pushers.sql deleted file mode 100644 index 712c454aa1..0000000000 --- a/synapse/storage/schema/delta/30/deleted_pushers.sql +++ /dev/null @@ -1,25 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -CREATE TABLE IF NOT EXISTS deleted_pushers( - stream_id BIGINT NOT NULL, - app_id TEXT NOT NULL, - pushkey TEXT NOT NULL, - user_id TEXT NOT NULL, - /* We only track the most recent delete for each app_id, pushkey and user_id. */ - UNIQUE (app_id, pushkey, user_id) -); - -CREATE INDEX deleted_pushers_stream_id ON deleted_pushers (stream_id); diff --git a/synapse/storage/schema/delta/30/presence_stream.sql b/synapse/storage/schema/delta/30/presence_stream.sql deleted file mode 100644 index 606bbb037d..0000000000 --- a/synapse/storage/schema/delta/30/presence_stream.sql +++ /dev/null @@ -1,30 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - - CREATE TABLE presence_stream( - stream_id BIGINT, - user_id TEXT, - state TEXT, - last_active_ts BIGINT, - last_federation_update_ts BIGINT, - last_user_sync_ts BIGINT, - status_msg TEXT, - currently_active BOOLEAN - ); - - CREATE INDEX presence_stream_id ON presence_stream(stream_id, user_id); - CREATE INDEX presence_stream_user_id ON presence_stream(user_id); - CREATE INDEX presence_stream_state ON presence_stream(state); diff --git a/synapse/storage/schema/delta/30/public_rooms.sql b/synapse/storage/schema/delta/30/public_rooms.sql deleted file mode 100644 index f09db4faa6..0000000000 --- a/synapse/storage/schema/delta/30/public_rooms.sql +++ /dev/null @@ -1,23 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - -/* This release removes the restriction that published rooms must have an alias, - * so we go back and ensure the only 'public' rooms are ones with an alias. - * We use (1 = 0) and (1 = 1) so that it works in both postgres and sqlite - */ -UPDATE rooms SET is_public = (1 = 0) WHERE is_public = (1 = 1) AND room_id not in ( - SELECT room_id FROM room_aliases -); diff --git a/synapse/storage/schema/delta/30/push_rule_stream.sql b/synapse/storage/schema/delta/30/push_rule_stream.sql deleted file mode 100644 index 735aa8d5f6..0000000000 --- a/synapse/storage/schema/delta/30/push_rule_stream.sql +++ /dev/null @@ -1,38 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - - -CREATE TABLE push_rules_stream( - stream_id BIGINT NOT NULL, - event_stream_ordering BIGINT NOT NULL, - user_id TEXT NOT NULL, - rule_id TEXT NOT NULL, - op TEXT NOT NULL, -- One of "ENABLE", "DISABLE", "ACTIONS", "ADD", "DELETE" - priority_class SMALLINT, - priority INTEGER, - conditions TEXT, - actions TEXT -); - --- The extra data for each operation is: --- * ENABLE, DISABLE, DELETE: [] --- * ACTIONS: ["actions"] --- * ADD: ["priority_class", "priority", "actions", "conditions"] - --- Index for replication queries. -CREATE INDEX push_rules_stream_id ON push_rules_stream(stream_id); --- Index for /sync queries. -CREATE INDEX push_rules_stream_user_stream_id on push_rules_stream(user_id, stream_id); diff --git a/synapse/storage/schema/delta/30/state_stream.sql b/synapse/storage/schema/delta/30/state_stream.sql deleted file mode 100644 index e85699e82e..0000000000 --- a/synapse/storage/schema/delta/30/state_stream.sql +++ /dev/null @@ -1,33 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - -/* We used to create a table called current_state_resets, but this is no - * longer used and is removed in delta 54. - */ - -/* The outlier events that have aquired a state group typically through - * backfill. This is tracked separately to the events table, as assigning a - * state group change the position of the existing event in the stream - * ordering. - * However since a stream_ordering is assigned in persist_event for the - * (event, state) pair, we can use that stream_ordering to identify when - * the new state was assigned for the event. - */ -CREATE TABLE IF NOT EXISTS ex_outlier_stream( - event_stream_ordering BIGINT PRIMARY KEY NOT NULL, - event_id TEXT NOT NULL, - state_group BIGINT NOT NULL -); diff --git a/synapse/storage/schema/delta/30/threepid_guest_access_tokens.sql b/synapse/storage/schema/delta/30/threepid_guest_access_tokens.sql deleted file mode 100644 index 0dd2f1360c..0000000000 --- a/synapse/storage/schema/delta/30/threepid_guest_access_tokens.sql +++ /dev/null @@ -1,24 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- Stores guest account access tokens generated for unbound 3pids. -CREATE TABLE threepid_guest_access_tokens( - medium TEXT, -- The medium of the 3pid. Must be "email". - address TEXT, -- The 3pid address. - guest_access_token TEXT, -- The access token for a guest user for this 3pid. - first_inviter TEXT -- User ID of the first user to invite this 3pid to a room. -); - -CREATE UNIQUE INDEX threepid_guest_access_tokens_index ON threepid_guest_access_tokens(medium, address); diff --git a/synapse/storage/schema/delta/31/invites.sql b/synapse/storage/schema/delta/31/invites.sql deleted file mode 100644 index 2c57846d5a..0000000000 --- a/synapse/storage/schema/delta/31/invites.sql +++ /dev/null @@ -1,42 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - -CREATE TABLE local_invites( - stream_id BIGINT NOT NULL, - inviter TEXT NOT NULL, - invitee TEXT NOT NULL, - event_id TEXT NOT NULL, - room_id TEXT NOT NULL, - locally_rejected TEXT, - replaced_by TEXT -); - --- Insert all invites for local users into new `invites` table -INSERT INTO local_invites SELECT - stream_ordering as stream_id, - sender as inviter, - state_key as invitee, - event_id, - room_id, - NULL as locally_rejected, - NULL as replaced_by - FROM events - NATURAL JOIN current_state_events - NATURAL JOIN room_memberships - WHERE membership = 'invite' AND state_key IN (SELECT name FROM users); - -CREATE INDEX local_invites_id ON local_invites(stream_id); -CREATE INDEX local_invites_for_user_idx ON local_invites(invitee, locally_rejected, replaced_by, room_id); diff --git a/synapse/storage/schema/delta/31/local_media_repository_url_cache.sql b/synapse/storage/schema/delta/31/local_media_repository_url_cache.sql deleted file mode 100644 index 9efb4280eb..0000000000 --- a/synapse/storage/schema/delta/31/local_media_repository_url_cache.sql +++ /dev/null @@ -1,27 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -CREATE TABLE local_media_repository_url_cache( - url TEXT, -- the URL being cached - response_code INTEGER, -- the HTTP response code of this download attempt - etag TEXT, -- the etag header of this response - expires INTEGER, -- the number of ms this response was valid for - og TEXT, -- cache of the OG metadata of this URL as JSON - media_id TEXT, -- the media_id, if any, of the URL's content in the repo - download_ts BIGINT -- the timestamp of this download attempt -); - -CREATE INDEX local_media_repository_url_cache_by_url_download_ts - ON local_media_repository_url_cache(url, download_ts); diff --git a/synapse/storage/schema/delta/31/pushers.py b/synapse/storage/schema/delta/31/pushers.py deleted file mode 100644 index 9bb504aad5..0000000000 --- a/synapse/storage/schema/delta/31/pushers.py +++ /dev/null @@ -1,87 +0,0 @@ -# Copyright 2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - - -# Change the last_token to last_stream_ordering now that pushers no longer -# listen on an event stream but instead select out of the event_push_actions -# table. - - -import logging - -logger = logging.getLogger(__name__) - - -def token_to_stream_ordering(token): - return int(token[1:].split("_")[0]) - - -def run_create(cur, database_engine, *args, **kwargs): - logger.info("Porting pushers table, delta 31...") - cur.execute( - """ - CREATE TABLE IF NOT EXISTS pushers2 ( - id BIGINT PRIMARY KEY, - user_name TEXT NOT NULL, - access_token BIGINT DEFAULT NULL, - profile_tag VARCHAR(32) NOT NULL, - kind VARCHAR(8) NOT NULL, - app_id VARCHAR(64) NOT NULL, - app_display_name VARCHAR(64) NOT NULL, - device_display_name VARCHAR(128) NOT NULL, - pushkey TEXT NOT NULL, - ts BIGINT NOT NULL, - lang VARCHAR(8), - data TEXT, - last_stream_ordering INTEGER, - last_success BIGINT, - failing_since BIGINT, - UNIQUE (app_id, pushkey, user_name) - ) - """ - ) - cur.execute( - """SELECT - id, user_name, access_token, profile_tag, kind, - app_id, app_display_name, device_display_name, - pushkey, ts, lang, data, last_token, last_success, - failing_since - FROM pushers - """ - ) - count = 0 - for row in cur.fetchall(): - row = list(row) - row[12] = token_to_stream_ordering(row[12]) - cur.execute( - database_engine.convert_param_style( - """ - INSERT into pushers2 ( - id, user_name, access_token, profile_tag, kind, - app_id, app_display_name, device_display_name, - pushkey, ts, lang, data, last_stream_ordering, last_success, - failing_since - ) values (%s)""" - % (",".join(["?" for _ in range(len(row))])) - ), - row, - ) - count += 1 - cur.execute("DROP TABLE pushers") - cur.execute("ALTER TABLE pushers2 RENAME TO pushers") - logger.info("Moved %d pushers to new table", count) - - -def run_upgrade(cur, database_engine, *args, **kwargs): - pass diff --git a/synapse/storage/schema/delta/31/pushers_index.sql b/synapse/storage/schema/delta/31/pushers_index.sql deleted file mode 100644 index a82add88fd..0000000000 --- a/synapse/storage/schema/delta/31/pushers_index.sql +++ /dev/null @@ -1,22 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -/** Using CREATE INDEX directly is deprecated in favour of using background - * update see synapse/storage/schema/delta/33/access_tokens_device_index.sql - * and synapse/storage/registration.py for an example using - * "access_tokens_device_index" **/ - CREATE INDEX event_push_actions_stream_ordering on event_push_actions( - stream_ordering, user_id - ); diff --git a/synapse/storage/schema/delta/31/search_update.py b/synapse/storage/schema/delta/31/search_update.py deleted file mode 100644 index 7d8ca5f93f..0000000000 --- a/synapse/storage/schema/delta/31/search_update.py +++ /dev/null @@ -1,66 +0,0 @@ -# Copyright 2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import logging - -import simplejson - -from synapse.storage.engines import PostgresEngine -from synapse.storage.prepare_database import get_statements - -logger = logging.getLogger(__name__) - - -ALTER_TABLE = """ -ALTER TABLE event_search ADD COLUMN origin_server_ts BIGINT; -ALTER TABLE event_search ADD COLUMN stream_ordering BIGINT; -""" - - -def run_create(cur, database_engine, *args, **kwargs): - if not isinstance(database_engine, PostgresEngine): - return - - for statement in get_statements(ALTER_TABLE.splitlines()): - cur.execute(statement) - - cur.execute("SELECT MIN(stream_ordering) FROM events") - rows = cur.fetchall() - min_stream_id = rows[0][0] - - cur.execute("SELECT MAX(stream_ordering) FROM events") - rows = cur.fetchall() - max_stream_id = rows[0][0] - - if min_stream_id is not None and max_stream_id is not None: - progress = { - "target_min_stream_id_inclusive": min_stream_id, - "max_stream_id_exclusive": max_stream_id + 1, - "rows_inserted": 0, - "have_added_indexes": False, - } - progress_json = simplejson.dumps(progress) - - sql = ( - "INSERT into background_updates (update_name, progress_json)" - " VALUES (?, ?)" - ) - - sql = database_engine.convert_param_style(sql) - - cur.execute(sql, ("event_search_order", progress_json)) - - -def run_upgrade(cur, database_engine, *args, **kwargs): - pass diff --git a/synapse/storage/schema/delta/32/events.sql b/synapse/storage/schema/delta/32/events.sql deleted file mode 100644 index 1dd0f9e170..0000000000 --- a/synapse/storage/schema/delta/32/events.sql +++ /dev/null @@ -1,16 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -ALTER TABLE events ADD COLUMN received_ts BIGINT; diff --git a/synapse/storage/schema/delta/32/openid.sql b/synapse/storage/schema/delta/32/openid.sql deleted file mode 100644 index 36f37b11c8..0000000000 --- a/synapse/storage/schema/delta/32/openid.sql +++ /dev/null @@ -1,9 +0,0 @@ - -CREATE TABLE open_id_tokens ( - token TEXT NOT NULL PRIMARY KEY, - ts_valid_until_ms bigint NOT NULL, - user_id TEXT NOT NULL, - UNIQUE (token) -); - -CREATE index open_id_tokens_ts_valid_until_ms ON open_id_tokens(ts_valid_until_ms); diff --git a/synapse/storage/schema/delta/32/pusher_throttle.sql b/synapse/storage/schema/delta/32/pusher_throttle.sql deleted file mode 100644 index d86d30c13c..0000000000 --- a/synapse/storage/schema/delta/32/pusher_throttle.sql +++ /dev/null @@ -1,23 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - -CREATE TABLE pusher_throttle( - pusher BIGINT NOT NULL, - room_id TEXT NOT NULL, - last_sent_ts BIGINT, - throttle_ms BIGINT, - PRIMARY KEY (pusher, room_id) -); diff --git a/synapse/storage/schema/delta/32/remove_indices.sql b/synapse/storage/schema/delta/32/remove_indices.sql deleted file mode 100644 index 4219cdd06a..0000000000 --- a/synapse/storage/schema/delta/32/remove_indices.sql +++ /dev/null @@ -1,34 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - --- The following indices are redundant, other indices are equivalent or --- supersets -DROP INDEX IF EXISTS events_room_id; -- Prefix of events_room_stream -DROP INDEX IF EXISTS events_order; -- Prefix of events_order_topo_stream_room -DROP INDEX IF EXISTS events_topological_ordering; -- Prefix of events_order_topo_stream_room -DROP INDEX IF EXISTS events_stream_ordering; -- Duplicate of PRIMARY KEY -DROP INDEX IF EXISTS state_groups_id; -- Duplicate of PRIMARY KEY -DROP INDEX IF EXISTS event_to_state_groups_id; -- Duplicate of PRIMARY KEY -DROP INDEX IF EXISTS event_push_actions_room_id_event_id_user_id_profile_tag; -- Duplicate of UNIQUE CONSTRAINT - -DROP INDEX IF EXISTS st_extrem_id; -- Prefix of UNIQUE CONSTRAINT -DROP INDEX IF EXISTS event_signatures_id; -- Prefix of UNIQUE CONSTRAINT -DROP INDEX IF EXISTS redactions_event_id; -- Duplicate of UNIQUE CONSTRAINT - --- The following indices were unused -DROP INDEX IF EXISTS remote_media_cache_thumbnails_media_id; -DROP INDEX IF EXISTS evauth_edges_auth_id; -DROP INDEX IF EXISTS presence_stream_state; diff --git a/synapse/storage/schema/delta/32/reports.sql b/synapse/storage/schema/delta/32/reports.sql deleted file mode 100644 index d13609776f..0000000000 --- a/synapse/storage/schema/delta/32/reports.sql +++ /dev/null @@ -1,25 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - -CREATE TABLE event_reports( - id BIGINT NOT NULL PRIMARY KEY, - received_ts BIGINT NOT NULL, - room_id TEXT NOT NULL, - event_id TEXT NOT NULL, - user_id TEXT NOT NULL, - reason TEXT, - content TEXT -); diff --git a/synapse/storage/schema/delta/33/access_tokens_device_index.sql b/synapse/storage/schema/delta/33/access_tokens_device_index.sql deleted file mode 100644 index 61ad3fe3e8..0000000000 --- a/synapse/storage/schema/delta/33/access_tokens_device_index.sql +++ /dev/null @@ -1,17 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -INSERT INTO background_updates (update_name, progress_json) VALUES - ('access_tokens_device_index', '{}'); diff --git a/synapse/storage/schema/delta/33/devices.sql b/synapse/storage/schema/delta/33/devices.sql deleted file mode 100644 index eca7268d82..0000000000 --- a/synapse/storage/schema/delta/33/devices.sql +++ /dev/null @@ -1,21 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -CREATE TABLE devices ( - user_id TEXT NOT NULL, - device_id TEXT NOT NULL, - display_name TEXT, - CONSTRAINT device_uniqueness UNIQUE (user_id, device_id) -); diff --git a/synapse/storage/schema/delta/33/devices_for_e2e_keys.sql b/synapse/storage/schema/delta/33/devices_for_e2e_keys.sql deleted file mode 100644 index aa4a3b9f2f..0000000000 --- a/synapse/storage/schema/delta/33/devices_for_e2e_keys.sql +++ /dev/null @@ -1,19 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- make sure that we have a device record for each set of E2E keys, so that the --- user can delete them if they like. -INSERT INTO devices - SELECT user_id, device_id, NULL FROM e2e_device_keys_json; diff --git a/synapse/storage/schema/delta/33/devices_for_e2e_keys_clear_unknown_device.sql b/synapse/storage/schema/delta/33/devices_for_e2e_keys_clear_unknown_device.sql deleted file mode 100644 index 6671573398..0000000000 --- a/synapse/storage/schema/delta/33/devices_for_e2e_keys_clear_unknown_device.sql +++ /dev/null @@ -1,20 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- a previous version of the "devices_for_e2e_keys" delta set all the device --- names to "unknown device". This wasn't terribly helpful -UPDATE devices - SET display_name = NULL - WHERE display_name = 'unknown device'; diff --git a/synapse/storage/schema/delta/33/event_fields.py b/synapse/storage/schema/delta/33/event_fields.py deleted file mode 100644 index bff1256a7b..0000000000 --- a/synapse/storage/schema/delta/33/event_fields.py +++ /dev/null @@ -1,61 +0,0 @@ -# Copyright 2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import logging - -import simplejson - -from synapse.storage.prepare_database import get_statements - -logger = logging.getLogger(__name__) - - -ALTER_TABLE = """ -ALTER TABLE events ADD COLUMN sender TEXT; -ALTER TABLE events ADD COLUMN contains_url BOOLEAN; -""" - - -def run_create(cur, database_engine, *args, **kwargs): - for statement in get_statements(ALTER_TABLE.splitlines()): - cur.execute(statement) - - cur.execute("SELECT MIN(stream_ordering) FROM events") - rows = cur.fetchall() - min_stream_id = rows[0][0] - - cur.execute("SELECT MAX(stream_ordering) FROM events") - rows = cur.fetchall() - max_stream_id = rows[0][0] - - if min_stream_id is not None and max_stream_id is not None: - progress = { - "target_min_stream_id_inclusive": min_stream_id, - "max_stream_id_exclusive": max_stream_id + 1, - "rows_inserted": 0, - } - progress_json = simplejson.dumps(progress) - - sql = ( - "INSERT into background_updates (update_name, progress_json)" - " VALUES (?, ?)" - ) - - sql = database_engine.convert_param_style(sql) - - cur.execute(sql, ("event_fields_sender_url", progress_json)) - - -def run_upgrade(cur, database_engine, *args, **kwargs): - pass diff --git a/synapse/storage/schema/delta/33/remote_media_ts.py b/synapse/storage/schema/delta/33/remote_media_ts.py deleted file mode 100644 index a26057dfb6..0000000000 --- a/synapse/storage/schema/delta/33/remote_media_ts.py +++ /dev/null @@ -1,30 +0,0 @@ -# Copyright 2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import time - -ALTER_TABLE = "ALTER TABLE remote_media_cache ADD COLUMN last_access_ts BIGINT" - - -def run_create(cur, database_engine, *args, **kwargs): - cur.execute(ALTER_TABLE) - - -def run_upgrade(cur, database_engine, *args, **kwargs): - cur.execute( - database_engine.convert_param_style( - "UPDATE remote_media_cache SET last_access_ts = ?" - ), - (int(time.time() * 1000),), - ) diff --git a/synapse/storage/schema/delta/33/user_ips_index.sql b/synapse/storage/schema/delta/33/user_ips_index.sql deleted file mode 100644 index 473f75a78e..0000000000 --- a/synapse/storage/schema/delta/33/user_ips_index.sql +++ /dev/null @@ -1,17 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -INSERT INTO background_updates (update_name, progress_json) VALUES - ('user_ips_device_index', '{}'); diff --git a/synapse/storage/schema/delta/34/appservice_stream.sql b/synapse/storage/schema/delta/34/appservice_stream.sql deleted file mode 100644 index 69e16eda0f..0000000000 --- a/synapse/storage/schema/delta/34/appservice_stream.sql +++ /dev/null @@ -1,23 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -CREATE TABLE IF NOT EXISTS appservice_stream_position( - Lock CHAR(1) NOT NULL DEFAULT 'X' UNIQUE, -- Makes sure this table only has one row. - stream_ordering BIGINT, - CHECK (Lock='X') -); - -INSERT INTO appservice_stream_position (stream_ordering) - SELECT COALESCE(MAX(stream_ordering), 0) FROM events; diff --git a/synapse/storage/schema/delta/34/cache_stream.py b/synapse/storage/schema/delta/34/cache_stream.py deleted file mode 100644 index cf09e43e2b..0000000000 --- a/synapse/storage/schema/delta/34/cache_stream.py +++ /dev/null @@ -1,46 +0,0 @@ -# Copyright 2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import logging - -from synapse.storage.engines import PostgresEngine -from synapse.storage.prepare_database import get_statements - -logger = logging.getLogger(__name__) - - -# This stream is used to notify replication slaves that some caches have -# been invalidated that they cannot infer from the other streams. -CREATE_TABLE = """ -CREATE TABLE cache_invalidation_stream ( - stream_id BIGINT, - cache_func TEXT, - keys TEXT[], - invalidation_ts BIGINT -); - -CREATE INDEX cache_invalidation_stream_id ON cache_invalidation_stream(stream_id); -""" - - -def run_create(cur, database_engine, *args, **kwargs): - if not isinstance(database_engine, PostgresEngine): - return - - for statement in get_statements(CREATE_TABLE.splitlines()): - cur.execute(statement) - - -def run_upgrade(cur, database_engine, *args, **kwargs): - pass diff --git a/synapse/storage/schema/delta/34/device_inbox.sql b/synapse/storage/schema/delta/34/device_inbox.sql deleted file mode 100644 index e68844c74a..0000000000 --- a/synapse/storage/schema/delta/34/device_inbox.sql +++ /dev/null @@ -1,24 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -CREATE TABLE device_inbox ( - user_id TEXT NOT NULL, - device_id TEXT NOT NULL, - stream_id BIGINT NOT NULL, - message_json TEXT NOT NULL -- {"type":, "sender":, "content",} -); - -CREATE INDEX device_inbox_user_stream_id ON device_inbox(user_id, device_id, stream_id); -CREATE INDEX device_inbox_stream_id ON device_inbox(stream_id); diff --git a/synapse/storage/schema/delta/34/push_display_name_rename.sql b/synapse/storage/schema/delta/34/push_display_name_rename.sql deleted file mode 100644 index 0d9fe1a99a..0000000000 --- a/synapse/storage/schema/delta/34/push_display_name_rename.sql +++ /dev/null @@ -1,20 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -DELETE FROM push_rules WHERE rule_id = 'global/override/.m.rule.contains_display_name'; -UPDATE push_rules SET rule_id = 'global/override/.m.rule.contains_display_name' WHERE rule_id = 'global/underride/.m.rule.contains_display_name'; - -DELETE FROM push_rules_enable WHERE rule_id = 'global/override/.m.rule.contains_display_name'; -UPDATE push_rules_enable SET rule_id = 'global/override/.m.rule.contains_display_name' WHERE rule_id = 'global/underride/.m.rule.contains_display_name'; diff --git a/synapse/storage/schema/delta/34/received_txn_purge.py b/synapse/storage/schema/delta/34/received_txn_purge.py deleted file mode 100644 index 67d505e68b..0000000000 --- a/synapse/storage/schema/delta/34/received_txn_purge.py +++ /dev/null @@ -1,32 +0,0 @@ -# Copyright 2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import logging - -from synapse.storage.engines import PostgresEngine - -logger = logging.getLogger(__name__) - - -def run_create(cur, database_engine, *args, **kwargs): - if isinstance(database_engine, PostgresEngine): - cur.execute("TRUNCATE received_transactions") - else: - cur.execute("DELETE FROM received_transactions") - - cur.execute("CREATE INDEX received_transactions_ts ON received_transactions(ts)") - - -def run_upgrade(cur, database_engine, *args, **kwargs): - pass diff --git a/synapse/storage/schema/delta/35/00background_updates_add_col.sql b/synapse/storage/schema/delta/35/00background_updates_add_col.sql new file mode 100644 index 0000000000..c2d2a4f836 --- /dev/null +++ b/synapse/storage/schema/delta/35/00background_updates_add_col.sql @@ -0,0 +1,17 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + +ALTER TABLE background_updates ADD COLUMN depends_on TEXT; diff --git a/synapse/storage/schema/delta/35/add_state_index.sql b/synapse/storage/schema/delta/35/add_state_index.sql deleted file mode 100644 index 0fce26345b..0000000000 --- a/synapse/storage/schema/delta/35/add_state_index.sql +++ /dev/null @@ -1,20 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - -ALTER TABLE background_updates ADD COLUMN depends_on TEXT; - -INSERT into background_updates (update_name, progress_json, depends_on) - VALUES ('state_group_state_type_index', '{}', 'state_group_state_deduplication'); diff --git a/synapse/storage/schema/delta/35/contains_url.sql b/synapse/storage/schema/delta/35/contains_url.sql deleted file mode 100644 index 6cd123027b..0000000000 --- a/synapse/storage/schema/delta/35/contains_url.sql +++ /dev/null @@ -1,17 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - INSERT into background_updates (update_name, progress_json) - VALUES ('event_contains_url_index', '{}'); diff --git a/synapse/storage/schema/delta/35/device_outbox.sql b/synapse/storage/schema/delta/35/device_outbox.sql deleted file mode 100644 index 17e6c43105..0000000000 --- a/synapse/storage/schema/delta/35/device_outbox.sql +++ /dev/null @@ -1,39 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -DROP TABLE IF EXISTS device_federation_outbox; -CREATE TABLE device_federation_outbox ( - destination TEXT NOT NULL, - stream_id BIGINT NOT NULL, - queued_ts BIGINT NOT NULL, - messages_json TEXT NOT NULL -); - - -DROP INDEX IF EXISTS device_federation_outbox_destination_id; -CREATE INDEX device_federation_outbox_destination_id - ON device_federation_outbox(destination, stream_id); - - -DROP TABLE IF EXISTS device_federation_inbox; -CREATE TABLE device_federation_inbox ( - origin TEXT NOT NULL, - message_id TEXT NOT NULL, - received_ts BIGINT NOT NULL -); - -DROP INDEX IF EXISTS device_federation_inbox_sender_id; -CREATE INDEX device_federation_inbox_sender_id - ON device_federation_inbox(origin, message_id); diff --git a/synapse/storage/schema/delta/35/device_stream_id.sql b/synapse/storage/schema/delta/35/device_stream_id.sql deleted file mode 100644 index 7ab7d942e2..0000000000 --- a/synapse/storage/schema/delta/35/device_stream_id.sql +++ /dev/null @@ -1,21 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -CREATE TABLE device_max_stream_id ( - stream_id BIGINT NOT NULL -); - -INSERT INTO device_max_stream_id (stream_id) - SELECT COALESCE(MAX(stream_id), 0) FROM device_inbox; diff --git a/synapse/storage/schema/delta/35/event_push_actions_index.sql b/synapse/storage/schema/delta/35/event_push_actions_index.sql deleted file mode 100644 index 2e836d8e9c..0000000000 --- a/synapse/storage/schema/delta/35/event_push_actions_index.sql +++ /dev/null @@ -1,17 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - INSERT into background_updates (update_name, progress_json) - VALUES ('epa_highlight_index', '{}'); diff --git a/synapse/storage/schema/delta/35/public_room_list_change_stream.sql b/synapse/storage/schema/delta/35/public_room_list_change_stream.sql deleted file mode 100644 index dd2bf2e28a..0000000000 --- a/synapse/storage/schema/delta/35/public_room_list_change_stream.sql +++ /dev/null @@ -1,33 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - -CREATE TABLE public_room_list_stream ( - stream_id BIGINT NOT NULL, - room_id TEXT NOT NULL, - visibility BOOLEAN NOT NULL -); - -INSERT INTO public_room_list_stream (stream_id, room_id, visibility) - SELECT 1, room_id, is_public FROM rooms - WHERE is_public = CAST(1 AS BOOLEAN); - -CREATE INDEX public_room_list_stream_idx on public_room_list_stream( - stream_id -); - -CREATE INDEX public_room_list_stream_rm_idx on public_room_list_stream( - room_id, stream_id -); diff --git a/synapse/storage/schema/delta/35/state.sql b/synapse/storage/schema/delta/35/state.sql deleted file mode 100644 index 0f1fa68a89..0000000000 --- a/synapse/storage/schema/delta/35/state.sql +++ /dev/null @@ -1,22 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -CREATE TABLE state_group_edges( - state_group BIGINT NOT NULL, - prev_state_group BIGINT NOT NULL -); - -CREATE INDEX state_group_edges_idx ON state_group_edges(state_group); -CREATE INDEX state_group_edges_prev_idx ON state_group_edges(prev_state_group); diff --git a/synapse/storage/schema/delta/35/state_dedupe.sql b/synapse/storage/schema/delta/35/state_dedupe.sql deleted file mode 100644 index 97e5067ef4..0000000000 --- a/synapse/storage/schema/delta/35/state_dedupe.sql +++ /dev/null @@ -1,17 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -INSERT into background_updates (update_name, progress_json) - VALUES ('state_group_state_deduplication', '{}'); diff --git a/synapse/storage/schema/delta/35/stream_order_to_extrem.sql b/synapse/storage/schema/delta/35/stream_order_to_extrem.sql deleted file mode 100644 index 2b945d8a57..0000000000 --- a/synapse/storage/schema/delta/35/stream_order_to_extrem.sql +++ /dev/null @@ -1,37 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - -CREATE TABLE stream_ordering_to_exterm ( - stream_ordering BIGINT NOT NULL, - room_id TEXT NOT NULL, - event_id TEXT NOT NULL -); - -INSERT INTO stream_ordering_to_exterm (stream_ordering, room_id, event_id) - SELECT stream_ordering, room_id, event_id FROM event_forward_extremities - INNER JOIN ( - SELECT room_id, max(stream_ordering) as stream_ordering FROM events - INNER JOIN event_forward_extremities USING (room_id, event_id) - GROUP BY room_id - ) AS rms USING (room_id); - -CREATE INDEX stream_ordering_to_exterm_idx on stream_ordering_to_exterm( - stream_ordering -); - -CREATE INDEX stream_ordering_to_exterm_rm_idx on stream_ordering_to_exterm( - room_id, stream_ordering -); diff --git a/synapse/storage/schema/delta/36/readd_public_rooms.sql b/synapse/storage/schema/delta/36/readd_public_rooms.sql deleted file mode 100644 index 90d8fd18f9..0000000000 --- a/synapse/storage/schema/delta/36/readd_public_rooms.sql +++ /dev/null @@ -1,26 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- Re-add some entries to stream_ordering_to_exterm that were incorrectly deleted -INSERT INTO stream_ordering_to_exterm (stream_ordering, room_id, event_id) - SELECT - (SELECT stream_ordering FROM events where event_id = e.event_id) AS stream_ordering, - room_id, - event_id - FROM event_forward_extremities AS e - WHERE NOT EXISTS ( - SELECT room_id FROM stream_ordering_to_exterm AS s - WHERE s.room_id = e.room_id - ); diff --git a/synapse/storage/schema/delta/37/remove_auth_idx.py b/synapse/storage/schema/delta/37/remove_auth_idx.py deleted file mode 100644 index a377884169..0000000000 --- a/synapse/storage/schema/delta/37/remove_auth_idx.py +++ /dev/null @@ -1,85 +0,0 @@ -# Copyright 2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import logging - -from synapse.storage.engines import PostgresEngine -from synapse.storage.prepare_database import get_statements - -logger = logging.getLogger(__name__) - -DROP_INDICES = """ --- We only ever query based on event_id -DROP INDEX IF EXISTS state_events_room_id; -DROP INDEX IF EXISTS state_events_type; -DROP INDEX IF EXISTS state_events_state_key; - --- room_id is indexed elsewhere -DROP INDEX IF EXISTS current_state_events_room_id; -DROP INDEX IF EXISTS current_state_events_state_key; -DROP INDEX IF EXISTS current_state_events_type; - -DROP INDEX IF EXISTS transactions_have_ref; - --- (topological_ordering, stream_ordering, room_id) seems like a strange index, --- and is used incredibly rarely. -DROP INDEX IF EXISTS events_order_topo_stream_room; - --- an equivalent index to this actually gets re-created in delta 41, because it --- turned out that deleting it wasn't a great plan :/. In any case, let's --- delete it here, and delta 41 will create a new one with an added UNIQUE --- constraint -DROP INDEX IF EXISTS event_search_ev_idx; -""" - -POSTGRES_DROP_CONSTRAINT = """ -ALTER TABLE event_auth DROP CONSTRAINT IF EXISTS event_auth_event_id_auth_id_room_id_key; -""" - -SQLITE_DROP_CONSTRAINT = """ -DROP INDEX IF EXISTS evauth_edges_id; - -CREATE TABLE IF NOT EXISTS event_auth_new( - event_id TEXT NOT NULL, - auth_id TEXT NOT NULL, - room_id TEXT NOT NULL -); - -INSERT INTO event_auth_new - SELECT event_id, auth_id, room_id - FROM event_auth; - -DROP TABLE event_auth; - -ALTER TABLE event_auth_new RENAME TO event_auth; - -CREATE INDEX evauth_edges_id ON event_auth(event_id); -""" - - -def run_create(cur, database_engine, *args, **kwargs): - for statement in get_statements(DROP_INDICES.splitlines()): - cur.execute(statement) - - if isinstance(database_engine, PostgresEngine): - drop_constraint = POSTGRES_DROP_CONSTRAINT - else: - drop_constraint = SQLITE_DROP_CONSTRAINT - - for statement in get_statements(drop_constraint.splitlines()): - cur.execute(statement) - - -def run_upgrade(cur, database_engine, *args, **kwargs): - pass diff --git a/synapse/storage/schema/delta/37/user_threepids.sql b/synapse/storage/schema/delta/37/user_threepids.sql deleted file mode 100644 index cf7a90dd10..0000000000 --- a/synapse/storage/schema/delta/37/user_threepids.sql +++ /dev/null @@ -1,52 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -/* - * Update any email addresses that were stored with mixed case into all - * lowercase - */ - - -- There may be "duplicate" emails (with different case) already in the table, - -- so we find them and move all but the most recently used account. - UPDATE user_threepids - SET medium = 'email_old' - WHERE medium = 'email' - AND address IN ( - -- We select all the addresses that are linked to the user_id that is NOT - -- the most recently created. - SELECT u.address - FROM - user_threepids AS u, - -- `duplicate_addresses` is a table of all the email addresses that - -- appear multiple times and when the binding was created - ( - SELECT lower(u1.address) AS address, max(u1.added_at) AS max_ts - FROM user_threepids AS u1 - INNER JOIN user_threepids AS u2 ON u1.medium = u2.medium AND lower(u1.address) = lower(u2.address) AND u1.address != u2.address - WHERE u1.medium = 'email' AND u2.medium = 'email' - GROUP BY lower(u1.address) - ) AS duplicate_addresses - WHERE - lower(u.address) = duplicate_addresses.address - AND u.added_at != max_ts -- NOT the most recently created - ); - - --- This update is now safe since we've removed the duplicate addresses. -UPDATE user_threepids SET address = LOWER(address) WHERE medium = 'email'; - - -/* Add an index for the select we do on passwored reset */ -CREATE INDEX user_threepids_medium_address on user_threepids (medium, address); diff --git a/synapse/storage/schema/delta/38/postgres_fts_gist.sql b/synapse/storage/schema/delta/38/postgres_fts_gist.sql deleted file mode 100644 index 515e6b8e84..0000000000 --- a/synapse/storage/schema/delta/38/postgres_fts_gist.sql +++ /dev/null @@ -1,19 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- We no longer do this given we back it out again in schema 47 - --- INSERT into background_updates (update_name, progress_json) --- VALUES ('event_search_postgres_gist', '{}'); diff --git a/synapse/storage/schema/delta/39/appservice_room_list.sql b/synapse/storage/schema/delta/39/appservice_room_list.sql deleted file mode 100644 index 74bdc49073..0000000000 --- a/synapse/storage/schema/delta/39/appservice_room_list.sql +++ /dev/null @@ -1,29 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -CREATE TABLE appservice_room_list( - appservice_id TEXT NOT NULL, - network_id TEXT NOT NULL, - room_id TEXT NOT NULL -); - --- Each appservice can have multiple published room lists associated with them, --- keyed of a particular network_id -CREATE UNIQUE INDEX appservice_room_list_idx ON appservice_room_list( - appservice_id, network_id, room_id -); - -ALTER TABLE public_room_list_stream ADD COLUMN appservice_id TEXT; -ALTER TABLE public_room_list_stream ADD COLUMN network_id TEXT; diff --git a/synapse/storage/schema/delta/39/device_federation_stream_idx.sql b/synapse/storage/schema/delta/39/device_federation_stream_idx.sql deleted file mode 100644 index 00be801e90..0000000000 --- a/synapse/storage/schema/delta/39/device_federation_stream_idx.sql +++ /dev/null @@ -1,16 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -CREATE INDEX device_federation_outbox_id ON device_federation_outbox(stream_id); diff --git a/synapse/storage/schema/delta/39/event_push_index.sql b/synapse/storage/schema/delta/39/event_push_index.sql deleted file mode 100644 index de2ad93e5c..0000000000 --- a/synapse/storage/schema/delta/39/event_push_index.sql +++ /dev/null @@ -1,17 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -INSERT INTO background_updates (update_name, progress_json) VALUES - ('event_push_actions_highlights_index', '{}'); diff --git a/synapse/storage/schema/delta/39/federation_out_position.sql b/synapse/storage/schema/delta/39/federation_out_position.sql deleted file mode 100644 index 5af814290b..0000000000 --- a/synapse/storage/schema/delta/39/federation_out_position.sql +++ /dev/null @@ -1,22 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - CREATE TABLE federation_stream_position( - type TEXT NOT NULL, - stream_id INTEGER NOT NULL - ); - - INSERT INTO federation_stream_position (type, stream_id) VALUES ('federation', -1); - INSERT INTO federation_stream_position (type, stream_id) SELECT 'events', coalesce(max(stream_ordering), -1) FROM events; diff --git a/synapse/storage/schema/delta/39/membership_profile.sql b/synapse/storage/schema/delta/39/membership_profile.sql deleted file mode 100644 index 1bf911c8ab..0000000000 --- a/synapse/storage/schema/delta/39/membership_profile.sql +++ /dev/null @@ -1,20 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -ALTER TABLE room_memberships ADD COLUMN display_name TEXT; -ALTER TABLE room_memberships ADD COLUMN avatar_url TEXT; - -INSERT into background_updates (update_name, progress_json) - VALUES ('room_membership_profile_update', '{}'); diff --git a/synapse/storage/schema/delta/40/current_state_idx.sql b/synapse/storage/schema/delta/40/current_state_idx.sql deleted file mode 100644 index 7ffa189f39..0000000000 --- a/synapse/storage/schema/delta/40/current_state_idx.sql +++ /dev/null @@ -1,17 +0,0 @@ -/* Copyright 2017 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -INSERT INTO background_updates (update_name, progress_json) VALUES - ('current_state_members_idx', '{}'); diff --git a/synapse/storage/schema/delta/40/device_inbox.sql b/synapse/storage/schema/delta/40/device_inbox.sql deleted file mode 100644 index b9fe1f0480..0000000000 --- a/synapse/storage/schema/delta/40/device_inbox.sql +++ /dev/null @@ -1,21 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- turn the pre-fill startup query into a index-only scan on postgresql. -INSERT into background_updates (update_name, progress_json) - VALUES ('device_inbox_stream_index', '{}'); - -INSERT into background_updates (update_name, progress_json, depends_on) - VALUES ('device_inbox_stream_drop', '{}', 'device_inbox_stream_index'); diff --git a/synapse/storage/schema/delta/40/device_list_streams.sql b/synapse/storage/schema/delta/40/device_list_streams.sql deleted file mode 100644 index dd6dcb65f1..0000000000 --- a/synapse/storage/schema/delta/40/device_list_streams.sql +++ /dev/null @@ -1,60 +0,0 @@ -/* Copyright 2017 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- Cache of remote devices. -CREATE TABLE device_lists_remote_cache ( - user_id TEXT NOT NULL, - device_id TEXT NOT NULL, - content TEXT NOT NULL -); - --- The last update we got for a user. Empty if we're not receiving updates for --- that user. -CREATE TABLE device_lists_remote_extremeties ( - user_id TEXT NOT NULL, - stream_id TEXT NOT NULL -); - --- we used to create non-unique indexes on these tables, but as of update 52 we create --- unique indexes concurrently: --- --- CREATE INDEX device_lists_remote_cache_id ON device_lists_remote_cache(user_id, device_id); --- CREATE INDEX device_lists_remote_extremeties_id ON device_lists_remote_extremeties(user_id, stream_id); - - --- Stream of device lists updates. Includes both local and remotes -CREATE TABLE device_lists_stream ( - stream_id BIGINT NOT NULL, - user_id TEXT NOT NULL, - device_id TEXT NOT NULL -); - -CREATE INDEX device_lists_stream_id ON device_lists_stream(stream_id, user_id); - - --- The stream of updates to send to other servers. We keep at least one row --- per user that was sent so that the prev_id for any new updates can be --- calculated -CREATE TABLE device_lists_outbound_pokes ( - destination TEXT NOT NULL, - stream_id BIGINT NOT NULL, - user_id TEXT NOT NULL, - device_id TEXT NOT NULL, - sent BOOLEAN NOT NULL, - ts BIGINT NOT NULL -- So that in future we can clear out pokes to dead servers -); - -CREATE INDEX device_lists_outbound_pokes_id ON device_lists_outbound_pokes(destination, stream_id); -CREATE INDEX device_lists_outbound_pokes_user ON device_lists_outbound_pokes(destination, user_id); diff --git a/synapse/storage/schema/delta/40/event_push_summary.sql b/synapse/storage/schema/delta/40/event_push_summary.sql deleted file mode 100644 index 3918f0b794..0000000000 --- a/synapse/storage/schema/delta/40/event_push_summary.sql +++ /dev/null @@ -1,37 +0,0 @@ -/* Copyright 2017 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- Aggregate of old notification counts that have been deleted out of the --- main event_push_actions table. This count does not include those that were --- highlights, as they remain in the event_push_actions table. -CREATE TABLE event_push_summary ( - user_id TEXT NOT NULL, - room_id TEXT NOT NULL, - notif_count BIGINT NOT NULL, - stream_ordering BIGINT NOT NULL -); - -CREATE INDEX event_push_summary_user_rm ON event_push_summary(user_id, room_id); - - --- The stream ordering up to which we have aggregated the event_push_actions --- table into event_push_summary -CREATE TABLE event_push_summary_stream_ordering ( - Lock CHAR(1) NOT NULL DEFAULT 'X' UNIQUE, -- Makes sure this table only has one row. - stream_ordering BIGINT NOT NULL, - CHECK (Lock='X') -); - -INSERT INTO event_push_summary_stream_ordering (stream_ordering) VALUES (0); diff --git a/synapse/storage/schema/delta/40/pushers.sql b/synapse/storage/schema/delta/40/pushers.sql deleted file mode 100644 index 054a223f14..0000000000 --- a/synapse/storage/schema/delta/40/pushers.sql +++ /dev/null @@ -1,39 +0,0 @@ -/* Copyright 2017 Vector Creations Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -CREATE TABLE IF NOT EXISTS pushers2 ( - id BIGINT PRIMARY KEY, - user_name TEXT NOT NULL, - access_token BIGINT DEFAULT NULL, - profile_tag TEXT NOT NULL, - kind TEXT NOT NULL, - app_id TEXT NOT NULL, - app_display_name TEXT NOT NULL, - device_display_name TEXT NOT NULL, - pushkey TEXT NOT NULL, - ts BIGINT NOT NULL, - lang TEXT, - data TEXT, - last_stream_ordering INTEGER, - last_success BIGINT, - failing_since BIGINT, - UNIQUE (app_id, pushkey, user_name) -); - -INSERT INTO pushers2 SELECT * FROM PUSHERS; - -DROP TABLE PUSHERS; - -ALTER TABLE pushers2 RENAME TO pushers; diff --git a/synapse/storage/schema/delta/41/device_list_stream_idx.sql b/synapse/storage/schema/delta/41/device_list_stream_idx.sql deleted file mode 100644 index b7bee8b692..0000000000 --- a/synapse/storage/schema/delta/41/device_list_stream_idx.sql +++ /dev/null @@ -1,17 +0,0 @@ -/* Copyright 2017 Vector Creations Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -INSERT into background_updates (update_name, progress_json) - VALUES ('device_lists_stream_idx', '{}'); diff --git a/synapse/storage/schema/delta/41/device_outbound_index.sql b/synapse/storage/schema/delta/41/device_outbound_index.sql deleted file mode 100644 index 62f0b9892b..0000000000 --- a/synapse/storage/schema/delta/41/device_outbound_index.sql +++ /dev/null @@ -1,16 +0,0 @@ -/* Copyright 2017 Vector Creations Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -CREATE INDEX device_lists_outbound_pokes_stream ON device_lists_outbound_pokes(stream_id); diff --git a/synapse/storage/schema/delta/41/event_search_event_id_idx.sql b/synapse/storage/schema/delta/41/event_search_event_id_idx.sql deleted file mode 100644 index 5d9cfecf36..0000000000 --- a/synapse/storage/schema/delta/41/event_search_event_id_idx.sql +++ /dev/null @@ -1,17 +0,0 @@ -/* Copyright 2017 Vector Creations Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -INSERT into background_updates (update_name, progress_json) - VALUES ('event_search_event_id_idx', '{}'); diff --git a/synapse/storage/schema/delta/41/ratelimit.sql b/synapse/storage/schema/delta/41/ratelimit.sql deleted file mode 100644 index a194bf0238..0000000000 --- a/synapse/storage/schema/delta/41/ratelimit.sql +++ /dev/null @@ -1,22 +0,0 @@ -/* Copyright 2017 Vector Creations Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -CREATE TABLE ratelimit_override ( - user_id TEXT NOT NULL, - messages_per_second BIGINT, - burst_count BIGINT -); - -CREATE UNIQUE INDEX ratelimit_override_idx ON ratelimit_override(user_id); diff --git a/synapse/storage/schema/delta/42/current_state_delta.sql b/synapse/storage/schema/delta/42/current_state_delta.sql deleted file mode 100644 index d28851aff8..0000000000 --- a/synapse/storage/schema/delta/42/current_state_delta.sql +++ /dev/null @@ -1,26 +0,0 @@ -/* Copyright 2017 Vector Creations Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - -CREATE TABLE current_state_delta_stream ( - stream_id BIGINT NOT NULL, - room_id TEXT NOT NULL, - type TEXT NOT NULL, - state_key TEXT NOT NULL, - event_id TEXT, -- Is null if the key was removed - prev_event_id TEXT -- Is null if the key was added -); - -CREATE INDEX current_state_delta_stream_idx ON current_state_delta_stream(stream_id); diff --git a/synapse/storage/schema/delta/42/device_list_last_id.sql b/synapse/storage/schema/delta/42/device_list_last_id.sql deleted file mode 100644 index 9ab8c14fa3..0000000000 --- a/synapse/storage/schema/delta/42/device_list_last_id.sql +++ /dev/null @@ -1,33 +0,0 @@ -/* Copyright 2017 Vector Creations Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - --- Table of last stream_id that we sent to destination for user_id. This is --- used to fill out the `prev_id` fields of outbound device list updates. -CREATE TABLE device_lists_outbound_last_success ( - destination TEXT NOT NULL, - user_id TEXT NOT NULL, - stream_id BIGINT NOT NULL -); - -INSERT INTO device_lists_outbound_last_success - SELECT destination, user_id, coalesce(max(stream_id), 0) as stream_id - FROM device_lists_outbound_pokes - WHERE sent = (1 = 1) -- sqlite doesn't have inbuilt boolean values - GROUP BY destination, user_id; - -CREATE INDEX device_lists_outbound_last_success_idx ON device_lists_outbound_last_success( - destination, user_id, stream_id -); diff --git a/synapse/storage/schema/delta/42/event_auth_state_only.sql b/synapse/storage/schema/delta/42/event_auth_state_only.sql deleted file mode 100644 index b8821ac759..0000000000 --- a/synapse/storage/schema/delta/42/event_auth_state_only.sql +++ /dev/null @@ -1,17 +0,0 @@ -/* Copyright 2017 Vector Creations Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -INSERT INTO background_updates (update_name, progress_json) VALUES - ('event_auth_state_only', '{}'); diff --git a/synapse/storage/schema/delta/42/user_dir.py b/synapse/storage/schema/delta/42/user_dir.py deleted file mode 100644 index 506f326f4d..0000000000 --- a/synapse/storage/schema/delta/42/user_dir.py +++ /dev/null @@ -1,84 +0,0 @@ -# Copyright 2017 Vector Creations Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import logging - -from synapse.storage.engines import PostgresEngine, Sqlite3Engine -from synapse.storage.prepare_database import get_statements - -logger = logging.getLogger(__name__) - - -BOTH_TABLES = """ -CREATE TABLE user_directory_stream_pos ( - Lock CHAR(1) NOT NULL DEFAULT 'X' UNIQUE, -- Makes sure this table only has one row. - stream_id BIGINT, - CHECK (Lock='X') -); - -INSERT INTO user_directory_stream_pos (stream_id) VALUES (null); - -CREATE TABLE user_directory ( - user_id TEXT NOT NULL, - room_id TEXT NOT NULL, -- A room_id that we know the user is joined to - display_name TEXT, - avatar_url TEXT -); - -CREATE INDEX user_directory_room_idx ON user_directory(room_id); -CREATE UNIQUE INDEX user_directory_user_idx ON user_directory(user_id); - -CREATE TABLE users_in_pubic_room ( - user_id TEXT NOT NULL, - room_id TEXT NOT NULL -- A room_id that we know is public -); - -CREATE INDEX users_in_pubic_room_room_idx ON users_in_pubic_room(room_id); -CREATE UNIQUE INDEX users_in_pubic_room_user_idx ON users_in_pubic_room(user_id); -""" - - -POSTGRES_TABLE = """ -CREATE TABLE user_directory_search ( - user_id TEXT NOT NULL, - vector tsvector -); - -CREATE INDEX user_directory_search_fts_idx ON user_directory_search USING gin(vector); -CREATE UNIQUE INDEX user_directory_search_user_idx ON user_directory_search(user_id); -""" - - -SQLITE_TABLE = """ -CREATE VIRTUAL TABLE user_directory_search - USING fts4 ( user_id, value ); -""" - - -def run_create(cur, database_engine, *args, **kwargs): - for statement in get_statements(BOTH_TABLES.splitlines()): - cur.execute(statement) - - if isinstance(database_engine, PostgresEngine): - for statement in get_statements(POSTGRES_TABLE.splitlines()): - cur.execute(statement) - elif isinstance(database_engine, Sqlite3Engine): - for statement in get_statements(SQLITE_TABLE.splitlines()): - cur.execute(statement) - else: - raise Exception("Unrecognized database engine") - - -def run_upgrade(*args, **kwargs): - pass diff --git a/synapse/storage/schema/delta/43/blocked_rooms.sql b/synapse/storage/schema/delta/43/blocked_rooms.sql deleted file mode 100644 index 0e3cd143ff..0000000000 --- a/synapse/storage/schema/delta/43/blocked_rooms.sql +++ /dev/null @@ -1,21 +0,0 @@ -/* Copyright 2017 Vector Creations Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -CREATE TABLE blocked_rooms ( - room_id TEXT NOT NULL, - user_id TEXT NOT NULL -- Admin who blocked the room -); - -CREATE UNIQUE INDEX blocked_rooms_idx ON blocked_rooms(room_id); diff --git a/synapse/storage/schema/delta/43/quarantine_media.sql b/synapse/storage/schema/delta/43/quarantine_media.sql deleted file mode 100644 index 630907ec4f..0000000000 --- a/synapse/storage/schema/delta/43/quarantine_media.sql +++ /dev/null @@ -1,17 +0,0 @@ -/* Copyright 2017 Vector Creations Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -ALTER TABLE local_media_repository ADD COLUMN quarantined_by TEXT; -ALTER TABLE remote_media_cache ADD COLUMN quarantined_by TEXT; diff --git a/synapse/storage/schema/delta/43/url_cache.sql b/synapse/storage/schema/delta/43/url_cache.sql deleted file mode 100644 index 45ebe020da..0000000000 --- a/synapse/storage/schema/delta/43/url_cache.sql +++ /dev/null @@ -1,16 +0,0 @@ -/* Copyright 2017 Vector Creations Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -ALTER TABLE local_media_repository ADD COLUMN url_cache TEXT; diff --git a/synapse/storage/schema/delta/43/user_share.sql b/synapse/storage/schema/delta/43/user_share.sql deleted file mode 100644 index ee7062abe4..0000000000 --- a/synapse/storage/schema/delta/43/user_share.sql +++ /dev/null @@ -1,33 +0,0 @@ -/* Copyright 2017 Vector Creations Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- Table keeping track of who shares a room with who. We only keep track --- of this for local users, so `user_id` is local users only (but we do keep track --- of which remote users share a room) -CREATE TABLE users_who_share_rooms ( - user_id TEXT NOT NULL, - other_user_id TEXT NOT NULL, - room_id TEXT NOT NULL, - share_private BOOLEAN NOT NULL -- is the shared room private? i.e. they share a private room -); - - -CREATE UNIQUE INDEX users_who_share_rooms_u_idx ON users_who_share_rooms(user_id, other_user_id); -CREATE INDEX users_who_share_rooms_r_idx ON users_who_share_rooms(room_id); -CREATE INDEX users_who_share_rooms_o_idx ON users_who_share_rooms(other_user_id); - - --- Make sure that we populate the table initially -UPDATE user_directory_stream_pos SET stream_id = NULL; diff --git a/synapse/storage/schema/delta/44/expire_url_cache.sql b/synapse/storage/schema/delta/44/expire_url_cache.sql deleted file mode 100644 index b12f9b2ebf..0000000000 --- a/synapse/storage/schema/delta/44/expire_url_cache.sql +++ /dev/null @@ -1,41 +0,0 @@ -/* Copyright 2017 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- this didn't work on SQLite 3.7 (because of lack of partial indexes), so was --- removed and replaced with 46/local_media_repository_url_idx.sql. --- --- CREATE INDEX local_media_repository_url_idx ON local_media_repository(created_ts) WHERE url_cache IS NOT NULL; - --- we need to change `expires` to `expires_ts` so that we can index on it. SQLite doesn't support --- indices on expressions until 3.9. -CREATE TABLE local_media_repository_url_cache_new( - url TEXT, - response_code INTEGER, - etag TEXT, - expires_ts BIGINT, - og TEXT, - media_id TEXT, - download_ts BIGINT -); - -INSERT INTO local_media_repository_url_cache_new - SELECT url, response_code, etag, expires + download_ts, og, media_id, download_ts FROM local_media_repository_url_cache; - -DROP TABLE local_media_repository_url_cache; -ALTER TABLE local_media_repository_url_cache_new RENAME TO local_media_repository_url_cache; - -CREATE INDEX local_media_repository_url_cache_expires_idx ON local_media_repository_url_cache(expires_ts); -CREATE INDEX local_media_repository_url_cache_by_url_download_ts ON local_media_repository_url_cache(url, download_ts); -CREATE INDEX local_media_repository_url_cache_media_idx ON local_media_repository_url_cache(media_id); diff --git a/synapse/storage/schema/delta/45/group_server.sql b/synapse/storage/schema/delta/45/group_server.sql deleted file mode 100644 index b2333848a0..0000000000 --- a/synapse/storage/schema/delta/45/group_server.sql +++ /dev/null @@ -1,167 +0,0 @@ -/* Copyright 2017 Vector Creations Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -CREATE TABLE groups ( - group_id TEXT NOT NULL, - name TEXT, -- the display name of the room - avatar_url TEXT, - short_description TEXT, - long_description TEXT -); - -CREATE UNIQUE INDEX groups_idx ON groups(group_id); - - --- list of users the group server thinks are joined -CREATE TABLE group_users ( - group_id TEXT NOT NULL, - user_id TEXT NOT NULL, - is_admin BOOLEAN NOT NULL, - is_public BOOLEAN NOT NULL -- whether the users membership can be seen by everyone -); - - -CREATE INDEX groups_users_g_idx ON group_users(group_id, user_id); -CREATE INDEX groups_users_u_idx ON group_users(user_id); - --- list of users the group server thinks are invited -CREATE TABLE group_invites ( - group_id TEXT NOT NULL, - user_id TEXT NOT NULL -); - -CREATE INDEX groups_invites_g_idx ON group_invites(group_id, user_id); -CREATE INDEX groups_invites_u_idx ON group_invites(user_id); - - -CREATE TABLE group_rooms ( - group_id TEXT NOT NULL, - room_id TEXT NOT NULL, - is_public BOOLEAN NOT NULL -- whether the room can be seen by everyone -); - -CREATE UNIQUE INDEX groups_rooms_g_idx ON group_rooms(group_id, room_id); -CREATE INDEX groups_rooms_r_idx ON group_rooms(room_id); - - --- Rooms to include in the summary -CREATE TABLE group_summary_rooms ( - group_id TEXT NOT NULL, - room_id TEXT NOT NULL, - category_id TEXT NOT NULL, - room_order BIGINT NOT NULL, - is_public BOOLEAN NOT NULL, -- whether the room should be show to everyone - UNIQUE (group_id, category_id, room_id, room_order), - CHECK (room_order > 0) -); - -CREATE UNIQUE INDEX group_summary_rooms_g_idx ON group_summary_rooms(group_id, room_id, category_id); - - --- Categories to include in the summary -CREATE TABLE group_summary_room_categories ( - group_id TEXT NOT NULL, - category_id TEXT NOT NULL, - cat_order BIGINT NOT NULL, - UNIQUE (group_id, category_id, cat_order), - CHECK (cat_order > 0) -); - --- The categories in the group -CREATE TABLE group_room_categories ( - group_id TEXT NOT NULL, - category_id TEXT NOT NULL, - profile TEXT NOT NULL, - is_public BOOLEAN NOT NULL, -- whether the category should be show to everyone - UNIQUE (group_id, category_id) -); - --- The users to include in the group summary -CREATE TABLE group_summary_users ( - group_id TEXT NOT NULL, - user_id TEXT NOT NULL, - role_id TEXT NOT NULL, - user_order BIGINT NOT NULL, - is_public BOOLEAN NOT NULL -- whether the user should be show to everyone -); - -CREATE INDEX group_summary_users_g_idx ON group_summary_users(group_id); - --- The roles to include in the group summary -CREATE TABLE group_summary_roles ( - group_id TEXT NOT NULL, - role_id TEXT NOT NULL, - role_order BIGINT NOT NULL, - UNIQUE (group_id, role_id, role_order), - CHECK (role_order > 0) -); - - --- The roles in a groups -CREATE TABLE group_roles ( - group_id TEXT NOT NULL, - role_id TEXT NOT NULL, - profile TEXT NOT NULL, - is_public BOOLEAN NOT NULL, -- whether the role should be show to everyone - UNIQUE (group_id, role_id) -); - - --- List of attestations we've given out and need to renew -CREATE TABLE group_attestations_renewals ( - group_id TEXT NOT NULL, - user_id TEXT NOT NULL, - valid_until_ms BIGINT NOT NULL -); - -CREATE INDEX group_attestations_renewals_g_idx ON group_attestations_renewals(group_id, user_id); -CREATE INDEX group_attestations_renewals_u_idx ON group_attestations_renewals(user_id); -CREATE INDEX group_attestations_renewals_v_idx ON group_attestations_renewals(valid_until_ms); - - --- List of attestations we've received from remotes and are interested in. -CREATE TABLE group_attestations_remote ( - group_id TEXT NOT NULL, - user_id TEXT NOT NULL, - valid_until_ms BIGINT NOT NULL, - attestation_json TEXT NOT NULL -); - -CREATE INDEX group_attestations_remote_g_idx ON group_attestations_remote(group_id, user_id); -CREATE INDEX group_attestations_remote_u_idx ON group_attestations_remote(user_id); -CREATE INDEX group_attestations_remote_v_idx ON group_attestations_remote(valid_until_ms); - - --- The group membership for the HS's users -CREATE TABLE local_group_membership ( - group_id TEXT NOT NULL, - user_id TEXT NOT NULL, - is_admin BOOLEAN NOT NULL, - membership TEXT NOT NULL, - is_publicised BOOLEAN NOT NULL, -- if the user is publicising their membership - content TEXT NOT NULL -); - -CREATE INDEX local_group_membership_u_idx ON local_group_membership(user_id, group_id); -CREATE INDEX local_group_membership_g_idx ON local_group_membership(group_id); - - -CREATE TABLE local_group_updates ( - stream_id BIGINT NOT NULL, - group_id TEXT NOT NULL, - user_id TEXT NOT NULL, - type TEXT NOT NULL, - content TEXT NOT NULL -); diff --git a/synapse/storage/schema/delta/45/profile_cache.sql b/synapse/storage/schema/delta/45/profile_cache.sql deleted file mode 100644 index e5ddc84df0..0000000000 --- a/synapse/storage/schema/delta/45/profile_cache.sql +++ /dev/null @@ -1,28 +0,0 @@ -/* Copyright 2017 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - --- A subset of remote users whose profiles we have cached. --- Whether a user is in this table or not is defined by the storage function --- `is_subscribed_remote_profile_for_user` -CREATE TABLE remote_profile_cache ( - user_id TEXT NOT NULL, - displayname TEXT, - avatar_url TEXT, - last_check BIGINT NOT NULL -); - -CREATE UNIQUE INDEX remote_profile_cache_user_id ON remote_profile_cache(user_id); -CREATE INDEX remote_profile_cache_time ON remote_profile_cache(last_check); diff --git a/synapse/storage/schema/delta/46/drop_refresh_tokens.sql b/synapse/storage/schema/delta/46/drop_refresh_tokens.sql deleted file mode 100644 index 68c48a89a9..0000000000 --- a/synapse/storage/schema/delta/46/drop_refresh_tokens.sql +++ /dev/null @@ -1,17 +0,0 @@ -/* Copyright 2017 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -/* we no longer use (or create) the refresh_tokens table */ -DROP TABLE IF EXISTS refresh_tokens; diff --git a/synapse/storage/schema/delta/46/drop_unique_deleted_pushers.sql b/synapse/storage/schema/delta/46/drop_unique_deleted_pushers.sql deleted file mode 100644 index bb307889c1..0000000000 --- a/synapse/storage/schema/delta/46/drop_unique_deleted_pushers.sql +++ /dev/null @@ -1,35 +0,0 @@ -/* Copyright 2017 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- drop the unique constraint on deleted_pushers so that we can just insert --- into it rather than upserting. - -CREATE TABLE deleted_pushers2 ( - stream_id BIGINT NOT NULL, - app_id TEXT NOT NULL, - pushkey TEXT NOT NULL, - user_id TEXT NOT NULL -); - -INSERT INTO deleted_pushers2 (stream_id, app_id, pushkey, user_id) - SELECT stream_id, app_id, pushkey, user_id from deleted_pushers; - -DROP TABLE deleted_pushers; -ALTER TABLE deleted_pushers2 RENAME TO deleted_pushers; - --- create the index after doing the inserts because that's more efficient. --- it also means we can give it the same name as the old one without renaming. -CREATE INDEX deleted_pushers_stream_id ON deleted_pushers (stream_id); - diff --git a/synapse/storage/schema/delta/46/group_server.sql b/synapse/storage/schema/delta/46/group_server.sql deleted file mode 100644 index 097679bc9a..0000000000 --- a/synapse/storage/schema/delta/46/group_server.sql +++ /dev/null @@ -1,32 +0,0 @@ -/* Copyright 2017 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -CREATE TABLE groups_new ( - group_id TEXT NOT NULL, - name TEXT, -- the display name of the room - avatar_url TEXT, - short_description TEXT, - long_description TEXT, - is_public BOOL NOT NULL -- whether non-members can access group APIs -); - --- NB: awful hack to get the default to be true on postgres and 1 on sqlite -INSERT INTO groups_new - SELECT group_id, name, avatar_url, short_description, long_description, (1=1) FROM groups; - -DROP TABLE groups; -ALTER TABLE groups_new RENAME TO groups; - -CREATE UNIQUE INDEX groups_idx ON groups(group_id); diff --git a/synapse/storage/schema/delta/46/local_media_repository_url_idx.sql b/synapse/storage/schema/delta/46/local_media_repository_url_idx.sql deleted file mode 100644 index bbfc7f5d1a..0000000000 --- a/synapse/storage/schema/delta/46/local_media_repository_url_idx.sql +++ /dev/null @@ -1,24 +0,0 @@ -/* Copyright 2017 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- register a background update which will recreate the --- local_media_repository_url_idx index. --- --- We do this as a bg update not because it is a particularly onerous --- operation, but because we'd like it to be a partial index if possible, and --- the background_index_update code will understand whether we are on --- postgres or sqlite and behave accordingly. -INSERT INTO background_updates (update_name, progress_json) VALUES - ('local_media_repository_url_idx', '{}'); diff --git a/synapse/storage/schema/delta/46/user_dir_null_room_ids.sql b/synapse/storage/schema/delta/46/user_dir_null_room_ids.sql deleted file mode 100644 index cb0d5a2576..0000000000 --- a/synapse/storage/schema/delta/46/user_dir_null_room_ids.sql +++ /dev/null @@ -1,35 +0,0 @@ -/* Copyright 2017 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- change the user_directory table to also cover global local user profiles --- rather than just profiles within specific rooms. - -CREATE TABLE user_directory2 ( - user_id TEXT NOT NULL, - room_id TEXT, - display_name TEXT, - avatar_url TEXT -); - -INSERT INTO user_directory2(user_id, room_id, display_name, avatar_url) - SELECT user_id, room_id, display_name, avatar_url from user_directory; - -DROP TABLE user_directory; -ALTER TABLE user_directory2 RENAME TO user_directory; - --- create indexes after doing the inserts because that's more efficient. --- it also means we can give it the same name as the old one without renaming. -CREATE INDEX user_directory_room_idx ON user_directory(room_id); -CREATE UNIQUE INDEX user_directory_user_idx ON user_directory(user_id); diff --git a/synapse/storage/schema/delta/46/user_dir_typos.sql b/synapse/storage/schema/delta/46/user_dir_typos.sql deleted file mode 100644 index d9505f8da1..0000000000 --- a/synapse/storage/schema/delta/46/user_dir_typos.sql +++ /dev/null @@ -1,24 +0,0 @@ -/* Copyright 2017 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- this is just embarassing :| -ALTER TABLE users_in_pubic_room RENAME TO users_in_public_rooms; - --- this is only 300K rows on matrix.org and takes ~3s to generate the index, --- so is hopefully not going to block anyone else for that long... -CREATE INDEX users_in_public_rooms_room_idx ON users_in_public_rooms(room_id); -CREATE UNIQUE INDEX users_in_public_rooms_user_idx ON users_in_public_rooms(user_id); -DROP INDEX users_in_pubic_room_room_idx; -DROP INDEX users_in_pubic_room_user_idx; diff --git a/synapse/storage/schema/delta/47/last_access_media.sql b/synapse/storage/schema/delta/47/last_access_media.sql deleted file mode 100644 index f505fb22b5..0000000000 --- a/synapse/storage/schema/delta/47/last_access_media.sql +++ /dev/null @@ -1,16 +0,0 @@ -/* Copyright 2018 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -ALTER TABLE local_media_repository ADD COLUMN last_access_ts BIGINT; diff --git a/synapse/storage/schema/delta/47/postgres_fts_gin.sql b/synapse/storage/schema/delta/47/postgres_fts_gin.sql deleted file mode 100644 index 31d7a817eb..0000000000 --- a/synapse/storage/schema/delta/47/postgres_fts_gin.sql +++ /dev/null @@ -1,17 +0,0 @@ -/* Copyright 2018 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -INSERT into background_updates (update_name, progress_json) - VALUES ('event_search_postgres_gin', '{}'); diff --git a/synapse/storage/schema/delta/47/push_actions_staging.sql b/synapse/storage/schema/delta/47/push_actions_staging.sql deleted file mode 100644 index edccf4a96f..0000000000 --- a/synapse/storage/schema/delta/47/push_actions_staging.sql +++ /dev/null @@ -1,28 +0,0 @@ -/* Copyright 2018 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- Temporary staging area for push actions that have been calculated for an --- event, but the event hasn't yet been persisted. --- When the event is persisted the rows are moved over to the --- event_push_actions table. -CREATE TABLE event_push_actions_staging ( - event_id TEXT NOT NULL, - user_id TEXT NOT NULL, - actions TEXT NOT NULL, - notif SMALLINT NOT NULL, - highlight SMALLINT NOT NULL -); - -CREATE INDEX event_push_actions_staging_id ON event_push_actions_staging(event_id); diff --git a/synapse/storage/schema/delta/47/state_group_seq.py b/synapse/storage/schema/delta/47/state_group_seq.py deleted file mode 100644 index 9fd1ccf6f7..0000000000 --- a/synapse/storage/schema/delta/47/state_group_seq.py +++ /dev/null @@ -1,34 +0,0 @@ -# Copyright 2018 New Vector Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -from synapse.storage.engines import PostgresEngine - - -def run_create(cur, database_engine, *args, **kwargs): - if isinstance(database_engine, PostgresEngine): - # if we already have some state groups, we want to start making new - # ones with a higher id. - cur.execute("SELECT max(id) FROM state_groups") - row = cur.fetchone() - - if row[0] is None: - start_val = 1 - else: - start_val = row[0] + 1 - - cur.execute("CREATE SEQUENCE state_group_id_seq START WITH %s", (start_val,)) - - -def run_upgrade(*args, **kwargs): - pass diff --git a/synapse/storage/schema/delta/48/add_user_consent.sql b/synapse/storage/schema/delta/48/add_user_consent.sql deleted file mode 100644 index 5237491506..0000000000 --- a/synapse/storage/schema/delta/48/add_user_consent.sql +++ /dev/null @@ -1,18 +0,0 @@ -/* Copyright 2018 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -/* record the version of the privacy policy the user has consented to - */ -ALTER TABLE users ADD COLUMN consent_version TEXT; diff --git a/synapse/storage/schema/delta/48/add_user_ips_last_seen_index.sql b/synapse/storage/schema/delta/48/add_user_ips_last_seen_index.sql deleted file mode 100644 index 9248b0b24a..0000000000 --- a/synapse/storage/schema/delta/48/add_user_ips_last_seen_index.sql +++ /dev/null @@ -1,17 +0,0 @@ -/* Copyright 2018 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -INSERT into background_updates (update_name, progress_json) - VALUES ('user_ips_last_seen_index', '{}'); diff --git a/synapse/storage/schema/delta/48/deactivated_users.sql b/synapse/storage/schema/delta/48/deactivated_users.sql deleted file mode 100644 index e9013a6969..0000000000 --- a/synapse/storage/schema/delta/48/deactivated_users.sql +++ /dev/null @@ -1,25 +0,0 @@ -/* Copyright 2018 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -/* - * Store any accounts that have been requested to be deactivated. - * We part the account from all the rooms its in when its - * deactivated. This can take some time and synapse may be restarted - * before it completes, so store the user IDs here until the process - * is complete. - */ -CREATE TABLE users_pending_deactivation ( - user_id TEXT NOT NULL -); diff --git a/synapse/storage/schema/delta/48/group_unique_indexes.py b/synapse/storage/schema/delta/48/group_unique_indexes.py deleted file mode 100644 index 49f5f2c003..0000000000 --- a/synapse/storage/schema/delta/48/group_unique_indexes.py +++ /dev/null @@ -1,63 +0,0 @@ -# Copyright 2018 New Vector Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -from synapse.storage.engines import PostgresEngine -from synapse.storage.prepare_database import get_statements - -FIX_INDEXES = """ --- rebuild indexes as uniques -DROP INDEX groups_invites_g_idx; -CREATE UNIQUE INDEX group_invites_g_idx ON group_invites(group_id, user_id); -DROP INDEX groups_users_g_idx; -CREATE UNIQUE INDEX group_users_g_idx ON group_users(group_id, user_id); - --- rename other indexes to actually match their table names.. -DROP INDEX groups_users_u_idx; -CREATE INDEX group_users_u_idx ON group_users(user_id); -DROP INDEX groups_invites_u_idx; -CREATE INDEX group_invites_u_idx ON group_invites(user_id); -DROP INDEX groups_rooms_g_idx; -CREATE UNIQUE INDEX group_rooms_g_idx ON group_rooms(group_id, room_id); -DROP INDEX groups_rooms_r_idx; -CREATE INDEX group_rooms_r_idx ON group_rooms(room_id); -""" - - -def run_create(cur, database_engine, *args, **kwargs): - rowid = "ctid" if isinstance(database_engine, PostgresEngine) else "rowid" - - # remove duplicates from group_users & group_invites tables - cur.execute( - """ - DELETE FROM group_users WHERE %s NOT IN ( - SELECT min(%s) FROM group_users GROUP BY group_id, user_id - ); - """ - % (rowid, rowid) - ) - cur.execute( - """ - DELETE FROM group_invites WHERE %s NOT IN ( - SELECT min(%s) FROM group_invites GROUP BY group_id, user_id - ); - """ - % (rowid, rowid) - ) - - for statement in get_statements(FIX_INDEXES.splitlines()): - cur.execute(statement) - - -def run_upgrade(*args, **kwargs): - pass diff --git a/synapse/storage/schema/delta/48/groups_joinable.sql b/synapse/storage/schema/delta/48/groups_joinable.sql deleted file mode 100644 index ce26eaf0c9..0000000000 --- a/synapse/storage/schema/delta/48/groups_joinable.sql +++ /dev/null @@ -1,22 +0,0 @@ -/* Copyright 2018 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -/* - * This isn't a real ENUM because sqlite doesn't support it - * and we use a default of NULL for inserted rows and interpret - * NULL at the python store level as necessary so that existing - * rows are given the correct default policy. - */ -ALTER TABLE groups ADD COLUMN join_policy TEXT NOT NULL DEFAULT 'invite'; diff --git a/synapse/storage/schema/delta/49/add_user_consent_server_notice_sent.sql b/synapse/storage/schema/delta/49/add_user_consent_server_notice_sent.sql deleted file mode 100644 index 14dcf18d73..0000000000 --- a/synapse/storage/schema/delta/49/add_user_consent_server_notice_sent.sql +++ /dev/null @@ -1,20 +0,0 @@ -/* Copyright 2018 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -/* record whether we have sent a server notice about consenting to the - * privacy policy. Specifically records the version of the policy we sent - * a message about. - */ -ALTER TABLE users ADD COLUMN consent_server_notice_sent TEXT; diff --git a/synapse/storage/schema/delta/49/add_user_daily_visits.sql b/synapse/storage/schema/delta/49/add_user_daily_visits.sql deleted file mode 100644 index 3dd478196f..0000000000 --- a/synapse/storage/schema/delta/49/add_user_daily_visits.sql +++ /dev/null @@ -1,21 +0,0 @@ -/* Copyright 2018 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - -CREATE TABLE user_daily_visits ( user_id TEXT NOT NULL, - device_id TEXT, - timestamp BIGINT NOT NULL ); -CREATE INDEX user_daily_visits_uts_idx ON user_daily_visits(user_id, timestamp); -CREATE INDEX user_daily_visits_ts_idx ON user_daily_visits(timestamp); diff --git a/synapse/storage/schema/delta/49/add_user_ips_last_seen_only_index.sql b/synapse/storage/schema/delta/49/add_user_ips_last_seen_only_index.sql deleted file mode 100644 index 3a4ed59b5b..0000000000 --- a/synapse/storage/schema/delta/49/add_user_ips_last_seen_only_index.sql +++ /dev/null @@ -1,17 +0,0 @@ -/* Copyright 2018 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -INSERT into background_updates (update_name, progress_json) - VALUES ('user_ips_last_seen_only_index', '{}'); diff --git a/synapse/storage/schema/delta/50/add_creation_ts_users_index.sql b/synapse/storage/schema/delta/50/add_creation_ts_users_index.sql deleted file mode 100644 index c93ae47532..0000000000 --- a/synapse/storage/schema/delta/50/add_creation_ts_users_index.sql +++ /dev/null @@ -1,19 +0,0 @@ -/* Copyright 2018 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - - -INSERT into background_updates (update_name, progress_json) - VALUES ('users_creation_ts', '{}'); diff --git a/synapse/storage/schema/delta/50/erasure_store.sql b/synapse/storage/schema/delta/50/erasure_store.sql deleted file mode 100644 index 5d8641a9ab..0000000000 --- a/synapse/storage/schema/delta/50/erasure_store.sql +++ /dev/null @@ -1,21 +0,0 @@ -/* Copyright 2018 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- a table of users who have requested that their details be erased -CREATE TABLE erased_users ( - user_id TEXT NOT NULL -); - -CREATE UNIQUE INDEX erased_users_user ON erased_users(user_id); diff --git a/synapse/storage/schema/delta/50/make_event_content_nullable.py b/synapse/storage/schema/delta/50/make_event_content_nullable.py deleted file mode 100644 index b1684a8441..0000000000 --- a/synapse/storage/schema/delta/50/make_event_content_nullable.py +++ /dev/null @@ -1,96 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2018 New Vector Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -""" -We want to stop populating 'event.content', so we need to make it nullable. - -If this has to be rolled back, then the following should populate the missing data: - -Postgres: - - UPDATE events SET content=(ej.json::json)->'content' FROM event_json ej - WHERE ej.event_id = events.event_id AND - stream_ordering < ( - SELECT stream_ordering FROM events WHERE content IS NOT NULL - ORDER BY stream_ordering LIMIT 1 - ); - - UPDATE events SET content=(ej.json::json)->'content' FROM event_json ej - WHERE ej.event_id = events.event_id AND - stream_ordering > ( - SELECT stream_ordering FROM events WHERE content IS NOT NULL - ORDER BY stream_ordering DESC LIMIT 1 - ); - -SQLite: - - UPDATE events SET content=( - SELECT json_extract(json,'$.content') FROM event_json ej - WHERE ej.event_id = events.event_id - ) - WHERE - stream_ordering < ( - SELECT stream_ordering FROM events WHERE content IS NOT NULL - ORDER BY stream_ordering LIMIT 1 - ) - OR stream_ordering > ( - SELECT stream_ordering FROM events WHERE content IS NOT NULL - ORDER BY stream_ordering DESC LIMIT 1 - ); - -""" - -import logging - -from synapse.storage.engines import PostgresEngine - -logger = logging.getLogger(__name__) - - -def run_create(cur, database_engine, *args, **kwargs): - pass - - -def run_upgrade(cur, database_engine, *args, **kwargs): - if isinstance(database_engine, PostgresEngine): - cur.execute( - """ - ALTER TABLE events ALTER COLUMN content DROP NOT NULL; - """ - ) - return - - # sqlite is an arse about this. ref: https://www.sqlite.org/lang_altertable.html - - cur.execute( - "SELECT sql FROM sqlite_master WHERE tbl_name='events' AND type='table'" - ) - (oldsql,) = cur.fetchone() - - sql = oldsql.replace("content TEXT NOT NULL", "content TEXT") - if sql == oldsql: - raise Exception("Couldn't find null constraint to drop in %s" % oldsql) - - logger.info("Replacing definition of 'events' with: %s", sql) - - cur.execute("PRAGMA schema_version") - (oldver,) = cur.fetchone() - cur.execute("PRAGMA writable_schema=ON") - cur.execute( - "UPDATE sqlite_master SET sql=? WHERE tbl_name='events' AND type='table'", - (sql,), - ) - cur.execute("PRAGMA schema_version=%i" % (oldver + 1,)) - cur.execute("PRAGMA writable_schema=OFF") diff --git a/synapse/storage/schema/delta/51/e2e_room_keys.sql b/synapse/storage/schema/delta/51/e2e_room_keys.sql deleted file mode 100644 index c0e66a697d..0000000000 --- a/synapse/storage/schema/delta/51/e2e_room_keys.sql +++ /dev/null @@ -1,39 +0,0 @@ -/* Copyright 2017 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- users' optionally backed up encrypted e2e sessions -CREATE TABLE e2e_room_keys ( - user_id TEXT NOT NULL, - room_id TEXT NOT NULL, - session_id TEXT NOT NULL, - version TEXT NOT NULL, - first_message_index INT, - forwarded_count INT, - is_verified BOOLEAN, - session_data TEXT NOT NULL -); - -CREATE UNIQUE INDEX e2e_room_keys_idx ON e2e_room_keys(user_id, room_id, session_id); - --- the metadata for each generation of encrypted e2e session backups -CREATE TABLE e2e_room_keys_versions ( - user_id TEXT NOT NULL, - version TEXT NOT NULL, - algorithm TEXT NOT NULL, - auth_data TEXT NOT NULL, - deleted SMALLINT DEFAULT 0 NOT NULL -); - -CREATE UNIQUE INDEX e2e_room_keys_versions_idx ON e2e_room_keys_versions(user_id, version); diff --git a/synapse/storage/schema/delta/51/monthly_active_users.sql b/synapse/storage/schema/delta/51/monthly_active_users.sql deleted file mode 100644 index c9d537d5a3..0000000000 --- a/synapse/storage/schema/delta/51/monthly_active_users.sql +++ /dev/null @@ -1,27 +0,0 @@ -/* Copyright 2018 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- a table of monthly active users, for use where blocking based on mau limits -CREATE TABLE monthly_active_users ( - user_id TEXT NOT NULL, - -- Last time we saw the user. Not guaranteed to be accurate due to rate limiting - -- on updates, Granularity of updates governed by - -- synapse.storage.monthly_active_users.LAST_SEEN_GRANULARITY - -- Measured in ms since epoch. - timestamp BIGINT NOT NULL -); - -CREATE UNIQUE INDEX monthly_active_users_users ON monthly_active_users(user_id); -CREATE INDEX monthly_active_users_time_stamp ON monthly_active_users(timestamp); diff --git a/synapse/storage/schema/delta/52/add_event_to_state_group_index.sql b/synapse/storage/schema/delta/52/add_event_to_state_group_index.sql deleted file mode 100644 index 91e03d13e1..0000000000 --- a/synapse/storage/schema/delta/52/add_event_to_state_group_index.sql +++ /dev/null @@ -1,19 +0,0 @@ -/* Copyright 2018 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- This is needed to efficiently check for unreferenced state groups during --- purge. Added events_to_state_group(state_group) index -INSERT into background_updates (update_name, progress_json) - VALUES ('event_to_state_groups_sg_index', '{}'); diff --git a/synapse/storage/schema/delta/52/device_list_streams_unique_idx.sql b/synapse/storage/schema/delta/52/device_list_streams_unique_idx.sql deleted file mode 100644 index bfa49e6f92..0000000000 --- a/synapse/storage/schema/delta/52/device_list_streams_unique_idx.sql +++ /dev/null @@ -1,36 +0,0 @@ -/* Copyright 2018 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- register a background update which will create a unique index on --- device_lists_remote_cache -INSERT into background_updates (update_name, progress_json) - VALUES ('device_lists_remote_cache_unique_idx', '{}'); - --- and one on device_lists_remote_extremeties -INSERT into background_updates (update_name, progress_json, depends_on) - VALUES ( - 'device_lists_remote_extremeties_unique_idx', '{}', - - -- doesn't really depend on this, but we need to make sure both happen - -- before we drop the old indexes. - 'device_lists_remote_cache_unique_idx' - ); - --- once they complete, we can drop the old indexes. -INSERT into background_updates (update_name, progress_json, depends_on) - VALUES ( - 'drop_device_list_streams_non_unique_indexes', '{}', - 'device_lists_remote_extremeties_unique_idx' - ); diff --git a/synapse/storage/schema/delta/52/e2e_room_keys.sql b/synapse/storage/schema/delta/52/e2e_room_keys.sql deleted file mode 100644 index db687cccae..0000000000 --- a/synapse/storage/schema/delta/52/e2e_room_keys.sql +++ /dev/null @@ -1,53 +0,0 @@ -/* Copyright 2018 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -/* Change version column to an integer so we can do MAX() sensibly - */ -CREATE TABLE e2e_room_keys_versions_new ( - user_id TEXT NOT NULL, - version BIGINT NOT NULL, - algorithm TEXT NOT NULL, - auth_data TEXT NOT NULL, - deleted SMALLINT DEFAULT 0 NOT NULL -); - -INSERT INTO e2e_room_keys_versions_new - SELECT user_id, CAST(version as BIGINT), algorithm, auth_data, deleted FROM e2e_room_keys_versions; - -DROP TABLE e2e_room_keys_versions; -ALTER TABLE e2e_room_keys_versions_new RENAME TO e2e_room_keys_versions; - -CREATE UNIQUE INDEX e2e_room_keys_versions_idx ON e2e_room_keys_versions(user_id, version); - -/* Change e2e_rooms_keys to match - */ -CREATE TABLE e2e_room_keys_new ( - user_id TEXT NOT NULL, - room_id TEXT NOT NULL, - session_id TEXT NOT NULL, - version BIGINT NOT NULL, - first_message_index INT, - forwarded_count INT, - is_verified BOOLEAN, - session_data TEXT NOT NULL -); - -INSERT INTO e2e_room_keys_new - SELECT user_id, room_id, session_id, CAST(version as BIGINT), first_message_index, forwarded_count, is_verified, session_data FROM e2e_room_keys; - -DROP TABLE e2e_room_keys; -ALTER TABLE e2e_room_keys_new RENAME TO e2e_room_keys; - -CREATE UNIQUE INDEX e2e_room_keys_idx ON e2e_room_keys(user_id, room_id, session_id); diff --git a/synapse/storage/schema/delta/53/add_user_type_to_users.sql b/synapse/storage/schema/delta/53/add_user_type_to_users.sql deleted file mode 100644 index 88ec2f83e5..0000000000 --- a/synapse/storage/schema/delta/53/add_user_type_to_users.sql +++ /dev/null @@ -1,19 +0,0 @@ -/* Copyright 2018 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -/* The type of the user: NULL for a regular user, or one of the constants in - * synapse.api.constants.UserTypes - */ -ALTER TABLE users ADD COLUMN user_type TEXT DEFAULT NULL; diff --git a/synapse/storage/schema/delta/53/drop_sent_transactions.sql b/synapse/storage/schema/delta/53/drop_sent_transactions.sql deleted file mode 100644 index e372f5a44a..0000000000 --- a/synapse/storage/schema/delta/53/drop_sent_transactions.sql +++ /dev/null @@ -1,16 +0,0 @@ -/* Copyright 2018 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -DROP TABLE IF EXISTS sent_transactions; diff --git a/synapse/storage/schema/delta/53/event_format_version.sql b/synapse/storage/schema/delta/53/event_format_version.sql deleted file mode 100644 index 1d977c2834..0000000000 --- a/synapse/storage/schema/delta/53/event_format_version.sql +++ /dev/null @@ -1,16 +0,0 @@ -/* Copyright 2019 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -ALTER TABLE event_json ADD COLUMN format_version INTEGER; diff --git a/synapse/storage/schema/delta/53/user_dir_populate.sql b/synapse/storage/schema/delta/53/user_dir_populate.sql deleted file mode 100644 index ffcc896b58..0000000000 --- a/synapse/storage/schema/delta/53/user_dir_populate.sql +++ /dev/null @@ -1,30 +0,0 @@ -/* Copyright 2019 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- Set up staging tables -INSERT INTO background_updates (update_name, progress_json) VALUES - ('populate_user_directory_createtables', '{}'); - --- Run through each room and update the user directory according to who is in it -INSERT INTO background_updates (update_name, progress_json, depends_on) VALUES - ('populate_user_directory_process_rooms', '{}', 'populate_user_directory_createtables'); - --- Insert all users, if search_all_users is on -INSERT INTO background_updates (update_name, progress_json, depends_on) VALUES - ('populate_user_directory_process_users', '{}', 'populate_user_directory_process_rooms'); - --- Clean up staging tables -INSERT INTO background_updates (update_name, progress_json, depends_on) VALUES - ('populate_user_directory_cleanup', '{}', 'populate_user_directory_process_users'); diff --git a/synapse/storage/schema/delta/53/user_ips_index.sql b/synapse/storage/schema/delta/53/user_ips_index.sql deleted file mode 100644 index b812c5794f..0000000000 --- a/synapse/storage/schema/delta/53/user_ips_index.sql +++ /dev/null @@ -1,30 +0,0 @@ -/* Copyright 2018 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - -- analyze user_ips, to help ensure the correct indices are used -INSERT INTO background_updates (update_name, progress_json) VALUES - ('user_ips_analyze', '{}'); - --- delete duplicates -INSERT INTO background_updates (update_name, progress_json, depends_on) VALUES - ('user_ips_remove_dupes', '{}', 'user_ips_analyze'); - --- add a new unique index to user_ips table -INSERT INTO background_updates (update_name, progress_json, depends_on) VALUES - ('user_ips_device_unique_index', '{}', 'user_ips_remove_dupes'); - --- drop the old original index -INSERT INTO background_updates (update_name, progress_json, depends_on) VALUES - ('user_ips_drop_nonunique_index', '{}', 'user_ips_device_unique_index'); diff --git a/synapse/storage/schema/delta/53/user_share.sql b/synapse/storage/schema/delta/53/user_share.sql deleted file mode 100644 index 5831b1a6f8..0000000000 --- a/synapse/storage/schema/delta/53/user_share.sql +++ /dev/null @@ -1,44 +0,0 @@ -/* Copyright 2017 Vector Creations Ltd, 2019 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- Old disused version of the tables below. -DROP TABLE IF EXISTS users_who_share_rooms; - --- Tables keeping track of what users share rooms. This is a map of local users --- to local or remote users, per room. Remote users cannot be in the user_id --- column, only the other_user_id column. There are two tables, one for public --- rooms and those for private rooms. -CREATE TABLE IF NOT EXISTS users_who_share_public_rooms ( - user_id TEXT NOT NULL, - other_user_id TEXT NOT NULL, - room_id TEXT NOT NULL -); - -CREATE TABLE IF NOT EXISTS users_who_share_private_rooms ( - user_id TEXT NOT NULL, - other_user_id TEXT NOT NULL, - room_id TEXT NOT NULL -); - -CREATE UNIQUE INDEX users_who_share_public_rooms_u_idx ON users_who_share_public_rooms(user_id, other_user_id, room_id); -CREATE INDEX users_who_share_public_rooms_r_idx ON users_who_share_public_rooms(room_id); -CREATE INDEX users_who_share_public_rooms_o_idx ON users_who_share_public_rooms(other_user_id); - -CREATE UNIQUE INDEX users_who_share_private_rooms_u_idx ON users_who_share_private_rooms(user_id, other_user_id, room_id); -CREATE INDEX users_who_share_private_rooms_r_idx ON users_who_share_private_rooms(room_id); -CREATE INDEX users_who_share_private_rooms_o_idx ON users_who_share_private_rooms(other_user_id); - --- Make sure that we populate the tables initially by resetting the stream ID -UPDATE user_directory_stream_pos SET stream_id = NULL; diff --git a/synapse/storage/schema/delta/53/user_threepid_id.sql b/synapse/storage/schema/delta/53/user_threepid_id.sql deleted file mode 100644 index 80c2c573b6..0000000000 --- a/synapse/storage/schema/delta/53/user_threepid_id.sql +++ /dev/null @@ -1,29 +0,0 @@ -/* Copyright 2019 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- Tracks which identity server a user bound their threepid via. -CREATE TABLE user_threepid_id_server ( - user_id TEXT NOT NULL, - medium TEXT NOT NULL, - address TEXT NOT NULL, - id_server TEXT NOT NULL -); - -CREATE UNIQUE INDEX user_threepid_id_server_idx ON user_threepid_id_server( - user_id, medium, address, id_server -); - -INSERT INTO background_updates (update_name, progress_json) VALUES - ('user_threepids_grandfather', '{}'); diff --git a/synapse/storage/schema/delta/53/users_in_public_rooms.sql b/synapse/storage/schema/delta/53/users_in_public_rooms.sql deleted file mode 100644 index f7827ca6d2..0000000000 --- a/synapse/storage/schema/delta/53/users_in_public_rooms.sql +++ /dev/null @@ -1,28 +0,0 @@ -/* Copyright 2019 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- We don't need the old version of this table. -DROP TABLE IF EXISTS users_in_public_rooms; - --- Old version of users_in_public_rooms -DROP TABLE IF EXISTS users_who_share_public_rooms; - --- Track what users are in public rooms. -CREATE TABLE IF NOT EXISTS users_in_public_rooms ( - user_id TEXT NOT NULL, - room_id TEXT NOT NULL -); - -CREATE UNIQUE INDEX users_in_public_rooms_u_idx ON users_in_public_rooms(user_id, room_id); diff --git a/synapse/storage/schema/delta/54/account_validity_with_renewal.sql b/synapse/storage/schema/delta/54/account_validity_with_renewal.sql deleted file mode 100644 index 0adb2ad55e..0000000000 --- a/synapse/storage/schema/delta/54/account_validity_with_renewal.sql +++ /dev/null @@ -1,30 +0,0 @@ -/* Copyright 2019 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- We previously changed the schema for this table without renaming the file, which means --- that some databases might still be using the old schema. This ensures Synapse uses the --- right schema for the table. -DROP TABLE IF EXISTS account_validity; - --- Track what users are in public rooms. -CREATE TABLE IF NOT EXISTS account_validity ( - user_id TEXT PRIMARY KEY, - expiration_ts_ms BIGINT NOT NULL, - email_sent BOOLEAN NOT NULL, - renewal_token TEXT -); - -CREATE INDEX account_validity_email_sent_idx ON account_validity(email_sent, expiration_ts_ms) -CREATE UNIQUE INDEX account_validity_renewal_string_idx ON account_validity(renewal_token) diff --git a/synapse/storage/schema/delta/54/add_validity_to_server_keys.sql b/synapse/storage/schema/delta/54/add_validity_to_server_keys.sql deleted file mode 100644 index c01aa9d2d9..0000000000 --- a/synapse/storage/schema/delta/54/add_validity_to_server_keys.sql +++ /dev/null @@ -1,23 +0,0 @@ -/* Copyright 2019 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -/* When we can use this key until, before we have to refresh it. */ -ALTER TABLE server_signature_keys ADD COLUMN ts_valid_until_ms BIGINT; - -UPDATE server_signature_keys SET ts_valid_until_ms = ( - SELECT MAX(ts_valid_until_ms) FROM server_keys_json skj WHERE - skj.server_name = server_signature_keys.server_name AND - skj.key_id = server_signature_keys.key_id -); diff --git a/synapse/storage/schema/delta/54/delete_forward_extremities.sql b/synapse/storage/schema/delta/54/delete_forward_extremities.sql deleted file mode 100644 index b062ec840c..0000000000 --- a/synapse/storage/schema/delta/54/delete_forward_extremities.sql +++ /dev/null @@ -1,23 +0,0 @@ -/* Copyright 2019 The Matrix.org Foundation C.I.C. - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- Start a background job to cleanup extremities that were incorrectly added --- by bug #5269. -INSERT INTO background_updates (update_name, progress_json) VALUES - ('delete_soft_failed_extremities', '{}'); - -DROP TABLE IF EXISTS _extremities_to_check; -- To make this delta schema file idempotent. -CREATE TABLE _extremities_to_check AS SELECT event_id FROM event_forward_extremities; -CREATE INDEX _extremities_to_check_id ON _extremities_to_check(event_id); diff --git a/synapse/storage/schema/delta/54/drop_legacy_tables.sql b/synapse/storage/schema/delta/54/drop_legacy_tables.sql deleted file mode 100644 index dbbe682697..0000000000 --- a/synapse/storage/schema/delta/54/drop_legacy_tables.sql +++ /dev/null @@ -1,30 +0,0 @@ -/* Copyright 2019 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- we need to do this first due to foreign constraints -DROP TABLE IF EXISTS application_services_regex; - -DROP TABLE IF EXISTS application_services; -DROP TABLE IF EXISTS transaction_id_to_pdu; -DROP TABLE IF EXISTS stats_reporting; -DROP TABLE IF EXISTS current_state_resets; -DROP TABLE IF EXISTS event_content_hashes; -DROP TABLE IF EXISTS event_destinations; -DROP TABLE IF EXISTS event_edge_hashes; -DROP TABLE IF EXISTS event_signatures; -DROP TABLE IF EXISTS feedback; -DROP TABLE IF EXISTS room_hosts; -DROP TABLE IF EXISTS server_tls_certificates; -DROP TABLE IF EXISTS state_forward_extremities; diff --git a/synapse/storage/schema/delta/54/drop_presence_list.sql b/synapse/storage/schema/delta/54/drop_presence_list.sql deleted file mode 100644 index e6ee70c623..0000000000 --- a/synapse/storage/schema/delta/54/drop_presence_list.sql +++ /dev/null @@ -1,16 +0,0 @@ -/* Copyright 2019 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -DROP TABLE IF EXISTS presence_list; diff --git a/synapse/storage/schema/delta/54/relations.sql b/synapse/storage/schema/delta/54/relations.sql deleted file mode 100644 index 134862b870..0000000000 --- a/synapse/storage/schema/delta/54/relations.sql +++ /dev/null @@ -1,27 +0,0 @@ -/* Copyright 2019 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- Tracks related events, like reactions, replies, edits, etc. Note that things --- in this table are not necessarily "valid", e.g. it may contain edits from --- people who don't have power to edit other peoples events. -CREATE TABLE IF NOT EXISTS event_relations ( - event_id TEXT NOT NULL, - relates_to_id TEXT NOT NULL, - relation_type TEXT NOT NULL, - aggregation_key TEXT -); - -CREATE UNIQUE INDEX event_relations_id ON event_relations(event_id); -CREATE INDEX event_relations_relates ON event_relations(relates_to_id, relation_type, aggregation_key); diff --git a/synapse/storage/schema/delta/54/stats.sql b/synapse/storage/schema/delta/54/stats.sql deleted file mode 100644 index 652e58308e..0000000000 --- a/synapse/storage/schema/delta/54/stats.sql +++ /dev/null @@ -1,80 +0,0 @@ -/* Copyright 2018 New Vector Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -CREATE TABLE stats_stream_pos ( - Lock CHAR(1) NOT NULL DEFAULT 'X' UNIQUE, -- Makes sure this table only has one row. - stream_id BIGINT, - CHECK (Lock='X') -); - -INSERT INTO stats_stream_pos (stream_id) VALUES (null); - -CREATE TABLE user_stats ( - user_id TEXT NOT NULL, - ts BIGINT NOT NULL, - bucket_size INT NOT NULL, - public_rooms INT NOT NULL, - private_rooms INT NOT NULL -); - -CREATE UNIQUE INDEX user_stats_user_ts ON user_stats(user_id, ts); - -CREATE TABLE room_stats ( - room_id TEXT NOT NULL, - ts BIGINT NOT NULL, - bucket_size INT NOT NULL, - current_state_events INT NOT NULL, - joined_members INT NOT NULL, - invited_members INT NOT NULL, - left_members INT NOT NULL, - banned_members INT NOT NULL, - state_events INT NOT NULL -); - -CREATE UNIQUE INDEX room_stats_room_ts ON room_stats(room_id, ts); - --- cache of current room state; useful for the publicRooms list -CREATE TABLE room_state ( - room_id TEXT NOT NULL, - join_rules TEXT, - history_visibility TEXT, - encryption TEXT, - name TEXT, - topic TEXT, - avatar TEXT, - canonical_alias TEXT - -- get aliases straight from the right table -); - -CREATE UNIQUE INDEX room_state_room ON room_state(room_id); - -CREATE TABLE room_stats_earliest_token ( - room_id TEXT NOT NULL, - token BIGINT NOT NULL -); - -CREATE UNIQUE INDEX room_stats_earliest_token_idx ON room_stats_earliest_token(room_id); - --- Set up staging tables -INSERT INTO background_updates (update_name, progress_json) VALUES - ('populate_stats_createtables', '{}'); - --- Run through each room and update stats -INSERT INTO background_updates (update_name, progress_json, depends_on) VALUES - ('populate_stats_process_rooms', '{}', 'populate_stats_createtables'); - --- Clean up staging tables -INSERT INTO background_updates (update_name, progress_json, depends_on) VALUES - ('populate_stats_cleanup', '{}', 'populate_stats_process_rooms'); diff --git a/synapse/storage/schema/delta/54/stats2.sql b/synapse/storage/schema/delta/54/stats2.sql deleted file mode 100644 index 3b2d48447f..0000000000 --- a/synapse/storage/schema/delta/54/stats2.sql +++ /dev/null @@ -1,28 +0,0 @@ -/* Copyright 2019 The Matrix.org Foundation C.I.C. - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- This delta file gets run after `54/stats.sql` delta. - --- We want to add some indices to the temporary stats table, so we re-insert --- 'populate_stats_createtables' if we are still processing the rooms update. -INSERT INTO background_updates (update_name, progress_json) - SELECT 'populate_stats_createtables', '{}' - WHERE - 'populate_stats_process_rooms' IN ( - SELECT update_name FROM background_updates - ) - AND 'populate_stats_createtables' NOT IN ( -- don't insert if already exists - SELECT update_name FROM background_updates - ); diff --git a/synapse/storage/schema/delta/55/access_token_expiry.sql b/synapse/storage/schema/delta/55/access_token_expiry.sql deleted file mode 100644 index 4590604bfd..0000000000 --- a/synapse/storage/schema/delta/55/access_token_expiry.sql +++ /dev/null @@ -1,18 +0,0 @@ -/* Copyright 2019 The Matrix.org Foundation C.I.C. - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- when this access token can be used until, in ms since the epoch. NULL means the token --- never expires. -ALTER TABLE access_tokens ADD COLUMN valid_until_ms BIGINT; diff --git a/synapse/storage/schema/delta/55/track_threepid_validations.sql b/synapse/storage/schema/delta/55/track_threepid_validations.sql deleted file mode 100644 index a8eced2e0a..0000000000 --- a/synapse/storage/schema/delta/55/track_threepid_validations.sql +++ /dev/null @@ -1,31 +0,0 @@ -/* Copyright 2019 The Matrix.org Foundation C.I.C. - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ -CREATE TABLE IF NOT EXISTS threepid_validation_session ( - session_id TEXT PRIMARY KEY, - medium TEXT NOT NULL, - address TEXT NOT NULL, - client_secret TEXT NOT NULL, - last_send_attempt BIGINT NOT NULL, - validated_at BIGINT -); - -CREATE TABLE IF NOT EXISTS threepid_validation_token ( - token TEXT PRIMARY KEY, - session_id TEXT NOT NULL, - next_link TEXT, - expires BIGINT NOT NULL -); - -CREATE INDEX threepid_validation_token_session_id ON threepid_validation_token(session_id); diff --git a/synapse/storage/schema/delta/55/users_alter_deactivated.sql b/synapse/storage/schema/delta/55/users_alter_deactivated.sql deleted file mode 100644 index dabdde489b..0000000000 --- a/synapse/storage/schema/delta/55/users_alter_deactivated.sql +++ /dev/null @@ -1,19 +0,0 @@ -/* Copyright 2019 The Matrix.org Foundation C.I.C. - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -ALTER TABLE users ADD deactivated SMALLINT DEFAULT 0 NOT NULL; - -INSERT INTO background_updates (update_name, progress_json) VALUES - ('users_set_deactivated_flag', '{}'); diff --git a/synapse/storage/schema/delta/56/add_spans_to_device_lists.sql b/synapse/storage/schema/delta/56/add_spans_to_device_lists.sql deleted file mode 100644 index 41807eb1e7..0000000000 --- a/synapse/storage/schema/delta/56/add_spans_to_device_lists.sql +++ /dev/null @@ -1,20 +0,0 @@ -/* Copyright 2019 The Matrix.org Foundation C.I.C - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -/* - * Opentracing context data for inclusion in the device_list_update EDUs, as a - * json-encoded dictionary. NULL if opentracing is disabled (or not enabled for this destination). - */ -ALTER TABLE device_lists_outbound_pokes ADD opentracing_context TEXT; diff --git a/synapse/storage/schema/delta/56/current_state_events_membership.sql b/synapse/storage/schema/delta/56/current_state_events_membership.sql deleted file mode 100644 index 473018676f..0000000000 --- a/synapse/storage/schema/delta/56/current_state_events_membership.sql +++ /dev/null @@ -1,22 +0,0 @@ -/* Copyright 2019 The Matrix.org Foundation C.I.C. - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- We add membership to current state so that we don't need to join against --- room_memberships, which can be surprisingly costly (we do such queries --- very frequently). --- This will be null for non-membership events and the content.membership key --- for membership events. (Will also be null for membership events until the --- background update job has finished). -ALTER TABLE current_state_events ADD membership TEXT; diff --git a/synapse/storage/schema/delta/56/current_state_events_membership_mk2.sql b/synapse/storage/schema/delta/56/current_state_events_membership_mk2.sql deleted file mode 100644 index 3133d42d4a..0000000000 --- a/synapse/storage/schema/delta/56/current_state_events_membership_mk2.sql +++ /dev/null @@ -1,24 +0,0 @@ -/* Copyright 2019 The Matrix.org Foundation C.I.C. - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- We add membership to current state so that we don't need to join against --- room_memberships, which can be surprisingly costly (we do such queries --- very frequently). --- This will be null for non-membership events and the content.membership key --- for membership events. (Will also be null for membership events until the --- background update job has finished). - -INSERT INTO background_updates (update_name, progress_json) VALUES - ('current_state_events_membership', '{}'); diff --git a/synapse/storage/schema/delta/56/destinations_failure_ts.sql b/synapse/storage/schema/delta/56/destinations_failure_ts.sql deleted file mode 100644 index f00889290b..0000000000 --- a/synapse/storage/schema/delta/56/destinations_failure_ts.sql +++ /dev/null @@ -1,25 +0,0 @@ -/* Copyright 2019 The Matrix.org Foundation C.I.C - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -/* - * Record the timestamp when a given server started failing - */ -ALTER TABLE destinations ADD failure_ts BIGINT; - -/* as a rough approximation, we assume that the server started failing at - * retry_interval before the last retry - */ -UPDATE destinations SET failure_ts = retry_last_ts - retry_interval - WHERE retry_last_ts > 0; diff --git a/synapse/storage/schema/delta/56/destinations_retry_interval_type.sql.postgres b/synapse/storage/schema/delta/56/destinations_retry_interval_type.sql.postgres deleted file mode 100644 index b9bbb18a91..0000000000 --- a/synapse/storage/schema/delta/56/destinations_retry_interval_type.sql.postgres +++ /dev/null @@ -1,18 +0,0 @@ -/* Copyright 2019 The Matrix.org Foundation C.I.C - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- We want to store large retry intervals so we upgrade the column from INT --- to BIGINT. We don't need to do this on SQLite. -ALTER TABLE destinations ALTER retry_interval SET DATA TYPE BIGINT; diff --git a/synapse/storage/schema/delta/56/devices_last_seen.sql b/synapse/storage/schema/delta/56/devices_last_seen.sql deleted file mode 100644 index dfa902d0ba..0000000000 --- a/synapse/storage/schema/delta/56/devices_last_seen.sql +++ /dev/null @@ -1,24 +0,0 @@ -/* Copyright 2019 Matrix.org Foundation CIC - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- Track last seen information for a device in the devices table, rather --- than relying on it being in the user_ips table (which we want to be able --- to purge old entries from) -ALTER TABLE devices ADD COLUMN last_seen BIGINT; -ALTER TABLE devices ADD COLUMN ip TEXT; -ALTER TABLE devices ADD COLUMN user_agent TEXT; - -INSERT INTO background_updates (update_name, progress_json) VALUES - ('devices_last_seen', '{}'); diff --git a/synapse/storage/schema/delta/56/drop_unused_event_tables.sql b/synapse/storage/schema/delta/56/drop_unused_event_tables.sql deleted file mode 100644 index 9f09922c67..0000000000 --- a/synapse/storage/schema/delta/56/drop_unused_event_tables.sql +++ /dev/null @@ -1,20 +0,0 @@ -/* Copyright 2019 The Matrix.org Foundation C.I.C. - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- these tables are never used. -DROP TABLE IF EXISTS room_names; -DROP TABLE IF EXISTS topics; -DROP TABLE IF EXISTS history_visibility; -DROP TABLE IF EXISTS guest_access; diff --git a/synapse/storage/schema/delta/56/fix_room_keys_index.sql b/synapse/storage/schema/delta/56/fix_room_keys_index.sql deleted file mode 100644 index 014cb3b538..0000000000 --- a/synapse/storage/schema/delta/56/fix_room_keys_index.sql +++ /dev/null @@ -1,18 +0,0 @@ -/* Copyright 2019 Matrix.org Foundation CIC - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- version is supposed to be part of the room keys index -CREATE UNIQUE INDEX e2e_room_keys_with_version_idx ON e2e_room_keys(user_id, version, room_id, session_id); -DROP INDEX IF EXISTS e2e_room_keys_idx; diff --git a/synapse/storage/schema/delta/56/public_room_list_idx.sql b/synapse/storage/schema/delta/56/public_room_list_idx.sql deleted file mode 100644 index 7be31ffebb..0000000000 --- a/synapse/storage/schema/delta/56/public_room_list_idx.sql +++ /dev/null @@ -1,16 +0,0 @@ -/* Copyright 2019 The Matrix.org Foundation C.I.C. - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -CREATE INDEX public_room_list_stream_network ON public_room_list_stream (appservice_id, network_id, room_id); diff --git a/synapse/storage/schema/delta/56/redaction_censor.sql b/synapse/storage/schema/delta/56/redaction_censor.sql deleted file mode 100644 index fe51b02309..0000000000 --- a/synapse/storage/schema/delta/56/redaction_censor.sql +++ /dev/null @@ -1,17 +0,0 @@ -/* Copyright 2019 The Matrix.org Foundation C.I.C. - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -ALTER TABLE redactions ADD COLUMN have_censored BOOL NOT NULL DEFAULT false; -CREATE INDEX redactions_have_censored ON redactions(event_id) WHERE not have_censored; diff --git a/synapse/storage/schema/delta/56/redaction_censor2.sql b/synapse/storage/schema/delta/56/redaction_censor2.sql deleted file mode 100644 index 77a5eca499..0000000000 --- a/synapse/storage/schema/delta/56/redaction_censor2.sql +++ /dev/null @@ -1,20 +0,0 @@ -/* Copyright 2019 The Matrix.org Foundation C.I.C. - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -ALTER TABLE redactions ADD COLUMN received_ts BIGINT; -CREATE INDEX redactions_have_censored_ts ON redactions(received_ts) WHERE not have_censored; - -INSERT INTO background_updates (update_name, progress_json) VALUES - ('redactions_received_ts', '{}'); diff --git a/synapse/storage/schema/delta/56/redaction_censor3_fix_update.sql.postgres b/synapse/storage/schema/delta/56/redaction_censor3_fix_update.sql.postgres deleted file mode 100644 index 67471f3ef5..0000000000 --- a/synapse/storage/schema/delta/56/redaction_censor3_fix_update.sql.postgres +++ /dev/null @@ -1,25 +0,0 @@ -/* Copyright 2019 The Matrix.org Foundation C.I.C. - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - --- There was a bug where we may have updated censored redactions as bytes, --- which can (somehow) cause json to be inserted hex encoded. These updates go --- and undoes any such hex encoded JSON. - -INSERT into background_updates (update_name, progress_json) - VALUES ('event_fix_redactions_bytes_create_index', '{}'); - -INSERT into background_updates (update_name, progress_json, depends_on) - VALUES ('event_fix_redactions_bytes', '{}', 'event_fix_redactions_bytes_create_index'); diff --git a/synapse/storage/schema/delta/56/room_membership_idx.sql b/synapse/storage/schema/delta/56/room_membership_idx.sql deleted file mode 100644 index 92ab1f5e65..0000000000 --- a/synapse/storage/schema/delta/56/room_membership_idx.sql +++ /dev/null @@ -1,18 +0,0 @@ -/* Copyright 2019 The Matrix.org Foundation C.I.C. - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- Adds an index on room_memberships for fetching all forgotten rooms for a user -INSERT INTO background_updates (update_name, progress_json) VALUES - ('room_membership_forgotten_idx', '{}'); diff --git a/synapse/storage/schema/delta/56/stats_separated.sql b/synapse/storage/schema/delta/56/stats_separated.sql deleted file mode 100644 index 163529c071..0000000000 --- a/synapse/storage/schema/delta/56/stats_separated.sql +++ /dev/null @@ -1,152 +0,0 @@ -/* Copyright 2018 New Vector Ltd - * Copyright 2019 The Matrix.org Foundation C.I.C. - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - ------ First clean up from previous versions of room stats. - --- First remove old stats stuff -DROP TABLE IF EXISTS room_stats; -DROP TABLE IF EXISTS room_state; -DROP TABLE IF EXISTS room_stats_state; -DROP TABLE IF EXISTS user_stats; -DROP TABLE IF EXISTS room_stats_earliest_tokens; -DROP TABLE IF EXISTS _temp_populate_stats_position; -DROP TABLE IF EXISTS _temp_populate_stats_rooms; -DROP TABLE IF EXISTS stats_stream_pos; - --- Unschedule old background updates if they're still scheduled -DELETE FROM background_updates WHERE update_name IN ( - 'populate_stats_createtables', - 'populate_stats_process_rooms', - 'populate_stats_process_users', - 'populate_stats_cleanup' -); - -INSERT INTO background_updates (update_name, progress_json, depends_on) VALUES - ('populate_stats_process_rooms', '{}', ''); - -INSERT INTO background_updates (update_name, progress_json, depends_on) VALUES - ('populate_stats_process_users', '{}', 'populate_stats_process_rooms'); - ------ Create tables for our version of room stats. - --- single-row table to track position of incremental updates -DROP TABLE IF EXISTS stats_incremental_position; -CREATE TABLE stats_incremental_position ( - Lock CHAR(1) NOT NULL DEFAULT 'X' UNIQUE, -- Makes sure this table only has one row. - stream_id BIGINT NOT NULL, - CHECK (Lock='X') -); - --- insert a null row and make sure it is the only one. -INSERT INTO stats_incremental_position ( - stream_id -) SELECT COALESCE(MAX(stream_ordering), 0) from events; - --- represents PRESENT room statistics for a room --- only holds absolute fields -DROP TABLE IF EXISTS room_stats_current; -CREATE TABLE room_stats_current ( - room_id TEXT NOT NULL PRIMARY KEY, - - -- These are absolute counts - current_state_events INT NOT NULL, - joined_members INT NOT NULL, - invited_members INT NOT NULL, - left_members INT NOT NULL, - banned_members INT NOT NULL, - - local_users_in_room INT NOT NULL, - - -- The maximum delta stream position that this row takes into account. - completed_delta_stream_id BIGINT NOT NULL -); - - --- represents HISTORICAL room statistics for a room -DROP TABLE IF EXISTS room_stats_historical; -CREATE TABLE room_stats_historical ( - room_id TEXT NOT NULL, - -- These stats cover the time from (end_ts - bucket_size)...end_ts (in ms). - -- Note that end_ts is quantised. - end_ts BIGINT NOT NULL, - bucket_size BIGINT NOT NULL, - - -- These stats are absolute counts - current_state_events BIGINT NOT NULL, - joined_members BIGINT NOT NULL, - invited_members BIGINT NOT NULL, - left_members BIGINT NOT NULL, - banned_members BIGINT NOT NULL, - local_users_in_room BIGINT NOT NULL, - - -- These stats are per time slice - total_events BIGINT NOT NULL, - total_event_bytes BIGINT NOT NULL, - - PRIMARY KEY (room_id, end_ts) -); - --- We use this index to speed up deletion of ancient room stats. -CREATE INDEX room_stats_historical_end_ts ON room_stats_historical (end_ts); - --- represents PRESENT statistics for a user --- only holds absolute fields -DROP TABLE IF EXISTS user_stats_current; -CREATE TABLE user_stats_current ( - user_id TEXT NOT NULL PRIMARY KEY, - - joined_rooms BIGINT NOT NULL, - - -- The maximum delta stream position that this row takes into account. - completed_delta_stream_id BIGINT NOT NULL -); - --- represents HISTORICAL statistics for a user -DROP TABLE IF EXISTS user_stats_historical; -CREATE TABLE user_stats_historical ( - user_id TEXT NOT NULL, - end_ts BIGINT NOT NULL, - bucket_size BIGINT NOT NULL, - - joined_rooms BIGINT NOT NULL, - - invites_sent BIGINT NOT NULL, - rooms_created BIGINT NOT NULL, - total_events BIGINT NOT NULL, - total_event_bytes BIGINT NOT NULL, - - PRIMARY KEY (user_id, end_ts) -); - --- We use this index to speed up deletion of ancient user stats. -CREATE INDEX user_stats_historical_end_ts ON user_stats_historical (end_ts); - - -CREATE TABLE room_stats_state ( - room_id TEXT NOT NULL, - name TEXT, - canonical_alias TEXT, - join_rules TEXT, - history_visibility TEXT, - encryption TEXT, - avatar TEXT, - guest_access TEXT, - is_federatable BOOLEAN, - topic TEXT -); - -CREATE UNIQUE INDEX room_stats_state_room ON room_stats_state(room_id); diff --git a/synapse/storage/schema/delta/56/unique_user_filter_index.py b/synapse/storage/schema/delta/56/unique_user_filter_index.py deleted file mode 100644 index 1de8b54961..0000000000 --- a/synapse/storage/schema/delta/56/unique_user_filter_index.py +++ /dev/null @@ -1,52 +0,0 @@ -import logging - -from synapse.storage.engines import PostgresEngine - -logger = logging.getLogger(__name__) - - -""" -This migration updates the user_filters table as follows: - - - drops any (user_id, filter_id) duplicates - - makes the columns NON-NULLable - - turns the index into a UNIQUE index -""" - - -def run_upgrade(cur, database_engine, *args, **kwargs): - pass - - -def run_create(cur, database_engine, *args, **kwargs): - if isinstance(database_engine, PostgresEngine): - select_clause = """ - SELECT DISTINCT ON (user_id, filter_id) user_id, filter_id, filter_json - FROM user_filters - """ - else: - select_clause = """ - SELECT * FROM user_filters GROUP BY user_id, filter_id - """ - sql = """ - DROP TABLE IF EXISTS user_filters_migration; - DROP INDEX IF EXISTS user_filters_unique; - CREATE TABLE user_filters_migration ( - user_id TEXT NOT NULL, - filter_id BIGINT NOT NULL, - filter_json BYTEA NOT NULL - ); - INSERT INTO user_filters_migration (user_id, filter_id, filter_json) - %s; - CREATE UNIQUE INDEX user_filters_unique ON user_filters_migration - (user_id, filter_id); - DROP TABLE user_filters; - ALTER TABLE user_filters_migration RENAME TO user_filters; - """ % ( - select_clause, - ) - - if isinstance(database_engine, PostgresEngine): - cur.execute(sql) - else: - cur.executescript(sql) diff --git a/synapse/storage/schema/delta/56/user_external_ids.sql b/synapse/storage/schema/delta/56/user_external_ids.sql deleted file mode 100644 index 91390c4527..0000000000 --- a/synapse/storage/schema/delta/56/user_external_ids.sql +++ /dev/null @@ -1,24 +0,0 @@ -/* Copyright 2019 The Matrix.org Foundation C.I.C. - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -/* - * a table which records mappings from external auth providers to mxids - */ -CREATE TABLE IF NOT EXISTS user_external_ids ( - auth_provider TEXT NOT NULL, - external_id TEXT NOT NULL, - user_id TEXT NOT NULL, - UNIQUE (auth_provider, external_id) -); diff --git a/synapse/storage/schema/delta/56/users_in_public_rooms_idx.sql b/synapse/storage/schema/delta/56/users_in_public_rooms_idx.sql deleted file mode 100644 index 149f8be8b6..0000000000 --- a/synapse/storage/schema/delta/56/users_in_public_rooms_idx.sql +++ /dev/null @@ -1,17 +0,0 @@ -/* Copyright 2019 Matrix.org Foundation CIC - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- this was apparently forgotten when the table was created back in delta 53. -CREATE INDEX users_in_public_rooms_r_idx ON users_in_public_rooms(room_id); diff --git a/synapse/storage/schema/full_schemas/16/application_services.sql b/synapse/storage/schema/full_schemas/16/application_services.sql deleted file mode 100644 index 883fcd10b2..0000000000 --- a/synapse/storage/schema/full_schemas/16/application_services.sql +++ /dev/null @@ -1,37 +0,0 @@ -/* Copyright 2015, 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -/* We used to create tables called application_services and - * application_services_regex, but these are no longer used and are removed in - * delta 54. - */ - - -CREATE TABLE IF NOT EXISTS application_services_state( - as_id TEXT PRIMARY KEY, - state VARCHAR(5), - last_txn INTEGER -); - -CREATE TABLE IF NOT EXISTS application_services_txns( - as_id TEXT NOT NULL, - txn_id INTEGER NOT NULL, - event_ids TEXT NOT NULL, - UNIQUE(as_id, txn_id) -); - -CREATE INDEX application_services_txns_id ON application_services_txns ( - as_id -); diff --git a/synapse/storage/schema/full_schemas/16/event_edges.sql b/synapse/storage/schema/full_schemas/16/event_edges.sql deleted file mode 100644 index 10ce2aa7a0..0000000000 --- a/synapse/storage/schema/full_schemas/16/event_edges.sql +++ /dev/null @@ -1,70 +0,0 @@ -/* Copyright 2014-2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -/* We used to create tables called event_destinations and - * state_forward_extremities, but these are no longer used and are removed in - * delta 54. - */ - -CREATE TABLE IF NOT EXISTS event_forward_extremities( - event_id TEXT NOT NULL, - room_id TEXT NOT NULL, - UNIQUE (event_id, room_id) -); - -CREATE INDEX ev_extrem_room ON event_forward_extremities(room_id); -CREATE INDEX ev_extrem_id ON event_forward_extremities(event_id); - - -CREATE TABLE IF NOT EXISTS event_backward_extremities( - event_id TEXT NOT NULL, - room_id TEXT NOT NULL, - UNIQUE (event_id, room_id) -); - -CREATE INDEX ev_b_extrem_room ON event_backward_extremities(room_id); -CREATE INDEX ev_b_extrem_id ON event_backward_extremities(event_id); - - -CREATE TABLE IF NOT EXISTS event_edges( - event_id TEXT NOT NULL, - prev_event_id TEXT NOT NULL, - room_id TEXT NOT NULL, - is_state BOOL NOT NULL, -- true if this is a prev_state edge rather than a regular - -- event dag edge. - UNIQUE (event_id, prev_event_id, room_id, is_state) -); - -CREATE INDEX ev_edges_id ON event_edges(event_id); -CREATE INDEX ev_edges_prev_id ON event_edges(prev_event_id); - - -CREATE TABLE IF NOT EXISTS room_depth( - room_id TEXT NOT NULL, - min_depth INTEGER NOT NULL, - UNIQUE (room_id) -); - -CREATE INDEX room_depth_room ON room_depth(room_id); - -CREATE TABLE IF NOT EXISTS event_auth( - event_id TEXT NOT NULL, - auth_id TEXT NOT NULL, - room_id TEXT NOT NULL, - UNIQUE (event_id, auth_id, room_id) -); - -CREATE INDEX evauth_edges_id ON event_auth(event_id); -CREATE INDEX evauth_edges_auth_id ON event_auth(auth_id); diff --git a/synapse/storage/schema/full_schemas/16/event_signatures.sql b/synapse/storage/schema/full_schemas/16/event_signatures.sql deleted file mode 100644 index 95826da431..0000000000 --- a/synapse/storage/schema/full_schemas/16/event_signatures.sql +++ /dev/null @@ -1,38 +0,0 @@ -/* Copyright 2014-2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - /* We used to create tables called event_content_hashes and event_edge_hashes, - * but these are no longer used and are removed in delta 54. - */ - -CREATE TABLE IF NOT EXISTS event_reference_hashes ( - event_id TEXT, - algorithm TEXT, - hash bytea, - UNIQUE (event_id, algorithm) -); - -CREATE INDEX event_reference_hashes_id ON event_reference_hashes(event_id); - - -CREATE TABLE IF NOT EXISTS event_signatures ( - event_id TEXT, - signature_name TEXT, - key_id TEXT, - signature bytea, - UNIQUE (event_id, signature_name, key_id) -); - -CREATE INDEX event_signatures_id ON event_signatures(event_id); diff --git a/synapse/storage/schema/full_schemas/16/im.sql b/synapse/storage/schema/full_schemas/16/im.sql deleted file mode 100644 index a1a2aa8e5b..0000000000 --- a/synapse/storage/schema/full_schemas/16/im.sql +++ /dev/null @@ -1,120 +0,0 @@ -/* Copyright 2014-2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -/* We used to create tables called room_hosts and feedback, - * but these are no longer used and are removed in delta 54. - */ - -CREATE TABLE IF NOT EXISTS events( - stream_ordering INTEGER PRIMARY KEY, - topological_ordering BIGINT NOT NULL, - event_id TEXT NOT NULL, - type TEXT NOT NULL, - room_id TEXT NOT NULL, - - -- 'content' used to be created NULLable, but as of delta 50 we drop that constraint. - -- the hack we use to drop the constraint doesn't work for an in-memory sqlite - -- database, which breaks the sytests. Hence, we no longer make it nullable. - content TEXT, - - unrecognized_keys TEXT, - processed BOOL NOT NULL, - outlier BOOL NOT NULL, - depth BIGINT DEFAULT 0 NOT NULL, - UNIQUE (event_id) -); - -CREATE INDEX events_stream_ordering ON events (stream_ordering); -CREATE INDEX events_topological_ordering ON events (topological_ordering); -CREATE INDEX events_order ON events (topological_ordering, stream_ordering); -CREATE INDEX events_room_id ON events (room_id); -CREATE INDEX events_order_room ON events ( - room_id, topological_ordering, stream_ordering -); - - -CREATE TABLE IF NOT EXISTS event_json( - event_id TEXT NOT NULL, - room_id TEXT NOT NULL, - internal_metadata TEXT NOT NULL, - json TEXT NOT NULL, - UNIQUE (event_id) -); - -CREATE INDEX event_json_room_id ON event_json(room_id); - - -CREATE TABLE IF NOT EXISTS state_events( - event_id TEXT NOT NULL, - room_id TEXT NOT NULL, - type TEXT NOT NULL, - state_key TEXT NOT NULL, - prev_state TEXT, - UNIQUE (event_id) -); - -CREATE INDEX state_events_room_id ON state_events (room_id); -CREATE INDEX state_events_type ON state_events (type); -CREATE INDEX state_events_state_key ON state_events (state_key); - - -CREATE TABLE IF NOT EXISTS current_state_events( - event_id TEXT NOT NULL, - room_id TEXT NOT NULL, - type TEXT NOT NULL, - state_key TEXT NOT NULL, - UNIQUE (event_id), - UNIQUE (room_id, type, state_key) -); - -CREATE INDEX current_state_events_room_id ON current_state_events (room_id); -CREATE INDEX current_state_events_type ON current_state_events (type); -CREATE INDEX current_state_events_state_key ON current_state_events (state_key); - -CREATE TABLE IF NOT EXISTS room_memberships( - event_id TEXT NOT NULL, - user_id TEXT NOT NULL, - sender TEXT NOT NULL, - room_id TEXT NOT NULL, - membership TEXT NOT NULL, - UNIQUE (event_id) -); - -CREATE INDEX room_memberships_room_id ON room_memberships (room_id); -CREATE INDEX room_memberships_user_id ON room_memberships (user_id); - -CREATE TABLE IF NOT EXISTS topics( - event_id TEXT NOT NULL, - room_id TEXT NOT NULL, - topic TEXT NOT NULL, - UNIQUE (event_id) -); - -CREATE INDEX topics_room_id ON topics(room_id); - -CREATE TABLE IF NOT EXISTS room_names( - event_id TEXT NOT NULL, - room_id TEXT NOT NULL, - name TEXT NOT NULL, - UNIQUE (event_id) -); - -CREATE INDEX room_names_room_id ON room_names(room_id); - -CREATE TABLE IF NOT EXISTS rooms( - room_id TEXT PRIMARY KEY NOT NULL, - is_public BOOL, - creator TEXT -); diff --git a/synapse/storage/schema/full_schemas/16/keys.sql b/synapse/storage/schema/full_schemas/16/keys.sql deleted file mode 100644 index 11cdffdbb3..0000000000 --- a/synapse/storage/schema/full_schemas/16/keys.sql +++ /dev/null @@ -1,26 +0,0 @@ -/* Copyright 2014-2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - --- we used to create a table called server_tls_certificates, but this is no --- longer used, and is removed in delta 54. - -CREATE TABLE IF NOT EXISTS server_signature_keys( - server_name TEXT, -- Server name. - key_id TEXT, -- Key version. - from_server TEXT, -- Which key server the key was fetched form. - ts_added_ms BIGINT, -- When the key was added. - verify_key bytea, -- NACL verification key. - UNIQUE (server_name, key_id) -); diff --git a/synapse/storage/schema/full_schemas/16/media_repository.sql b/synapse/storage/schema/full_schemas/16/media_repository.sql deleted file mode 100644 index 8f3759bb2a..0000000000 --- a/synapse/storage/schema/full_schemas/16/media_repository.sql +++ /dev/null @@ -1,68 +0,0 @@ -/* Copyright 2014-2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -CREATE TABLE IF NOT EXISTS local_media_repository ( - media_id TEXT, -- The id used to refer to the media. - media_type TEXT, -- The MIME-type of the media. - media_length INTEGER, -- Length of the media in bytes. - created_ts BIGINT, -- When the content was uploaded in ms. - upload_name TEXT, -- The name the media was uploaded with. - user_id TEXT, -- The user who uploaded the file. - UNIQUE (media_id) -); - -CREATE TABLE IF NOT EXISTS local_media_repository_thumbnails ( - media_id TEXT, -- The id used to refer to the media. - thumbnail_width INTEGER, -- The width of the thumbnail in pixels. - thumbnail_height INTEGER, -- The height of the thumbnail in pixels. - thumbnail_type TEXT, -- The MIME-type of the thumbnail. - thumbnail_method TEXT, -- The method used to make the thumbnail. - thumbnail_length INTEGER, -- The length of the thumbnail in bytes. - UNIQUE ( - media_id, thumbnail_width, thumbnail_height, thumbnail_type - ) -); - -CREATE INDEX local_media_repository_thumbnails_media_id - ON local_media_repository_thumbnails (media_id); - -CREATE TABLE IF NOT EXISTS remote_media_cache ( - media_origin TEXT, -- The remote HS the media came from. - media_id TEXT, -- The id used to refer to the media on that server. - media_type TEXT, -- The MIME-type of the media. - created_ts BIGINT, -- When the content was uploaded in ms. - upload_name TEXT, -- The name the media was uploaded with. - media_length INTEGER, -- Length of the media in bytes. - filesystem_id TEXT, -- The name used to store the media on disk. - UNIQUE (media_origin, media_id) -); - -CREATE TABLE IF NOT EXISTS remote_media_cache_thumbnails ( - media_origin TEXT, -- The remote HS the media came from. - media_id TEXT, -- The id used to refer to the media. - thumbnail_width INTEGER, -- The width of the thumbnail in pixels. - thumbnail_height INTEGER, -- The height of the thumbnail in pixels. - thumbnail_method TEXT, -- The method used to make the thumbnail - thumbnail_type TEXT, -- The MIME-type of the thumbnail. - thumbnail_length INTEGER, -- The length of the thumbnail in bytes. - filesystem_id TEXT, -- The name used to store the media on disk. - UNIQUE ( - media_origin, media_id, thumbnail_width, thumbnail_height, - thumbnail_type - ) -); - -CREATE INDEX remote_media_cache_thumbnails_media_id - ON remote_media_cache_thumbnails (media_id); diff --git a/synapse/storage/schema/full_schemas/16/presence.sql b/synapse/storage/schema/full_schemas/16/presence.sql deleted file mode 100644 index 01d2d8f833..0000000000 --- a/synapse/storage/schema/full_schemas/16/presence.sql +++ /dev/null @@ -1,32 +0,0 @@ -/* Copyright 2014-2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ -CREATE TABLE IF NOT EXISTS presence( - user_id TEXT NOT NULL, - state VARCHAR(20), - status_msg TEXT, - mtime BIGINT, -- miliseconds since last state change - UNIQUE (user_id) -); - --- For each of /my/ users which possibly-remote users are allowed to see their --- presence state -CREATE TABLE IF NOT EXISTS presence_allow_inbound( - observed_user_id TEXT NOT NULL, - observer_user_id TEXT NOT NULL, -- a UserID, - UNIQUE (observed_user_id, observer_user_id) -); - --- We used to create a table called presence_list, but this is no longer used --- and is removed in delta 54. \ No newline at end of file diff --git a/synapse/storage/schema/full_schemas/16/profiles.sql b/synapse/storage/schema/full_schemas/16/profiles.sql deleted file mode 100644 index c04f4747d9..0000000000 --- a/synapse/storage/schema/full_schemas/16/profiles.sql +++ /dev/null @@ -1,20 +0,0 @@ -/* Copyright 2014-2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ -CREATE TABLE IF NOT EXISTS profiles( - user_id TEXT NOT NULL, - displayname TEXT, - avatar_url TEXT, - UNIQUE(user_id) -); diff --git a/synapse/storage/schema/full_schemas/16/push.sql b/synapse/storage/schema/full_schemas/16/push.sql deleted file mode 100644 index e44465cf45..0000000000 --- a/synapse/storage/schema/full_schemas/16/push.sql +++ /dev/null @@ -1,74 +0,0 @@ -/* Copyright 2015, 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -CREATE TABLE IF NOT EXISTS rejections( - event_id TEXT NOT NULL, - reason TEXT NOT NULL, - last_check TEXT NOT NULL, - UNIQUE (event_id) -); - --- Push notification endpoints that users have configured -CREATE TABLE IF NOT EXISTS pushers ( - id BIGINT PRIMARY KEY, - user_name TEXT NOT NULL, - access_token BIGINT DEFAULT NULL, - profile_tag VARCHAR(32) NOT NULL, - kind VARCHAR(8) NOT NULL, - app_id VARCHAR(64) NOT NULL, - app_display_name VARCHAR(64) NOT NULL, - device_display_name VARCHAR(128) NOT NULL, - pushkey bytea NOT NULL, - ts BIGINT NOT NULL, - lang VARCHAR(8), - data bytea, - last_token TEXT, - last_success BIGINT, - failing_since BIGINT, - UNIQUE (app_id, pushkey) -); - -CREATE TABLE IF NOT EXISTS push_rules ( - id BIGINT PRIMARY KEY, - user_name TEXT NOT NULL, - rule_id TEXT NOT NULL, - priority_class SMALLINT NOT NULL, - priority INTEGER NOT NULL DEFAULT 0, - conditions TEXT NOT NULL, - actions TEXT NOT NULL, - UNIQUE(user_name, rule_id) -); - -CREATE INDEX push_rules_user_name on push_rules (user_name); - -CREATE TABLE IF NOT EXISTS user_filters( - user_id TEXT, - filter_id BIGINT, - filter_json bytea -); - -CREATE INDEX user_filters_by_user_id_filter_id ON user_filters( - user_id, filter_id -); - -CREATE TABLE IF NOT EXISTS push_rules_enable ( - id BIGINT PRIMARY KEY, - user_name TEXT NOT NULL, - rule_id TEXT NOT NULL, - enabled SMALLINT, - UNIQUE(user_name, rule_id) -); - -CREATE INDEX push_rules_enable_user_name on push_rules_enable (user_name); diff --git a/synapse/storage/schema/full_schemas/16/redactions.sql b/synapse/storage/schema/full_schemas/16/redactions.sql deleted file mode 100644 index 318f0d9aa5..0000000000 --- a/synapse/storage/schema/full_schemas/16/redactions.sql +++ /dev/null @@ -1,22 +0,0 @@ -/* Copyright 2014-2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ -CREATE TABLE IF NOT EXISTS redactions ( - event_id TEXT NOT NULL, - redacts TEXT NOT NULL, - UNIQUE (event_id) -); - -CREATE INDEX redactions_event_id ON redactions (event_id); -CREATE INDEX redactions_redacts ON redactions (redacts); diff --git a/synapse/storage/schema/full_schemas/16/room_aliases.sql b/synapse/storage/schema/full_schemas/16/room_aliases.sql deleted file mode 100644 index d47da3b12f..0000000000 --- a/synapse/storage/schema/full_schemas/16/room_aliases.sql +++ /dev/null @@ -1,29 +0,0 @@ -/* Copyright 2014-2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -CREATE TABLE IF NOT EXISTS room_aliases( - room_alias TEXT NOT NULL, - room_id TEXT NOT NULL, - UNIQUE (room_alias) -); - -CREATE INDEX room_aliases_id ON room_aliases(room_id); - -CREATE TABLE IF NOT EXISTS room_alias_servers( - room_alias TEXT NOT NULL, - server TEXT NOT NULL -); - -CREATE INDEX room_alias_servers_alias ON room_alias_servers(room_alias); diff --git a/synapse/storage/schema/full_schemas/16/state.sql b/synapse/storage/schema/full_schemas/16/state.sql deleted file mode 100644 index 96391a8f0e..0000000000 --- a/synapse/storage/schema/full_schemas/16/state.sql +++ /dev/null @@ -1,40 +0,0 @@ -/* Copyright 2014-2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -CREATE TABLE IF NOT EXISTS state_groups( - id BIGINT PRIMARY KEY, - room_id TEXT NOT NULL, - event_id TEXT NOT NULL -); - -CREATE TABLE IF NOT EXISTS state_groups_state( - state_group BIGINT NOT NULL, - room_id TEXT NOT NULL, - type TEXT NOT NULL, - state_key TEXT NOT NULL, - event_id TEXT NOT NULL -); - -CREATE TABLE IF NOT EXISTS event_to_state_groups( - event_id TEXT NOT NULL, - state_group BIGINT NOT NULL, - UNIQUE (event_id) -); - -CREATE INDEX state_groups_id ON state_groups(id); - -CREATE INDEX state_groups_state_id ON state_groups_state(state_group); -CREATE INDEX state_groups_state_tuple ON state_groups_state(room_id, type, state_key); -CREATE INDEX event_to_state_groups_id ON event_to_state_groups(event_id); diff --git a/synapse/storage/schema/full_schemas/16/transactions.sql b/synapse/storage/schema/full_schemas/16/transactions.sql deleted file mode 100644 index 17e67bedac..0000000000 --- a/synapse/storage/schema/full_schemas/16/transactions.sql +++ /dev/null @@ -1,44 +0,0 @@ -/* Copyright 2014-2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ --- Stores what transaction ids we have received and what our response was -CREATE TABLE IF NOT EXISTS received_transactions( - transaction_id TEXT, - origin TEXT, - ts BIGINT, - response_code INTEGER, - response_json bytea, - has_been_referenced smallint default 0, -- Whether thishas been referenced by a prev_tx - UNIQUE (transaction_id, origin) -); - -CREATE INDEX transactions_have_ref ON received_transactions(origin, has_been_referenced);-- WHERE has_been_referenced = 0; - --- For sent transactions only. -CREATE TABLE IF NOT EXISTS transaction_id_to_pdu( - transaction_id INTEGER, - destination TEXT, - pdu_id TEXT, - pdu_origin TEXT, - UNIQUE (transaction_id, destination) -); - -CREATE INDEX transaction_id_to_pdu_dest ON transaction_id_to_pdu(destination); - --- To track destination health -CREATE TABLE IF NOT EXISTS destinations( - destination TEXT PRIMARY KEY, - retry_last_ts BIGINT, - retry_interval INTEGER -); diff --git a/synapse/storage/schema/full_schemas/16/users.sql b/synapse/storage/schema/full_schemas/16/users.sql deleted file mode 100644 index f013aa8b18..0000000000 --- a/synapse/storage/schema/full_schemas/16/users.sql +++ /dev/null @@ -1,42 +0,0 @@ -/* Copyright 2014-2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ -CREATE TABLE IF NOT EXISTS users( - name TEXT, - password_hash TEXT, - creation_ts BIGINT, - admin SMALLINT DEFAULT 0 NOT NULL, - UNIQUE(name) -); - -CREATE TABLE IF NOT EXISTS access_tokens( - id BIGINT PRIMARY KEY, - user_id TEXT NOT NULL, - device_id TEXT, - token TEXT NOT NULL, - last_used BIGINT, - UNIQUE(token) -); - -CREATE TABLE IF NOT EXISTS user_ips ( - user_id TEXT NOT NULL, - access_token TEXT NOT NULL, - device_id TEXT, - ip TEXT NOT NULL, - user_agent TEXT NOT NULL, - last_seen BIGINT NOT NULL -); - -CREATE INDEX user_ips_user ON user_ips(user_id); -CREATE INDEX user_ips_user_ip ON user_ips(user_id, access_token, ip); diff --git a/synapse/storage/schema/full_schemas/54/full.sql b/synapse/storage/schema/full_schemas/54/full.sql new file mode 100644 index 0000000000..1005880466 --- /dev/null +++ b/synapse/storage/schema/full_schemas/54/full.sql @@ -0,0 +1,8 @@ + + +CREATE TABLE background_updates ( + update_name text NOT NULL, + progress_json text NOT NULL, + depends_on text, + CONSTRAINT background_updates_uniqueness UNIQUE (update_name) +); diff --git a/synapse/storage/schema/full_schemas/54/full.sql.postgres b/synapse/storage/schema/full_schemas/54/full.sql.postgres deleted file mode 100644 index 098434356f..0000000000 --- a/synapse/storage/schema/full_schemas/54/full.sql.postgres +++ /dev/null @@ -1,2052 +0,0 @@ - - - - - -CREATE TABLE access_tokens ( - id bigint NOT NULL, - user_id text NOT NULL, - device_id text, - token text NOT NULL, - last_used bigint -); - - - -CREATE TABLE account_data ( - user_id text NOT NULL, - account_data_type text NOT NULL, - stream_id bigint NOT NULL, - content text NOT NULL -); - - - -CREATE TABLE account_data_max_stream_id ( - lock character(1) DEFAULT 'X'::bpchar NOT NULL, - stream_id bigint NOT NULL, - CONSTRAINT private_user_data_max_stream_id_lock_check CHECK ((lock = 'X'::bpchar)) -); - - - -CREATE TABLE account_validity ( - user_id text NOT NULL, - expiration_ts_ms bigint NOT NULL, - email_sent boolean NOT NULL, - renewal_token text -); - - - -CREATE TABLE application_services_state ( - as_id text NOT NULL, - state character varying(5), - last_txn integer -); - - - -CREATE TABLE application_services_txns ( - as_id text NOT NULL, - txn_id integer NOT NULL, - event_ids text NOT NULL -); - - - -CREATE TABLE appservice_room_list ( - appservice_id text NOT NULL, - network_id text NOT NULL, - room_id text NOT NULL -); - - - -CREATE TABLE appservice_stream_position ( - lock character(1) DEFAULT 'X'::bpchar NOT NULL, - stream_ordering bigint, - CONSTRAINT appservice_stream_position_lock_check CHECK ((lock = 'X'::bpchar)) -); - - - -CREATE TABLE background_updates ( - update_name text NOT NULL, - progress_json text NOT NULL, - depends_on text -); - - - -CREATE TABLE blocked_rooms ( - room_id text NOT NULL, - user_id text NOT NULL -); - - - -CREATE TABLE cache_invalidation_stream ( - stream_id bigint, - cache_func text, - keys text[], - invalidation_ts bigint -); - - - -CREATE TABLE current_state_delta_stream ( - stream_id bigint NOT NULL, - room_id text NOT NULL, - type text NOT NULL, - state_key text NOT NULL, - event_id text, - prev_event_id text -); - - - -CREATE TABLE current_state_events ( - event_id text NOT NULL, - room_id text NOT NULL, - type text NOT NULL, - state_key text NOT NULL -); - - - -CREATE TABLE deleted_pushers ( - stream_id bigint NOT NULL, - app_id text NOT NULL, - pushkey text NOT NULL, - user_id text NOT NULL -); - - - -CREATE TABLE destinations ( - destination text NOT NULL, - retry_last_ts bigint, - retry_interval integer -); - - - -CREATE TABLE device_federation_inbox ( - origin text NOT NULL, - message_id text NOT NULL, - received_ts bigint NOT NULL -); - - - -CREATE TABLE device_federation_outbox ( - destination text NOT NULL, - stream_id bigint NOT NULL, - queued_ts bigint NOT NULL, - messages_json text NOT NULL -); - - - -CREATE TABLE device_inbox ( - user_id text NOT NULL, - device_id text NOT NULL, - stream_id bigint NOT NULL, - message_json text NOT NULL -); - - - -CREATE TABLE device_lists_outbound_last_success ( - destination text NOT NULL, - user_id text NOT NULL, - stream_id bigint NOT NULL -); - - - -CREATE TABLE device_lists_outbound_pokes ( - destination text NOT NULL, - stream_id bigint NOT NULL, - user_id text NOT NULL, - device_id text NOT NULL, - sent boolean NOT NULL, - ts bigint NOT NULL -); - - - -CREATE TABLE device_lists_remote_cache ( - user_id text NOT NULL, - device_id text NOT NULL, - content text NOT NULL -); - - - -CREATE TABLE device_lists_remote_extremeties ( - user_id text NOT NULL, - stream_id text NOT NULL -); - - - -CREATE TABLE device_lists_stream ( - stream_id bigint NOT NULL, - user_id text NOT NULL, - device_id text NOT NULL -); - - - -CREATE TABLE device_max_stream_id ( - stream_id bigint NOT NULL -); - - - -CREATE TABLE devices ( - user_id text NOT NULL, - device_id text NOT NULL, - display_name text -); - - - -CREATE TABLE e2e_device_keys_json ( - user_id text NOT NULL, - device_id text NOT NULL, - ts_added_ms bigint NOT NULL, - key_json text NOT NULL -); - - - -CREATE TABLE e2e_one_time_keys_json ( - user_id text NOT NULL, - device_id text NOT NULL, - algorithm text NOT NULL, - key_id text NOT NULL, - ts_added_ms bigint NOT NULL, - key_json text NOT NULL -); - - - -CREATE TABLE e2e_room_keys ( - user_id text NOT NULL, - room_id text NOT NULL, - session_id text NOT NULL, - version bigint NOT NULL, - first_message_index integer, - forwarded_count integer, - is_verified boolean, - session_data text NOT NULL -); - - - -CREATE TABLE e2e_room_keys_versions ( - user_id text NOT NULL, - version bigint NOT NULL, - algorithm text NOT NULL, - auth_data text NOT NULL, - deleted smallint DEFAULT 0 NOT NULL -); - - - -CREATE TABLE erased_users ( - user_id text NOT NULL -); - - - -CREATE TABLE event_auth ( - event_id text NOT NULL, - auth_id text NOT NULL, - room_id text NOT NULL -); - - - -CREATE TABLE event_backward_extremities ( - event_id text NOT NULL, - room_id text NOT NULL -); - - - -CREATE TABLE event_edges ( - event_id text NOT NULL, - prev_event_id text NOT NULL, - room_id text NOT NULL, - is_state boolean NOT NULL -); - - - -CREATE TABLE event_forward_extremities ( - event_id text NOT NULL, - room_id text NOT NULL -); - - - -CREATE TABLE event_json ( - event_id text NOT NULL, - room_id text NOT NULL, - internal_metadata text NOT NULL, - json text NOT NULL, - format_version integer -); - - - -CREATE TABLE event_push_actions ( - room_id text NOT NULL, - event_id text NOT NULL, - user_id text NOT NULL, - profile_tag character varying(32), - actions text NOT NULL, - topological_ordering bigint, - stream_ordering bigint, - notif smallint, - highlight smallint -); - - - -CREATE TABLE event_push_actions_staging ( - event_id text NOT NULL, - user_id text NOT NULL, - actions text NOT NULL, - notif smallint NOT NULL, - highlight smallint NOT NULL -); - - - -CREATE TABLE event_push_summary ( - user_id text NOT NULL, - room_id text NOT NULL, - notif_count bigint NOT NULL, - stream_ordering bigint NOT NULL -); - - - -CREATE TABLE event_push_summary_stream_ordering ( - lock character(1) DEFAULT 'X'::bpchar NOT NULL, - stream_ordering bigint NOT NULL, - CONSTRAINT event_push_summary_stream_ordering_lock_check CHECK ((lock = 'X'::bpchar)) -); - - - -CREATE TABLE event_reference_hashes ( - event_id text, - algorithm text, - hash bytea -); - - - -CREATE TABLE event_relations ( - event_id text NOT NULL, - relates_to_id text NOT NULL, - relation_type text NOT NULL, - aggregation_key text -); - - - -CREATE TABLE event_reports ( - id bigint NOT NULL, - received_ts bigint NOT NULL, - room_id text NOT NULL, - event_id text NOT NULL, - user_id text NOT NULL, - reason text, - content text -); - - - -CREATE TABLE event_search ( - event_id text, - room_id text, - sender text, - key text, - vector tsvector, - origin_server_ts bigint, - stream_ordering bigint -); - - - -CREATE TABLE event_to_state_groups ( - event_id text NOT NULL, - state_group bigint NOT NULL -); - - - -CREATE TABLE events ( - stream_ordering integer NOT NULL, - topological_ordering bigint NOT NULL, - event_id text NOT NULL, - type text NOT NULL, - room_id text NOT NULL, - content text, - unrecognized_keys text, - processed boolean NOT NULL, - outlier boolean NOT NULL, - depth bigint DEFAULT 0 NOT NULL, - origin_server_ts bigint, - received_ts bigint, - sender text, - contains_url boolean -); - - - -CREATE TABLE ex_outlier_stream ( - event_stream_ordering bigint NOT NULL, - event_id text NOT NULL, - state_group bigint NOT NULL -); - - - -CREATE TABLE federation_stream_position ( - type text NOT NULL, - stream_id integer NOT NULL -); - - - -CREATE TABLE group_attestations_remote ( - group_id text NOT NULL, - user_id text NOT NULL, - valid_until_ms bigint NOT NULL, - attestation_json text NOT NULL -); - - - -CREATE TABLE group_attestations_renewals ( - group_id text NOT NULL, - user_id text NOT NULL, - valid_until_ms bigint NOT NULL -); - - - -CREATE TABLE group_invites ( - group_id text NOT NULL, - user_id text NOT NULL -); - - - -CREATE TABLE group_roles ( - group_id text NOT NULL, - role_id text NOT NULL, - profile text NOT NULL, - is_public boolean NOT NULL -); - - - -CREATE TABLE group_room_categories ( - group_id text NOT NULL, - category_id text NOT NULL, - profile text NOT NULL, - is_public boolean NOT NULL -); - - - -CREATE TABLE group_rooms ( - group_id text NOT NULL, - room_id text NOT NULL, - is_public boolean NOT NULL -); - - - -CREATE TABLE group_summary_roles ( - group_id text NOT NULL, - role_id text NOT NULL, - role_order bigint NOT NULL, - CONSTRAINT group_summary_roles_role_order_check CHECK ((role_order > 0)) -); - - - -CREATE TABLE group_summary_room_categories ( - group_id text NOT NULL, - category_id text NOT NULL, - cat_order bigint NOT NULL, - CONSTRAINT group_summary_room_categories_cat_order_check CHECK ((cat_order > 0)) -); - - - -CREATE TABLE group_summary_rooms ( - group_id text NOT NULL, - room_id text NOT NULL, - category_id text NOT NULL, - room_order bigint NOT NULL, - is_public boolean NOT NULL, - CONSTRAINT group_summary_rooms_room_order_check CHECK ((room_order > 0)) -); - - - -CREATE TABLE group_summary_users ( - group_id text NOT NULL, - user_id text NOT NULL, - role_id text NOT NULL, - user_order bigint NOT NULL, - is_public boolean NOT NULL -); - - - -CREATE TABLE group_users ( - group_id text NOT NULL, - user_id text NOT NULL, - is_admin boolean NOT NULL, - is_public boolean NOT NULL -); - - - -CREATE TABLE groups ( - group_id text NOT NULL, - name text, - avatar_url text, - short_description text, - long_description text, - is_public boolean NOT NULL, - join_policy text DEFAULT 'invite'::text NOT NULL -); - - - -CREATE TABLE guest_access ( - event_id text NOT NULL, - room_id text NOT NULL, - guest_access text NOT NULL -); - - - -CREATE TABLE history_visibility ( - event_id text NOT NULL, - room_id text NOT NULL, - history_visibility text NOT NULL -); - - - -CREATE TABLE local_group_membership ( - group_id text NOT NULL, - user_id text NOT NULL, - is_admin boolean NOT NULL, - membership text NOT NULL, - is_publicised boolean NOT NULL, - content text NOT NULL -); - - - -CREATE TABLE local_group_updates ( - stream_id bigint NOT NULL, - group_id text NOT NULL, - user_id text NOT NULL, - type text NOT NULL, - content text NOT NULL -); - - - -CREATE TABLE local_invites ( - stream_id bigint NOT NULL, - inviter text NOT NULL, - invitee text NOT NULL, - event_id text NOT NULL, - room_id text NOT NULL, - locally_rejected text, - replaced_by text -); - - - -CREATE TABLE local_media_repository ( - media_id text, - media_type text, - media_length integer, - created_ts bigint, - upload_name text, - user_id text, - quarantined_by text, - url_cache text, - last_access_ts bigint -); - - - -CREATE TABLE local_media_repository_thumbnails ( - media_id text, - thumbnail_width integer, - thumbnail_height integer, - thumbnail_type text, - thumbnail_method text, - thumbnail_length integer -); - - - -CREATE TABLE local_media_repository_url_cache ( - url text, - response_code integer, - etag text, - expires_ts bigint, - og text, - media_id text, - download_ts bigint -); - - - -CREATE TABLE monthly_active_users ( - user_id text NOT NULL, - "timestamp" bigint NOT NULL -); - - - -CREATE TABLE open_id_tokens ( - token text NOT NULL, - ts_valid_until_ms bigint NOT NULL, - user_id text NOT NULL -); - - - -CREATE TABLE presence ( - user_id text NOT NULL, - state character varying(20), - status_msg text, - mtime bigint -); - - - -CREATE TABLE presence_allow_inbound ( - observed_user_id text NOT NULL, - observer_user_id text NOT NULL -); - - - -CREATE TABLE presence_stream ( - stream_id bigint, - user_id text, - state text, - last_active_ts bigint, - last_federation_update_ts bigint, - last_user_sync_ts bigint, - status_msg text, - currently_active boolean -); - - - -CREATE TABLE profiles ( - user_id text NOT NULL, - displayname text, - avatar_url text -); - - - -CREATE TABLE public_room_list_stream ( - stream_id bigint NOT NULL, - room_id text NOT NULL, - visibility boolean NOT NULL, - appservice_id text, - network_id text -); - - - -CREATE TABLE push_rules ( - id bigint NOT NULL, - user_name text NOT NULL, - rule_id text NOT NULL, - priority_class smallint NOT NULL, - priority integer DEFAULT 0 NOT NULL, - conditions text NOT NULL, - actions text NOT NULL -); - - - -CREATE TABLE push_rules_enable ( - id bigint NOT NULL, - user_name text NOT NULL, - rule_id text NOT NULL, - enabled smallint -); - - - -CREATE TABLE push_rules_stream ( - stream_id bigint NOT NULL, - event_stream_ordering bigint NOT NULL, - user_id text NOT NULL, - rule_id text NOT NULL, - op text NOT NULL, - priority_class smallint, - priority integer, - conditions text, - actions text -); - - - -CREATE TABLE pusher_throttle ( - pusher bigint NOT NULL, - room_id text NOT NULL, - last_sent_ts bigint, - throttle_ms bigint -); - - - -CREATE TABLE pushers ( - id bigint NOT NULL, - user_name text NOT NULL, - access_token bigint, - profile_tag text NOT NULL, - kind text NOT NULL, - app_id text NOT NULL, - app_display_name text NOT NULL, - device_display_name text NOT NULL, - pushkey text NOT NULL, - ts bigint NOT NULL, - lang text, - data text, - last_stream_ordering integer, - last_success bigint, - failing_since bigint -); - - - -CREATE TABLE ratelimit_override ( - user_id text NOT NULL, - messages_per_second bigint, - burst_count bigint -); - - - -CREATE TABLE receipts_graph ( - room_id text NOT NULL, - receipt_type text NOT NULL, - user_id text NOT NULL, - event_ids text NOT NULL, - data text NOT NULL -); - - - -CREATE TABLE receipts_linearized ( - stream_id bigint NOT NULL, - room_id text NOT NULL, - receipt_type text NOT NULL, - user_id text NOT NULL, - event_id text NOT NULL, - data text NOT NULL -); - - - -CREATE TABLE received_transactions ( - transaction_id text, - origin text, - ts bigint, - response_code integer, - response_json bytea, - has_been_referenced smallint DEFAULT 0 -); - - - -CREATE TABLE redactions ( - event_id text NOT NULL, - redacts text NOT NULL -); - - - -CREATE TABLE rejections ( - event_id text NOT NULL, - reason text NOT NULL, - last_check text NOT NULL -); - - - -CREATE TABLE remote_media_cache ( - media_origin text, - media_id text, - media_type text, - created_ts bigint, - upload_name text, - media_length integer, - filesystem_id text, - last_access_ts bigint, - quarantined_by text -); - - - -CREATE TABLE remote_media_cache_thumbnails ( - media_origin text, - media_id text, - thumbnail_width integer, - thumbnail_height integer, - thumbnail_method text, - thumbnail_type text, - thumbnail_length integer, - filesystem_id text -); - - - -CREATE TABLE remote_profile_cache ( - user_id text NOT NULL, - displayname text, - avatar_url text, - last_check bigint NOT NULL -); - - - -CREATE TABLE room_account_data ( - user_id text NOT NULL, - room_id text NOT NULL, - account_data_type text NOT NULL, - stream_id bigint NOT NULL, - content text NOT NULL -); - - - -CREATE TABLE room_alias_servers ( - room_alias text NOT NULL, - server text NOT NULL -); - - - -CREATE TABLE room_aliases ( - room_alias text NOT NULL, - room_id text NOT NULL, - creator text -); - - - -CREATE TABLE room_depth ( - room_id text NOT NULL, - min_depth integer NOT NULL -); - - - -CREATE TABLE room_memberships ( - event_id text NOT NULL, - user_id text NOT NULL, - sender text NOT NULL, - room_id text NOT NULL, - membership text NOT NULL, - forgotten integer DEFAULT 0, - display_name text, - avatar_url text -); - - - -CREATE TABLE room_names ( - event_id text NOT NULL, - room_id text NOT NULL, - name text NOT NULL -); - - - -CREATE TABLE room_state ( - room_id text NOT NULL, - join_rules text, - history_visibility text, - encryption text, - name text, - topic text, - avatar text, - canonical_alias text -); - - - -CREATE TABLE room_stats ( - room_id text NOT NULL, - ts bigint NOT NULL, - bucket_size integer NOT NULL, - current_state_events integer NOT NULL, - joined_members integer NOT NULL, - invited_members integer NOT NULL, - left_members integer NOT NULL, - banned_members integer NOT NULL, - state_events integer NOT NULL -); - - - -CREATE TABLE room_stats_earliest_token ( - room_id text NOT NULL, - token bigint NOT NULL -); - - - -CREATE TABLE room_tags ( - user_id text NOT NULL, - room_id text NOT NULL, - tag text NOT NULL, - content text NOT NULL -); - - - -CREATE TABLE room_tags_revisions ( - user_id text NOT NULL, - room_id text NOT NULL, - stream_id bigint NOT NULL -); - - - -CREATE TABLE rooms ( - room_id text NOT NULL, - is_public boolean, - creator text -); - - - -CREATE TABLE server_keys_json ( - server_name text NOT NULL, - key_id text NOT NULL, - from_server text NOT NULL, - ts_added_ms bigint NOT NULL, - ts_valid_until_ms bigint NOT NULL, - key_json bytea NOT NULL -); - - - -CREATE TABLE server_signature_keys ( - server_name text, - key_id text, - from_server text, - ts_added_ms bigint, - verify_key bytea, - ts_valid_until_ms bigint -); - - - -CREATE TABLE state_events ( - event_id text NOT NULL, - room_id text NOT NULL, - type text NOT NULL, - state_key text NOT NULL, - prev_state text -); - - - -CREATE TABLE state_group_edges ( - state_group bigint NOT NULL, - prev_state_group bigint NOT NULL -); - - - -CREATE SEQUENCE state_group_id_seq - START WITH 1 - INCREMENT BY 1 - NO MINVALUE - NO MAXVALUE - CACHE 1; - - - -CREATE TABLE state_groups ( - id bigint NOT NULL, - room_id text NOT NULL, - event_id text NOT NULL -); - - - -CREATE TABLE state_groups_state ( - state_group bigint NOT NULL, - room_id text NOT NULL, - type text NOT NULL, - state_key text NOT NULL, - event_id text NOT NULL -); - - - -CREATE TABLE stats_stream_pos ( - lock character(1) DEFAULT 'X'::bpchar NOT NULL, - stream_id bigint, - CONSTRAINT stats_stream_pos_lock_check CHECK ((lock = 'X'::bpchar)) -); - - - -CREATE TABLE stream_ordering_to_exterm ( - stream_ordering bigint NOT NULL, - room_id text NOT NULL, - event_id text NOT NULL -); - - - -CREATE TABLE threepid_guest_access_tokens ( - medium text, - address text, - guest_access_token text, - first_inviter text -); - - - -CREATE TABLE topics ( - event_id text NOT NULL, - room_id text NOT NULL, - topic text NOT NULL -); - - - -CREATE TABLE user_daily_visits ( - user_id text NOT NULL, - device_id text, - "timestamp" bigint NOT NULL -); - - - -CREATE TABLE user_directory ( - user_id text NOT NULL, - room_id text, - display_name text, - avatar_url text -); - - - -CREATE TABLE user_directory_search ( - user_id text NOT NULL, - vector tsvector -); - - - -CREATE TABLE user_directory_stream_pos ( - lock character(1) DEFAULT 'X'::bpchar NOT NULL, - stream_id bigint, - CONSTRAINT user_directory_stream_pos_lock_check CHECK ((lock = 'X'::bpchar)) -); - - - -CREATE TABLE user_filters ( - user_id text, - filter_id bigint, - filter_json bytea -); - - - -CREATE TABLE user_ips ( - user_id text NOT NULL, - access_token text NOT NULL, - device_id text, - ip text NOT NULL, - user_agent text NOT NULL, - last_seen bigint NOT NULL -); - - - -CREATE TABLE user_stats ( - user_id text NOT NULL, - ts bigint NOT NULL, - bucket_size integer NOT NULL, - public_rooms integer NOT NULL, - private_rooms integer NOT NULL -); - - - -CREATE TABLE user_threepid_id_server ( - user_id text NOT NULL, - medium text NOT NULL, - address text NOT NULL, - id_server text NOT NULL -); - - - -CREATE TABLE user_threepids ( - user_id text NOT NULL, - medium text NOT NULL, - address text NOT NULL, - validated_at bigint NOT NULL, - added_at bigint NOT NULL -); - - - -CREATE TABLE users ( - name text, - password_hash text, - creation_ts bigint, - admin smallint DEFAULT 0 NOT NULL, - upgrade_ts bigint, - is_guest smallint DEFAULT 0 NOT NULL, - appservice_id text, - consent_version text, - consent_server_notice_sent text, - user_type text -); - - - -CREATE TABLE users_in_public_rooms ( - user_id text NOT NULL, - room_id text NOT NULL -); - - - -CREATE TABLE users_pending_deactivation ( - user_id text NOT NULL -); - - - -CREATE TABLE users_who_share_private_rooms ( - user_id text NOT NULL, - other_user_id text NOT NULL, - room_id text NOT NULL -); - - - -ALTER TABLE ONLY access_tokens - ADD CONSTRAINT access_tokens_pkey PRIMARY KEY (id); - - - -ALTER TABLE ONLY access_tokens - ADD CONSTRAINT access_tokens_token_key UNIQUE (token); - - - -ALTER TABLE ONLY account_data - ADD CONSTRAINT account_data_uniqueness UNIQUE (user_id, account_data_type); - - - -ALTER TABLE ONLY account_validity - ADD CONSTRAINT account_validity_pkey PRIMARY KEY (user_id); - - - -ALTER TABLE ONLY application_services_state - ADD CONSTRAINT application_services_state_pkey PRIMARY KEY (as_id); - - - -ALTER TABLE ONLY application_services_txns - ADD CONSTRAINT application_services_txns_as_id_txn_id_key UNIQUE (as_id, txn_id); - - - -ALTER TABLE ONLY appservice_stream_position - ADD CONSTRAINT appservice_stream_position_lock_key UNIQUE (lock); - - - -ALTER TABLE ONLY background_updates - ADD CONSTRAINT background_updates_uniqueness UNIQUE (update_name); - - - -ALTER TABLE ONLY current_state_events - ADD CONSTRAINT current_state_events_event_id_key UNIQUE (event_id); - - - -ALTER TABLE ONLY current_state_events - ADD CONSTRAINT current_state_events_room_id_type_state_key_key UNIQUE (room_id, type, state_key); - - - -ALTER TABLE ONLY destinations - ADD CONSTRAINT destinations_pkey PRIMARY KEY (destination); - - - -ALTER TABLE ONLY devices - ADD CONSTRAINT device_uniqueness UNIQUE (user_id, device_id); - - - -ALTER TABLE ONLY e2e_device_keys_json - ADD CONSTRAINT e2e_device_keys_json_uniqueness UNIQUE (user_id, device_id); - - - -ALTER TABLE ONLY e2e_one_time_keys_json - ADD CONSTRAINT e2e_one_time_keys_json_uniqueness UNIQUE (user_id, device_id, algorithm, key_id); - - - -ALTER TABLE ONLY event_backward_extremities - ADD CONSTRAINT event_backward_extremities_event_id_room_id_key UNIQUE (event_id, room_id); - - - -ALTER TABLE ONLY event_edges - ADD CONSTRAINT event_edges_event_id_prev_event_id_room_id_is_state_key UNIQUE (event_id, prev_event_id, room_id, is_state); - - - -ALTER TABLE ONLY event_forward_extremities - ADD CONSTRAINT event_forward_extremities_event_id_room_id_key UNIQUE (event_id, room_id); - - - -ALTER TABLE ONLY event_push_actions - ADD CONSTRAINT event_id_user_id_profile_tag_uniqueness UNIQUE (room_id, event_id, user_id, profile_tag); - - - -ALTER TABLE ONLY event_json - ADD CONSTRAINT event_json_event_id_key UNIQUE (event_id); - - - -ALTER TABLE ONLY event_push_summary_stream_ordering - ADD CONSTRAINT event_push_summary_stream_ordering_lock_key UNIQUE (lock); - - - -ALTER TABLE ONLY event_reference_hashes - ADD CONSTRAINT event_reference_hashes_event_id_algorithm_key UNIQUE (event_id, algorithm); - - - -ALTER TABLE ONLY event_reports - ADD CONSTRAINT event_reports_pkey PRIMARY KEY (id); - - - -ALTER TABLE ONLY event_to_state_groups - ADD CONSTRAINT event_to_state_groups_event_id_key UNIQUE (event_id); - - - -ALTER TABLE ONLY events - ADD CONSTRAINT events_event_id_key UNIQUE (event_id); - - - -ALTER TABLE ONLY events - ADD CONSTRAINT events_pkey PRIMARY KEY (stream_ordering); - - - -ALTER TABLE ONLY ex_outlier_stream - ADD CONSTRAINT ex_outlier_stream_pkey PRIMARY KEY (event_stream_ordering); - - - -ALTER TABLE ONLY group_roles - ADD CONSTRAINT group_roles_group_id_role_id_key UNIQUE (group_id, role_id); - - - -ALTER TABLE ONLY group_room_categories - ADD CONSTRAINT group_room_categories_group_id_category_id_key UNIQUE (group_id, category_id); - - - -ALTER TABLE ONLY group_summary_roles - ADD CONSTRAINT group_summary_roles_group_id_role_id_role_order_key UNIQUE (group_id, role_id, role_order); - - - -ALTER TABLE ONLY group_summary_room_categories - ADD CONSTRAINT group_summary_room_categories_group_id_category_id_cat_orde_key UNIQUE (group_id, category_id, cat_order); - - - -ALTER TABLE ONLY group_summary_rooms - ADD CONSTRAINT group_summary_rooms_group_id_category_id_room_id_room_order_key UNIQUE (group_id, category_id, room_id, room_order); - - - -ALTER TABLE ONLY guest_access - ADD CONSTRAINT guest_access_event_id_key UNIQUE (event_id); - - - -ALTER TABLE ONLY history_visibility - ADD CONSTRAINT history_visibility_event_id_key UNIQUE (event_id); - - - -ALTER TABLE ONLY local_media_repository - ADD CONSTRAINT local_media_repository_media_id_key UNIQUE (media_id); - - - -ALTER TABLE ONLY local_media_repository_thumbnails - ADD CONSTRAINT local_media_repository_thumbn_media_id_thumbnail_width_thum_key UNIQUE (media_id, thumbnail_width, thumbnail_height, thumbnail_type); - - - -ALTER TABLE ONLY user_threepids - ADD CONSTRAINT medium_address UNIQUE (medium, address); - - - -ALTER TABLE ONLY open_id_tokens - ADD CONSTRAINT open_id_tokens_pkey PRIMARY KEY (token); - - - -ALTER TABLE ONLY presence_allow_inbound - ADD CONSTRAINT presence_allow_inbound_observed_user_id_observer_user_id_key UNIQUE (observed_user_id, observer_user_id); - - - -ALTER TABLE ONLY presence - ADD CONSTRAINT presence_user_id_key UNIQUE (user_id); - - - -ALTER TABLE ONLY account_data_max_stream_id - ADD CONSTRAINT private_user_data_max_stream_id_lock_key UNIQUE (lock); - - - -ALTER TABLE ONLY profiles - ADD CONSTRAINT profiles_user_id_key UNIQUE (user_id); - - - -ALTER TABLE ONLY push_rules_enable - ADD CONSTRAINT push_rules_enable_pkey PRIMARY KEY (id); - - - -ALTER TABLE ONLY push_rules_enable - ADD CONSTRAINT push_rules_enable_user_name_rule_id_key UNIQUE (user_name, rule_id); - - - -ALTER TABLE ONLY push_rules - ADD CONSTRAINT push_rules_pkey PRIMARY KEY (id); - - - -ALTER TABLE ONLY push_rules - ADD CONSTRAINT push_rules_user_name_rule_id_key UNIQUE (user_name, rule_id); - - - -ALTER TABLE ONLY pusher_throttle - ADD CONSTRAINT pusher_throttle_pkey PRIMARY KEY (pusher, room_id); - - - -ALTER TABLE ONLY pushers - ADD CONSTRAINT pushers2_app_id_pushkey_user_name_key UNIQUE (app_id, pushkey, user_name); - - - -ALTER TABLE ONLY pushers - ADD CONSTRAINT pushers2_pkey PRIMARY KEY (id); - - - -ALTER TABLE ONLY receipts_graph - ADD CONSTRAINT receipts_graph_uniqueness UNIQUE (room_id, receipt_type, user_id); - - - -ALTER TABLE ONLY receipts_linearized - ADD CONSTRAINT receipts_linearized_uniqueness UNIQUE (room_id, receipt_type, user_id); - - - -ALTER TABLE ONLY received_transactions - ADD CONSTRAINT received_transactions_transaction_id_origin_key UNIQUE (transaction_id, origin); - - - -ALTER TABLE ONLY redactions - ADD CONSTRAINT redactions_event_id_key UNIQUE (event_id); - - - -ALTER TABLE ONLY rejections - ADD CONSTRAINT rejections_event_id_key UNIQUE (event_id); - - - -ALTER TABLE ONLY remote_media_cache - ADD CONSTRAINT remote_media_cache_media_origin_media_id_key UNIQUE (media_origin, media_id); - - - -ALTER TABLE ONLY remote_media_cache_thumbnails - ADD CONSTRAINT remote_media_cache_thumbnails_media_origin_media_id_thumbna_key UNIQUE (media_origin, media_id, thumbnail_width, thumbnail_height, thumbnail_type); - - - -ALTER TABLE ONLY room_account_data - ADD CONSTRAINT room_account_data_uniqueness UNIQUE (user_id, room_id, account_data_type); - - - -ALTER TABLE ONLY room_aliases - ADD CONSTRAINT room_aliases_room_alias_key UNIQUE (room_alias); - - - -ALTER TABLE ONLY room_depth - ADD CONSTRAINT room_depth_room_id_key UNIQUE (room_id); - - - -ALTER TABLE ONLY room_memberships - ADD CONSTRAINT room_memberships_event_id_key UNIQUE (event_id); - - - -ALTER TABLE ONLY room_names - ADD CONSTRAINT room_names_event_id_key UNIQUE (event_id); - - - -ALTER TABLE ONLY room_tags_revisions - ADD CONSTRAINT room_tag_revisions_uniqueness UNIQUE (user_id, room_id); - - - -ALTER TABLE ONLY room_tags - ADD CONSTRAINT room_tag_uniqueness UNIQUE (user_id, room_id, tag); - - - -ALTER TABLE ONLY rooms - ADD CONSTRAINT rooms_pkey PRIMARY KEY (room_id); - - - -ALTER TABLE ONLY server_keys_json - ADD CONSTRAINT server_keys_json_uniqueness UNIQUE (server_name, key_id, from_server); - - - -ALTER TABLE ONLY server_signature_keys - ADD CONSTRAINT server_signature_keys_server_name_key_id_key UNIQUE (server_name, key_id); - - - -ALTER TABLE ONLY state_events - ADD CONSTRAINT state_events_event_id_key UNIQUE (event_id); - - - -ALTER TABLE ONLY state_groups - ADD CONSTRAINT state_groups_pkey PRIMARY KEY (id); - - - -ALTER TABLE ONLY stats_stream_pos - ADD CONSTRAINT stats_stream_pos_lock_key UNIQUE (lock); - - - -ALTER TABLE ONLY topics - ADD CONSTRAINT topics_event_id_key UNIQUE (event_id); - - - -ALTER TABLE ONLY user_directory_stream_pos - ADD CONSTRAINT user_directory_stream_pos_lock_key UNIQUE (lock); - - - -ALTER TABLE ONLY users - ADD CONSTRAINT users_name_key UNIQUE (name); - - - -CREATE INDEX access_tokens_device_id ON access_tokens USING btree (user_id, device_id); - - - -CREATE INDEX account_data_stream_id ON account_data USING btree (user_id, stream_id); - - - -CREATE INDEX application_services_txns_id ON application_services_txns USING btree (as_id); - - - -CREATE UNIQUE INDEX appservice_room_list_idx ON appservice_room_list USING btree (appservice_id, network_id, room_id); - - - -CREATE UNIQUE INDEX blocked_rooms_idx ON blocked_rooms USING btree (room_id); - - - -CREATE INDEX cache_invalidation_stream_id ON cache_invalidation_stream USING btree (stream_id); - - - -CREATE INDEX current_state_delta_stream_idx ON current_state_delta_stream USING btree (stream_id); - - - -CREATE INDEX current_state_events_member_index ON current_state_events USING btree (state_key) WHERE (type = 'm.room.member'::text); - - - -CREATE INDEX deleted_pushers_stream_id ON deleted_pushers USING btree (stream_id); - - - -CREATE INDEX device_federation_inbox_sender_id ON device_federation_inbox USING btree (origin, message_id); - - - -CREATE INDEX device_federation_outbox_destination_id ON device_federation_outbox USING btree (destination, stream_id); - - - -CREATE INDEX device_federation_outbox_id ON device_federation_outbox USING btree (stream_id); - - - -CREATE INDEX device_inbox_stream_id_user_id ON device_inbox USING btree (stream_id, user_id); - - - -CREATE INDEX device_inbox_user_stream_id ON device_inbox USING btree (user_id, device_id, stream_id); - - - -CREATE INDEX device_lists_outbound_last_success_idx ON device_lists_outbound_last_success USING btree (destination, user_id, stream_id); - - - -CREATE INDEX device_lists_outbound_pokes_id ON device_lists_outbound_pokes USING btree (destination, stream_id); - - - -CREATE INDEX device_lists_outbound_pokes_stream ON device_lists_outbound_pokes USING btree (stream_id); - - - -CREATE INDEX device_lists_outbound_pokes_user ON device_lists_outbound_pokes USING btree (destination, user_id); - - - -CREATE UNIQUE INDEX device_lists_remote_cache_unique_id ON device_lists_remote_cache USING btree (user_id, device_id); - - - -CREATE UNIQUE INDEX device_lists_remote_extremeties_unique_idx ON device_lists_remote_extremeties USING btree (user_id); - - - -CREATE INDEX device_lists_stream_id ON device_lists_stream USING btree (stream_id, user_id); - - - -CREATE INDEX device_lists_stream_user_id ON device_lists_stream USING btree (user_id, device_id); - - - -CREATE UNIQUE INDEX e2e_room_keys_idx ON e2e_room_keys USING btree (user_id, room_id, session_id); - - - -CREATE UNIQUE INDEX e2e_room_keys_versions_idx ON e2e_room_keys_versions USING btree (user_id, version); - - - -CREATE UNIQUE INDEX erased_users_user ON erased_users USING btree (user_id); - - - -CREATE INDEX ev_b_extrem_id ON event_backward_extremities USING btree (event_id); - - - -CREATE INDEX ev_b_extrem_room ON event_backward_extremities USING btree (room_id); - - - -CREATE INDEX ev_edges_id ON event_edges USING btree (event_id); - - - -CREATE INDEX ev_edges_prev_id ON event_edges USING btree (prev_event_id); - - - -CREATE INDEX ev_extrem_id ON event_forward_extremities USING btree (event_id); - - - -CREATE INDEX ev_extrem_room ON event_forward_extremities USING btree (room_id); - - - -CREATE INDEX evauth_edges_id ON event_auth USING btree (event_id); - - - -CREATE INDEX event_contains_url_index ON events USING btree (room_id, topological_ordering, stream_ordering) WHERE ((contains_url = true) AND (outlier = false)); - - - -CREATE INDEX event_json_room_id ON event_json USING btree (room_id); - - - -CREATE INDEX event_push_actions_highlights_index ON event_push_actions USING btree (user_id, room_id, topological_ordering, stream_ordering) WHERE (highlight = 1); - - - -CREATE INDEX event_push_actions_rm_tokens ON event_push_actions USING btree (user_id, room_id, topological_ordering, stream_ordering); - - - -CREATE INDEX event_push_actions_room_id_user_id ON event_push_actions USING btree (room_id, user_id); - - - -CREATE INDEX event_push_actions_staging_id ON event_push_actions_staging USING btree (event_id); - - - -CREATE INDEX event_push_actions_stream_ordering ON event_push_actions USING btree (stream_ordering, user_id); - - - -CREATE INDEX event_push_actions_u_highlight ON event_push_actions USING btree (user_id, stream_ordering); - - - -CREATE INDEX event_push_summary_user_rm ON event_push_summary USING btree (user_id, room_id); - - - -CREATE INDEX event_reference_hashes_id ON event_reference_hashes USING btree (event_id); - - - -CREATE UNIQUE INDEX event_relations_id ON event_relations USING btree (event_id); - - - -CREATE INDEX event_relations_relates ON event_relations USING btree (relates_to_id, relation_type, aggregation_key); - - - -CREATE INDEX event_search_ev_ridx ON event_search USING btree (room_id); - - - -CREATE UNIQUE INDEX event_search_event_id_idx ON event_search USING btree (event_id); - - - -CREATE INDEX event_search_fts_idx ON event_search USING gin (vector); - - - -CREATE INDEX event_to_state_groups_sg_index ON event_to_state_groups USING btree (state_group); - - - -CREATE INDEX events_order_room ON events USING btree (room_id, topological_ordering, stream_ordering); - - - -CREATE INDEX events_room_stream ON events USING btree (room_id, stream_ordering); - - - -CREATE INDEX events_ts ON events USING btree (origin_server_ts, stream_ordering); - - - -CREATE INDEX group_attestations_remote_g_idx ON group_attestations_remote USING btree (group_id, user_id); - - - -CREATE INDEX group_attestations_remote_u_idx ON group_attestations_remote USING btree (user_id); - - - -CREATE INDEX group_attestations_remote_v_idx ON group_attestations_remote USING btree (valid_until_ms); - - - -CREATE INDEX group_attestations_renewals_g_idx ON group_attestations_renewals USING btree (group_id, user_id); - - - -CREATE INDEX group_attestations_renewals_u_idx ON group_attestations_renewals USING btree (user_id); - - - -CREATE INDEX group_attestations_renewals_v_idx ON group_attestations_renewals USING btree (valid_until_ms); - - - -CREATE UNIQUE INDEX group_invites_g_idx ON group_invites USING btree (group_id, user_id); - - - -CREATE INDEX group_invites_u_idx ON group_invites USING btree (user_id); - - - -CREATE UNIQUE INDEX group_rooms_g_idx ON group_rooms USING btree (group_id, room_id); - - - -CREATE INDEX group_rooms_r_idx ON group_rooms USING btree (room_id); - - - -CREATE UNIQUE INDEX group_summary_rooms_g_idx ON group_summary_rooms USING btree (group_id, room_id, category_id); - - - -CREATE INDEX group_summary_users_g_idx ON group_summary_users USING btree (group_id); - - - -CREATE UNIQUE INDEX group_users_g_idx ON group_users USING btree (group_id, user_id); - - - -CREATE INDEX group_users_u_idx ON group_users USING btree (user_id); - - - -CREATE UNIQUE INDEX groups_idx ON groups USING btree (group_id); - - - -CREATE INDEX local_group_membership_g_idx ON local_group_membership USING btree (group_id); - - - -CREATE INDEX local_group_membership_u_idx ON local_group_membership USING btree (user_id, group_id); - - - -CREATE INDEX local_invites_for_user_idx ON local_invites USING btree (invitee, locally_rejected, replaced_by, room_id); - - - -CREATE INDEX local_invites_id ON local_invites USING btree (stream_id); - - - -CREATE INDEX local_media_repository_thumbnails_media_id ON local_media_repository_thumbnails USING btree (media_id); - - - -CREATE INDEX local_media_repository_url_cache_by_url_download_ts ON local_media_repository_url_cache USING btree (url, download_ts); - - - -CREATE INDEX local_media_repository_url_cache_expires_idx ON local_media_repository_url_cache USING btree (expires_ts); - - - -CREATE INDEX local_media_repository_url_cache_media_idx ON local_media_repository_url_cache USING btree (media_id); - - - -CREATE INDEX local_media_repository_url_idx ON local_media_repository USING btree (created_ts) WHERE (url_cache IS NOT NULL); - - - -CREATE INDEX monthly_active_users_time_stamp ON monthly_active_users USING btree ("timestamp"); - - - -CREATE UNIQUE INDEX monthly_active_users_users ON monthly_active_users USING btree (user_id); - - - -CREATE INDEX open_id_tokens_ts_valid_until_ms ON open_id_tokens USING btree (ts_valid_until_ms); - - - -CREATE INDEX presence_stream_id ON presence_stream USING btree (stream_id, user_id); - - - -CREATE INDEX presence_stream_user_id ON presence_stream USING btree (user_id); - - - -CREATE INDEX public_room_index ON rooms USING btree (is_public); - - - -CREATE INDEX public_room_list_stream_idx ON public_room_list_stream USING btree (stream_id); - - - -CREATE INDEX public_room_list_stream_rm_idx ON public_room_list_stream USING btree (room_id, stream_id); - - - -CREATE INDEX push_rules_enable_user_name ON push_rules_enable USING btree (user_name); - - - -CREATE INDEX push_rules_stream_id ON push_rules_stream USING btree (stream_id); - - - -CREATE INDEX push_rules_stream_user_stream_id ON push_rules_stream USING btree (user_id, stream_id); - - - -CREATE INDEX push_rules_user_name ON push_rules USING btree (user_name); - - - -CREATE UNIQUE INDEX ratelimit_override_idx ON ratelimit_override USING btree (user_id); - - - -CREATE INDEX receipts_linearized_id ON receipts_linearized USING btree (stream_id); - - - -CREATE INDEX receipts_linearized_room_stream ON receipts_linearized USING btree (room_id, stream_id); - - - -CREATE INDEX receipts_linearized_user ON receipts_linearized USING btree (user_id); - - - -CREATE INDEX received_transactions_ts ON received_transactions USING btree (ts); - - - -CREATE INDEX redactions_redacts ON redactions USING btree (redacts); - - - -CREATE INDEX remote_profile_cache_time ON remote_profile_cache USING btree (last_check); - - - -CREATE UNIQUE INDEX remote_profile_cache_user_id ON remote_profile_cache USING btree (user_id); - - - -CREATE INDEX room_account_data_stream_id ON room_account_data USING btree (user_id, stream_id); - - - -CREATE INDEX room_alias_servers_alias ON room_alias_servers USING btree (room_alias); - - - -CREATE INDEX room_aliases_id ON room_aliases USING btree (room_id); - - - -CREATE INDEX room_depth_room ON room_depth USING btree (room_id); - - - -CREATE INDEX room_memberships_room_id ON room_memberships USING btree (room_id); - - - -CREATE INDEX room_memberships_user_id ON room_memberships USING btree (user_id); - - - -CREATE INDEX room_names_room_id ON room_names USING btree (room_id); - - - -CREATE UNIQUE INDEX room_state_room ON room_state USING btree (room_id); - - - -CREATE UNIQUE INDEX room_stats_earliest_token_idx ON room_stats_earliest_token USING btree (room_id); - - - -CREATE UNIQUE INDEX room_stats_room_ts ON room_stats USING btree (room_id, ts); - - - -CREATE INDEX state_group_edges_idx ON state_group_edges USING btree (state_group); - - - -CREATE INDEX state_group_edges_prev_idx ON state_group_edges USING btree (prev_state_group); - - - -CREATE INDEX state_groups_state_type_idx ON state_groups_state USING btree (state_group, type, state_key); - - - -CREATE INDEX stream_ordering_to_exterm_idx ON stream_ordering_to_exterm USING btree (stream_ordering); - - - -CREATE INDEX stream_ordering_to_exterm_rm_idx ON stream_ordering_to_exterm USING btree (room_id, stream_ordering); - - - -CREATE UNIQUE INDEX threepid_guest_access_tokens_index ON threepid_guest_access_tokens USING btree (medium, address); - - - -CREATE INDEX topics_room_id ON topics USING btree (room_id); - - - -CREATE INDEX user_daily_visits_ts_idx ON user_daily_visits USING btree ("timestamp"); - - - -CREATE INDEX user_daily_visits_uts_idx ON user_daily_visits USING btree (user_id, "timestamp"); - - - -CREATE INDEX user_directory_room_idx ON user_directory USING btree (room_id); - - - -CREATE INDEX user_directory_search_fts_idx ON user_directory_search USING gin (vector); - - - -CREATE UNIQUE INDEX user_directory_search_user_idx ON user_directory_search USING btree (user_id); - - - -CREATE UNIQUE INDEX user_directory_user_idx ON user_directory USING btree (user_id); - - - -CREATE INDEX user_filters_by_user_id_filter_id ON user_filters USING btree (user_id, filter_id); - - - -CREATE INDEX user_ips_device_id ON user_ips USING btree (user_id, device_id, last_seen); - - - -CREATE INDEX user_ips_last_seen ON user_ips USING btree (user_id, last_seen); - - - -CREATE INDEX user_ips_last_seen_only ON user_ips USING btree (last_seen); - - - -CREATE UNIQUE INDEX user_ips_user_token_ip_unique_index ON user_ips USING btree (user_id, access_token, ip); - - - -CREATE UNIQUE INDEX user_stats_user_ts ON user_stats USING btree (user_id, ts); - - - -CREATE UNIQUE INDEX user_threepid_id_server_idx ON user_threepid_id_server USING btree (user_id, medium, address, id_server); - - - -CREATE INDEX user_threepids_medium_address ON user_threepids USING btree (medium, address); - - - -CREATE INDEX user_threepids_user_id ON user_threepids USING btree (user_id); - - - -CREATE INDEX users_creation_ts ON users USING btree (creation_ts); - - - -CREATE UNIQUE INDEX users_in_public_rooms_u_idx ON users_in_public_rooms USING btree (user_id, room_id); - - - -CREATE INDEX users_who_share_private_rooms_o_idx ON users_who_share_private_rooms USING btree (other_user_id); - - - -CREATE INDEX users_who_share_private_rooms_r_idx ON users_who_share_private_rooms USING btree (room_id); - - - -CREATE UNIQUE INDEX users_who_share_private_rooms_u_idx ON users_who_share_private_rooms USING btree (user_id, other_user_id, room_id); - - - diff --git a/synapse/storage/schema/full_schemas/54/full.sql.sqlite b/synapse/storage/schema/full_schemas/54/full.sql.sqlite deleted file mode 100644 index be9295e4c9..0000000000 --- a/synapse/storage/schema/full_schemas/54/full.sql.sqlite +++ /dev/null @@ -1,260 +0,0 @@ -CREATE TABLE application_services_state( as_id TEXT PRIMARY KEY, state VARCHAR(5), last_txn INTEGER ); -CREATE TABLE application_services_txns( as_id TEXT NOT NULL, txn_id INTEGER NOT NULL, event_ids TEXT NOT NULL, UNIQUE(as_id, txn_id) ); -CREATE INDEX application_services_txns_id ON application_services_txns ( as_id ); -CREATE TABLE presence( user_id TEXT NOT NULL, state VARCHAR(20), status_msg TEXT, mtime BIGINT, UNIQUE (user_id) ); -CREATE TABLE presence_allow_inbound( observed_user_id TEXT NOT NULL, observer_user_id TEXT NOT NULL, UNIQUE (observed_user_id, observer_user_id) ); -CREATE TABLE users( name TEXT, password_hash TEXT, creation_ts BIGINT, admin SMALLINT DEFAULT 0 NOT NULL, upgrade_ts BIGINT, is_guest SMALLINT DEFAULT 0 NOT NULL, appservice_id TEXT, consent_version TEXT, consent_server_notice_sent TEXT, user_type TEXT DEFAULT NULL, UNIQUE(name) ); -CREATE TABLE access_tokens( id BIGINT PRIMARY KEY, user_id TEXT NOT NULL, device_id TEXT, token TEXT NOT NULL, last_used BIGINT, UNIQUE(token) ); -CREATE TABLE user_ips ( user_id TEXT NOT NULL, access_token TEXT NOT NULL, device_id TEXT, ip TEXT NOT NULL, user_agent TEXT NOT NULL, last_seen BIGINT NOT NULL ); -CREATE TABLE profiles( user_id TEXT NOT NULL, displayname TEXT, avatar_url TEXT, UNIQUE(user_id) ); -CREATE TABLE received_transactions( transaction_id TEXT, origin TEXT, ts BIGINT, response_code INTEGER, response_json bytea, has_been_referenced smallint default 0, UNIQUE (transaction_id, origin) ); -CREATE TABLE destinations( destination TEXT PRIMARY KEY, retry_last_ts BIGINT, retry_interval INTEGER ); -CREATE TABLE events( stream_ordering INTEGER PRIMARY KEY, topological_ordering BIGINT NOT NULL, event_id TEXT NOT NULL, type TEXT NOT NULL, room_id TEXT NOT NULL, content TEXT, unrecognized_keys TEXT, processed BOOL NOT NULL, outlier BOOL NOT NULL, depth BIGINT DEFAULT 0 NOT NULL, origin_server_ts BIGINT, received_ts BIGINT, sender TEXT, contains_url BOOLEAN, UNIQUE (event_id) ); -CREATE INDEX events_order_room ON events ( room_id, topological_ordering, stream_ordering ); -CREATE TABLE event_json( event_id TEXT NOT NULL, room_id TEXT NOT NULL, internal_metadata TEXT NOT NULL, json TEXT NOT NULL, format_version INTEGER, UNIQUE (event_id) ); -CREATE INDEX event_json_room_id ON event_json(room_id); -CREATE TABLE state_events( event_id TEXT NOT NULL, room_id TEXT NOT NULL, type TEXT NOT NULL, state_key TEXT NOT NULL, prev_state TEXT, UNIQUE (event_id) ); -CREATE TABLE current_state_events( event_id TEXT NOT NULL, room_id TEXT NOT NULL, type TEXT NOT NULL, state_key TEXT NOT NULL, UNIQUE (event_id), UNIQUE (room_id, type, state_key) ); -CREATE TABLE room_memberships( event_id TEXT NOT NULL, user_id TEXT NOT NULL, sender TEXT NOT NULL, room_id TEXT NOT NULL, membership TEXT NOT NULL, forgotten INTEGER DEFAULT 0, display_name TEXT, avatar_url TEXT, UNIQUE (event_id) ); -CREATE INDEX room_memberships_room_id ON room_memberships (room_id); -CREATE INDEX room_memberships_user_id ON room_memberships (user_id); -CREATE TABLE topics( event_id TEXT NOT NULL, room_id TEXT NOT NULL, topic TEXT NOT NULL, UNIQUE (event_id) ); -CREATE INDEX topics_room_id ON topics(room_id); -CREATE TABLE room_names( event_id TEXT NOT NULL, room_id TEXT NOT NULL, name TEXT NOT NULL, UNIQUE (event_id) ); -CREATE INDEX room_names_room_id ON room_names(room_id); -CREATE TABLE rooms( room_id TEXT PRIMARY KEY NOT NULL, is_public BOOL, creator TEXT ); -CREATE TABLE server_signature_keys( server_name TEXT, key_id TEXT, from_server TEXT, ts_added_ms BIGINT, verify_key bytea, ts_valid_until_ms BIGINT, UNIQUE (server_name, key_id) ); -CREATE TABLE rejections( event_id TEXT NOT NULL, reason TEXT NOT NULL, last_check TEXT NOT NULL, UNIQUE (event_id) ); -CREATE TABLE push_rules ( id BIGINT PRIMARY KEY, user_name TEXT NOT NULL, rule_id TEXT NOT NULL, priority_class SMALLINT NOT NULL, priority INTEGER NOT NULL DEFAULT 0, conditions TEXT NOT NULL, actions TEXT NOT NULL, UNIQUE(user_name, rule_id) ); -CREATE INDEX push_rules_user_name on push_rules (user_name); -CREATE TABLE user_filters( user_id TEXT, filter_id BIGINT, filter_json bytea ); -CREATE INDEX user_filters_by_user_id_filter_id ON user_filters( user_id, filter_id ); -CREATE TABLE push_rules_enable ( id BIGINT PRIMARY KEY, user_name TEXT NOT NULL, rule_id TEXT NOT NULL, enabled SMALLINT, UNIQUE(user_name, rule_id) ); -CREATE INDEX push_rules_enable_user_name on push_rules_enable (user_name); -CREATE TABLE event_forward_extremities( event_id TEXT NOT NULL, room_id TEXT NOT NULL, UNIQUE (event_id, room_id) ); -CREATE INDEX ev_extrem_room ON event_forward_extremities(room_id); -CREATE INDEX ev_extrem_id ON event_forward_extremities(event_id); -CREATE TABLE event_backward_extremities( event_id TEXT NOT NULL, room_id TEXT NOT NULL, UNIQUE (event_id, room_id) ); -CREATE INDEX ev_b_extrem_room ON event_backward_extremities(room_id); -CREATE INDEX ev_b_extrem_id ON event_backward_extremities(event_id); -CREATE TABLE event_edges( event_id TEXT NOT NULL, prev_event_id TEXT NOT NULL, room_id TEXT NOT NULL, is_state BOOL NOT NULL, UNIQUE (event_id, prev_event_id, room_id, is_state) ); -CREATE INDEX ev_edges_id ON event_edges(event_id); -CREATE INDEX ev_edges_prev_id ON event_edges(prev_event_id); -CREATE TABLE room_depth( room_id TEXT NOT NULL, min_depth INTEGER NOT NULL, UNIQUE (room_id) ); -CREATE INDEX room_depth_room ON room_depth(room_id); -CREATE TABLE state_groups( id BIGINT PRIMARY KEY, room_id TEXT NOT NULL, event_id TEXT NOT NULL ); -CREATE TABLE state_groups_state( state_group BIGINT NOT NULL, room_id TEXT NOT NULL, type TEXT NOT NULL, state_key TEXT NOT NULL, event_id TEXT NOT NULL ); -CREATE TABLE event_to_state_groups( event_id TEXT NOT NULL, state_group BIGINT NOT NULL, UNIQUE (event_id) ); -CREATE TABLE local_media_repository ( media_id TEXT, media_type TEXT, media_length INTEGER, created_ts BIGINT, upload_name TEXT, user_id TEXT, quarantined_by TEXT, url_cache TEXT, last_access_ts BIGINT, UNIQUE (media_id) ); -CREATE TABLE local_media_repository_thumbnails ( media_id TEXT, thumbnail_width INTEGER, thumbnail_height INTEGER, thumbnail_type TEXT, thumbnail_method TEXT, thumbnail_length INTEGER, UNIQUE ( media_id, thumbnail_width, thumbnail_height, thumbnail_type ) ); -CREATE INDEX local_media_repository_thumbnails_media_id ON local_media_repository_thumbnails (media_id); -CREATE TABLE remote_media_cache ( media_origin TEXT, media_id TEXT, media_type TEXT, created_ts BIGINT, upload_name TEXT, media_length INTEGER, filesystem_id TEXT, last_access_ts BIGINT, quarantined_by TEXT, UNIQUE (media_origin, media_id) ); -CREATE TABLE remote_media_cache_thumbnails ( media_origin TEXT, media_id TEXT, thumbnail_width INTEGER, thumbnail_height INTEGER, thumbnail_method TEXT, thumbnail_type TEXT, thumbnail_length INTEGER, filesystem_id TEXT, UNIQUE ( media_origin, media_id, thumbnail_width, thumbnail_height, thumbnail_type ) ); -CREATE TABLE redactions ( event_id TEXT NOT NULL, redacts TEXT NOT NULL, UNIQUE (event_id) ); -CREATE INDEX redactions_redacts ON redactions (redacts); -CREATE TABLE room_aliases( room_alias TEXT NOT NULL, room_id TEXT NOT NULL, creator TEXT, UNIQUE (room_alias) ); -CREATE INDEX room_aliases_id ON room_aliases(room_id); -CREATE TABLE room_alias_servers( room_alias TEXT NOT NULL, server TEXT NOT NULL ); -CREATE INDEX room_alias_servers_alias ON room_alias_servers(room_alias); -CREATE TABLE event_reference_hashes ( event_id TEXT, algorithm TEXT, hash bytea, UNIQUE (event_id, algorithm) ); -CREATE INDEX event_reference_hashes_id ON event_reference_hashes(event_id); -CREATE TABLE IF NOT EXISTS "server_keys_json" ( server_name TEXT NOT NULL, key_id TEXT NOT NULL, from_server TEXT NOT NULL, ts_added_ms BIGINT NOT NULL, ts_valid_until_ms BIGINT NOT NULL, key_json bytea NOT NULL, CONSTRAINT server_keys_json_uniqueness UNIQUE (server_name, key_id, from_server) ); -CREATE TABLE e2e_device_keys_json ( user_id TEXT NOT NULL, device_id TEXT NOT NULL, ts_added_ms BIGINT NOT NULL, key_json TEXT NOT NULL, CONSTRAINT e2e_device_keys_json_uniqueness UNIQUE (user_id, device_id) ); -CREATE TABLE e2e_one_time_keys_json ( user_id TEXT NOT NULL, device_id TEXT NOT NULL, algorithm TEXT NOT NULL, key_id TEXT NOT NULL, ts_added_ms BIGINT NOT NULL, key_json TEXT NOT NULL, CONSTRAINT e2e_one_time_keys_json_uniqueness UNIQUE (user_id, device_id, algorithm, key_id) ); -CREATE TABLE receipts_graph( room_id TEXT NOT NULL, receipt_type TEXT NOT NULL, user_id TEXT NOT NULL, event_ids TEXT NOT NULL, data TEXT NOT NULL, CONSTRAINT receipts_graph_uniqueness UNIQUE (room_id, receipt_type, user_id) ); -CREATE TABLE receipts_linearized ( stream_id BIGINT NOT NULL, room_id TEXT NOT NULL, receipt_type TEXT NOT NULL, user_id TEXT NOT NULL, event_id TEXT NOT NULL, data TEXT NOT NULL, CONSTRAINT receipts_linearized_uniqueness UNIQUE (room_id, receipt_type, user_id) ); -CREATE INDEX receipts_linearized_id ON receipts_linearized( stream_id ); -CREATE INDEX receipts_linearized_room_stream ON receipts_linearized( room_id, stream_id ); -CREATE TABLE IF NOT EXISTS "user_threepids" ( user_id TEXT NOT NULL, medium TEXT NOT NULL, address TEXT NOT NULL, validated_at BIGINT NOT NULL, added_at BIGINT NOT NULL, CONSTRAINT medium_address UNIQUE (medium, address) ); -CREATE INDEX user_threepids_user_id ON user_threepids(user_id); -CREATE TABLE background_updates( update_name TEXT NOT NULL, progress_json TEXT NOT NULL, depends_on TEXT, CONSTRAINT background_updates_uniqueness UNIQUE (update_name) ); -CREATE VIRTUAL TABLE event_search USING fts4 ( event_id, room_id, sender, key, value ) -/* event_search(event_id,room_id,sender,"key",value) */; -CREATE TABLE IF NOT EXISTS 'event_search_content'(docid INTEGER PRIMARY KEY, 'c0event_id', 'c1room_id', 'c2sender', 'c3key', 'c4value'); -CREATE TABLE IF NOT EXISTS 'event_search_segments'(blockid INTEGER PRIMARY KEY, block BLOB); -CREATE TABLE IF NOT EXISTS 'event_search_segdir'(level INTEGER,idx INTEGER,start_block INTEGER,leaves_end_block INTEGER,end_block INTEGER,root BLOB,PRIMARY KEY(level, idx)); -CREATE TABLE IF NOT EXISTS 'event_search_docsize'(docid INTEGER PRIMARY KEY, size BLOB); -CREATE TABLE IF NOT EXISTS 'event_search_stat'(id INTEGER PRIMARY KEY, value BLOB); -CREATE TABLE guest_access( event_id TEXT NOT NULL, room_id TEXT NOT NULL, guest_access TEXT NOT NULL, UNIQUE (event_id) ); -CREATE TABLE history_visibility( event_id TEXT NOT NULL, room_id TEXT NOT NULL, history_visibility TEXT NOT NULL, UNIQUE (event_id) ); -CREATE TABLE room_tags( user_id TEXT NOT NULL, room_id TEXT NOT NULL, tag TEXT NOT NULL, content TEXT NOT NULL, CONSTRAINT room_tag_uniqueness UNIQUE (user_id, room_id, tag) ); -CREATE TABLE room_tags_revisions ( user_id TEXT NOT NULL, room_id TEXT NOT NULL, stream_id BIGINT NOT NULL, CONSTRAINT room_tag_revisions_uniqueness UNIQUE (user_id, room_id) ); -CREATE TABLE IF NOT EXISTS "account_data_max_stream_id"( Lock CHAR(1) NOT NULL DEFAULT 'X' UNIQUE, stream_id BIGINT NOT NULL, CHECK (Lock='X') ); -CREATE TABLE account_data( user_id TEXT NOT NULL, account_data_type TEXT NOT NULL, stream_id BIGINT NOT NULL, content TEXT NOT NULL, CONSTRAINT account_data_uniqueness UNIQUE (user_id, account_data_type) ); -CREATE TABLE room_account_data( user_id TEXT NOT NULL, room_id TEXT NOT NULL, account_data_type TEXT NOT NULL, stream_id BIGINT NOT NULL, content TEXT NOT NULL, CONSTRAINT room_account_data_uniqueness UNIQUE (user_id, room_id, account_data_type) ); -CREATE INDEX account_data_stream_id on account_data(user_id, stream_id); -CREATE INDEX room_account_data_stream_id on room_account_data(user_id, stream_id); -CREATE INDEX events_ts ON events(origin_server_ts, stream_ordering); -CREATE TABLE event_push_actions( room_id TEXT NOT NULL, event_id TEXT NOT NULL, user_id TEXT NOT NULL, profile_tag VARCHAR(32), actions TEXT NOT NULL, topological_ordering BIGINT, stream_ordering BIGINT, notif SMALLINT, highlight SMALLINT, CONSTRAINT event_id_user_id_profile_tag_uniqueness UNIQUE (room_id, event_id, user_id, profile_tag) ); -CREATE INDEX event_push_actions_room_id_user_id on event_push_actions(room_id, user_id); -CREATE INDEX events_room_stream on events(room_id, stream_ordering); -CREATE INDEX public_room_index on rooms(is_public); -CREATE INDEX receipts_linearized_user ON receipts_linearized( user_id ); -CREATE INDEX event_push_actions_rm_tokens on event_push_actions( user_id, room_id, topological_ordering, stream_ordering ); -CREATE TABLE presence_stream( stream_id BIGINT, user_id TEXT, state TEXT, last_active_ts BIGINT, last_federation_update_ts BIGINT, last_user_sync_ts BIGINT, status_msg TEXT, currently_active BOOLEAN ); -CREATE INDEX presence_stream_id ON presence_stream(stream_id, user_id); -CREATE INDEX presence_stream_user_id ON presence_stream(user_id); -CREATE TABLE push_rules_stream( stream_id BIGINT NOT NULL, event_stream_ordering BIGINT NOT NULL, user_id TEXT NOT NULL, rule_id TEXT NOT NULL, op TEXT NOT NULL, priority_class SMALLINT, priority INTEGER, conditions TEXT, actions TEXT ); -CREATE INDEX push_rules_stream_id ON push_rules_stream(stream_id); -CREATE INDEX push_rules_stream_user_stream_id on push_rules_stream(user_id, stream_id); -CREATE TABLE ex_outlier_stream( event_stream_ordering BIGINT PRIMARY KEY NOT NULL, event_id TEXT NOT NULL, state_group BIGINT NOT NULL ); -CREATE TABLE threepid_guest_access_tokens( medium TEXT, address TEXT, guest_access_token TEXT, first_inviter TEXT ); -CREATE UNIQUE INDEX threepid_guest_access_tokens_index ON threepid_guest_access_tokens(medium, address); -CREATE TABLE local_invites( stream_id BIGINT NOT NULL, inviter TEXT NOT NULL, invitee TEXT NOT NULL, event_id TEXT NOT NULL, room_id TEXT NOT NULL, locally_rejected TEXT, replaced_by TEXT ); -CREATE INDEX local_invites_id ON local_invites(stream_id); -CREATE INDEX local_invites_for_user_idx ON local_invites(invitee, locally_rejected, replaced_by, room_id); -CREATE INDEX event_push_actions_stream_ordering on event_push_actions( stream_ordering, user_id ); -CREATE TABLE open_id_tokens ( token TEXT NOT NULL PRIMARY KEY, ts_valid_until_ms bigint NOT NULL, user_id TEXT NOT NULL, UNIQUE (token) ); -CREATE INDEX open_id_tokens_ts_valid_until_ms ON open_id_tokens(ts_valid_until_ms); -CREATE TABLE pusher_throttle( pusher BIGINT NOT NULL, room_id TEXT NOT NULL, last_sent_ts BIGINT, throttle_ms BIGINT, PRIMARY KEY (pusher, room_id) ); -CREATE TABLE event_reports( id BIGINT NOT NULL PRIMARY KEY, received_ts BIGINT NOT NULL, room_id TEXT NOT NULL, event_id TEXT NOT NULL, user_id TEXT NOT NULL, reason TEXT, content TEXT ); -CREATE TABLE devices ( user_id TEXT NOT NULL, device_id TEXT NOT NULL, display_name TEXT, CONSTRAINT device_uniqueness UNIQUE (user_id, device_id) ); -CREATE TABLE appservice_stream_position( Lock CHAR(1) NOT NULL DEFAULT 'X' UNIQUE, stream_ordering BIGINT, CHECK (Lock='X') ); -CREATE TABLE device_inbox ( user_id TEXT NOT NULL, device_id TEXT NOT NULL, stream_id BIGINT NOT NULL, message_json TEXT NOT NULL ); -CREATE INDEX device_inbox_user_stream_id ON device_inbox(user_id, device_id, stream_id); -CREATE INDEX received_transactions_ts ON received_transactions(ts); -CREATE TABLE device_federation_outbox ( destination TEXT NOT NULL, stream_id BIGINT NOT NULL, queued_ts BIGINT NOT NULL, messages_json TEXT NOT NULL ); -CREATE INDEX device_federation_outbox_destination_id ON device_federation_outbox(destination, stream_id); -CREATE TABLE device_federation_inbox ( origin TEXT NOT NULL, message_id TEXT NOT NULL, received_ts BIGINT NOT NULL ); -CREATE INDEX device_federation_inbox_sender_id ON device_federation_inbox(origin, message_id); -CREATE TABLE device_max_stream_id ( stream_id BIGINT NOT NULL ); -CREATE TABLE public_room_list_stream ( stream_id BIGINT NOT NULL, room_id TEXT NOT NULL, visibility BOOLEAN NOT NULL , appservice_id TEXT, network_id TEXT); -CREATE INDEX public_room_list_stream_idx on public_room_list_stream( stream_id ); -CREATE INDEX public_room_list_stream_rm_idx on public_room_list_stream( room_id, stream_id ); -CREATE TABLE state_group_edges( state_group BIGINT NOT NULL, prev_state_group BIGINT NOT NULL ); -CREATE INDEX state_group_edges_idx ON state_group_edges(state_group); -CREATE INDEX state_group_edges_prev_idx ON state_group_edges(prev_state_group); -CREATE TABLE stream_ordering_to_exterm ( stream_ordering BIGINT NOT NULL, room_id TEXT NOT NULL, event_id TEXT NOT NULL ); -CREATE INDEX stream_ordering_to_exterm_idx on stream_ordering_to_exterm( stream_ordering ); -CREATE INDEX stream_ordering_to_exterm_rm_idx on stream_ordering_to_exterm( room_id, stream_ordering ); -CREATE TABLE IF NOT EXISTS "event_auth"( event_id TEXT NOT NULL, auth_id TEXT NOT NULL, room_id TEXT NOT NULL ); -CREATE INDEX evauth_edges_id ON event_auth(event_id); -CREATE INDEX user_threepids_medium_address on user_threepids (medium, address); -CREATE TABLE appservice_room_list( appservice_id TEXT NOT NULL, network_id TEXT NOT NULL, room_id TEXT NOT NULL ); -CREATE UNIQUE INDEX appservice_room_list_idx ON appservice_room_list( appservice_id, network_id, room_id ); -CREATE INDEX device_federation_outbox_id ON device_federation_outbox(stream_id); -CREATE TABLE federation_stream_position( type TEXT NOT NULL, stream_id INTEGER NOT NULL ); -CREATE TABLE device_lists_remote_cache ( user_id TEXT NOT NULL, device_id TEXT NOT NULL, content TEXT NOT NULL ); -CREATE TABLE device_lists_remote_extremeties ( user_id TEXT NOT NULL, stream_id TEXT NOT NULL ); -CREATE TABLE device_lists_stream ( stream_id BIGINT NOT NULL, user_id TEXT NOT NULL, device_id TEXT NOT NULL ); -CREATE INDEX device_lists_stream_id ON device_lists_stream(stream_id, user_id); -CREATE TABLE device_lists_outbound_pokes ( destination TEXT NOT NULL, stream_id BIGINT NOT NULL, user_id TEXT NOT NULL, device_id TEXT NOT NULL, sent BOOLEAN NOT NULL, ts BIGINT NOT NULL ); -CREATE INDEX device_lists_outbound_pokes_id ON device_lists_outbound_pokes(destination, stream_id); -CREATE INDEX device_lists_outbound_pokes_user ON device_lists_outbound_pokes(destination, user_id); -CREATE TABLE event_push_summary ( user_id TEXT NOT NULL, room_id TEXT NOT NULL, notif_count BIGINT NOT NULL, stream_ordering BIGINT NOT NULL ); -CREATE INDEX event_push_summary_user_rm ON event_push_summary(user_id, room_id); -CREATE TABLE event_push_summary_stream_ordering ( Lock CHAR(1) NOT NULL DEFAULT 'X' UNIQUE, stream_ordering BIGINT NOT NULL, CHECK (Lock='X') ); -CREATE TABLE IF NOT EXISTS "pushers" ( id BIGINT PRIMARY KEY, user_name TEXT NOT NULL, access_token BIGINT DEFAULT NULL, profile_tag TEXT NOT NULL, kind TEXT NOT NULL, app_id TEXT NOT NULL, app_display_name TEXT NOT NULL, device_display_name TEXT NOT NULL, pushkey TEXT NOT NULL, ts BIGINT NOT NULL, lang TEXT, data TEXT, last_stream_ordering INTEGER, last_success BIGINT, failing_since BIGINT, UNIQUE (app_id, pushkey, user_name) ); -CREATE INDEX device_lists_outbound_pokes_stream ON device_lists_outbound_pokes(stream_id); -CREATE TABLE ratelimit_override ( user_id TEXT NOT NULL, messages_per_second BIGINT, burst_count BIGINT ); -CREATE UNIQUE INDEX ratelimit_override_idx ON ratelimit_override(user_id); -CREATE TABLE current_state_delta_stream ( stream_id BIGINT NOT NULL, room_id TEXT NOT NULL, type TEXT NOT NULL, state_key TEXT NOT NULL, event_id TEXT, prev_event_id TEXT ); -CREATE INDEX current_state_delta_stream_idx ON current_state_delta_stream(stream_id); -CREATE TABLE device_lists_outbound_last_success ( destination TEXT NOT NULL, user_id TEXT NOT NULL, stream_id BIGINT NOT NULL ); -CREATE INDEX device_lists_outbound_last_success_idx ON device_lists_outbound_last_success( destination, user_id, stream_id ); -CREATE TABLE user_directory_stream_pos ( Lock CHAR(1) NOT NULL DEFAULT 'X' UNIQUE, stream_id BIGINT, CHECK (Lock='X') ); -CREATE VIRTUAL TABLE user_directory_search USING fts4 ( user_id, value ) -/* user_directory_search(user_id,value) */; -CREATE TABLE IF NOT EXISTS 'user_directory_search_content'(docid INTEGER PRIMARY KEY, 'c0user_id', 'c1value'); -CREATE TABLE IF NOT EXISTS 'user_directory_search_segments'(blockid INTEGER PRIMARY KEY, block BLOB); -CREATE TABLE IF NOT EXISTS 'user_directory_search_segdir'(level INTEGER,idx INTEGER,start_block INTEGER,leaves_end_block INTEGER,end_block INTEGER,root BLOB,PRIMARY KEY(level, idx)); -CREATE TABLE IF NOT EXISTS 'user_directory_search_docsize'(docid INTEGER PRIMARY KEY, size BLOB); -CREATE TABLE IF NOT EXISTS 'user_directory_search_stat'(id INTEGER PRIMARY KEY, value BLOB); -CREATE TABLE blocked_rooms ( room_id TEXT NOT NULL, user_id TEXT NOT NULL ); -CREATE UNIQUE INDEX blocked_rooms_idx ON blocked_rooms(room_id); -CREATE TABLE IF NOT EXISTS "local_media_repository_url_cache"( url TEXT, response_code INTEGER, etag TEXT, expires_ts BIGINT, og TEXT, media_id TEXT, download_ts BIGINT ); -CREATE INDEX local_media_repository_url_cache_expires_idx ON local_media_repository_url_cache(expires_ts); -CREATE INDEX local_media_repository_url_cache_by_url_download_ts ON local_media_repository_url_cache(url, download_ts); -CREATE INDEX local_media_repository_url_cache_media_idx ON local_media_repository_url_cache(media_id); -CREATE TABLE group_users ( group_id TEXT NOT NULL, user_id TEXT NOT NULL, is_admin BOOLEAN NOT NULL, is_public BOOLEAN NOT NULL ); -CREATE TABLE group_invites ( group_id TEXT NOT NULL, user_id TEXT NOT NULL ); -CREATE TABLE group_rooms ( group_id TEXT NOT NULL, room_id TEXT NOT NULL, is_public BOOLEAN NOT NULL ); -CREATE TABLE group_summary_rooms ( group_id TEXT NOT NULL, room_id TEXT NOT NULL, category_id TEXT NOT NULL, room_order BIGINT NOT NULL, is_public BOOLEAN NOT NULL, UNIQUE (group_id, category_id, room_id, room_order), CHECK (room_order > 0) ); -CREATE UNIQUE INDEX group_summary_rooms_g_idx ON group_summary_rooms(group_id, room_id, category_id); -CREATE TABLE group_summary_room_categories ( group_id TEXT NOT NULL, category_id TEXT NOT NULL, cat_order BIGINT NOT NULL, UNIQUE (group_id, category_id, cat_order), CHECK (cat_order > 0) ); -CREATE TABLE group_room_categories ( group_id TEXT NOT NULL, category_id TEXT NOT NULL, profile TEXT NOT NULL, is_public BOOLEAN NOT NULL, UNIQUE (group_id, category_id) ); -CREATE TABLE group_summary_users ( group_id TEXT NOT NULL, user_id TEXT NOT NULL, role_id TEXT NOT NULL, user_order BIGINT NOT NULL, is_public BOOLEAN NOT NULL ); -CREATE INDEX group_summary_users_g_idx ON group_summary_users(group_id); -CREATE TABLE group_summary_roles ( group_id TEXT NOT NULL, role_id TEXT NOT NULL, role_order BIGINT NOT NULL, UNIQUE (group_id, role_id, role_order), CHECK (role_order > 0) ); -CREATE TABLE group_roles ( group_id TEXT NOT NULL, role_id TEXT NOT NULL, profile TEXT NOT NULL, is_public BOOLEAN NOT NULL, UNIQUE (group_id, role_id) ); -CREATE TABLE group_attestations_renewals ( group_id TEXT NOT NULL, user_id TEXT NOT NULL, valid_until_ms BIGINT NOT NULL ); -CREATE INDEX group_attestations_renewals_g_idx ON group_attestations_renewals(group_id, user_id); -CREATE INDEX group_attestations_renewals_u_idx ON group_attestations_renewals(user_id); -CREATE INDEX group_attestations_renewals_v_idx ON group_attestations_renewals(valid_until_ms); -CREATE TABLE group_attestations_remote ( group_id TEXT NOT NULL, user_id TEXT NOT NULL, valid_until_ms BIGINT NOT NULL, attestation_json TEXT NOT NULL ); -CREATE INDEX group_attestations_remote_g_idx ON group_attestations_remote(group_id, user_id); -CREATE INDEX group_attestations_remote_u_idx ON group_attestations_remote(user_id); -CREATE INDEX group_attestations_remote_v_idx ON group_attestations_remote(valid_until_ms); -CREATE TABLE local_group_membership ( group_id TEXT NOT NULL, user_id TEXT NOT NULL, is_admin BOOLEAN NOT NULL, membership TEXT NOT NULL, is_publicised BOOLEAN NOT NULL, content TEXT NOT NULL ); -CREATE INDEX local_group_membership_u_idx ON local_group_membership(user_id, group_id); -CREATE INDEX local_group_membership_g_idx ON local_group_membership(group_id); -CREATE TABLE local_group_updates ( stream_id BIGINT NOT NULL, group_id TEXT NOT NULL, user_id TEXT NOT NULL, type TEXT NOT NULL, content TEXT NOT NULL ); -CREATE TABLE remote_profile_cache ( user_id TEXT NOT NULL, displayname TEXT, avatar_url TEXT, last_check BIGINT NOT NULL ); -CREATE UNIQUE INDEX remote_profile_cache_user_id ON remote_profile_cache(user_id); -CREATE INDEX remote_profile_cache_time ON remote_profile_cache(last_check); -CREATE TABLE IF NOT EXISTS "deleted_pushers" ( stream_id BIGINT NOT NULL, app_id TEXT NOT NULL, pushkey TEXT NOT NULL, user_id TEXT NOT NULL ); -CREATE INDEX deleted_pushers_stream_id ON deleted_pushers (stream_id); -CREATE TABLE IF NOT EXISTS "groups" ( group_id TEXT NOT NULL, name TEXT, avatar_url TEXT, short_description TEXT, long_description TEXT, is_public BOOL NOT NULL , join_policy TEXT NOT NULL DEFAULT 'invite'); -CREATE UNIQUE INDEX groups_idx ON groups(group_id); -CREATE TABLE IF NOT EXISTS "user_directory" ( user_id TEXT NOT NULL, room_id TEXT, display_name TEXT, avatar_url TEXT ); -CREATE INDEX user_directory_room_idx ON user_directory(room_id); -CREATE UNIQUE INDEX user_directory_user_idx ON user_directory(user_id); -CREATE TABLE event_push_actions_staging ( event_id TEXT NOT NULL, user_id TEXT NOT NULL, actions TEXT NOT NULL, notif SMALLINT NOT NULL, highlight SMALLINT NOT NULL ); -CREATE INDEX event_push_actions_staging_id ON event_push_actions_staging(event_id); -CREATE TABLE users_pending_deactivation ( user_id TEXT NOT NULL ); -CREATE UNIQUE INDEX group_invites_g_idx ON group_invites(group_id, user_id); -CREATE UNIQUE INDEX group_users_g_idx ON group_users(group_id, user_id); -CREATE INDEX group_users_u_idx ON group_users(user_id); -CREATE INDEX group_invites_u_idx ON group_invites(user_id); -CREATE UNIQUE INDEX group_rooms_g_idx ON group_rooms(group_id, room_id); -CREATE INDEX group_rooms_r_idx ON group_rooms(room_id); -CREATE TABLE user_daily_visits ( user_id TEXT NOT NULL, device_id TEXT, timestamp BIGINT NOT NULL ); -CREATE INDEX user_daily_visits_uts_idx ON user_daily_visits(user_id, timestamp); -CREATE INDEX user_daily_visits_ts_idx ON user_daily_visits(timestamp); -CREATE TABLE erased_users ( user_id TEXT NOT NULL ); -CREATE UNIQUE INDEX erased_users_user ON erased_users(user_id); -CREATE TABLE monthly_active_users ( user_id TEXT NOT NULL, timestamp BIGINT NOT NULL ); -CREATE UNIQUE INDEX monthly_active_users_users ON monthly_active_users(user_id); -CREATE INDEX monthly_active_users_time_stamp ON monthly_active_users(timestamp); -CREATE TABLE IF NOT EXISTS "e2e_room_keys_versions" ( user_id TEXT NOT NULL, version BIGINT NOT NULL, algorithm TEXT NOT NULL, auth_data TEXT NOT NULL, deleted SMALLINT DEFAULT 0 NOT NULL ); -CREATE UNIQUE INDEX e2e_room_keys_versions_idx ON e2e_room_keys_versions(user_id, version); -CREATE TABLE IF NOT EXISTS "e2e_room_keys" ( user_id TEXT NOT NULL, room_id TEXT NOT NULL, session_id TEXT NOT NULL, version BIGINT NOT NULL, first_message_index INT, forwarded_count INT, is_verified BOOLEAN, session_data TEXT NOT NULL ); -CREATE UNIQUE INDEX e2e_room_keys_idx ON e2e_room_keys(user_id, room_id, session_id); -CREATE TABLE users_who_share_private_rooms ( user_id TEXT NOT NULL, other_user_id TEXT NOT NULL, room_id TEXT NOT NULL ); -CREATE UNIQUE INDEX users_who_share_private_rooms_u_idx ON users_who_share_private_rooms(user_id, other_user_id, room_id); -CREATE INDEX users_who_share_private_rooms_r_idx ON users_who_share_private_rooms(room_id); -CREATE INDEX users_who_share_private_rooms_o_idx ON users_who_share_private_rooms(other_user_id); -CREATE TABLE user_threepid_id_server ( user_id TEXT NOT NULL, medium TEXT NOT NULL, address TEXT NOT NULL, id_server TEXT NOT NULL ); -CREATE UNIQUE INDEX user_threepid_id_server_idx ON user_threepid_id_server( user_id, medium, address, id_server ); -CREATE TABLE users_in_public_rooms ( user_id TEXT NOT NULL, room_id TEXT NOT NULL ); -CREATE UNIQUE INDEX users_in_public_rooms_u_idx ON users_in_public_rooms(user_id, room_id); -CREATE TABLE account_validity ( user_id TEXT PRIMARY KEY, expiration_ts_ms BIGINT NOT NULL, email_sent BOOLEAN NOT NULL, renewal_token TEXT ); -CREATE TABLE event_relations ( event_id TEXT NOT NULL, relates_to_id TEXT NOT NULL, relation_type TEXT NOT NULL, aggregation_key TEXT ); -CREATE UNIQUE INDEX event_relations_id ON event_relations(event_id); -CREATE INDEX event_relations_relates ON event_relations(relates_to_id, relation_type, aggregation_key); -CREATE TABLE stats_stream_pos ( Lock CHAR(1) NOT NULL DEFAULT 'X' UNIQUE, stream_id BIGINT, CHECK (Lock='X') ); -CREATE TABLE user_stats ( user_id TEXT NOT NULL, ts BIGINT NOT NULL, bucket_size INT NOT NULL, public_rooms INT NOT NULL, private_rooms INT NOT NULL ); -CREATE UNIQUE INDEX user_stats_user_ts ON user_stats(user_id, ts); -CREATE TABLE room_stats ( room_id TEXT NOT NULL, ts BIGINT NOT NULL, bucket_size INT NOT NULL, current_state_events INT NOT NULL, joined_members INT NOT NULL, invited_members INT NOT NULL, left_members INT NOT NULL, banned_members INT NOT NULL, state_events INT NOT NULL ); -CREATE UNIQUE INDEX room_stats_room_ts ON room_stats(room_id, ts); -CREATE TABLE room_state ( room_id TEXT NOT NULL, join_rules TEXT, history_visibility TEXT, encryption TEXT, name TEXT, topic TEXT, avatar TEXT, canonical_alias TEXT ); -CREATE UNIQUE INDEX room_state_room ON room_state(room_id); -CREATE TABLE room_stats_earliest_token ( room_id TEXT NOT NULL, token BIGINT NOT NULL ); -CREATE UNIQUE INDEX room_stats_earliest_token_idx ON room_stats_earliest_token(room_id); -CREATE INDEX access_tokens_device_id ON access_tokens (user_id, device_id); -CREATE INDEX user_ips_device_id ON user_ips (user_id, device_id, last_seen); -CREATE INDEX event_contains_url_index ON events (room_id, topological_ordering, stream_ordering); -CREATE INDEX event_push_actions_u_highlight ON event_push_actions (user_id, stream_ordering); -CREATE INDEX event_push_actions_highlights_index ON event_push_actions (user_id, room_id, topological_ordering, stream_ordering); -CREATE INDEX current_state_events_member_index ON current_state_events (state_key); -CREATE INDEX device_inbox_stream_id_user_id ON device_inbox (stream_id, user_id); -CREATE INDEX device_lists_stream_user_id ON device_lists_stream (user_id, device_id); -CREATE INDEX local_media_repository_url_idx ON local_media_repository (created_ts); -CREATE INDEX user_ips_last_seen ON user_ips (user_id, last_seen); -CREATE INDEX user_ips_last_seen_only ON user_ips (last_seen); -CREATE INDEX users_creation_ts ON users (creation_ts); -CREATE INDEX event_to_state_groups_sg_index ON event_to_state_groups (state_group); -CREATE UNIQUE INDEX device_lists_remote_cache_unique_id ON device_lists_remote_cache (user_id, device_id); -CREATE INDEX state_groups_state_type_idx ON state_groups_state(state_group, type, state_key); -CREATE UNIQUE INDEX device_lists_remote_extremeties_unique_idx ON device_lists_remote_extremeties (user_id); -CREATE UNIQUE INDEX user_ips_user_token_ip_unique_index ON user_ips (user_id, access_token, ip); diff --git a/synapse/storage/schema/full_schemas/54/stream_positions.sql b/synapse/storage/schema/full_schemas/54/stream_positions.sql deleted file mode 100644 index c265fd20e2..0000000000 --- a/synapse/storage/schema/full_schemas/54/stream_positions.sql +++ /dev/null @@ -1,7 +0,0 @@ - -INSERT INTO appservice_stream_position (stream_ordering) SELECT COALESCE(MAX(stream_ordering), 0) FROM events; -INSERT INTO federation_stream_position (type, stream_id) VALUES ('federation', -1); -INSERT INTO federation_stream_position (type, stream_id) SELECT 'events', coalesce(max(stream_ordering), -1) FROM events; -INSERT INTO user_directory_stream_pos (stream_id) VALUES (0); -INSERT INTO stats_stream_pos (stream_id) VALUES (0); -INSERT INTO event_push_summary_stream_ordering (stream_ordering) VALUES (0); diff --git a/synapse/storage/schema/full_schemas/README.txt b/synapse/storage/schema/full_schemas/README.txt deleted file mode 100644 index d3f6401344..0000000000 --- a/synapse/storage/schema/full_schemas/README.txt +++ /dev/null @@ -1,19 +0,0 @@ -Building full schema dumps -========================== - -These schemas need to be made from a database that has had all background updates run. - -Postgres --------- - -$ pg_dump --format=plain --schema-only --no-tablespaces --no-acl --no-owner $DATABASE_NAME| sed -e '/^--/d' -e 's/public\.//g' -e '/^SET /d' -e '/^SELECT /d' > full.sql.postgres - -SQLite ------- - -$ sqlite3 $DATABASE_FILE ".schema" > full.sql.sqlite - -After ------ - -Delete the CREATE statements for "sqlite_stat1", "schema_version", "applied_schema_deltas", and "applied_module_schemas". \ No newline at end of file diff --git a/synapse/storage/search.py b/synapse/storage/search.py deleted file mode 100644 index 7695bf09fc..0000000000 --- a/synapse/storage/search.py +++ /dev/null @@ -1,713 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2015, 2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import logging -import re -from collections import namedtuple - -from six import string_types - -from canonicaljson import json - -from twisted.internet import defer - -from synapse.api.errors import SynapseError -from synapse.storage._base import make_in_list_sql_clause -from synapse.storage.engines import PostgresEngine, Sqlite3Engine - -from .background_updates import BackgroundUpdateStore - -logger = logging.getLogger(__name__) - -SearchEntry = namedtuple( - "SearchEntry", - ["key", "value", "event_id", "room_id", "stream_ordering", "origin_server_ts"], -) - - -class SearchBackgroundUpdateStore(BackgroundUpdateStore): - - EVENT_SEARCH_UPDATE_NAME = "event_search" - EVENT_SEARCH_ORDER_UPDATE_NAME = "event_search_order" - EVENT_SEARCH_USE_GIST_POSTGRES_NAME = "event_search_postgres_gist" - EVENT_SEARCH_USE_GIN_POSTGRES_NAME = "event_search_postgres_gin" - - def __init__(self, db_conn, hs): - super(SearchBackgroundUpdateStore, self).__init__(db_conn, hs) - - if not hs.config.enable_search: - return - - self.register_background_update_handler( - self.EVENT_SEARCH_UPDATE_NAME, self._background_reindex_search - ) - self.register_background_update_handler( - self.EVENT_SEARCH_ORDER_UPDATE_NAME, self._background_reindex_search_order - ) - - # we used to have a background update to turn the GIN index into a - # GIST one; we no longer do that (obviously) because we actually want - # a GIN index. However, it's possible that some people might still have - # the background update queued, so we register a handler to clear the - # background update. - self.register_noop_background_update(self.EVENT_SEARCH_USE_GIST_POSTGRES_NAME) - - self.register_background_update_handler( - self.EVENT_SEARCH_USE_GIN_POSTGRES_NAME, self._background_reindex_gin_search - ) - - @defer.inlineCallbacks - def _background_reindex_search(self, progress, batch_size): - # we work through the events table from highest stream id to lowest - target_min_stream_id = progress["target_min_stream_id_inclusive"] - max_stream_id = progress["max_stream_id_exclusive"] - rows_inserted = progress.get("rows_inserted", 0) - - TYPES = ["m.room.name", "m.room.message", "m.room.topic"] - - def reindex_search_txn(txn): - sql = ( - "SELECT stream_ordering, event_id, room_id, type, json, " - " origin_server_ts FROM events" - " JOIN event_json USING (room_id, event_id)" - " WHERE ? <= stream_ordering AND stream_ordering < ?" - " AND (%s)" - " ORDER BY stream_ordering DESC" - " LIMIT ?" - ) % (" OR ".join("type = '%s'" % (t,) for t in TYPES),) - - txn.execute(sql, (target_min_stream_id, max_stream_id, batch_size)) - - # we could stream straight from the results into - # store_search_entries_txn with a generator function, but that - # would mean having two cursors open on the database at once. - # Instead we just build a list of results. - rows = self.cursor_to_dict(txn) - if not rows: - return 0 - - min_stream_id = rows[-1]["stream_ordering"] - - event_search_rows = [] - for row in rows: - try: - event_id = row["event_id"] - room_id = row["room_id"] - etype = row["type"] - stream_ordering = row["stream_ordering"] - origin_server_ts = row["origin_server_ts"] - try: - event_json = json.loads(row["json"]) - content = event_json["content"] - except Exception: - continue - - if etype == "m.room.message": - key = "content.body" - value = content["body"] - elif etype == "m.room.topic": - key = "content.topic" - value = content["topic"] - elif etype == "m.room.name": - key = "content.name" - value = content["name"] - else: - raise Exception("unexpected event type %s" % etype) - except (KeyError, AttributeError): - # If the event is missing a necessary field then - # skip over it. - continue - - if not isinstance(value, string_types): - # If the event body, name or topic isn't a string - # then skip over it - continue - - event_search_rows.append( - SearchEntry( - key=key, - value=value, - event_id=event_id, - room_id=room_id, - stream_ordering=stream_ordering, - origin_server_ts=origin_server_ts, - ) - ) - - self.store_search_entries_txn(txn, event_search_rows) - - progress = { - "target_min_stream_id_inclusive": target_min_stream_id, - "max_stream_id_exclusive": min_stream_id, - "rows_inserted": rows_inserted + len(event_search_rows), - } - - self._background_update_progress_txn( - txn, self.EVENT_SEARCH_UPDATE_NAME, progress - ) - - return len(event_search_rows) - - result = yield self.runInteraction( - self.EVENT_SEARCH_UPDATE_NAME, reindex_search_txn - ) - - if not result: - yield self._end_background_update(self.EVENT_SEARCH_UPDATE_NAME) - - return result - - @defer.inlineCallbacks - def _background_reindex_gin_search(self, progress, batch_size): - """This handles old synapses which used GIST indexes, if any; - converting them back to be GIN as per the actual schema. - """ - - def create_index(conn): - conn.rollback() - - # we have to set autocommit, because postgres refuses to - # CREATE INDEX CONCURRENTLY without it. - conn.set_session(autocommit=True) - - try: - c = conn.cursor() - - # if we skipped the conversion to GIST, we may already/still - # have an event_search_fts_idx; unfortunately postgres 9.4 - # doesn't support CREATE INDEX IF EXISTS so we just catch the - # exception and ignore it. - import psycopg2 - - try: - c.execute( - "CREATE INDEX CONCURRENTLY event_search_fts_idx" - " ON event_search USING GIN (vector)" - ) - except psycopg2.ProgrammingError as e: - logger.warn( - "Ignoring error %r when trying to switch from GIST to GIN", e - ) - - # we should now be able to delete the GIST index. - c.execute("DROP INDEX IF EXISTS event_search_fts_idx_gist") - finally: - conn.set_session(autocommit=False) - - if isinstance(self.database_engine, PostgresEngine): - yield self.runWithConnection(create_index) - - yield self._end_background_update(self.EVENT_SEARCH_USE_GIN_POSTGRES_NAME) - return 1 - - @defer.inlineCallbacks - def _background_reindex_search_order(self, progress, batch_size): - target_min_stream_id = progress["target_min_stream_id_inclusive"] - max_stream_id = progress["max_stream_id_exclusive"] - rows_inserted = progress.get("rows_inserted", 0) - have_added_index = progress["have_added_indexes"] - - if not have_added_index: - - def create_index(conn): - conn.rollback() - conn.set_session(autocommit=True) - c = conn.cursor() - - # We create with NULLS FIRST so that when we search *backwards* - # we get the ones with non null origin_server_ts *first* - c.execute( - "CREATE INDEX CONCURRENTLY event_search_room_order ON event_search(" - "room_id, origin_server_ts NULLS FIRST, stream_ordering NULLS FIRST)" - ) - c.execute( - "CREATE INDEX CONCURRENTLY event_search_order ON event_search(" - "origin_server_ts NULLS FIRST, stream_ordering NULLS FIRST)" - ) - conn.set_session(autocommit=False) - - yield self.runWithConnection(create_index) - - pg = dict(progress) - pg["have_added_indexes"] = True - - yield self.runInteraction( - self.EVENT_SEARCH_ORDER_UPDATE_NAME, - self._background_update_progress_txn, - self.EVENT_SEARCH_ORDER_UPDATE_NAME, - pg, - ) - - def reindex_search_txn(txn): - sql = ( - "UPDATE event_search AS es SET stream_ordering = e.stream_ordering," - " origin_server_ts = e.origin_server_ts" - " FROM events AS e" - " WHERE e.event_id = es.event_id" - " AND ? <= e.stream_ordering AND e.stream_ordering < ?" - " RETURNING es.stream_ordering" - ) - - min_stream_id = max_stream_id - batch_size - txn.execute(sql, (min_stream_id, max_stream_id)) - rows = txn.fetchall() - - if min_stream_id < target_min_stream_id: - # We've recached the end. - return len(rows), False - - progress = { - "target_min_stream_id_inclusive": target_min_stream_id, - "max_stream_id_exclusive": min_stream_id, - "rows_inserted": rows_inserted + len(rows), - "have_added_indexes": True, - } - - self._background_update_progress_txn( - txn, self.EVENT_SEARCH_ORDER_UPDATE_NAME, progress - ) - - return len(rows), True - - num_rows, finished = yield self.runInteraction( - self.EVENT_SEARCH_ORDER_UPDATE_NAME, reindex_search_txn - ) - - if not finished: - yield self._end_background_update(self.EVENT_SEARCH_ORDER_UPDATE_NAME) - - return num_rows - - def store_search_entries_txn(self, txn, entries): - """Add entries to the search table - - Args: - txn (cursor): - entries (iterable[SearchEntry]): - entries to be added to the table - """ - if not self.hs.config.enable_search: - return - if isinstance(self.database_engine, PostgresEngine): - sql = ( - "INSERT INTO event_search" - " (event_id, room_id, key, vector, stream_ordering, origin_server_ts)" - " VALUES (?,?,?,to_tsvector('english', ?),?,?)" - ) - - args = ( - ( - entry.event_id, - entry.room_id, - entry.key, - entry.value, - entry.stream_ordering, - entry.origin_server_ts, - ) - for entry in entries - ) - - txn.executemany(sql, args) - - elif isinstance(self.database_engine, Sqlite3Engine): - sql = ( - "INSERT INTO event_search (event_id, room_id, key, value)" - " VALUES (?,?,?,?)" - ) - args = ( - (entry.event_id, entry.room_id, entry.key, entry.value) - for entry in entries - ) - - txn.executemany(sql, args) - else: - # This should be unreachable. - raise Exception("Unrecognized database engine") - - -class SearchStore(SearchBackgroundUpdateStore): - def __init__(self, db_conn, hs): - super(SearchStore, self).__init__(db_conn, hs) - - def store_event_search_txn(self, txn, event, key, value): - """Add event to the search table - - Args: - txn (cursor): - event (EventBase): - key (str): - value (str): - """ - self.store_search_entries_txn( - txn, - ( - SearchEntry( - key=key, - value=value, - event_id=event.event_id, - room_id=event.room_id, - stream_ordering=event.internal_metadata.stream_ordering, - origin_server_ts=event.origin_server_ts, - ), - ), - ) - - @defer.inlineCallbacks - def search_msgs(self, room_ids, search_term, keys): - """Performs a full text search over events with given keys. - - Args: - room_ids (list): List of room ids to search in - search_term (str): Search term to search for - keys (list): List of keys to search in, currently supports - "content.body", "content.name", "content.topic" - - Returns: - list of dicts - """ - clauses = [] - - search_query = search_query = _parse_query(self.database_engine, search_term) - - args = [] - - # Make sure we don't explode because the person is in too many rooms. - # We filter the results below regardless. - if len(room_ids) < 500: - clause, args = make_in_list_sql_clause( - self.database_engine, "room_id", room_ids - ) - clauses = [clause] - - local_clauses = [] - for key in keys: - local_clauses.append("key = ?") - args.append(key) - - clauses.append("(%s)" % (" OR ".join(local_clauses),)) - - count_args = args - count_clauses = clauses - - if isinstance(self.database_engine, PostgresEngine): - sql = ( - "SELECT ts_rank_cd(vector, to_tsquery('english', ?)) AS rank," - " room_id, event_id" - " FROM event_search" - " WHERE vector @@ to_tsquery('english', ?)" - ) - args = [search_query, search_query] + args - - count_sql = ( - "SELECT room_id, count(*) as count FROM event_search" - " WHERE vector @@ to_tsquery('english', ?)" - ) - count_args = [search_query] + count_args - elif isinstance(self.database_engine, Sqlite3Engine): - sql = ( - "SELECT rank(matchinfo(event_search)) as rank, room_id, event_id" - " FROM event_search" - " WHERE value MATCH ?" - ) - args = [search_query] + args - - count_sql = ( - "SELECT room_id, count(*) as count FROM event_search" - " WHERE value MATCH ?" - ) - count_args = [search_term] + count_args - else: - # This should be unreachable. - raise Exception("Unrecognized database engine") - - for clause in clauses: - sql += " AND " + clause - - for clause in count_clauses: - count_sql += " AND " + clause - - # We add an arbitrary limit here to ensure we don't try to pull the - # entire table from the database. - sql += " ORDER BY rank DESC LIMIT 500" - - results = yield self._execute("search_msgs", self.cursor_to_dict, sql, *args) - - results = list(filter(lambda row: row["room_id"] in room_ids, results)) - - events = yield self.get_events_as_list([r["event_id"] for r in results]) - - event_map = {ev.event_id: ev for ev in events} - - highlights = None - if isinstance(self.database_engine, PostgresEngine): - highlights = yield self._find_highlights_in_postgres(search_query, events) - - count_sql += " GROUP BY room_id" - - count_results = yield self._execute( - "search_rooms_count", self.cursor_to_dict, count_sql, *count_args - ) - - count = sum(row["count"] for row in count_results if row["room_id"] in room_ids) - - return { - "results": [ - {"event": event_map[r["event_id"]], "rank": r["rank"]} - for r in results - if r["event_id"] in event_map - ], - "highlights": highlights, - "count": count, - } - - @defer.inlineCallbacks - def search_rooms(self, room_ids, search_term, keys, limit, pagination_token=None): - """Performs a full text search over events with given keys. - - Args: - room_id (list): The room_ids to search in - search_term (str): Search term to search for - keys (list): List of keys to search in, currently supports - "content.body", "content.name", "content.topic" - pagination_token (str): A pagination token previously returned - - Returns: - list of dicts - """ - clauses = [] - - search_query = search_query = _parse_query(self.database_engine, search_term) - - args = [] - - # Make sure we don't explode because the person is in too many rooms. - # We filter the results below regardless. - if len(room_ids) < 500: - clause, args = make_in_list_sql_clause( - self.database_engine, "room_id", room_ids - ) - clauses = [clause] - - local_clauses = [] - for key in keys: - local_clauses.append("key = ?") - args.append(key) - - clauses.append("(%s)" % (" OR ".join(local_clauses),)) - - # take copies of the current args and clauses lists, before adding - # pagination clauses to main query. - count_args = list(args) - count_clauses = list(clauses) - - if pagination_token: - try: - origin_server_ts, stream = pagination_token.split(",") - origin_server_ts = int(origin_server_ts) - stream = int(stream) - except Exception: - raise SynapseError(400, "Invalid pagination token") - - clauses.append( - "(origin_server_ts < ?" - " OR (origin_server_ts = ? AND stream_ordering < ?))" - ) - args.extend([origin_server_ts, origin_server_ts, stream]) - - if isinstance(self.database_engine, PostgresEngine): - sql = ( - "SELECT ts_rank_cd(vector, to_tsquery('english', ?)) as rank," - " origin_server_ts, stream_ordering, room_id, event_id" - " FROM event_search" - " WHERE vector @@ to_tsquery('english', ?) AND " - ) - args = [search_query, search_query] + args - - count_sql = ( - "SELECT room_id, count(*) as count FROM event_search" - " WHERE vector @@ to_tsquery('english', ?) AND " - ) - count_args = [search_query] + count_args - elif isinstance(self.database_engine, Sqlite3Engine): - # We use CROSS JOIN here to ensure we use the right indexes. - # https://sqlite.org/optoverview.html#crossjoin - # - # We want to use the full text search index on event_search to - # extract all possible matches first, then lookup those matches - # in the events table to get the topological ordering. We need - # to use the indexes in this order because sqlite refuses to - # MATCH unless it uses the full text search index - sql = ( - "SELECT rank(matchinfo) as rank, room_id, event_id," - " origin_server_ts, stream_ordering" - " FROM (SELECT key, event_id, matchinfo(event_search) as matchinfo" - " FROM event_search" - " WHERE value MATCH ?" - " )" - " CROSS JOIN events USING (event_id)" - " WHERE " - ) - args = [search_query] + args - - count_sql = ( - "SELECT room_id, count(*) as count FROM event_search" - " WHERE value MATCH ? AND " - ) - count_args = [search_term] + count_args - else: - # This should be unreachable. - raise Exception("Unrecognized database engine") - - sql += " AND ".join(clauses) - count_sql += " AND ".join(count_clauses) - - # We add an arbitrary limit here to ensure we don't try to pull the - # entire table from the database. - if isinstance(self.database_engine, PostgresEngine): - sql += ( - " ORDER BY origin_server_ts DESC NULLS LAST," - " stream_ordering DESC NULLS LAST LIMIT ?" - ) - elif isinstance(self.database_engine, Sqlite3Engine): - sql += " ORDER BY origin_server_ts DESC, stream_ordering DESC LIMIT ?" - else: - raise Exception("Unrecognized database engine") - - args.append(limit) - - results = yield self._execute("search_rooms", self.cursor_to_dict, sql, *args) - - results = list(filter(lambda row: row["room_id"] in room_ids, results)) - - events = yield self.get_events_as_list([r["event_id"] for r in results]) - - event_map = {ev.event_id: ev for ev in events} - - highlights = None - if isinstance(self.database_engine, PostgresEngine): - highlights = yield self._find_highlights_in_postgres(search_query, events) - - count_sql += " GROUP BY room_id" - - count_results = yield self._execute( - "search_rooms_count", self.cursor_to_dict, count_sql, *count_args - ) - - count = sum(row["count"] for row in count_results if row["room_id"] in room_ids) - - return { - "results": [ - { - "event": event_map[r["event_id"]], - "rank": r["rank"], - "pagination_token": "%s,%s" - % (r["origin_server_ts"], r["stream_ordering"]), - } - for r in results - if r["event_id"] in event_map - ], - "highlights": highlights, - "count": count, - } - - def _find_highlights_in_postgres(self, search_query, events): - """Given a list of events and a search term, return a list of words - that match from the content of the event. - - This is used to give a list of words that clients can match against to - highlight the matching parts. - - Args: - search_query (str) - events (list): A list of events - - Returns: - deferred : A set of strings. - """ - - def f(txn): - highlight_words = set() - for event in events: - # As a hack we simply join values of all possible keys. This is - # fine since we're only using them to find possible highlights. - values = [] - for key in ("body", "name", "topic"): - v = event.content.get(key, None) - if v: - values.append(v) - - if not values: - continue - - value = " ".join(values) - - # We need to find some values for StartSel and StopSel that - # aren't in the value so that we can pick results out. - start_sel = "<" - stop_sel = ">" - - while start_sel in value: - start_sel += "<" - while stop_sel in value: - stop_sel += ">" - - query = "SELECT ts_headline(?, to_tsquery('english', ?), %s)" % ( - _to_postgres_options( - { - "StartSel": start_sel, - "StopSel": stop_sel, - "MaxFragments": "50", - } - ) - ) - txn.execute(query, (value, search_query)) - headline, = txn.fetchall()[0] - - # Now we need to pick the possible highlights out of the haedline - # result. - matcher_regex = "%s(.*?)%s" % ( - re.escape(start_sel), - re.escape(stop_sel), - ) - - res = re.findall(matcher_regex, headline) - highlight_words.update([r.lower() for r in res]) - - return highlight_words - - return self.runInteraction("_find_highlights", f) - - -def _to_postgres_options(options_dict): - return "'%s'" % (",".join("%s=%s" % (k, v) for k, v in options_dict.items()),) - - -def _parse_query(database_engine, search_term): - """Takes a plain unicode string from the user and converts it into a form - that can be passed to database. - We use this so that we can add prefix matching, which isn't something - that is supported by default. - """ - - # Pull out the individual words, discarding any non-word characters. - results = re.findall(r"([\w\-]+)", search_term, re.UNICODE) - - if isinstance(database_engine, PostgresEngine): - return " & ".join(result + ":*" for result in results) - elif isinstance(database_engine, Sqlite3Engine): - return " & ".join(result + "*" for result in results) - else: - # This should be unreachable. - raise Exception("Unrecognized database engine") diff --git a/synapse/storage/signatures.py b/synapse/storage/signatures.py deleted file mode 100644 index fb83218f90..0000000000 --- a/synapse/storage/signatures.py +++ /dev/null @@ -1,102 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2014-2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import six - -from unpaddedbase64 import encode_base64 - -from twisted.internet import defer - -from synapse.crypto.event_signing import compute_event_reference_hash -from synapse.util.caches.descriptors import cached, cachedList - -from ._base import SQLBaseStore - -# py2 sqlite has buffer hardcoded as only binary type, so we must use it, -# despite being deprecated and removed in favor of memoryview -if six.PY2: - db_binary_type = six.moves.builtins.buffer -else: - db_binary_type = memoryview - - -class SignatureWorkerStore(SQLBaseStore): - @cached() - def get_event_reference_hash(self, event_id): - # This is a dummy function to allow get_event_reference_hashes - # to use its cache - raise NotImplementedError() - - @cachedList( - cached_method_name="get_event_reference_hash", list_name="event_ids", num_args=1 - ) - def get_event_reference_hashes(self, event_ids): - def f(txn): - return { - event_id: self._get_event_reference_hashes_txn(txn, event_id) - for event_id in event_ids - } - - return self.runInteraction("get_event_reference_hashes", f) - - @defer.inlineCallbacks - def add_event_hashes(self, event_ids): - hashes = yield self.get_event_reference_hashes(event_ids) - hashes = { - e_id: {k: encode_base64(v) for k, v in h.items() if k == "sha256"} - for e_id, h in hashes.items() - } - - return list(hashes.items()) - - def _get_event_reference_hashes_txn(self, txn, event_id): - """Get all the hashes for a given PDU. - Args: - txn (cursor): - event_id (str): Id for the Event. - Returns: - A dict[unicode, bytes] of algorithm -> hash. - """ - query = ( - "SELECT algorithm, hash" - " FROM event_reference_hashes" - " WHERE event_id = ?" - ) - txn.execute(query, (event_id,)) - return {k: v for k, v in txn} - - -class SignatureStore(SignatureWorkerStore): - """Persistence for event signatures and hashes""" - - def _store_event_reference_hashes_txn(self, txn, events): - """Store a hash for a PDU - Args: - txn (cursor): - events (list): list of Events. - """ - - vals = [] - for event in events: - ref_alg, ref_hash_bytes = compute_event_reference_hash(event) - vals.append( - { - "event_id": event.event_id, - "algorithm": ref_alg, - "hash": db_binary_type(ref_hash_bytes), - } - ) - - self._simple_insert_many_txn(txn, table="event_reference_hashes", values=vals) diff --git a/synapse/storage/state.py b/synapse/storage/state.py index a941a5ae3f..a2df8fa827 100644 --- a/synapse/storage/state.py +++ b/synapse/storage/state.py @@ -14,45 +14,16 @@ # limitations under the License. import logging -from collections import namedtuple from six import iteritems, itervalues -from six.moves import range import attr -from twisted.internet import defer - from synapse.api.constants import EventTypes -from synapse.api.errors import NotFoundError -from synapse.storage._base import SQLBaseStore -from synapse.storage.background_updates import BackgroundUpdateStore -from synapse.storage.engines import PostgresEngine -from synapse.storage.events_worker import EventsWorkerStore -from synapse.util.caches import get_cache_factor_for, intern_string -from synapse.util.caches.descriptors import cached, cachedList -from synapse.util.caches.dictionary_cache import DictionaryCache -from synapse.util.stringutils import to_ascii logger = logging.getLogger(__name__) -MAX_STATE_DELTA_HOPS = 100 - - -class _GetStateGroupDelta( - namedtuple("_GetStateGroupDelta", ("prev_group", "delta_ids")) -): - """Return type of get_state_group_delta that implements __len__, which lets - us use the itrable flag when caching - """ - - __slots__ = [] - - def __len__(self): - return len(self.delta_ids) if self.delta_ids else 0 - - @attr.s(slots=True) class StateFilter(object): """A filter used when querying for state. @@ -351,1195 +322,3 @@ class StateFilter(object): ) return member_filter, non_member_filter - - -class StateGroupBackgroundUpdateStore(SQLBaseStore): - """Defines functions related to state groups needed to run the state backgroud - updates. - """ - - def _count_state_group_hops_txn(self, txn, state_group): - """Given a state group, count how many hops there are in the tree. - - This is used to ensure the delta chains don't get too long. - """ - if isinstance(self.database_engine, PostgresEngine): - sql = """ - WITH RECURSIVE state(state_group) AS ( - VALUES(?::bigint) - UNION ALL - SELECT prev_state_group FROM state_group_edges e, state s - WHERE s.state_group = e.state_group - ) - SELECT count(*) FROM state; - """ - - txn.execute(sql, (state_group,)) - row = txn.fetchone() - if row and row[0]: - return row[0] - else: - return 0 - else: - # We don't use WITH RECURSIVE on sqlite3 as there are distributions - # that ship with an sqlite3 version that doesn't support it (e.g. wheezy) - next_group = state_group - count = 0 - - while next_group: - next_group = self._simple_select_one_onecol_txn( - txn, - table="state_group_edges", - keyvalues={"state_group": next_group}, - retcol="prev_state_group", - allow_none=True, - ) - if next_group: - count += 1 - - return count - - def _get_state_groups_from_groups_txn( - self, txn, groups, state_filter=StateFilter.all() - ): - results = {group: {} for group in groups} - - where_clause, where_args = state_filter.make_sql_filter_clause() - - # Unless the filter clause is empty, we're going to append it after an - # existing where clause - if where_clause: - where_clause = " AND (%s)" % (where_clause,) - - if isinstance(self.database_engine, PostgresEngine): - # Temporarily disable sequential scans in this transaction. This is - # a temporary hack until we can add the right indices in - txn.execute("SET LOCAL enable_seqscan=off") - - # The below query walks the state_group tree so that the "state" - # table includes all state_groups in the tree. It then joins - # against `state_groups_state` to fetch the latest state. - # It assumes that previous state groups are always numerically - # lesser. - # The PARTITION is used to get the event_id in the greatest state - # group for the given type, state_key. - # This may return multiple rows per (type, state_key), but last_value - # should be the same. - sql = """ - WITH RECURSIVE state(state_group) AS ( - VALUES(?::bigint) - UNION ALL - SELECT prev_state_group FROM state_group_edges e, state s - WHERE s.state_group = e.state_group - ) - SELECT DISTINCT type, state_key, last_value(event_id) OVER ( - PARTITION BY type, state_key ORDER BY state_group ASC - ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING - ) AS event_id FROM state_groups_state - WHERE state_group IN ( - SELECT state_group FROM state - ) - """ - - for group in groups: - args = [group] - args.extend(where_args) - - txn.execute(sql + where_clause, args) - for row in txn: - typ, state_key, event_id = row - key = (typ, state_key) - results[group][key] = event_id - else: - max_entries_returned = state_filter.max_entries_returned() - - # We don't use WITH RECURSIVE on sqlite3 as there are distributions - # that ship with an sqlite3 version that doesn't support it (e.g. wheezy) - for group in groups: - next_group = group - - while next_group: - # We did this before by getting the list of group ids, and - # then passing that list to sqlite to get latest event for - # each (type, state_key). However, that was terribly slow - # without the right indices (which we can't add until - # after we finish deduping state, which requires this func) - args = [next_group] - args.extend(where_args) - - txn.execute( - "SELECT type, state_key, event_id FROM state_groups_state" - " WHERE state_group = ? " + where_clause, - args, - ) - results[group].update( - ((typ, state_key), event_id) - for typ, state_key, event_id in txn - if (typ, state_key) not in results[group] - ) - - # If the number of entries in the (type,state_key)->event_id dict - # matches the number of (type,state_keys) types we were searching - # for, then we must have found them all, so no need to go walk - # further down the tree... UNLESS our types filter contained - # wildcards (i.e. Nones) in which case we have to do an exhaustive - # search - if ( - max_entries_returned is not None - and len(results[group]) == max_entries_returned - ): - break - - next_group = self._simple_select_one_onecol_txn( - txn, - table="state_group_edges", - keyvalues={"state_group": next_group}, - retcol="prev_state_group", - allow_none=True, - ) - - return results - - -# this inherits from EventsWorkerStore because it calls self.get_events -class StateGroupWorkerStore( - EventsWorkerStore, StateGroupBackgroundUpdateStore, SQLBaseStore -): - """The parts of StateGroupStore that can be called from workers. - """ - - STATE_GROUP_DEDUPLICATION_UPDATE_NAME = "state_group_state_deduplication" - STATE_GROUP_INDEX_UPDATE_NAME = "state_group_state_type_index" - CURRENT_STATE_INDEX_UPDATE_NAME = "current_state_members_idx" - - def __init__(self, db_conn, hs): - super(StateGroupWorkerStore, self).__init__(db_conn, hs) - - # Originally the state store used a single DictionaryCache to cache the - # event IDs for the state types in a given state group to avoid hammering - # on the state_group* tables. - # - # The point of using a DictionaryCache is that it can cache a subset - # of the state events for a given state group (i.e. a subset of the keys for a - # given dict which is an entry in the cache for a given state group ID). - # - # However, this poses problems when performing complicated queries - # on the store - for instance: "give me all the state for this group, but - # limit members to this subset of users", as DictionaryCache's API isn't - # rich enough to say "please cache any of these fields, apart from this subset". - # This is problematic when lazy loading members, which requires this behaviour, - # as without it the cache has no choice but to speculatively load all - # state events for the group, which negates the efficiency being sought. - # - # Rather than overcomplicating DictionaryCache's API, we instead split the - # state_group_cache into two halves - one for tracking non-member events, - # and the other for tracking member_events. This means that lazy loading - # queries can be made in a cache-friendly manner by querying both caches - # separately and then merging the result. So for the example above, you - # would query the members cache for a specific subset of state keys - # (which DictionaryCache will handle efficiently and fine) and the non-members - # cache for all state (which DictionaryCache will similarly handle fine) - # and then just merge the results together. - # - # We size the non-members cache to be smaller than the members cache as the - # vast majority of state in Matrix (today) is member events. - - self._state_group_cache = DictionaryCache( - "*stateGroupCache*", - # TODO: this hasn't been tuned yet - 50000 * get_cache_factor_for("stateGroupCache"), - ) - self._state_group_members_cache = DictionaryCache( - "*stateGroupMembersCache*", - 500000 * get_cache_factor_for("stateGroupMembersCache"), - ) - - @defer.inlineCallbacks - def get_room_version(self, room_id): - """Get the room_version of a given room - - Args: - room_id (str) - - Returns: - Deferred[str] - - Raises: - NotFoundError if the room is unknown - """ - # for now we do this by looking at the create event. We may want to cache this - # more intelligently in future. - - # Retrieve the room's create event - create_event = yield self.get_create_event_for_room(room_id) - return create_event.content.get("room_version", "1") - - @defer.inlineCallbacks - def get_room_predecessor(self, room_id): - """Get the predecessor room of an upgraded room if one exists. - Otherwise return None. - - Args: - room_id (str) - - Returns: - Deferred[unicode|None]: predecessor room id - - Raises: - NotFoundError if the room is unknown - """ - # Retrieve the room's create event - create_event = yield self.get_create_event_for_room(room_id) - - # Return predecessor if present - return create_event.content.get("predecessor", None) - - @defer.inlineCallbacks - def get_create_event_for_room(self, room_id): - """Get the create state event for a room. - - Args: - room_id (str) - - Returns: - Deferred[EventBase]: The room creation event. - - Raises: - NotFoundError if the room is unknown - """ - state_ids = yield self.get_current_state_ids(room_id) - create_id = state_ids.get((EventTypes.Create, "")) - - # If we can't find the create event, assume we've hit a dead end - if not create_id: - raise NotFoundError("Unknown room %s" % (room_id)) - - # Retrieve the room's create event and return - create_event = yield self.get_event(create_id) - return create_event - - @cached(max_entries=100000, iterable=True) - def get_current_state_ids(self, room_id): - """Get the current state event ids for a room based on the - current_state_events table. - - Args: - room_id (str) - - Returns: - deferred: dict of (type, state_key) -> event_id - """ - - def _get_current_state_ids_txn(txn): - txn.execute( - """SELECT type, state_key, event_id FROM current_state_events - WHERE room_id = ? - """, - (room_id,), - ) - - return { - (intern_string(r[0]), intern_string(r[1])): to_ascii(r[2]) for r in txn - } - - return self.runInteraction("get_current_state_ids", _get_current_state_ids_txn) - - # FIXME: how should this be cached? - def get_filtered_current_state_ids(self, room_id, state_filter=StateFilter.all()): - """Get the current state event of a given type for a room based on the - current_state_events table. This may not be as up-to-date as the result - of doing a fresh state resolution as per state_handler.get_current_state - - Args: - room_id (str) - state_filter (StateFilter): The state filter used to fetch state - from the database. - - Returns: - Deferred[dict[tuple[str, str], str]]: Map from type/state_key to - event ID. - """ - - where_clause, where_args = state_filter.make_sql_filter_clause() - - if not where_clause: - # We delegate to the cached version - return self.get_current_state_ids(room_id) - - def _get_filtered_current_state_ids_txn(txn): - results = {} - sql = """ - SELECT type, state_key, event_id FROM current_state_events - WHERE room_id = ? - """ - - if where_clause: - sql += " AND (%s)" % (where_clause,) - - args = [room_id] - args.extend(where_args) - txn.execute(sql, args) - for row in txn: - typ, state_key, event_id = row - key = (intern_string(typ), intern_string(state_key)) - results[key] = event_id - - return results - - return self.runInteraction( - "get_filtered_current_state_ids", _get_filtered_current_state_ids_txn - ) - - @defer.inlineCallbacks - def get_canonical_alias_for_room(self, room_id): - """Get canonical alias for room, if any - - Args: - room_id (str) - - Returns: - Deferred[str|None]: The canonical alias, if any - """ - - state = yield self.get_filtered_current_state_ids( - room_id, StateFilter.from_types([(EventTypes.CanonicalAlias, "")]) - ) - - event_id = state.get((EventTypes.CanonicalAlias, "")) - if not event_id: - return - - event = yield self.get_event(event_id, allow_none=True) - if not event: - return - - return event.content.get("canonical_alias") - - @cached(max_entries=10000, iterable=True) - def get_state_group_delta(self, state_group): - """Given a state group try to return a previous group and a delta between - the old and the new. - - Returns: - (prev_group, delta_ids), where both may be None. - """ - - def _get_state_group_delta_txn(txn): - prev_group = self._simple_select_one_onecol_txn( - txn, - table="state_group_edges", - keyvalues={"state_group": state_group}, - retcol="prev_state_group", - allow_none=True, - ) - - if not prev_group: - return _GetStateGroupDelta(None, None) - - delta_ids = self._simple_select_list_txn( - txn, - table="state_groups_state", - keyvalues={"state_group": state_group}, - retcols=("type", "state_key", "event_id"), - ) - - return _GetStateGroupDelta( - prev_group, - {(row["type"], row["state_key"]): row["event_id"] for row in delta_ids}, - ) - - return self.runInteraction("get_state_group_delta", _get_state_group_delta_txn) - - @defer.inlineCallbacks - def get_state_groups_ids(self, _room_id, event_ids): - """Get the event IDs of all the state for the state groups for the given events - - Args: - _room_id (str): id of the room for these events - event_ids (iterable[str]): ids of the events - - Returns: - Deferred[dict[int, dict[tuple[str, str], str]]]: - dict of state_group_id -> (dict of (type, state_key) -> event id) - """ - if not event_ids: - return {} - - event_to_groups = yield self._get_state_group_for_events(event_ids) - - groups = set(itervalues(event_to_groups)) - group_to_state = yield self._get_state_for_groups(groups) - - return group_to_state - - @defer.inlineCallbacks - def get_state_ids_for_group(self, state_group): - """Get the event IDs of all the state in the given state group - - Args: - state_group (int) - - Returns: - Deferred[dict]: Resolves to a map of (type, state_key) -> event_id - """ - group_to_state = yield self._get_state_for_groups((state_group,)) - - return group_to_state[state_group] - - @defer.inlineCallbacks - def get_state_groups(self, room_id, event_ids): - """ Get the state groups for the given list of event_ids - - Returns: - Deferred[dict[int, list[EventBase]]]: - dict of state_group_id -> list of state events. - """ - if not event_ids: - return {} - - group_to_ids = yield self.get_state_groups_ids(room_id, event_ids) - - state_event_map = yield self.get_events( - [ - ev_id - for group_ids in itervalues(group_to_ids) - for ev_id in itervalues(group_ids) - ], - get_prev_content=False, - ) - - return { - group: [ - state_event_map[v] - for v in itervalues(event_id_map) - if v in state_event_map - ] - for group, event_id_map in iteritems(group_to_ids) - } - - @defer.inlineCallbacks - def _get_state_groups_from_groups(self, groups, state_filter): - """Returns the state groups for a given set of groups, filtering on - types of state events. - - Args: - groups(list[int]): list of state group IDs to query - state_filter (StateFilter): The state filter used to fetch state - from the database. - Returns: - Deferred[dict[int, dict[tuple[str, str], str]]]: - dict of state_group_id -> (dict of (type, state_key) -> event id) - """ - results = {} - - chunks = [groups[i : i + 100] for i in range(0, len(groups), 100)] - for chunk in chunks: - res = yield self.runInteraction( - "_get_state_groups_from_groups", - self._get_state_groups_from_groups_txn, - chunk, - state_filter, - ) - results.update(res) - - return results - - @defer.inlineCallbacks - def get_state_for_events(self, event_ids, state_filter=StateFilter.all()): - """Given a list of event_ids and type tuples, return a list of state - dicts for each event. - - Args: - event_ids (list[string]) - state_filter (StateFilter): The state filter used to fetch state - from the database. - - Returns: - deferred: A dict of (event_id) -> (type, state_key) -> [state_events] - """ - event_to_groups = yield self._get_state_group_for_events(event_ids) - - groups = set(itervalues(event_to_groups)) - group_to_state = yield self._get_state_for_groups(groups, state_filter) - - state_event_map = yield self.get_events( - [ev_id for sd in itervalues(group_to_state) for ev_id in itervalues(sd)], - get_prev_content=False, - ) - - event_to_state = { - event_id: { - k: state_event_map[v] - for k, v in iteritems(group_to_state[group]) - if v in state_event_map - } - for event_id, group in iteritems(event_to_groups) - } - - return {event: event_to_state[event] for event in event_ids} - - @defer.inlineCallbacks - def get_state_ids_for_events(self, event_ids, state_filter=StateFilter.all()): - """ - Get the state dicts corresponding to a list of events, containing the event_ids - of the state events (as opposed to the events themselves) - - Args: - event_ids(list(str)): events whose state should be returned - state_filter (StateFilter): The state filter used to fetch state - from the database. - - Returns: - A deferred dict from event_id -> (type, state_key) -> event_id - """ - event_to_groups = yield self._get_state_group_for_events(event_ids) - - groups = set(itervalues(event_to_groups)) - group_to_state = yield self._get_state_for_groups(groups, state_filter) - - event_to_state = { - event_id: group_to_state[group] - for event_id, group in iteritems(event_to_groups) - } - - return {event: event_to_state[event] for event in event_ids} - - @defer.inlineCallbacks - def get_state_for_event(self, event_id, state_filter=StateFilter.all()): - """ - Get the state dict corresponding to a particular event - - Args: - event_id(str): event whose state should be returned - state_filter (StateFilter): The state filter used to fetch state - from the database. - - Returns: - A deferred dict from (type, state_key) -> state_event - """ - state_map = yield self.get_state_for_events([event_id], state_filter) - return state_map[event_id] - - @defer.inlineCallbacks - def get_state_ids_for_event(self, event_id, state_filter=StateFilter.all()): - """ - Get the state dict corresponding to a particular event - - Args: - event_id(str): event whose state should be returned - state_filter (StateFilter): The state filter used to fetch state - from the database. - - Returns: - A deferred dict from (type, state_key) -> state_event - """ - state_map = yield self.get_state_ids_for_events([event_id], state_filter) - return state_map[event_id] - - @cached(max_entries=50000) - def _get_state_group_for_event(self, event_id): - return self._simple_select_one_onecol( - table="event_to_state_groups", - keyvalues={"event_id": event_id}, - retcol="state_group", - allow_none=True, - desc="_get_state_group_for_event", - ) - - @cachedList( - cached_method_name="_get_state_group_for_event", - list_name="event_ids", - num_args=1, - inlineCallbacks=True, - ) - def _get_state_group_for_events(self, event_ids): - """Returns mapping event_id -> state_group - """ - rows = yield self._simple_select_many_batch( - table="event_to_state_groups", - column="event_id", - iterable=event_ids, - keyvalues={}, - retcols=("event_id", "state_group"), - desc="_get_state_group_for_events", - ) - - return {row["event_id"]: row["state_group"] for row in rows} - - def _get_state_for_group_using_cache(self, cache, group, state_filter): - """Checks if group is in cache. See `_get_state_for_groups` - - Args: - cache(DictionaryCache): the state group cache to use - group(int): The state group to lookup - state_filter (StateFilter): The state filter used to fetch state - from the database. - - Returns 2-tuple (`state_dict`, `got_all`). - `got_all` is a bool indicating if we successfully retrieved all - requests state from the cache, if False we need to query the DB for the - missing state. - """ - is_all, known_absent, state_dict_ids = cache.get(group) - - if is_all or state_filter.is_full(): - # Either we have everything or want everything, either way - # `is_all` tells us whether we've gotten everything. - return state_filter.filter_state(state_dict_ids), is_all - - # tracks whether any of our requested types are missing from the cache - missing_types = False - - if state_filter.has_wildcards(): - # We don't know if we fetched all the state keys for the types in - # the filter that are wildcards, so we have to assume that we may - # have missed some. - missing_types = True - else: - # There aren't any wild cards, so `concrete_types()` returns the - # complete list of event types we're wanting. - for key in state_filter.concrete_types(): - if key not in state_dict_ids and key not in known_absent: - missing_types = True - break - - return state_filter.filter_state(state_dict_ids), not missing_types - - @defer.inlineCallbacks - def _get_state_for_groups(self, groups, state_filter=StateFilter.all()): - """Gets the state at each of a list of state groups, optionally - filtering by type/state_key - - Args: - groups (iterable[int]): list of state groups for which we want - to get the state. - state_filter (StateFilter): The state filter used to fetch state - from the database. - Returns: - Deferred[dict[int, dict[tuple[str, str], str]]]: - dict of state_group_id -> (dict of (type, state_key) -> event id) - """ - - member_filter, non_member_filter = state_filter.get_member_split() - - # Now we look them up in the member and non-member caches - non_member_state, incomplete_groups_nm, = ( - yield self._get_state_for_groups_using_cache( - groups, self._state_group_cache, state_filter=non_member_filter - ) - ) - - member_state, incomplete_groups_m, = ( - yield self._get_state_for_groups_using_cache( - groups, self._state_group_members_cache, state_filter=member_filter - ) - ) - - state = dict(non_member_state) - for group in groups: - state[group].update(member_state[group]) - - # Now fetch any missing groups from the database - - incomplete_groups = incomplete_groups_m | incomplete_groups_nm - - if not incomplete_groups: - return state - - cache_sequence_nm = self._state_group_cache.sequence - cache_sequence_m = self._state_group_members_cache.sequence - - # Help the cache hit ratio by expanding the filter a bit - db_state_filter = state_filter.return_expanded() - - group_to_state_dict = yield self._get_state_groups_from_groups( - list(incomplete_groups), state_filter=db_state_filter - ) - - # Now lets update the caches - self._insert_into_cache( - group_to_state_dict, - db_state_filter, - cache_seq_num_members=cache_sequence_m, - cache_seq_num_non_members=cache_sequence_nm, - ) - - # And finally update the result dict, by filtering out any extra - # stuff we pulled out of the database. - for group, group_state_dict in iteritems(group_to_state_dict): - # We just replace any existing entries, as we will have loaded - # everything we need from the database anyway. - state[group] = state_filter.filter_state(group_state_dict) - - return state - - def _get_state_for_groups_using_cache(self, groups, cache, state_filter): - """Gets the state at each of a list of state groups, optionally - filtering by type/state_key, querying from a specific cache. - - Args: - groups (iterable[int]): list of state groups for which we want - to get the state. - cache (DictionaryCache): the cache of group ids to state dicts which - we will pass through - either the normal state cache or the specific - members state cache. - state_filter (StateFilter): The state filter used to fetch state - from the database. - - Returns: - tuple[dict[int, dict[tuple[str, str], str]], set[int]]: Tuple of - dict of state_group_id -> (dict of (type, state_key) -> event id) - of entries in the cache, and the state group ids either missing - from the cache or incomplete. - """ - results = {} - incomplete_groups = set() - for group in set(groups): - state_dict_ids, got_all = self._get_state_for_group_using_cache( - cache, group, state_filter - ) - results[group] = state_dict_ids - - if not got_all: - incomplete_groups.add(group) - - return results, incomplete_groups - - def _insert_into_cache( - self, - group_to_state_dict, - state_filter, - cache_seq_num_members, - cache_seq_num_non_members, - ): - """Inserts results from querying the database into the relevant cache. - - Args: - group_to_state_dict (dict): The new entries pulled from database. - Map from state group to state dict - state_filter (StateFilter): The state filter used to fetch state - from the database. - cache_seq_num_members (int): Sequence number of member cache since - last lookup in cache - cache_seq_num_non_members (int): Sequence number of member cache since - last lookup in cache - """ - - # We need to work out which types we've fetched from the DB for the - # member vs non-member caches. This should be as accurate as possible, - # but can be an underestimate (e.g. when we have wild cards) - - member_filter, non_member_filter = state_filter.get_member_split() - if member_filter.is_full(): - # We fetched all member events - member_types = None - else: - # `concrete_types()` will only return a subset when there are wild - # cards in the filter, but that's fine. - member_types = member_filter.concrete_types() - - if non_member_filter.is_full(): - # We fetched all non member events - non_member_types = None - else: - non_member_types = non_member_filter.concrete_types() - - for group, group_state_dict in iteritems(group_to_state_dict): - state_dict_members = {} - state_dict_non_members = {} - - for k, v in iteritems(group_state_dict): - if k[0] == EventTypes.Member: - state_dict_members[k] = v - else: - state_dict_non_members[k] = v - - self._state_group_members_cache.update( - cache_seq_num_members, - key=group, - value=state_dict_members, - fetched_keys=member_types, - ) - - self._state_group_cache.update( - cache_seq_num_non_members, - key=group, - value=state_dict_non_members, - fetched_keys=non_member_types, - ) - - def store_state_group( - self, event_id, room_id, prev_group, delta_ids, current_state_ids - ): - """Store a new set of state, returning a newly assigned state group. - - Args: - event_id (str): The event ID for which the state was calculated - room_id (str) - prev_group (int|None): A previous state group for the room, optional. - delta_ids (dict|None): The delta between state at `prev_group` and - `current_state_ids`, if `prev_group` was given. Same format as - `current_state_ids`. - current_state_ids (dict): The state to store. Map of (type, state_key) - to event_id. - - Returns: - Deferred[int]: The state group ID - """ - - def _store_state_group_txn(txn): - if current_state_ids is None: - # AFAIK, this can never happen - raise Exception("current_state_ids cannot be None") - - state_group = self.database_engine.get_next_state_group_id(txn) - - self._simple_insert_txn( - txn, - table="state_groups", - values={"id": state_group, "room_id": room_id, "event_id": event_id}, - ) - - # We persist as a delta if we can, while also ensuring the chain - # of deltas isn't tooo long, as otherwise read performance degrades. - if prev_group: - is_in_db = self._simple_select_one_onecol_txn( - txn, - table="state_groups", - keyvalues={"id": prev_group}, - retcol="id", - allow_none=True, - ) - if not is_in_db: - raise Exception( - "Trying to persist state with unpersisted prev_group: %r" - % (prev_group,) - ) - - potential_hops = self._count_state_group_hops_txn(txn, prev_group) - if prev_group and potential_hops < MAX_STATE_DELTA_HOPS: - self._simple_insert_txn( - txn, - table="state_group_edges", - values={"state_group": state_group, "prev_state_group": prev_group}, - ) - - self._simple_insert_many_txn( - txn, - table="state_groups_state", - values=[ - { - "state_group": state_group, - "room_id": room_id, - "type": key[0], - "state_key": key[1], - "event_id": state_id, - } - for key, state_id in iteritems(delta_ids) - ], - ) - else: - self._simple_insert_many_txn( - txn, - table="state_groups_state", - values=[ - { - "state_group": state_group, - "room_id": room_id, - "type": key[0], - "state_key": key[1], - "event_id": state_id, - } - for key, state_id in iteritems(current_state_ids) - ], - ) - - # Prefill the state group caches with this group. - # It's fine to use the sequence like this as the state group map - # is immutable. (If the map wasn't immutable then this prefill could - # race with another update) - - current_member_state_ids = { - s: ev - for (s, ev) in iteritems(current_state_ids) - if s[0] == EventTypes.Member - } - txn.call_after( - self._state_group_members_cache.update, - self._state_group_members_cache.sequence, - key=state_group, - value=dict(current_member_state_ids), - ) - - current_non_member_state_ids = { - s: ev - for (s, ev) in iteritems(current_state_ids) - if s[0] != EventTypes.Member - } - txn.call_after( - self._state_group_cache.update, - self._state_group_cache.sequence, - key=state_group, - value=dict(current_non_member_state_ids), - ) - - return state_group - - return self.runInteraction("store_state_group", _store_state_group_txn) - - -class StateBackgroundUpdateStore( - StateGroupBackgroundUpdateStore, BackgroundUpdateStore -): - - STATE_GROUP_DEDUPLICATION_UPDATE_NAME = "state_group_state_deduplication" - STATE_GROUP_INDEX_UPDATE_NAME = "state_group_state_type_index" - CURRENT_STATE_INDEX_UPDATE_NAME = "current_state_members_idx" - EVENT_STATE_GROUP_INDEX_UPDATE_NAME = "event_to_state_groups_sg_index" - - def __init__(self, db_conn, hs): - super(StateBackgroundUpdateStore, self).__init__(db_conn, hs) - self.register_background_update_handler( - self.STATE_GROUP_DEDUPLICATION_UPDATE_NAME, - self._background_deduplicate_state, - ) - self.register_background_update_handler( - self.STATE_GROUP_INDEX_UPDATE_NAME, self._background_index_state - ) - self.register_background_index_update( - self.CURRENT_STATE_INDEX_UPDATE_NAME, - index_name="current_state_events_member_index", - table="current_state_events", - columns=["state_key"], - where_clause="type='m.room.member'", - ) - self.register_background_index_update( - self.EVENT_STATE_GROUP_INDEX_UPDATE_NAME, - index_name="event_to_state_groups_sg_index", - table="event_to_state_groups", - columns=["state_group"], - ) - - @defer.inlineCallbacks - def _background_deduplicate_state(self, progress, batch_size): - """This background update will slowly deduplicate state by reencoding - them as deltas. - """ - last_state_group = progress.get("last_state_group", 0) - rows_inserted = progress.get("rows_inserted", 0) - max_group = progress.get("max_group", None) - - BATCH_SIZE_SCALE_FACTOR = 100 - - batch_size = max(1, int(batch_size / BATCH_SIZE_SCALE_FACTOR)) - - if max_group is None: - rows = yield self._execute( - "_background_deduplicate_state", - None, - "SELECT coalesce(max(id), 0) FROM state_groups", - ) - max_group = rows[0][0] - - def reindex_txn(txn): - new_last_state_group = last_state_group - for count in range(batch_size): - txn.execute( - "SELECT id, room_id FROM state_groups" - " WHERE ? < id AND id <= ?" - " ORDER BY id ASC" - " LIMIT 1", - (new_last_state_group, max_group), - ) - row = txn.fetchone() - if row: - state_group, room_id = row - - if not row or not state_group: - return True, count - - txn.execute( - "SELECT state_group FROM state_group_edges" - " WHERE state_group = ?", - (state_group,), - ) - - # If we reach a point where we've already started inserting - # edges we should stop. - if txn.fetchall(): - return True, count - - txn.execute( - "SELECT coalesce(max(id), 0) FROM state_groups" - " WHERE id < ? AND room_id = ?", - (state_group, room_id), - ) - prev_group, = txn.fetchone() - new_last_state_group = state_group - - if prev_group: - potential_hops = self._count_state_group_hops_txn(txn, prev_group) - if potential_hops >= MAX_STATE_DELTA_HOPS: - # We want to ensure chains are at most this long,# - # otherwise read performance degrades. - continue - - prev_state = self._get_state_groups_from_groups_txn( - txn, [prev_group] - ) - prev_state = prev_state[prev_group] - - curr_state = self._get_state_groups_from_groups_txn( - txn, [state_group] - ) - curr_state = curr_state[state_group] - - if not set(prev_state.keys()) - set(curr_state.keys()): - # We can only do a delta if the current has a strict super set - # of keys - - delta_state = { - key: value - for key, value in iteritems(curr_state) - if prev_state.get(key, None) != value - } - - self._simple_delete_txn( - txn, - table="state_group_edges", - keyvalues={"state_group": state_group}, - ) - - self._simple_insert_txn( - txn, - table="state_group_edges", - values={ - "state_group": state_group, - "prev_state_group": prev_group, - }, - ) - - self._simple_delete_txn( - txn, - table="state_groups_state", - keyvalues={"state_group": state_group}, - ) - - self._simple_insert_many_txn( - txn, - table="state_groups_state", - values=[ - { - "state_group": state_group, - "room_id": room_id, - "type": key[0], - "state_key": key[1], - "event_id": state_id, - } - for key, state_id in iteritems(delta_state) - ], - ) - - progress = { - "last_state_group": state_group, - "rows_inserted": rows_inserted + batch_size, - "max_group": max_group, - } - - self._background_update_progress_txn( - txn, self.STATE_GROUP_DEDUPLICATION_UPDATE_NAME, progress - ) - - return False, batch_size - - finished, result = yield self.runInteraction( - self.STATE_GROUP_DEDUPLICATION_UPDATE_NAME, reindex_txn - ) - - if finished: - yield self._end_background_update( - self.STATE_GROUP_DEDUPLICATION_UPDATE_NAME - ) - - return result * BATCH_SIZE_SCALE_FACTOR - - @defer.inlineCallbacks - def _background_index_state(self, progress, batch_size): - def reindex_txn(conn): - conn.rollback() - if isinstance(self.database_engine, PostgresEngine): - # postgres insists on autocommit for the index - conn.set_session(autocommit=True) - try: - txn = conn.cursor() - txn.execute( - "CREATE INDEX CONCURRENTLY state_groups_state_type_idx" - " ON state_groups_state(state_group, type, state_key)" - ) - txn.execute("DROP INDEX IF EXISTS state_groups_state_id") - finally: - conn.set_session(autocommit=False) - else: - txn = conn.cursor() - txn.execute( - "CREATE INDEX state_groups_state_type_idx" - " ON state_groups_state(state_group, type, state_key)" - ) - txn.execute("DROP INDEX IF EXISTS state_groups_state_id") - - yield self.runWithConnection(reindex_txn) - - yield self._end_background_update(self.STATE_GROUP_INDEX_UPDATE_NAME) - - return 1 - - -class StateStore(StateGroupWorkerStore, StateBackgroundUpdateStore): - """ Keeps track of the state at a given event. - - This is done by the concept of `state groups`. Every event is a assigned - a state group (identified by an arbitrary string), which references a - collection of state events. The current state of an event is then the - collection of state events referenced by the event's state group. - - Hence, every change in the current state causes a new state group to be - generated. However, if no change happens (e.g., if we get a message event - with only one parent it inherits the state group from its parent.) - - There are three tables: - * `state_groups`: Stores group name, first event with in the group and - room id. - * `event_to_state_groups`: Maps events to state groups. - * `state_groups_state`: Maps state group to state events. - """ - - def __init__(self, db_conn, hs): - super(StateStore, self).__init__(db_conn, hs) - - def _store_event_state_mappings_txn(self, txn, events_and_contexts): - state_groups = {} - for event, context in events_and_contexts: - if event.internal_metadata.is_outlier(): - continue - - # if the event was rejected, just give it the same state as its - # predecessor. - if context.rejected: - state_groups[event.event_id] = context.prev_group - continue - - state_groups[event.event_id] = context.state_group - - self._simple_insert_many_txn( - txn, - table="event_to_state_groups", - values=[ - {"state_group": state_group_id, "event_id": event_id} - for event_id, state_group_id in iteritems(state_groups) - ], - ) - - for event_id, state_group_id in iteritems(state_groups): - txn.call_after( - self._get_state_group_for_event.prefill, (event_id,), state_group_id - ) diff --git a/synapse/storage/state_deltas.py b/synapse/storage/state_deltas.py deleted file mode 100644 index 28f33ec18f..0000000000 --- a/synapse/storage/state_deltas.py +++ /dev/null @@ -1,119 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2018 Vector Creations Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import logging - -from synapse.storage._base import SQLBaseStore - -logger = logging.getLogger(__name__) - - -class StateDeltasStore(SQLBaseStore): - def get_current_state_deltas(self, prev_stream_id: int, max_stream_id: int): - """Fetch a list of room state changes since the given stream id - - Each entry in the result contains the following fields: - - stream_id (int) - - room_id (str) - - type (str): event type - - state_key (str): - - event_id (str|None): new event_id for this state key. None if the - state has been deleted. - - prev_event_id (str|None): previous event_id for this state key. None - if it's new state. - - Args: - prev_stream_id (int): point to get changes since (exclusive) - max_stream_id (int): the point that we know has been correctly persisted - - ie, an upper limit to return changes from. - - Returns: - Deferred[tuple[int, list[dict]]: A tuple consisting of: - - the stream id which these results go up to - - list of current_state_delta_stream rows. If it is empty, we are - up to date. - """ - prev_stream_id = int(prev_stream_id) - - # check we're not going backwards - assert prev_stream_id <= max_stream_id - - if not self._curr_state_delta_stream_cache.has_any_entity_changed( - prev_stream_id - ): - # if the CSDs haven't changed between prev_stream_id and now, we - # know for certain that they haven't changed between prev_stream_id and - # max_stream_id. - return max_stream_id, [] - - def get_current_state_deltas_txn(txn): - # First we calculate the max stream id that will give us less than - # N results. - # We arbitarily limit to 100 stream_id entries to ensure we don't - # select toooo many. - sql = """ - SELECT stream_id, count(*) - FROM current_state_delta_stream - WHERE stream_id > ? AND stream_id <= ? - GROUP BY stream_id - ORDER BY stream_id ASC - LIMIT 100 - """ - txn.execute(sql, (prev_stream_id, max_stream_id)) - - total = 0 - - for stream_id, count in txn: - total += count - if total > 100: - # We arbitarily limit to 100 entries to ensure we don't - # select toooo many. - logger.debug( - "Clipping current_state_delta_stream rows to stream_id %i", - stream_id, - ) - clipped_stream_id = stream_id - break - else: - # if there's no problem, we may as well go right up to the max_stream_id - clipped_stream_id = max_stream_id - - # Now actually get the deltas - sql = """ - SELECT stream_id, room_id, type, state_key, event_id, prev_event_id - FROM current_state_delta_stream - WHERE ? < stream_id AND stream_id <= ? - ORDER BY stream_id ASC - """ - txn.execute(sql, (prev_stream_id, clipped_stream_id)) - return clipped_stream_id, self.cursor_to_dict(txn) - - return self.runInteraction( - "get_current_state_deltas", get_current_state_deltas_txn - ) - - def _get_max_stream_id_in_current_state_deltas_txn(self, txn): - return self._simple_select_one_onecol_txn( - txn, - table="current_state_delta_stream", - keyvalues={}, - retcol="COALESCE(MAX(stream_id), -1)", - ) - - def get_max_stream_id_in_current_state_deltas(self): - return self.runInteraction( - "get_max_stream_id_in_current_state_deltas", - self._get_max_stream_id_in_current_state_deltas_txn, - ) diff --git a/synapse/storage/stats.py b/synapse/storage/stats.py deleted file mode 100644 index 7c224cd3d9..0000000000 --- a/synapse/storage/stats.py +++ /dev/null @@ -1,881 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2018, 2019 New Vector Ltd -# Copyright 2019 The Matrix.org Foundation C.I.C. -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import logging -from itertools import chain - -from twisted.internet import defer -from twisted.internet.defer import DeferredLock - -from synapse.api.constants import EventTypes, Membership -from synapse.storage import PostgresEngine -from synapse.storage.state_deltas import StateDeltasStore -from synapse.util.caches.descriptors import cached - -logger = logging.getLogger(__name__) - -# these fields track absolutes (e.g. total number of rooms on the server) -# You can think of these as Prometheus Gauges. -# You can draw these stats on a line graph. -# Example: number of users in a room -ABSOLUTE_STATS_FIELDS = { - "room": ( - "current_state_events", - "joined_members", - "invited_members", - "left_members", - "banned_members", - "local_users_in_room", - ), - "user": ("joined_rooms",), -} - -# these fields are per-timeslice and so should be reset to 0 upon a new slice -# You can draw these stats on a histogram. -# Example: number of events sent locally during a time slice -PER_SLICE_FIELDS = { - "room": ("total_events", "total_event_bytes"), - "user": ("invites_sent", "rooms_created", "total_events", "total_event_bytes"), -} - -TYPE_TO_TABLE = {"room": ("room_stats", "room_id"), "user": ("user_stats", "user_id")} - -# these are the tables (& ID columns) which contain our actual subjects -TYPE_TO_ORIGIN_TABLE = {"room": ("rooms", "room_id"), "user": ("users", "name")} - - -class StatsStore(StateDeltasStore): - def __init__(self, db_conn, hs): - super(StatsStore, self).__init__(db_conn, hs) - - self.server_name = hs.hostname - self.clock = self.hs.get_clock() - self.stats_enabled = hs.config.stats_enabled - self.stats_bucket_size = hs.config.stats_bucket_size - - self.stats_delta_processing_lock = DeferredLock() - - self.register_background_update_handler( - "populate_stats_process_rooms", self._populate_stats_process_rooms - ) - self.register_background_update_handler( - "populate_stats_process_users", self._populate_stats_process_users - ) - # we no longer need to perform clean-up, but we will give ourselves - # the potential to reintroduce it in the future – so documentation - # will still encourage the use of this no-op handler. - self.register_noop_background_update("populate_stats_cleanup") - self.register_noop_background_update("populate_stats_prepare") - - def quantise_stats_time(self, ts): - """ - Quantises a timestamp to be a multiple of the bucket size. - - Args: - ts (int): the timestamp to quantise, in milliseconds since the Unix - Epoch - - Returns: - int: a timestamp which - - is divisible by the bucket size; - - is no later than `ts`; and - - is the largest such timestamp. - """ - return (ts // self.stats_bucket_size) * self.stats_bucket_size - - @defer.inlineCallbacks - def _populate_stats_process_users(self, progress, batch_size): - """ - This is a background update which regenerates statistics for users. - """ - if not self.stats_enabled: - yield self._end_background_update("populate_stats_process_users") - return 1 - - last_user_id = progress.get("last_user_id", "") - - def _get_next_batch(txn): - sql = """ - SELECT DISTINCT name FROM users - WHERE name > ? - ORDER BY name ASC - LIMIT ? - """ - txn.execute(sql, (last_user_id, batch_size)) - return [r for r, in txn] - - users_to_work_on = yield self.runInteraction( - "_populate_stats_process_users", _get_next_batch - ) - - # No more rooms -- complete the transaction. - if not users_to_work_on: - yield self._end_background_update("populate_stats_process_users") - return 1 - - for user_id in users_to_work_on: - yield self._calculate_and_set_initial_state_for_user(user_id) - progress["last_user_id"] = user_id - - yield self.runInteraction( - "populate_stats_process_users", - self._background_update_progress_txn, - "populate_stats_process_users", - progress, - ) - - return len(users_to_work_on) - - @defer.inlineCallbacks - def _populate_stats_process_rooms(self, progress, batch_size): - """ - This is a background update which regenerates statistics for rooms. - """ - if not self.stats_enabled: - yield self._end_background_update("populate_stats_process_rooms") - return 1 - - last_room_id = progress.get("last_room_id", "") - - def _get_next_batch(txn): - sql = """ - SELECT DISTINCT room_id FROM current_state_events - WHERE room_id > ? - ORDER BY room_id ASC - LIMIT ? - """ - txn.execute(sql, (last_room_id, batch_size)) - return [r for r, in txn] - - rooms_to_work_on = yield self.runInteraction( - "populate_stats_rooms_get_batch", _get_next_batch - ) - - # No more rooms -- complete the transaction. - if not rooms_to_work_on: - yield self._end_background_update("populate_stats_process_rooms") - return 1 - - for room_id in rooms_to_work_on: - yield self._calculate_and_set_initial_state_for_room(room_id) - progress["last_room_id"] = room_id - - yield self.runInteraction( - "_populate_stats_process_rooms", - self._background_update_progress_txn, - "populate_stats_process_rooms", - progress, - ) - - return len(rooms_to_work_on) - - def get_stats_positions(self): - """ - Returns the stats processor positions. - """ - return self._simple_select_one_onecol( - table="stats_incremental_position", - keyvalues={}, - retcol="stream_id", - desc="stats_incremental_position", - ) - - def update_room_state(self, room_id, fields): - """ - Args: - room_id (str) - fields (dict[str:Any]) - """ - - # For whatever reason some of the fields may contain null bytes, which - # postgres isn't a fan of, so we replace those fields with null. - for col in ( - "join_rules", - "history_visibility", - "encryption", - "name", - "topic", - "avatar", - "canonical_alias", - ): - field = fields.get(col) - if field and "\0" in field: - fields[col] = None - - return self._simple_upsert( - table="room_stats_state", - keyvalues={"room_id": room_id}, - values=fields, - desc="update_room_state", - ) - - def get_statistics_for_subject(self, stats_type, stats_id, start, size=100): - """ - Get statistics for a given subject. - - Args: - stats_type (str): The type of subject - stats_id (str): The ID of the subject (e.g. room_id or user_id) - start (int): Pagination start. Number of entries, not timestamp. - size (int): How many entries to return. - - Returns: - Deferred[list[dict]], where the dict has the keys of - ABSOLUTE_STATS_FIELDS[stats_type], and "bucket_size" and "end_ts". - """ - return self.runInteraction( - "get_statistics_for_subject", - self._get_statistics_for_subject_txn, - stats_type, - stats_id, - start, - size, - ) - - def _get_statistics_for_subject_txn( - self, txn, stats_type, stats_id, start, size=100 - ): - """ - Transaction-bound version of L{get_statistics_for_subject}. - """ - - table, id_col = TYPE_TO_TABLE[stats_type] - selected_columns = list( - ABSOLUTE_STATS_FIELDS[stats_type] + PER_SLICE_FIELDS[stats_type] - ) - - slice_list = self._simple_select_list_paginate_txn( - txn, - table + "_historical", - {id_col: stats_id}, - "end_ts", - start, - size, - retcols=selected_columns + ["bucket_size", "end_ts"], - order_direction="DESC", - ) - - return slice_list - - def get_room_stats_state(self, room_id): - """ - Returns the current room_stats_state for a room. - - Args: - room_id (str): The ID of the room to return state for. - - Returns (dict): - Dictionary containing these keys: - "name", "topic", "canonical_alias", "avatar", "join_rules", - "history_visibility" - """ - return self._simple_select_one( - "room_stats_state", - {"room_id": room_id}, - retcols=( - "name", - "topic", - "canonical_alias", - "avatar", - "join_rules", - "history_visibility", - ), - ) - - @cached() - def get_earliest_token_for_stats(self, stats_type, id): - """ - Fetch the "earliest token". This is used by the room stats delta - processor to ignore deltas that have been processed between the - start of the background task and any particular room's stats - being calculated. - - Returns: - Deferred[int] - """ - table, id_col = TYPE_TO_TABLE[stats_type] - - return self._simple_select_one_onecol( - "%s_current" % (table,), - keyvalues={id_col: id}, - retcol="completed_delta_stream_id", - allow_none=True, - ) - - def bulk_update_stats_delta(self, ts, updates, stream_id): - """Bulk update stats tables for a given stream_id and updates the stats - incremental position. - - Args: - ts (int): Current timestamp in ms - updates(dict[str, dict[str, dict[str, Counter]]]): The updates to - commit as a mapping stats_type -> stats_id -> field -> delta. - stream_id (int): Current position. - - Returns: - Deferred - """ - - def _bulk_update_stats_delta_txn(txn): - for stats_type, stats_updates in updates.items(): - for stats_id, fields in stats_updates.items(): - logger.info( - "Updating %s stats for %s: %s", stats_type, stats_id, fields - ) - self._update_stats_delta_txn( - txn, - ts=ts, - stats_type=stats_type, - stats_id=stats_id, - fields=fields, - complete_with_stream_id=stream_id, - ) - - self._simple_update_one_txn( - txn, - table="stats_incremental_position", - keyvalues={}, - updatevalues={"stream_id": stream_id}, - ) - - return self.runInteraction( - "bulk_update_stats_delta", _bulk_update_stats_delta_txn - ) - - def update_stats_delta( - self, - ts, - stats_type, - stats_id, - fields, - complete_with_stream_id, - absolute_field_overrides=None, - ): - """ - Updates the statistics for a subject, with a delta (difference/relative - change). - - Args: - ts (int): timestamp of the change - stats_type (str): "room" or "user" – the kind of subject - stats_id (str): the subject's ID (room ID or user ID) - fields (dict[str, int]): Deltas of stats values. - complete_with_stream_id (int, optional): - If supplied, converts an incomplete row into a complete row, - with the supplied stream_id marked as the stream_id where the - row was completed. - absolute_field_overrides (dict[str, int]): Current stats values - (i.e. not deltas) of absolute fields. - Does not work with per-slice fields. - """ - - return self.runInteraction( - "update_stats_delta", - self._update_stats_delta_txn, - ts, - stats_type, - stats_id, - fields, - complete_with_stream_id=complete_with_stream_id, - absolute_field_overrides=absolute_field_overrides, - ) - - def _update_stats_delta_txn( - self, - txn, - ts, - stats_type, - stats_id, - fields, - complete_with_stream_id, - absolute_field_overrides=None, - ): - if absolute_field_overrides is None: - absolute_field_overrides = {} - - table, id_col = TYPE_TO_TABLE[stats_type] - - quantised_ts = self.quantise_stats_time(int(ts)) - end_ts = quantised_ts + self.stats_bucket_size - - # Lets be paranoid and check that all the given field names are known - abs_field_names = ABSOLUTE_STATS_FIELDS[stats_type] - slice_field_names = PER_SLICE_FIELDS[stats_type] - for field in chain(fields.keys(), absolute_field_overrides.keys()): - if field not in abs_field_names and field not in slice_field_names: - # guard against potential SQL injection dodginess - raise ValueError( - "%s is not a recognised field" - " for stats type %s" % (field, stats_type) - ) - - # Per slice fields do not get added to the _current table - - # This calculates the deltas (`field = field + ?` values) - # for absolute fields, - # * defaulting to 0 if not specified - # (required for the INSERT part of upserting to work) - # * omitting overrides specified in `absolute_field_overrides` - deltas_of_absolute_fields = { - key: fields.get(key, 0) - for key in abs_field_names - if key not in absolute_field_overrides - } - - # Keep the delta stream ID field up to date - absolute_field_overrides = absolute_field_overrides.copy() - absolute_field_overrides["completed_delta_stream_id"] = complete_with_stream_id - - # first upsert the `_current` table - self._upsert_with_additive_relatives_txn( - txn=txn, - table=table + "_current", - keyvalues={id_col: stats_id}, - absolutes=absolute_field_overrides, - additive_relatives=deltas_of_absolute_fields, - ) - - per_slice_additive_relatives = { - key: fields.get(key, 0) for key in slice_field_names - } - self._upsert_copy_from_table_with_additive_relatives_txn( - txn=txn, - into_table=table + "_historical", - keyvalues={id_col: stats_id}, - extra_dst_insvalues={"bucket_size": self.stats_bucket_size}, - extra_dst_keyvalues={"end_ts": end_ts}, - additive_relatives=per_slice_additive_relatives, - src_table=table + "_current", - copy_columns=abs_field_names, - ) - - def _upsert_with_additive_relatives_txn( - self, txn, table, keyvalues, absolutes, additive_relatives - ): - """Used to update values in the stats tables. - - This is basically a slightly convoluted upsert that *adds* to any - existing rows. - - Args: - txn - table (str): Table name - keyvalues (dict[str, any]): Row-identifying key values - absolutes (dict[str, any]): Absolute (set) fields - additive_relatives (dict[str, int]): Fields that will be added onto - if existing row present. - """ - if self.database_engine.can_native_upsert: - absolute_updates = [ - "%(field)s = EXCLUDED.%(field)s" % {"field": field} - for field in absolutes.keys() - ] - - relative_updates = [ - "%(field)s = EXCLUDED.%(field)s + %(table)s.%(field)s" - % {"table": table, "field": field} - for field in additive_relatives.keys() - ] - - insert_cols = [] - qargs = [] - - for (key, val) in chain( - keyvalues.items(), absolutes.items(), additive_relatives.items() - ): - insert_cols.append(key) - qargs.append(val) - - sql = """ - INSERT INTO %(table)s (%(insert_cols_cs)s) - VALUES (%(insert_vals_qs)s) - ON CONFLICT (%(key_columns)s) DO UPDATE SET %(updates)s - """ % { - "table": table, - "insert_cols_cs": ", ".join(insert_cols), - "insert_vals_qs": ", ".join( - ["?"] * (len(keyvalues) + len(absolutes) + len(additive_relatives)) - ), - "key_columns": ", ".join(keyvalues), - "updates": ", ".join(chain(absolute_updates, relative_updates)), - } - - txn.execute(sql, qargs) - else: - self.database_engine.lock_table(txn, table) - retcols = list(chain(absolutes.keys(), additive_relatives.keys())) - current_row = self._simple_select_one_txn( - txn, table, keyvalues, retcols, allow_none=True - ) - if current_row is None: - merged_dict = {**keyvalues, **absolutes, **additive_relatives} - self._simple_insert_txn(txn, table, merged_dict) - else: - for (key, val) in additive_relatives.items(): - current_row[key] += val - current_row.update(absolutes) - self._simple_update_one_txn(txn, table, keyvalues, current_row) - - def _upsert_copy_from_table_with_additive_relatives_txn( - self, - txn, - into_table, - keyvalues, - extra_dst_keyvalues, - extra_dst_insvalues, - additive_relatives, - src_table, - copy_columns, - ): - """Updates the historic stats table with latest updates. - - This involves copying "absolute" fields from the `_current` table, and - adding relative fields to any existing values. - - Args: - txn: Transaction - into_table (str): The destination table to UPSERT the row into - keyvalues (dict[str, any]): Row-identifying key values - extra_dst_keyvalues (dict[str, any]): Additional keyvalues - for `into_table`. - extra_dst_insvalues (dict[str, any]): Additional values to insert - on new row creation for `into_table`. - additive_relatives (dict[str, any]): Fields that will be added onto - if existing row present. (Must be disjoint from copy_columns.) - src_table (str): The source table to copy from - copy_columns (iterable[str]): The list of columns to copy - """ - if self.database_engine.can_native_upsert: - ins_columns = chain( - keyvalues, - copy_columns, - additive_relatives, - extra_dst_keyvalues, - extra_dst_insvalues, - ) - sel_exprs = chain( - keyvalues, - copy_columns, - ( - "?" - for _ in chain( - additive_relatives, extra_dst_keyvalues, extra_dst_insvalues - ) - ), - ) - keyvalues_where = ("%s = ?" % f for f in keyvalues) - - sets_cc = ("%s = EXCLUDED.%s" % (f, f) for f in copy_columns) - sets_ar = ( - "%s = EXCLUDED.%s + %s.%s" % (f, f, into_table, f) - for f in additive_relatives - ) - - sql = """ - INSERT INTO %(into_table)s (%(ins_columns)s) - SELECT %(sel_exprs)s - FROM %(src_table)s - WHERE %(keyvalues_where)s - ON CONFLICT (%(keyvalues)s) - DO UPDATE SET %(sets)s - """ % { - "into_table": into_table, - "ins_columns": ", ".join(ins_columns), - "sel_exprs": ", ".join(sel_exprs), - "keyvalues_where": " AND ".join(keyvalues_where), - "src_table": src_table, - "keyvalues": ", ".join( - chain(keyvalues.keys(), extra_dst_keyvalues.keys()) - ), - "sets": ", ".join(chain(sets_cc, sets_ar)), - } - - qargs = list( - chain( - additive_relatives.values(), - extra_dst_keyvalues.values(), - extra_dst_insvalues.values(), - keyvalues.values(), - ) - ) - txn.execute(sql, qargs) - else: - self.database_engine.lock_table(txn, into_table) - src_row = self._simple_select_one_txn( - txn, src_table, keyvalues, copy_columns - ) - all_dest_keyvalues = {**keyvalues, **extra_dst_keyvalues} - dest_current_row = self._simple_select_one_txn( - txn, - into_table, - keyvalues=all_dest_keyvalues, - retcols=list(chain(additive_relatives.keys(), copy_columns)), - allow_none=True, - ) - - if dest_current_row is None: - merged_dict = { - **keyvalues, - **extra_dst_keyvalues, - **extra_dst_insvalues, - **src_row, - **additive_relatives, - } - self._simple_insert_txn(txn, into_table, merged_dict) - else: - for (key, val) in additive_relatives.items(): - src_row[key] = dest_current_row[key] + val - self._simple_update_txn(txn, into_table, all_dest_keyvalues, src_row) - - def get_changes_room_total_events_and_bytes(self, min_pos, max_pos): - """Fetches the counts of events in the given range of stream IDs. - - Args: - min_pos (int) - max_pos (int) - - Returns: - Deferred[dict[str, dict[str, int]]]: Mapping of room ID to field - changes. - """ - - return self.runInteraction( - "stats_incremental_total_events_and_bytes", - self.get_changes_room_total_events_and_bytes_txn, - min_pos, - max_pos, - ) - - def get_changes_room_total_events_and_bytes_txn(self, txn, low_pos, high_pos): - """Gets the total_events and total_event_bytes counts for rooms and - senders, in a range of stream_orderings (including backfilled events). - - Args: - txn - low_pos (int): Low stream ordering - high_pos (int): High stream ordering - - Returns: - tuple[dict[str, dict[str, int]], dict[str, dict[str, int]]]: The - room and user deltas for total_events/total_event_bytes in the - format of `stats_id` -> fields - """ - - if low_pos >= high_pos: - # nothing to do here. - return {}, {} - - if isinstance(self.database_engine, PostgresEngine): - new_bytes_expression = "OCTET_LENGTH(json)" - else: - new_bytes_expression = "LENGTH(CAST(json AS BLOB))" - - sql = """ - SELECT events.room_id, COUNT(*) AS new_events, SUM(%s) AS new_bytes - FROM events INNER JOIN event_json USING (event_id) - WHERE (? < stream_ordering AND stream_ordering <= ?) - OR (? <= stream_ordering AND stream_ordering <= ?) - GROUP BY events.room_id - """ % ( - new_bytes_expression, - ) - - txn.execute(sql, (low_pos, high_pos, -high_pos, -low_pos)) - - room_deltas = { - room_id: {"total_events": new_events, "total_event_bytes": new_bytes} - for room_id, new_events, new_bytes in txn - } - - sql = """ - SELECT events.sender, COUNT(*) AS new_events, SUM(%s) AS new_bytes - FROM events INNER JOIN event_json USING (event_id) - WHERE (? < stream_ordering AND stream_ordering <= ?) - OR (? <= stream_ordering AND stream_ordering <= ?) - GROUP BY events.sender - """ % ( - new_bytes_expression, - ) - - txn.execute(sql, (low_pos, high_pos, -high_pos, -low_pos)) - - user_deltas = { - user_id: {"total_events": new_events, "total_event_bytes": new_bytes} - for user_id, new_events, new_bytes in txn - if self.hs.is_mine_id(user_id) - } - - return room_deltas, user_deltas - - @defer.inlineCallbacks - def _calculate_and_set_initial_state_for_room(self, room_id): - """Calculate and insert an entry into room_stats_current. - - Args: - room_id (str) - - Returns: - Deferred[tuple[dict, dict, int]]: A tuple of room state, membership - counts and stream position. - """ - - def _fetch_current_state_stats(txn): - pos = self.get_room_max_stream_ordering() - - rows = self._simple_select_many_txn( - txn, - table="current_state_events", - column="type", - iterable=[ - EventTypes.Create, - EventTypes.JoinRules, - EventTypes.RoomHistoryVisibility, - EventTypes.Encryption, - EventTypes.Name, - EventTypes.Topic, - EventTypes.RoomAvatar, - EventTypes.CanonicalAlias, - ], - keyvalues={"room_id": room_id, "state_key": ""}, - retcols=["event_id"], - ) - - event_ids = [row["event_id"] for row in rows] - - txn.execute( - """ - SELECT membership, count(*) FROM current_state_events - WHERE room_id = ? AND type = 'm.room.member' - GROUP BY membership - """, - (room_id,), - ) - membership_counts = {membership: cnt for membership, cnt in txn} - - txn.execute( - """ - SELECT COALESCE(count(*), 0) FROM current_state_events - WHERE room_id = ? - """, - (room_id,), - ) - - current_state_events_count, = txn.fetchone() - - users_in_room = self.get_users_in_room_txn(txn, room_id) - - return ( - event_ids, - membership_counts, - current_state_events_count, - users_in_room, - pos, - ) - - ( - event_ids, - membership_counts, - current_state_events_count, - users_in_room, - pos, - ) = yield self.runInteraction( - "get_initial_state_for_room", _fetch_current_state_stats - ) - - state_event_map = yield self.get_events(event_ids, get_prev_content=False) - - room_state = { - "join_rules": None, - "history_visibility": None, - "encryption": None, - "name": None, - "topic": None, - "avatar": None, - "canonical_alias": None, - "is_federatable": True, - } - - for event in state_event_map.values(): - if event.type == EventTypes.JoinRules: - room_state["join_rules"] = event.content.get("join_rule") - elif event.type == EventTypes.RoomHistoryVisibility: - room_state["history_visibility"] = event.content.get( - "history_visibility" - ) - elif event.type == EventTypes.Encryption: - room_state["encryption"] = event.content.get("algorithm") - elif event.type == EventTypes.Name: - room_state["name"] = event.content.get("name") - elif event.type == EventTypes.Topic: - room_state["topic"] = event.content.get("topic") - elif event.type == EventTypes.RoomAvatar: - room_state["avatar"] = event.content.get("url") - elif event.type == EventTypes.CanonicalAlias: - room_state["canonical_alias"] = event.content.get("alias") - elif event.type == EventTypes.Create: - room_state["is_federatable"] = ( - event.content.get("m.federate", True) is True - ) - - yield self.update_room_state(room_id, room_state) - - local_users_in_room = [u for u in users_in_room if self.hs.is_mine_id(u)] - - yield self.update_stats_delta( - ts=self.clock.time_msec(), - stats_type="room", - stats_id=room_id, - fields={}, - complete_with_stream_id=pos, - absolute_field_overrides={ - "current_state_events": current_state_events_count, - "joined_members": membership_counts.get(Membership.JOIN, 0), - "invited_members": membership_counts.get(Membership.INVITE, 0), - "left_members": membership_counts.get(Membership.LEAVE, 0), - "banned_members": membership_counts.get(Membership.BAN, 0), - "local_users_in_room": len(local_users_in_room), - }, - ) - - @defer.inlineCallbacks - def _calculate_and_set_initial_state_for_user(self, user_id): - def _calculate_and_set_initial_state_for_user_txn(txn): - pos = self._get_max_stream_id_in_current_state_deltas_txn(txn) - - txn.execute( - """ - SELECT COUNT(distinct room_id) FROM current_state_events - WHERE type = 'm.room.member' AND state_key = ? - AND membership = 'join' - """, - (user_id,), - ) - count, = txn.fetchone() - return count, pos - - joined_rooms, pos = yield self.runInteraction( - "calculate_and_set_initial_state_for_user", - _calculate_and_set_initial_state_for_user_txn, - ) - - yield self.update_stats_delta( - ts=self.clock.time_msec(), - stats_type="user", - stats_id=user_id, - fields={}, - complete_with_stream_id=pos, - absolute_field_overrides={"joined_rooms": joined_rooms}, - ) diff --git a/synapse/storage/stream.py b/synapse/storage/stream.py deleted file mode 100644 index 490454f19a..0000000000 --- a/synapse/storage/stream.py +++ /dev/null @@ -1,948 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2014-2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -""" This module is responsible for getting events from the DB for pagination -and event streaming. - -The order it returns events in depend on whether we are streaming forwards or -are paginating backwards. We do this because we want to handle out of order -messages nicely, while still returning them in the correct order when we -paginate bacwards. - -This is implemented by keeping two ordering columns: stream_ordering and -topological_ordering. Stream ordering is basically insertion/received order -(except for events from backfill requests). The topological_ordering is a -weak ordering of events based on the pdu graph. - -This means that we have to have two different types of tokens, depending on -what sort order was used: - - stream tokens are of the form: "s%d", which maps directly to the column - - topological tokems: "t%d-%d", where the integers map to the topological - and stream ordering columns respectively. -""" - -import abc -import logging -from collections import namedtuple - -from six.moves import range - -from twisted.internet import defer - -from synapse.logging.context import make_deferred_yieldable, run_in_background -from synapse.storage._base import SQLBaseStore -from synapse.storage.engines import PostgresEngine -from synapse.storage.events_worker import EventsWorkerStore -from synapse.types import RoomStreamToken -from synapse.util.caches.stream_change_cache import StreamChangeCache - -logger = logging.getLogger(__name__) - - -MAX_STREAM_SIZE = 1000 - - -_STREAM_TOKEN = "stream" -_TOPOLOGICAL_TOKEN = "topological" - - -# Used as return values for pagination APIs -_EventDictReturn = namedtuple( - "_EventDictReturn", ("event_id", "topological_ordering", "stream_ordering") -) - - -def generate_pagination_where_clause( - direction, column_names, from_token, to_token, engine -): - """Creates an SQL expression to bound the columns by the pagination - tokens. - - For example creates an SQL expression like: - - (6, 7) >= (topological_ordering, stream_ordering) - AND (5, 3) < (topological_ordering, stream_ordering) - - would be generated for dir=b, from_token=(6, 7) and to_token=(5, 3). - - Note that tokens are considered to be after the row they are in, e.g. if - a row A has a token T, then we consider A to be before T. This convention - is important when figuring out inequalities for the generated SQL, and - produces the following result: - - If paginating forwards then we exclude any rows matching the from - token, but include those that match the to token. - - If paginating backwards then we include any rows matching the from - token, but include those that match the to token. - - Args: - direction (str): Whether we're paginating backwards("b") or - forwards ("f"). - column_names (tuple[str, str]): The column names to bound. Must *not* - be user defined as these get inserted directly into the SQL - statement without escapes. - from_token (tuple[int, int]|None): The start point for the pagination. - This is an exclusive minimum bound if direction is "f", and an - inclusive maximum bound if direction is "b". - to_token (tuple[int, int]|None): The endpoint point for the pagination. - This is an inclusive maximum bound if direction is "f", and an - exclusive minimum bound if direction is "b". - engine: The database engine to generate the clauses for - - Returns: - str: The sql expression - """ - assert direction in ("b", "f") - - where_clause = [] - if from_token: - where_clause.append( - _make_generic_sql_bound( - bound=">=" if direction == "b" else "<", - column_names=column_names, - values=from_token, - engine=engine, - ) - ) - - if to_token: - where_clause.append( - _make_generic_sql_bound( - bound="<" if direction == "b" else ">=", - column_names=column_names, - values=to_token, - engine=engine, - ) - ) - - return " AND ".join(where_clause) - - -def _make_generic_sql_bound(bound, column_names, values, engine): - """Create an SQL expression that bounds the given column names by the - values, e.g. create the equivalent of `(1, 2) < (col1, col2)`. - - Only works with two columns. - - Older versions of SQLite don't support that syntax so we have to expand it - out manually. - - Args: - bound (str): The comparison operator to use. One of ">", "<", ">=", - "<=", where the values are on the left and columns on the right. - names (tuple[str, str]): The column names. Must *not* be user defined - as these get inserted directly into the SQL statement without - escapes. - values (tuple[int|None, int]): The values to bound the columns by. If - the first value is None then only creates a bound on the second - column. - engine: The database engine to generate the SQL for - - Returns: - str - """ - - assert bound in (">", "<", ">=", "<=") - - name1, name2 = column_names - val1, val2 = values - - if val1 is None: - val2 = int(val2) - return "(%d %s %s)" % (val2, bound, name2) - - val1 = int(val1) - val2 = int(val2) - - if isinstance(engine, PostgresEngine): - # Postgres doesn't optimise ``(x < a) OR (x=a AND y ? AND stream_ordering <= ?" - " ORDER BY stream_ordering %s LIMIT ?" - ) % (order,) - txn.execute(sql, (room_id, from_id, to_id, limit)) - - rows = [_EventDictReturn(row[0], None, row[1]) for row in txn] - return rows - - rows = yield self.runInteraction("get_room_events_stream_for_room", f) - - ret = yield self.get_events_as_list( - [r.event_id for r in rows], get_prev_content=True - ) - - self._set_before_and_after(ret, rows, topo_order=from_id is None) - - if order.lower() == "desc": - ret.reverse() - - if rows: - key = "s%d" % min(r.stream_ordering for r in rows) - else: - # Assume we didn't get anything because there was nothing to - # get. - key = from_key - - return ret, key - - @defer.inlineCallbacks - def get_membership_changes_for_user(self, user_id, from_key, to_key): - from_id = RoomStreamToken.parse_stream_token(from_key).stream - to_id = RoomStreamToken.parse_stream_token(to_key).stream - - if from_key == to_key: - return [] - - if from_id: - has_changed = self._membership_stream_cache.has_entity_changed( - user_id, int(from_id) - ) - if not has_changed: - return [] - - def f(txn): - sql = ( - "SELECT m.event_id, stream_ordering FROM events AS e," - " room_memberships AS m" - " WHERE e.event_id = m.event_id" - " AND m.user_id = ?" - " AND e.stream_ordering > ? AND e.stream_ordering <= ?" - " ORDER BY e.stream_ordering ASC" - ) - txn.execute(sql, (user_id, from_id, to_id)) - - rows = [_EventDictReturn(row[0], None, row[1]) for row in txn] - - return rows - - rows = yield self.runInteraction("get_membership_changes_for_user", f) - - ret = yield self.get_events_as_list( - [r.event_id for r in rows], get_prev_content=True - ) - - self._set_before_and_after(ret, rows, topo_order=False) - - return ret - - @defer.inlineCallbacks - def get_recent_events_for_room(self, room_id, limit, end_token): - """Get the most recent events in the room in topological ordering. - - Args: - room_id (str) - limit (int) - end_token (str): The stream token representing now. - - Returns: - Deferred[tuple[list[FrozenEvent], str]]: Returns a list of - events and a token pointing to the start of the returned - events. - The events returned are in ascending order. - """ - - rows, token = yield self.get_recent_event_ids_for_room( - room_id, limit, end_token - ) - - logger.debug("stream before") - events = yield self.get_events_as_list( - [r.event_id for r in rows], get_prev_content=True - ) - logger.debug("stream after") - - self._set_before_and_after(events, rows) - - return (events, token) - - @defer.inlineCallbacks - def get_recent_event_ids_for_room(self, room_id, limit, end_token): - """Get the most recent events in the room in topological ordering. - - Args: - room_id (str) - limit (int) - end_token (str): The stream token representing now. - - Returns: - Deferred[tuple[list[_EventDictReturn], str]]: Returns a list of - _EventDictReturn and a token pointing to the start of the returned - events. - The events returned are in ascending order. - """ - # Allow a zero limit here, and no-op. - if limit == 0: - return [], end_token - - end_token = RoomStreamToken.parse(end_token) - - rows, token = yield self.runInteraction( - "get_recent_event_ids_for_room", - self._paginate_room_events_txn, - room_id, - from_token=end_token, - limit=limit, - ) - - # We want to return the results in ascending order. - rows.reverse() - - return rows, token - - def get_room_event_after_stream_ordering(self, room_id, stream_ordering): - """Gets details of the first event in a room at or after a stream ordering - - Args: - room_id (str): - stream_ordering (int): - - Returns: - Deferred[(int, int, str)]: - (stream ordering, topological ordering, event_id) - """ - - def _f(txn): - sql = ( - "SELECT stream_ordering, topological_ordering, event_id" - " FROM events" - " WHERE room_id = ? AND stream_ordering >= ?" - " AND NOT outlier" - " ORDER BY stream_ordering" - " LIMIT 1" - ) - txn.execute(sql, (room_id, stream_ordering)) - return txn.fetchone() - - return self.runInteraction("get_room_event_after_stream_ordering", _f) - - @defer.inlineCallbacks - def get_room_events_max_id(self, room_id=None): - """Returns the current token for rooms stream. - - By default, it returns the current global stream token. Specifying a - `room_id` causes it to return the current room specific topological - token. - """ - token = yield self.get_room_max_stream_ordering() - if room_id is None: - return "s%d" % (token,) - else: - topo = yield self.runInteraction( - "_get_max_topological_txn", self._get_max_topological_txn, room_id - ) - return "t%d-%d" % (topo, token) - - def get_stream_token_for_event(self, event_id): - """The stream token for an event - Args: - event_id(str): The id of the event to look up a stream token for. - Raises: - StoreError if the event wasn't in the database. - Returns: - A deferred "s%d" stream token. - """ - return self._simple_select_one_onecol( - table="events", keyvalues={"event_id": event_id}, retcol="stream_ordering" - ).addCallback(lambda row: "s%d" % (row,)) - - def get_topological_token_for_event(self, event_id): - """The stream token for an event - Args: - event_id(str): The id of the event to look up a stream token for. - Raises: - StoreError if the event wasn't in the database. - Returns: - A deferred "t%d-%d" topological token. - """ - return self._simple_select_one( - table="events", - keyvalues={"event_id": event_id}, - retcols=("stream_ordering", "topological_ordering"), - desc="get_topological_token_for_event", - ).addCallback( - lambda row: "t%d-%d" % (row["topological_ordering"], row["stream_ordering"]) - ) - - def get_max_topological_token(self, room_id, stream_key): - """Get the max topological token in a room before the given stream - ordering. - - Args: - room_id (str) - stream_key (int) - - Returns: - Deferred[int] - """ - sql = ( - "SELECT coalesce(max(topological_ordering), 0) FROM events" - " WHERE room_id = ? AND stream_ordering < ?" - ) - return self._execute( - "get_max_topological_token", None, sql, room_id, stream_key - ).addCallback(lambda r: r[0][0] if r else 0) - - def _get_max_topological_txn(self, txn, room_id): - txn.execute( - "SELECT MAX(topological_ordering) FROM events" " WHERE room_id = ?", - (room_id,), - ) - - rows = txn.fetchall() - return rows[0][0] if rows else 0 - - @staticmethod - def _set_before_and_after(events, rows, topo_order=True): - """Inserts ordering information to events' internal metadata from - the DB rows. - - Args: - events (list[FrozenEvent]) - rows (list[_EventDictReturn]) - topo_order (bool): Whether the events were ordered topologically - or by stream ordering. If true then all rows should have a non - null topological_ordering. - """ - for event, row in zip(events, rows): - stream = row.stream_ordering - if topo_order and row.topological_ordering: - topo = row.topological_ordering - else: - topo = None - internal = event.internal_metadata - internal.before = str(RoomStreamToken(topo, stream - 1)) - internal.after = str(RoomStreamToken(topo, stream)) - internal.order = (int(topo) if topo else 0, int(stream)) - - @defer.inlineCallbacks - def get_events_around( - self, room_id, event_id, before_limit, after_limit, event_filter=None - ): - """Retrieve events and pagination tokens around a given event in a - room. - - Args: - room_id (str) - event_id (str) - before_limit (int) - after_limit (int) - event_filter (Filter|None) - - Returns: - dict - """ - - results = yield self.runInteraction( - "get_events_around", - self._get_events_around_txn, - room_id, - event_id, - before_limit, - after_limit, - event_filter, - ) - - events_before = yield self.get_events_as_list( - [e for e in results["before"]["event_ids"]], get_prev_content=True - ) - - events_after = yield self.get_events_as_list( - [e for e in results["after"]["event_ids"]], get_prev_content=True - ) - - return { - "events_before": events_before, - "events_after": events_after, - "start": results["before"]["token"], - "end": results["after"]["token"], - } - - def _get_events_around_txn( - self, txn, room_id, event_id, before_limit, after_limit, event_filter - ): - """Retrieves event_ids and pagination tokens around a given event in a - room. - - Args: - room_id (str) - event_id (str) - before_limit (int) - after_limit (int) - event_filter (Filter|None) - - Returns: - dict - """ - - results = self._simple_select_one_txn( - txn, - "events", - keyvalues={"event_id": event_id, "room_id": room_id}, - retcols=["stream_ordering", "topological_ordering"], - ) - - # Paginating backwards includes the event at the token, but paginating - # forward doesn't. - before_token = RoomStreamToken( - results["topological_ordering"] - 1, results["stream_ordering"] - ) - - after_token = RoomStreamToken( - results["topological_ordering"], results["stream_ordering"] - ) - - rows, start_token = self._paginate_room_events_txn( - txn, - room_id, - before_token, - direction="b", - limit=before_limit, - event_filter=event_filter, - ) - events_before = [r.event_id for r in rows] - - rows, end_token = self._paginate_room_events_txn( - txn, - room_id, - after_token, - direction="f", - limit=after_limit, - event_filter=event_filter, - ) - events_after = [r.event_id for r in rows] - - return { - "before": {"event_ids": events_before, "token": start_token}, - "after": {"event_ids": events_after, "token": end_token}, - } - - @defer.inlineCallbacks - def get_all_new_events_stream(self, from_id, current_id, limit): - """Get all new events - - Returns all events with from_id < stream_ordering <= current_id. - - Args: - from_id (int): the stream_ordering of the last event we processed - current_id (int): the stream_ordering of the most recently processed event - limit (int): the maximum number of events to return - - Returns: - Deferred[Tuple[int, list[FrozenEvent]]]: A tuple of (next_id, events), where - `next_id` is the next value to pass as `from_id` (it will either be the - stream_ordering of the last returned event, or, if fewer than `limit` events - were found, `current_id`. - """ - - def get_all_new_events_stream_txn(txn): - sql = ( - "SELECT e.stream_ordering, e.event_id" - " FROM events AS e" - " WHERE" - " ? < e.stream_ordering AND e.stream_ordering <= ?" - " ORDER BY e.stream_ordering ASC" - " LIMIT ?" - ) - - txn.execute(sql, (from_id, current_id, limit)) - rows = txn.fetchall() - - upper_bound = current_id - if len(rows) == limit: - upper_bound = rows[-1][0] - - return upper_bound, [row[1] for row in rows] - - upper_bound, event_ids = yield self.runInteraction( - "get_all_new_events_stream", get_all_new_events_stream_txn - ) - - events = yield self.get_events_as_list(event_ids) - - return upper_bound, events - - def get_federation_out_pos(self, typ): - return self._simple_select_one_onecol( - table="federation_stream_position", - retcol="stream_id", - keyvalues={"type": typ}, - desc="get_federation_out_pos", - ) - - def update_federation_out_pos(self, typ, stream_id): - return self._simple_update_one( - table="federation_stream_position", - keyvalues={"type": typ}, - updatevalues={"stream_id": stream_id}, - desc="update_federation_out_pos", - ) - - def has_room_changed_since(self, room_id, stream_id): - return self._events_stream_cache.has_entity_changed(room_id, stream_id) - - def _paginate_room_events_txn( - self, - txn, - room_id, - from_token, - to_token=None, - direction="b", - limit=-1, - event_filter=None, - ): - """Returns list of events before or after a given token. - - Args: - txn - room_id (str) - from_token (RoomStreamToken): The token used to stream from - to_token (RoomStreamToken|None): A token which if given limits the - results to only those before - direction(char): Either 'b' or 'f' to indicate whether we are - paginating forwards or backwards from `from_key`. - limit (int): The maximum number of events to return. - event_filter (Filter|None): If provided filters the events to - those that match the filter. - - Returns: - Deferred[tuple[list[_EventDictReturn], str]]: Returns the results - as a list of _EventDictReturn and a token that points to the end - of the result set. If no events are returned then the end of the - stream has been reached (i.e. there are no events between - `from_token` and `to_token`), or `limit` is zero. - """ - - assert int(limit) >= 0 - - # Tokens really represent positions between elements, but we use - # the convention of pointing to the event before the gap. Hence - # we have a bit of asymmetry when it comes to equalities. - args = [False, room_id] - if direction == "b": - order = "DESC" - else: - order = "ASC" - - bounds = generate_pagination_where_clause( - direction=direction, - column_names=("topological_ordering", "stream_ordering"), - from_token=from_token, - to_token=to_token, - engine=self.database_engine, - ) - - filter_clause, filter_args = filter_to_clause(event_filter) - - if filter_clause: - bounds += " AND " + filter_clause - args.extend(filter_args) - - args.append(int(limit)) - - sql = ( - "SELECT event_id, topological_ordering, stream_ordering" - " FROM events" - " WHERE outlier = ? AND room_id = ? AND %(bounds)s" - " ORDER BY topological_ordering %(order)s," - " stream_ordering %(order)s LIMIT ?" - ) % {"bounds": bounds, "order": order} - - txn.execute(sql, args) - - rows = [_EventDictReturn(row[0], row[1], row[2]) for row in txn] - - if rows: - topo = rows[-1].topological_ordering - toke = rows[-1].stream_ordering - if direction == "b": - # Tokens are positions between events. - # This token points *after* the last event in the chunk. - # We need it to point to the event before it in the chunk - # when we are going backwards so we subtract one from the - # stream part. - toke -= 1 - next_token = RoomStreamToken(topo, toke) - else: - # TODO (erikj): We should work out what to do here instead. - next_token = to_token if to_token else from_token - - return rows, str(next_token) - - @defer.inlineCallbacks - def paginate_room_events( - self, room_id, from_key, to_key=None, direction="b", limit=-1, event_filter=None - ): - """Returns list of events before or after a given token. - - Args: - room_id (str) - from_key (str): The token used to stream from - to_key (str|None): A token which if given limits the results to - only those before - direction(char): Either 'b' or 'f' to indicate whether we are - paginating forwards or backwards from `from_key`. - limit (int): The maximum number of events to return. - event_filter (Filter|None): If provided filters the events to - those that match the filter. - - Returns: - tuple[list[FrozenEvent], str]: Returns the results as a list of - events and a token that points to the end of the result set. If no - events are returned then the end of the stream has been reached - (i.e. there are no events between `from_key` and `to_key`). - """ - - from_key = RoomStreamToken.parse(from_key) - if to_key: - to_key = RoomStreamToken.parse(to_key) - - rows, token = yield self.runInteraction( - "paginate_room_events", - self._paginate_room_events_txn, - room_id, - from_key, - to_key, - direction, - limit, - event_filter, - ) - - events = yield self.get_events_as_list( - [r.event_id for r in rows], get_prev_content=True - ) - - self._set_before_and_after(events, rows) - - return (events, token) - - -class StreamStore(StreamWorkerStore): - def get_room_max_stream_ordering(self): - return self._stream_id_gen.get_current_token() - - def get_room_min_stream_ordering(self): - return self._backfill_id_gen.get_current_token() diff --git a/synapse/storage/tags.py b/synapse/storage/tags.py deleted file mode 100644 index 20dd6bd53d..0000000000 --- a/synapse/storage/tags.py +++ /dev/null @@ -1,265 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2014-2016 OpenMarket Ltd -# Copyright 2018 New Vector Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import logging - -from six.moves import range - -from canonicaljson import json - -from twisted.internet import defer - -from synapse.storage.account_data import AccountDataWorkerStore -from synapse.util.caches.descriptors import cached - -logger = logging.getLogger(__name__) - - -class TagsWorkerStore(AccountDataWorkerStore): - @cached() - def get_tags_for_user(self, user_id): - """Get all the tags for a user. - - - Args: - user_id(str): The user to get the tags for. - Returns: - A deferred dict mapping from room_id strings to dicts mapping from - tag strings to tag content. - """ - - deferred = self._simple_select_list( - "room_tags", {"user_id": user_id}, ["room_id", "tag", "content"] - ) - - @deferred.addCallback - def tags_by_room(rows): - tags_by_room = {} - for row in rows: - room_tags = tags_by_room.setdefault(row["room_id"], {}) - room_tags[row["tag"]] = json.loads(row["content"]) - return tags_by_room - - return deferred - - @defer.inlineCallbacks - def get_all_updated_tags(self, last_id, current_id, limit): - """Get all the client tags that have changed on the server - Args: - last_id(int): The position to fetch from. - current_id(int): The position to fetch up to. - Returns: - A deferred list of tuples of stream_id int, user_id string, - room_id string, tag string and content string. - """ - if last_id == current_id: - return [] - - def get_all_updated_tags_txn(txn): - sql = ( - "SELECT stream_id, user_id, room_id" - " FROM room_tags_revisions as r" - " WHERE ? < stream_id AND stream_id <= ?" - " ORDER BY stream_id ASC LIMIT ?" - ) - txn.execute(sql, (last_id, current_id, limit)) - return txn.fetchall() - - tag_ids = yield self.runInteraction( - "get_all_updated_tags", get_all_updated_tags_txn - ) - - def get_tag_content(txn, tag_ids): - sql = ( - "SELECT tag, content" " FROM room_tags" " WHERE user_id=? AND room_id=?" - ) - results = [] - for stream_id, user_id, room_id in tag_ids: - txn.execute(sql, (user_id, room_id)) - tags = [] - for tag, content in txn: - tags.append(json.dumps(tag) + ":" + content) - tag_json = "{" + ",".join(tags) + "}" - results.append((stream_id, user_id, room_id, tag_json)) - - return results - - batch_size = 50 - results = [] - for i in range(0, len(tag_ids), batch_size): - tags = yield self.runInteraction( - "get_all_updated_tag_content", - get_tag_content, - tag_ids[i : i + batch_size], - ) - results.extend(tags) - - return results - - @defer.inlineCallbacks - def get_updated_tags(self, user_id, stream_id): - """Get all the tags for the rooms where the tags have changed since the - given version - - Args: - user_id(str): The user to get the tags for. - stream_id(int): The earliest update to get for the user. - Returns: - A deferred dict mapping from room_id strings to lists of tag - strings for all the rooms that changed since the stream_id token. - """ - - def get_updated_tags_txn(txn): - sql = ( - "SELECT room_id from room_tags_revisions" - " WHERE user_id = ? AND stream_id > ?" - ) - txn.execute(sql, (user_id, stream_id)) - room_ids = [row[0] for row in txn] - return room_ids - - changed = self._account_data_stream_cache.has_entity_changed( - user_id, int(stream_id) - ) - if not changed: - return {} - - room_ids = yield self.runInteraction("get_updated_tags", get_updated_tags_txn) - - results = {} - if room_ids: - tags_by_room = yield self.get_tags_for_user(user_id) - for room_id in room_ids: - results[room_id] = tags_by_room.get(room_id, {}) - - return results - - def get_tags_for_room(self, user_id, room_id): - """Get all the tags for the given room - Args: - user_id(str): The user to get tags for - room_id(str): The room to get tags for - Returns: - A deferred list of string tags. - """ - return self._simple_select_list( - table="room_tags", - keyvalues={"user_id": user_id, "room_id": room_id}, - retcols=("tag", "content"), - desc="get_tags_for_room", - ).addCallback( - lambda rows: {row["tag"]: json.loads(row["content"]) for row in rows} - ) - - -class TagsStore(TagsWorkerStore): - @defer.inlineCallbacks - def add_tag_to_room(self, user_id, room_id, tag, content): - """Add a tag to a room for a user. - Args: - user_id(str): The user to add a tag for. - room_id(str): The room to add a tag for. - tag(str): The tag name to add. - content(dict): A json object to associate with the tag. - Returns: - A deferred that completes once the tag has been added. - """ - content_json = json.dumps(content) - - def add_tag_txn(txn, next_id): - self._simple_upsert_txn( - txn, - table="room_tags", - keyvalues={"user_id": user_id, "room_id": room_id, "tag": tag}, - values={"content": content_json}, - ) - self._update_revision_txn(txn, user_id, room_id, next_id) - - with self._account_data_id_gen.get_next() as next_id: - yield self.runInteraction("add_tag", add_tag_txn, next_id) - - self.get_tags_for_user.invalidate((user_id,)) - - result = self._account_data_id_gen.get_current_token() - return result - - @defer.inlineCallbacks - def remove_tag_from_room(self, user_id, room_id, tag): - """Remove a tag from a room for a user. - Returns: - A deferred that completes once the tag has been removed - """ - - def remove_tag_txn(txn, next_id): - sql = ( - "DELETE FROM room_tags " - " WHERE user_id = ? AND room_id = ? AND tag = ?" - ) - txn.execute(sql, (user_id, room_id, tag)) - self._update_revision_txn(txn, user_id, room_id, next_id) - - with self._account_data_id_gen.get_next() as next_id: - yield self.runInteraction("remove_tag", remove_tag_txn, next_id) - - self.get_tags_for_user.invalidate((user_id,)) - - result = self._account_data_id_gen.get_current_token() - return result - - def _update_revision_txn(self, txn, user_id, room_id, next_id): - """Update the latest revision of the tags for the given user and room. - - Args: - txn: The database cursor - user_id(str): The ID of the user. - room_id(str): The ID of the room. - next_id(int): The the revision to advance to. - """ - - txn.call_after( - self._account_data_stream_cache.entity_has_changed, user_id, next_id - ) - - update_max_id_sql = ( - "UPDATE account_data_max_stream_id" - " SET stream_id = ?" - " WHERE stream_id < ?" - ) - txn.execute(update_max_id_sql, (next_id, next_id)) - - update_sql = ( - "UPDATE room_tags_revisions" - " SET stream_id = ?" - " WHERE user_id = ?" - " AND room_id = ?" - ) - txn.execute(update_sql, (next_id, user_id, room_id)) - - if txn.rowcount == 0: - insert_sql = ( - "INSERT INTO room_tags_revisions (user_id, room_id, stream_id)" - " VALUES (?, ?, ?)" - ) - try: - txn.execute(insert_sql, (user_id, room_id, next_id)) - except self.database_engine.module.IntegrityError: - # Ignore insertion errors. It doesn't matter if the row wasn't - # inserted because if two updates happend concurrently the one - # with the higher stream_id will not be reported to a client - # unless the previous update has completed. It doesn't matter - # which stream_id ends up in the table, as long as it is higher - # than the id that the client has. - pass diff --git a/synapse/storage/transactions.py b/synapse/storage/transactions.py deleted file mode 100644 index 289c117396..0000000000 --- a/synapse/storage/transactions.py +++ /dev/null @@ -1,274 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2014-2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import logging -from collections import namedtuple - -import six - -from canonicaljson import encode_canonical_json - -from twisted.internet import defer - -from synapse.metrics.background_process_metrics import run_as_background_process -from synapse.util.caches.expiringcache import ExpiringCache - -from ._base import SQLBaseStore, db_to_json - -# py2 sqlite has buffer hardcoded as only binary type, so we must use it, -# despite being deprecated and removed in favor of memoryview -if six.PY2: - db_binary_type = six.moves.builtins.buffer -else: - db_binary_type = memoryview - -logger = logging.getLogger(__name__) - - -_TransactionRow = namedtuple( - "_TransactionRow", - ("id", "transaction_id", "destination", "ts", "response_code", "response_json"), -) - -_UpdateTransactionRow = namedtuple( - "_TransactionRow", ("response_code", "response_json") -) - -SENTINEL = object() - - -class TransactionStore(SQLBaseStore): - """A collection of queries for handling PDUs. - """ - - def __init__(self, db_conn, hs): - super(TransactionStore, self).__init__(db_conn, hs) - - self._clock.looping_call(self._start_cleanup_transactions, 30 * 60 * 1000) - - self._destination_retry_cache = ExpiringCache( - cache_name="get_destination_retry_timings", - clock=self._clock, - expiry_ms=5 * 60 * 1000, - ) - - def get_received_txn_response(self, transaction_id, origin): - """For an incoming transaction from a given origin, check if we have - already responded to it. If so, return the response code and response - body (as a dict). - - Args: - transaction_id (str) - origin(str) - - Returns: - tuple: None if we have not previously responded to - this transaction or a 2-tuple of (int, dict) - """ - - return self.runInteraction( - "get_received_txn_response", - self._get_received_txn_response, - transaction_id, - origin, - ) - - def _get_received_txn_response(self, txn, transaction_id, origin): - result = self._simple_select_one_txn( - txn, - table="received_transactions", - keyvalues={"transaction_id": transaction_id, "origin": origin}, - retcols=( - "transaction_id", - "origin", - "ts", - "response_code", - "response_json", - "has_been_referenced", - ), - allow_none=True, - ) - - if result and result["response_code"]: - return result["response_code"], db_to_json(result["response_json"]) - - else: - return None - - def set_received_txn_response(self, transaction_id, origin, code, response_dict): - """Persist the response we returened for an incoming transaction, and - should return for subsequent transactions with the same transaction_id - and origin. - - Args: - txn - transaction_id (str) - origin (str) - code (int) - response_json (str) - """ - - return self._simple_insert( - table="received_transactions", - values={ - "transaction_id": transaction_id, - "origin": origin, - "response_code": code, - "response_json": db_binary_type(encode_canonical_json(response_dict)), - "ts": self._clock.time_msec(), - }, - or_ignore=True, - desc="set_received_txn_response", - ) - - @defer.inlineCallbacks - def get_destination_retry_timings(self, destination): - """Gets the current retry timings (if any) for a given destination. - - Args: - destination (str) - - Returns: - None if not retrying - Otherwise a dict for the retry scheme - """ - - result = self._destination_retry_cache.get(destination, SENTINEL) - if result is not SENTINEL: - return result - - result = yield self.runInteraction( - "get_destination_retry_timings", - self._get_destination_retry_timings, - destination, - ) - - # We don't hugely care about race conditions between getting and - # invalidating the cache, since we time out fairly quickly anyway. - self._destination_retry_cache[destination] = result - return result - - def _get_destination_retry_timings(self, txn, destination): - result = self._simple_select_one_txn( - txn, - table="destinations", - keyvalues={"destination": destination}, - retcols=("destination", "failure_ts", "retry_last_ts", "retry_interval"), - allow_none=True, - ) - - if result and result["retry_last_ts"] > 0: - return result - else: - return None - - def set_destination_retry_timings( - self, destination, failure_ts, retry_last_ts, retry_interval - ): - """Sets the current retry timings for a given destination. - Both timings should be zero if retrying is no longer occuring. - - Args: - destination (str) - failure_ts (int|None) - when the server started failing (ms since epoch) - retry_last_ts (int) - time of last retry attempt in unix epoch ms - retry_interval (int) - how long until next retry in ms - """ - - self._destination_retry_cache.pop(destination, None) - return self.runInteraction( - "set_destination_retry_timings", - self._set_destination_retry_timings, - destination, - failure_ts, - retry_last_ts, - retry_interval, - ) - - def _set_destination_retry_timings( - self, txn, destination, failure_ts, retry_last_ts, retry_interval - ): - - if self.database_engine.can_native_upsert: - # Upsert retry time interval if retry_interval is zero (i.e. we're - # resetting it) or greater than the existing retry interval. - - sql = """ - INSERT INTO destinations ( - destination, failure_ts, retry_last_ts, retry_interval - ) - VALUES (?, ?, ?, ?) - ON CONFLICT (destination) DO UPDATE SET - failure_ts = EXCLUDED.failure_ts, - retry_last_ts = EXCLUDED.retry_last_ts, - retry_interval = EXCLUDED.retry_interval - WHERE - EXCLUDED.retry_interval = 0 - OR destinations.retry_interval < EXCLUDED.retry_interval - """ - - txn.execute(sql, (destination, failure_ts, retry_last_ts, retry_interval)) - - return - - self.database_engine.lock_table(txn, "destinations") - - # We need to be careful here as the data may have changed from under us - # due to a worker setting the timings. - - prev_row = self._simple_select_one_txn( - txn, - table="destinations", - keyvalues={"destination": destination}, - retcols=("failure_ts", "retry_last_ts", "retry_interval"), - allow_none=True, - ) - - if not prev_row: - self._simple_insert_txn( - txn, - table="destinations", - values={ - "destination": destination, - "failure_ts": failure_ts, - "retry_last_ts": retry_last_ts, - "retry_interval": retry_interval, - }, - ) - elif retry_interval == 0 or prev_row["retry_interval"] < retry_interval: - self._simple_update_one_txn( - txn, - "destinations", - keyvalues={"destination": destination}, - updatevalues={ - "failure_ts": failure_ts, - "retry_last_ts": retry_last_ts, - "retry_interval": retry_interval, - }, - ) - - def _start_cleanup_transactions(self): - return run_as_background_process( - "cleanup_transactions", self._cleanup_transactions - ) - - def _cleanup_transactions(self): - now = self._clock.time_msec() - month_ago = now - 30 * 24 * 60 * 60 * 1000 - - def _cleanup_transactions_txn(txn): - txn.execute("DELETE FROM received_transactions WHERE ts < ?", (month_ago,)) - - return self.runInteraction("_cleanup_transactions", _cleanup_transactions_txn) diff --git a/synapse/storage/user_directory.py b/synapse/storage/user_directory.py deleted file mode 100644 index 1b1e4751b9..0000000000 --- a/synapse/storage/user_directory.py +++ /dev/null @@ -1,827 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2017 Vector Creations Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import logging -import re - -from twisted.internet import defer - -from synapse.api.constants import EventTypes, JoinRules -from synapse.storage.background_updates import BackgroundUpdateStore -from synapse.storage.engines import PostgresEngine, Sqlite3Engine -from synapse.storage.state import StateFilter -from synapse.storage.state_deltas import StateDeltasStore -from synapse.types import get_domain_from_id, get_localpart_from_id -from synapse.util.caches.descriptors import cached - -logger = logging.getLogger(__name__) - - -TEMP_TABLE = "_temp_populate_user_directory" - - -class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore): - - # How many records do we calculate before sending it to - # add_users_who_share_private_rooms? - SHARE_PRIVATE_WORKING_SET = 500 - - def __init__(self, db_conn, hs): - super(UserDirectoryBackgroundUpdateStore, self).__init__(db_conn, hs) - - self.server_name = hs.hostname - - self.register_background_update_handler( - "populate_user_directory_createtables", - self._populate_user_directory_createtables, - ) - self.register_background_update_handler( - "populate_user_directory_process_rooms", - self._populate_user_directory_process_rooms, - ) - self.register_background_update_handler( - "populate_user_directory_process_users", - self._populate_user_directory_process_users, - ) - self.register_background_update_handler( - "populate_user_directory_cleanup", self._populate_user_directory_cleanup - ) - - @defer.inlineCallbacks - def _populate_user_directory_createtables(self, progress, batch_size): - - # Get all the rooms that we want to process. - def _make_staging_area(txn): - sql = ( - "CREATE TABLE IF NOT EXISTS " - + TEMP_TABLE - + "_rooms(room_id TEXT NOT NULL, events BIGINT NOT NULL)" - ) - txn.execute(sql) - - sql = ( - "CREATE TABLE IF NOT EXISTS " - + TEMP_TABLE - + "_position(position TEXT NOT NULL)" - ) - txn.execute(sql) - - # Get rooms we want to process from the database - sql = """ - SELECT room_id, count(*) FROM current_state_events - GROUP BY room_id - """ - txn.execute(sql) - rooms = [{"room_id": x[0], "events": x[1]} for x in txn.fetchall()] - self._simple_insert_many_txn(txn, TEMP_TABLE + "_rooms", rooms) - del rooms - - # If search all users is on, get all the users we want to add. - if self.hs.config.user_directory_search_all_users: - sql = ( - "CREATE TABLE IF NOT EXISTS " - + TEMP_TABLE - + "_users(user_id TEXT NOT NULL)" - ) - txn.execute(sql) - - txn.execute("SELECT name FROM users") - users = [{"user_id": x[0]} for x in txn.fetchall()] - - self._simple_insert_many_txn(txn, TEMP_TABLE + "_users", users) - - new_pos = yield self.get_max_stream_id_in_current_state_deltas() - yield self.runInteraction( - "populate_user_directory_temp_build", _make_staging_area - ) - yield self._simple_insert(TEMP_TABLE + "_position", {"position": new_pos}) - - yield self._end_background_update("populate_user_directory_createtables") - return 1 - - @defer.inlineCallbacks - def _populate_user_directory_cleanup(self, progress, batch_size): - """ - Update the user directory stream position, then clean up the old tables. - """ - position = yield self._simple_select_one_onecol( - TEMP_TABLE + "_position", None, "position" - ) - yield self.update_user_directory_stream_pos(position) - - def _delete_staging_area(txn): - txn.execute("DROP TABLE IF EXISTS " + TEMP_TABLE + "_rooms") - txn.execute("DROP TABLE IF EXISTS " + TEMP_TABLE + "_users") - txn.execute("DROP TABLE IF EXISTS " + TEMP_TABLE + "_position") - - yield self.runInteraction( - "populate_user_directory_cleanup", _delete_staging_area - ) - - yield self._end_background_update("populate_user_directory_cleanup") - return 1 - - @defer.inlineCallbacks - def _populate_user_directory_process_rooms(self, progress, batch_size): - """ - Args: - progress (dict) - batch_size (int): Maximum number of state events to process - per cycle. - """ - state = self.hs.get_state_handler() - - # If we don't have progress filed, delete everything. - if not progress: - yield self.delete_all_from_user_dir() - - def _get_next_batch(txn): - # Only fetch 250 rooms, so we don't fetch too many at once, even - # if those 250 rooms have less than batch_size state events. - sql = """ - SELECT room_id, events FROM %s - ORDER BY events DESC - LIMIT 250 - """ % ( - TEMP_TABLE + "_rooms", - ) - txn.execute(sql) - rooms_to_work_on = txn.fetchall() - - if not rooms_to_work_on: - return None - - # Get how many are left to process, so we can give status on how - # far we are in processing - txn.execute("SELECT COUNT(*) FROM " + TEMP_TABLE + "_rooms") - progress["remaining"] = txn.fetchone()[0] - - return rooms_to_work_on - - rooms_to_work_on = yield self.runInteraction( - "populate_user_directory_temp_read", _get_next_batch - ) - - # No more rooms -- complete the transaction. - if not rooms_to_work_on: - yield self._end_background_update("populate_user_directory_process_rooms") - return 1 - - logger.info( - "Processing the next %d rooms of %d remaining" - % (len(rooms_to_work_on), progress["remaining"]) - ) - - processed_event_count = 0 - - for room_id, event_count in rooms_to_work_on: - is_in_room = yield self.is_host_joined(room_id, self.server_name) - - if is_in_room: - is_public = yield self.is_room_world_readable_or_publicly_joinable( - room_id - ) - - users_with_profile = yield state.get_current_users_in_room(room_id) - user_ids = set(users_with_profile) - - # Update each user in the user directory. - for user_id, profile in users_with_profile.items(): - yield self.update_profile_in_user_dir( - user_id, profile.display_name, profile.avatar_url - ) - - to_insert = set() - - if is_public: - for user_id in user_ids: - if self.get_if_app_services_interested_in_user(user_id): - continue - - to_insert.add(user_id) - - if to_insert: - yield self.add_users_in_public_rooms(room_id, to_insert) - to_insert.clear() - else: - for user_id in user_ids: - if not self.hs.is_mine_id(user_id): - continue - - if self.get_if_app_services_interested_in_user(user_id): - continue - - for other_user_id in user_ids: - if user_id == other_user_id: - continue - - user_set = (user_id, other_user_id) - to_insert.add(user_set) - - # If it gets too big, stop and write to the database - # to prevent storing too much in RAM. - if len(to_insert) >= self.SHARE_PRIVATE_WORKING_SET: - yield self.add_users_who_share_private_room( - room_id, to_insert - ) - to_insert.clear() - - if to_insert: - yield self.add_users_who_share_private_room(room_id, to_insert) - to_insert.clear() - - # We've finished a room. Delete it from the table. - yield self._simple_delete_one(TEMP_TABLE + "_rooms", {"room_id": room_id}) - # Update the remaining counter. - progress["remaining"] -= 1 - yield self.runInteraction( - "populate_user_directory", - self._background_update_progress_txn, - "populate_user_directory_process_rooms", - progress, - ) - - processed_event_count += event_count - - if processed_event_count > batch_size: - # Don't process any more rooms, we've hit our batch size. - return processed_event_count - - return processed_event_count - - @defer.inlineCallbacks - def _populate_user_directory_process_users(self, progress, batch_size): - """ - If search_all_users is enabled, add all of the users to the user directory. - """ - if not self.hs.config.user_directory_search_all_users: - yield self._end_background_update("populate_user_directory_process_users") - return 1 - - def _get_next_batch(txn): - sql = "SELECT user_id FROM %s LIMIT %s" % ( - TEMP_TABLE + "_users", - str(batch_size), - ) - txn.execute(sql) - users_to_work_on = txn.fetchall() - - if not users_to_work_on: - return None - - users_to_work_on = [x[0] for x in users_to_work_on] - - # Get how many are left to process, so we can give status on how - # far we are in processing - sql = "SELECT COUNT(*) FROM " + TEMP_TABLE + "_users" - txn.execute(sql) - progress["remaining"] = txn.fetchone()[0] - - return users_to_work_on - - users_to_work_on = yield self.runInteraction( - "populate_user_directory_temp_read", _get_next_batch - ) - - # No more users -- complete the transaction. - if not users_to_work_on: - yield self._end_background_update("populate_user_directory_process_users") - return 1 - - logger.info( - "Processing the next %d users of %d remaining" - % (len(users_to_work_on), progress["remaining"]) - ) - - for user_id in users_to_work_on: - profile = yield self.get_profileinfo(get_localpart_from_id(user_id)) - yield self.update_profile_in_user_dir( - user_id, profile.display_name, profile.avatar_url - ) - - # We've finished processing a user. Delete it from the table. - yield self._simple_delete_one(TEMP_TABLE + "_users", {"user_id": user_id}) - # Update the remaining counter. - progress["remaining"] -= 1 - yield self.runInteraction( - "populate_user_directory", - self._background_update_progress_txn, - "populate_user_directory_process_users", - progress, - ) - - return len(users_to_work_on) - - @defer.inlineCallbacks - def is_room_world_readable_or_publicly_joinable(self, room_id): - """Check if the room is either world_readable or publically joinable - """ - - # Create a state filter that only queries join and history state event - types_to_filter = ( - (EventTypes.JoinRules, ""), - (EventTypes.RoomHistoryVisibility, ""), - ) - - current_state_ids = yield self.get_filtered_current_state_ids( - room_id, StateFilter.from_types(types_to_filter) - ) - - join_rules_id = current_state_ids.get((EventTypes.JoinRules, "")) - if join_rules_id: - join_rule_ev = yield self.get_event(join_rules_id, allow_none=True) - if join_rule_ev: - if join_rule_ev.content.get("join_rule") == JoinRules.PUBLIC: - return True - - hist_vis_id = current_state_ids.get((EventTypes.RoomHistoryVisibility, "")) - if hist_vis_id: - hist_vis_ev = yield self.get_event(hist_vis_id, allow_none=True) - if hist_vis_ev: - if hist_vis_ev.content.get("history_visibility") == "world_readable": - return True - - return False - - def update_profile_in_user_dir(self, user_id, display_name, avatar_url): - """ - Update or add a user's profile in the user directory. - """ - - def _update_profile_in_user_dir_txn(txn): - new_entry = self._simple_upsert_txn( - txn, - table="user_directory", - keyvalues={"user_id": user_id}, - values={"display_name": display_name, "avatar_url": avatar_url}, - lock=False, # We're only inserter - ) - - if isinstance(self.database_engine, PostgresEngine): - # We weight the localpart most highly, then display name and finally - # server name - if self.database_engine.can_native_upsert: - sql = """ - INSERT INTO user_directory_search(user_id, vector) - VALUES (?, - setweight(to_tsvector('english', ?), 'A') - || setweight(to_tsvector('english', ?), 'D') - || setweight(to_tsvector('english', COALESCE(?, '')), 'B') - ) ON CONFLICT (user_id) DO UPDATE SET vector=EXCLUDED.vector - """ - txn.execute( - sql, - ( - user_id, - get_localpart_from_id(user_id), - get_domain_from_id(user_id), - display_name, - ), - ) - else: - # TODO: Remove this code after we've bumped the minimum version - # of postgres to always support upserts, so we can get rid of - # `new_entry` usage - if new_entry is True: - sql = """ - INSERT INTO user_directory_search(user_id, vector) - VALUES (?, - setweight(to_tsvector('english', ?), 'A') - || setweight(to_tsvector('english', ?), 'D') - || setweight(to_tsvector('english', COALESCE(?, '')), 'B') - ) - """ - txn.execute( - sql, - ( - user_id, - get_localpart_from_id(user_id), - get_domain_from_id(user_id), - display_name, - ), - ) - elif new_entry is False: - sql = """ - UPDATE user_directory_search - SET vector = setweight(to_tsvector('english', ?), 'A') - || setweight(to_tsvector('english', ?), 'D') - || setweight(to_tsvector('english', COALESCE(?, '')), 'B') - WHERE user_id = ? - """ - txn.execute( - sql, - ( - get_localpart_from_id(user_id), - get_domain_from_id(user_id), - display_name, - user_id, - ), - ) - else: - raise RuntimeError( - "upsert returned None when 'can_native_upsert' is False" - ) - elif isinstance(self.database_engine, Sqlite3Engine): - value = "%s %s" % (user_id, display_name) if display_name else user_id - self._simple_upsert_txn( - txn, - table="user_directory_search", - keyvalues={"user_id": user_id}, - values={"value": value}, - lock=False, # We're only inserter - ) - else: - # This should be unreachable. - raise Exception("Unrecognized database engine") - - txn.call_after(self.get_user_in_directory.invalidate, (user_id,)) - - return self.runInteraction( - "update_profile_in_user_dir", _update_profile_in_user_dir_txn - ) - - def add_users_who_share_private_room(self, room_id, user_id_tuples): - """Insert entries into the users_who_share_private_rooms table. The first - user should be a local user. - - Args: - room_id (str) - user_id_tuples([(str, str)]): iterable of 2-tuple of user IDs. - """ - - def _add_users_who_share_room_txn(txn): - self._simple_upsert_many_txn( - txn, - table="users_who_share_private_rooms", - key_names=["user_id", "other_user_id", "room_id"], - key_values=[ - (user_id, other_user_id, room_id) - for user_id, other_user_id in user_id_tuples - ], - value_names=(), - value_values=None, - ) - - return self.runInteraction( - "add_users_who_share_room", _add_users_who_share_room_txn - ) - - def add_users_in_public_rooms(self, room_id, user_ids): - """Insert entries into the users_who_share_private_rooms table. The first - user should be a local user. - - Args: - room_id (str) - user_ids (list[str]) - """ - - def _add_users_in_public_rooms_txn(txn): - - self._simple_upsert_many_txn( - txn, - table="users_in_public_rooms", - key_names=["user_id", "room_id"], - key_values=[(user_id, room_id) for user_id in user_ids], - value_names=(), - value_values=None, - ) - - return self.runInteraction( - "add_users_in_public_rooms", _add_users_in_public_rooms_txn - ) - - def delete_all_from_user_dir(self): - """Delete the entire user directory - """ - - def _delete_all_from_user_dir_txn(txn): - txn.execute("DELETE FROM user_directory") - txn.execute("DELETE FROM user_directory_search") - txn.execute("DELETE FROM users_in_public_rooms") - txn.execute("DELETE FROM users_who_share_private_rooms") - txn.call_after(self.get_user_in_directory.invalidate_all) - - return self.runInteraction( - "delete_all_from_user_dir", _delete_all_from_user_dir_txn - ) - - @cached() - def get_user_in_directory(self, user_id): - return self._simple_select_one( - table="user_directory", - keyvalues={"user_id": user_id}, - retcols=("display_name", "avatar_url"), - allow_none=True, - desc="get_user_in_directory", - ) - - def update_user_directory_stream_pos(self, stream_id): - return self._simple_update_one( - table="user_directory_stream_pos", - keyvalues={}, - updatevalues={"stream_id": stream_id}, - desc="update_user_directory_stream_pos", - ) - - -class UserDirectoryStore(UserDirectoryBackgroundUpdateStore): - - # How many records do we calculate before sending it to - # add_users_who_share_private_rooms? - SHARE_PRIVATE_WORKING_SET = 500 - - def __init__(self, db_conn, hs): - super(UserDirectoryStore, self).__init__(db_conn, hs) - - def remove_from_user_dir(self, user_id): - def _remove_from_user_dir_txn(txn): - self._simple_delete_txn( - txn, table="user_directory", keyvalues={"user_id": user_id} - ) - self._simple_delete_txn( - txn, table="user_directory_search", keyvalues={"user_id": user_id} - ) - self._simple_delete_txn( - txn, table="users_in_public_rooms", keyvalues={"user_id": user_id} - ) - self._simple_delete_txn( - txn, - table="users_who_share_private_rooms", - keyvalues={"user_id": user_id}, - ) - self._simple_delete_txn( - txn, - table="users_who_share_private_rooms", - keyvalues={"other_user_id": user_id}, - ) - txn.call_after(self.get_user_in_directory.invalidate, (user_id,)) - - return self.runInteraction("remove_from_user_dir", _remove_from_user_dir_txn) - - @defer.inlineCallbacks - def get_users_in_dir_due_to_room(self, room_id): - """Get all user_ids that are in the room directory because they're - in the given room_id - """ - user_ids_share_pub = yield self._simple_select_onecol( - table="users_in_public_rooms", - keyvalues={"room_id": room_id}, - retcol="user_id", - desc="get_users_in_dir_due_to_room", - ) - - user_ids_share_priv = yield self._simple_select_onecol( - table="users_who_share_private_rooms", - keyvalues={"room_id": room_id}, - retcol="other_user_id", - desc="get_users_in_dir_due_to_room", - ) - - user_ids = set(user_ids_share_pub) - user_ids.update(user_ids_share_priv) - - return user_ids - - def remove_user_who_share_room(self, user_id, room_id): - """ - Deletes entries in the users_who_share_*_rooms table. The first - user should be a local user. - - Args: - user_id (str) - room_id (str) - """ - - def _remove_user_who_share_room_txn(txn): - self._simple_delete_txn( - txn, - table="users_who_share_private_rooms", - keyvalues={"user_id": user_id, "room_id": room_id}, - ) - self._simple_delete_txn( - txn, - table="users_who_share_private_rooms", - keyvalues={"other_user_id": user_id, "room_id": room_id}, - ) - self._simple_delete_txn( - txn, - table="users_in_public_rooms", - keyvalues={"user_id": user_id, "room_id": room_id}, - ) - - return self.runInteraction( - "remove_user_who_share_room", _remove_user_who_share_room_txn - ) - - @defer.inlineCallbacks - def get_user_dir_rooms_user_is_in(self, user_id): - """ - Returns the rooms that a user is in. - - Args: - user_id(str): Must be a local user - - Returns: - list: user_id - """ - rows = yield self._simple_select_onecol( - table="users_who_share_private_rooms", - keyvalues={"user_id": user_id}, - retcol="room_id", - desc="get_rooms_user_is_in", - ) - - pub_rows = yield self._simple_select_onecol( - table="users_in_public_rooms", - keyvalues={"user_id": user_id}, - retcol="room_id", - desc="get_rooms_user_is_in", - ) - - users = set(pub_rows) - users.update(rows) - return list(users) - - @defer.inlineCallbacks - def get_rooms_in_common_for_users(self, user_id, other_user_id): - """Given two user_ids find out the list of rooms they share. - """ - sql = """ - SELECT room_id FROM ( - SELECT c.room_id FROM current_state_events AS c - INNER JOIN room_memberships AS m USING (event_id) - WHERE type = 'm.room.member' - AND m.membership = 'join' - AND state_key = ? - ) AS f1 INNER JOIN ( - SELECT c.room_id FROM current_state_events AS c - INNER JOIN room_memberships AS m USING (event_id) - WHERE type = 'm.room.member' - AND m.membership = 'join' - AND state_key = ? - ) f2 USING (room_id) - """ - - rows = yield self._execute( - "get_rooms_in_common_for_users", None, sql, user_id, other_user_id - ) - - return [room_id for room_id, in rows] - - def get_user_directory_stream_pos(self): - return self._simple_select_one_onecol( - table="user_directory_stream_pos", - keyvalues={}, - retcol="stream_id", - desc="get_user_directory_stream_pos", - ) - - @defer.inlineCallbacks - def search_user_dir(self, user_id, search_term, limit): - """Searches for users in directory - - Returns: - dict of the form:: - - { - "limited": , # whether there were more results or not - "results": [ # Ordered by best match first - { - "user_id": , - "display_name": , - "avatar_url": - } - ] - } - """ - - if self.hs.config.user_directory_search_all_users: - join_args = (user_id,) - where_clause = "user_id != ?" - else: - join_args = (user_id,) - where_clause = """ - ( - EXISTS (select 1 from users_in_public_rooms WHERE user_id = t.user_id) - OR EXISTS ( - SELECT 1 FROM users_who_share_private_rooms - WHERE user_id = ? AND other_user_id = t.user_id - ) - ) - """ - - if isinstance(self.database_engine, PostgresEngine): - full_query, exact_query, prefix_query = _parse_query_postgres(search_term) - - # We order by rank and then if they have profile info - # The ranking algorithm is hand tweaked for "best" results. Broadly - # the idea is we give a higher weight to exact matches. - # The array of numbers are the weights for the various part of the - # search: (domain, _, display name, localpart) - sql = """ - SELECT d.user_id AS user_id, display_name, avatar_url - FROM user_directory_search as t - INNER JOIN user_directory AS d USING (user_id) - WHERE - %s - AND vector @@ to_tsquery('english', ?) - ORDER BY - (CASE WHEN d.user_id IS NOT NULL THEN 4.0 ELSE 1.0 END) - * (CASE WHEN display_name IS NOT NULL THEN 1.2 ELSE 1.0 END) - * (CASE WHEN avatar_url IS NOT NULL THEN 1.2 ELSE 1.0 END) - * ( - 3 * ts_rank_cd( - '{0.1, 0.1, 0.9, 1.0}', - vector, - to_tsquery('english', ?), - 8 - ) - + ts_rank_cd( - '{0.1, 0.1, 0.9, 1.0}', - vector, - to_tsquery('english', ?), - 8 - ) - ) - DESC, - display_name IS NULL, - avatar_url IS NULL - LIMIT ? - """ % ( - where_clause, - ) - args = join_args + (full_query, exact_query, prefix_query, limit + 1) - elif isinstance(self.database_engine, Sqlite3Engine): - search_query = _parse_query_sqlite(search_term) - - sql = """ - SELECT d.user_id AS user_id, display_name, avatar_url - FROM user_directory_search as t - INNER JOIN user_directory AS d USING (user_id) - WHERE - %s - AND value MATCH ? - ORDER BY - rank(matchinfo(user_directory_search)) DESC, - display_name IS NULL, - avatar_url IS NULL - LIMIT ? - """ % ( - where_clause, - ) - args = join_args + (search_query, limit + 1) - else: - # This should be unreachable. - raise Exception("Unrecognized database engine") - - results = yield self._execute( - "search_user_dir", self.cursor_to_dict, sql, *args - ) - - limited = len(results) > limit - - return {"limited": limited, "results": results} - - -def _parse_query_sqlite(search_term): - """Takes a plain unicode string from the user and converts it into a form - that can be passed to database. - We use this so that we can add prefix matching, which isn't something - that is supported by default. - - We specifically add both a prefix and non prefix matching term so that - exact matches get ranked higher. - """ - - # Pull out the individual words, discarding any non-word characters. - results = re.findall(r"([\w\-]+)", search_term, re.UNICODE) - return " & ".join("(%s* OR %s)" % (result, result) for result in results) - - -def _parse_query_postgres(search_term): - """Takes a plain unicode string from the user and converts it into a form - that can be passed to database. - We use this so that we can add prefix matching, which isn't something - that is supported by default. - """ - - # Pull out the individual words, discarding any non-word characters. - results = re.findall(r"([\w\-]+)", search_term, re.UNICODE) - - both = " & ".join("(%s:* | %s)" % (result, result) for result in results) - exact = " & ".join("%s" % (result,) for result in results) - prefix = " & ".join("%s:*" % (result,) for result in results) - - return both, exact, prefix diff --git a/synapse/storage/user_erasure_store.py b/synapse/storage/user_erasure_store.py deleted file mode 100644 index aa4f0da5f0..0000000000 --- a/synapse/storage/user_erasure_store.py +++ /dev/null @@ -1,91 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2018 New Vector Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import operator - -from synapse.storage._base import SQLBaseStore -from synapse.util.caches.descriptors import cached, cachedList - - -class UserErasureWorkerStore(SQLBaseStore): - @cached() - def is_user_erased(self, user_id): - """ - Check if the given user id has requested erasure - - Args: - user_id (str): full user id to check - - Returns: - Deferred[bool]: True if the user has requested erasure - """ - return self._simple_select_onecol( - table="erased_users", - keyvalues={"user_id": user_id}, - retcol="1", - desc="is_user_erased", - ).addCallback(operator.truth) - - @cachedList( - cached_method_name="is_user_erased", list_name="user_ids", inlineCallbacks=True - ) - def are_users_erased(self, user_ids): - """ - Checks which users in a list have requested erasure - - Args: - user_ids (iterable[str]): full user id to check - - Returns: - Deferred[dict[str, bool]]: - for each user, whether the user has requested erasure. - """ - # this serves the dual purpose of (a) making sure we can do len and - # iterate it multiple times, and (b) avoiding duplicates. - user_ids = tuple(set(user_ids)) - - rows = yield self._simple_select_many_batch( - table="erased_users", - column="user_id", - iterable=user_ids, - retcols=("user_id",), - desc="are_users_erased", - ) - erased_users = set(row["user_id"] for row in rows) - - res = dict((u, u in erased_users) for u in user_ids) - return res - - -class UserErasureStore(UserErasureWorkerStore): - def mark_user_erased(self, user_id): - """Indicate that user_id wishes their message history to be erased. - - Args: - user_id (str): full user_id to be erased - """ - - def f(txn): - # first check if they are already in the list - txn.execute("SELECT 1 FROM erased_users WHERE user_id = ?", (user_id,)) - if txn.fetchone(): - return - - # they are not already there: do the insert. - txn.execute("INSERT INTO erased_users (user_id) VALUES (?)", (user_id,)) - - self._invalidate_cache_and_stream(txn, self.is_user_erased, (user_id,)) - - return self.runInteraction("mark_user_erased", f) diff --git a/tests/handlers/test_stats.py b/tests/handlers/test_stats.py index 7569b6fab5..d5c8bd7612 100644 --- a/tests/handlers/test_stats.py +++ b/tests/handlers/test_stats.py @@ -13,9 +13,9 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse import storage from synapse.rest import admin from synapse.rest.client.v1 import login, room +from synapse.storage.data_stores.main import stats from tests import unittest @@ -87,10 +87,10 @@ class StatsRoomTests(unittest.HomeserverTestCase): ) def _get_current_stats(self, stats_type, stat_id): - table, id_col = storage.stats.TYPE_TO_TABLE[stats_type] + table, id_col = stats.TYPE_TO_TABLE[stats_type] - cols = list(storage.stats.ABSOLUTE_STATS_FIELDS[stats_type]) + list( - storage.stats.PER_SLICE_FIELDS[stats_type] + cols = list(stats.ABSOLUTE_STATS_FIELDS[stats_type]) + list( + stats.PER_SLICE_FIELDS[stats_type] ) end_ts = self.store.quantise_stats_time(self.reactor.seconds() * 1000) diff --git a/tests/storage/test_appservice.py b/tests/storage/test_appservice.py index 622b16a071..dfeea24599 100644 --- a/tests/storage/test_appservice.py +++ b/tests/storage/test_appservice.py @@ -24,7 +24,7 @@ from twisted.internet import defer from synapse.appservice import ApplicationService, ApplicationServiceState from synapse.config._base import ConfigError -from synapse.storage.appservice import ( +from synapse.storage.data_stores.main.appservice import ( ApplicationServiceStore, ApplicationServiceTransactionStore, ) diff --git a/tests/storage/test_cleanup_extrems.py b/tests/storage/test_cleanup_extrems.py index 34f9c72709..69dcaa63d5 100644 --- a/tests/storage/test_cleanup_extrems.py +++ b/tests/storage/test_cleanup_extrems.py @@ -50,6 +50,8 @@ class CleanupExtremBackgroundUpdateStoreTestCase(HomeserverTestCase): schema_path = os.path.join( prepare_database.dir_path, + "data_stores", + "main", "schema", "delta", "54", diff --git a/tests/storage/test_profile.py b/tests/storage/test_profile.py index 45824bd3b2..24c7fe16c3 100644 --- a/tests/storage/test_profile.py +++ b/tests/storage/test_profile.py @@ -16,7 +16,7 @@ from twisted.internet import defer -from synapse.storage.profile import ProfileStore +from synapse.storage.data_stores.main.profile import ProfileStore from synapse.types import UserID from tests import unittest diff --git a/tests/storage/test_user_directory.py b/tests/storage/test_user_directory.py index d7d244ce97..7eea57c0e2 100644 --- a/tests/storage/test_user_directory.py +++ b/tests/storage/test_user_directory.py @@ -15,7 +15,7 @@ from twisted.internet import defer -from synapse.storage import UserDirectoryStore +from synapse.storage.data_stores.main.user_directory import UserDirectoryStore from tests import unittest from tests.utils import setup_test_homeserver -- cgit 1.4.1 From 336eeea3ffd14dbd879459cde50f1b7f32e9a325 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Tue, 22 Oct 2019 11:02:01 +0100 Subject: Fix postgres unit tests to use prepare_database --- tests/utils.py | 12 ++---------- 1 file changed, 2 insertions(+), 10 deletions(-) (limited to 'tests') diff --git a/tests/utils.py b/tests/utils.py index 46ef2959f2..0a64f75d04 100644 --- a/tests/utils.py +++ b/tests/utils.py @@ -38,11 +38,7 @@ from synapse.logging.context import LoggingContext from synapse.server import HomeServer from synapse.storage import DataStore from synapse.storage.engines import PostgresEngine, create_engine -from synapse.storage.prepare_database import ( - _get_or_create_schema_state, - _setup_new_database, - prepare_database, -) +from synapse.storage.prepare_database import prepare_database from synapse.util.ratelimitutils import FederationRateLimiter # set this to True to run the tests against postgres instead of sqlite. @@ -88,11 +84,7 @@ def setupdb(): host=POSTGRES_HOST, password=POSTGRES_PASSWORD, ) - cur = db_conn.cursor() - _get_or_create_schema_state(cur, db_engine) - _setup_new_database(cur, db_engine) - db_conn.commit() - cur.close() + prepare_database(db_conn, db_engine, None) db_conn.close() def _cleanup(): -- cgit 1.4.1 From 3e3f9b684e2853217d86349f289780b397afa88a Mon Sep 17 00:00:00 2001 From: Hubert Chathi Date: Tue, 22 Oct 2019 22:26:30 -0400 Subject: fix unit test --- tests/storage/test_devices.py | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'tests') diff --git a/tests/storage/test_devices.py b/tests/storage/test_devices.py index 3cc18f9f1c..039cc79357 100644 --- a/tests/storage/test_devices.py +++ b/tests/storage/test_devices.py @@ -137,7 +137,9 @@ class DeviceStoreTestCase(tests.unittest.TestCase): """Check that an specific device ids exist in a list of device update EDUs""" self.assertEqual(len(device_updates), len(expected_device_ids)) - received_device_ids = {update["device_id"] for update in device_updates} + received_device_ids = { + update["device_id"] for edu_type, update in device_updates + } self.assertEqual(received_device_ids, set(expected_device_ids)) @defer.inlineCallbacks -- cgit 1.4.1 From 3ca4c7c516a349cbb9dace0ace33bb889955eda1 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 23 Oct 2019 12:02:36 +0100 Subject: Use new EventPersistenceStore --- synapse/handlers/federation.py | 3 ++- synapse/handlers/message.py | 3 ++- synapse/server.py | 9 ++++++++- tests/replication/slave/storage/_base.py | 1 + tests/replication/slave/storage/test_events.py | 10 +++++++--- tests/storage/test_redaction.py | 11 ++++++----- tests/storage/test_room.py | 3 ++- tests/storage/test_roommember.py | 3 ++- tests/storage/test_state.py | 3 ++- tests/test_visibility.py | 7 ++++--- tests/utils.py | 10 ++++++++-- 11 files changed, 44 insertions(+), 19 deletions(-) (limited to 'tests') diff --git a/synapse/handlers/federation.py b/synapse/handlers/federation.py index 4b4c6c15f9..7e9c9aee13 100644 --- a/synapse/handlers/federation.py +++ b/synapse/handlers/federation.py @@ -109,6 +109,7 @@ class FederationHandler(BaseHandler): self.hs = hs self.store = hs.get_datastore() + self.storage = hs.get_storage() self.federation_client = hs.get_federation_client() self.state_handler = hs.get_state_handler() self.server_name = hs.hostname @@ -2648,7 +2649,7 @@ class FederationHandler(BaseHandler): backfilled=backfilled, ) else: - max_stream_id = yield self.store.persist_events( + max_stream_id = yield self.storage.persistence.persist_events( event_and_contexts, backfilled=backfilled ) diff --git a/synapse/handlers/message.py b/synapse/handlers/message.py index 0f8cce8ffe..7908a2d52c 100644 --- a/synapse/handlers/message.py +++ b/synapse/handlers/message.py @@ -234,6 +234,7 @@ class EventCreationHandler(object): self.hs = hs self.auth = hs.get_auth() self.store = hs.get_datastore() + self.storage = hs.get_storage() self.state = hs.get_state_handler() self.clock = hs.get_clock() self.validator = EventValidator() @@ -868,7 +869,7 @@ class EventCreationHandler(object): if prev_state_ids: raise AuthError(403, "Changing the room create event is forbidden") - (event_stream_id, max_stream_id) = yield self.store.persist_event( + event_stream_id, max_stream_id = yield self.storage.persistence.persist_event( event, context=context ) diff --git a/synapse/server.py b/synapse/server.py index 1fcc7375d3..54a7f4aa5f 100644 --- a/synapse/server.py +++ b/synapse/server.py @@ -95,6 +95,7 @@ from synapse.server_notices.worker_server_notices_sender import ( WorkerServerNoticesSender, ) from synapse.state import StateHandler, StateResolutionHandler +from synapse.storage import DataStores, Storage from synapse.streams.events import EventSources from synapse.util import Clock from synapse.util.distributor import Distributor @@ -196,6 +197,7 @@ class HomeServer(object): "account_validity_handler", "saml_handler", "event_client_serializer", + "storage", ] REQUIRED_ON_MASTER_STARTUP = ["user_directory_handler", "stats_handler"] @@ -225,6 +227,7 @@ class HomeServer(object): self.registration_ratelimiter = Ratelimiter() self.datastore = None + self.datastores = None # Other kwargs are explicit dependencies for depname in kwargs: @@ -234,6 +237,7 @@ class HomeServer(object): logger.info("Setting up.") with self.get_db_conn() as conn: self.datastore = self.DATASTORE_CLASS(conn, self) + self.datastores = DataStores(self.datastore, conn, self) conn.commit() logger.info("Finished setting up.") @@ -266,7 +270,7 @@ class HomeServer(object): return self.clock def get_datastore(self): - return self.datastore + return self.datastores.main def get_config(self): return self.config @@ -537,6 +541,9 @@ class HomeServer(object): def build_event_client_serializer(self): return EventClientSerializer(self) + def build_storage(self) -> Storage: + return Storage(self, self.datastores) + def remove_pusher(self, app_id, push_key, user_id): return self.get_pusherpool().remove_pusher(app_id, push_key, user_id) diff --git a/tests/replication/slave/storage/_base.py b/tests/replication/slave/storage/_base.py index 104349cdbd..4f924ce451 100644 --- a/tests/replication/slave/storage/_base.py +++ b/tests/replication/slave/storage/_base.py @@ -41,6 +41,7 @@ class BaseSlavedStoreTestCase(unittest.HomeserverTestCase): def prepare(self, reactor, clock, hs): self.master_store = self.hs.get_datastore() + self.storage = hs.get_storage() self.slaved_store = self.STORE_TYPE(self.hs.get_db_conn(), self.hs) self.event_id = 0 diff --git a/tests/replication/slave/storage/test_events.py b/tests/replication/slave/storage/test_events.py index a368117b43..b68e9fe082 100644 --- a/tests/replication/slave/storage/test_events.py +++ b/tests/replication/slave/storage/test_events.py @@ -234,7 +234,9 @@ class SlavedEventStoreTestCase(BaseSlavedStoreTestCase): type="m.room.member", sender=USER_ID_2, key=USER_ID_2, membership="join" ) msg, msgctx = self.build_event() - self.get_success(self.master_store.persist_events([(j2, j2ctx), (msg, msgctx)])) + self.get_success( + self.storage.persistence.persist_events([(j2, j2ctx), (msg, msgctx)]) + ) self.replicate() event_source = RoomEventSource(self.hs) @@ -290,10 +292,12 @@ class SlavedEventStoreTestCase(BaseSlavedStoreTestCase): if backfill: self.get_success( - self.master_store.persist_events([(event, context)], backfilled=True) + self.storage.persistence.persist_events( + [(event, context)], backfilled=True + ) ) else: - self.get_success(self.master_store.persist_event(event, context)) + self.get_success(self.storage.persistence.persist_event(event, context)) return event diff --git a/tests/storage/test_redaction.py b/tests/storage/test_redaction.py index 427d3c49c5..4561c3e383 100644 --- a/tests/storage/test_redaction.py +++ b/tests/storage/test_redaction.py @@ -39,6 +39,7 @@ class RedactionTestCase(unittest.HomeserverTestCase): def prepare(self, reactor, clock, hs): self.store = hs.get_datastore() + self.storage = hs.get_storage() self.event_builder_factory = hs.get_event_builder_factory() self.event_creation_handler = hs.get_event_creation_handler() @@ -73,7 +74,7 @@ class RedactionTestCase(unittest.HomeserverTestCase): self.event_creation_handler.create_new_client_event(builder) ) - self.get_success(self.store.persist_event(event, context)) + self.get_success(self.storage.persistence.persist_event(event, context)) return event @@ -95,7 +96,7 @@ class RedactionTestCase(unittest.HomeserverTestCase): self.event_creation_handler.create_new_client_event(builder) ) - self.get_success(self.store.persist_event(event, context)) + self.get_success(self.storage.persistence.persist_event(event, context)) return event @@ -116,7 +117,7 @@ class RedactionTestCase(unittest.HomeserverTestCase): self.event_creation_handler.create_new_client_event(builder) ) - self.get_success(self.store.persist_event(event, context)) + self.get_success(self.storage.persistence.persist_event(event, context)) return event @@ -263,7 +264,7 @@ class RedactionTestCase(unittest.HomeserverTestCase): ) ) - self.get_success(self.store.persist_event(event_1, context_1)) + self.get_success(self.storage.persistence.persist_event(event_1, context_1)) event_2, context_2 = self.get_success( self.event_creation_handler.create_new_client_event( @@ -282,7 +283,7 @@ class RedactionTestCase(unittest.HomeserverTestCase): ) ) ) - self.get_success(self.store.persist_event(event_2, context_2)) + self.get_success(self.storage.persistence.persist_event(event_2, context_2)) # fetch one of the redactions fetched = self.get_success(self.store.get_event(redaction_event_id1)) diff --git a/tests/storage/test_room.py b/tests/storage/test_room.py index 1bee45706f..3ddaa151fe 100644 --- a/tests/storage/test_room.py +++ b/tests/storage/test_room.py @@ -62,6 +62,7 @@ class RoomEventsStoreTestCase(unittest.TestCase): # Room events need the full datastore, for persist_event() and # get_room_state() self.store = hs.get_datastore() + self.storage = hs.get_storage() self.event_factory = hs.get_event_factory() self.room = RoomID.from_string("!abcde:test") @@ -72,7 +73,7 @@ class RoomEventsStoreTestCase(unittest.TestCase): @defer.inlineCallbacks def inject_room_event(self, **kwargs): - yield self.store.persist_event( + yield self.storage.persistence.persist_event( self.event_factory.create_event(room_id=self.room.to_string(), **kwargs) ) diff --git a/tests/storage/test_roommember.py b/tests/storage/test_roommember.py index 447a3c6ffb..9ddd17f73d 100644 --- a/tests/storage/test_roommember.py +++ b/tests/storage/test_roommember.py @@ -44,6 +44,7 @@ class RoomMemberStoreTestCase(unittest.HomeserverTestCase): # We can't test the RoomMemberStore on its own without the other event # storage logic self.store = hs.get_datastore() + self.storage = hs.get_storage() self.event_builder_factory = hs.get_event_builder_factory() self.event_creation_handler = hs.get_event_creation_handler() @@ -70,7 +71,7 @@ class RoomMemberStoreTestCase(unittest.HomeserverTestCase): self.event_creation_handler.create_new_client_event(builder) ) - self.get_success(self.store.persist_event(event, context)) + self.get_success(self.storage.persistence.persist_event(event, context)) return event diff --git a/tests/storage/test_state.py b/tests/storage/test_state.py index 5c2cf3c2db..d573a3e07b 100644 --- a/tests/storage/test_state.py +++ b/tests/storage/test_state.py @@ -34,6 +34,7 @@ class StateStoreTestCase(tests.unittest.TestCase): hs = yield tests.utils.setup_test_homeserver(self.addCleanup) self.store = hs.get_datastore() + self.storage = hs.get_storage() self.event_builder_factory = hs.get_event_builder_factory() self.event_creation_handler = hs.get_event_creation_handler() @@ -63,7 +64,7 @@ class StateStoreTestCase(tests.unittest.TestCase): builder ) - yield self.store.persist_event(event, context) + yield self.storage.persistence.persist_event(event, context) return event diff --git a/tests/test_visibility.py b/tests/test_visibility.py index 18f1a0035d..6ae1ea9b04 100644 --- a/tests/test_visibility.py +++ b/tests/test_visibility.py @@ -36,6 +36,7 @@ class FilterEventsForServerTestCase(tests.unittest.TestCase): self.event_creation_handler = self.hs.get_event_creation_handler() self.event_builder_factory = self.hs.get_event_builder_factory() self.store = self.hs.get_datastore() + self.storage = self.hs.get_storage() yield create_room(self.hs, TEST_ROOM_ID, "@someone:ROOM") @@ -137,7 +138,7 @@ class FilterEventsForServerTestCase(tests.unittest.TestCase): event, context = yield self.event_creation_handler.create_new_client_event( builder ) - yield self.hs.get_datastore().persist_event(event, context) + yield self.storage.persistence.persist_event(event, context) return event @defer.inlineCallbacks @@ -159,7 +160,7 @@ class FilterEventsForServerTestCase(tests.unittest.TestCase): builder ) - yield self.hs.get_datastore().persist_event(event, context) + yield self.storage.persistence.persist_event(event, context) return event @defer.inlineCallbacks @@ -180,7 +181,7 @@ class FilterEventsForServerTestCase(tests.unittest.TestCase): builder ) - yield self.hs.get_datastore().persist_event(event, context) + yield self.storage.persistence.persist_event(event, context) return event @defer.inlineCallbacks diff --git a/tests/utils.py b/tests/utils.py index 0a64f75d04..2808eea933 100644 --- a/tests/utils.py +++ b/tests/utils.py @@ -326,10 +326,16 @@ def setup_test_homeserver( if homeserverToUse.__name__ == "TestHomeServer": hs.setup_master() else: + # If we have been given an explicit datastore we probably want to mock + # out the DataStores somehow too. This all feels a bit wrong, but then + # mocking the stores feels wrong too. + datastores = Mock(datastore=datastore) + hs = homeserverToUse( name, db_pool=None, datastore=datastore, + datastores=datastores, config=config, version_string="Synapse/tests", database_engine=db_engine, @@ -647,7 +653,7 @@ def create_room(hs, room_id, creator_id): creator_id (str) """ - store = hs.get_datastore() + persistence_store = hs.get_storage().persistence event_builder_factory = hs.get_event_builder_factory() event_creation_handler = hs.get_event_creation_handler() @@ -664,4 +670,4 @@ def create_room(hs, room_id, creator_id): event, context = yield event_creation_handler.create_new_client_event(builder) - yield store.persist_event(event, context) + yield persistence_store.persist_event(event, context) -- cgit 1.4.1 From 2794b79052f96b8103ce2b710959853313a82e90 Mon Sep 17 00:00:00 2001 From: Neil Johnson Date: Thu, 24 Oct 2019 11:48:46 +0100 Subject: Option to suppress resource exceeded alerting (#6173) The expected use case is to suppress MAU limiting on small instances --- changelog.d/6173.feature | 1 + docs/sample_config.yaml | 8 +- synapse/api/auth.py | 12 ++- synapse/api/constants.py | 7 ++ synapse/config/server.py | 10 +- .../resource_limits_server_notices.py | 110 ++++++++++++++------- .../test_resource_limits_server_notices.py | 59 ++++++++++- tests/utils.py | 1 - 8 files changed, 161 insertions(+), 47 deletions(-) create mode 100644 changelog.d/6173.feature (limited to 'tests') diff --git a/changelog.d/6173.feature b/changelog.d/6173.feature new file mode 100644 index 0000000000..b1cabc322b --- /dev/null +++ b/changelog.d/6173.feature @@ -0,0 +1 @@ +Add config option to suppress client side resource limit alerting. diff --git a/docs/sample_config.yaml b/docs/sample_config.yaml index b4dd146f06..6c81c0db75 100644 --- a/docs/sample_config.yaml +++ b/docs/sample_config.yaml @@ -241,7 +241,6 @@ listeners: # #hs_disabled: false #hs_disabled_message: 'Human readable reason for why the HS is blocked' -#hs_disabled_limit_type: 'error code(str), to help clients decode reason' # Monthly Active User Blocking # @@ -261,9 +260,16 @@ listeners: # sign up in a short space of time never to return after their initial # session. # +# 'mau_limit_alerting' is a means of limiting client side alerting +# should the mau limit be reached. This is useful for small instances +# where the admin has 5 mau seats (say) for 5 specific people and no +# interest increasing the mau limit further. Defaults to True, which +# means that alerting is enabled +# #limit_usage_by_mau: false #max_mau_value: 50 #mau_trial_days: 2 +#mau_limit_alerting: false # If enabled, the metrics for the number of monthly active users will # be populated, however no one will be limited. If limit_usage_by_mau diff --git a/synapse/api/auth.py b/synapse/api/auth.py index cd347fbe1b..53f3bb0fa8 100644 --- a/synapse/api/auth.py +++ b/synapse/api/auth.py @@ -25,7 +25,13 @@ from twisted.internet import defer import synapse.logging.opentracing as opentracing import synapse.types from synapse import event_auth -from synapse.api.constants import EventTypes, JoinRules, Membership, UserTypes +from synapse.api.constants import ( + EventTypes, + JoinRules, + LimitBlockingTypes, + Membership, + UserTypes, +) from synapse.api.errors import ( AuthError, Codes, @@ -726,7 +732,7 @@ class Auth(object): self.hs.config.hs_disabled_message, errcode=Codes.RESOURCE_LIMIT_EXCEEDED, admin_contact=self.hs.config.admin_contact, - limit_type=self.hs.config.hs_disabled_limit_type, + limit_type=LimitBlockingTypes.HS_DISABLED, ) if self.hs.config.limit_usage_by_mau is True: assert not (user_id and threepid) @@ -759,5 +765,5 @@ class Auth(object): "Monthly Active User Limit Exceeded", admin_contact=self.hs.config.admin_contact, errcode=Codes.RESOURCE_LIMIT_EXCEEDED, - limit_type="monthly_active_user", + limit_type=LimitBlockingTypes.MONTHLY_ACTIVE_USER, ) diff --git a/synapse/api/constants.py b/synapse/api/constants.py index 60e99e4663..312196675e 100644 --- a/synapse/api/constants.py +++ b/synapse/api/constants.py @@ -131,3 +131,10 @@ class RelationTypes(object): ANNOTATION = "m.annotation" REPLACE = "m.replace" REFERENCE = "m.reference" + + +class LimitBlockingTypes(object): + """Reasons that a server may be blocked""" + + MONTHLY_ACTIVE_USER = "monthly_active_user" + HS_DISABLED = "hs_disabled" diff --git a/synapse/config/server.py b/synapse/config/server.py index c942841578..d556df308d 100644 --- a/synapse/config/server.py +++ b/synapse/config/server.py @@ -171,6 +171,7 @@ class ServerConfig(Config): ) self.mau_trial_days = config.get("mau_trial_days", 0) + self.mau_limit_alerting = config.get("mau_limit_alerting", True) # How long to keep redacted events in the database in unredacted form # before redacting them. @@ -192,7 +193,6 @@ class ServerConfig(Config): # Options to disable HS self.hs_disabled = config.get("hs_disabled", False) self.hs_disabled_message = config.get("hs_disabled_message", "") - self.hs_disabled_limit_type = config.get("hs_disabled_limit_type", "") # Admin uri to direct users at should their instance become blocked # due to resource constraints @@ -675,7 +675,6 @@ class ServerConfig(Config): # #hs_disabled: false #hs_disabled_message: 'Human readable reason for why the HS is blocked' - #hs_disabled_limit_type: 'error code(str), to help clients decode reason' # Monthly Active User Blocking # @@ -695,9 +694,16 @@ class ServerConfig(Config): # sign up in a short space of time never to return after their initial # session. # + # 'mau_limit_alerting' is a means of limiting client side alerting + # should the mau limit be reached. This is useful for small instances + # where the admin has 5 mau seats (say) for 5 specific people and no + # interest increasing the mau limit further. Defaults to True, which + # means that alerting is enabled + # #limit_usage_by_mau: false #max_mau_value: 50 #mau_trial_days: 2 + #mau_limit_alerting: false # If enabled, the metrics for the number of monthly active users will # be populated, however no one will be limited. If limit_usage_by_mau diff --git a/synapse/server_notices/resource_limits_server_notices.py b/synapse/server_notices/resource_limits_server_notices.py index 81c4aff496..c0e7f475c9 100644 --- a/synapse/server_notices/resource_limits_server_notices.py +++ b/synapse/server_notices/resource_limits_server_notices.py @@ -20,6 +20,7 @@ from twisted.internet import defer from synapse.api.constants import ( EventTypes, + LimitBlockingTypes, ServerNoticeLimitReached, ServerNoticeMsgType, ) @@ -70,7 +71,7 @@ class ResourceLimitsServerNotices(object): return if not self._server_notices_manager.is_enabled(): - # Don't try and send server notices unles they've been enabled + # Don't try and send server notices unless they've been enabled return timestamp = yield self._store.user_last_seen_monthly_active(user_id) @@ -79,8 +80,6 @@ class ResourceLimitsServerNotices(object): # In practice, not sure we can ever get here return - # Determine current state of room - room_id = yield self._server_notices_manager.get_notice_room_for_user(user_id) if not room_id: @@ -88,50 +87,85 @@ class ResourceLimitsServerNotices(object): return yield self._check_and_set_tags(user_id, room_id) + + # Determine current state of room currently_blocked, ref_events = yield self._is_room_currently_blocked(room_id) + limit_msg = None + limit_type = None try: - # Normally should always pass in user_id if you have it, but in - # this case are checking what would happen to other users if they - # were to arrive. - try: - yield self._auth.check_auth_blocking() - is_auth_blocking = False - except ResourceLimitError as e: - is_auth_blocking = True - event_content = e.msg - event_limit_type = e.limit_type - - if currently_blocked and not is_auth_blocking: - # Room is notifying of a block, when it ought not to be. - # Remove block notification - content = {"pinned": ref_events} - yield self._server_notices_manager.send_notice( - user_id, content, EventTypes.Pinned, "" - ) + # Normally should always pass in user_id to check_auth_blocking + # if you have it, but in this case are checking what would happen + # to other users if they were to arrive. + yield self._auth.check_auth_blocking() + except ResourceLimitError as e: + limit_msg = e.msg + limit_type = e.limit_type - elif not currently_blocked and is_auth_blocking: + try: + if ( + limit_type == LimitBlockingTypes.MONTHLY_ACTIVE_USER + and not self._config.mau_limit_alerting + ): + # We have hit the MAU limit, but MAU alerting is disabled: + # reset room if necessary and return + if currently_blocked: + self._remove_limit_block_notification(user_id, ref_events) + return + + if currently_blocked and not limit_msg: + # Room is notifying of a block, when it ought not to be. + yield self._remove_limit_block_notification(user_id, ref_events) + elif not currently_blocked and limit_msg: # Room is not notifying of a block, when it ought to be. - # Add block notification - content = { - "body": event_content, - "msgtype": ServerNoticeMsgType, - "server_notice_type": ServerNoticeLimitReached, - "admin_contact": self._config.admin_contact, - "limit_type": event_limit_type, - } - event = yield self._server_notices_manager.send_notice( - user_id, content, EventTypes.Message + yield self._apply_limit_block_notification( + user_id, limit_msg, limit_type ) - - content = {"pinned": [event.event_id]} - yield self._server_notices_manager.send_notice( - user_id, content, EventTypes.Pinned, "" - ) - except SynapseError as e: logger.error("Error sending resource limits server notice: %s", e) + @defer.inlineCallbacks + def _remove_limit_block_notification(self, user_id, ref_events): + """Utility method to remove limit block notifications from the server + notices room. + + Args: + user_id (str): user to notify + ref_events (list[str]): The event_ids of pinned events that are unrelated to + limit blocking and need to be preserved. + """ + content = {"pinned": ref_events} + yield self._server_notices_manager.send_notice( + user_id, content, EventTypes.Pinned, "" + ) + + @defer.inlineCallbacks + def _apply_limit_block_notification(self, user_id, event_body, event_limit_type): + """Utility method to apply limit block notifications in the server + notices room. + + Args: + user_id (str): user to notify + event_body(str): The human readable text that describes the block. + event_limit_type(str): Specifies the type of block e.g. monthly active user + limit has been exceeded. + """ + content = { + "body": event_body, + "msgtype": ServerNoticeMsgType, + "server_notice_type": ServerNoticeLimitReached, + "admin_contact": self._config.admin_contact, + "limit_type": event_limit_type, + } + event = yield self._server_notices_manager.send_notice( + user_id, content, EventTypes.Message + ) + + content = {"pinned": [event.event_id]} + yield self._server_notices_manager.send_notice( + user_id, content, EventTypes.Pinned, "" + ) + @defer.inlineCallbacks def _check_and_set_tags(self, user_id, room_id): """ diff --git a/tests/server_notices/test_resource_limits_server_notices.py b/tests/server_notices/test_resource_limits_server_notices.py index cdf89e3383..eb540e34f6 100644 --- a/tests/server_notices/test_resource_limits_server_notices.py +++ b/tests/server_notices/test_resource_limits_server_notices.py @@ -17,7 +17,7 @@ from mock import Mock from twisted.internet import defer -from synapse.api.constants import EventTypes, ServerNoticeMsgType +from synapse.api.constants import EventTypes, LimitBlockingTypes, ServerNoticeMsgType from synapse.api.errors import ResourceLimitError from synapse.server_notices.resource_limits_server_notices import ( ResourceLimitsServerNotices, @@ -133,7 +133,7 @@ class TestResourceLimitsServerNotices(unittest.HomeserverTestCase): self.get_success(self._rlsn.maybe_send_server_notice_to_user(self.user_id)) # Would be better to check contents, but 2 calls == set blocking event - self.assertTrue(self._send_notice.call_count == 2) + self.assertEqual(self._send_notice.call_count, 2) def test_maybe_send_server_notice_to_user_add_blocked_notice_noop(self): """ @@ -158,6 +158,61 @@ class TestResourceLimitsServerNotices(unittest.HomeserverTestCase): self._send_notice.assert_not_called() + def test_maybe_send_server_notice_when_alerting_suppressed_room_unblocked(self): + """ + Test that when server is over MAU limit and alerting is suppressed, then + an alert message is not sent into the room + """ + self.hs.config.mau_limit_alerting = False + self._rlsn._auth.check_auth_blocking = Mock( + side_effect=ResourceLimitError( + 403, "foo", limit_type=LimitBlockingTypes.MONTHLY_ACTIVE_USER + ) + ) + self.get_success(self._rlsn.maybe_send_server_notice_to_user(self.user_id)) + + self.assertTrue(self._send_notice.call_count == 0) + + def test_check_hs_disabled_unaffected_by_mau_alert_suppression(self): + """ + Test that when a server is disabled, that MAU limit alerting is ignored. + """ + self.hs.config.mau_limit_alerting = False + self._rlsn._auth.check_auth_blocking = Mock( + side_effect=ResourceLimitError( + 403, "foo", limit_type=LimitBlockingTypes.HS_DISABLED + ) + ) + self.get_success(self._rlsn.maybe_send_server_notice_to_user(self.user_id)) + + # Would be better to check contents, but 2 calls == set blocking event + self.assertEqual(self._send_notice.call_count, 2) + + def test_maybe_send_server_notice_when_alerting_suppressed_room_blocked(self): + """ + When the room is already in a blocked state, test that when alerting + is suppressed that the room is returned to an unblocked state. + """ + self.hs.config.mau_limit_alerting = False + self._rlsn._auth.check_auth_blocking = Mock( + side_effect=ResourceLimitError( + 403, "foo", limit_type=LimitBlockingTypes.MONTHLY_ACTIVE_USER + ) + ) + self._rlsn._server_notices_manager.__is_room_currently_blocked = Mock( + return_value=defer.succeed((True, [])) + ) + + mock_event = Mock( + type=EventTypes.Message, content={"msgtype": ServerNoticeMsgType} + ) + self._rlsn._store.get_events = Mock( + return_value=defer.succeed({"123": mock_event}) + ) + self.get_success(self._rlsn.maybe_send_server_notice_to_user(self.user_id)) + + self._send_notice.assert_called_once() + class TestResourceLimitsServerNoticesWithRealRooms(unittest.HomeserverTestCase): def prepare(self, reactor, clock, hs): diff --git a/tests/utils.py b/tests/utils.py index 0a64f75d04..8cced4b7e8 100644 --- a/tests/utils.py +++ b/tests/utils.py @@ -137,7 +137,6 @@ def default_config(name, parse=False): "limit_usage_by_mau": False, "hs_disabled": False, "hs_disabled_message": "", - "hs_disabled_limit_type": "", "max_mau_value": 50, "mau_trial_days": 0, "mau_stats_only": False, -- cgit 1.4.1 From 848cd388d96ec95b2598f1eaaf8967b8f064c08c Mon Sep 17 00:00:00 2001 From: Hubert Chathi Date: Thu, 24 Oct 2019 21:13:01 -0400 Subject: delete keys when deleting backups --- synapse/storage/data_stores/main/e2e_room_keys.py | 8 +++ .../delta/56/delete_keys_from_deleted_backups.sql | 25 +++++++ tests/storage/test_e2e_room_keys.py | 76 ++++++++++++++++++++++ 3 files changed, 109 insertions(+) create mode 100644 synapse/storage/data_stores/main/schema/delta/56/delete_keys_from_deleted_backups.sql create mode 100644 tests/storage/test_e2e_room_keys.py (limited to 'tests') diff --git a/synapse/storage/data_stores/main/e2e_room_keys.py b/synapse/storage/data_stores/main/e2e_room_keys.py index ef88e79293..1cbbae5b63 100644 --- a/synapse/storage/data_stores/main/e2e_room_keys.py +++ b/synapse/storage/data_stores/main/e2e_room_keys.py @@ -321,9 +321,17 @@ class EndToEndRoomKeyStore(SQLBaseStore): def _delete_e2e_room_keys_version_txn(txn): if version is None: this_version = self._get_current_version(txn, user_id) + if this_version is None: + raise StoreError(404, "No current backup version") else: this_version = version + self._simple_delete_txn( + txn, + table="e2e_room_keys", + keyvalues={"user_id": user_id, "version": this_version}, + ) + return self._simple_update_one_txn( txn, table="e2e_room_keys_versions", diff --git a/synapse/storage/data_stores/main/schema/delta/56/delete_keys_from_deleted_backups.sql b/synapse/storage/data_stores/main/schema/delta/56/delete_keys_from_deleted_backups.sql new file mode 100644 index 0000000000..1d2ddb1b1a --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/56/delete_keys_from_deleted_backups.sql @@ -0,0 +1,25 @@ +/* Copyright 2019 The Matrix.org Foundation C.I.C + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* delete room keys that belong to deleted room key version, or to room key + * versions that don't exist (anymore) + */ +DELETE FROM e2e_room_keys +WHERE version NOT IN ( + SELECT version + FROM e2e_room_keys_versions + WHERE e2e_room_keys.user_id = e2e_room_keys_versions.user_id + AND e2e_room_keys_versions.deleted = 0 +); diff --git a/tests/storage/test_e2e_room_keys.py b/tests/storage/test_e2e_room_keys.py new file mode 100644 index 0000000000..ef4e7ce9d6 --- /dev/null +++ b/tests/storage/test_e2e_room_keys.py @@ -0,0 +1,76 @@ +# -*- coding: utf-8 -*- +# Copyright 2019 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from twisted.internet import defer + +from tests import unittest, utils + +# sample room_key data for use in the tests +room_key = { + "first_message_index": 1, + "forwarded_count": 1, + "is_verified": False, + "session_data": "SSBBTSBBIEZJU0gK", +} + + +class E2eRoomKeysHandlerTestCase(unittest.TestCase): + def __init__(self, *args, **kwargs): + super(E2eRoomKeysHandlerTestCase, self).__init__(*args, **kwargs) + self.hs = None # type: synapse.server.HomeServer + self.store = None # type: synapse.storage.DataStore + + @defer.inlineCallbacks + def setUp(self): + hs = yield utils.setup_test_homeserver(self.addCleanup) + + self.store = hs.get_datastore() + + @defer.inlineCallbacks + def test_room_keys_version_delete(self): + # test that deleting a room key backup deletes the keys + version1 = yield self.store.create_e2e_room_keys_version( + "user_id", {"algorithm": "rot13", "auth_data": {}} + ) + + yield self.store.set_e2e_room_key( + "user_id", version1, "room", "session", room_key + ) + + version2 = yield self.store.create_e2e_room_keys_version( + "user_id", {"algorithm": "rot13", "auth_data": {}} + ) + + yield self.store.set_e2e_room_key( + "user_id", version2, "room", "session", room_key + ) + + # make sure the keys were stored properly + keys = yield self.store.get_e2e_room_keys("user_id", version1) + self.assertEqual(len(keys["rooms"]), 1) + + keys = yield self.store.get_e2e_room_keys("user_id", version2) + self.assertEqual(len(keys["rooms"]), 1) + + # delete version1 + yield self.store.delete_e2e_room_keys_version("user_id", version1) + + # make sure the key from version1 is gone, and the key from version2 is + # still there + keys = yield self.store.get_e2e_room_keys("user_id", version1) + self.assertEqual(len(keys["rooms"]), 0) + + keys = yield self.store.get_e2e_room_keys("user_id", version2) + self.assertEqual(len(keys["rooms"]), 1) -- cgit 1.4.1 From 29a0bc5637e6811220f44ee727370a190b5be1ab Mon Sep 17 00:00:00 2001 From: Hubert Chathi Date: Thu, 24 Oct 2019 21:43:02 -0400 Subject: remove some unnecessary lines --- tests/storage/test_e2e_room_keys.py | 5 ----- 1 file changed, 5 deletions(-) (limited to 'tests') diff --git a/tests/storage/test_e2e_room_keys.py b/tests/storage/test_e2e_room_keys.py index ef4e7ce9d6..6658dbda94 100644 --- a/tests/storage/test_e2e_room_keys.py +++ b/tests/storage/test_e2e_room_keys.py @@ -27,11 +27,6 @@ room_key = { class E2eRoomKeysHandlerTestCase(unittest.TestCase): - def __init__(self, *args, **kwargs): - super(E2eRoomKeysHandlerTestCase, self).__init__(*args, **kwargs) - self.hs = None # type: synapse.server.HomeServer - self.store = None # type: synapse.storage.DataStore - @defer.inlineCallbacks def setUp(self): hs = yield utils.setup_test_homeserver(self.addCleanup) -- cgit 1.4.1 From 7e7a1461f64888fa71c8f06fb12f29cc3d233179 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Fri, 25 Oct 2019 10:57:37 +0100 Subject: Fix tests --- tests/handlers/test_stats.py | 2 ++ 1 file changed, 2 insertions(+) (limited to 'tests') diff --git a/tests/handlers/test_stats.py b/tests/handlers/test_stats.py index d5c8bd7612..e0075ccd32 100644 --- a/tests/handlers/test_stats.py +++ b/tests/handlers/test_stats.py @@ -607,6 +607,7 @@ class StatsRoomTests(unittest.HomeserverTestCase): """ self.hs.config.stats_enabled = False + self.handler.stats_enabled = False u1 = self.register_user("u1", "pass") u1token = self.login("u1", "pass") @@ -618,6 +619,7 @@ class StatsRoomTests(unittest.HomeserverTestCase): self.assertIsNone(self._get_current_stats("user", u1)) self.hs.config.stats_enabled = True + self.handler.stats_enabled = True self._perform_background_initial_update() -- cgit 1.4.1 From 4cf3a30a20c64c3939135b00b3eb5b06f273c9f9 Mon Sep 17 00:00:00 2001 From: Hubert Chathi Date: Fri, 25 Oct 2019 10:42:07 -0400 Subject: switch to using HomeserverTestCase --- tests/storage/test_e2e_room_keys.py | 44 +++++++++++++++++++++---------------- 1 file changed, 25 insertions(+), 19 deletions(-) (limited to 'tests') diff --git a/tests/storage/test_e2e_room_keys.py b/tests/storage/test_e2e_room_keys.py index 6658dbda94..9935ac59ce 100644 --- a/tests/storage/test_e2e_room_keys.py +++ b/tests/storage/test_e2e_room_keys.py @@ -26,46 +26,52 @@ room_key = { } -class E2eRoomKeysHandlerTestCase(unittest.TestCase): - @defer.inlineCallbacks - def setUp(self): - hs = yield utils.setup_test_homeserver(self.addCleanup) - +class E2eRoomKeysHandlerTestCase(unittest.HomeserverTestCase): + def make_homeserver(self, reactor, clock): + hs = self.setup_test_homeserver("server", http_client=None) self.store = hs.get_datastore() + return hs - @defer.inlineCallbacks def test_room_keys_version_delete(self): # test that deleting a room key backup deletes the keys - version1 = yield self.store.create_e2e_room_keys_version( - "user_id", {"algorithm": "rot13", "auth_data": {}} + version1 = self.get_success( + self.store.create_e2e_room_keys_version( + "user_id", {"algorithm": "rot13", "auth_data": {}} + ) ) - yield self.store.set_e2e_room_key( - "user_id", version1, "room", "session", room_key + self.get_success( + self.store.set_e2e_room_key( + "user_id", version1, "room", "session", room_key + ) ) - version2 = yield self.store.create_e2e_room_keys_version( - "user_id", {"algorithm": "rot13", "auth_data": {}} + version2 = self.get_success( + self.store.create_e2e_room_keys_version( + "user_id", {"algorithm": "rot13", "auth_data": {}} + ) ) - yield self.store.set_e2e_room_key( - "user_id", version2, "room", "session", room_key + self.get_success( + self.store.set_e2e_room_key( + "user_id", version2, "room", "session", room_key + ) ) # make sure the keys were stored properly - keys = yield self.store.get_e2e_room_keys("user_id", version1) + keys = self.get_success(self.store.get_e2e_room_keys("user_id", version1)) self.assertEqual(len(keys["rooms"]), 1) - keys = yield self.store.get_e2e_room_keys("user_id", version2) + keys = self.get_success(self.store.get_e2e_room_keys("user_id", version2)) self.assertEqual(len(keys["rooms"]), 1) # delete version1 - yield self.store.delete_e2e_room_keys_version("user_id", version1) + self.get_success(self.store.delete_e2e_room_keys_version("user_id", version1)) # make sure the key from version1 is gone, and the key from version2 is # still there - keys = yield self.store.get_e2e_room_keys("user_id", version1) + keys = self.get_success(self.store.get_e2e_room_keys("user_id", version1)) self.assertEqual(len(keys["rooms"]), 0) - keys = yield self.store.get_e2e_room_keys("user_id", version2) + keys = self.get_success(self.store.get_e2e_room_keys("user_id", version2)) self.assertEqual(len(keys["rooms"]), 1) -- cgit 1.4.1 From 4697c0de0b0b51b7b5791f3a842b174931261a47 Mon Sep 17 00:00:00 2001 From: Hubert Chathi Date: Fri, 25 Oct 2019 10:47:02 -0400 Subject: remove unneeded imports --- tests/storage/test_e2e_room_keys.py | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) (limited to 'tests') diff --git a/tests/storage/test_e2e_room_keys.py b/tests/storage/test_e2e_room_keys.py index 9935ac59ce..d128fde441 100644 --- a/tests/storage/test_e2e_room_keys.py +++ b/tests/storage/test_e2e_room_keys.py @@ -13,9 +13,7 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.internet import defer - -from tests import unittest, utils +from tests import unittest # sample room_key data for use in the tests room_key = { -- cgit 1.4.1 From d0d8a22c13427cce341dbb7ae1d92d2c0ae709c3 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Mon, 28 Oct 2019 13:33:04 +0000 Subject: Quick fix to ensure cache descriptors always return deferreds --- synapse/push/bulk_push_rule_evaluator.py | 2 +- synapse/storage/data_stores/main/roommember.py | 2 +- synapse/util/caches/descriptors.py | 4 ++-- tests/util/caches/test_descriptors.py | 4 ++-- 4 files changed, 6 insertions(+), 6 deletions(-) (limited to 'tests') diff --git a/synapse/push/bulk_push_rule_evaluator.py b/synapse/push/bulk_push_rule_evaluator.py index 22491f3700..2bbdd11941 100644 --- a/synapse/push/bulk_push_rule_evaluator.py +++ b/synapse/push/bulk_push_rule_evaluator.py @@ -79,7 +79,7 @@ class BulkPushRuleEvaluator(object): dict of user_id -> push_rules """ room_id = event.room_id - rules_for_room = self._get_rules_for_room(room_id) + rules_for_room = yield self._get_rules_for_room(room_id) rules_by_user = yield rules_for_room.get_rules(event, context) diff --git a/synapse/storage/data_stores/main/roommember.py b/synapse/storage/data_stores/main/roommember.py index e47ab604dd..bc04bfd7d4 100644 --- a/synapse/storage/data_stores/main/roommember.py +++ b/synapse/storage/data_stores/main/roommember.py @@ -720,7 +720,7 @@ class RoomMemberWorkerStore(EventsWorkerStore): # See bulk_get_push_rules_for_room for how we work around this. assert state_group is not None - cache = self._get_joined_hosts_cache(room_id) + cache = yield self._get_joined_hosts_cache(room_id) joined_hosts = yield cache.get_destinations(state_entry) return joined_hosts diff --git a/synapse/util/caches/descriptors.py b/synapse/util/caches/descriptors.py index 5ac2530a6a..5a8da449b2 100644 --- a/synapse/util/caches/descriptors.py +++ b/synapse/util/caches/descriptors.py @@ -438,7 +438,7 @@ class CacheDescriptor(_CacheDescriptorBase): if isinstance(cached_result_d, ObservableDeferred): observer = cached_result_d.observe() else: - observer = cached_result_d + observer = defer.succeed(cached_result_d) except KeyError: ret = defer.maybeDeferred( @@ -618,7 +618,7 @@ class CacheListDescriptor(_CacheDescriptorBase): ) return make_deferred_yieldable(d) else: - return results + return defer.succeed(results) obj.__dict__[self.orig.__name__] = wrapped diff --git a/tests/util/caches/test_descriptors.py b/tests/util/caches/test_descriptors.py index 5713870f48..f907903511 100644 --- a/tests/util/caches/test_descriptors.py +++ b/tests/util/caches/test_descriptors.py @@ -325,9 +325,9 @@ class DescriptorTestCase(unittest.TestCase): self.assertEqual(len(obj.fn.cache.cache), 3) r = obj.fn(1, 2) - self.assertEqual(r, ["spam", "eggs"]) + self.assertEqual(r.result, ["spam", "eggs"]) r = obj.fn(1, 3) - self.assertEqual(r, ["chips"]) + self.assertEqual(r.result, ["chips"]) obj.mock.assert_not_called() def test_cache_iterable_with_sync_exception(self): -- cgit 1.4.1 From 3f33879be4697666a304a472d090d4def0c671d0 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Tue, 29 Oct 2019 14:05:32 +0000 Subject: Port federation_server to async/await --- synapse/federation/federation_server.py | 205 ++++++++++++++------------------ tests/handlers/test_typing.py | 3 + 2 files changed, 90 insertions(+), 118 deletions(-) (limited to 'tests') diff --git a/synapse/federation/federation_server.py b/synapse/federation/federation_server.py index 5fc7c1d67b..15c1fa0a51 100644 --- a/synapse/federation/federation_server.py +++ b/synapse/federation/federation_server.py @@ -21,7 +21,6 @@ from six import iteritems from canonicaljson import json from prometheus_client import Counter -from twisted.internet import defer from twisted.internet.abstract import isIPAddress from twisted.python import failure @@ -86,14 +85,12 @@ class FederationServer(FederationBase): # come in waves. self._state_resp_cache = ResponseCache(hs, "state_resp", timeout_ms=30000) - @defer.inlineCallbacks - @log_function - def on_backfill_request(self, origin, room_id, versions, limit): - with (yield self._server_linearizer.queue((origin, room_id))): + async def on_backfill_request(self, origin, room_id, versions, limit): + with (await self._server_linearizer.queue((origin, room_id))): origin_host, _ = parse_server_name(origin) - yield self.check_server_matches_acl(origin_host, room_id) + await self.check_server_matches_acl(origin_host, room_id) - pdus = yield self.handler.on_backfill_request( + pdus = await self.handler.on_backfill_request( origin, room_id, versions, limit ) @@ -101,9 +98,7 @@ class FederationServer(FederationBase): return 200, res - @defer.inlineCallbacks - @log_function - def on_incoming_transaction(self, origin, transaction_data): + async def on_incoming_transaction(self, origin, transaction_data): # keep this as early as possible to make the calculated origin ts as # accurate as possible. request_time = self._clock.time_msec() @@ -118,18 +113,17 @@ class FederationServer(FederationBase): # use a linearizer to ensure that we don't process the same transaction # multiple times in parallel. with ( - yield self._transaction_linearizer.queue( + await self._transaction_linearizer.queue( (origin, transaction.transaction_id) ) ): - result = yield self._handle_incoming_transaction( + result = await self._handle_incoming_transaction( origin, transaction, request_time ) return result - @defer.inlineCallbacks - def _handle_incoming_transaction(self, origin, transaction, request_time): + async def _handle_incoming_transaction(self, origin, transaction, request_time): """ Process an incoming transaction and return the HTTP response Args: @@ -140,7 +134,7 @@ class FederationServer(FederationBase): Returns: Deferred[(int, object)]: http response code and body """ - response = yield self.transaction_actions.have_responded(origin, transaction) + response = await self.transaction_actions.have_responded(origin, transaction) if response: logger.debug( @@ -159,7 +153,7 @@ class FederationServer(FederationBase): logger.info("Transaction PDU or EDU count too large. Returning 400") response = {} - yield self.transaction_actions.set_response( + await self.transaction_actions.set_response( origin, transaction, 400, response ) return 400, response @@ -195,7 +189,7 @@ class FederationServer(FederationBase): continue try: - room_version = yield self.store.get_room_version(room_id) + room_version = await self.store.get_room_version(room_id) except NotFoundError: logger.info("Ignoring PDU for unknown room_id: %s", room_id) continue @@ -221,11 +215,10 @@ class FederationServer(FederationBase): # require callouts to other servers to fetch missing events), but # impose a limit to avoid going too crazy with ram/cpu. - @defer.inlineCallbacks - def process_pdus_for_room(room_id): + async def process_pdus_for_room(room_id): logger.debug("Processing PDUs for %s", room_id) try: - yield self.check_server_matches_acl(origin_host, room_id) + await self.check_server_matches_acl(origin_host, room_id) except AuthError as e: logger.warn("Ignoring PDUs for room %s from banned server", room_id) for pdu in pdus_by_room[room_id]: @@ -237,7 +230,7 @@ class FederationServer(FederationBase): event_id = pdu.event_id with nested_logging_context(event_id): try: - yield self._handle_received_pdu(origin, pdu) + await self._handle_received_pdu(origin, pdu) pdu_results[event_id] = {} except FederationError as e: logger.warn("Error handling PDU %s: %s", event_id, e) @@ -251,36 +244,33 @@ class FederationServer(FederationBase): exc_info=(f.type, f.value, f.getTracebackObject()), ) - yield concurrently_execute( + await concurrently_execute( process_pdus_for_room, pdus_by_room.keys(), TRANSACTION_CONCURRENCY_LIMIT ) if hasattr(transaction, "edus"): for edu in (Edu(**x) for x in transaction.edus): - yield self.received_edu(origin, edu.edu_type, edu.content) + await self.received_edu(origin, edu.edu_type, edu.content) response = {"pdus": pdu_results} logger.debug("Returning: %s", str(response)) - yield self.transaction_actions.set_response(origin, transaction, 200, response) + await self.transaction_actions.set_response(origin, transaction, 200, response) return 200, response - @defer.inlineCallbacks - def received_edu(self, origin, edu_type, content): + async def received_edu(self, origin, edu_type, content): received_edus_counter.inc() - yield self.registry.on_edu(edu_type, origin, content) + await self.registry.on_edu(edu_type, origin, content) - @defer.inlineCallbacks - @log_function - def on_context_state_request(self, origin, room_id, event_id): + async def on_context_state_request(self, origin, room_id, event_id): if not event_id: raise NotImplementedError("Specify an event") origin_host, _ = parse_server_name(origin) - yield self.check_server_matches_acl(origin_host, room_id) + await self.check_server_matches_acl(origin_host, room_id) - in_room = yield self.auth.check_host_in_room(room_id, origin) + in_room = await self.auth.check_host_in_room(room_id, origin) if not in_room: raise AuthError(403, "Host not in room.") @@ -289,8 +279,8 @@ class FederationServer(FederationBase): # in the cache so we could return it without waiting for the linearizer # - but that's non-trivial to get right, and anyway somewhat defeats # the point of the linearizer. - with (yield self._server_linearizer.queue((origin, room_id))): - resp = yield self._state_resp_cache.wrap( + with (await self._server_linearizer.queue((origin, room_id))): + resp = await self._state_resp_cache.wrap( (room_id, event_id), self._on_context_state_request_compute, room_id, @@ -299,65 +289,58 @@ class FederationServer(FederationBase): return 200, resp - @defer.inlineCallbacks - def on_state_ids_request(self, origin, room_id, event_id): + async def on_state_ids_request(self, origin, room_id, event_id): if not event_id: raise NotImplementedError("Specify an event") origin_host, _ = parse_server_name(origin) - yield self.check_server_matches_acl(origin_host, room_id) + await self.check_server_matches_acl(origin_host, room_id) - in_room = yield self.auth.check_host_in_room(room_id, origin) + in_room = await self.auth.check_host_in_room(room_id, origin) if not in_room: raise AuthError(403, "Host not in room.") - state_ids = yield self.handler.get_state_ids_for_pdu(room_id, event_id) - auth_chain_ids = yield self.store.get_auth_chain_ids(state_ids) + state_ids = await self.handler.get_state_ids_for_pdu(room_id, event_id) + auth_chain_ids = await self.store.get_auth_chain_ids(state_ids) return 200, {"pdu_ids": state_ids, "auth_chain_ids": auth_chain_ids} - @defer.inlineCallbacks - def _on_context_state_request_compute(self, room_id, event_id): - pdus = yield self.handler.get_state_for_pdu(room_id, event_id) - auth_chain = yield self.store.get_auth_chain([pdu.event_id for pdu in pdus]) + async def _on_context_state_request_compute(self, room_id, event_id): + pdus = await self.handler.get_state_for_pdu(room_id, event_id) + auth_chain = await self.store.get_auth_chain([pdu.event_id for pdu in pdus]) return { "pdus": [pdu.get_pdu_json() for pdu in pdus], "auth_chain": [pdu.get_pdu_json() for pdu in auth_chain], } - @defer.inlineCallbacks - @log_function - def on_pdu_request(self, origin, event_id): - pdu = yield self.handler.get_persisted_pdu(origin, event_id) + async def on_pdu_request(self, origin, event_id): + pdu = await self.handler.get_persisted_pdu(origin, event_id) if pdu: return 200, self._transaction_from_pdus([pdu]).get_dict() else: return 404, "" - @defer.inlineCallbacks - def on_query_request(self, query_type, args): + async def on_query_request(self, query_type, args): received_queries_counter.labels(query_type).inc() - resp = yield self.registry.on_query(query_type, args) + resp = await self.registry.on_query(query_type, args) return 200, resp - @defer.inlineCallbacks - def on_make_join_request(self, origin, room_id, user_id, supported_versions): + async def on_make_join_request(self, origin, room_id, user_id, supported_versions): origin_host, _ = parse_server_name(origin) - yield self.check_server_matches_acl(origin_host, room_id) + await self.check_server_matches_acl(origin_host, room_id) - room_version = yield self.store.get_room_version(room_id) + room_version = await self.store.get_room_version(room_id) if room_version not in supported_versions: logger.warn("Room version %s not in %s", room_version, supported_versions) raise IncompatibleRoomVersionError(room_version=room_version) - pdu = yield self.handler.on_make_join_request(origin, room_id, user_id) + pdu = await self.handler.on_make_join_request(origin, room_id, user_id) time_now = self._clock.time_msec() return {"event": pdu.get_pdu_json(time_now), "room_version": room_version} - @defer.inlineCallbacks - def on_invite_request(self, origin, content, room_version): + async def on_invite_request(self, origin, content, room_version): if room_version not in KNOWN_ROOM_VERSIONS: raise SynapseError( 400, @@ -369,28 +352,27 @@ class FederationServer(FederationBase): pdu = event_from_pdu_json(content, format_ver) origin_host, _ = parse_server_name(origin) - yield self.check_server_matches_acl(origin_host, pdu.room_id) - pdu = yield self._check_sigs_and_hash(room_version, pdu) - ret_pdu = yield self.handler.on_invite_request(origin, pdu) + await self.check_server_matches_acl(origin_host, pdu.room_id) + pdu = await self._check_sigs_and_hash(room_version, pdu) + ret_pdu = await self.handler.on_invite_request(origin, pdu) time_now = self._clock.time_msec() return {"event": ret_pdu.get_pdu_json(time_now)} - @defer.inlineCallbacks - def on_send_join_request(self, origin, content, room_id): + async def on_send_join_request(self, origin, content, room_id): logger.debug("on_send_join_request: content: %s", content) - room_version = yield self.store.get_room_version(room_id) + room_version = await self.store.get_room_version(room_id) format_ver = room_version_to_event_format(room_version) pdu = event_from_pdu_json(content, format_ver) origin_host, _ = parse_server_name(origin) - yield self.check_server_matches_acl(origin_host, pdu.room_id) + await self.check_server_matches_acl(origin_host, pdu.room_id) logger.debug("on_send_join_request: pdu sigs: %s", pdu.signatures) - pdu = yield self._check_sigs_and_hash(room_version, pdu) + pdu = await self._check_sigs_and_hash(room_version, pdu) - res_pdus = yield self.handler.on_send_join_request(origin, pdu) + res_pdus = await self.handler.on_send_join_request(origin, pdu) time_now = self._clock.time_msec() return ( 200, @@ -402,48 +384,44 @@ class FederationServer(FederationBase): }, ) - @defer.inlineCallbacks - def on_make_leave_request(self, origin, room_id, user_id): + async def on_make_leave_request(self, origin, room_id, user_id): origin_host, _ = parse_server_name(origin) - yield self.check_server_matches_acl(origin_host, room_id) - pdu = yield self.handler.on_make_leave_request(origin, room_id, user_id) + await self.check_server_matches_acl(origin_host, room_id) + pdu = await self.handler.on_make_leave_request(origin, room_id, user_id) - room_version = yield self.store.get_room_version(room_id) + room_version = await self.store.get_room_version(room_id) time_now = self._clock.time_msec() return {"event": pdu.get_pdu_json(time_now), "room_version": room_version} - @defer.inlineCallbacks - def on_send_leave_request(self, origin, content, room_id): + async def on_send_leave_request(self, origin, content, room_id): logger.debug("on_send_leave_request: content: %s", content) - room_version = yield self.store.get_room_version(room_id) + room_version = await self.store.get_room_version(room_id) format_ver = room_version_to_event_format(room_version) pdu = event_from_pdu_json(content, format_ver) origin_host, _ = parse_server_name(origin) - yield self.check_server_matches_acl(origin_host, pdu.room_id) + await self.check_server_matches_acl(origin_host, pdu.room_id) logger.debug("on_send_leave_request: pdu sigs: %s", pdu.signatures) - pdu = yield self._check_sigs_and_hash(room_version, pdu) + pdu = await self._check_sigs_and_hash(room_version, pdu) - yield self.handler.on_send_leave_request(origin, pdu) + await self.handler.on_send_leave_request(origin, pdu) return 200, {} - @defer.inlineCallbacks - def on_event_auth(self, origin, room_id, event_id): - with (yield self._server_linearizer.queue((origin, room_id))): + async def on_event_auth(self, origin, room_id, event_id): + with (await self._server_linearizer.queue((origin, room_id))): origin_host, _ = parse_server_name(origin) - yield self.check_server_matches_acl(origin_host, room_id) + await self.check_server_matches_acl(origin_host, room_id) time_now = self._clock.time_msec() - auth_pdus = yield self.handler.on_event_auth(event_id) + auth_pdus = await self.handler.on_event_auth(event_id) res = {"auth_chain": [a.get_pdu_json(time_now) for a in auth_pdus]} return 200, res - @defer.inlineCallbacks - def on_query_auth_request(self, origin, content, room_id, event_id): + async def on_query_auth_request(self, origin, content, room_id, event_id): """ Content is a dict with keys:: auth_chain (list): A list of events that give the auth chain. @@ -462,22 +440,22 @@ class FederationServer(FederationBase): Returns: Deferred: Results in `dict` with the same format as `content` """ - with (yield self._server_linearizer.queue((origin, room_id))): + with (await self._server_linearizer.queue((origin, room_id))): origin_host, _ = parse_server_name(origin) - yield self.check_server_matches_acl(origin_host, room_id) + await self.check_server_matches_acl(origin_host, room_id) - room_version = yield self.store.get_room_version(room_id) + room_version = await self.store.get_room_version(room_id) format_ver = room_version_to_event_format(room_version) auth_chain = [ event_from_pdu_json(e, format_ver) for e in content["auth_chain"] ] - signed_auth = yield self._check_sigs_and_hash_and_fetch( + signed_auth = await self._check_sigs_and_hash_and_fetch( origin, auth_chain, outlier=True, room_version=room_version ) - ret = yield self.handler.on_query_auth( + ret = await self.handler.on_query_auth( origin, event_id, room_id, @@ -503,16 +481,14 @@ class FederationServer(FederationBase): return self.on_query_request("user_devices", user_id) @trace - @defer.inlineCallbacks - @log_function - def on_claim_client_keys(self, origin, content): + async def on_claim_client_keys(self, origin, content): query = [] for user_id, device_keys in content.get("one_time_keys", {}).items(): for device_id, algorithm in device_keys.items(): query.append((user_id, device_id, algorithm)) log_kv({"message": "Claiming one time keys.", "user, device pairs": query}) - results = yield self.store.claim_e2e_one_time_keys(query) + results = await self.store.claim_e2e_one_time_keys(query) json_result = {} for user_id, device_keys in results.items(): @@ -536,14 +512,12 @@ class FederationServer(FederationBase): return {"one_time_keys": json_result} - @defer.inlineCallbacks - @log_function - def on_get_missing_events( + async def on_get_missing_events( self, origin, room_id, earliest_events, latest_events, limit ): - with (yield self._server_linearizer.queue((origin, room_id))): + with (await self._server_linearizer.queue((origin, room_id))): origin_host, _ = parse_server_name(origin) - yield self.check_server_matches_acl(origin_host, room_id) + await self.check_server_matches_acl(origin_host, room_id) logger.info( "on_get_missing_events: earliest_events: %r, latest_events: %r," @@ -553,7 +527,7 @@ class FederationServer(FederationBase): limit, ) - missing_events = yield self.handler.on_get_missing_events( + missing_events = await self.handler.on_get_missing_events( origin, room_id, earliest_events, latest_events, limit ) @@ -586,8 +560,7 @@ class FederationServer(FederationBase): destination=None, ) - @defer.inlineCallbacks - def _handle_received_pdu(self, origin, pdu): + async def _handle_received_pdu(self, origin, pdu): """ Process a PDU received in a federation /send/ transaction. If the event is invalid, then this method throws a FederationError. @@ -640,37 +613,34 @@ class FederationServer(FederationBase): logger.info("Accepting join PDU %s from %s", pdu.event_id, origin) # We've already checked that we know the room version by this point - room_version = yield self.store.get_room_version(pdu.room_id) + room_version = await self.store.get_room_version(pdu.room_id) # Check signature. try: - pdu = yield self._check_sigs_and_hash(room_version, pdu) + pdu = await self._check_sigs_and_hash(room_version, pdu) except SynapseError as e: raise FederationError("ERROR", e.code, e.msg, affected=pdu.event_id) - yield self.handler.on_receive_pdu(origin, pdu, sent_to_us_directly=True) + await self.handler.on_receive_pdu(origin, pdu, sent_to_us_directly=True) def __str__(self): return "" % self.server_name - @defer.inlineCallbacks - def exchange_third_party_invite( + async def exchange_third_party_invite( self, sender_user_id, target_user_id, room_id, signed ): - ret = yield self.handler.exchange_third_party_invite( + ret = await self.handler.exchange_third_party_invite( sender_user_id, target_user_id, room_id, signed ) return ret - @defer.inlineCallbacks - def on_exchange_third_party_invite_request(self, room_id, event_dict): - ret = yield self.handler.on_exchange_third_party_invite_request( + async def on_exchange_third_party_invite_request(self, room_id, event_dict): + ret = await self.handler.on_exchange_third_party_invite_request( room_id, event_dict ) return ret - @defer.inlineCallbacks - def check_server_matches_acl(self, server_name, room_id): + async def check_server_matches_acl(self, server_name, room_id): """Check if the given server is allowed by the server ACLs in the room Args: @@ -680,13 +650,13 @@ class FederationServer(FederationBase): Raises: AuthError if the server does not match the ACL """ - state_ids = yield self.store.get_current_state_ids(room_id) + state_ids = await self.store.get_current_state_ids(room_id) acl_event_id = state_ids.get((EventTypes.ServerACL, "")) if not acl_event_id: return - acl_event = yield self.store.get_event(acl_event_id) + acl_event = await self.store.get_event(acl_event_id) if server_matches_acl_event(server_name, acl_event): return @@ -799,15 +769,14 @@ class FederationHandlerRegistry(object): self.query_handlers[query_type] = handler - @defer.inlineCallbacks - def on_edu(self, edu_type, origin, content): + async def on_edu(self, edu_type, origin, content): handler = self.edu_handlers.get(edu_type) if not handler: logger.warn("No handler registered for EDU type %s", edu_type) with start_active_span_from_edu(content, "handle_edu"): try: - yield handler(origin, content) + await handler(origin, content) except SynapseError as e: logger.info("Failed to handle edu %r: %r", edu_type, e) except Exception: diff --git a/tests/handlers/test_typing.py b/tests/handlers/test_typing.py index 67f1013051..f360c8e965 100644 --- a/tests/handlers/test_typing.py +++ b/tests/handlers/test_typing.py @@ -144,6 +144,9 @@ class TypingNotificationsTestCase(unittest.HomeserverTestCase): self.datastore.get_to_device_stream_token = lambda: 0 self.datastore.get_new_device_msgs_for_remote = lambda *args, **kargs: ([], 0) self.datastore.delete_device_msgs_for_remote = lambda *args, **kargs: None + self.datastore.set_received_txn_response = lambda *args, **kwargs: defer.succeed( + None + ) def test_started_typing_local(self): self.room_members = [U_APPLE, U_BANANA] -- cgit 1.4.1 From 326b3dace77aeb36e516ea9b04ba1baa171bcb47 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 30 Oct 2019 11:35:46 +0000 Subject: Make ObservableDeferred.observe() always return deferred. This makes it easier to use in an async/await world. Also fixes a bug where cache descriptors would occaisonally return a raw value rather than a deferred. --- synapse/util/async_helpers.py | 7 ++----- tests/storage/test__base.py | 2 +- tests/util/caches/test_descriptors.py | 4 ++-- 3 files changed, 5 insertions(+), 8 deletions(-) (limited to 'tests') diff --git a/synapse/util/async_helpers.py b/synapse/util/async_helpers.py index 7659eaeb42..fd75ba27ad 100644 --- a/synapse/util/async_helpers.py +++ b/synapse/util/async_helpers.py @@ -86,11 +86,8 @@ class ObservableDeferred(object): deferred.addCallbacks(callback, errback) - def observe(self): + def observe(self) -> defer.Deferred: """Observe the underlying deferred. - - Can return either a deferred if the underlying deferred is still pending - (or has failed), or the actual value. Callers may need to use maybeDeferred. """ if not self._result: d = defer.Deferred() @@ -105,7 +102,7 @@ class ObservableDeferred(object): return d else: success, res = self._result - return res if success else defer.fail(res) + return defer.succeed(res) if success else defer.fail(res) def observers(self): return self._observers diff --git a/tests/storage/test__base.py b/tests/storage/test__base.py index dd49a14524..9b81b536f5 100644 --- a/tests/storage/test__base.py +++ b/tests/storage/test__base.py @@ -197,7 +197,7 @@ class CacheDecoratorTestCase(unittest.TestCase): a.func.prefill(("foo",), ObservableDeferred(d)) - self.assertEquals(a.func("foo"), d.result) + self.assertEquals(a.func("foo").result, d.result) self.assertEquals(callcount[0], 0) @defer.inlineCallbacks diff --git a/tests/util/caches/test_descriptors.py b/tests/util/caches/test_descriptors.py index f907903511..39e360fe24 100644 --- a/tests/util/caches/test_descriptors.py +++ b/tests/util/caches/test_descriptors.py @@ -310,14 +310,14 @@ class DescriptorTestCase(unittest.TestCase): obj.mock.return_value = ["spam", "eggs"] r = obj.fn(1, 2) - self.assertEqual(r, ["spam", "eggs"]) + self.assertEqual(r.result, ["spam", "eggs"]) obj.mock.assert_called_once_with(1, 2) obj.mock.reset_mock() # a call with different params should call the mock again obj.mock.return_value = ["chips"] r = obj.fn(1, 3) - self.assertEqual(r, ["chips"]) + self.assertEqual(r.result, ["chips"]) obj.mock.assert_called_once_with(1, 3) obj.mock.reset_mock() -- cgit 1.4.1 From a8d16f6c00d5adb204af5fa73ffaea40eea4b632 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 30 Oct 2019 13:33:38 +0000 Subject: Review comments --- synapse/server.py | 5 ++--- synapse/storage/__init__.py | 4 ++-- synapse/storage/data_stores/main/events.py | 19 ++++++++++++++----- synapse/storage/persist_events.py | 13 ++++++++++--- tests/crypto/test_keyring.py | 4 ++-- tests/test_federation.py | 11 +++++++---- 6 files changed, 37 insertions(+), 19 deletions(-) (limited to 'tests') diff --git a/synapse/server.py b/synapse/server.py index 54a7f4aa5f..0b81af646c 100644 --- a/synapse/server.py +++ b/synapse/server.py @@ -226,7 +226,6 @@ class HomeServer(object): self.admin_redaction_ratelimiter = Ratelimiter() self.registration_ratelimiter = Ratelimiter() - self.datastore = None self.datastores = None # Other kwargs are explicit dependencies @@ -236,8 +235,8 @@ class HomeServer(object): def setup(self): logger.info("Setting up.") with self.get_db_conn() as conn: - self.datastore = self.DATASTORE_CLASS(conn, self) - self.datastores = DataStores(self.datastore, conn, self) + datastore = self.DATASTORE_CLASS(conn, self) + self.datastores = DataStores(datastore, conn, self) conn.commit() logger.info("Finished setting up.") diff --git a/synapse/storage/__init__.py b/synapse/storage/__init__.py index a58187a76f..a6429d17ed 100644 --- a/synapse/storage/__init__.py +++ b/synapse/storage/__init__.py @@ -29,7 +29,7 @@ stored in `synapse.storage.schema`. from synapse.storage.data_stores import DataStores from synapse.storage.data_stores.main import DataStore -from synapse.storage.persist_events import EventsPersistenceStore +from synapse.storage.persist_events import EventsPersistenceStorage __all__ = ["DataStores", "DataStore"] @@ -44,7 +44,7 @@ class Storage(object): # interfaces. self.main = stores.main - self.persistence = EventsPersistenceStore(hs, stores) + self.persistence = EventsPersistenceStorage(hs, stores) def are_all_users_on_domain(txn, database_engine, domain): diff --git a/synapse/storage/data_stores/main/events.py b/synapse/storage/data_stores/main/events.py index 6304531cd5..813f34528c 100644 --- a/synapse/storage/data_stores/main/events.py +++ b/synapse/storage/data_stores/main/events.py @@ -146,7 +146,7 @@ class EventsStore( @_retry_on_integrity_error @defer.inlineCallbacks - def _persist_events( + def _persist_events_and_state_updates( self, events_and_contexts, current_state_for_room, @@ -155,18 +155,27 @@ class EventsStore( backfilled=False, delete_existing=False, ): - """Persist events to db + """Persist a set of events alongside updates to the current state and + forward extremities tables. Args: events_and_contexts (list[(EventBase, EventContext)]): - backfilled (bool): + current_state_for_room (dict[str, dict]): Map from room_id to the + current state of the room based on forward extremities + state_delta_for_room (dict[str, tuple]): Map from room_id to tuple + of `(to_delete, to_insert)` where to_delete is a list + of type/state keys to remove from current state, and to_insert + is a map (type,key)->event_id giving the state delta in each + room. + new_forward_extremities (dict[str, list[str]]): Map from room_id + to list of event IDs that are the new forward extremities of + the room. + backfilled (bool) delete_existing (bool): Returns: Deferred: resolves when the events have been persisted """ - if not events_and_contexts: - return # We want to calculate the stream orderings as late as possible, as # we only notify after all events with a lesser stream ordering have diff --git a/synapse/storage/persist_events.py b/synapse/storage/persist_events.py index 9a63953d4d..cf66225574 100644 --- a/synapse/storage/persist_events.py +++ b/synapse/storage/persist_events.py @@ -171,7 +171,13 @@ class _EventPeristenceQueue(object): pass -class EventsPersistenceStore(object): +class EventsPersistenceStorage(object): + """High level interface for handling persisting newly received events. + + Takes care of batching up events by room, and calculating the necessary + current state and forward extremity changes. + """ + def __init__(self, hs, stores: DataStores): # We ultimately want to split out the state store from the main store, # so we use separate variables here even though they point to the same @@ -257,7 +263,8 @@ class EventsPersistenceStore(object): def _persist_events( self, events_and_contexts, backfilled=False, delete_existing=False ): - """Persist events to db + """Calculates the change to current state and forward extremities, and + persists the given events and with those updates. Args: events_and_contexts (list[(EventBase, EventContext)]): @@ -399,7 +406,7 @@ class EventsPersistenceStore(object): if current_state is not None: current_state_for_room[room_id] = current_state - yield self.main_store._persist_events( + yield self.main_store._persist_events_and_state_updates( chunk, current_state_for_room=current_state_for_room, state_delta_for_room=state_delta_for_room, diff --git a/tests/crypto/test_keyring.py b/tests/crypto/test_keyring.py index c4f0bbd3dd..8efd39c7f7 100644 --- a/tests/crypto/test_keyring.py +++ b/tests/crypto/test_keyring.py @@ -178,7 +178,7 @@ class KeyringTestCase(unittest.HomeserverTestCase): kr = keyring.Keyring(self.hs) key1 = signedjson.key.generate_signing_key(1) - r = self.hs.datastore.store_server_verify_keys( + r = self.hs.get_datastore().store_server_verify_keys( "server9", time.time() * 1000, [("server9", get_key_id(key1), FetchKeyResult(get_verify_key(key1), 1000))], @@ -209,7 +209,7 @@ class KeyringTestCase(unittest.HomeserverTestCase): ) key1 = signedjson.key.generate_signing_key(1) - r = self.hs.datastore.store_server_verify_keys( + r = self.hs.get_datastore().store_server_verify_keys( "server9", time.time() * 1000, [("server9", get_key_id(key1), FetchKeyResult(get_verify_key(key1), None))], diff --git a/tests/test_federation.py b/tests/test_federation.py index a73f18f88e..d1acb16f30 100644 --- a/tests/test_federation.py +++ b/tests/test_federation.py @@ -36,7 +36,8 @@ class MessageAcceptTests(unittest.TestCase): # Figure out what the most recent event is most_recent = self.successResultOf( maybeDeferred( - self.homeserver.datastore.get_latest_event_ids_in_room, self.room_id + self.homeserver.get_datastore().get_latest_event_ids_in_room, + self.room_id, ) )[0] @@ -75,7 +76,8 @@ class MessageAcceptTests(unittest.TestCase): self.assertEqual( self.successResultOf( maybeDeferred( - self.homeserver.datastore.get_latest_event_ids_in_room, self.room_id + self.homeserver.get_datastore().get_latest_event_ids_in_room, + self.room_id, ) )[0], "$join:test.serv", @@ -97,7 +99,8 @@ class MessageAcceptTests(unittest.TestCase): # Figure out what the most recent event is most_recent = self.successResultOf( maybeDeferred( - self.homeserver.datastore.get_latest_event_ids_in_room, self.room_id + self.homeserver.get_datastore().get_latest_event_ids_in_room, + self.room_id, ) )[0] @@ -137,6 +140,6 @@ class MessageAcceptTests(unittest.TestCase): # Make sure the invalid event isn't there extrem = maybeDeferred( - self.homeserver.datastore.get_latest_event_ids_in_room, self.room_id + self.homeserver.get_datastore().get_latest_event_ids_in_room, self.room_id ) self.assertEqual(self.successResultOf(extrem)[0], "$join:test.serv") -- cgit 1.4.1 From 69f0054ce675bd9d35104c39af9fae9a908b7f33 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 23 Oct 2019 17:25:54 +0100 Subject: Port to use state storage --- synapse/handlers/admin.py | 7 +- synapse/handlers/device.py | 3 +- synapse/handlers/events.py | 6 +- synapse/handlers/federation.py | 19 ++--- synapse/handlers/initial_sync.py | 14 ++-- synapse/handlers/message.py | 10 +-- synapse/handlers/pagination.py | 6 +- synapse/handlers/room.py | 6 +- synapse/handlers/search.py | 12 ++-- synapse/handlers/sync.py | 20 +++--- synapse/notifier.py | 6 +- synapse/push/httppusher.py | 3 +- synapse/push/mailer.py | 3 +- synapse/push/push_tools.py | 9 +-- synapse/state/__init__.py | 13 ++-- synapse/visibility.py | 30 ++++---- tests/storage/test_state.py | 150 ++++++++++++++++++++++++++------------- tests/test_state.py | 3 + tests/test_visibility.py | 11 ++- 19 files changed, 216 insertions(+), 115 deletions(-) (limited to 'tests') diff --git a/synapse/handlers/admin.py b/synapse/handlers/admin.py index 1a87b58838..6407d56f8e 100644 --- a/synapse/handlers/admin.py +++ b/synapse/handlers/admin.py @@ -30,6 +30,9 @@ class AdminHandler(BaseHandler): def __init__(self, hs): super(AdminHandler, self).__init__(hs) + self.storage = hs.get_storage() + self.state_store = self.storage.state + @defer.inlineCallbacks def get_whois(self, user): connections = [] @@ -205,7 +208,7 @@ class AdminHandler(BaseHandler): from_key = events[-1].internal_metadata.after - events = yield filter_events_for_client(self.store, user_id, events) + events = yield filter_events_for_client(self.storage, user_id, events) writer.write_events(room_id, events) @@ -241,7 +244,7 @@ class AdminHandler(BaseHandler): for event_id in extremities: if not event_to_unseen_prevs[event_id]: continue - state = yield self.store.get_state_for_event(event_id) + state = yield self.state_store.get_state_for_event(event_id) writer.write_state(room_id, event_id, state) return writer.finished() diff --git a/synapse/handlers/device.py b/synapse/handlers/device.py index 5f23ee4488..b3fd7e6249 100644 --- a/synapse/handlers/device.py +++ b/synapse/handlers/device.py @@ -46,6 +46,7 @@ class DeviceWorkerHandler(BaseHandler): self.hs = hs self.state = hs.get_state_handler() + self.state_store = hs.get_storage().state self._auth_handler = hs.get_auth_handler() @trace @@ -178,7 +179,7 @@ class DeviceWorkerHandler(BaseHandler): continue # mapping from event_id -> state_dict - prev_state_ids = yield self.store.get_state_ids_for_events(event_ids) + prev_state_ids = yield self.state_store.get_state_ids_for_events(event_ids) # Check if we've joined the room? If so we just blindly add all the users to # the "possibly changed" users. diff --git a/synapse/handlers/events.py b/synapse/handlers/events.py index 5e748687e3..45fe13c62f 100644 --- a/synapse/handlers/events.py +++ b/synapse/handlers/events.py @@ -147,6 +147,10 @@ class EventStreamHandler(BaseHandler): class EventHandler(BaseHandler): + def __init__(self, hs): + super(EventHandler, self).__init__(hs) + self.storage = hs.get_storage() + @defer.inlineCallbacks def get_event(self, user, room_id, event_id): """Retrieve a single specified event. @@ -172,7 +176,7 @@ class EventHandler(BaseHandler): is_peeking = user.to_string() not in users filtered = yield filter_events_for_client( - self.store, user.to_string(), [event], is_peeking=is_peeking + self.storage, user.to_string(), [event], is_peeking=is_peeking ) if not filtered: diff --git a/synapse/handlers/federation.py b/synapse/handlers/federation.py index 08276fdebf..4d9e33346d 100644 --- a/synapse/handlers/federation.py +++ b/synapse/handlers/federation.py @@ -110,6 +110,7 @@ class FederationHandler(BaseHandler): self.store = hs.get_datastore() self.storage = hs.get_storage() + self.state_store = self.storage.state self.federation_client = hs.get_federation_client() self.state_handler = hs.get_state_handler() self.server_name = hs.hostname @@ -325,7 +326,7 @@ class FederationHandler(BaseHandler): event_map = {event_id: pdu} try: # Get the state of the events we know about - ours = yield self.store.get_state_groups_ids(room_id, seen) + ours = yield self.state_store.get_state_groups_ids(room_id, seen) # state_maps is a list of mappings from (type, state_key) to event_id state_maps = list( @@ -889,7 +890,7 @@ class FederationHandler(BaseHandler): # We set `check_history_visibility_only` as we might otherwise get false # positives from users having been erased. filtered_extremities = yield filter_events_for_server( - self.store, + self.storage, self.server_name, list(extremities_events.values()), redact=False, @@ -1550,7 +1551,7 @@ class FederationHandler(BaseHandler): event_id, allow_none=False, check_room_id=room_id ) - state_groups = yield self.store.get_state_groups(room_id, [event_id]) + state_groups = yield self.state_store.get_state_groups(room_id, [event_id]) if state_groups: _, state = list(iteritems(state_groups)).pop() @@ -1579,7 +1580,7 @@ class FederationHandler(BaseHandler): event_id, allow_none=False, check_room_id=room_id ) - state_groups = yield self.store.get_state_groups_ids(room_id, [event_id]) + state_groups = yield self.state_store.get_state_groups_ids(room_id, [event_id]) if state_groups: _, state = list(state_groups.items()).pop() @@ -1607,7 +1608,7 @@ class FederationHandler(BaseHandler): events = yield self.store.get_backfill_events(room_id, pdu_list, limit) - events = yield filter_events_for_server(self.store, origin, events) + events = yield filter_events_for_server(self.storage, origin, events) return events @@ -1637,7 +1638,7 @@ class FederationHandler(BaseHandler): if not in_room: raise AuthError(403, "Host not in room.") - events = yield filter_events_for_server(self.store, origin, [event]) + events = yield filter_events_for_server(self.storage, origin, [event]) event = events[0] return event else: @@ -1903,7 +1904,7 @@ class FederationHandler(BaseHandler): # given state at the event. This should correctly handle cases # like bans, especially with state res v2. - state_sets = yield self.store.get_state_groups( + state_sets = yield self.state_store.get_state_groups( event.room_id, extrem_ids ) state_sets = list(state_sets.values()) @@ -1994,7 +1995,7 @@ class FederationHandler(BaseHandler): ) missing_events = yield filter_events_for_server( - self.store, origin, missing_events + self.storage, origin, missing_events ) return missing_events @@ -2235,7 +2236,7 @@ class FederationHandler(BaseHandler): # create a new state group as a delta from the existing one. prev_group = context.state_group - state_group = yield self.store.store_state_group( + state_group = yield self.state_store.store_state_group( event.event_id, event.room_id, prev_group=prev_group, diff --git a/synapse/handlers/initial_sync.py b/synapse/handlers/initial_sync.py index f991efeee3..49c9e031f9 100644 --- a/synapse/handlers/initial_sync.py +++ b/synapse/handlers/initial_sync.py @@ -43,6 +43,8 @@ class InitialSyncHandler(BaseHandler): self.validator = EventValidator() self.snapshot_cache = SnapshotCache() self._event_serializer = hs.get_event_client_serializer() + self.storage = hs.get_storage() + self.state_store = self.storage.state def snapshot_all_rooms( self, @@ -169,7 +171,7 @@ class InitialSyncHandler(BaseHandler): elif event.membership == Membership.LEAVE: room_end_token = "s%d" % (event.stream_ordering,) deferred_room_state = run_in_background( - self.store.get_state_for_events, [event.event_id] + self.state_store.get_state_for_events, [event.event_id] ) deferred_room_state.addCallback( lambda states: states[event.event_id] @@ -189,7 +191,9 @@ class InitialSyncHandler(BaseHandler): ) ).addErrback(unwrapFirstError) - messages = yield filter_events_for_client(self.store, user_id, messages) + messages = yield filter_events_for_client( + self.storage, user_id, messages + ) start_token = now_token.copy_and_replace("room_key", token) end_token = now_token.copy_and_replace("room_key", room_end_token) @@ -307,7 +311,7 @@ class InitialSyncHandler(BaseHandler): def _room_initial_sync_parted( self, user_id, room_id, pagin_config, membership, member_event_id, is_peeking ): - room_state = yield self.store.get_state_for_events([member_event_id]) + room_state = yield self.state_store.get_state_for_events([member_event_id]) room_state = room_state[member_event_id] @@ -322,7 +326,7 @@ class InitialSyncHandler(BaseHandler): ) messages = yield filter_events_for_client( - self.store, user_id, messages, is_peeking=is_peeking + self.storage, user_id, messages, is_peeking=is_peeking ) start_token = StreamToken.START.copy_and_replace("room_key", token) @@ -414,7 +418,7 @@ class InitialSyncHandler(BaseHandler): ) messages = yield filter_events_for_client( - self.store, user_id, messages, is_peeking=is_peeking + self.storage, user_id, messages, is_peeking=is_peeking ) start_token = now_token.copy_and_replace("room_key", token) diff --git a/synapse/handlers/message.py b/synapse/handlers/message.py index 7908a2d52c..6e2a360262 100644 --- a/synapse/handlers/message.py +++ b/synapse/handlers/message.py @@ -59,6 +59,8 @@ class MessageHandler(object): self.clock = hs.get_clock() self.state = hs.get_state_handler() self.store = hs.get_datastore() + self.storage = hs.get_storage() + self.state_store = self.storage.state self._event_serializer = hs.get_event_client_serializer() @defer.inlineCallbacks @@ -82,7 +84,7 @@ class MessageHandler(object): data = yield self.state.get_current_state(room_id, event_type, state_key) elif membership == Membership.LEAVE: key = (event_type, state_key) - room_state = yield self.store.get_state_for_events( + room_state = yield self.state_store.get_state_for_events( [membership_event_id], StateFilter.from_types([key]) ) data = room_state[membership_event_id].get(key) @@ -135,12 +137,12 @@ class MessageHandler(object): raise NotFoundError("Can't find event for token %s" % (at_token,)) visible_events = yield filter_events_for_client( - self.store, user_id, last_events + self.storage, user_id, last_events ) event = last_events[0] if visible_events: - room_state = yield self.store.get_state_for_events( + room_state = yield self.state_store.get_state_for_events( [event.event_id], state_filter=state_filter ) room_state = room_state[event.event_id] @@ -161,7 +163,7 @@ class MessageHandler(object): ) room_state = yield self.store.get_events(state_ids.values()) elif membership == Membership.LEAVE: - room_state = yield self.store.get_state_for_events( + room_state = yield self.state_store.get_state_for_events( [membership_event_id], state_filter=state_filter ) room_state = room_state[membership_event_id] diff --git a/synapse/handlers/pagination.py b/synapse/handlers/pagination.py index 5744f4579d..b7185fe7a0 100644 --- a/synapse/handlers/pagination.py +++ b/synapse/handlers/pagination.py @@ -69,6 +69,8 @@ class PaginationHandler(object): self.hs = hs self.auth = hs.get_auth() self.store = hs.get_datastore() + self.storage = hs.get_storage() + self.state_store = self.storage.state self.clock = hs.get_clock() self._server_name = hs.hostname @@ -255,7 +257,7 @@ class PaginationHandler(object): events = event_filter.filter(events) events = yield filter_events_for_client( - self.store, user_id, events, is_peeking=(member_event_id is None) + self.storage, user_id, events, is_peeking=(member_event_id is None) ) if not events: @@ -274,7 +276,7 @@ class PaginationHandler(object): (EventTypes.Member, event.sender) for event in events ) - state_ids = yield self.store.get_state_ids_for_event( + state_ids = yield self.state_store.get_state_ids_for_event( events[0].event_id, state_filter=state_filter ) diff --git a/synapse/handlers/room.py b/synapse/handlers/room.py index 2816bd8f87..84bad39815 100644 --- a/synapse/handlers/room.py +++ b/synapse/handlers/room.py @@ -822,6 +822,8 @@ class RoomContextHandler(object): def __init__(self, hs): self.hs = hs self.store = hs.get_datastore() + self.storage = hs.get_storage() + self.state_store = self.storage.state @defer.inlineCallbacks def get_event_context(self, user, room_id, event_id, limit, event_filter): @@ -848,7 +850,7 @@ class RoomContextHandler(object): def filter_evts(events): return filter_events_for_client( - self.store, user.to_string(), events, is_peeking=is_peeking + self.storage, user.to_string(), events, is_peeking=is_peeking ) event = yield self.store.get_event( @@ -890,7 +892,7 @@ class RoomContextHandler(object): # first? Shouldn't we be consistent with /sync? # https://github.com/matrix-org/matrix-doc/issues/687 - state = yield self.store.get_state_for_events( + state = yield self.state_store.get_state_for_events( [last_event_id], state_filter=state_filter ) results["state"] = list(state[last_event_id].values()) diff --git a/synapse/handlers/search.py b/synapse/handlers/search.py index cd5e90bacb..f4d8a60774 100644 --- a/synapse/handlers/search.py +++ b/synapse/handlers/search.py @@ -35,6 +35,8 @@ class SearchHandler(BaseHandler): def __init__(self, hs): super(SearchHandler, self).__init__(hs) self._event_serializer = hs.get_event_client_serializer() + self.storage = hs.get_storage() + self.state_store = self.storage.state @defer.inlineCallbacks def get_old_rooms_from_upgraded_room(self, room_id): @@ -221,7 +223,7 @@ class SearchHandler(BaseHandler): filtered_events = search_filter.filter([r["event"] for r in results]) events = yield filter_events_for_client( - self.store, user.to_string(), filtered_events + self.storage, user.to_string(), filtered_events ) events.sort(key=lambda e: -rank_map[e.event_id]) @@ -271,7 +273,7 @@ class SearchHandler(BaseHandler): filtered_events = search_filter.filter([r["event"] for r in results]) events = yield filter_events_for_client( - self.store, user.to_string(), filtered_events + self.storage, user.to_string(), filtered_events ) room_events.extend(events) @@ -340,11 +342,11 @@ class SearchHandler(BaseHandler): ) res["events_before"] = yield filter_events_for_client( - self.store, user.to_string(), res["events_before"] + self.storage, user.to_string(), res["events_before"] ) res["events_after"] = yield filter_events_for_client( - self.store, user.to_string(), res["events_after"] + self.storage, user.to_string(), res["events_after"] ) res["start"] = now_token.copy_and_replace( @@ -372,7 +374,7 @@ class SearchHandler(BaseHandler): [(EventTypes.Member, sender) for sender in senders] ) - state = yield self.store.get_state_for_event( + state = yield self.state_store.get_state_for_event( last_event_id, state_filter ) diff --git a/synapse/handlers/sync.py b/synapse/handlers/sync.py index d99160e9d7..43a082dcda 100644 --- a/synapse/handlers/sync.py +++ b/synapse/handlers/sync.py @@ -230,6 +230,8 @@ class SyncHandler(object): self.response_cache = ResponseCache(hs, "sync") self.state = hs.get_state_handler() self.auth = hs.get_auth() + self.storage = hs.get_storage() + self.state_store = self.storage.state # ExpiringCache((User, Device)) -> LruCache(state_key => event_id) self.lazy_loaded_members_cache = ExpiringCache( @@ -417,7 +419,7 @@ class SyncHandler(object): current_state_ids = frozenset(itervalues(current_state_ids)) recents = yield filter_events_for_client( - self.store, + self.storage, sync_config.user.to_string(), recents, always_include_ids=current_state_ids, @@ -470,7 +472,7 @@ class SyncHandler(object): current_state_ids = frozenset(itervalues(current_state_ids)) loaded_recents = yield filter_events_for_client( - self.store, + self.storage, sync_config.user.to_string(), loaded_recents, always_include_ids=current_state_ids, @@ -509,7 +511,7 @@ class SyncHandler(object): Returns: A Deferred map from ((type, state_key)->Event) """ - state_ids = yield self.store.get_state_ids_for_event( + state_ids = yield self.state_store.get_state_ids_for_event( event.event_id, state_filter=state_filter ) if event.is_state(): @@ -580,7 +582,7 @@ class SyncHandler(object): return None last_event = last_events[-1] - state_ids = yield self.store.get_state_ids_for_event( + state_ids = yield self.state_store.get_state_ids_for_event( last_event.event_id, state_filter=StateFilter.from_types( [(EventTypes.Name, ""), (EventTypes.CanonicalAlias, "")] @@ -757,11 +759,11 @@ class SyncHandler(object): if full_state: if batch: - current_state_ids = yield self.store.get_state_ids_for_event( + current_state_ids = yield self.state_store.get_state_ids_for_event( batch.events[-1].event_id, state_filter=state_filter ) - state_ids = yield self.store.get_state_ids_for_event( + state_ids = yield self.state_store.get_state_ids_for_event( batch.events[0].event_id, state_filter=state_filter ) @@ -781,7 +783,7 @@ class SyncHandler(object): ) elif batch.limited: if batch: - state_at_timeline_start = yield self.store.get_state_ids_for_event( + state_at_timeline_start = yield self.state_store.get_state_ids_for_event( batch.events[0].event_id, state_filter=state_filter ) else: @@ -810,7 +812,7 @@ class SyncHandler(object): ) if batch: - current_state_ids = yield self.store.get_state_ids_for_event( + current_state_ids = yield self.state_store.get_state_ids_for_event( batch.events[-1].event_id, state_filter=state_filter ) else: @@ -841,7 +843,7 @@ class SyncHandler(object): # So we fish out all the member events corresponding to the # timeline here, and then dedupe any redundant ones below. - state_ids = yield self.store.get_state_ids_for_event( + state_ids = yield self.state_store.get_state_ids_for_event( batch.events[0].event_id, # we only want members! state_filter=StateFilter.from_types( diff --git a/synapse/notifier.py b/synapse/notifier.py index 4e091314e6..af161a81d7 100644 --- a/synapse/notifier.py +++ b/synapse/notifier.py @@ -159,6 +159,7 @@ class Notifier(object): self.room_to_user_streams = {} self.hs = hs + self.storage = hs.get_storage() self.event_sources = hs.get_event_sources() self.store = hs.get_datastore() self.pending_new_room_events = [] @@ -425,7 +426,10 @@ class Notifier(object): if name == "room": new_events = yield filter_events_for_client( - self.store, user.to_string(), new_events, is_peeking=is_peeking + self.storage, + user.to_string(), + new_events, + is_peeking=is_peeking, ) elif name == "presence": now = self.clock.time_msec() diff --git a/synapse/push/httppusher.py b/synapse/push/httppusher.py index 6299587808..36e26032a1 100644 --- a/synapse/push/httppusher.py +++ b/synapse/push/httppusher.py @@ -64,6 +64,7 @@ class HttpPusher(object): def __init__(self, hs, pusherdict): self.hs = hs self.store = self.hs.get_datastore() + self.storage = self.hs.get_storage() self.clock = self.hs.get_clock() self.state_handler = self.hs.get_state_handler() self.user_id = pusherdict["user_name"] @@ -329,7 +330,7 @@ class HttpPusher(object): return d ctx = yield push_tools.get_context_for_event( - self.store, self.state_handler, event, self.user_id + self.storage, self.state_handler, event, self.user_id ) d = { diff --git a/synapse/push/mailer.py b/synapse/push/mailer.py index 5b16ab4ae8..1d15a06a58 100644 --- a/synapse/push/mailer.py +++ b/synapse/push/mailer.py @@ -119,6 +119,7 @@ class Mailer(object): self.store = self.hs.get_datastore() self.macaroon_gen = self.hs.get_macaroon_generator() self.state_handler = self.hs.get_state_handler() + self.storage = hs.get_storage() self.app_name = app_name logger.info("Created Mailer for app_name %s" % app_name) @@ -389,7 +390,7 @@ class Mailer(object): } the_events = yield filter_events_for_client( - self.store, user_id, results["events_before"] + self.storage, user_id, results["events_before"] ) the_events.append(notif_event) diff --git a/synapse/push/push_tools.py b/synapse/push/push_tools.py index a54051a726..de5c101a58 100644 --- a/synapse/push/push_tools.py +++ b/synapse/push/push_tools.py @@ -16,6 +16,7 @@ from twisted.internet import defer from synapse.push.presentable_names import calculate_room_name, name_from_member_event +from synapse.storage import Storage @defer.inlineCallbacks @@ -43,22 +44,22 @@ def get_badge_count(store, user_id): @defer.inlineCallbacks -def get_context_for_event(store, state_handler, ev, user_id): +def get_context_for_event(storage: Storage, state_handler, ev, user_id): ctx = {} - room_state_ids = yield store.get_state_ids_for_event(ev.event_id) + room_state_ids = yield storage.state.get_state_ids_for_event(ev.event_id) # we no longer bother setting room_alias, and make room_name the # human-readable name instead, be that m.room.name, an alias or # a list of people in the room name = yield calculate_room_name( - store, room_state_ids, user_id, fallback_to_single_member=False + storage.main, room_state_ids, user_id, fallback_to_single_member=False ) if name: ctx["name"] = name sender_state_event_id = room_state_ids[("m.room.member", ev.sender)] - sender_state_event = yield store.get_event(sender_state_event_id) + sender_state_event = yield storage.main.get_event(sender_state_event_id) ctx["sender_display_name"] = name_from_member_event(sender_state_event) return ctx diff --git a/synapse/state/__init__.py b/synapse/state/__init__.py index dc9f5a9008..4e91eb66fe 100644 --- a/synapse/state/__init__.py +++ b/synapse/state/__init__.py @@ -103,6 +103,7 @@ class StateHandler(object): def __init__(self, hs): self.clock = hs.get_clock() self.store = hs.get_datastore() + self.state_store = hs.get_storage().state self.hs = hs self._state_resolution_handler = hs.get_state_resolution_handler() @@ -271,7 +272,7 @@ class StateHandler(object): else: current_state_ids = prev_state_ids - state_group = yield self.store.store_state_group( + state_group = yield self.state_store.store_state_group( event.event_id, event.room_id, prev_group=None, @@ -321,7 +322,7 @@ class StateHandler(object): delta_ids = dict(entry.delta_ids) delta_ids[key] = event.event_id - state_group = yield self.store.store_state_group( + state_group = yield self.state_store.store_state_group( event.event_id, event.room_id, prev_group=prev_group, @@ -334,7 +335,7 @@ class StateHandler(object): delta_ids = entry.delta_ids if entry.state_group is None: - entry.state_group = yield self.store.store_state_group( + entry.state_group = yield self.state_store.store_state_group( event.event_id, event.room_id, prev_group=entry.prev_group, @@ -376,14 +377,16 @@ class StateHandler(object): # map from state group id to the state in that state group (where # 'state' is a map from state key to event id) # dict[int, dict[(str, str), str]] - state_groups_ids = yield self.store.get_state_groups_ids(room_id, event_ids) + state_groups_ids = yield self.state_store.get_state_groups_ids( + room_id, event_ids + ) if len(state_groups_ids) == 0: return _StateCacheEntry(state={}, state_group=None) elif len(state_groups_ids) == 1: name, state_list = list(state_groups_ids.items()).pop() - prev_group, delta_ids = yield self.store.get_state_group_delta(name) + prev_group, delta_ids = yield self.state_store.get_state_group_delta(name) return _StateCacheEntry( state=state_list, diff --git a/synapse/visibility.py b/synapse/visibility.py index bf0f1eebd8..8c843febd8 100644 --- a/synapse/visibility.py +++ b/synapse/visibility.py @@ -23,6 +23,7 @@ from twisted.internet import defer from synapse.api.constants import EventTypes, Membership from synapse.events.utils import prune_event +from synapse.storage import Storage from synapse.storage.state import StateFilter from synapse.types import get_domain_from_id @@ -43,14 +44,13 @@ MEMBERSHIP_PRIORITY = ( @defer.inlineCallbacks def filter_events_for_client( - store, user_id, events, is_peeking=False, always_include_ids=frozenset() + storage: Storage, user_id, events, is_peeking=False, always_include_ids=frozenset() ): """ Check which events a user is allowed to see Args: - store (synapse.storage.DataStore): our datastore (can also be a worker - store) + storage user_id(str): user id to be checked events(list[synapse.events.EventBase]): sequence of events to be checked is_peeking(bool): should be True if: @@ -68,12 +68,12 @@ def filter_events_for_client( events = list(e for e in events if not e.internal_metadata.is_soft_failed()) types = ((EventTypes.RoomHistoryVisibility, ""), (EventTypes.Member, user_id)) - event_id_to_state = yield store.get_state_for_events( + event_id_to_state = yield storage.state.get_state_for_events( frozenset(e.event_id for e in events), state_filter=StateFilter.from_types(types), ) - ignore_dict_content = yield store.get_global_account_data_by_type_for_user( + ignore_dict_content = yield storage.main.get_global_account_data_by_type_for_user( "m.ignored_user_list", user_id ) @@ -84,7 +84,7 @@ def filter_events_for_client( else [] ) - erased_senders = yield store.are_users_erased((e.sender for e in events)) + erased_senders = yield storage.main.are_users_erased((e.sender for e in events)) def allowed(event): """ @@ -213,13 +213,17 @@ def filter_events_for_client( @defer.inlineCallbacks def filter_events_for_server( - store, server_name, events, redact=True, check_history_visibility_only=False + storage: Storage, + server_name, + events, + redact=True, + check_history_visibility_only=False, ): """Filter a list of events based on whether given server is allowed to see them. Args: - store (DataStore) + storage server_name (str) events (iterable[FrozenEvent]) redact (bool): Whether to return a redacted version of the event, or @@ -274,7 +278,7 @@ def filter_events_for_server( # Lets check to see if all the events have a history visibility # of "shared" or "world_readable". If thats the case then we don't # need to check membership (as we know the server is in the room). - event_to_state_ids = yield store.get_state_ids_for_events( + event_to_state_ids = yield storage.state.get_state_ids_for_events( frozenset(e.event_id for e in events), state_filter=StateFilter.from_types( types=((EventTypes.RoomHistoryVisibility, ""),) @@ -292,14 +296,14 @@ def filter_events_for_server( if not visibility_ids: all_open = True else: - event_map = yield store.get_events(visibility_ids) + event_map = yield storage.main.get_events(visibility_ids) all_open = all( e.content.get("history_visibility") in (None, "shared", "world_readable") for e in itervalues(event_map) ) if not check_history_visibility_only: - erased_senders = yield store.are_users_erased((e.sender for e in events)) + erased_senders = yield storage.main.are_users_erased((e.sender for e in events)) else: # We don't want to check whether users are erased, which is equivalent # to no users having been erased. @@ -328,7 +332,7 @@ def filter_events_for_server( # first, for each event we're wanting to return, get the event_ids # of the history vis and membership state at those events. - event_to_state_ids = yield store.get_state_ids_for_events( + event_to_state_ids = yield storage.state.get_state_ids_for_events( frozenset(e.event_id for e in events), state_filter=StateFilter.from_types( types=((EventTypes.RoomHistoryVisibility, ""), (EventTypes.Member, None)) @@ -358,7 +362,7 @@ def filter_events_for_server( return False return state_key[idx + 1 :] == server_name - event_map = yield store.get_events( + event_map = yield storage.main.get_events( [ e_id for e_id, key in iteritems(event_id_to_state_key) diff --git a/tests/storage/test_state.py b/tests/storage/test_state.py index d573a3e07b..43200654f1 100644 --- a/tests/storage/test_state.py +++ b/tests/storage/test_state.py @@ -35,6 +35,7 @@ class StateStoreTestCase(tests.unittest.TestCase): self.store = hs.get_datastore() self.storage = hs.get_storage() + self.state_datastore = self.store self.event_builder_factory = hs.get_event_builder_factory() self.event_creation_handler = hs.get_event_creation_handler() @@ -83,7 +84,7 @@ class StateStoreTestCase(tests.unittest.TestCase): self.room, self.u_alice, EventTypes.Name, "", {"name": "test room"} ) - state_group_map = yield self.store.get_state_groups_ids( + state_group_map = yield self.storage.state.get_state_groups_ids( self.room, [e2.event_id] ) self.assertEqual(len(state_group_map), 1) @@ -102,7 +103,9 @@ class StateStoreTestCase(tests.unittest.TestCase): self.room, self.u_alice, EventTypes.Name, "", {"name": "test room"} ) - state_group_map = yield self.store.get_state_groups(self.room, [e2.event_id]) + state_group_map = yield self.storage.state.get_state_groups( + self.room, [e2.event_id] + ) self.assertEqual(len(state_group_map), 1) state_list = list(state_group_map.values())[0] @@ -142,7 +145,7 @@ class StateStoreTestCase(tests.unittest.TestCase): ) # check we get the full state as of the final event - state = yield self.store.get_state_for_event(e5.event_id) + state = yield self.storage.state.get_state_for_event(e5.event_id) self.assertIsNotNone(e4) @@ -158,21 +161,21 @@ class StateStoreTestCase(tests.unittest.TestCase): ) # check we can filter to the m.room.name event (with a '' state key) - state = yield self.store.get_state_for_event( + state = yield self.storage.state.get_state_for_event( e5.event_id, StateFilter.from_types([(EventTypes.Name, "")]) ) self.assertStateMapEqual({(e2.type, e2.state_key): e2}, state) # check we can filter to the m.room.name event (with a wildcard None state key) - state = yield self.store.get_state_for_event( + state = yield self.storage.state.get_state_for_event( e5.event_id, StateFilter.from_types([(EventTypes.Name, None)]) ) self.assertStateMapEqual({(e2.type, e2.state_key): e2}, state) # check we can grab the m.room.member events (with a wildcard None state key) - state = yield self.store.get_state_for_event( + state = yield self.storage.state.get_state_for_event( e5.event_id, StateFilter.from_types([(EventTypes.Member, None)]) ) @@ -182,7 +185,7 @@ class StateStoreTestCase(tests.unittest.TestCase): # check we can grab a specific room member without filtering out the # other event types - state = yield self.store.get_state_for_event( + state = yield self.storage.state.get_state_for_event( e5.event_id, state_filter=StateFilter( types={EventTypes.Member: {self.u_alice.to_string()}}, @@ -200,7 +203,7 @@ class StateStoreTestCase(tests.unittest.TestCase): ) # check that we can grab everything except members - state = yield self.store.get_state_for_event( + state = yield self.storage.state.get_state_for_event( e5.event_id, state_filter=StateFilter( types={EventTypes.Member: set()}, include_others=True @@ -216,13 +219,18 @@ class StateStoreTestCase(tests.unittest.TestCase): ####################################################### room_id = self.room.to_string() - group_ids = yield self.store.get_state_groups_ids(room_id, [e5.event_id]) + group_ids = yield self.storage.state.get_state_groups_ids( + room_id, [e5.event_id] + ) group = list(group_ids.keys())[0] # test _get_state_for_group_using_cache correctly filters out members # with types=[] - (state_dict, is_all) = yield self.store._get_state_for_group_using_cache( - self.store._state_group_cache, + ( + state_dict, + is_all, + ) = yield self.state_datastore._get_state_for_group_using_cache( + self.state_datastore._state_group_cache, group, state_filter=StateFilter( types={EventTypes.Member: set()}, include_others=True @@ -238,8 +246,11 @@ class StateStoreTestCase(tests.unittest.TestCase): state_dict, ) - (state_dict, is_all) = yield self.store._get_state_for_group_using_cache( - self.store._state_group_members_cache, + ( + state_dict, + is_all, + ) = yield self.state_datastore._get_state_for_group_using_cache( + self.state_datastore._state_group_members_cache, group, state_filter=StateFilter( types={EventTypes.Member: set()}, include_others=True @@ -251,8 +262,11 @@ class StateStoreTestCase(tests.unittest.TestCase): # test _get_state_for_group_using_cache correctly filters in members # with wildcard types - (state_dict, is_all) = yield self.store._get_state_for_group_using_cache( - self.store._state_group_cache, + ( + state_dict, + is_all, + ) = yield self.state_datastore._get_state_for_group_using_cache( + self.state_datastore._state_group_cache, group, state_filter=StateFilter( types={EventTypes.Member: None}, include_others=True @@ -268,8 +282,11 @@ class StateStoreTestCase(tests.unittest.TestCase): state_dict, ) - (state_dict, is_all) = yield self.store._get_state_for_group_using_cache( - self.store._state_group_members_cache, + ( + state_dict, + is_all, + ) = yield self.state_datastore._get_state_for_group_using_cache( + self.state_datastore._state_group_members_cache, group, state_filter=StateFilter( types={EventTypes.Member: None}, include_others=True @@ -288,8 +305,11 @@ class StateStoreTestCase(tests.unittest.TestCase): # test _get_state_for_group_using_cache correctly filters in members # with specific types - (state_dict, is_all) = yield self.store._get_state_for_group_using_cache( - self.store._state_group_cache, + ( + state_dict, + is_all, + ) = yield self.state_datastore._get_state_for_group_using_cache( + self.state_datastore._state_group_cache, group, state_filter=StateFilter( types={EventTypes.Member: {e5.state_key}}, include_others=True @@ -305,8 +325,11 @@ class StateStoreTestCase(tests.unittest.TestCase): state_dict, ) - (state_dict, is_all) = yield self.store._get_state_for_group_using_cache( - self.store._state_group_members_cache, + ( + state_dict, + is_all, + ) = yield self.state_datastore._get_state_for_group_using_cache( + self.state_datastore._state_group_members_cache, group, state_filter=StateFilter( types={EventTypes.Member: {e5.state_key}}, include_others=True @@ -318,8 +341,11 @@ class StateStoreTestCase(tests.unittest.TestCase): # test _get_state_for_group_using_cache correctly filters in members # with specific types - (state_dict, is_all) = yield self.store._get_state_for_group_using_cache( - self.store._state_group_members_cache, + ( + state_dict, + is_all, + ) = yield self.state_datastore._get_state_for_group_using_cache( + self.state_datastore._state_group_members_cache, group, state_filter=StateFilter( types={EventTypes.Member: {e5.state_key}}, include_others=False @@ -332,9 +358,11 @@ class StateStoreTestCase(tests.unittest.TestCase): ####################################################### # deliberately remove e2 (room name) from the _state_group_cache - (is_all, known_absent, state_dict_ids) = self.store._state_group_cache.get( - group - ) + ( + is_all, + known_absent, + state_dict_ids, + ) = self.state_datastore._state_group_cache.get(group) self.assertEqual(is_all, True) self.assertEqual(known_absent, set()) @@ -347,18 +375,20 @@ class StateStoreTestCase(tests.unittest.TestCase): ) state_dict_ids.pop((e2.type, e2.state_key)) - self.store._state_group_cache.invalidate(group) - self.store._state_group_cache.update( - sequence=self.store._state_group_cache.sequence, + self.state_datastore._state_group_cache.invalidate(group) + self.state_datastore._state_group_cache.update( + sequence=self.state_datastore._state_group_cache.sequence, key=group, value=state_dict_ids, # list fetched keys so it knows it's partial fetched_keys=((e1.type, e1.state_key),), ) - (is_all, known_absent, state_dict_ids) = self.store._state_group_cache.get( - group - ) + ( + is_all, + known_absent, + state_dict_ids, + ) = self.state_datastore._state_group_cache.get(group) self.assertEqual(is_all, False) self.assertEqual(known_absent, set([(e1.type, e1.state_key)])) @@ -370,8 +400,11 @@ class StateStoreTestCase(tests.unittest.TestCase): # test _get_state_for_group_using_cache correctly filters out members # with types=[] room_id = self.room.to_string() - (state_dict, is_all) = yield self.store._get_state_for_group_using_cache( - self.store._state_group_cache, + ( + state_dict, + is_all, + ) = yield self.state_datastore._get_state_for_group_using_cache( + self.state_datastore._state_group_cache, group, state_filter=StateFilter( types={EventTypes.Member: set()}, include_others=True @@ -382,8 +415,11 @@ class StateStoreTestCase(tests.unittest.TestCase): self.assertDictEqual({(e1.type, e1.state_key): e1.event_id}, state_dict) room_id = self.room.to_string() - (state_dict, is_all) = yield self.store._get_state_for_group_using_cache( - self.store._state_group_members_cache, + ( + state_dict, + is_all, + ) = yield self.state_datastore._get_state_for_group_using_cache( + self.state_datastore._state_group_members_cache, group, state_filter=StateFilter( types={EventTypes.Member: set()}, include_others=True @@ -395,8 +431,11 @@ class StateStoreTestCase(tests.unittest.TestCase): # test _get_state_for_group_using_cache correctly filters in members # wildcard types - (state_dict, is_all) = yield self.store._get_state_for_group_using_cache( - self.store._state_group_cache, + ( + state_dict, + is_all, + ) = yield self.state_datastore._get_state_for_group_using_cache( + self.state_datastore._state_group_cache, group, state_filter=StateFilter( types={EventTypes.Member: None}, include_others=True @@ -406,8 +445,11 @@ class StateStoreTestCase(tests.unittest.TestCase): self.assertEqual(is_all, False) self.assertDictEqual({(e1.type, e1.state_key): e1.event_id}, state_dict) - (state_dict, is_all) = yield self.store._get_state_for_group_using_cache( - self.store._state_group_members_cache, + ( + state_dict, + is_all, + ) = yield self.state_datastore._get_state_for_group_using_cache( + self.state_datastore._state_group_members_cache, group, state_filter=StateFilter( types={EventTypes.Member: None}, include_others=True @@ -425,8 +467,11 @@ class StateStoreTestCase(tests.unittest.TestCase): # test _get_state_for_group_using_cache correctly filters in members # with specific types - (state_dict, is_all) = yield self.store._get_state_for_group_using_cache( - self.store._state_group_cache, + ( + state_dict, + is_all, + ) = yield self.state_datastore._get_state_for_group_using_cache( + self.state_datastore._state_group_cache, group, state_filter=StateFilter( types={EventTypes.Member: {e5.state_key}}, include_others=True @@ -436,8 +481,11 @@ class StateStoreTestCase(tests.unittest.TestCase): self.assertEqual(is_all, False) self.assertDictEqual({(e1.type, e1.state_key): e1.event_id}, state_dict) - (state_dict, is_all) = yield self.store._get_state_for_group_using_cache( - self.store._state_group_members_cache, + ( + state_dict, + is_all, + ) = yield self.state_datastore._get_state_for_group_using_cache( + self.state_datastore._state_group_members_cache, group, state_filter=StateFilter( types={EventTypes.Member: {e5.state_key}}, include_others=True @@ -449,8 +497,11 @@ class StateStoreTestCase(tests.unittest.TestCase): # test _get_state_for_group_using_cache correctly filters in members # with specific types - (state_dict, is_all) = yield self.store._get_state_for_group_using_cache( - self.store._state_group_cache, + ( + state_dict, + is_all, + ) = yield self.state_datastore._get_state_for_group_using_cache( + self.state_datastore._state_group_cache, group, state_filter=StateFilter( types={EventTypes.Member: {e5.state_key}}, include_others=False @@ -460,8 +511,11 @@ class StateStoreTestCase(tests.unittest.TestCase): self.assertEqual(is_all, False) self.assertDictEqual({}, state_dict) - (state_dict, is_all) = yield self.store._get_state_for_group_using_cache( - self.store._state_group_members_cache, + ( + state_dict, + is_all, + ) = yield self.state_datastore._get_state_for_group_using_cache( + self.state_datastore._state_group_members_cache, group, state_filter=StateFilter( types={EventTypes.Member: {e5.state_key}}, include_others=False diff --git a/tests/test_state.py b/tests/test_state.py index 610ec9fb46..38246555bd 100644 --- a/tests/test_state.py +++ b/tests/test_state.py @@ -158,10 +158,12 @@ class Graph(object): class StateTestCase(unittest.TestCase): def setUp(self): self.store = StateGroupStore() + storage = Mock(main=self.store, state=self.store) hs = Mock( spec_set=[ "config", "get_datastore", + "get_storage", "get_auth", "get_state_handler", "get_clock", @@ -174,6 +176,7 @@ class StateTestCase(unittest.TestCase): hs.get_clock.return_value = MockClock() hs.get_auth.return_value = Auth(hs) hs.get_state_resolution_handler = lambda: StateResolutionHandler(hs) + hs.get_storage.return_value = storage self.state = StateHandler(hs) self.event_id = 0 diff --git a/tests/test_visibility.py b/tests/test_visibility.py index 6ae1ea9b04..f7381b2885 100644 --- a/tests/test_visibility.py +++ b/tests/test_visibility.py @@ -14,6 +14,8 @@ # limitations under the License. import logging +from mock import Mock + from twisted.internet import defer from twisted.internet.defer import succeed @@ -63,7 +65,7 @@ class FilterEventsForServerTestCase(tests.unittest.TestCase): events_to_filter.append(evt) filtered = yield filter_events_for_server( - self.store, "test_server", events_to_filter + self.storage, "test_server", events_to_filter ) # the result should be 5 redacted events, and 5 unredacted events. @@ -101,7 +103,7 @@ class FilterEventsForServerTestCase(tests.unittest.TestCase): # ... and the filtering happens. filtered = yield filter_events_for_server( - self.store, "test_server", events_to_filter + self.storage, "test_server", events_to_filter ) for i in range(0, len(events_to_filter)): @@ -258,6 +260,11 @@ class FilterEventsForServerTestCase(tests.unittest.TestCase): logger.info("Starting filtering") start = time.time() + + storage = Mock() + storage.main = test_store + storage.state = test_store + filtered = yield filter_events_for_server( test_store, "test_server", events_to_filter ) -- cgit 1.4.1 From 7c8c97e635811609c5a7ae4c0bb94e6573c30753 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 30 Oct 2019 15:12:49 +0000 Subject: Split purge API into events vs state --- synapse/handlers/pagination.py | 7 +- synapse/storage/__init__.py | 2 + synapse/storage/data_stores/main/events.py | 328 ++++++++++++++--------------- synapse/storage/data_stores/main/state.py | 23 ++ synapse/storage/purge_events.py | 117 ++++++++++ tests/storage/test_purge.py | 15 +- 6 files changed, 308 insertions(+), 184 deletions(-) create mode 100644 synapse/storage/purge_events.py (limited to 'tests') diff --git a/synapse/handlers/pagination.py b/synapse/handlers/pagination.py index 5744f4579d..9088ba14cd 100644 --- a/synapse/handlers/pagination.py +++ b/synapse/handlers/pagination.py @@ -69,6 +69,7 @@ class PaginationHandler(object): self.hs = hs self.auth = hs.get_auth() self.store = hs.get_datastore() + self.storage = hs.get_storage() self.clock = hs.get_clock() self._server_name = hs.hostname @@ -125,7 +126,9 @@ class PaginationHandler(object): self._purges_in_progress_by_room.add(room_id) try: with (yield self.pagination_lock.write(room_id)): - yield self.store.purge_history(room_id, token, delete_local_events) + yield self.storage.purge_events.purge_history( + room_id, token, delete_local_events + ) logger.info("[purge] complete") self._purges_by_id[purge_id].status = PurgeStatus.STATUS_COMPLETE except Exception: @@ -168,7 +171,7 @@ class PaginationHandler(object): if joined: raise SynapseError(400, "Users are still joined to this room") - await self.store.purge_room(room_id) + await self.storage.purge_events.purge_room(room_id) @defer.inlineCallbacks def get_messages( diff --git a/synapse/storage/__init__.py b/synapse/storage/__init__.py index a6429d17ed..3646ebd007 100644 --- a/synapse/storage/__init__.py +++ b/synapse/storage/__init__.py @@ -30,6 +30,7 @@ stored in `synapse.storage.schema`. from synapse.storage.data_stores import DataStores from synapse.storage.data_stores.main import DataStore from synapse.storage.persist_events import EventsPersistenceStorage +from synapse.storage.purge_events import PurgeEventsStorage __all__ = ["DataStores", "DataStore"] @@ -45,6 +46,7 @@ class Storage(object): self.main = stores.main self.persistence = EventsPersistenceStorage(hs, stores) + self.purge_events = PurgeEventsStorage(hs, stores) def are_all_users_on_domain(txn, database_engine, domain): diff --git a/synapse/storage/data_stores/main/events.py b/synapse/storage/data_stores/main/events.py index 7c3607f308..4eacba8058 100644 --- a/synapse/storage/data_stores/main/events.py +++ b/synapse/storage/data_stores/main/events.py @@ -1368,6 +1368,10 @@ class EventsStore( if True, we will delete local events as well as remote ones (instead of just marking them as outliers and deleting their state groups). + + Returns: + Deferred[set[int]]: The set of state groups that reference deleted + events. """ return self.runInteraction( @@ -1521,60 +1525,6 @@ class EventsStore( "[purge] found %i referenced state groups", len(referenced_state_groups) ) - logger.info("[purge] finding state groups that can be deleted") - - _ = self._find_unreferenced_groups_during_purge(txn, referenced_state_groups) - state_groups_to_delete, remaining_state_groups = _ - - logger.info( - "[purge] found %i state groups to delete", len(state_groups_to_delete) - ) - - logger.info( - "[purge] de-delta-ing %i remaining state groups", - len(remaining_state_groups), - ) - - # Now we turn the state groups that reference to-be-deleted state - # groups to non delta versions. - for sg in remaining_state_groups: - logger.info("[purge] de-delta-ing remaining state group %s", sg) - curr_state = self._get_state_groups_from_groups_txn(txn, [sg]) - curr_state = curr_state[sg] - - self._simple_delete_txn( - txn, table="state_groups_state", keyvalues={"state_group": sg} - ) - - self._simple_delete_txn( - txn, table="state_group_edges", keyvalues={"state_group": sg} - ) - - self._simple_insert_many_txn( - txn, - table="state_groups_state", - values=[ - { - "state_group": sg, - "room_id": room_id, - "type": key[0], - "state_key": key[1], - "event_id": state_id, - } - for key, state_id in iteritems(curr_state) - ], - ) - - logger.info("[purge] removing redundant state groups") - txn.executemany( - "DELETE FROM state_groups_state WHERE state_group = ?", - ((sg,) for sg in state_groups_to_delete), - ) - txn.executemany( - "DELETE FROM state_groups WHERE id = ?", - ((sg,) for sg in state_groups_to_delete), - ) - logger.info("[purge] removing events from event_to_state_groups") txn.execute( "DELETE FROM event_to_state_groups " @@ -1661,87 +1611,7 @@ class EventsStore( logger.info("[purge] done") - def _find_unreferenced_groups_during_purge(self, txn, state_groups): - """Used when purging history to figure out which state groups can be - deleted and which need to be de-delta'ed (due to one of its prev groups - being scheduled for deletion). - - Args: - txn - state_groups (set[int]): Set of state groups referenced by events - that are going to be deleted. - - Returns: - tuple[set[int], set[int]]: The set of state groups that can be - deleted and the set of state groups that need to be de-delta'ed - """ - # Graph of state group -> previous group - graph = {} - - # Set of events that we have found to be referenced by events - referenced_groups = set() - - # Set of state groups we've already seen - state_groups_seen = set(state_groups) - - # Set of state groups to handle next. - next_to_search = set(state_groups) - while next_to_search: - # We bound size of groups we're looking up at once, to stop the - # SQL query getting too big - if len(next_to_search) < 100: - current_search = next_to_search - next_to_search = set() - else: - current_search = set(itertools.islice(next_to_search, 100)) - next_to_search -= current_search - - # Check if state groups are referenced - sql = """ - SELECT DISTINCT state_group FROM event_to_state_groups - LEFT JOIN events_to_purge AS ep USING (event_id) - WHERE ep.event_id IS NULL AND - """ - clause, args = make_in_list_sql_clause( - txn.database_engine, "state_group", current_search - ) - txn.execute(sql + clause, list(args)) - - referenced = set(sg for sg, in txn) - referenced_groups |= referenced - - # We don't continue iterating up the state group graphs for state - # groups that are referenced. - current_search -= referenced - - rows = self._simple_select_many_txn( - txn, - table="state_group_edges", - column="prev_state_group", - iterable=current_search, - keyvalues={}, - retcols=("prev_state_group", "state_group"), - ) - - prevs = set(row["state_group"] for row in rows) - # We don't bother re-handling groups we've already seen - prevs -= state_groups_seen - next_to_search |= prevs - state_groups_seen |= prevs - - for row in rows: - # Note: Each state group can have at most one prev group - graph[row["state_group"]] = row["prev_state_group"] - - to_delete = state_groups_seen - referenced_groups - - to_dedelta = set() - for sg in referenced_groups: - prev_sg = graph.get(sg) - if prev_sg and prev_sg in to_delete: - to_dedelta.add(sg) - - return to_delete, to_dedelta + return referenced_state_groups def purge_room(self, room_id): """Deletes all record of a room @@ -1753,46 +1623,7 @@ class EventsStore( return self.runInteraction("purge_room", self._purge_room_txn, room_id) def _purge_room_txn(self, txn, room_id): - # first we have to delete the state groups states - logger.info("[purge] removing %s from state_groups_state", room_id) - - txn.execute( - """ - DELETE FROM state_groups_state WHERE state_group IN ( - SELECT state_group FROM events JOIN event_to_state_groups USING(event_id) - WHERE events.room_id=? - ) - """, - (room_id,), - ) - - # ... and the state group edges - logger.info("[purge] removing %s from state_group_edges", room_id) - - txn.execute( - """ - DELETE FROM state_group_edges WHERE state_group IN ( - SELECT state_group FROM events JOIN event_to_state_groups USING(event_id) - WHERE events.room_id=? - ) - """, - (room_id,), - ) - - # ... and the state groups - logger.info("[purge] removing %s from state_groups", room_id) - - txn.execute( - """ - DELETE FROM state_groups WHERE id IN ( - SELECT state_group FROM events JOIN event_to_state_groups USING(event_id) - WHERE events.room_id=? - ) - """, - (room_id,), - ) - - # and then tables which lack an index on room_id but have one on event_id + # First delete tables which lack an index on room_id but have one on event_id for table in ( "event_auth", "event_edges", @@ -1881,6 +1712,153 @@ class EventsStore( logger.info("[purge] done") + def purge_unreferenced_state_groups(self, room_id, state_groups_to_delete): + """Deletes no longer referenced state groups and de-deltas any state + groups that reference them. + """ + + return self.runInteraction( + "purge_unreferenced_state_groups", + self._purge_unreferenced_state_groups, + room_id, + state_groups_to_delete, + ) + + def _purge_unreferenced_state_groups(self, txn, room_id, state_groups_to_delete): + logger.info( + "[purge] found %i state groups to delete", len(state_groups_to_delete) + ) + + rows = self._simple_select_many_txn( + txn, + table="state_group_edges", + column="prev_state_group", + iterable=state_groups_to_delete, + keyvalues={}, + retcols=("state_group",), + ) + + remaining_state_groups = set( + row["state_group"] + for row in rows + if row["state_group"] not in state_groups_to_delete + ) + + logger.info( + "[purge] de-delta-ing %i remaining state groups", + len(remaining_state_groups), + ) + + # Now we turn the state groups that reference to-be-deleted state + # groups to non delta versions. + for sg in remaining_state_groups: + logger.info("[purge] de-delta-ing remaining state group %s", sg) + curr_state = self._get_state_groups_from_groups_txn(txn, [sg]) + curr_state = curr_state[sg] + + self._simple_delete_txn( + txn, table="state_groups_state", keyvalues={"state_group": sg} + ) + + self._simple_delete_txn( + txn, table="state_group_edges", keyvalues={"state_group": sg} + ) + + self._simple_insert_many_txn( + txn, + table="state_groups_state", + values=[ + { + "state_group": sg, + "room_id": room_id, + "type": key[0], + "state_key": key[1], + "event_id": state_id, + } + for key, state_id in iteritems(curr_state) + ], + ) + + logger.info("[purge] removing redundant state groups") + txn.executemany( + "DELETE FROM state_groups_state WHERE state_group = ?", + ((sg,) for sg in state_groups_to_delete), + ) + txn.executemany( + "DELETE FROM state_groups WHERE id = ?", + ((sg,) for sg in state_groups_to_delete), + ) + + @defer.inlineCallbacks + def get_previous_state_groups(self, state_groups): + """Fetch the previous groups of the given state groups. + + Args: + state_groups (Iterable[int]) + + Returns: + Deferred[dict[int, int]]: mapping from state group to previous + state group. + """ + + rows = yield self._simple_select_many_batch( + table="state_group_edges", + column="prev_state_group", + iterable=state_groups, + keyvalues={}, + retcols=("prev_state_group", "state_group"), + desc="get_previous_state_groups", + ) + + return {row["state_group"]: row["prev_state_group"] for row in rows} + + def purge_room_state(self, room_id): + """Deletes all record of a room from state tables + + Args: + room_id (str): + """ + + return self.runInteraction( + "purge_room_state", self._purge_room_state_txn, room_id + ) + + def _purge_room_state_txn(self, txn, room_id): + # first we have to delete the state groups states + logger.info("[purge] removing %s from state_groups_state", room_id) + + txn.execute( + """ + DELETE FROM state_groups_state + INNER JOIN state_groups USING (event_id) + WHEREE state_groups.room_id = ? + """, + (room_id,), + ) + + # ... and the state group edges + logger.info("[purge] removing %s from state_group_edges", room_id) + + txn.execute( + """ + DELETE FROM state_group_edges + INNER JOIN state_groups USING (event_id) + WHEREE state_groups.room_id = ? + ) + """, + (room_id,), + ) + + # ... and the state groups + logger.info("[purge] removing %s from state_groups", room_id) + + txn.execute( + """ + DELETE FROM state_groups WHEREE room_id = ? + """, + (room_id,), + ) + async def is_event_after(self, event_id1, event_id2): """Returns True if event_id1 is after event_id2 in the stream """ diff --git a/synapse/storage/data_stores/main/state.py b/synapse/storage/data_stores/main/state.py index 9b2207075b..36be8f0a9d 100644 --- a/synapse/storage/data_stores/main/state.py +++ b/synapse/storage/data_stores/main/state.py @@ -989,6 +989,29 @@ class StateGroupWorkerStore( return self.runInteraction("store_state_group", _store_state_group_txn) + @defer.inlineCallbacks + def get_referenced_state_groups(self, state_groups): + """Check if the state groups are referenced by events. + + Args: + state_groups (Iterable[int]) + + Returns: + Deferred[set[int]]: The subset of state groups that are + referenced. + """ + + rows = yield self._simple_select_many_batch( + table="event_to_state_groups", + column="state_group", + iterable=state_groups, + keyvalues={}, + retcols=("DISTINCT state_group",), + desc="get_referenced_state_groups", + ) + + return set(row["state_group"] for row in rows) + class StateBackgroundUpdateStore( StateGroupBackgroundUpdateStore, BackgroundUpdateStore diff --git a/synapse/storage/purge_events.py b/synapse/storage/purge_events.py new file mode 100644 index 0000000000..dd45df0c88 --- /dev/null +++ b/synapse/storage/purge_events.py @@ -0,0 +1,117 @@ +# -*- coding: utf-8 -*- +# Copyright 2019 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import itertools +import logging + +from twisted.internet import defer + +logger = logging.getLogger(__name__) + + +class PurgeEventsStorage(object): + """High level interface for purging rooms and event history. + """ + + def __init__(self, hs, stores): + self.stores = stores + + @defer.inlineCallbacks + def purge_room(self, room_id: str): + """Deletes all record of a room + """ + + yield self.stores.main.purge_room(room_id) + yield self.stores.main.purge_room_state(room_id) + + @defer.inlineCallbacks + def purge_history(self, room_id, token, delete_local_events): + """Deletes room history before a certain point + + Args: + room_id (str): + + token (str): A topological token to delete events before + + delete_local_events (bool): + if True, we will delete local events as well as remote ones + (instead of just marking them as outliers and deleting their + state groups). + """ + state_groups = yield self.stores.main.purge_history( + room_id, token, delete_local_events + ) + + logger.info("[purge] finding state groups that can be deleted") + + sg_to_delete = yield self._find_unreferenced_groups(state_groups) + + yield self.stores.main.purge_unreferenced_state_groups(room_id, sg_to_delete) + + @defer.inlineCallbacks + def _find_unreferenced_groups(self, state_groups): + """Used when purging history to figure out which state groups can be + deleted. + + Args: + state_groups (set[int]): Set of state groups referenced by events + that are going to be deleted. + + Returns: + Deferred[set[int]] The set of state groups that can be deleted. + """ + # Graph of state group -> previous group + graph = {} + + # Set of events that we have found to be referenced by events + referenced_groups = set() + + # Set of state groups we've already seen + state_groups_seen = set(state_groups) + + # Set of state groups to handle next. + next_to_search = set(state_groups) + while next_to_search: + # We bound size of groups we're looking up at once, to stop the + # SQL query getting too big + if len(next_to_search) < 100: + current_search = next_to_search + next_to_search = set() + else: + current_search = set(itertools.islice(next_to_search, 100)) + next_to_search -= current_search + + referenced = yield self.stores.main.get_referenced_state_groups( + current_search + ) + referenced_groups |= referenced + + # We don't continue iterating up the state group graphs for state + # groups that are referenced. + current_search -= referenced + + edges = yield self.stores.main.get_previous_state_groups(current_search) + + prevs = set(edges.values()) + # We don't bother re-handling groups we've already seen + prevs -= state_groups_seen + next_to_search |= prevs + state_groups_seen |= prevs + + graph.update(edges) + + to_delete = state_groups_seen - referenced_groups + + return to_delete diff --git a/tests/storage/test_purge.py b/tests/storage/test_purge.py index f671599cb8..b9fafaa1a6 100644 --- a/tests/storage/test_purge.py +++ b/tests/storage/test_purge.py @@ -40,23 +40,24 @@ class PurgeTests(HomeserverTestCase): third = self.helper.send(self.room_id, body="test3") last = self.helper.send(self.room_id, body="test4") - storage = self.hs.get_datastore() + store = self.hs.get_datastore() + storage = self.hs.get_storage() # Get the topological token - event = storage.get_topological_token_for_event(last["event_id"]) + event = store.get_topological_token_for_event(last["event_id"]) self.pump() event = self.successResultOf(event) # Purge everything before this topological token - purge = storage.purge_history(self.room_id, event, True) + purge = storage.purge_events.purge_history(self.room_id, event, True) self.pump() self.assertEqual(self.successResultOf(purge), None) # Try and get the events - get_first = storage.get_event(first["event_id"]) - get_second = storage.get_event(second["event_id"]) - get_third = storage.get_event(third["event_id"]) - get_last = storage.get_event(last["event_id"]) + get_first = store.get_event(first["event_id"]) + get_second = store.get_event(second["event_id"]) + get_third = store.get_event(third["event_id"]) + get_last = store.get_event(last["event_id"]) self.pump() # 1-3 should fail and last will succeed, meaning that 1-3 are deleted -- cgit 1.4.1 From e7943f660add8b602ea5225060bd0d74e6440017 Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Wed, 30 Oct 2019 16:15:04 +0000 Subject: Add unit tests --- synapse/api/filtering.py | 2 +- tests/api/test_filtering.py | 51 +++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 52 insertions(+), 1 deletion(-) (limited to 'tests') diff --git a/synapse/api/filtering.py b/synapse/api/filtering.py index a27029c678..bd91b9f018 100644 --- a/synapse/api/filtering.py +++ b/synapse/api/filtering.py @@ -307,7 +307,7 @@ class Filter(object): content = event.get("content", {}) # check if there is a string url field in the content for filtering purposes contains_url = isinstance(content.get("url"), text_type) - labels = content.get(LabelsField) + labels = content.get(LabelsField, []) return self.check_fields(room_id, sender, ev_type, labels, contains_url) diff --git a/tests/api/test_filtering.py b/tests/api/test_filtering.py index 6ba623de13..66b3c828db 100644 --- a/tests/api/test_filtering.py +++ b/tests/api/test_filtering.py @@ -19,6 +19,7 @@ import jsonschema from twisted.internet import defer +from synapse.api.constants import LabelsField from synapse.api.errors import SynapseError from synapse.api.filtering import Filter from synapse.events import FrozenEvent @@ -95,6 +96,8 @@ class FilteringTestCase(unittest.TestCase): "types": ["m.room.message"], "not_rooms": ["!726s6s6q:example.com"], "not_senders": ["@spam:example.com"], + "org.matrix.labels": ["#fun"], + "org.matrix.not_labels": ["#work"], }, "ephemeral": { "types": ["m.receipt", "m.typing"], @@ -320,6 +323,54 @@ class FilteringTestCase(unittest.TestCase): ) self.assertFalse(Filter(definition).check(event)) + def test_filter_labels(self): + definition = {"org.matrix.labels": ["#fun"]} + event = MockEvent( + sender="@foo:bar", + type="m.room.message", + room_id="!secretbase:unknown", + content={ + LabelsField: ["#fun"] + }, + ) + + self.assertTrue(Filter(definition).check(event)) + + event = MockEvent( + sender="@foo:bar", + type="m.room.message", + room_id="!secretbase:unknown", + content={ + LabelsField: ["#notfun"] + }, + ) + + self.assertFalse(Filter(definition).check(event)) + + def test_filter_not_labels(self): + definition = {"org.matrix.not_labels": ["#fun"]} + event = MockEvent( + sender="@foo:bar", + type="m.room.message", + room_id="!secretbase:unknown", + content={ + LabelsField: ["#fun"] + }, + ) + + self.assertFalse(Filter(definition).check(event)) + + event = MockEvent( + sender="@foo:bar", + type="m.room.message", + room_id="!secretbase:unknown", + content={ + LabelsField: ["#notfun"] + }, + ) + + self.assertTrue(Filter(definition).check(event)) + @defer.inlineCallbacks def test_filter_presence_match(self): user_filter_json = {"presence": {"types": ["m.*"]}} -- cgit 1.4.1 From 395683add1d569c0fdfd83d279551a3ba926f4d5 Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Wed, 30 Oct 2019 16:47:37 +0000 Subject: Add integration tests for sync --- tests/rest/client/v1/utils.py | 15 ++++- tests/rest/client/v2_alpha/test_sync.py | 112 +++++++++++++++++++++++++++++++- 2 files changed, 122 insertions(+), 5 deletions(-) (limited to 'tests') diff --git a/tests/rest/client/v1/utils.py b/tests/rest/client/v1/utils.py index cdded88b7f..8ea0cb05ea 100644 --- a/tests/rest/client/v1/utils.py +++ b/tests/rest/client/v1/utils.py @@ -106,13 +106,22 @@ class RestHelper(object): self.auth_user_id = temp_id def send(self, room_id, body=None, txn_id=None, tok=None, expect_code=200): - if txn_id is None: - txn_id = "m%s" % (str(time.time())) if body is None: body = "body_text_here" - path = "/_matrix/client/r0/rooms/%s/send/m.room.message/%s" % (room_id, txn_id) content = {"msgtype": "m.text", "body": body} + + return self.send_event( + room_id, "m.room.message", content, txn_id, tok, expect_code + ) + + def send_event( + self, room_id, type, content={}, txn_id=None, tok=None, expect_code=200 + ): + if txn_id is None: + txn_id = "m%s" % (str(time.time())) + + path = "/_matrix/client/r0/rooms/%s/send/%s/%s" % (room_id, type, txn_id) if tok: path = path + "?access_token=%s" % tok diff --git a/tests/rest/client/v2_alpha/test_sync.py b/tests/rest/client/v2_alpha/test_sync.py index 71895094bd..0263be010f 100644 --- a/tests/rest/client/v2_alpha/test_sync.py +++ b/tests/rest/client/v2_alpha/test_sync.py @@ -12,9 +12,10 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. - +import json from mock import Mock +from synapse.api.constants import EventTypes, LabelsField import synapse.rest.admin from synapse.rest.client.v1 import login, room from synapse.rest.client.v2_alpha import sync @@ -26,7 +27,12 @@ from tests.server import TimedOutException class FilterTestCase(unittest.HomeserverTestCase): user_id = "@apple:test" - servlets = [sync.register_servlets] + servlets = [ + synapse.rest.admin.register_servlets_for_client_rest_resource, + room.register_servlets, + login.register_servlets, + sync.register_servlets, + ] def make_homeserver(self, reactor, clock): @@ -70,6 +76,108 @@ class FilterTestCase(unittest.HomeserverTestCase): ) +class SyncFilterTestCase(unittest.HomeserverTestCase): + servlets = [ + synapse.rest.admin.register_servlets_for_client_rest_resource, + room.register_servlets, + login.register_servlets, + sync.register_servlets, + ] + + def test_sync_filter_labels(self): + sync_filter = json.dumps( + { + "room": { + "timeline": { + "types": [EventTypes.Message], + "org.matrix.labels": ["#fun"], + } + } + } + ) + + events = self._test_sync_filter_labels(sync_filter) + + self.assertEqual(len(events), 2, events) + self.assertEqual(events[0]["content"]["body"], "with label", events[0]) + self.assertEqual(events[1]["content"]["body"], "with label", events[1]) + + def test_sync_filter_not_labels(self): + sync_filter = json.dumps( + { + "room": { + "timeline": { + "types": [EventTypes.Message], + "org.matrix.not_labels": ["#fun"], + } + } + } + ) + + events = self._test_sync_filter_labels(sync_filter) + + self.assertEqual(len(events), 2, events) + self.assertEqual(events[0]["content"]["body"], "without label", events[0]) + self.assertEqual(events[1]["content"]["body"], "with wrong label", events[1]) + + def _test_sync_filter_labels(self, sync_filter): + user_id = self.register_user("kermit", "test") + tok = self.login("kermit", "test") + + room_id = self.helper.create_room_as(user_id, tok=tok) + + self.helper.send_event( + room_id=room_id, + type=EventTypes.Message, + content={ + "msgtype": "m.text", + "body": "with label", + LabelsField: ["#fun"], + }, + tok=tok, + ) + + self.helper.send_event( + room_id=room_id, + type=EventTypes.Message, + content={ + "msgtype": "m.text", + "body": "without label", + }, + tok=tok, + ) + + self.helper.send_event( + room_id=room_id, + type=EventTypes.Message, + content={ + "msgtype": "m.text", + "body": "with wrong label", + LabelsField: ["#work"], + }, + tok=tok, + ) + + self.helper.send_event( + room_id=room_id, + type=EventTypes.Message, + content={ + "msgtype": "m.text", + "body": "with label", + LabelsField: ["#fun"], + }, + tok=tok, + ) + + request, channel = self.make_request( + "GET", "/sync?filter=%s" % sync_filter, access_token=tok + ) + self.render(request) + self.assertEqual(channel.code, 200, channel.result) + + return channel.json_body["rooms"]["join"][room_id]["timeline"]["events"] + + class SyncTypingTests(unittest.HomeserverTestCase): servlets = [ -- cgit 1.4.1 From fe51d6cacf6e1a2da5fc3589d0bc4118342b33dd Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Wed, 30 Oct 2019 17:28:41 +0000 Subject: Add more integration testing --- synapse/storage/data_stores/main/stream.py | 2 +- tests/rest/client/v2_alpha/test_sync.py | 45 ++++++++++++++++++++++++++---- 2 files changed, 40 insertions(+), 7 deletions(-) (limited to 'tests') diff --git a/synapse/storage/data_stores/main/stream.py b/synapse/storage/data_stores/main/stream.py index 907d7f20ba..cfa34ba1e7 100644 --- a/synapse/storage/data_stores/main/stream.py +++ b/synapse/storage/data_stores/main/stream.py @@ -872,7 +872,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): args.append(int(limit)) sql = ( - "SELECT event_id, topological_ordering, stream_ordering" + "SELECT DISTINCT event_id, topological_ordering, stream_ordering" " FROM events" " LEFT JOIN event_labels USING (event_id)" " WHERE outlier = ? AND room_id = ? AND %(bounds)s" diff --git a/tests/rest/client/v2_alpha/test_sync.py b/tests/rest/client/v2_alpha/test_sync.py index 0263be010f..a1aa7d87bd 100644 --- a/tests/rest/client/v2_alpha/test_sync.py +++ b/tests/rest/client/v2_alpha/test_sync.py @@ -85,6 +85,7 @@ class SyncFilterTestCase(unittest.HomeserverTestCase): ] def test_sync_filter_labels(self): + """Test that we can filter by a label.""" sync_filter = json.dumps( { "room": { @@ -98,11 +99,12 @@ class SyncFilterTestCase(unittest.HomeserverTestCase): events = self._test_sync_filter_labels(sync_filter) - self.assertEqual(len(events), 2, events) - self.assertEqual(events[0]["content"]["body"], "with label", events[0]) - self.assertEqual(events[1]["content"]["body"], "with label", events[1]) + self.assertEqual(len(events), 2, [event["content"] for event in events]) + self.assertEqual(events[0]["content"]["body"], "with right label", events[0]) + self.assertEqual(events[1]["content"]["body"], "with right label", events[1]) def test_sync_filter_not_labels(self): + """Test that we can filter by the absence of a label.""" sync_filter = json.dumps( { "room": { @@ -116,9 +118,29 @@ class SyncFilterTestCase(unittest.HomeserverTestCase): events = self._test_sync_filter_labels(sync_filter) - self.assertEqual(len(events), 2, events) + self.assertEqual(len(events), 3, [event["content"] for event in events]) self.assertEqual(events[0]["content"]["body"], "without label", events[0]) self.assertEqual(events[1]["content"]["body"], "with wrong label", events[1]) + self.assertEqual(events[2]["content"]["body"], "with two wrong labels", events[2]) + + def test_sync_filter_labels_not_labels(self): + """Test that we can filter by both a label and the absence of another label.""" + sync_filter = json.dumps( + { + "room": { + "timeline": { + "types": [EventTypes.Message], + "org.matrix.labels": ["#work"], + "org.matrix.not_labels": ["#notfun"], + } + } + } + ) + + events = self._test_sync_filter_labels(sync_filter) + + self.assertEqual(len(events), 1, [event["content"] for event in events]) + self.assertEqual(events[0]["content"]["body"], "with wrong label", events[0]) def _test_sync_filter_labels(self, sync_filter): user_id = self.register_user("kermit", "test") @@ -131,7 +153,7 @@ class SyncFilterTestCase(unittest.HomeserverTestCase): type=EventTypes.Message, content={ "msgtype": "m.text", - "body": "with label", + "body": "with right label", LabelsField: ["#fun"], }, tok=tok, @@ -163,7 +185,18 @@ class SyncFilterTestCase(unittest.HomeserverTestCase): type=EventTypes.Message, content={ "msgtype": "m.text", - "body": "with label", + "body": "with two wrong labels", + LabelsField: ["#work", "#notfun"], + }, + tok=tok, + ) + + self.helper.send_event( + room_id=room_id, + type=EventTypes.Message, + content={ + "msgtype": "m.text", + "body": "with right label", LabelsField: ["#fun"], }, tok=tok, -- cgit 1.4.1 From d8c9109aeee58950f0fd4d9865836b82aa7aafb6 Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Wed, 30 Oct 2019 17:48:22 +0000 Subject: Add integration tests for /messages --- tests/rest/client/v1/test_rooms.py | 102 ++++++++++++++++++++++++++++++++++++- 1 file changed, 101 insertions(+), 1 deletion(-) (limited to 'tests') diff --git a/tests/rest/client/v1/test_rooms.py b/tests/rest/client/v1/test_rooms.py index 2f2ca74611..ba2008497e 100644 --- a/tests/rest/client/v1/test_rooms.py +++ b/tests/rest/client/v1/test_rooms.py @@ -24,7 +24,7 @@ from six.moves.urllib import parse as urlparse from twisted.internet import defer import synapse.rest.admin -from synapse.api.constants import Membership +from synapse.api.constants import EventTypes, LabelsField, Membership from synapse.rest.client.v1 import login, profile, room from tests import unittest @@ -811,6 +811,106 @@ class RoomMessageListTestCase(RoomBase): self.assertTrue("chunk" in channel.json_body) self.assertTrue("end" in channel.json_body) + def test_filter_labels(self): + """Test that we can filter by a label.""" + message_filter = json.dumps({ + "types": [EventTypes.Message], + "org.matrix.labels": ["#fun"], + }) + + events = self._test_filter_labels(message_filter) + + self.assertEqual(len(events), 2, [event["content"] for event in events]) + self.assertEqual(events[0]["content"]["body"], "with right label", events[0]) + self.assertEqual(events[1]["content"]["body"], "with right label", events[1]) + + def test_filter_not_labels(self): + """Test that we can filter by the absence of a label.""" + message_filter = json.dumps({ + "types": [EventTypes.Message], + "org.matrix.not_labels": ["#fun"], + }) + + events = self._test_filter_labels(message_filter) + + self.assertEqual(len(events), 3, [event["content"] for event in events]) + self.assertEqual(events[0]["content"]["body"], "without label", events[0]) + self.assertEqual(events[1]["content"]["body"], "with wrong label", events[1]) + self.assertEqual(events[2]["content"]["body"], "with two wrong labels", events[2]) + + def test_filter_labels_not_labels(self): + """Test that we can filter by both a label and the absence of another label.""" + sync_filter = json.dumps({ + "types": [EventTypes.Message], + "org.matrix.labels": ["#work"], + "org.matrix.not_labels": ["#notfun"], + }) + + events = self._test_filter_labels(sync_filter) + + self.assertEqual(len(events), 1, [event["content"] for event in events]) + self.assertEqual(events[0]["content"]["body"], "with wrong label", events[0]) + + def _test_filter_labels(self, message_filter): + self.helper.send_event( + room_id=self.room_id, + type=EventTypes.Message, + content={ + "msgtype": "m.text", + "body": "with right label", + LabelsField: ["#fun"], + } + ) + + self.helper.send_event( + room_id=self.room_id, + type=EventTypes.Message, + content={ + "msgtype": "m.text", + "body": "without label", + } + ) + + self.helper.send_event( + room_id=self.room_id, + type=EventTypes.Message, + content={ + "msgtype": "m.text", + "body": "with wrong label", + LabelsField: ["#work"], + } + ) + + self.helper.send_event( + room_id=self.room_id, + type=EventTypes.Message, + content={ + "msgtype": "m.text", + "body": "with two wrong labels", + LabelsField: ["#work", "#notfun"], + } + ) + + self.helper.send_event( + room_id=self.room_id, + type=EventTypes.Message, + content={ + "msgtype": "m.text", + "body": "with right label", + LabelsField: ["#fun"], + } + ) + + token = "s0_0_0_0_0_0_0_0_0" + request, channel = self.make_request( + "GET", "/rooms/%s/messages?access_token=x&from=%s&filter=%s" % ( + self.room_id, token, message_filter + ) + ) + self.render(request) + + return channel.json_body["chunk"] + class RoomSearchTestCase(unittest.HomeserverTestCase): servlets = [ -- cgit 1.4.1 From dcc069a2e2540862c233a20037e3e59591a42431 Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Wed, 30 Oct 2019 18:01:56 +0000 Subject: Lint --- synapse/storage/data_stores/main/events.py | 8 +---- tests/api/test_filtering.py | 16 +++------- tests/rest/client/v1/test_rooms.py | 49 +++++++++++++++--------------- tests/rest/client/v2_alpha/test_sync.py | 12 ++++---- 4 files changed, 35 insertions(+), 50 deletions(-) (limited to 'tests') diff --git a/synapse/storage/data_stores/main/events.py b/synapse/storage/data_stores/main/events.py index f80b5f1a3f..2b900f1ce1 100644 --- a/synapse/storage/data_stores/main/events.py +++ b/synapse/storage/data_stores/main/events.py @@ -2486,13 +2486,7 @@ class EventsStore( return self._simple_insert_many_txn( txn=txn, table="event_labels", - values=[ - { - "event_id": event_id, - "label": label, - } - for label in labels - ], + values=[{"event_id": event_id, "label": label} for label in labels], ) diff --git a/tests/api/test_filtering.py b/tests/api/test_filtering.py index 66b3c828db..e004ab1ee5 100644 --- a/tests/api/test_filtering.py +++ b/tests/api/test_filtering.py @@ -329,9 +329,7 @@ class FilteringTestCase(unittest.TestCase): sender="@foo:bar", type="m.room.message", room_id="!secretbase:unknown", - content={ - LabelsField: ["#fun"] - }, + content={LabelsField: ["#fun"]}, ) self.assertTrue(Filter(definition).check(event)) @@ -340,9 +338,7 @@ class FilteringTestCase(unittest.TestCase): sender="@foo:bar", type="m.room.message", room_id="!secretbase:unknown", - content={ - LabelsField: ["#notfun"] - }, + content={LabelsField: ["#notfun"]}, ) self.assertFalse(Filter(definition).check(event)) @@ -353,9 +349,7 @@ class FilteringTestCase(unittest.TestCase): sender="@foo:bar", type="m.room.message", room_id="!secretbase:unknown", - content={ - LabelsField: ["#fun"] - }, + content={LabelsField: ["#fun"]}, ) self.assertFalse(Filter(definition).check(event)) @@ -364,9 +358,7 @@ class FilteringTestCase(unittest.TestCase): sender="@foo:bar", type="m.room.message", room_id="!secretbase:unknown", - content={ - LabelsField: ["#notfun"] - }, + content={LabelsField: ["#notfun"]}, ) self.assertTrue(Filter(definition).check(event)) diff --git a/tests/rest/client/v1/test_rooms.py b/tests/rest/client/v1/test_rooms.py index ba2008497e..188f47bd7d 100644 --- a/tests/rest/client/v1/test_rooms.py +++ b/tests/rest/client/v1/test_rooms.py @@ -813,10 +813,9 @@ class RoomMessageListTestCase(RoomBase): def test_filter_labels(self): """Test that we can filter by a label.""" - message_filter = json.dumps({ - "types": [EventTypes.Message], - "org.matrix.labels": ["#fun"], - }) + message_filter = json.dumps( + {"types": [EventTypes.Message], "org.matrix.labels": ["#fun"]} + ) events = self._test_filter_labels(message_filter) @@ -826,25 +825,28 @@ class RoomMessageListTestCase(RoomBase): def test_filter_not_labels(self): """Test that we can filter by the absence of a label.""" - message_filter = json.dumps({ - "types": [EventTypes.Message], - "org.matrix.not_labels": ["#fun"], - }) + message_filter = json.dumps( + {"types": [EventTypes.Message], "org.matrix.not_labels": ["#fun"]} + ) events = self._test_filter_labels(message_filter) self.assertEqual(len(events), 3, [event["content"] for event in events]) self.assertEqual(events[0]["content"]["body"], "without label", events[0]) self.assertEqual(events[1]["content"]["body"], "with wrong label", events[1]) - self.assertEqual(events[2]["content"]["body"], "with two wrong labels", events[2]) + self.assertEqual( + events[2]["content"]["body"], "with two wrong labels", events[2] + ) def test_filter_labels_not_labels(self): """Test that we can filter by both a label and the absence of another label.""" - sync_filter = json.dumps({ - "types": [EventTypes.Message], - "org.matrix.labels": ["#work"], - "org.matrix.not_labels": ["#notfun"], - }) + sync_filter = json.dumps( + { + "types": [EventTypes.Message], + "org.matrix.labels": ["#work"], + "org.matrix.not_labels": ["#notfun"], + } + ) events = self._test_filter_labels(sync_filter) @@ -859,16 +861,13 @@ class RoomMessageListTestCase(RoomBase): "msgtype": "m.text", "body": "with right label", LabelsField: ["#fun"], - } + }, ) self.helper.send_event( room_id=self.room_id, type=EventTypes.Message, - content={ - "msgtype": "m.text", - "body": "without label", - } + content={"msgtype": "m.text", "body": "without label"}, ) self.helper.send_event( @@ -878,7 +877,7 @@ class RoomMessageListTestCase(RoomBase): "msgtype": "m.text", "body": "with wrong label", LabelsField: ["#work"], - } + }, ) self.helper.send_event( @@ -888,7 +887,7 @@ class RoomMessageListTestCase(RoomBase): "msgtype": "m.text", "body": "with two wrong labels", LabelsField: ["#work", "#notfun"], - } + }, ) self.helper.send_event( @@ -898,14 +897,14 @@ class RoomMessageListTestCase(RoomBase): "msgtype": "m.text", "body": "with right label", LabelsField: ["#fun"], - } + }, ) token = "s0_0_0_0_0_0_0_0_0" request, channel = self.make_request( - "GET", "/rooms/%s/messages?access_token=x&from=%s&filter=%s" % ( - self.room_id, token, message_filter - ) + "GET", + "/rooms/%s/messages?access_token=x&from=%s&filter=%s" + % (self.room_id, token, message_filter), ) self.render(request) diff --git a/tests/rest/client/v2_alpha/test_sync.py b/tests/rest/client/v2_alpha/test_sync.py index a1aa7d87bd..c5c199d412 100644 --- a/tests/rest/client/v2_alpha/test_sync.py +++ b/tests/rest/client/v2_alpha/test_sync.py @@ -13,10 +13,11 @@ # See the License for the specific language governing permissions and # limitations under the License. import json + from mock import Mock -from synapse.api.constants import EventTypes, LabelsField import synapse.rest.admin +from synapse.api.constants import EventTypes, LabelsField from synapse.rest.client.v1 import login, room from synapse.rest.client.v2_alpha import sync @@ -121,7 +122,9 @@ class SyncFilterTestCase(unittest.HomeserverTestCase): self.assertEqual(len(events), 3, [event["content"] for event in events]) self.assertEqual(events[0]["content"]["body"], "without label", events[0]) self.assertEqual(events[1]["content"]["body"], "with wrong label", events[1]) - self.assertEqual(events[2]["content"]["body"], "with two wrong labels", events[2]) + self.assertEqual( + events[2]["content"]["body"], "with two wrong labels", events[2] + ) def test_sync_filter_labels_not_labels(self): """Test that we can filter by both a label and the absence of another label.""" @@ -162,10 +165,7 @@ class SyncFilterTestCase(unittest.HomeserverTestCase): self.helper.send_event( room_id=room_id, type=EventTypes.Message, - content={ - "msgtype": "m.text", - "body": "without label", - }, + content={"msgtype": "m.text", "body": "without label"}, tok=tok, ) -- cgit 1.4.1 From bb6cec27a5ac6d5d6d5f67df21610a63745ac0a9 Mon Sep 17 00:00:00 2001 From: Hubert Chathi Date: Wed, 30 Oct 2019 14:57:34 -0400 Subject: rename get_devices_by_remote to get_device_updates_by_remote --- synapse/federation/sender/per_destination_queue.py | 4 ++-- synapse/storage/data_stores/main/devices.py | 8 ++++---- tests/handlers/test_typing.py | 4 ++-- tests/storage/test_devices.py | 12 ++++++------ 4 files changed, 14 insertions(+), 14 deletions(-) (limited to 'tests') diff --git a/synapse/federation/sender/per_destination_queue.py b/synapse/federation/sender/per_destination_queue.py index d5d4a60c88..6e3012cd41 100644 --- a/synapse/federation/sender/per_destination_queue.py +++ b/synapse/federation/sender/per_destination_queue.py @@ -359,7 +359,7 @@ class PerDestinationQueue(object): last_device_list = self._last_device_list_stream_id # Retrieve list of new device updates to send to the destination - now_stream_id, results = yield self._store.get_devices_by_remote( + now_stream_id, results = yield self._store.get_device_updates_by_remote( self._destination, last_device_list, limit=limit ) edus = [ @@ -372,7 +372,7 @@ class PerDestinationQueue(object): for (edu_type, content) in results ] - assert len(edus) <= limit, "get_devices_by_remote returned too many EDUs" + assert len(edus) <= limit, "get_device_updates_by_remote returned too many EDUs" return (edus, now_stream_id) diff --git a/synapse/storage/data_stores/main/devices.py b/synapse/storage/data_stores/main/devices.py index 0b12bc58c4..717eab4159 100644 --- a/synapse/storage/data_stores/main/devices.py +++ b/synapse/storage/data_stores/main/devices.py @@ -91,7 +91,7 @@ class DeviceWorkerStore(SQLBaseStore): @trace @defer.inlineCallbacks - def get_devices_by_remote(self, destination, from_stream_id, limit): + def get_device_updates_by_remote(self, destination, from_stream_id, limit): """Get a stream of device updates to send to the given remote server. Args: @@ -123,8 +123,8 @@ class DeviceWorkerStore(SQLBaseStore): # stream_id; the rationale being that such a large device list update # is likely an error. updates = yield self.runInteraction( - "get_devices_by_remote", - self._get_devices_by_remote_txn, + "get_device_updates_by_remote", + self._get_device_updates_by_remote_txn, destination, from_stream_id, now_stream_id, @@ -241,7 +241,7 @@ class DeviceWorkerStore(SQLBaseStore): return now_stream_id, results - def _get_devices_by_remote_txn( + def _get_device_updates_by_remote_txn( self, txn, destination, from_stream_id, now_stream_id, limit ): """Return device update information for a given remote destination diff --git a/tests/handlers/test_typing.py b/tests/handlers/test_typing.py index f360c8e965..5ec568f4e6 100644 --- a/tests/handlers/test_typing.py +++ b/tests/handlers/test_typing.py @@ -73,7 +73,7 @@ class TypingNotificationsTestCase(unittest.HomeserverTestCase): "get_received_txn_response", "set_received_txn_response", "get_destination_retry_timings", - "get_devices_by_remote", + "get_device_updates_by_remote", # Bits that user_directory needs "get_user_directory_stream_pos", "get_current_state_deltas", @@ -109,7 +109,7 @@ class TypingNotificationsTestCase(unittest.HomeserverTestCase): retry_timings_res ) - self.datastore.get_devices_by_remote.return_value = (0, []) + self.datastore.get_device_updates_by_remote.return_value = (0, []) def get_received_txn_response(*args): return defer.succeed(None) diff --git a/tests/storage/test_devices.py b/tests/storage/test_devices.py index 039cc79357..6f8d990959 100644 --- a/tests/storage/test_devices.py +++ b/tests/storage/test_devices.py @@ -72,7 +72,7 @@ class DeviceStoreTestCase(tests.unittest.TestCase): ) @defer.inlineCallbacks - def test_get_devices_by_remote(self): + def test_get_device_updates_by_remote(self): device_ids = ["device_id1", "device_id2"] # Add two device updates with a single stream_id @@ -81,7 +81,7 @@ class DeviceStoreTestCase(tests.unittest.TestCase): ) # Get all device updates ever meant for this remote - now_stream_id, device_updates = yield self.store.get_devices_by_remote( + now_stream_id, device_updates = yield self.store.get_device_updates_by_remote( "somehost", -1, limit=100 ) @@ -89,7 +89,7 @@ class DeviceStoreTestCase(tests.unittest.TestCase): self._check_devices_in_updates(device_ids, device_updates) @defer.inlineCallbacks - def test_get_devices_by_remote_limited(self): + def test_get_device_updates_by_remote_limited(self): # Test breaking the update limit in 1, 101, and 1 device_id segments # first add one device @@ -115,20 +115,20 @@ class DeviceStoreTestCase(tests.unittest.TestCase): # # first we should get a single update - now_stream_id, device_updates = yield self.store.get_devices_by_remote( + now_stream_id, device_updates = yield self.store.get_device_updates_by_remote( "someotherhost", -1, limit=100 ) self._check_devices_in_updates(device_ids1, device_updates) # Then we should get an empty list back as the 101 devices broke the limit - now_stream_id, device_updates = yield self.store.get_devices_by_remote( + now_stream_id, device_updates = yield self.store.get_device_updates_by_remote( "someotherhost", now_stream_id, limit=100 ) self.assertEqual(len(device_updates), 0) # The 101 devices should've been cleared, so we should now just get one device # update - now_stream_id, device_updates = yield self.store.get_devices_by_remote( + now_stream_id, device_updates = yield self.store.get_device_updates_by_remote( "someotherhost", now_stream_id, limit=100 ) self._check_devices_in_updates(device_ids3, device_updates) -- cgit 1.4.1 From 97c60ccaa35059a55866304f6850b24e99912036 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Thu, 31 Oct 2019 11:30:25 +0000 Subject: Add unit test for /purge_room API --- tests/rest/admin/test_admin.py | 78 ++++++++++++++++++++++++++++++++++++++++++ tests/server.py | 6 +++- 2 files changed, 83 insertions(+), 1 deletion(-) (limited to 'tests') diff --git a/tests/rest/admin/test_admin.py b/tests/rest/admin/test_admin.py index d3a4f717f7..8e1ca8b738 100644 --- a/tests/rest/admin/test_admin.py +++ b/tests/rest/admin/test_admin.py @@ -561,3 +561,81 @@ class DeleteGroupTestCase(unittest.HomeserverTestCase): self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) return channel.json_body["groups"] + + +class PurgeRoomTestCase(unittest.HomeserverTestCase): + """Test /purge_room admin API. + """ + + servlets = [ + synapse.rest.admin.register_servlets, + login.register_servlets, + room.register_servlets, + ] + + def prepare(self, reactor, clock, hs): + self.store = hs.get_datastore() + + self.admin_user = self.register_user("admin", "pass", admin=True) + self.admin_user_tok = self.login("admin", "pass") + + def test_purge_room(self): + room_id = self.helper.create_room_as(self.admin_user, tok=self.admin_user_tok) + + # All users have to have left the room. + self.helper.leave(room_id, user=self.admin_user, tok=self.admin_user_tok) + + url = "/_synapse/admin/v1/purge_room" + request, channel = self.make_request( + "POST", + url.encode("ascii"), + {"room_id": room_id}, + access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + + # Test that the following tables have been purged of all rows related to the room. + for table in ( + "current_state_events", + "event_backward_extremities", + "event_forward_extremities", + "event_json", + "event_push_actions", + "event_search", + "events", + "group_rooms", + "public_room_list_stream", + "receipts_graph", + "receipts_linearized", + "room_aliases", + "room_depth", + "room_memberships", + "room_stats_state", + "room_stats_current", + "room_stats_historical", + "room_stats_earliest_token", + "rooms", + "stream_ordering_to_exterm", + "users_in_public_rooms", + "users_who_share_private_rooms", + "appservice_room_list", + "e2e_room_keys", + "event_push_summary", + "pusher_throttle", + "group_summary_rooms", + "local_invites", + "room_account_data", + "room_tags", + ): + count = self.get_success( + self.store._simple_select_one_onecol( + table="events", + keyvalues={"room_id": room_id}, + retcol="COUNT(*)", + desc="test_purge_room", + ) + ) + + self.assertEqual(count, 0, msg="Rows not purged in {}".format(table)) diff --git a/tests/server.py b/tests/server.py index e397ebe8fa..469efb4edb 100644 --- a/tests/server.py +++ b/tests/server.py @@ -161,7 +161,11 @@ def make_request( path = path.encode("ascii") # Decorate it to be the full path, if we're using shorthand - if shorthand and not path.startswith(b"/_matrix"): + if ( + shorthand + and not path.startswith(b"/_matrix") + and not path.startswith(b"/_synapse") + ): path = b"/_matrix/client/r0/" + path path = path.replace(b"//", b"/") -- cgit 1.4.1 From c6dbca2422bf77ccbf0b52d9245d28c258dac4f3 Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Fri, 1 Nov 2019 10:30:51 +0000 Subject: Incorporate review --- changelog.d/6301.feature | 2 +- synapse/api/constants.py | 5 ++++- synapse/api/filtering.py | 6 ++++-- synapse/storage/data_stores/main/events.py | 12 ++++++++++-- tests/api/test_filtering.py | 10 +++++----- tests/rest/client/v1/test_rooms.py | 10 +++++----- tests/rest/client/v2_alpha/test_sync.py | 10 +++++----- 7 files changed, 34 insertions(+), 21 deletions(-) (limited to 'tests') diff --git a/changelog.d/6301.feature b/changelog.d/6301.feature index b7ff3fad3b..78a187a1dc 100644 --- a/changelog.d/6301.feature +++ b/changelog.d/6301.feature @@ -1 +1 @@ -Implement label-based filtering. +Implement label-based filtering on `/sync` and `/messages` ([MSC2326](https://github.com/matrix-org/matrix-doc/pull/2326)). diff --git a/synapse/api/constants.py b/synapse/api/constants.py index 999ec02fd9..cf4ce5f5a2 100644 --- a/synapse/api/constants.py +++ b/synapse/api/constants.py @@ -140,4 +140,7 @@ class LimitBlockingTypes(object): HS_DISABLED = "hs_disabled" -LabelsField = "org.matrix.labels" +class EventContentFields(object): + """Fields found in events' content, regardless of type.""" + # Labels for the event, cf https://github.com/matrix-org/matrix-doc/pull/2326 + Labels = "org.matrix.labels" diff --git a/synapse/api/filtering.py b/synapse/api/filtering.py index bd91b9f018..30a7ee0a7a 100644 --- a/synapse/api/filtering.py +++ b/synapse/api/filtering.py @@ -20,7 +20,7 @@ from jsonschema import FormatChecker from twisted.internet import defer -from synapse.api.constants import LabelsField +from synapse.api.constants import EventContentFields from synapse.api.errors import SynapseError from synapse.storage.presence import UserPresenceState from synapse.types import RoomID, UserID @@ -67,6 +67,8 @@ ROOM_EVENT_FILTER_SCHEMA = { "contains_url": {"type": "boolean"}, "lazy_load_members": {"type": "boolean"}, "include_redundant_members": {"type": "boolean"}, + # Include or exclude events with the provided labels. + # cf https://github.com/matrix-org/matrix-doc/pull/2326 "org.matrix.labels": {"type": "array", "items": {"type": "string"}}, "org.matrix.not_labels": {"type": "array", "items": {"type": "string"}}, }, @@ -307,7 +309,7 @@ class Filter(object): content = event.get("content", {}) # check if there is a string url field in the content for filtering purposes contains_url = isinstance(content.get("url"), text_type) - labels = content.get(LabelsField, []) + labels = content.get(EventContentFields.Labels, []) return self.check_fields(room_id, sender, ev_type, labels, contains_url) diff --git a/synapse/storage/data_stores/main/events.py b/synapse/storage/data_stores/main/events.py index 2b900f1ce1..42ffa9066a 100644 --- a/synapse/storage/data_stores/main/events.py +++ b/synapse/storage/data_stores/main/events.py @@ -29,7 +29,7 @@ from prometheus_client import Counter, Histogram from twisted.internet import defer import synapse.metrics -from synapse.api.constants import EventTypes, LabelsField +from synapse.api.constants import EventTypes, EventContentFields from synapse.api.errors import SynapseError from synapse.events import EventBase # noqa: F401 from synapse.events.snapshot import EventContext # noqa: F401 @@ -1491,7 +1491,7 @@ class EventsStore( self._handle_event_relations(txn, event) # Store the labels for this event. - labels = event.content.get(LabelsField) + labels = event.content.get(EventContentFields.Labels) if labels: self.insert_labels_for_event_txn(txn, event.event_id, labels) @@ -2483,6 +2483,14 @@ class EventsStore( ) def insert_labels_for_event_txn(self, txn, event_id, labels): + """Store the mapping between an event's ID and its labels, with one row per + (event_id, label) tuple. + + Args: + txn (LoggingTransaction): The transaction to execute. + event_id (str): The event's ID. + labels (list[str]): A list of text labels. + """ return self._simple_insert_many_txn( txn=txn, table="event_labels", diff --git a/tests/api/test_filtering.py b/tests/api/test_filtering.py index e004ab1ee5..8ec48c4154 100644 --- a/tests/api/test_filtering.py +++ b/tests/api/test_filtering.py @@ -19,7 +19,7 @@ import jsonschema from twisted.internet import defer -from synapse.api.constants import LabelsField +from synapse.api.constants import EventContentFields from synapse.api.errors import SynapseError from synapse.api.filtering import Filter from synapse.events import FrozenEvent @@ -329,7 +329,7 @@ class FilteringTestCase(unittest.TestCase): sender="@foo:bar", type="m.room.message", room_id="!secretbase:unknown", - content={LabelsField: ["#fun"]}, + content={EventContentFields.Labels: ["#fun"]}, ) self.assertTrue(Filter(definition).check(event)) @@ -338,7 +338,7 @@ class FilteringTestCase(unittest.TestCase): sender="@foo:bar", type="m.room.message", room_id="!secretbase:unknown", - content={LabelsField: ["#notfun"]}, + content={EventContentFields.Labels: ["#notfun"]}, ) self.assertFalse(Filter(definition).check(event)) @@ -349,7 +349,7 @@ class FilteringTestCase(unittest.TestCase): sender="@foo:bar", type="m.room.message", room_id="!secretbase:unknown", - content={LabelsField: ["#fun"]}, + content={EventContentFields.Labels: ["#fun"]}, ) self.assertFalse(Filter(definition).check(event)) @@ -358,7 +358,7 @@ class FilteringTestCase(unittest.TestCase): sender="@foo:bar", type="m.room.message", room_id="!secretbase:unknown", - content={LabelsField: ["#notfun"]}, + content={EventContentFields.Labels: ["#notfun"]}, ) self.assertTrue(Filter(definition).check(event)) diff --git a/tests/rest/client/v1/test_rooms.py b/tests/rest/client/v1/test_rooms.py index 188f47bd7d..0dc0faa0e5 100644 --- a/tests/rest/client/v1/test_rooms.py +++ b/tests/rest/client/v1/test_rooms.py @@ -24,7 +24,7 @@ from six.moves.urllib import parse as urlparse from twisted.internet import defer import synapse.rest.admin -from synapse.api.constants import EventTypes, LabelsField, Membership +from synapse.api.constants import EventContentFields, EventTypes, Membership from synapse.rest.client.v1 import login, profile, room from tests import unittest @@ -860,7 +860,7 @@ class RoomMessageListTestCase(RoomBase): content={ "msgtype": "m.text", "body": "with right label", - LabelsField: ["#fun"], + EventContentFields.Labels: ["#fun"], }, ) @@ -876,7 +876,7 @@ class RoomMessageListTestCase(RoomBase): content={ "msgtype": "m.text", "body": "with wrong label", - LabelsField: ["#work"], + EventContentFields.Labels: ["#work"], }, ) @@ -886,7 +886,7 @@ class RoomMessageListTestCase(RoomBase): content={ "msgtype": "m.text", "body": "with two wrong labels", - LabelsField: ["#work", "#notfun"], + EventContentFields.Labels: ["#work", "#notfun"], }, ) @@ -896,7 +896,7 @@ class RoomMessageListTestCase(RoomBase): content={ "msgtype": "m.text", "body": "with right label", - LabelsField: ["#fun"], + EventContentFields.Labels: ["#fun"], }, ) diff --git a/tests/rest/client/v2_alpha/test_sync.py b/tests/rest/client/v2_alpha/test_sync.py index c5c199d412..c3c6f75ced 100644 --- a/tests/rest/client/v2_alpha/test_sync.py +++ b/tests/rest/client/v2_alpha/test_sync.py @@ -17,7 +17,7 @@ import json from mock import Mock import synapse.rest.admin -from synapse.api.constants import EventTypes, LabelsField +from synapse.api.constants import EventContentFields, EventTypes from synapse.rest.client.v1 import login, room from synapse.rest.client.v2_alpha import sync @@ -157,7 +157,7 @@ class SyncFilterTestCase(unittest.HomeserverTestCase): content={ "msgtype": "m.text", "body": "with right label", - LabelsField: ["#fun"], + EventContentFields.Labels: ["#fun"], }, tok=tok, ) @@ -175,7 +175,7 @@ class SyncFilterTestCase(unittest.HomeserverTestCase): content={ "msgtype": "m.text", "body": "with wrong label", - LabelsField: ["#work"], + EventContentFields.Labels: ["#work"], }, tok=tok, ) @@ -186,7 +186,7 @@ class SyncFilterTestCase(unittest.HomeserverTestCase): content={ "msgtype": "m.text", "body": "with two wrong labels", - LabelsField: ["#work", "#notfun"], + EventContentFields.Labels: ["#work", "#notfun"], }, tok=tok, ) @@ -197,7 +197,7 @@ class SyncFilterTestCase(unittest.HomeserverTestCase): content={ "msgtype": "m.text", "body": "with right label", - LabelsField: ["#fun"], + EventContentFields.Labels: ["#fun"], }, tok=tok, ) -- cgit 1.4.1 From 1cb84c6486a5131dd284f341bb657434becda255 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Fri, 1 Nov 2019 14:07:44 +0000 Subject: Support for routing outbound HTTP requests via a proxy (#6239) The `http_proxy` and `HTTPS_PROXY` env vars can be set to a `host[:port]` value which should point to a proxy. The address of the proxy should be excluded from IP blacklists such as the `url_preview_ip_range_blacklist`. The proxy will then be used for * push * url previews * phone-home stats * recaptcha validation * CAS auth validation It will *not* be used for: * Application Services * Identity servers * Outbound federation * In worker configurations, connections from workers to masters Fixes #4198. --- changelog.d/6238.feature | 1 + synapse/app/homeserver.py | 2 +- synapse/handlers/ui_auth/checkers.py | 2 +- synapse/http/client.py | 17 +- synapse/http/connectproxyclient.py | 195 ++++++++++++ synapse/http/proxyagent.py | 195 ++++++++++++ synapse/push/httppusher.py | 2 +- synapse/rest/client/v1/login.py | 2 +- synapse/rest/media/v1/preview_url_resource.py | 2 + synapse/server.py | 9 + synapse/server.pyi | 9 + tests/http/__init__.py | 17 ++ .../federation/test_matrix_federation_agent.py | 11 +- tests/http/test_proxyagent.py | 334 +++++++++++++++++++++ tests/push/test_http.py | 2 +- tests/server.py | 24 +- 16 files changed, 812 insertions(+), 12 deletions(-) create mode 100644 changelog.d/6238.feature create mode 100644 synapse/http/connectproxyclient.py create mode 100644 synapse/http/proxyagent.py create mode 100644 tests/http/test_proxyagent.py (limited to 'tests') diff --git a/changelog.d/6238.feature b/changelog.d/6238.feature new file mode 100644 index 0000000000..d225ac33b6 --- /dev/null +++ b/changelog.d/6238.feature @@ -0,0 +1 @@ +Add support for outbound http proxying via http_proxy/HTTPS_PROXY env vars. diff --git a/synapse/app/homeserver.py b/synapse/app/homeserver.py index 8997c1f9e7..8d28076d92 100644 --- a/synapse/app/homeserver.py +++ b/synapse/app/homeserver.py @@ -565,7 +565,7 @@ def run(hs): "Reporting stats to %s: %s" % (hs.config.report_stats_endpoint, stats) ) try: - yield hs.get_simple_http_client().put_json( + yield hs.get_proxied_http_client().put_json( hs.config.report_stats_endpoint, stats ) except Exception as e: diff --git a/synapse/handlers/ui_auth/checkers.py b/synapse/handlers/ui_auth/checkers.py index 29aa1e5aaf..8363d887a9 100644 --- a/synapse/handlers/ui_auth/checkers.py +++ b/synapse/handlers/ui_auth/checkers.py @@ -81,7 +81,7 @@ class RecaptchaAuthChecker(UserInteractiveAuthChecker): def __init__(self, hs): super().__init__(hs) self._enabled = bool(hs.config.recaptcha_private_key) - self._http_client = hs.get_simple_http_client() + self._http_client = hs.get_proxied_http_client() self._url = hs.config.recaptcha_siteverify_api self._secret = hs.config.recaptcha_private_key diff --git a/synapse/http/client.py b/synapse/http/client.py index 2df5b383b5..d4c285445e 100644 --- a/synapse/http/client.py +++ b/synapse/http/client.py @@ -45,6 +45,7 @@ from synapse.http import ( cancelled_to_request_timed_out_error, redact_uri, ) +from synapse.http.proxyagent import ProxyAgent from synapse.logging.context import make_deferred_yieldable from synapse.logging.opentracing import set_tag, start_active_span, tags from synapse.util.async_helpers import timeout_deferred @@ -183,7 +184,15 @@ class SimpleHttpClient(object): using HTTP in Matrix """ - def __init__(self, hs, treq_args={}, ip_whitelist=None, ip_blacklist=None): + def __init__( + self, + hs, + treq_args={}, + ip_whitelist=None, + ip_blacklist=None, + http_proxy=None, + https_proxy=None, + ): """ Args: hs (synapse.server.HomeServer) @@ -192,6 +201,8 @@ class SimpleHttpClient(object): we may not request. ip_whitelist (netaddr.IPSet): The whitelisted IP addresses, that we can request if it were otherwise caught in a blacklist. + http_proxy (bytes): proxy server to use for http connections. host[:port] + https_proxy (bytes): proxy server to use for https connections. host[:port] """ self.hs = hs @@ -236,11 +247,13 @@ class SimpleHttpClient(object): # The default context factory in Twisted 14.0.0 (which we require) is # BrowserLikePolicyForHTTPS which will do regular cert validation # 'like a browser' - self.agent = Agent( + self.agent = ProxyAgent( self.reactor, connectTimeout=15, contextFactory=self.hs.get_http_client_context_factory(), pool=pool, + http_proxy=http_proxy, + https_proxy=https_proxy, ) if self._ip_blacklist: diff --git a/synapse/http/connectproxyclient.py b/synapse/http/connectproxyclient.py new file mode 100644 index 0000000000..be7b2ceb8e --- /dev/null +++ b/synapse/http/connectproxyclient.py @@ -0,0 +1,195 @@ +# -*- coding: utf-8 -*- +# Copyright 2019 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging + +from zope.interface import implementer + +from twisted.internet import defer, protocol +from twisted.internet.error import ConnectError +from twisted.internet.interfaces import IStreamClientEndpoint +from twisted.internet.protocol import connectionDone +from twisted.web import http + +logger = logging.getLogger(__name__) + + +class ProxyConnectError(ConnectError): + pass + + +@implementer(IStreamClientEndpoint) +class HTTPConnectProxyEndpoint(object): + """An Endpoint implementation which will send a CONNECT request to an http proxy + + Wraps an existing HostnameEndpoint for the proxy. + + When we get the connect() request from the connection pool (via the TLS wrapper), + we'll first connect to the proxy endpoint with a ProtocolFactory which will make the + CONNECT request. Once that completes, we invoke the protocolFactory which was passed + in. + + Args: + reactor: the Twisted reactor to use for the connection + proxy_endpoint (IStreamClientEndpoint): the endpoint to use to connect to the + proxy + host (bytes): hostname that we want to CONNECT to + port (int): port that we want to connect to + """ + + def __init__(self, reactor, proxy_endpoint, host, port): + self._reactor = reactor + self._proxy_endpoint = proxy_endpoint + self._host = host + self._port = port + + def __repr__(self): + return "" % (self._proxy_endpoint,) + + def connect(self, protocolFactory): + f = HTTPProxiedClientFactory(self._host, self._port, protocolFactory) + d = self._proxy_endpoint.connect(f) + # once the tcp socket connects successfully, we need to wait for the + # CONNECT to complete. + d.addCallback(lambda conn: f.on_connection) + return d + + +class HTTPProxiedClientFactory(protocol.ClientFactory): + """ClientFactory wrapper that triggers an HTTP proxy CONNECT on connect. + + Once the CONNECT completes, invokes the original ClientFactory to build the + HTTP Protocol object and run the rest of the connection. + + Args: + dst_host (bytes): hostname that we want to CONNECT to + dst_port (int): port that we want to connect to + wrapped_factory (protocol.ClientFactory): The original Factory + """ + + def __init__(self, dst_host, dst_port, wrapped_factory): + self.dst_host = dst_host + self.dst_port = dst_port + self.wrapped_factory = wrapped_factory + self.on_connection = defer.Deferred() + + def startedConnecting(self, connector): + return self.wrapped_factory.startedConnecting(connector) + + def buildProtocol(self, addr): + wrapped_protocol = self.wrapped_factory.buildProtocol(addr) + + return HTTPConnectProtocol( + self.dst_host, self.dst_port, wrapped_protocol, self.on_connection + ) + + def clientConnectionFailed(self, connector, reason): + logger.debug("Connection to proxy failed: %s", reason) + if not self.on_connection.called: + self.on_connection.errback(reason) + return self.wrapped_factory.clientConnectionFailed(connector, reason) + + def clientConnectionLost(self, connector, reason): + logger.debug("Connection to proxy lost: %s", reason) + if not self.on_connection.called: + self.on_connection.errback(reason) + return self.wrapped_factory.clientConnectionLost(connector, reason) + + +class HTTPConnectProtocol(protocol.Protocol): + """Protocol that wraps an existing Protocol to do a CONNECT handshake at connect + + Args: + host (bytes): The original HTTP(s) hostname or IPv4 or IPv6 address literal + to put in the CONNECT request + + port (int): The original HTTP(s) port to put in the CONNECT request + + wrapped_protocol (interfaces.IProtocol): the original protocol (probably + HTTPChannel or TLSMemoryBIOProtocol, but could be anything really) + + connected_deferred (Deferred): a Deferred which will be callbacked with + wrapped_protocol when the CONNECT completes + """ + + def __init__(self, host, port, wrapped_protocol, connected_deferred): + self.host = host + self.port = port + self.wrapped_protocol = wrapped_protocol + self.connected_deferred = connected_deferred + self.http_setup_client = HTTPConnectSetupClient(self.host, self.port) + self.http_setup_client.on_connected.addCallback(self.proxyConnected) + + def connectionMade(self): + self.http_setup_client.makeConnection(self.transport) + + def connectionLost(self, reason=connectionDone): + if self.wrapped_protocol.connected: + self.wrapped_protocol.connectionLost(reason) + + self.http_setup_client.connectionLost(reason) + + if not self.connected_deferred.called: + self.connected_deferred.errback(reason) + + def proxyConnected(self, _): + self.wrapped_protocol.makeConnection(self.transport) + + self.connected_deferred.callback(self.wrapped_protocol) + + # Get any pending data from the http buf and forward it to the original protocol + buf = self.http_setup_client.clearLineBuffer() + if buf: + self.wrapped_protocol.dataReceived(buf) + + def dataReceived(self, data): + # if we've set up the HTTP protocol, we can send the data there + if self.wrapped_protocol.connected: + return self.wrapped_protocol.dataReceived(data) + + # otherwise, we must still be setting up the connection: send the data to the + # setup client + return self.http_setup_client.dataReceived(data) + + +class HTTPConnectSetupClient(http.HTTPClient): + """HTTPClient protocol to send a CONNECT message for proxies and read the response. + + Args: + host (bytes): The hostname to send in the CONNECT message + port (int): The port to send in the CONNECT message + """ + + def __init__(self, host, port): + self.host = host + self.port = port + self.on_connected = defer.Deferred() + + def connectionMade(self): + logger.debug("Connected to proxy, sending CONNECT") + self.sendCommand(b"CONNECT", b"%s:%d" % (self.host, self.port)) + self.endHeaders() + + def handleStatus(self, version, status, message): + logger.debug("Got Status: %s %s %s", status, message, version) + if status != b"200": + raise ProxyConnectError("Unexpected status on CONNECT: %s" % status) + + def handleEndHeaders(self): + logger.debug("End Headers") + self.on_connected.callback(None) + + def handleResponse(self, body): + pass diff --git a/synapse/http/proxyagent.py b/synapse/http/proxyagent.py new file mode 100644 index 0000000000..332da02a8d --- /dev/null +++ b/synapse/http/proxyagent.py @@ -0,0 +1,195 @@ +# -*- coding: utf-8 -*- +# Copyright 2019 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import logging +import re + +from zope.interface import implementer + +from twisted.internet import defer +from twisted.internet.endpoints import HostnameEndpoint, wrapClientTLS +from twisted.python.failure import Failure +from twisted.web.client import URI, BrowserLikePolicyForHTTPS, _AgentBase +from twisted.web.error import SchemeNotSupported +from twisted.web.iweb import IAgent + +from synapse.http.connectproxyclient import HTTPConnectProxyEndpoint + +logger = logging.getLogger(__name__) + +_VALID_URI = re.compile(br"\A[\x21-\x7e]+\Z") + + +@implementer(IAgent) +class ProxyAgent(_AgentBase): + """An Agent implementation which will use an HTTP proxy if one was requested + + Args: + reactor: twisted reactor to place outgoing + connections. + + contextFactory (IPolicyForHTTPS): A factory for TLS contexts, to control the + verification parameters of OpenSSL. The default is to use a + `BrowserLikePolicyForHTTPS`, so unless you have special + requirements you can leave this as-is. + + connectTimeout (float): The amount of time that this Agent will wait + for the peer to accept a connection. + + bindAddress (bytes): The local address for client sockets to bind to. + + pool (HTTPConnectionPool|None): connection pool to be used. If None, a + non-persistent pool instance will be created. + """ + + def __init__( + self, + reactor, + contextFactory=BrowserLikePolicyForHTTPS(), + connectTimeout=None, + bindAddress=None, + pool=None, + http_proxy=None, + https_proxy=None, + ): + _AgentBase.__init__(self, reactor, pool) + + self._endpoint_kwargs = {} + if connectTimeout is not None: + self._endpoint_kwargs["timeout"] = connectTimeout + if bindAddress is not None: + self._endpoint_kwargs["bindAddress"] = bindAddress + + self.http_proxy_endpoint = _http_proxy_endpoint( + http_proxy, reactor, **self._endpoint_kwargs + ) + + self.https_proxy_endpoint = _http_proxy_endpoint( + https_proxy, reactor, **self._endpoint_kwargs + ) + + self._policy_for_https = contextFactory + self._reactor = reactor + + def request(self, method, uri, headers=None, bodyProducer=None): + """ + Issue a request to the server indicated by the given uri. + + Supports `http` and `https` schemes. + + An existing connection from the connection pool may be used or a new one may be + created. + + See also: twisted.web.iweb.IAgent.request + + Args: + method (bytes): The request method to use, such as `GET`, `POST`, etc + + uri (bytes): The location of the resource to request. + + headers (Headers|None): Extra headers to send with the request + + bodyProducer (IBodyProducer|None): An object which can generate bytes to + make up the body of this request (for example, the properly encoded + contents of a file for a file upload). Or, None if the request is to + have no body. + + Returns: + Deferred[IResponse]: completes when the header of the response has + been received (regardless of the response status code). + """ + uri = uri.strip() + if not _VALID_URI.match(uri): + raise ValueError("Invalid URI {!r}".format(uri)) + + parsed_uri = URI.fromBytes(uri) + pool_key = (parsed_uri.scheme, parsed_uri.host, parsed_uri.port) + request_path = parsed_uri.originForm + + if parsed_uri.scheme == b"http" and self.http_proxy_endpoint: + # Cache *all* connections under the same key, since we are only + # connecting to a single destination, the proxy: + pool_key = ("http-proxy", self.http_proxy_endpoint) + endpoint = self.http_proxy_endpoint + request_path = uri + elif parsed_uri.scheme == b"https" and self.https_proxy_endpoint: + endpoint = HTTPConnectProxyEndpoint( + self._reactor, + self.https_proxy_endpoint, + parsed_uri.host, + parsed_uri.port, + ) + else: + # not using a proxy + endpoint = HostnameEndpoint( + self._reactor, parsed_uri.host, parsed_uri.port, **self._endpoint_kwargs + ) + + logger.debug("Requesting %s via %s", uri, endpoint) + + if parsed_uri.scheme == b"https": + tls_connection_creator = self._policy_for_https.creatorForNetloc( + parsed_uri.host, parsed_uri.port + ) + endpoint = wrapClientTLS(tls_connection_creator, endpoint) + elif parsed_uri.scheme == b"http": + pass + else: + return defer.fail( + Failure( + SchemeNotSupported("Unsupported scheme: %r" % (parsed_uri.scheme,)) + ) + ) + + return self._requestWithEndpoint( + pool_key, endpoint, method, parsed_uri, headers, bodyProducer, request_path + ) + + +def _http_proxy_endpoint(proxy, reactor, **kwargs): + """Parses an http proxy setting and returns an endpoint for the proxy + + Args: + proxy (bytes|None): the proxy setting + reactor: reactor to be used to connect to the proxy + kwargs: other args to be passed to HostnameEndpoint + + Returns: + interfaces.IStreamClientEndpoint|None: endpoint to use to connect to the proxy, + or None + """ + if proxy is None: + return None + + # currently we only support hostname:port. Some apps also support + # protocol://[:port], which allows a way of requiring a TLS connection to the + # proxy. + + host, port = parse_host_port(proxy, default_port=1080) + return HostnameEndpoint(reactor, host, port, **kwargs) + + +def parse_host_port(hostport, default_port=None): + # could have sworn we had one of these somewhere else... + if b":" in hostport: + host, port = hostport.rsplit(b":", 1) + try: + port = int(port) + return host, port + except ValueError: + # the thing after the : wasn't a valid port; presumably this is an + # IPv6 address. + pass + + return hostport, default_port diff --git a/synapse/push/httppusher.py b/synapse/push/httppusher.py index 7dde2ad055..e994037be6 100644 --- a/synapse/push/httppusher.py +++ b/synapse/push/httppusher.py @@ -103,7 +103,7 @@ class HttpPusher(object): if "url" not in self.data: raise PusherConfigException("'url' required in data for HTTP pusher") self.url = self.data["url"] - self.http_client = hs.get_simple_http_client() + self.http_client = hs.get_proxied_http_client() self.data_minus_url = {} self.data_minus_url.update(self.data) del self.data_minus_url["url"] diff --git a/synapse/rest/client/v1/login.py b/synapse/rest/client/v1/login.py index 00a7dd6d09..24a0ce74f2 100644 --- a/synapse/rest/client/v1/login.py +++ b/synapse/rest/client/v1/login.py @@ -381,7 +381,7 @@ class CasTicketServlet(RestServlet): self.cas_displayname_attribute = hs.config.cas_displayname_attribute self.cas_required_attributes = hs.config.cas_required_attributes self._sso_auth_handler = SSOAuthHandler(hs) - self._http_client = hs.get_simple_http_client() + self._http_client = hs.get_proxied_http_client() @defer.inlineCallbacks def on_GET(self, request): diff --git a/synapse/rest/media/v1/preview_url_resource.py b/synapse/rest/media/v1/preview_url_resource.py index 5a25b6b3fc..531d923f76 100644 --- a/synapse/rest/media/v1/preview_url_resource.py +++ b/synapse/rest/media/v1/preview_url_resource.py @@ -74,6 +74,8 @@ class PreviewUrlResource(DirectServeResource): treq_args={"browser_like_redirects": True}, ip_whitelist=hs.config.url_preview_ip_range_whitelist, ip_blacklist=hs.config.url_preview_ip_range_blacklist, + http_proxy=os.getenv("http_proxy"), + https_proxy=os.getenv("HTTPS_PROXY"), ) self.media_repo = media_repo self.primary_base_path = media_repo.primary_base_path diff --git a/synapse/server.py b/synapse/server.py index 0b81af646c..f8aeebcff8 100644 --- a/synapse/server.py +++ b/synapse/server.py @@ -23,6 +23,7 @@ # Imports required for the default HomeServer() implementation import abc import logging +import os from twisted.enterprise import adbapi from twisted.mail.smtp import sendmail @@ -168,6 +169,7 @@ class HomeServer(object): "filtering", "http_client_context_factory", "simple_http_client", + "proxied_http_client", "media_repository", "media_repository_resource", "federation_transport_client", @@ -311,6 +313,13 @@ class HomeServer(object): def build_simple_http_client(self): return SimpleHttpClient(self) + def build_proxied_http_client(self): + return SimpleHttpClient( + self, + http_proxy=os.getenv("http_proxy"), + https_proxy=os.getenv("HTTPS_PROXY"), + ) + def build_room_creation_handler(self): return RoomCreationHandler(self) diff --git a/synapse/server.pyi b/synapse/server.pyi index 83d1f11283..b5e0b57095 100644 --- a/synapse/server.pyi +++ b/synapse/server.pyi @@ -12,6 +12,7 @@ import synapse.handlers.message import synapse.handlers.room import synapse.handlers.room_member import synapse.handlers.set_password +import synapse.http.client import synapse.rest.media.v1.media_repository import synapse.server_notices.server_notices_manager import synapse.server_notices.server_notices_sender @@ -38,6 +39,14 @@ class HomeServer(object): pass def get_state_resolution_handler(self) -> synapse.state.StateResolutionHandler: pass + def get_simple_http_client(self) -> synapse.http.client.SimpleHttpClient: + """Fetch an HTTP client implementation which doesn't do any blacklisting + or support any HTTP_PROXY settings""" + pass + def get_proxied_http_client(self) -> synapse.http.client.SimpleHttpClient: + """Fetch an HTTP client implementation which doesn't do any blacklisting + but does support HTTP_PROXY settings""" + pass def get_deactivate_account_handler( self, ) -> synapse.handlers.deactivate_account.DeactivateAccountHandler: diff --git a/tests/http/__init__.py b/tests/http/__init__.py index 2d5dba6464..2096ba3c91 100644 --- a/tests/http/__init__.py +++ b/tests/http/__init__.py @@ -20,6 +20,23 @@ from zope.interface import implementer from OpenSSL import SSL from OpenSSL.SSL import Connection from twisted.internet.interfaces import IOpenSSLServerConnectionCreator +from twisted.internet.ssl import Certificate, trustRootFromCertificates +from twisted.web.client import BrowserLikePolicyForHTTPS # noqa: F401 +from twisted.web.iweb import IPolicyForHTTPS # noqa: F401 + + +def get_test_https_policy(): + """Get a test IPolicyForHTTPS which trusts the test CA cert + + Returns: + IPolicyForHTTPS + """ + ca_file = get_test_ca_cert_file() + with open(ca_file) as stream: + content = stream.read() + cert = Certificate.loadPEM(content) + trust_root = trustRootFromCertificates([cert]) + return BrowserLikePolicyForHTTPS(trustRoot=trust_root) def get_test_ca_cert_file(): diff --git a/tests/http/federation/test_matrix_federation_agent.py b/tests/http/federation/test_matrix_federation_agent.py index 71d7025264..cfcd98ff7d 100644 --- a/tests/http/federation/test_matrix_federation_agent.py +++ b/tests/http/federation/test_matrix_federation_agent.py @@ -124,19 +124,24 @@ class MatrixFederationAgentTests(unittest.TestCase): FakeTransport(client_protocol, self.reactor, server_tls_protocol) ) + # grab a hold of the TLS connection, in case it gets torn down + server_tls_connection = server_tls_protocol._tlsConnection + + # fish the test server back out of the server-side TLS protocol. + http_protocol = server_tls_protocol.wrappedProtocol + # give the reactor a pump to get the TLS juices flowing. self.reactor.pump((0.1,)) # check the SNI - server_name = server_tls_protocol._tlsConnection.get_servername() + server_name = server_tls_connection.get_servername() self.assertEqual( server_name, expected_sni, "Expected SNI %s but got %s" % (expected_sni, server_name), ) - # fish the test server back out of the server-side TLS protocol. - return server_tls_protocol.wrappedProtocol + return http_protocol @defer.inlineCallbacks def _make_get_request(self, uri): diff --git a/tests/http/test_proxyagent.py b/tests/http/test_proxyagent.py new file mode 100644 index 0000000000..22abf76515 --- /dev/null +++ b/tests/http/test_proxyagent.py @@ -0,0 +1,334 @@ +# -*- coding: utf-8 -*- +# Copyright 2019 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import logging + +import treq + +from twisted.internet import interfaces # noqa: F401 +from twisted.internet.protocol import Factory +from twisted.protocols.tls import TLSMemoryBIOFactory +from twisted.web.http import HTTPChannel + +from synapse.http.proxyagent import ProxyAgent + +from tests.http import TestServerTLSConnectionFactory, get_test_https_policy +from tests.server import FakeTransport, ThreadedMemoryReactorClock +from tests.unittest import TestCase + +logger = logging.getLogger(__name__) + +HTTPFactory = Factory.forProtocol(HTTPChannel) + + +class MatrixFederationAgentTests(TestCase): + def setUp(self): + self.reactor = ThreadedMemoryReactorClock() + + def _make_connection( + self, client_factory, server_factory, ssl=False, expected_sni=None + ): + """Builds a test server, and completes the outgoing client connection + + Args: + client_factory (interfaces.IProtocolFactory): the the factory that the + application is trying to use to make the outbound connection. We will + invoke it to build the client Protocol + + server_factory (interfaces.IProtocolFactory): a factory to build the + server-side protocol + + ssl (bool): If true, we will expect an ssl connection and wrap + server_factory with a TLSMemoryBIOFactory + + expected_sni (bytes|None): the expected SNI value + + Returns: + IProtocol: the server Protocol returned by server_factory + """ + if ssl: + server_factory = _wrap_server_factory_for_tls(server_factory) + + server_protocol = server_factory.buildProtocol(None) + + # now, tell the client protocol factory to build the client protocol, + # and wire the output of said protocol up to the server via + # a FakeTransport. + # + # Normally this would be done by the TCP socket code in Twisted, but we are + # stubbing that out here. + client_protocol = client_factory.buildProtocol(None) + client_protocol.makeConnection( + FakeTransport(server_protocol, self.reactor, client_protocol) + ) + + # tell the server protocol to send its stuff back to the client, too + server_protocol.makeConnection( + FakeTransport(client_protocol, self.reactor, server_protocol) + ) + + if ssl: + http_protocol = server_protocol.wrappedProtocol + tls_connection = server_protocol._tlsConnection + else: + http_protocol = server_protocol + tls_connection = None + + # give the reactor a pump to get the TLS juices flowing (if needed) + self.reactor.advance(0) + + if expected_sni is not None: + server_name = tls_connection.get_servername() + self.assertEqual( + server_name, + expected_sni, + "Expected SNI %s but got %s" % (expected_sni, server_name), + ) + + return http_protocol + + def test_http_request(self): + agent = ProxyAgent(self.reactor) + + self.reactor.lookups["test.com"] = "1.2.3.4" + d = agent.request(b"GET", b"http://test.com") + + # there should be a pending TCP connection + clients = self.reactor.tcpClients + self.assertEqual(len(clients), 1) + (host, port, client_factory, _timeout, _bindAddress) = clients[0] + self.assertEqual(host, "1.2.3.4") + self.assertEqual(port, 80) + + # make a test server, and wire up the client + http_server = self._make_connection( + client_factory, _get_test_protocol_factory() + ) + + # the FakeTransport is async, so we need to pump the reactor + self.reactor.advance(0) + + # now there should be a pending request + self.assertEqual(len(http_server.requests), 1) + + request = http_server.requests[0] + self.assertEqual(request.method, b"GET") + self.assertEqual(request.path, b"/") + self.assertEqual(request.requestHeaders.getRawHeaders(b"host"), [b"test.com"]) + request.write(b"result") + request.finish() + + self.reactor.advance(0) + + resp = self.successResultOf(d) + body = self.successResultOf(treq.content(resp)) + self.assertEqual(body, b"result") + + def test_https_request(self): + agent = ProxyAgent(self.reactor, contextFactory=get_test_https_policy()) + + self.reactor.lookups["test.com"] = "1.2.3.4" + d = agent.request(b"GET", b"https://test.com/abc") + + # there should be a pending TCP connection + clients = self.reactor.tcpClients + self.assertEqual(len(clients), 1) + (host, port, client_factory, _timeout, _bindAddress) = clients[0] + self.assertEqual(host, "1.2.3.4") + self.assertEqual(port, 443) + + # make a test server, and wire up the client + http_server = self._make_connection( + client_factory, + _get_test_protocol_factory(), + ssl=True, + expected_sni=b"test.com", + ) + + # the FakeTransport is async, so we need to pump the reactor + self.reactor.advance(0) + + # now there should be a pending request + self.assertEqual(len(http_server.requests), 1) + + request = http_server.requests[0] + self.assertEqual(request.method, b"GET") + self.assertEqual(request.path, b"/abc") + self.assertEqual(request.requestHeaders.getRawHeaders(b"host"), [b"test.com"]) + request.write(b"result") + request.finish() + + self.reactor.advance(0) + + resp = self.successResultOf(d) + body = self.successResultOf(treq.content(resp)) + self.assertEqual(body, b"result") + + def test_http_request_via_proxy(self): + agent = ProxyAgent(self.reactor, http_proxy=b"proxy.com:8888") + + self.reactor.lookups["proxy.com"] = "1.2.3.5" + d = agent.request(b"GET", b"http://test.com") + + # there should be a pending TCP connection + clients = self.reactor.tcpClients + self.assertEqual(len(clients), 1) + (host, port, client_factory, _timeout, _bindAddress) = clients[0] + self.assertEqual(host, "1.2.3.5") + self.assertEqual(port, 8888) + + # make a test server, and wire up the client + http_server = self._make_connection( + client_factory, _get_test_protocol_factory() + ) + + # the FakeTransport is async, so we need to pump the reactor + self.reactor.advance(0) + + # now there should be a pending request + self.assertEqual(len(http_server.requests), 1) + + request = http_server.requests[0] + self.assertEqual(request.method, b"GET") + self.assertEqual(request.path, b"http://test.com") + self.assertEqual(request.requestHeaders.getRawHeaders(b"host"), [b"test.com"]) + request.write(b"result") + request.finish() + + self.reactor.advance(0) + + resp = self.successResultOf(d) + body = self.successResultOf(treq.content(resp)) + self.assertEqual(body, b"result") + + def test_https_request_via_proxy(self): + agent = ProxyAgent( + self.reactor, + contextFactory=get_test_https_policy(), + https_proxy=b"proxy.com", + ) + + self.reactor.lookups["proxy.com"] = "1.2.3.5" + d = agent.request(b"GET", b"https://test.com/abc") + + # there should be a pending TCP connection + clients = self.reactor.tcpClients + self.assertEqual(len(clients), 1) + (host, port, client_factory, _timeout, _bindAddress) = clients[0] + self.assertEqual(host, "1.2.3.5") + self.assertEqual(port, 1080) + + # make a test HTTP server, and wire up the client + proxy_server = self._make_connection( + client_factory, _get_test_protocol_factory() + ) + + # fish the transports back out so that we can do the old switcheroo + s2c_transport = proxy_server.transport + client_protocol = s2c_transport.other + c2s_transport = client_protocol.transport + + # the FakeTransport is async, so we need to pump the reactor + self.reactor.advance(0) + + # now there should be a pending CONNECT request + self.assertEqual(len(proxy_server.requests), 1) + + request = proxy_server.requests[0] + self.assertEqual(request.method, b"CONNECT") + self.assertEqual(request.path, b"test.com:443") + + # tell the proxy server not to close the connection + proxy_server.persistent = True + + # this just stops the http Request trying to do a chunked response + # request.setHeader(b"Content-Length", b"0") + request.finish() + + # now we can replace the proxy channel with a new, SSL-wrapped HTTP channel + ssl_factory = _wrap_server_factory_for_tls(_get_test_protocol_factory()) + ssl_protocol = ssl_factory.buildProtocol(None) + http_server = ssl_protocol.wrappedProtocol + + ssl_protocol.makeConnection( + FakeTransport(client_protocol, self.reactor, ssl_protocol) + ) + c2s_transport.other = ssl_protocol + + self.reactor.advance(0) + + server_name = ssl_protocol._tlsConnection.get_servername() + expected_sni = b"test.com" + self.assertEqual( + server_name, + expected_sni, + "Expected SNI %s but got %s" % (expected_sni, server_name), + ) + + # now there should be a pending request + self.assertEqual(len(http_server.requests), 1) + + request = http_server.requests[0] + self.assertEqual(request.method, b"GET") + self.assertEqual(request.path, b"/abc") + self.assertEqual(request.requestHeaders.getRawHeaders(b"host"), [b"test.com"]) + request.write(b"result") + request.finish() + + self.reactor.advance(0) + + resp = self.successResultOf(d) + body = self.successResultOf(treq.content(resp)) + self.assertEqual(body, b"result") + + +def _wrap_server_factory_for_tls(factory, sanlist=None): + """Wrap an existing Protocol Factory with a test TLSMemoryBIOFactory + + The resultant factory will create a TLS server which presents a certificate + signed by our test CA, valid for the domains in `sanlist` + + Args: + factory (interfaces.IProtocolFactory): protocol factory to wrap + sanlist (iterable[bytes]): list of domains the cert should be valid for + + Returns: + interfaces.IProtocolFactory + """ + if sanlist is None: + sanlist = [b"DNS:test.com"] + + connection_creator = TestServerTLSConnectionFactory(sanlist=sanlist) + return TLSMemoryBIOFactory( + connection_creator, isClient=False, wrappedFactory=factory + ) + + +def _get_test_protocol_factory(): + """Get a protocol Factory which will build an HTTPChannel + + Returns: + interfaces.IProtocolFactory + """ + server_factory = Factory.forProtocol(HTTPChannel) + + # Request.finish expects the factory to have a 'log' method. + server_factory.log = _log_request + + return server_factory + + +def _log_request(request): + """Implements Factory.log, which is expected by Request.finish""" + logger.info("Completed request %s", request) diff --git a/tests/push/test_http.py b/tests/push/test_http.py index 8ce6bb62da..af2327fb66 100644 --- a/tests/push/test_http.py +++ b/tests/push/test_http.py @@ -50,7 +50,7 @@ class HTTPPusherTests(HomeserverTestCase): config = self.default_config() config["start_pushers"] = True - hs = self.setup_test_homeserver(config=config, simple_http_client=m) + hs = self.setup_test_homeserver(config=config, proxied_http_client=m) return hs diff --git a/tests/server.py b/tests/server.py index 469efb4edb..f878aeaada 100644 --- a/tests/server.py +++ b/tests/server.py @@ -395,11 +395,24 @@ class FakeTransport(object): self.disconnecting = True if self._protocol: self._protocol.connectionLost(reason) - self.disconnected = True + + # if we still have data to write, delay until that is done + if self.buffer: + logger.info( + "FakeTransport: Delaying disconnect until buffer is flushed" + ) + else: + self.disconnected = True def abortConnection(self): logger.info("FakeTransport: abortConnection()") - self.loseConnection() + + if not self.disconnecting: + self.disconnecting = True + if self._protocol: + self._protocol.connectionLost(None) + + self.disconnected = True def pauseProducing(self): if not self.producer: @@ -430,6 +443,9 @@ class FakeTransport(object): self._reactor.callLater(0.0, _produce) def write(self, byt): + if self.disconnecting: + raise Exception("Writing to disconnecting FakeTransport") + self.buffer = self.buffer + byt # always actually do the write asynchronously. Some protocols (notably the @@ -474,6 +490,10 @@ class FakeTransport(object): if self.buffer and self.autoflush: self._reactor.callLater(0.0, self.flush) + if not self.buffer and self.disconnecting: + logger.info("FakeTransport: Buffer now empty, completing disconnect") + self.disconnected = True + def connect_client(reactor: IReactorTCP, client_id: int) -> AccumulatingProtocol: """ -- cgit 1.4.1 From c6516adbe03a0acdd614ba6eb9d6f447dd4259e9 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Fri, 1 Nov 2019 16:19:09 +0000 Subject: Factor out an _AsyncEventContextImpl (#6298) The intention here is to make it clearer which fields we can expect to be populated when: notably, that the _event_type etc aren't used for the synchronous impl of EventContext. --- changelog.d/6298.misc | 1 + synapse/events/snapshot.py | 107 ++++++++++++++++------------------------- synapse/handlers/federation.py | 38 +++++++-------- tests/test_federation.py | 4 +- 4 files changed, 65 insertions(+), 85 deletions(-) create mode 100644 changelog.d/6298.misc (limited to 'tests') diff --git a/changelog.d/6298.misc b/changelog.d/6298.misc new file mode 100644 index 0000000000..d4190730b2 --- /dev/null +++ b/changelog.d/6298.misc @@ -0,0 +1 @@ +Refactor EventContext for clarity. \ No newline at end of file diff --git a/synapse/events/snapshot.py b/synapse/events/snapshot.py index 27cd8a63ff..a269de5482 100644 --- a/synapse/events/snapshot.py +++ b/synapse/events/snapshot.py @@ -37,9 +37,6 @@ class EventContext: delta_ids (dict[(str, str), str]): Delta from ``prev_group``. (type, state_key) -> event_id. ``None`` for an outlier. - prev_state_events (?): XXX: is this ever set to anything other than - the empty list? - app_service: FIXME _current_state_ids (dict[(str, str), str]|None): @@ -51,36 +48,16 @@ class EventContext: The current state map excluding the current event. None if outlier or we haven't fetched the state from DB yet. (type, state_key) -> event_id - - _fetching_state_deferred (Deferred|None): Resolves when *_state_ids have - been calculated. None if we haven't started calculating yet - - _event_type (str): The type of the event the context is associated with. - Only set when state has not been fetched yet. - - _event_state_key (str|None): The state_key of the event the context is - associated with. Only set when state has not been fetched yet. - - _prev_state_id (str|None): If the event associated with the context is - a state event, then `_prev_state_id` is the event_id of the state - that was replaced. - Only set when state has not been fetched yet. """ state_group = attr.ib(default=None) rejected = attr.ib(default=False) prev_group = attr.ib(default=None) delta_ids = attr.ib(default=None) - prev_state_events = attr.ib(default=attr.Factory(list)) app_service = attr.ib(default=None) - _current_state_ids = attr.ib(default=None) _prev_state_ids = attr.ib(default=None) - _prev_state_id = attr.ib(default=None) - - _event_type = attr.ib(default=None) - _event_state_key = attr.ib(default=None) - _fetching_state_deferred = attr.ib(default=None) + _current_state_ids = attr.ib(default=None) @staticmethod def with_state( @@ -90,7 +67,6 @@ class EventContext: current_state_ids=current_state_ids, prev_state_ids=prev_state_ids, state_group=state_group, - fetching_state_deferred=defer.succeed(None), prev_group=prev_group, delta_ids=delta_ids, ) @@ -125,7 +101,6 @@ class EventContext: "rejected": self.rejected, "prev_group": self.prev_group, "delta_ids": _encode_state_dict(self.delta_ids), - "prev_state_events": self.prev_state_events, "app_service_id": self.app_service.id if self.app_service else None, } @@ -141,7 +116,7 @@ class EventContext: Returns: EventContext """ - context = EventContext( + context = _AsyncEventContextImpl( # We use the state_group and prev_state_id stuff to pull the # current_state_ids out of the DB and construct prev_state_ids. prev_state_id=input["prev_state_id"], @@ -151,7 +126,6 @@ class EventContext: prev_group=input["prev_group"], delta_ids=_decode_state_dict(input["delta_ids"]), rejected=input["rejected"], - prev_state_events=input["prev_state_events"], ) app_service_id = input["app_service_id"] @@ -170,14 +144,7 @@ class EventContext: Maps a (type, state_key) to the event ID of the state event matching this tuple. """ - - if not self._fetching_state_deferred: - self._fetching_state_deferred = run_in_background( - self._fill_out_state, store - ) - - yield make_deferred_yieldable(self._fetching_state_deferred) - + yield self._ensure_fetched(store) return self._current_state_ids @defer.inlineCallbacks @@ -190,14 +157,7 @@ class EventContext: Maps a (type, state_key) to the event ID of the state event matching this tuple. """ - - if not self._fetching_state_deferred: - self._fetching_state_deferred = run_in_background( - self._fill_out_state, store - ) - - yield make_deferred_yieldable(self._fetching_state_deferred) - + yield self._ensure_fetched(store) return self._prev_state_ids def get_cached_current_state_ids(self): @@ -211,6 +171,44 @@ class EventContext: return self._current_state_ids + def _ensure_fetched(self, store): + return defer.succeed(None) + + +@attr.s(slots=True) +class _AsyncEventContextImpl(EventContext): + """ + An implementation of EventContext which fetches _current_state_ids and + _prev_state_ids from the database on demand. + + Attributes: + + _fetching_state_deferred (Deferred|None): Resolves when *_state_ids have + been calculated. None if we haven't started calculating yet + + _event_type (str): The type of the event the context is associated with. + + _event_state_key (str): The state_key of the event the context is + associated with. + + _prev_state_id (str|None): If the event associated with the context is + a state event, then `_prev_state_id` is the event_id of the state + that was replaced. + """ + + _prev_state_id = attr.ib(default=None) + _event_type = attr.ib(default=None) + _event_state_key = attr.ib(default=None) + _fetching_state_deferred = attr.ib(default=None) + + def _ensure_fetched(self, store): + if not self._fetching_state_deferred: + self._fetching_state_deferred = run_in_background( + self._fill_out_state, store + ) + + return make_deferred_yieldable(self._fetching_state_deferred) + @defer.inlineCallbacks def _fill_out_state(self, store): """Called to populate the _current_state_ids and _prev_state_ids @@ -228,27 +226,6 @@ class EventContext: else: self._prev_state_ids = self._current_state_ids - @defer.inlineCallbacks - def update_state( - self, state_group, prev_state_ids, current_state_ids, prev_group, delta_ids - ): - """Replace the state in the context - """ - - # We need to make sure we wait for any ongoing fetching of state - # to complete so that the updated state doesn't get clobbered - if self._fetching_state_deferred: - yield make_deferred_yieldable(self._fetching_state_deferred) - - self.state_group = state_group - self._prev_state_ids = prev_state_ids - self.prev_group = prev_group - self._current_state_ids = current_state_ids - self.delta_ids = delta_ids - - # We need to ensure that that we've marked as having fetched the state - self._fetching_state_deferred = defer.succeed(None) - def _encode_state_dict(state_dict): """Since dicts of (type, state_key) -> event_id cannot be serialized in diff --git a/synapse/handlers/federation.py b/synapse/handlers/federation.py index dab6be9573..8cafcfdab0 100644 --- a/synapse/handlers/federation.py +++ b/synapse/handlers/federation.py @@ -45,6 +45,7 @@ from synapse.api.errors import ( from synapse.api.room_versions import KNOWN_ROOM_VERSIONS, RoomVersions from synapse.crypto.event_signing import compute_event_signature from synapse.event_auth import auth_types_for_event +from synapse.events.snapshot import EventContext from synapse.events.validator import EventValidator from synapse.logging.context import ( make_deferred_yieldable, @@ -1871,14 +1872,7 @@ class FederationHandler(BaseHandler): if c and c.type == EventTypes.Create: auth_events[(c.type, c.state_key)] = c - try: - yield self.do_auth(origin, event, context, auth_events=auth_events) - except AuthError as e: - logger.warning( - "[%s %s] Rejecting: %s", event.room_id, event.event_id, e.msg - ) - - context.rejected = RejectedReason.AUTH_ERROR + context = yield self.do_auth(origin, event, context, auth_events=auth_events) if not context.rejected: yield self._check_for_soft_fail(event, state, backfilled) @@ -2047,12 +2041,12 @@ class FederationHandler(BaseHandler): Also NB that this function adds entries to it. Returns: - defer.Deferred[None] + defer.Deferred[EventContext]: updated context object """ room_version = yield self.store.get_room_version(event.room_id) try: - yield self._update_auth_events_and_context_for_auth( + context = yield self._update_auth_events_and_context_for_auth( origin, event, context, auth_events ) except Exception: @@ -2070,7 +2064,9 @@ class FederationHandler(BaseHandler): event_auth.check(room_version, event, auth_events=auth_events) except AuthError as e: logger.warning("Failed auth resolution for %r because %s", event, e) - raise e + context.rejected = RejectedReason.AUTH_ERROR + + return context @defer.inlineCallbacks def _update_auth_events_and_context_for_auth( @@ -2094,7 +2090,7 @@ class FederationHandler(BaseHandler): auth_events (dict[(str, str)->synapse.events.EventBase]): Returns: - defer.Deferred[None] + defer.Deferred[EventContext]: updated context """ event_auth_events = set(event.auth_event_ids()) @@ -2133,7 +2129,7 @@ class FederationHandler(BaseHandler): # The other side isn't around or doesn't implement the # endpoint, so lets just bail out. logger.info("Failed to get event auth from remote: %s", e) - return + return context seen_remotes = yield self.store.have_seen_events( [e.event_id for e in remote_auth_chain] @@ -2174,7 +2170,7 @@ class FederationHandler(BaseHandler): if event.internal_metadata.is_outlier(): logger.info("Skipping auth_event fetch for outlier") - return + return context # FIXME: Assumes we have and stored all the state for all the # prev_events @@ -2183,7 +2179,7 @@ class FederationHandler(BaseHandler): ) if not different_auth: - return + return context logger.info( "auth_events refers to events which are not in our calculated auth " @@ -2230,10 +2226,12 @@ class FederationHandler(BaseHandler): auth_events.update(new_state) - yield self._update_context_for_auth_events( + context = yield self._update_context_for_auth_events( event, context, auth_events, event_key ) + return context + @defer.inlineCallbacks def _update_context_for_auth_events(self, event, context, auth_events, event_key): """Update the state_ids in an event context after auth event resolution, @@ -2242,14 +2240,16 @@ class FederationHandler(BaseHandler): Args: event (Event): The event we're handling the context for - context (synapse.events.snapshot.EventContext): event context - to be updated + context (synapse.events.snapshot.EventContext): initial event context auth_events (dict[(str, str)->str]): Events to update in the event context. event_key ((str, str)): (type, state_key) for the current event. this will not be included in the current_state in the context. + + Returns: + Deferred[EventContext]: new event context """ state_updates = { k: a.event_id for k, a in iteritems(auth_events) if k != event_key @@ -2274,7 +2274,7 @@ class FederationHandler(BaseHandler): current_state_ids=current_state_ids, ) - yield context.update_state( + return EventContext.with_state( state_group=state_group, current_state_ids=current_state_ids, prev_state_ids=prev_state_ids, diff --git a/tests/test_federation.py b/tests/test_federation.py index d1acb16f30..7d82b58466 100644 --- a/tests/test_federation.py +++ b/tests/test_federation.py @@ -59,7 +59,9 @@ class MessageAcceptTests(unittest.TestCase): ) self.handler = self.homeserver.get_handlers().federation_handler - self.handler.do_auth = lambda *a, **b: succeed(True) + self.handler.do_auth = lambda origin, event, context, auth_events: succeed( + context + ) self.client = self.homeserver.get_federation_client() self.client._check_sigs_and_hash_and_fetch = lambda dest, pdus, **k: succeed( pdus -- cgit 1.4.1 From 988d8d6507a0e8b34f2c352c77b5742197762190 Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Fri, 1 Nov 2019 16:22:44 +0000 Subject: Incorporate review --- synapse/api/constants.py | 2 +- synapse/api/filtering.py | 2 +- synapse/storage/data_stores/main/events.py | 2 +- synapse/storage/data_stores/main/schema/delta/56/event_labels.sql | 6 ++++++ tests/api/test_filtering.py | 8 ++++---- tests/rest/client/v1/test_rooms.py | 8 ++++---- tests/rest/client/v2_alpha/test_sync.py | 8 ++++---- 7 files changed, 21 insertions(+), 15 deletions(-) (limited to 'tests') diff --git a/synapse/api/constants.py b/synapse/api/constants.py index 066ce18704..49c4b85054 100644 --- a/synapse/api/constants.py +++ b/synapse/api/constants.py @@ -144,4 +144,4 @@ class EventContentFields(object): """Fields found in events' content, regardless of type.""" # Labels for the event, cf https://github.com/matrix-org/matrix-doc/pull/2326 - Labels = "org.matrix.labels" + LABELS = "org.matrix.labels" diff --git a/synapse/api/filtering.py b/synapse/api/filtering.py index 30a7ee0a7a..bec13f08d8 100644 --- a/synapse/api/filtering.py +++ b/synapse/api/filtering.py @@ -309,7 +309,7 @@ class Filter(object): content = event.get("content", {}) # check if there is a string url field in the content for filtering purposes contains_url = isinstance(content.get("url"), text_type) - labels = content.get(EventContentFields.Labels, []) + labels = content.get(EventContentFields.LABELS, []) return self.check_fields(room_id, sender, ev_type, labels, contains_url) diff --git a/synapse/storage/data_stores/main/events.py b/synapse/storage/data_stores/main/events.py index 577e79bcf9..1045c7fa2e 100644 --- a/synapse/storage/data_stores/main/events.py +++ b/synapse/storage/data_stores/main/events.py @@ -1491,7 +1491,7 @@ class EventsStore( self._handle_event_relations(txn, event) # Store the labels for this event. - labels = event.content.get(EventContentFields.Labels) + labels = event.content.get(EventContentFields.LABELS) if labels: self.insert_labels_for_event_txn( txn, event.event_id, labels, event.room_id, event.depth diff --git a/synapse/storage/data_stores/main/schema/delta/56/event_labels.sql b/synapse/storage/data_stores/main/schema/delta/56/event_labels.sql index 2acd8e1be5..5e29c1da19 100644 --- a/synapse/storage/data_stores/main/schema/delta/56/event_labels.sql +++ b/synapse/storage/data_stores/main/schema/delta/56/event_labels.sql @@ -13,6 +13,8 @@ * limitations under the License. */ +-- room_id and topoligical_ordering are denormalised from the events table in order to +-- make the index work. CREATE TABLE IF NOT EXISTS event_labels ( event_id TEXT, label TEXT, @@ -21,4 +23,8 @@ CREATE TABLE IF NOT EXISTS event_labels ( PRIMARY KEY(event_id, label) ); + +-- This index enables an event pagination looking for a particular label to index the +-- event_labels table first, which is much quicker than scanning the events table and then +-- filtering by label, if the label is rarely used relative to the size of the room. CREATE INDEX event_labels_room_id_label_idx ON event_labels(room_id, label, topological_ordering); diff --git a/tests/api/test_filtering.py b/tests/api/test_filtering.py index 8ec48c4154..2dc5052249 100644 --- a/tests/api/test_filtering.py +++ b/tests/api/test_filtering.py @@ -329,7 +329,7 @@ class FilteringTestCase(unittest.TestCase): sender="@foo:bar", type="m.room.message", room_id="!secretbase:unknown", - content={EventContentFields.Labels: ["#fun"]}, + content={EventContentFields.LABELS: ["#fun"]}, ) self.assertTrue(Filter(definition).check(event)) @@ -338,7 +338,7 @@ class FilteringTestCase(unittest.TestCase): sender="@foo:bar", type="m.room.message", room_id="!secretbase:unknown", - content={EventContentFields.Labels: ["#notfun"]}, + content={EventContentFields.LABELS: ["#notfun"]}, ) self.assertFalse(Filter(definition).check(event)) @@ -349,7 +349,7 @@ class FilteringTestCase(unittest.TestCase): sender="@foo:bar", type="m.room.message", room_id="!secretbase:unknown", - content={EventContentFields.Labels: ["#fun"]}, + content={EventContentFields.LABELS: ["#fun"]}, ) self.assertFalse(Filter(definition).check(event)) @@ -358,7 +358,7 @@ class FilteringTestCase(unittest.TestCase): sender="@foo:bar", type="m.room.message", room_id="!secretbase:unknown", - content={EventContentFields.Labels: ["#notfun"]}, + content={EventContentFields.LABELS: ["#notfun"]}, ) self.assertTrue(Filter(definition).check(event)) diff --git a/tests/rest/client/v1/test_rooms.py b/tests/rest/client/v1/test_rooms.py index 0dc0faa0e5..5e38fd6ced 100644 --- a/tests/rest/client/v1/test_rooms.py +++ b/tests/rest/client/v1/test_rooms.py @@ -860,7 +860,7 @@ class RoomMessageListTestCase(RoomBase): content={ "msgtype": "m.text", "body": "with right label", - EventContentFields.Labels: ["#fun"], + EventContentFields.LABELS: ["#fun"], }, ) @@ -876,7 +876,7 @@ class RoomMessageListTestCase(RoomBase): content={ "msgtype": "m.text", "body": "with wrong label", - EventContentFields.Labels: ["#work"], + EventContentFields.LABELS: ["#work"], }, ) @@ -886,7 +886,7 @@ class RoomMessageListTestCase(RoomBase): content={ "msgtype": "m.text", "body": "with two wrong labels", - EventContentFields.Labels: ["#work", "#notfun"], + EventContentFields.LABELS: ["#work", "#notfun"], }, ) @@ -896,7 +896,7 @@ class RoomMessageListTestCase(RoomBase): content={ "msgtype": "m.text", "body": "with right label", - EventContentFields.Labels: ["#fun"], + EventContentFields.LABELS: ["#fun"], }, ) diff --git a/tests/rest/client/v2_alpha/test_sync.py b/tests/rest/client/v2_alpha/test_sync.py index c3c6f75ced..3283c0e47b 100644 --- a/tests/rest/client/v2_alpha/test_sync.py +++ b/tests/rest/client/v2_alpha/test_sync.py @@ -157,7 +157,7 @@ class SyncFilterTestCase(unittest.HomeserverTestCase): content={ "msgtype": "m.text", "body": "with right label", - EventContentFields.Labels: ["#fun"], + EventContentFields.LABELS: ["#fun"], }, tok=tok, ) @@ -175,7 +175,7 @@ class SyncFilterTestCase(unittest.HomeserverTestCase): content={ "msgtype": "m.text", "body": "with wrong label", - EventContentFields.Labels: ["#work"], + EventContentFields.LABELS: ["#work"], }, tok=tok, ) @@ -186,7 +186,7 @@ class SyncFilterTestCase(unittest.HomeserverTestCase): content={ "msgtype": "m.text", "body": "with two wrong labels", - EventContentFields.Labels: ["#work", "#notfun"], + EventContentFields.LABELS: ["#work", "#notfun"], }, tok=tok, ) @@ -197,7 +197,7 @@ class SyncFilterTestCase(unittest.HomeserverTestCase): content={ "msgtype": "m.text", "body": "with right label", - EventContentFields.Labels: ["#fun"], + EventContentFields.LABELS: ["#fun"], }, tok=tok, ) -- cgit 1.4.1 From 09957ce0e4dcfd84c2de4039653059faae03065b Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Mon, 4 Nov 2019 17:09:22 +0000 Subject: Implement per-room message retention policies --- changelog.d/5815.feature | 1 + docs/sample_config.yaml | 63 ++++ synapse/api/constants.py | 2 + synapse/config/server.py | 172 +++++++++++ synapse/events/validator.py | 100 ++++++- synapse/handlers/federation.py | 2 +- synapse/handlers/message.py | 4 +- synapse/handlers/pagination.py | 111 +++++++ synapse/storage/data_stores/main/events.py | 3 + synapse/storage/data_stores/main/room.py | 252 ++++++++++++++++ .../main/schema/delta/56/room_retention.sql | 33 +++ synapse/visibility.py | 17 ++ tests/rest/client/test_retention.py | 320 +++++++++++++++++++++ 13 files changed, 1074 insertions(+), 6 deletions(-) create mode 100644 changelog.d/5815.feature create mode 100644 synapse/storage/data_stores/main/schema/delta/56/room_retention.sql create mode 100644 tests/rest/client/test_retention.py (limited to 'tests') diff --git a/changelog.d/5815.feature b/changelog.d/5815.feature new file mode 100644 index 0000000000..ca4df4e7f6 --- /dev/null +++ b/changelog.d/5815.feature @@ -0,0 +1 @@ +Implement per-room message retention policies. diff --git a/docs/sample_config.yaml b/docs/sample_config.yaml index d2f4aff826..87fba27d13 100644 --- a/docs/sample_config.yaml +++ b/docs/sample_config.yaml @@ -328,6 +328,69 @@ listeners: # #user_ips_max_age: 14d +# Message retention policy at the server level. +# +# Room admins and mods can define a retention period for their rooms using the +# 'm.room.retention' state event, and server admins can cap this period by setting +# the 'allowed_lifetime_min' and 'allowed_lifetime_max' config options. +# +# If this feature is enabled, Synapse will regularly look for and purge events +# which are older than the room's maximum retention period. Synapse will also +# filter events received over federation so that events that should have been +# purged are ignored and not stored again. +# +retention: + # The message retention policies feature is disabled by default. Uncomment the + # following line to enable it. + # + #enabled: true + + # Default retention policy. If set, Synapse will apply it to rooms that lack the + # 'm.room.retention' state event. Currently, the value of 'min_lifetime' doesn't + # matter much because Synapse doesn't take it into account yet. + # + #default_policy: + # min_lifetime: 1d + # max_lifetime: 1y + + # Retention policy limits. If set, a user won't be able to send a + # 'm.room.retention' event which features a 'min_lifetime' or a 'max_lifetime' + # that's not within this range. This is especially useful in closed federations, + # in which server admins can make sure every federating server applies the same + # rules. + # + #allowed_lifetime_min: 1d + #allowed_lifetime_max: 1y + + # Server admins can define the settings of the background jobs purging the + # events which lifetime has expired under the 'purge_jobs' section. + # + # If no configuration is provided, a single job will be set up to delete expired + # events in every room daily. + # + # Each job's configuration defines which range of message lifetimes the job + # takes care of. For example, if 'shortest_max_lifetime' is '2d' and + # 'longest_max_lifetime' is '3d', the job will handle purging expired events in + # rooms whose state defines a 'max_lifetime' that's both higher than 2 days, and + # lower than or equal to 3 days. Both the minimum and the maximum value of a + # range are optional, e.g. a job with no 'shortest_max_lifetime' and a + # 'longest_max_lifetime' of '3d' will handle every room with a retention policy + # which 'max_lifetime' is lower than or equal to three days. + # + # The rationale for this per-job configuration is that some rooms might have a + # retention policy with a low 'max_lifetime', where history needs to be purged + # of outdated messages on a very frequent basis (e.g. every 5min), but not want + # that purge to be performed by a job that's iterating over every room it knows, + # which would be quite heavy on the server. + # + #purge_jobs: + # - shortest_max_lifetime: 1d + # longest_max_lifetime: 3d + # interval: 5m: + # - shortest_max_lifetime: 3d + # longest_max_lifetime: 1y + # interval: 24h + ## TLS ## diff --git a/synapse/api/constants.py b/synapse/api/constants.py index 49c4b85054..e3f086f1c3 100644 --- a/synapse/api/constants.py +++ b/synapse/api/constants.py @@ -94,6 +94,8 @@ class EventTypes(object): ServerACL = "m.room.server_acl" Pinned = "m.room.pinned_events" + Retention = "m.room.retention" + class RejectedReason(object): AUTH_ERROR = "auth_error" diff --git a/synapse/config/server.py b/synapse/config/server.py index d556df308d..aa93a416f1 100644 --- a/synapse/config/server.py +++ b/synapse/config/server.py @@ -246,6 +246,115 @@ class ServerConfig(Config): # events with profile information that differ from the target's global profile. self.allow_per_room_profiles = config.get("allow_per_room_profiles", True) + retention_config = config.get("retention") + if retention_config is None: + retention_config = {} + + self.retention_enabled = retention_config.get("enabled", False) + + retention_default_policy = retention_config.get("default_policy") + + if retention_default_policy is not None: + self.retention_default_min_lifetime = retention_default_policy.get( + "min_lifetime" + ) + if self.retention_default_min_lifetime is not None: + self.retention_default_min_lifetime = self.parse_duration( + self.retention_default_min_lifetime + ) + + self.retention_default_max_lifetime = retention_default_policy.get( + "max_lifetime" + ) + if self.retention_default_max_lifetime is not None: + self.retention_default_max_lifetime = self.parse_duration( + self.retention_default_max_lifetime + ) + + if ( + self.retention_default_min_lifetime is not None + and self.retention_default_max_lifetime is not None + and ( + self.retention_default_min_lifetime + > self.retention_default_max_lifetime + ) + ): + raise ConfigError( + "The default retention policy's 'min_lifetime' can not be greater" + " than its 'max_lifetime'" + ) + else: + self.retention_default_min_lifetime = None + self.retention_default_max_lifetime = None + + self.retention_allowed_lifetime_min = retention_config.get("allowed_lifetime_min") + if self.retention_allowed_lifetime_min is not None: + self.retention_allowed_lifetime_min = self.parse_duration( + self.retention_allowed_lifetime_min + ) + + self.retention_allowed_lifetime_max = retention_config.get("allowed_lifetime_max") + if self.retention_allowed_lifetime_max is not None: + self.retention_allowed_lifetime_max = self.parse_duration( + self.retention_allowed_lifetime_max + ) + + if ( + self.retention_allowed_lifetime_min is not None + and self.retention_allowed_lifetime_max is not None + and self.retention_allowed_lifetime_min > self.retention_allowed_lifetime_max + ): + raise ConfigError( + "Invalid retention policy limits: 'allowed_lifetime_min' can not be" + " greater than 'allowed_lifetime_max'" + ) + + self.retention_purge_jobs = [] + for purge_job_config in retention_config.get("purge_jobs", []): + interval_config = purge_job_config.get("interval") + + if interval_config is None: + raise ConfigError( + "A retention policy's purge jobs configuration must have the" + " 'interval' key set." + ) + + interval = self.parse_duration(interval_config) + + shortest_max_lifetime = purge_job_config.get("shortest_max_lifetime") + + if shortest_max_lifetime is not None: + shortest_max_lifetime = self.parse_duration(shortest_max_lifetime) + + longest_max_lifetime = purge_job_config.get("longest_max_lifetime") + + if longest_max_lifetime is not None: + longest_max_lifetime = self.parse_duration(longest_max_lifetime) + + if ( + shortest_max_lifetime is not None + and longest_max_lifetime is not None + and shortest_max_lifetime > longest_max_lifetime + ): + raise ConfigError( + "A retention policy's purge jobs configuration's" + " 'shortest_max_lifetime' value can not be greater than its" + " 'longest_max_lifetime' value." + ) + + self.retention_purge_jobs.append({ + "interval": interval, + "shortest_max_lifetime": shortest_max_lifetime, + "longest_max_lifetime": longest_max_lifetime, + }) + + if not self.retention_purge_jobs: + self.retention_purge_jobs = [{ + "interval": self.parse_duration("1d"), + "shortest_max_lifetime": None, + "longest_max_lifetime": None, + }] + self.listeners = [] # type: List[dict] for listener in config.get("listeners", []): if not isinstance(listener.get("port", None), int): @@ -761,6 +870,69 @@ class ServerConfig(Config): # Defaults to `28d`. Set to `null` to disable clearing out of old rows. # #user_ips_max_age: 14d + + # Message retention policy at the server level. + # + # Room admins and mods can define a retention period for their rooms using the + # 'm.room.retention' state event, and server admins can cap this period by setting + # the 'allowed_lifetime_min' and 'allowed_lifetime_max' config options. + # + # If this feature is enabled, Synapse will regularly look for and purge events + # which are older than the room's maximum retention period. Synapse will also + # filter events received over federation so that events that should have been + # purged are ignored and not stored again. + # + retention: + # The message retention policies feature is disabled by default. Uncomment the + # following line to enable it. + # + #enabled: true + + # Default retention policy. If set, Synapse will apply it to rooms that lack the + # 'm.room.retention' state event. Currently, the value of 'min_lifetime' doesn't + # matter much because Synapse doesn't take it into account yet. + # + #default_policy: + # min_lifetime: 1d + # max_lifetime: 1y + + # Retention policy limits. If set, a user won't be able to send a + # 'm.room.retention' event which features a 'min_lifetime' or a 'max_lifetime' + # that's not within this range. This is especially useful in closed federations, + # in which server admins can make sure every federating server applies the same + # rules. + # + #allowed_lifetime_min: 1d + #allowed_lifetime_max: 1y + + # Server admins can define the settings of the background jobs purging the + # events which lifetime has expired under the 'purge_jobs' section. + # + # If no configuration is provided, a single job will be set up to delete expired + # events in every room daily. + # + # Each job's configuration defines which range of message lifetimes the job + # takes care of. For example, if 'shortest_max_lifetime' is '2d' and + # 'longest_max_lifetime' is '3d', the job will handle purging expired events in + # rooms whose state defines a 'max_lifetime' that's both higher than 2 days, and + # lower than or equal to 3 days. Both the minimum and the maximum value of a + # range are optional, e.g. a job with no 'shortest_max_lifetime' and a + # 'longest_max_lifetime' of '3d' will handle every room with a retention policy + # which 'max_lifetime' is lower than or equal to three days. + # + # The rationale for this per-job configuration is that some rooms might have a + # retention policy with a low 'max_lifetime', where history needs to be purged + # of outdated messages on a very frequent basis (e.g. every 5min), but not want + # that purge to be performed by a job that's iterating over every room it knows, + # which would be quite heavy on the server. + # + #purge_jobs: + # - shortest_max_lifetime: 1d + # longest_max_lifetime: 3d + # interval: 5m: + # - shortest_max_lifetime: 3d + # longest_max_lifetime: 1y + # interval: 24h """ % locals() ) diff --git a/synapse/events/validator.py b/synapse/events/validator.py index 272426e105..9b90c9ce04 100644 --- a/synapse/events/validator.py +++ b/synapse/events/validator.py @@ -13,7 +13,7 @@ # See the License for the specific language governing permissions and # limitations under the License. -from six import string_types +from six import integer_types, string_types from synapse.api.constants import MAX_ALIAS_LENGTH, EventTypes, Membership from synapse.api.errors import Codes, SynapseError @@ -22,11 +22,12 @@ from synapse.types import EventID, RoomID, UserID class EventValidator(object): - def validate_new(self, event): + def validate_new(self, event, config): """Validates the event has roughly the right format Args: - event (FrozenEvent) + event (FrozenEvent): The event to validate. + config (Config): The homeserver's configuration. """ self.validate_builder(event) @@ -67,6 +68,99 @@ class EventValidator(object): Codes.INVALID_PARAM, ) + if event.type == EventTypes.Retention: + self._validate_retention(event, config) + + def _validate_retention(self, event, config): + """Checks that an event that defines the retention policy for a room respects the + boundaries imposed by the server's administrator. + + Args: + event (FrozenEvent): The event to validate. + config (Config): The homeserver's configuration. + """ + min_lifetime = event.content.get("min_lifetime") + max_lifetime = event.content.get("max_lifetime") + + if min_lifetime is not None: + if not isinstance(min_lifetime, integer_types): + raise SynapseError( + code=400, + msg="'min_lifetime' must be an integer", + errcode=Codes.BAD_JSON, + ) + + if ( + config.retention_allowed_lifetime_min is not None + and min_lifetime < config.retention_allowed_lifetime_min + ): + raise SynapseError( + code=400, + msg=( + "'min_lifetime' can't be lower than the minimum allowed" + " value enforced by the server's administrator" + ), + errcode=Codes.BAD_JSON, + ) + + if ( + config.retention_allowed_lifetime_max is not None + and min_lifetime > config.retention_allowed_lifetime_max + ): + raise SynapseError( + code=400, + msg=( + "'min_lifetime' can't be greater than the maximum allowed" + " value enforced by the server's administrator" + ), + errcode=Codes.BAD_JSON, + ) + + if max_lifetime is not None: + if not isinstance(max_lifetime, integer_types): + raise SynapseError( + code=400, + msg="'max_lifetime' must be an integer", + errcode=Codes.BAD_JSON, + ) + + if ( + config.retention_allowed_lifetime_min is not None + and max_lifetime < config.retention_allowed_lifetime_min + ): + raise SynapseError( + code=400, + msg=( + "'max_lifetime' can't be lower than the minimum allowed value" + " enforced by the server's administrator" + ), + errcode=Codes.BAD_JSON, + ) + + if ( + config.retention_allowed_lifetime_max is not None + and max_lifetime > config.retention_allowed_lifetime_max + ): + raise SynapseError( + code=400, + msg=( + "'max_lifetime' can't be greater than the maximum allowed" + " value enforced by the server's administrator" + ), + errcode=Codes.BAD_JSON, + ) + + if ( + min_lifetime is not None + and max_lifetime is not None + and min_lifetime > max_lifetime + ): + raise SynapseError( + code=400, + msg="'min_lifetime' can't be greater than 'max_lifetime", + errcode=Codes.BAD_JSON, + ) + def validate_builder(self, event): """Validates that the builder/event has roughly the right format. Only checks values that we expect a proto event to have, rather than all the diff --git a/synapse/handlers/federation.py b/synapse/handlers/federation.py index 8cafcfdab0..3994137d18 100644 --- a/synapse/handlers/federation.py +++ b/synapse/handlers/federation.py @@ -2454,7 +2454,7 @@ class FederationHandler(BaseHandler): room_version, event_dict, event, context ) - EventValidator().validate_new(event) + EventValidator().validate_new(event, self.config) # We need to tell the transaction queue to send this out, even # though the sender isn't a local user. diff --git a/synapse/handlers/message.py b/synapse/handlers/message.py index d682dc2b7a..155ed6e06a 100644 --- a/synapse/handlers/message.py +++ b/synapse/handlers/message.py @@ -417,7 +417,7 @@ class EventCreationHandler(object): 403, "You must be in the room to create an alias for it" ) - self.validator.validate_new(event) + self.validator.validate_new(event, self.config) return (event, context) @@ -634,7 +634,7 @@ class EventCreationHandler(object): if requester: context.app_service = requester.app_service - self.validator.validate_new(event) + self.validator.validate_new(event, self.config) # If this event is an annotation then we check that that the sender # can't annotate the same way twice (e.g. stops users from liking an diff --git a/synapse/handlers/pagination.py b/synapse/handlers/pagination.py index 97f15a1c32..e1800177fa 100644 --- a/synapse/handlers/pagination.py +++ b/synapse/handlers/pagination.py @@ -15,12 +15,15 @@ # limitations under the License. import logging +from six import iteritems + from twisted.internet import defer from twisted.python.failure import Failure from synapse.api.constants import EventTypes, Membership from synapse.api.errors import SynapseError from synapse.logging.context import run_in_background +from synapse.metrics.background_process_metrics import run_as_background_process from synapse.storage.state import StateFilter from synapse.types import RoomStreamToken from synapse.util.async_helpers import ReadWriteLock @@ -80,6 +83,114 @@ class PaginationHandler(object): self._purges_by_id = {} self._event_serializer = hs.get_event_client_serializer() + self._retention_default_max_lifetime = hs.config.retention_default_max_lifetime + + if hs.config.retention_enabled: + # Run the purge jobs described in the configuration file. + for job in hs.config.retention_purge_jobs: + self.clock.looping_call( + run_as_background_process, + job["interval"], + "purge_history_for_rooms_in_range", + self.purge_history_for_rooms_in_range, + job["shortest_max_lifetime"], + job["longest_max_lifetime"], + ) + + @defer.inlineCallbacks + def purge_history_for_rooms_in_range(self, min_ms, max_ms): + """Purge outdated events from rooms within the given retention range. + + If a default retention policy is defined in the server's configuration and its + 'max_lifetime' is within this range, also targets rooms which don't have a + retention policy. + + Args: + min_ms (int|None): Duration in milliseconds that define the lower limit of + the range to handle (exclusive). If None, it means that the range has no + lower limit. + max_ms (int|None): Duration in milliseconds that define the upper limit of + the range to handle (inclusive). If None, it means that the range has no + upper limit. + """ + # We want the storage layer to to include rooms with no retention policy in its + # return value only if a default retention policy is defined in the server's + # configuration and that policy's 'max_lifetime' is either lower (or equal) than + # max_ms or higher than min_ms (or both). + if self._retention_default_max_lifetime is not None: + include_null = True + + if min_ms is not None and min_ms >= self._retention_default_max_lifetime: + # The default max_lifetime is lower than (or equal to) min_ms. + include_null = False + + if max_ms is not None and max_ms < self._retention_default_max_lifetime: + # The default max_lifetime is higher than max_ms. + include_null = False + else: + include_null = False + + rooms = yield self.store.get_rooms_for_retention_period_in_range( + min_ms, max_ms, include_null + ) + + for room_id, retention_policy in iteritems(rooms): + if room_id in self._purges_in_progress_by_room: + logger.warning( + "[purge] not purging room %s as there's an ongoing purge running" + " for this room", + room_id, + ) + continue + + max_lifetime = retention_policy["max_lifetime"] + + if max_lifetime is None: + # If max_lifetime is None, it means that include_null equals True, + # therefore we can safely assume that there is a default policy defined + # in the server's configuration. + max_lifetime = self._retention_default_max_lifetime + + # Figure out what token we should start purging at. + ts = self.clock.time_msec() - max_lifetime + + stream_ordering = ( + yield self.store.find_first_stream_ordering_after_ts(ts) + ) + + r = ( + yield self.store.get_room_event_after_stream_ordering( + room_id, stream_ordering, + ) + ) + if not r: + logger.warning( + "[purge] purging events not possible: No event found " + "(ts %i => stream_ordering %i)", + ts, stream_ordering, + ) + continue + + (stream, topo, _event_id) = r + token = "t%d-%d" % (topo, stream) + + purge_id = random_string(16) + + self._purges_by_id[purge_id] = PurgeStatus() + + logger.info( + "Starting purging events in room %s (purge_id %s)" % (room_id, purge_id) + ) + + # We want to purge everything, including local events, and to run the purge in + # the background so that it's not blocking any other operation apart from + # other purges in the same room. + run_as_background_process( + "_purge_history", + self._purge_history, + purge_id, room_id, token, True, + ) + def start_purge_history(self, room_id, token, delete_local_events=False): """Start off a history purge on a room. diff --git a/synapse/storage/data_stores/main/events.py b/synapse/storage/data_stores/main/events.py index 301f8ea128..b332a42d82 100644 --- a/synapse/storage/data_stores/main/events.py +++ b/synapse/storage/data_stores/main/events.py @@ -929,6 +929,9 @@ class EventsStore( elif event.type == EventTypes.Redaction: # Insert into the redactions table. self._store_redaction(txn, event) + elif event.type == EventTypes.Retention: + # Update the room_retention table. + self._store_retention_policy_for_room_txn(txn, event) self._handle_event_relations(txn, event) diff --git a/synapse/storage/data_stores/main/room.py b/synapse/storage/data_stores/main/room.py index 67bb1b6f60..54a7d24c73 100644 --- a/synapse/storage/data_stores/main/room.py +++ b/synapse/storage/data_stores/main/room.py @@ -19,10 +19,13 @@ import logging import re from typing import Optional, Tuple +from six import integer_types + from canonicaljson import json from twisted.internet import defer +from synapse.api.constants import EventTypes from synapse.api.errors import StoreError from synapse.storage._base import SQLBaseStore from synapse.storage.data_stores.main.search import SearchStore @@ -302,6 +305,85 @@ class RoomWorkerStore(SQLBaseStore): class RoomStore(RoomWorkerStore, SearchStore): + def __init__(self, db_conn, hs): + super(RoomStore, self).__init__(db_conn, hs) + + self.config = hs.config + + self.register_background_update_handler( + "insert_room_retention", self._background_insert_retention, + ) + + @defer.inlineCallbacks + def _background_insert_retention(self, progress, batch_size): + """Retrieves a list of all rooms within a range and inserts an entry for each of + them into the room_retention table. + NULLs the property's columns if missing from the retention event in the room's + state (or NULLs all of them if there's no retention event in the room's state), + so that we fall back to the server's retention policy. + """ + + last_room = progress.get("room_id", "") + + def _background_insert_retention_txn(txn): + txn.execute( + """ + SELECT state.room_id, state.event_id, events.json + FROM current_state_events as state + LEFT JOIN event_json AS events ON (state.event_id = events.event_id) + WHERE state.room_id > ? AND state.type = '%s' + ORDER BY state.room_id ASC + LIMIT ?; + """ % EventTypes.Retention, + (last_room, batch_size) + ) + + rows = self.cursor_to_dict(txn) + + if not rows: + return True + + for row in rows: + if not row["json"]: + retention_policy = {} + else: + ev = json.loads(row["json"]) + retention_policy = json.dumps(ev["content"]) + + self._simple_insert_txn( + txn=txn, + table="room_retention", + values={ + "room_id": row["room_id"], + "event_id": row["event_id"], + "min_lifetime": retention_policy.get("min_lifetime"), + "max_lifetime": retention_policy.get("max_lifetime"), + } + ) + + logger.info("Inserted %d rows into room_retention", len(rows)) + + self._background_update_progress_txn( + txn, "insert_room_retention", { + "room_id": rows[-1]["room_id"], + } + ) + + if batch_size > len(rows): + return True + else: + return False + + end = yield self.runInteraction( + "insert_room_retention", + _background_insert_retention_txn, + ) + + if end: + yield self._end_background_update("insert_room_retention") + + defer.returnValue(batch_size) + @defer.inlineCallbacks def store_room(self, room_id, room_creator_user_id, is_public): """Stores a room. @@ -502,6 +584,37 @@ class RoomStore(RoomWorkerStore, SearchStore): txn, event, "content.body", event.content["body"] ) + def _store_retention_policy_for_room_txn(self, txn, event): + if ( + hasattr(event, "content") + and ("min_lifetime" in event.content or "max_lifetime" in event.content) + ): + if ( + ("min_lifetime" in event.content and not isinstance( + event.content.get("min_lifetime"), integer_types + )) + or ("max_lifetime" in event.content and not isinstance( + event.content.get("max_lifetime"), integer_types + )) + ): + # Ignore the event if one of the value isn't an integer. + return + + self._simple_insert_txn( + txn=txn, + table="room_retention", + values={ + "room_id": event.room_id, + "event_id": event.event_id, + "min_lifetime": event.content.get("min_lifetime"), + "max_lifetime": event.content.get("max_lifetime"), + }, + ) + + self._invalidate_cache_and_stream( + txn, self.get_retention_policy_for_room, (event.room_id,) + ) + def add_event_report( self, room_id, event_id, user_id, reason, content, received_ts ): @@ -683,3 +796,142 @@ class RoomStore(RoomWorkerStore, SearchStore): remote_media_mxcs.append((hostname, media_id)) return local_media_mxcs, remote_media_mxcs + + @defer.inlineCallbacks + def get_rooms_for_retention_period_in_range(self, min_ms, max_ms, include_null=False): + """Retrieves all of the rooms within the given retention range. + + Optionally includes the rooms which don't have a retention policy. + + Args: + min_ms (int|None): Duration in milliseconds that define the lower limit of + the range to handle (exclusive). If None, doesn't set a lower limit. + max_ms (int|None): Duration in milliseconds that define the upper limit of + the range to handle (inclusive). If None, doesn't set an upper limit. + include_null (bool): Whether to include rooms which retention policy is NULL + in the returned set. + + Returns: + dict[str, dict]: The rooms within this range, along with their retention + policy. The key is "room_id", and maps to a dict describing the retention + policy associated with this room ID. The keys for this nested dict are + "min_lifetime" (int|None), and "max_lifetime" (int|None). + """ + + def get_rooms_for_retention_period_in_range_txn(txn): + range_conditions = [] + args = [] + + if min_ms is not None: + range_conditions.append("max_lifetime > ?") + args.append(min_ms) + + if max_ms is not None: + range_conditions.append("max_lifetime <= ?") + args.append(max_ms) + + # Do a first query which will retrieve the rooms that have a retention policy + # in their current state. + sql = """ + SELECT room_id, min_lifetime, max_lifetime FROM room_retention + INNER JOIN current_state_events USING (event_id, room_id) + """ + + if len(range_conditions): + sql += " WHERE (" + " AND ".join(range_conditions) + ")" + + if include_null: + sql += " OR max_lifetime IS NULL" + + txn.execute(sql, args) + + rows = self.cursor_to_dict(txn) + rooms_dict = {} + + for row in rows: + rooms_dict[row["room_id"]] = { + "min_lifetime": row["min_lifetime"], + "max_lifetime": row["max_lifetime"], + } + + if include_null: + # If required, do a second query that retrieves all of the rooms we know + # of so we can handle rooms with no retention policy. + sql = "SELECT DISTINCT room_id FROM current_state_events" + + txn.execute(sql) + + rows = self.cursor_to_dict(txn) + + # If a room isn't already in the dict (i.e. it doesn't have a retention + # policy in its state), add it with a null policy. + for row in rows: + if row["room_id"] not in rooms_dict: + rooms_dict[row["room_id"]] = { + "min_lifetime": None, + "max_lifetime": None, + } + + return rooms_dict + + rooms = yield self.runInteraction( + "get_rooms_for_retention_period_in_range", + get_rooms_for_retention_period_in_range_txn, + ) + + defer.returnValue(rooms) + + @cachedInlineCallbacks() + def get_retention_policy_for_room(self, room_id): + """Get the retention policy for a given room. + + If no retention policy has been found for this room, returns a policy defined + by the configured default policy (which has None as both the 'min_lifetime' and + the 'max_lifetime' if no default policy has been defined in the server's + configuration). + + Args: + room_id (str): The ID of the room to get the retention policy of. + + Returns: + dict[int, int]: "min_lifetime" and "max_lifetime" for this room. + """ + + def get_retention_policy_for_room_txn(txn): + txn.execute( + """ + SELECT min_lifetime, max_lifetime FROM room_retention + INNER JOIN current_state_events USING (event_id, room_id) + WHERE room_id = ?; + """, + (room_id,) + ) + + return self.cursor_to_dict(txn) + + ret = yield self.runInteraction( + "get_retention_policy_for_room", + get_retention_policy_for_room_txn, + ) + + # If we don't know this room ID, ret will be None, in this case return the default + # policy. + if not ret: + defer.returnValue({ + "min_lifetime": self.config.retention_default_min_lifetime, + "max_lifetime": self.config.retention_default_max_lifetime, + }) + + row = ret[0] + + # If one of the room's policy's attributes isn't defined, use the matching + # attribute from the default policy. + # The default values will be None if no default policy has been defined, or if one + # of the attributes is missing from the default policy. + if row["min_lifetime"] is None: + row["min_lifetime"] = self.config.retention_default_min_lifetime + + if row["max_lifetime"] is None: + row["max_lifetime"] = self.config.retention_default_max_lifetime + + defer.returnValue(row) diff --git a/synapse/storage/data_stores/main/schema/delta/56/room_retention.sql b/synapse/storage/data_stores/main/schema/delta/56/room_retention.sql new file mode 100644 index 0000000000..ee6cdf7a14 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/56/room_retention.sql @@ -0,0 +1,33 @@ +/* Copyright 2019 New Vector Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- Tracks the retention policy of a room. +-- A NULL max_lifetime or min_lifetime means that the matching property is not defined in +-- the room's retention policy state event. +-- If a room doesn't have a retention policy state event in its state, both max_lifetime +-- and min_lifetime are NULL. +CREATE TABLE IF NOT EXISTS room_retention( + room_id TEXT, + event_id TEXT, + min_lifetime BIGINT, + max_lifetime BIGINT, + + PRIMARY KEY(room_id, event_id) +); + +CREATE INDEX room_retention_max_lifetime_idx on room_retention(max_lifetime); + +INSERT INTO background_updates (update_name, progress_json) VALUES + ('insert_room_retention', '{}'); diff --git a/synapse/visibility.py b/synapse/visibility.py index 8c843febd8..4498c156bc 100644 --- a/synapse/visibility.py +++ b/synapse/visibility.py @@ -86,6 +86,14 @@ def filter_events_for_client( erased_senders = yield storage.main.are_users_erased((e.sender for e in events)) + room_ids = set(e.room_id for e in events) + retention_policies = {} + + for room_id in room_ids: + retention_policies[room_id] = yield storage.main.get_retention_policy_for_room( + room_id + ) + def allowed(event): """ Args: @@ -103,6 +111,15 @@ def filter_events_for_client( if not event.is_state() and event.sender in ignore_list: return None + retention_policy = retention_policies[event.room_id] + max_lifetime = retention_policy.get("max_lifetime") + + if max_lifetime is not None: + oldest_allowed_ts = storage.main.clock.time_msec() - max_lifetime + + if event.origin_server_ts < oldest_allowed_ts: + return None + if event.event_id in always_include_ids: return event diff --git a/tests/rest/client/test_retention.py b/tests/rest/client/test_retention.py new file mode 100644 index 0000000000..41ea9db689 --- /dev/null +++ b/tests/rest/client/test_retention.py @@ -0,0 +1,320 @@ +# -*- coding: utf-8 -*- +# Copyright 2019 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +from mock import Mock + +from synapse.api.constants import EventTypes +from synapse.rest import admin +from synapse.rest.client.v1 import login, room +from synapse.visibility import filter_events_for_client + +from tests import unittest + +one_hour_ms = 3600000 +one_day_ms = one_hour_ms * 24 + + +class RetentionTestCase(unittest.HomeserverTestCase): + servlets = [ + admin.register_servlets, + login.register_servlets, + room.register_servlets, + ] + + def make_homeserver(self, reactor, clock): + config = self.default_config() + config["default_room_version"] = "1" + config["retention"] = { + "enabled": True, + "default_policy": { + "min_lifetime": one_day_ms, + "max_lifetime": one_day_ms * 3, + }, + "allowed_lifetime_min": one_day_ms, + "allowed_lifetime_max": one_day_ms * 3, + } + + self.hs = self.setup_test_homeserver(config=config) + return self.hs + + def prepare(self, reactor, clock, homeserver): + self.user_id = self.register_user("user", "password") + self.token = self.login("user", "password") + + def test_retention_state_event(self): + """Tests that the server configuration can limit the values a user can set to the + room's retention policy. + """ + room_id = self.helper.create_room_as(self.user_id, tok=self.token) + + self.helper.send_state( + room_id=room_id, + event_type=EventTypes.Retention, + body={ + "max_lifetime": one_day_ms * 4, + }, + tok=self.token, + expect_code=400, + ) + + self.helper.send_state( + room_id=room_id, + event_type=EventTypes.Retention, + body={ + "max_lifetime": one_hour_ms, + }, + tok=self.token, + expect_code=400, + ) + + def test_retention_event_purged_with_state_event(self): + """Tests that expired events are correctly purged when the room's retention policy + is defined by a state event. + """ + room_id = self.helper.create_room_as(self.user_id, tok=self.token) + + # Set the room's retention period to 2 days. + lifetime = one_day_ms * 2 + self.helper.send_state( + room_id=room_id, + event_type=EventTypes.Retention, + body={ + "max_lifetime": lifetime, + }, + tok=self.token, + ) + + self._test_retention_event_purged(room_id, one_day_ms * 1.5) + + def test_retention_event_purged_without_state_event(self): + """Tests that expired events are correctly purged when the room's retention policy + is defined by the server's configuration's default retention policy. + """ + room_id = self.helper.create_room_as(self.user_id, tok=self.token) + + self._test_retention_event_purged(room_id, one_day_ms * 2) + + def test_visibility(self): + """Tests that synapse.visibility.filter_events_for_client correctly filters out + outdated events + """ + store = self.hs.get_datastore() + storage = self.hs.get_storage() + room_id = self.helper.create_room_as(self.user_id, tok=self.token) + events = [] + + # Send a first event, which should be filtered out at the end of the test. + resp = self.helper.send( + room_id=room_id, + body="1", + tok=self.token, + ) + + # Get the event from the store so that we end up with a FrozenEvent that we can + # give to filter_events_for_client. We need to do this now because the event won't + # be in the database anymore after it has expired. + events.append(self.get_success( + store.get_event( + resp.get("event_id") + ) + )) + + # Advance the time by 2 days. We're using the default retention policy, therefore + # after this the first event will still be valid. + self.reactor.advance(one_day_ms * 2 / 1000) + + # Send another event, which shouldn't get filtered out. + resp = self.helper.send( + room_id=room_id, + body="2", + tok=self.token, + ) + + valid_event_id = resp.get("event_id") + + events.append(self.get_success( + store.get_event( + valid_event_id + ) + )) + + # Advance the time by anothe 2 days. After this, the first event should be + # outdated but not the second one. + self.reactor.advance(one_day_ms * 2 / 1000) + + # Run filter_events_for_client with our list of FrozenEvents. + filtered_events = self.get_success(filter_events_for_client( + storage, self.user_id, events + )) + + # We should only get one event back. + self.assertEqual(len(filtered_events), 1, filtered_events) + # That event should be the second, not outdated event. + self.assertEqual(filtered_events[0].event_id, valid_event_id, filtered_events) + + def _test_retention_event_purged(self, room_id, increment): + # Send a first event to the room. This is the event we'll want to be purged at the + # end of the test. + resp = self.helper.send( + room_id=room_id, + body="1", + tok=self.token, + ) + + expired_event_id = resp.get("event_id") + + # Check that we can retrieve the event. + expired_event = self.get_event(room_id, expired_event_id) + self.assertEqual(expired_event.get("content", {}).get("body"), "1", expired_event) + + # Advance the time. + self.reactor.advance(increment / 1000) + + # Send another event. We need this because the purge job won't purge the most + # recent event in the room. + resp = self.helper.send( + room_id=room_id, + body="2", + tok=self.token, + ) + + valid_event_id = resp.get("event_id") + + # Advance the time again. Now our first event should have expired but our second + # one should still be kept. + self.reactor.advance(increment / 1000) + + # Check that the event has been purged from the database. + self.get_event(room_id, expired_event_id, expected_code=404) + + # Check that the event that hasn't been purged can still be retrieved. + valid_event = self.get_event(room_id, valid_event_id) + self.assertEqual(valid_event.get("content", {}).get("body"), "2", valid_event) + + def get_event(self, room_id, event_id, expected_code=200): + url = "/_matrix/client/r0/rooms/%s/event/%s" % (room_id, event_id) + + request, channel = self.make_request("GET", url, access_token=self.token) + self.render(request) + + self.assertEqual(channel.code, expected_code, channel.result) + + return channel.json_body + + +class RetentionNoDefaultPolicyTestCase(unittest.HomeserverTestCase): + servlets = [ + admin.register_servlets, + login.register_servlets, + room.register_servlets, + ] + + def make_homeserver(self, reactor, clock): + config = self.default_config() + config["default_room_version"] = "1" + config["retention"] = { + "enabled": True, + } + + mock_federation_client = Mock(spec=["backfill"]) + + self.hs = self.setup_test_homeserver( + config=config, + federation_client=mock_federation_client, + ) + return self.hs + + def prepare(self, reactor, clock, homeserver): + self.user_id = self.register_user("user", "password") + self.token = self.login("user", "password") + + def test_no_default_policy(self): + """Tests that an event doesn't get expired if there is neither a default retention + policy nor a policy specific to the room. + """ + room_id = self.helper.create_room_as(self.user_id, tok=self.token) + + self._test_retention(room_id) + + def test_state_policy(self): + """Tests that an event gets correctly expired if there is no default retention + policy but there's a policy specific to the room. + """ + room_id = self.helper.create_room_as(self.user_id, tok=self.token) + + # Set the maximum lifetime to 35 days so that the first event gets expired but not + # the second one. + self.helper.send_state( + room_id=room_id, + event_type=EventTypes.Retention, + body={ + "max_lifetime": one_day_ms * 35, + }, + tok=self.token, + ) + + self._test_retention(room_id, expected_code_for_first_event=404) + + def _test_retention(self, room_id, expected_code_for_first_event=200): + # Send a first event to the room. This is the event we'll want to be purged at the + # end of the test. + resp = self.helper.send( + room_id=room_id, + body="1", + tok=self.token, + ) + + first_event_id = resp.get("event_id") + + # Check that we can retrieve the event. + expired_event = self.get_event(room_id, first_event_id) + self.assertEqual(expired_event.get("content", {}).get("body"), "1", expired_event) + + # Advance the time by a month. + self.reactor.advance(one_day_ms * 30 / 1000) + + # Send another event. We need this because the purge job won't purge the most + # recent event in the room. + resp = self.helper.send( + room_id=room_id, + body="2", + tok=self.token, + ) + + second_event_id = resp.get("event_id") + + # Advance the time by another month. + self.reactor.advance(one_day_ms * 30 / 1000) + + # Check if the event has been purged from the database. + first_event = self.get_event( + room_id, first_event_id, expected_code=expected_code_for_first_event + ) + + if expected_code_for_first_event == 200: + self.assertEqual(first_event.get("content", {}).get("body"), "1", first_event) + + # Check that the event that hasn't been purged can still be retrieved. + second_event = self.get_event(room_id, second_event_id) + self.assertEqual(second_event.get("content", {}).get("body"), "2", second_event) + + def get_event(self, room_id, event_id, expected_code=200): + url = "/_matrix/client/r0/rooms/%s/event/%s" % (room_id, event_id) + + request, channel = self.make_request("GET", url, access_token=self.token) + self.render(request) + + self.assertEqual(channel.code, expected_code, channel.result) + + return channel.json_body -- cgit 1.4.1 From 4e1c7b79fa3498c48106c17d0edbab2f7bcc0c38 Mon Sep 17 00:00:00 2001 From: Amber Brown Date: Tue, 5 Nov 2019 05:05:48 +1100 Subject: Remove the psutil dependency (#6318) * remove psutil and replace with resource --- changelog.d/6318.misc | 1 + synapse/app/homeserver.py | 174 ++++++++++++++++++++++------------------- synapse/python_dependencies.py | 1 - synapse/server.py | 2 + tests/test_phone_home.py | 51 ++++++++++++ 5 files changed, 146 insertions(+), 83 deletions(-) create mode 100644 changelog.d/6318.misc create mode 100644 tests/test_phone_home.py (limited to 'tests') diff --git a/changelog.d/6318.misc b/changelog.d/6318.misc new file mode 100644 index 0000000000..63527ccef4 --- /dev/null +++ b/changelog.d/6318.misc @@ -0,0 +1 @@ +Remove the dependency on psutil and replace functionality with the stdlib `resource` module. diff --git a/synapse/app/homeserver.py b/synapse/app/homeserver.py index 8d28076d92..00a7f8330e 100644 --- a/synapse/app/homeserver.py +++ b/synapse/app/homeserver.py @@ -19,12 +19,13 @@ from __future__ import print_function import gc import logging +import math import os +import resource import sys from six import iteritems -import psutil from prometheus_client import Gauge from twisted.application import service @@ -471,6 +472,87 @@ class SynapseService(service.Service): return self._port.stopListening() +# Contains the list of processes we will be monitoring +# currently either 0 or 1 +_stats_process = [] + + +@defer.inlineCallbacks +def phone_stats_home(hs, stats, stats_process=_stats_process): + logger.info("Gathering stats for reporting") + now = int(hs.get_clock().time()) + uptime = int(now - hs.start_time) + if uptime < 0: + uptime = 0 + + stats["homeserver"] = hs.config.server_name + stats["server_context"] = hs.config.server_context + stats["timestamp"] = now + stats["uptime_seconds"] = uptime + version = sys.version_info + stats["python_version"] = "{}.{}.{}".format( + version.major, version.minor, version.micro + ) + stats["total_users"] = yield hs.get_datastore().count_all_users() + + total_nonbridged_users = yield hs.get_datastore().count_nonbridged_users() + stats["total_nonbridged_users"] = total_nonbridged_users + + daily_user_type_results = yield hs.get_datastore().count_daily_user_type() + for name, count in iteritems(daily_user_type_results): + stats["daily_user_type_" + name] = count + + room_count = yield hs.get_datastore().get_room_count() + stats["total_room_count"] = room_count + + stats["daily_active_users"] = yield hs.get_datastore().count_daily_users() + stats["monthly_active_users"] = yield hs.get_datastore().count_monthly_users() + stats["daily_active_rooms"] = yield hs.get_datastore().count_daily_active_rooms() + stats["daily_messages"] = yield hs.get_datastore().count_daily_messages() + + r30_results = yield hs.get_datastore().count_r30_users() + for name, count in iteritems(r30_results): + stats["r30_users_" + name] = count + + daily_sent_messages = yield hs.get_datastore().count_daily_sent_messages() + stats["daily_sent_messages"] = daily_sent_messages + stats["cache_factor"] = CACHE_SIZE_FACTOR + stats["event_cache_size"] = hs.config.event_cache_size + + # + # Performance statistics + # + old = stats_process[0] + new = (now, resource.getrusage(resource.RUSAGE_SELF)) + stats_process[0] = new + + # Get RSS in bytes + stats["memory_rss"] = new[1].ru_maxrss + + # Get CPU time in % of a single core, not % of all cores + used_cpu_time = (new[1].ru_utime + new[1].ru_stime) - ( + old[1].ru_utime + old[1].ru_stime + ) + if used_cpu_time == 0 or new[0] == old[0]: + stats["cpu_average"] = 0 + else: + stats["cpu_average"] = math.floor(used_cpu_time / (new[0] - old[0]) * 100) + + # + # Database version + # + + stats["database_engine"] = hs.get_datastore().database_engine_name + stats["database_server_version"] = hs.get_datastore().get_server_version() + logger.info("Reporting stats to %s: %s" % (hs.config.report_stats_endpoint, stats)) + try: + yield hs.get_proxied_http_client().put_json( + hs.config.report_stats_endpoint, stats + ) + except Exception as e: + logger.warning("Error reporting stats: %s", e) + + def run(hs): PROFILE_SYNAPSE = False if PROFILE_SYNAPSE: @@ -497,91 +579,19 @@ def run(hs): reactor.run = profile(reactor.run) clock = hs.get_clock() - start_time = clock.time() stats = {} - # Contains the list of processes we will be monitoring - # currently either 0 or 1 - stats_process = [] + def performance_stats_init(): + _stats_process.clear() + _stats_process.append( + (int(hs.get_clock().time(), resource.getrusage(resource.RUSAGE_SELF))) + ) def start_phone_stats_home(): - return run_as_background_process("phone_stats_home", phone_stats_home) - - @defer.inlineCallbacks - def phone_stats_home(): - logger.info("Gathering stats for reporting") - now = int(hs.get_clock().time()) - uptime = int(now - start_time) - if uptime < 0: - uptime = 0 - - stats["homeserver"] = hs.config.server_name - stats["server_context"] = hs.config.server_context - stats["timestamp"] = now - stats["uptime_seconds"] = uptime - version = sys.version_info - stats["python_version"] = "{}.{}.{}".format( - version.major, version.minor, version.micro - ) - stats["total_users"] = yield hs.get_datastore().count_all_users() - - total_nonbridged_users = yield hs.get_datastore().count_nonbridged_users() - stats["total_nonbridged_users"] = total_nonbridged_users - - daily_user_type_results = yield hs.get_datastore().count_daily_user_type() - for name, count in iteritems(daily_user_type_results): - stats["daily_user_type_" + name] = count - - room_count = yield hs.get_datastore().get_room_count() - stats["total_room_count"] = room_count - - stats["daily_active_users"] = yield hs.get_datastore().count_daily_users() - stats["monthly_active_users"] = yield hs.get_datastore().count_monthly_users() - stats[ - "daily_active_rooms" - ] = yield hs.get_datastore().count_daily_active_rooms() - stats["daily_messages"] = yield hs.get_datastore().count_daily_messages() - - r30_results = yield hs.get_datastore().count_r30_users() - for name, count in iteritems(r30_results): - stats["r30_users_" + name] = count - - daily_sent_messages = yield hs.get_datastore().count_daily_sent_messages() - stats["daily_sent_messages"] = daily_sent_messages - stats["cache_factor"] = CACHE_SIZE_FACTOR - stats["event_cache_size"] = hs.config.event_cache_size - - if len(stats_process) > 0: - stats["memory_rss"] = 0 - stats["cpu_average"] = 0 - for process in stats_process: - stats["memory_rss"] += process.memory_info().rss - stats["cpu_average"] += int(process.cpu_percent(interval=None)) - - stats["database_engine"] = hs.get_datastore().database_engine_name - stats["database_server_version"] = hs.get_datastore().get_server_version() - logger.info( - "Reporting stats to %s: %s" % (hs.config.report_stats_endpoint, stats) + return run_as_background_process( + "phone_stats_home", phone_stats_home, hs, stats ) - try: - yield hs.get_proxied_http_client().put_json( - hs.config.report_stats_endpoint, stats - ) - except Exception as e: - logger.warning("Error reporting stats: %s", e) - - def performance_stats_init(): - try: - process = psutil.Process() - # Ensure we can fetch both, and make the initial request for cpu_percent - # so the next request will use this as the initial point. - process.memory_info().rss - process.cpu_percent(interval=None) - logger.info("report_stats can use psutil") - stats_process.append(process) - except (AttributeError): - logger.warning("Unable to read memory/cpu stats. Disabling reporting.") def generate_user_daily_visit_stats(): return run_as_background_process( @@ -626,7 +636,7 @@ def run(hs): if hs.config.report_stats: logger.info("Scheduling stats reporting for 3 hour intervals") - clock.looping_call(start_phone_stats_home, 3 * 60 * 60 * 1000) + clock.looping_call(start_phone_stats_home, 3 * 60 * 60 * 1000, hs, stats) # We need to defer this init for the cases that we daemonize # otherwise the process ID we get is that of the non-daemon process @@ -634,7 +644,7 @@ def run(hs): # We wait 5 minutes to send the first set of stats as the server can # be quite busy the first few minutes - clock.call_later(5 * 60, start_phone_stats_home) + clock.call_later(5 * 60, start_phone_stats_home, hs, stats) _base.start_reactor( "synapse-homeserver", diff --git a/synapse/python_dependencies.py b/synapse/python_dependencies.py index aa7da1c543..5871feaafd 100644 --- a/synapse/python_dependencies.py +++ b/synapse/python_dependencies.py @@ -61,7 +61,6 @@ REQUIREMENTS = [ "bcrypt>=3.1.0", "pillow>=4.3.0", "sortedcontainers>=1.4.4", - "psutil>=2.0.0", "pymacaroons>=0.13.0", "msgpack>=0.5.2", "phonenumbers>=8.2.0", diff --git a/synapse/server.py b/synapse/server.py index f8aeebcff8..90c3b072e8 100644 --- a/synapse/server.py +++ b/synapse/server.py @@ -221,6 +221,7 @@ class HomeServer(object): self.hostname = hostname self._building = {} self._listening_services = [] + self.start_time = None self.clock = Clock(reactor) self.distributor = Distributor() @@ -240,6 +241,7 @@ class HomeServer(object): datastore = self.DATASTORE_CLASS(conn, self) self.datastores = DataStores(datastore, conn, self) conn.commit() + self.start_time = int(self.get_clock().time()) logger.info("Finished setting up.") def setup_master(self): diff --git a/tests/test_phone_home.py b/tests/test_phone_home.py new file mode 100644 index 0000000000..7657bddea5 --- /dev/null +++ b/tests/test_phone_home.py @@ -0,0 +1,51 @@ +# -*- coding: utf-8 -*- +# Copyright 2019 Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import resource + +import mock + +from synapse.app.homeserver import phone_stats_home + +from tests.unittest import HomeserverTestCase + + +class PhoneHomeStatsTestCase(HomeserverTestCase): + def test_performance_frozen_clock(self): + """ + If time doesn't move, don't error out. + """ + past_stats = [ + (self.hs.get_clock().time(), resource.getrusage(resource.RUSAGE_SELF)) + ] + stats = {} + self.get_success(phone_stats_home(self.hs, stats, past_stats)) + self.assertEqual(stats["cpu_average"], 0) + + def test_performance_100(self): + """ + 1 second of usage over 1 second is 100% CPU usage. + """ + real_res = resource.getrusage(resource.RUSAGE_SELF) + old_resource = mock.Mock(spec=real_res) + old_resource.ru_utime = real_res.ru_utime - 1 + old_resource.ru_stime = real_res.ru_stime + old_resource.ru_maxrss = real_res.ru_maxrss + + past_stats = [(self.hs.get_clock().time(), old_resource)] + stats = {} + self.reactor.advance(1) + self.get_success(phone_stats_home(self.hs, stats, past_stats)) + self.assertApproximates(stats["cpu_average"], 100, tolerance=2.5) -- cgit 1.4.1 From a7c818c79b70d6b70abc5b26f0e1e78fd60c087e Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Tue, 5 Nov 2019 13:21:26 +0000 Subject: Add test case --- tests/rest/client/v1/test_rooms.py | 182 +++++++++++++++++++++++++++++++++++++ 1 file changed, 182 insertions(+) (limited to 'tests') diff --git a/tests/rest/client/v1/test_rooms.py b/tests/rest/client/v1/test_rooms.py index 5e38fd6ced..621c894e35 100644 --- a/tests/rest/client/v1/test_rooms.py +++ b/tests/rest/client/v1/test_rooms.py @@ -1106,3 +1106,185 @@ class PerRoomProfilesForbiddenTestCase(unittest.HomeserverTestCase): res_displayname = channel.json_body["content"]["displayname"] self.assertEqual(res_displayname, self.displayname, channel.result) + + +class ContextTestCase(unittest.HomeserverTestCase): + servlets = [ + synapse.rest.admin.register_servlets_for_client_rest_resource, + room.register_servlets, + login.register_servlets, + profile.register_servlets, + ] + + def test_context_filter_labels(self): + """Test that we can filter by a label.""" + context_filter = json.dumps( + { + "types": [EventTypes.Message], + "org.matrix.labels": ["#fun"], + } + ) + + res = self._test_context_filter_labels(context_filter) + + self.assertEqual( + res["event"]["content"]["body"], "with right label", res["event"] + ) + + events_before = res["events_before"] + + self.assertEqual( + len(events_before), 1, [event["content"] for event in events_before] + ) + self.assertEqual( + events_before[0]["content"]["body"], "with right label", events_before[0] + ) + + events_after = res["events_before"] + + self.assertEqual( + len(events_after), 1, [event["content"] for event in events_after] + ) + self.assertEqual( + events_after[0]["content"]["body"], "with right label", events_after[0] + ) + + def test_context_filter_not_labels(self): + """Test that we can filter by the absence of a label.""" + context_filter = json.dumps( + { + "types": [EventTypes.Message], + "org.matrix.not_labels": ["#fun"], + } + ) + + res = self._test_context_filter_labels(context_filter) + + events_before = res["events_before"] + + self.assertEqual( + len(events_before), 1, [event["content"] for event in events_before] + ) + self.assertEqual( + events_before[0]["content"]["body"], "without label", events_before[0] + ) + + events_after = res["events_after"] + + self.assertEqual( + len(events_after), 2, [event["content"] for event in events_after] + ) + self.assertEqual( + events_after[0]["content"]["body"], "with wrong label", events_after[0] + ) + self.assertEqual( + events_after[1]["content"]["body"], "with two wrong labels", events_after[1] + ) + + def test_context_filter_labels_not_labels(self): + """Test that we can filter by both a label and the absence of another label.""" + context_filter = json.dumps( + { + "types": [EventTypes.Message], + "org.matrix.labels": ["#work"], + "org.matrix.not_labels": ["#notfun"], + } + ) + + res = self._test_context_filter_labels(context_filter) + + events_before = res["events_before"] + + self.assertEqual( + len(events_before), 0, [event["content"] for event in events_before] + ) + + events_after = res["events_after"] + + self.assertEqual( + len(events_after), 1, [event["content"] for event in events_after] + ) + self.assertEqual( + events_after[0]["content"]["body"], "with wrong label", events_after[0] + ) + + def _test_context_filter_labels(self, context_filter): + user_id = self.register_user("kermit", "test") + tok = self.login("kermit", "test") + + room_id = self.helper.create_room_as(user_id, tok=tok) + + self.helper.send_event( + room_id=room_id, + type=EventTypes.Message, + content={ + "msgtype": "m.text", + "body": "with right label", + EventContentFields.LABELS: ["#fun"], + }, + tok=tok, + ) + + self.helper.send_event( + room_id=room_id, + type=EventTypes.Message, + content={"msgtype": "m.text", "body": "without label"}, + tok=tok, + ) + + # The event we'll look up the context for. + res = self.helper.send_event( + room_id=room_id, + type=EventTypes.Message, + content={ + "msgtype": "m.text", + "body": "with right label", + EventContentFields.LABELS: ["#fun"], + }, + tok=tok, + ) + event_id = res["event_id"] + + self.helper.send_event( + room_id=room_id, + type=EventTypes.Message, + content={ + "msgtype": "m.text", + "body": "with wrong label", + EventContentFields.LABELS: ["#work"], + }, + tok=tok, + ) + + self.helper.send_event( + room_id=room_id, + type=EventTypes.Message, + content={ + "msgtype": "m.text", + "body": "with two wrong labels", + EventContentFields.LABELS: ["#work", "#notfun"], + }, + tok=tok, + ) + + self.helper.send_event( + room_id=room_id, + type=EventTypes.Message, + content={ + "msgtype": "m.text", + "body": "with right label", + EventContentFields.LABELS: ["#fun"], + }, + tok=tok, + ) + + request, channel = self.make_request( + "GET", + "/rooms/%s/context/%s?filter=%s" % (room_id, event_id, context_filter), + access_token=tok, + ) + self.render(request) + self.assertEqual(channel.code, 200, channel.result) + + return channel.json_body + -- cgit 1.4.1 From c9e4748cb75271a2178d0cae05d551829249ada3 Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Tue, 5 Nov 2019 13:47:47 +0000 Subject: Merge labels tests for /context and /messages --- tests/rest/client/v1/test_rooms.py | 276 +++++++++++++++++-------------------- 1 file changed, 130 insertions(+), 146 deletions(-) (limited to 'tests') diff --git a/tests/rest/client/v1/test_rooms.py b/tests/rest/client/v1/test_rooms.py index 621c894e35..fe327d1bf8 100644 --- a/tests/rest/client/v1/test_rooms.py +++ b/tests/rest/client/v1/test_rooms.py @@ -811,105 +811,6 @@ class RoomMessageListTestCase(RoomBase): self.assertTrue("chunk" in channel.json_body) self.assertTrue("end" in channel.json_body) - def test_filter_labels(self): - """Test that we can filter by a label.""" - message_filter = json.dumps( - {"types": [EventTypes.Message], "org.matrix.labels": ["#fun"]} - ) - - events = self._test_filter_labels(message_filter) - - self.assertEqual(len(events), 2, [event["content"] for event in events]) - self.assertEqual(events[0]["content"]["body"], "with right label", events[0]) - self.assertEqual(events[1]["content"]["body"], "with right label", events[1]) - - def test_filter_not_labels(self): - """Test that we can filter by the absence of a label.""" - message_filter = json.dumps( - {"types": [EventTypes.Message], "org.matrix.not_labels": ["#fun"]} - ) - - events = self._test_filter_labels(message_filter) - - self.assertEqual(len(events), 3, [event["content"] for event in events]) - self.assertEqual(events[0]["content"]["body"], "without label", events[0]) - self.assertEqual(events[1]["content"]["body"], "with wrong label", events[1]) - self.assertEqual( - events[2]["content"]["body"], "with two wrong labels", events[2] - ) - - def test_filter_labels_not_labels(self): - """Test that we can filter by both a label and the absence of another label.""" - sync_filter = json.dumps( - { - "types": [EventTypes.Message], - "org.matrix.labels": ["#work"], - "org.matrix.not_labels": ["#notfun"], - } - ) - - events = self._test_filter_labels(sync_filter) - - self.assertEqual(len(events), 1, [event["content"] for event in events]) - self.assertEqual(events[0]["content"]["body"], "with wrong label", events[0]) - - def _test_filter_labels(self, message_filter): - self.helper.send_event( - room_id=self.room_id, - type=EventTypes.Message, - content={ - "msgtype": "m.text", - "body": "with right label", - EventContentFields.LABELS: ["#fun"], - }, - ) - - self.helper.send_event( - room_id=self.room_id, - type=EventTypes.Message, - content={"msgtype": "m.text", "body": "without label"}, - ) - - self.helper.send_event( - room_id=self.room_id, - type=EventTypes.Message, - content={ - "msgtype": "m.text", - "body": "with wrong label", - EventContentFields.LABELS: ["#work"], - }, - ) - - self.helper.send_event( - room_id=self.room_id, - type=EventTypes.Message, - content={ - "msgtype": "m.text", - "body": "with two wrong labels", - EventContentFields.LABELS: ["#work", "#notfun"], - }, - ) - - self.helper.send_event( - room_id=self.room_id, - type=EventTypes.Message, - content={ - "msgtype": "m.text", - "body": "with right label", - EventContentFields.LABELS: ["#fun"], - }, - ) - - token = "s0_0_0_0_0_0_0_0_0" - request, channel = self.make_request( - "GET", - "/rooms/%s/messages?access_token=x&from=%s&filter=%s" - % (self.room_id, token, message_filter), - ) - self.render(request) - - return channel.json_body["chunk"] - class RoomSearchTestCase(unittest.HomeserverTestCase): servlets = [ @@ -1108,7 +1009,7 @@ class PerRoomProfilesForbiddenTestCase(unittest.HomeserverTestCase): self.assertEqual(res_displayname, self.displayname, channel.result) -class ContextTestCase(unittest.HomeserverTestCase): +class LabelsTestCase(unittest.HomeserverTestCase): servlets = [ synapse.rest.admin.register_servlets_for_client_rest_resource, room.register_servlets, @@ -1116,8 +1017,13 @@ class ContextTestCase(unittest.HomeserverTestCase): profile.register_servlets, ] + def prepare(self, reactor, clock, homeserver): + self.user_id = self.register_user("test", "test") + self.tok = self.login("test", "test") + self.room_id = self.helper.create_room_as(self.user_id, tok=self.tok) + def test_context_filter_labels(self): - """Test that we can filter by a label.""" + """Test that we can filter by a label on a /context request.""" context_filter = json.dumps( { "types": [EventTypes.Message], @@ -1125,13 +1031,17 @@ class ContextTestCase(unittest.HomeserverTestCase): } ) - res = self._test_context_filter_labels(context_filter) + event_id = self._send_labelled_messages_in_room() - self.assertEqual( - res["event"]["content"]["body"], "with right label", res["event"] + request, channel = self.make_request( + "GET", + "/rooms/%s/context/%s?filter=%s" % (self.room_id, event_id, context_filter), + access_token=self.tok, ) + self.render(request) + self.assertEqual(channel.code, 200, channel.result) - events_before = res["events_before"] + events_before = channel.json_body["events_before"] self.assertEqual( len(events_before), 1, [event["content"] for event in events_before] @@ -1140,7 +1050,7 @@ class ContextTestCase(unittest.HomeserverTestCase): events_before[0]["content"]["body"], "with right label", events_before[0] ) - events_after = res["events_before"] + events_after = channel.json_body["events_before"] self.assertEqual( len(events_after), 1, [event["content"] for event in events_after] @@ -1150,7 +1060,7 @@ class ContextTestCase(unittest.HomeserverTestCase): ) def test_context_filter_not_labels(self): - """Test that we can filter by the absence of a label.""" + """Test that we can filter by the absence of a label on a /context request.""" context_filter = json.dumps( { "types": [EventTypes.Message], @@ -1158,9 +1068,17 @@ class ContextTestCase(unittest.HomeserverTestCase): } ) - res = self._test_context_filter_labels(context_filter) + event_id = self._send_labelled_messages_in_room() + + request, channel = self.make_request( + "GET", + "/rooms/%s/context/%s?filter=%s" % (self.room_id, event_id, context_filter), + access_token=self.tok, + ) + self.render(request) + self.assertEqual(channel.code, 200, channel.result) - events_before = res["events_before"] + events_before = channel.json_body["events_before"] self.assertEqual( len(events_before), 1, [event["content"] for event in events_before] @@ -1169,7 +1087,7 @@ class ContextTestCase(unittest.HomeserverTestCase): events_before[0]["content"]["body"], "without label", events_before[0] ) - events_after = res["events_after"] + events_after = channel.json_body["events_after"] self.assertEqual( len(events_after), 2, [event["content"] for event in events_after] @@ -1182,7 +1100,9 @@ class ContextTestCase(unittest.HomeserverTestCase): ) def test_context_filter_labels_not_labels(self): - """Test that we can filter by both a label and the absence of another label.""" + """Test that we can filter by both a label and the absence of another label on a + /context request. + """ context_filter = json.dumps( { "types": [EventTypes.Message], @@ -1191,15 +1111,23 @@ class ContextTestCase(unittest.HomeserverTestCase): } ) - res = self._test_context_filter_labels(context_filter) + event_id = self._send_labelled_messages_in_room() - events_before = res["events_before"] + request, channel = self.make_request( + "GET", + "/rooms/%s/context/%s?filter=%s" % (self.room_id, event_id, context_filter), + access_token=self.tok, + ) + self.render(request) + self.assertEqual(channel.code, 200, channel.result) + + events_before = channel.json_body["events_before"] self.assertEqual( len(events_before), 0, [event["content"] for event in events_before] ) - events_after = res["events_after"] + events_after = channel.json_body["events_after"] self.assertEqual( len(events_after), 1, [event["content"] for event in events_after] @@ -1208,83 +1136,139 @@ class ContextTestCase(unittest.HomeserverTestCase): events_after[0]["content"]["body"], "with wrong label", events_after[0] ) - def _test_context_filter_labels(self, context_filter): - user_id = self.register_user("kermit", "test") - tok = self.login("kermit", "test") + def test_messages_filter_labels(self): + """Test that we can filter by a label on a /messages request.""" + message_filter = json.dumps( + {"types": [EventTypes.Message], "org.matrix.labels": ["#fun"]} + ) + + self._send_labelled_messages_in_room() + + token = "s0_0_0_0_0_0_0_0_0" + request, channel = self.make_request( + "GET", + "/rooms/%s/messages?access_token=%s&from=%s&filter=%s" + % (self.room_id, self.tok, token, message_filter), + ) + self.render(request) + + events = channel.json_body["chunk"] + + self.assertEqual(len(events), 2, [event["content"] for event in events]) + self.assertEqual(events[0]["content"]["body"], "with right label", events[0]) + self.assertEqual(events[1]["content"]["body"], "with right label", events[1]) + + def test_messages_filter_not_labels(self): + """Test that we can filter by the absence of a label on a /messages request.""" + message_filter = json.dumps( + {"types": [EventTypes.Message], "org.matrix.not_labels": ["#fun"]} + ) + + self._send_labelled_messages_in_room() + + token = "s0_0_0_0_0_0_0_0_0" + request, channel = self.make_request( + "GET", + "/rooms/%s/messages?access_token=%s&from=%s&filter=%s" + % (self.room_id, self.tok, token, message_filter), + ) + self.render(request) + + events = channel.json_body["chunk"] + + self.assertEqual(len(events), 4, [event["content"] for event in events]) + self.assertEqual(events[0]["content"]["body"], "without label", events[0]) + self.assertEqual(events[1]["content"]["body"], "without label", events[1]) + self.assertEqual(events[2]["content"]["body"], "with wrong label", events[2]) + self.assertEqual( + events[3]["content"]["body"], "with two wrong labels", events[3] + ) + + def test_messages_filter_labels_not_labels(self): + """Test that we can filter by both a label and the absence of another label on a + /messages request. + """ + message_filter = json.dumps( + { + "types": [EventTypes.Message], + "org.matrix.labels": ["#work"], + "org.matrix.not_labels": ["#notfun"], + } + ) + + self._send_labelled_messages_in_room() + + token = "s0_0_0_0_0_0_0_0_0" + request, channel = self.make_request( + "GET", + "/rooms/%s/messages?access_token=%s&from=%s&filter=%s" + % (self.room_id, self.tok, token, message_filter), + ) + self.render(request) + + events = channel.json_body["chunk"] - room_id = self.helper.create_room_as(user_id, tok=tok) + self.assertEqual(len(events), 1, [event["content"] for event in events]) + self.assertEqual(events[0]["content"]["body"], "with wrong label", events[0]) + def _send_labelled_messages_in_room(self): self.helper.send_event( - room_id=room_id, + room_id=self.room_id, type=EventTypes.Message, content={ "msgtype": "m.text", "body": "with right label", EventContentFields.LABELS: ["#fun"], }, - tok=tok, + tok=self.tok, ) self.helper.send_event( - room_id=room_id, + room_id=self.room_id, type=EventTypes.Message, content={"msgtype": "m.text", "body": "without label"}, - tok=tok, + tok=self.tok, ) - # The event we'll look up the context for. res = self.helper.send_event( - room_id=room_id, + room_id=self.room_id, type=EventTypes.Message, - content={ - "msgtype": "m.text", - "body": "with right label", - EventContentFields.LABELS: ["#fun"], - }, - tok=tok, + content={"msgtype": "m.text", "body": "without label"}, + tok=self.tok, ) event_id = res["event_id"] self.helper.send_event( - room_id=room_id, + room_id=self.room_id, type=EventTypes.Message, content={ "msgtype": "m.text", "body": "with wrong label", EventContentFields.LABELS: ["#work"], }, - tok=tok, + tok=self.tok, ) self.helper.send_event( - room_id=room_id, + room_id=self.room_id, type=EventTypes.Message, content={ "msgtype": "m.text", "body": "with two wrong labels", EventContentFields.LABELS: ["#work", "#notfun"], }, - tok=tok, + tok=self.tok, ) self.helper.send_event( - room_id=room_id, + room_id=self.room_id, type=EventTypes.Message, content={ "msgtype": "m.text", "body": "with right label", EventContentFields.LABELS: ["#fun"], }, - tok=tok, + tok=self.tok, ) - request, channel = self.make_request( - "GET", - "/rooms/%s/context/%s?filter=%s" % (room_id, event_id, context_filter), - access_token=tok, - ) - self.render(request) - self.assertEqual(channel.code, 200, channel.result) - - return channel.json_body - + return event_id -- cgit 1.4.1 From 037360e6cf2ca181b7cf03884375d4a4d52ad64e Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Tue, 5 Nov 2019 14:33:18 +0000 Subject: Add tests for /search --- tests/rest/client/v1/test_rooms.py | 187 ++++++++++++++++++++++++++++--------- 1 file changed, 143 insertions(+), 44 deletions(-) (limited to 'tests') diff --git a/tests/rest/client/v1/test_rooms.py b/tests/rest/client/v1/test_rooms.py index fe327d1bf8..cc7499dcc0 100644 --- a/tests/rest/client/v1/test_rooms.py +++ b/tests/rest/client/v1/test_rooms.py @@ -1017,6 +1017,18 @@ class LabelsTestCase(unittest.HomeserverTestCase): profile.register_servlets, ] + # Filter that should only catch messages with the label "#fun". + FILTER_LABELS = {"types": [EventTypes.Message], "org.matrix.labels": ["#fun"]} + # Filter that should only catch messages without the label "#fun". + FILTER_NOT_LABELS = {"types": [EventTypes.Message], "org.matrix.not_labels": ["#fun"]} + # Filter that should only catch messages with the label "#work" but without the label + # "#notfun". + FILTER_LABELS_NOT_LABELS = { + "types": [EventTypes.Message], + "org.matrix.labels": ["#work"], + "org.matrix.not_labels": ["#notfun"], + } + def prepare(self, reactor, clock, homeserver): self.user_id = self.register_user("test", "test") self.tok = self.login("test", "test") @@ -1024,18 +1036,12 @@ class LabelsTestCase(unittest.HomeserverTestCase): def test_context_filter_labels(self): """Test that we can filter by a label on a /context request.""" - context_filter = json.dumps( - { - "types": [EventTypes.Message], - "org.matrix.labels": ["#fun"], - } - ) - event_id = self._send_labelled_messages_in_room() request, channel = self.make_request( "GET", - "/rooms/%s/context/%s?filter=%s" % (self.room_id, event_id, context_filter), + "/rooms/%s/context/%s?filter=%s" + % (self.room_id, event_id, json.dumps(self.FILTER_LABELS)), access_token=self.tok, ) self.render(request) @@ -1061,18 +1067,12 @@ class LabelsTestCase(unittest.HomeserverTestCase): def test_context_filter_not_labels(self): """Test that we can filter by the absence of a label on a /context request.""" - context_filter = json.dumps( - { - "types": [EventTypes.Message], - "org.matrix.not_labels": ["#fun"], - } - ) - event_id = self._send_labelled_messages_in_room() request, channel = self.make_request( "GET", - "/rooms/%s/context/%s?filter=%s" % (self.room_id, event_id, context_filter), + "/rooms/%s/context/%s?filter=%s" + % (self.room_id, event_id, json.dumps(self.FILTER_NOT_LABELS)), access_token=self.tok, ) self.render(request) @@ -1103,19 +1103,12 @@ class LabelsTestCase(unittest.HomeserverTestCase): """Test that we can filter by both a label and the absence of another label on a /context request. """ - context_filter = json.dumps( - { - "types": [EventTypes.Message], - "org.matrix.labels": ["#work"], - "org.matrix.not_labels": ["#notfun"], - } - ) - event_id = self._send_labelled_messages_in_room() request, channel = self.make_request( "GET", - "/rooms/%s/context/%s?filter=%s" % (self.room_id, event_id, context_filter), + "/rooms/%s/context/%s?filter=%s" + % (self.room_id, event_id, json.dumps(self.FILTER_LABELS_NOT_LABELS)), access_token=self.tok, ) self.render(request) @@ -1138,17 +1131,13 @@ class LabelsTestCase(unittest.HomeserverTestCase): def test_messages_filter_labels(self): """Test that we can filter by a label on a /messages request.""" - message_filter = json.dumps( - {"types": [EventTypes.Message], "org.matrix.labels": ["#fun"]} - ) - self._send_labelled_messages_in_room() token = "s0_0_0_0_0_0_0_0_0" request, channel = self.make_request( "GET", "/rooms/%s/messages?access_token=%s&from=%s&filter=%s" - % (self.room_id, self.tok, token, message_filter), + % (self.room_id, self.tok, token, json.dumps(self.FILTER_LABELS)), ) self.render(request) @@ -1160,17 +1149,13 @@ class LabelsTestCase(unittest.HomeserverTestCase): def test_messages_filter_not_labels(self): """Test that we can filter by the absence of a label on a /messages request.""" - message_filter = json.dumps( - {"types": [EventTypes.Message], "org.matrix.not_labels": ["#fun"]} - ) - self._send_labelled_messages_in_room() token = "s0_0_0_0_0_0_0_0_0" request, channel = self.make_request( "GET", "/rooms/%s/messages?access_token=%s&from=%s&filter=%s" - % (self.room_id, self.tok, token, message_filter), + % (self.room_id, self.tok, token, json.dumps(self.FILTER_NOT_LABELS)), ) self.render(request) @@ -1188,21 +1173,13 @@ class LabelsTestCase(unittest.HomeserverTestCase): """Test that we can filter by both a label and the absence of another label on a /messages request. """ - message_filter = json.dumps( - { - "types": [EventTypes.Message], - "org.matrix.labels": ["#work"], - "org.matrix.not_labels": ["#notfun"], - } - ) - self._send_labelled_messages_in_room() token = "s0_0_0_0_0_0_0_0_0" request, channel = self.make_request( "GET", "/rooms/%s/messages?access_token=%s&from=%s&filter=%s" - % (self.room_id, self.tok, token, message_filter), + % (self.room_id, self.tok, token, json.dumps(self.FILTER_LABELS_NOT_LABELS)), ) self.render(request) @@ -1211,7 +1188,128 @@ class LabelsTestCase(unittest.HomeserverTestCase): self.assertEqual(len(events), 1, [event["content"] for event in events]) self.assertEqual(events[0]["content"]["body"], "with wrong label", events[0]) + def test_search_filter_labels(self): + """Test that we can filter by a label on a /search request.""" + request_data = json.dumps({ + "search_categories": { + "room_events": { + "search_term": "label", + "filter": self.FILTER_LABELS, + } + } + }) + + self._send_labelled_messages_in_room() + + request, channel = self.make_request( + "POST", "/search?access_token=%s" % self.tok, request_data + ) + self.render(request) + + results = channel.json_body["search_categories"]["room_events"]["results"] + + self.assertEqual( + len(results), + 2, + [result["result"]["content"] for result in results], + ) + self.assertEqual( + results[0]["result"]["content"]["body"], + "with right label", + results[0]["result"]["content"]["body"], + ) + self.assertEqual( + results[1]["result"]["content"]["body"], + "with right label", + results[1]["result"]["content"]["body"], + ) + + def test_search_filter_not_labels(self): + """Test that we can filter by the absence of a label on a /search request.""" + request_data = json.dumps({ + "search_categories": { + "room_events": { + "search_term": "label", + "filter": self.FILTER_NOT_LABELS, + } + } + }) + + self._send_labelled_messages_in_room() + + request, channel = self.make_request( + "POST", "/search?access_token=%s" % self.tok, request_data + ) + self.render(request) + + results = channel.json_body["search_categories"]["room_events"]["results"] + + self.assertEqual( + len(results), + 4, + [result["result"]["content"] for result in results], + ) + self.assertEqual( + results[0]["result"]["content"]["body"], + "without label", + results[0]["result"]["content"]["body"], + ) + self.assertEqual( + results[1]["result"]["content"]["body"], + "without label", + results[1]["result"]["content"]["body"], + ) + self.assertEqual( + results[2]["result"]["content"]["body"], + "with wrong label", + results[2]["result"]["content"]["body"], + ) + self.assertEqual( + results[3]["result"]["content"]["body"], + "with two wrong labels", + results[3]["result"]["content"]["body"], + ) + + def test_search_filter_labels_not_labels(self): + """Test that we can filter by both a label and the absence of another label on a + /search request. + """ + request_data = json.dumps({ + "search_categories": { + "room_events": { + "search_term": "label", + "filter": self.FILTER_LABELS_NOT_LABELS, + } + } + }) + + self._send_labelled_messages_in_room() + + request, channel = self.make_request( + "POST", "/search?access_token=%s" % self.tok, request_data + ) + self.render(request) + + results = channel.json_body["search_categories"]["room_events"]["results"] + + self.assertEqual( + len(results), + 1, + [result["result"]["content"] for result in results], + ) + self.assertEqual( + results[0]["result"]["content"]["body"], + "with wrong label", + results[0]["result"]["content"]["body"], + ) + def _send_labelled_messages_in_room(self): + """Sends several messages to a room with different labels (or without any) to test + filtering by label. + + Returns: + The ID of the event to use if we're testing filtering on /context. + """ self.helper.send_event( room_id=self.room_id, type=EventTypes.Message, @@ -1236,6 +1334,7 @@ class LabelsTestCase(unittest.HomeserverTestCase): content={"msgtype": "m.text", "body": "without label"}, tok=self.tok, ) + # Return this event's ID when we test filtering in /context requests. event_id = res["event_id"] self.helper.send_event( -- cgit 1.4.1 From 8822b331114a2f6fdcd5916f0c91991c0acae07e Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Tue, 5 Nov 2019 10:56:39 +0000 Subject: Update copyrights --- synapse/api/constants.py | 3 ++- synapse/api/filtering.py | 3 +++ synapse/rest/client/versions.py | 3 +++ synapse/storage/data_stores/main/stream.py | 3 +++ tests/api/test_filtering.py | 3 +++ tests/rest/client/v1/test_rooms.py | 2 ++ tests/rest/client/v1/utils.py | 3 +++ tests/rest/client/v2_alpha/test_sync.py | 3 ++- 8 files changed, 21 insertions(+), 2 deletions(-) (limited to 'tests') diff --git a/synapse/api/constants.py b/synapse/api/constants.py index 49c4b85054..312acff3d6 100644 --- a/synapse/api/constants.py +++ b/synapse/api/constants.py @@ -1,7 +1,8 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd # Copyright 2017 Vector Creations Ltd -# Copyright 2018 New Vector Ltd +# Copyright 2018-2019 New Vector Ltd +# Copyright 2019 The Matrix.org Foundation C.I.C. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/synapse/api/filtering.py b/synapse/api/filtering.py index bec13f08d8..6eab1f13f0 100644 --- a/synapse/api/filtering.py +++ b/synapse/api/filtering.py @@ -1,5 +1,8 @@ # -*- coding: utf-8 -*- # Copyright 2015, 2016 OpenMarket Ltd +# Copyright 2017 Vector Creations Ltd +# Copyright 2018-2019 New Vector Ltd +# Copyright 2019 The Matrix.org Foundation C.I.C. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/synapse/rest/client/versions.py b/synapse/rest/client/versions.py index bb30ce3f34..2a477ad22e 100644 --- a/synapse/rest/client/versions.py +++ b/synapse/rest/client/versions.py @@ -1,5 +1,8 @@ # -*- coding: utf-8 -*- # Copyright 2016 OpenMarket Ltd +# Copyright 2017 Vector Creations Ltd +# Copyright 2018-2019 New Vector Ltd +# Copyright 2019 The Matrix.org Foundation C.I.C. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/synapse/storage/data_stores/main/stream.py b/synapse/storage/data_stores/main/stream.py index 616ef91d4e..9cac664880 100644 --- a/synapse/storage/data_stores/main/stream.py +++ b/synapse/storage/data_stores/main/stream.py @@ -1,5 +1,8 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2017 Vector Creations Ltd +# Copyright 2018-2019 New Vector Ltd +# Copyright 2019 The Matrix.org Foundation C.I.C. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/tests/api/test_filtering.py b/tests/api/test_filtering.py index 2dc5052249..63d8633582 100644 --- a/tests/api/test_filtering.py +++ b/tests/api/test_filtering.py @@ -1,5 +1,8 @@ # -*- coding: utf-8 -*- # Copyright 2015, 2016 OpenMarket Ltd +# Copyright 2017 Vector Creations Ltd +# Copyright 2018-2019 New Vector Ltd +# Copyright 2019 The Matrix.org Foundation C.I.C. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/tests/rest/client/v1/test_rooms.py b/tests/rest/client/v1/test_rooms.py index cc7499dcc0..b2c1ef6f0e 100644 --- a/tests/rest/client/v1/test_rooms.py +++ b/tests/rest/client/v1/test_rooms.py @@ -1,5 +1,7 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2017 Vector Creations Ltd +# Copyright 2018-2019 New Vector Ltd # Copyright 2019 The Matrix.org Foundation C.I.C. # # Licensed under the Apache License, Version 2.0 (the "License"); diff --git a/tests/rest/client/v1/utils.py b/tests/rest/client/v1/utils.py index 8ea0cb05ea..e7417b3d14 100644 --- a/tests/rest/client/v1/utils.py +++ b/tests/rest/client/v1/utils.py @@ -1,5 +1,8 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2017 Vector Creations Ltd +# Copyright 2018-2019 New Vector Ltd +# Copyright 2019 The Matrix.org Foundation C.I.C. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. diff --git a/tests/rest/client/v2_alpha/test_sync.py b/tests/rest/client/v2_alpha/test_sync.py index 3283c0e47b..661c1f88b9 100644 --- a/tests/rest/client/v2_alpha/test_sync.py +++ b/tests/rest/client/v2_alpha/test_sync.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- -# Copyright 2018 New Vector +# Copyright 2018-2019 New Vector Ltd +# Copyright 2019 The Matrix.org Foundation C.I.C. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. -- cgit 1.4.1 From a6863da24934dcbb2ae09a9e0b6e37140ef390ff Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Tue, 5 Nov 2019 14:50:19 +0000 Subject: Lint --- tests/rest/client/v1/test_rooms.py | 71 ++++++++++++++++++++++---------------- 1 file changed, 41 insertions(+), 30 deletions(-) (limited to 'tests') diff --git a/tests/rest/client/v1/test_rooms.py b/tests/rest/client/v1/test_rooms.py index b2c1ef6f0e..c5d67fc1cd 100644 --- a/tests/rest/client/v1/test_rooms.py +++ b/tests/rest/client/v1/test_rooms.py @@ -1020,9 +1020,15 @@ class LabelsTestCase(unittest.HomeserverTestCase): ] # Filter that should only catch messages with the label "#fun". - FILTER_LABELS = {"types": [EventTypes.Message], "org.matrix.labels": ["#fun"]} + FILTER_LABELS = { + "types": [EventTypes.Message], + "org.matrix.labels": ["#fun"], + } # Filter that should only catch messages without the label "#fun". - FILTER_NOT_LABELS = {"types": [EventTypes.Message], "org.matrix.not_labels": ["#fun"]} + FILTER_NOT_LABELS = { + "types": [EventTypes.Message], + "org.matrix.not_labels": ["#fun"], + } # Filter that should only catch messages with the label "#work" but without the label # "#notfun". FILTER_LABELS_NOT_LABELS = { @@ -1181,7 +1187,12 @@ class LabelsTestCase(unittest.HomeserverTestCase): request, channel = self.make_request( "GET", "/rooms/%s/messages?access_token=%s&from=%s&filter=%s" - % (self.room_id, self.tok, token, json.dumps(self.FILTER_LABELS_NOT_LABELS)), + % ( + self.room_id, + self.tok, + token, + json.dumps(self.FILTER_LABELS_NOT_LABELS), + ), ) self.render(request) @@ -1192,14 +1203,16 @@ class LabelsTestCase(unittest.HomeserverTestCase): def test_search_filter_labels(self): """Test that we can filter by a label on a /search request.""" - request_data = json.dumps({ - "search_categories": { - "room_events": { - "search_term": "label", - "filter": self.FILTER_LABELS, + request_data = json.dumps( + { + "search_categories": { + "room_events": { + "search_term": "label", + "filter": self.FILTER_LABELS, + } } } - }) + ) self._send_labelled_messages_in_room() @@ -1211,9 +1224,7 @@ class LabelsTestCase(unittest.HomeserverTestCase): results = channel.json_body["search_categories"]["room_events"]["results"] self.assertEqual( - len(results), - 2, - [result["result"]["content"] for result in results], + len(results), 2, [result["result"]["content"] for result in results], ) self.assertEqual( results[0]["result"]["content"]["body"], @@ -1228,14 +1239,16 @@ class LabelsTestCase(unittest.HomeserverTestCase): def test_search_filter_not_labels(self): """Test that we can filter by the absence of a label on a /search request.""" - request_data = json.dumps({ - "search_categories": { - "room_events": { - "search_term": "label", - "filter": self.FILTER_NOT_LABELS, + request_data = json.dumps( + { + "search_categories": { + "room_events": { + "search_term": "label", + "filter": self.FILTER_NOT_LABELS, + } } } - }) + ) self._send_labelled_messages_in_room() @@ -1247,9 +1260,7 @@ class LabelsTestCase(unittest.HomeserverTestCase): results = channel.json_body["search_categories"]["room_events"]["results"] self.assertEqual( - len(results), - 4, - [result["result"]["content"] for result in results], + len(results), 4, [result["result"]["content"] for result in results], ) self.assertEqual( results[0]["result"]["content"]["body"], @@ -1276,14 +1287,16 @@ class LabelsTestCase(unittest.HomeserverTestCase): """Test that we can filter by both a label and the absence of another label on a /search request. """ - request_data = json.dumps({ - "search_categories": { - "room_events": { - "search_term": "label", - "filter": self.FILTER_LABELS_NOT_LABELS, + request_data = json.dumps( + { + "search_categories": { + "room_events": { + "search_term": "label", + "filter": self.FILTER_LABELS_NOT_LABELS, + } } } - }) + ) self._send_labelled_messages_in_room() @@ -1295,9 +1308,7 @@ class LabelsTestCase(unittest.HomeserverTestCase): results = channel.json_body["search_categories"]["room_events"]["results"] self.assertEqual( - len(results), - 1, - [result["result"]["content"] for result in results], + len(results), 1, [result["result"]["content"] for result in results], ) self.assertEqual( results[0]["result"]["content"]["body"], -- cgit 1.4.1 From e9bfe719ba1928dc191cea93120c5c8a89584434 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Tue, 5 Nov 2019 15:45:17 +0000 Subject: Strip overlong OpenGraph data from url preview ... to stop people causing DoSes with malicious web pages --- changelog.d/6331.feature | 1 + synapse/rest/media/v1/preview_url_resource.py | 20 +++++++++++++++- tests/rest/media/v1/test_url_preview.py | 34 +++++++++++++++++++++++++++ 3 files changed, 54 insertions(+), 1 deletion(-) create mode 100644 changelog.d/6331.feature (limited to 'tests') diff --git a/changelog.d/6331.feature b/changelog.d/6331.feature new file mode 100644 index 0000000000..eaf69ef3f6 --- /dev/null +++ b/changelog.d/6331.feature @@ -0,0 +1 @@ +Limit the length of data returned by url previews, to prevent DoS attacks. diff --git a/synapse/rest/media/v1/preview_url_resource.py b/synapse/rest/media/v1/preview_url_resource.py index 0c68c3aad5..6d8c39a410 100644 --- a/synapse/rest/media/v1/preview_url_resource.py +++ b/synapse/rest/media/v1/preview_url_resource.py @@ -56,6 +56,9 @@ logger = logging.getLogger(__name__) _charset_match = re.compile(br"<\s*meta[^>]*charset\s*=\s*([a-z0-9-]+)", flags=re.I) _content_type_match = re.compile(r'.*; *charset="?(.*?)"?(;|$)', flags=re.I) +OG_TAG_NAME_MAXLEN = 50 +OG_TAG_VALUE_MAXLEN = 1000 + class PreviewUrlResource(DirectServeResource): isLeaf = True @@ -167,7 +170,7 @@ class PreviewUrlResource(DirectServeResource): ts (int): Returns: - Deferred[str]: json-encoded og data + Deferred[bytes]: json-encoded og data """ # check the URL cache in the DB (which will also provide us with # historical previews, if we have any) @@ -268,6 +271,17 @@ class PreviewUrlResource(DirectServeResource): logger.warn("Failed to find any OG data in %s", url) og = {} + # filter out any stupidly long values + keys_to_remove = [] + for k, v in og.items(): + if len(k) > OG_TAG_NAME_MAXLEN or len(v) > OG_TAG_VALUE_MAXLEN: + logger.warning( + "Pruning overlong tag %s from OG data", k[:OG_TAG_NAME_MAXLEN] + ) + keys_to_remove.append(k) + for k in keys_to_remove: + del og[k] + logger.debug("Calculated OG for %s as %s" % (url, og)) jsonog = json.dumps(og) @@ -502,6 +516,10 @@ def _calc_og(tree, media_uri): og = {} for tag in tree.xpath("//*/meta[starts-with(@property, 'og:')]"): if "content" in tag.attrib: + # if we've got more than 50 tags, someone is taking the piss + if len(og) >= 50: + logger.warning("skipping OG for page with too many og: tags") + return {} og[tag.attrib["property"]] = tag.attrib["content"] # TODO: grab article: meta tags too, e.g.: diff --git a/tests/rest/media/v1/test_url_preview.py b/tests/rest/media/v1/test_url_preview.py index 976652aee8..da19a8e86f 100644 --- a/tests/rest/media/v1/test_url_preview.py +++ b/tests/rest/media/v1/test_url_preview.py @@ -247,6 +247,40 @@ class URLPreviewTests(unittest.HomeserverTestCase): self.assertEqual(channel.code, 200) self.assertEqual(channel.json_body["og:title"], "\u0434\u043a\u0430") + def test_overlong_title(self): + self.lookups["matrix.org"] = [(IPv4Address, "8.8.8.8")] + + end_content = ( + b"" + b"" + b"x" * 2000 + b"" + b'' + b"" + ) + + request, channel = self.make_request( + "GET", "url_preview?url=http://matrix.org", shorthand=False + ) + request.render(self.preview_url) + self.pump() + + client = self.reactor.tcpClients[0][2].buildProtocol(None) + server = AccumulatingProtocol() + server.makeConnection(FakeTransport(client, self.reactor)) + client.makeConnection(FakeTransport(server, self.reactor)) + client.dataReceived( + ( + b"HTTP/1.0 200 OK\r\nContent-Length: %d\r\n" + b'Content-Type: text/html; charset="windows-1251"\r\n\r\n' + ) + % (len(end_content),) + + end_content + ) + + self.pump() + self.assertEqual(channel.code, 200) + res = channel.json_body + self.assertCountEqual(["og:description"], res.keys()) + def test_ipaddr(self): """ IP addresses can be previewed directly. -- cgit 1.4.1 From e78167c94b3f63136f7d0e4f32a05ad1befdc0ec Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Tue, 5 Nov 2019 16:46:39 +0000 Subject: Apply suggestions from code review Co-Authored-By: Brendan Abolivier Co-Authored-By: Erik Johnston --- synapse/rest/media/v1/preview_url_resource.py | 2 +- tests/rest/media/v1/test_url_preview.py | 1 + 2 files changed, 2 insertions(+), 1 deletion(-) (limited to 'tests') diff --git a/synapse/rest/media/v1/preview_url_resource.py b/synapse/rest/media/v1/preview_url_resource.py index 6d8c39a410..4d4b3c1462 100644 --- a/synapse/rest/media/v1/preview_url_resource.py +++ b/synapse/rest/media/v1/preview_url_resource.py @@ -518,7 +518,7 @@ def _calc_og(tree, media_uri): if "content" in tag.attrib: # if we've got more than 50 tags, someone is taking the piss if len(og) >= 50: - logger.warning("skipping OG for page with too many og: tags") + logger.warning("Skipping OG for page with too many 'og:' tags") return {} og[tag.attrib["property"]] = tag.attrib["content"] diff --git a/tests/rest/media/v1/test_url_preview.py b/tests/rest/media/v1/test_url_preview.py index da19a8e86f..852b8ab11c 100644 --- a/tests/rest/media/v1/test_url_preview.py +++ b/tests/rest/media/v1/test_url_preview.py @@ -279,6 +279,7 @@ class URLPreviewTests(unittest.HomeserverTestCase): self.pump() self.assertEqual(channel.code, 200) res = channel.json_body + # We should only see the `og:description` field, as `title` is too long and should be stripped out self.assertCountEqual(["og:description"], res.keys()) def test_ipaddr(self): -- cgit 1.4.1 From 807ec3bd99908d2991d2b3d0615b0862610c6dc3 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Wed, 6 Nov 2019 10:01:39 +0000 Subject: Fix bug which caused rejected events to be stored with the wrong room state (#6320) Fixes a bug where rejected events were persisted with the wrong state group. Also fixes an occasional internal-server-error when receiving events over federation which are rejected and (possibly because they are backwards-extremities) have no prev_group. Fixes #6289. --- changelog.d/6320.bugfix | 1 + synapse/events/snapshot.py | 25 ++++- synapse/handlers/federation.py | 1 + synapse/state/__init__.py | 172 ++++++++++++++---------------- synapse/storage/data_stores/main/state.py | 2 +- tests/handlers/test_federation.py | 126 ++++++++++++++++++++++ tests/test_state.py | 61 +++++++++-- 7 files changed, 285 insertions(+), 103 deletions(-) create mode 100644 changelog.d/6320.bugfix (limited to 'tests') diff --git a/changelog.d/6320.bugfix b/changelog.d/6320.bugfix new file mode 100644 index 0000000000..2c3fad5655 --- /dev/null +++ b/changelog.d/6320.bugfix @@ -0,0 +1 @@ +Fix bug which casued rejected events to be persisted with the wrong room state. diff --git a/synapse/events/snapshot.py b/synapse/events/snapshot.py index 0f3c5989cb..64e898f40c 100644 --- a/synapse/events/snapshot.py +++ b/synapse/events/snapshot.py @@ -48,10 +48,21 @@ class EventContext: Note that this is a private attribute: it should be accessed via the ``state_group`` property. + state_group_before_event: The ID of the state group representing the state + of the room before this event. + + If this is a non-state event, this will be the same as ``state_group``. If + it's a state event, it will be the same as ``prev_group``. + + If ``state_group`` is None (ie, the event is an outlier), + ``state_group_before_event`` will always also be ``None``. + prev_group: If it is known, ``state_group``'s prev_group. Note that this being None does not necessarily mean that ``state_group`` does not have a prev_group! + If the event is a state event, this is normally the same as ``prev_group``. + If ``state_group`` is None (ie, the event is an outlier), ``prev_group`` will always also be ``None``. @@ -77,7 +88,8 @@ class EventContext: ``get_current_state_ids``. _AsyncEventContext impl calculates this on-demand: it will be None until that happens. - _prev_state_ids: The room state map, excluding this event. For a non-state + _prev_state_ids: The room state map, excluding this event - ie, the state + in ``state_group_before_event``. For a non-state event, this will be the same as _current_state_events. Note that it is a completely different thing to prev_group! @@ -92,6 +104,7 @@ class EventContext: rejected = attr.ib(default=False, type=Union[bool, str]) _state_group = attr.ib(default=None, type=Optional[int]) + state_group_before_event = attr.ib(default=None, type=Optional[int]) prev_group = attr.ib(default=None, type=Optional[int]) delta_ids = attr.ib(default=None, type=Optional[Dict[Tuple[str, str], str]]) app_service = attr.ib(default=None, type=Optional[ApplicationService]) @@ -103,12 +116,18 @@ class EventContext: @staticmethod def with_state( - state_group, current_state_ids, prev_state_ids, prev_group=None, delta_ids=None + state_group, + state_group_before_event, + current_state_ids, + prev_state_ids, + prev_group=None, + delta_ids=None, ): return EventContext( current_state_ids=current_state_ids, prev_state_ids=prev_state_ids, state_group=state_group, + state_group_before_event=state_group_before_event, prev_group=prev_group, delta_ids=delta_ids, ) @@ -140,6 +159,7 @@ class EventContext: "event_type": event.type, "event_state_key": event.state_key if event.is_state() else None, "state_group": self._state_group, + "state_group_before_event": self.state_group_before_event, "rejected": self.rejected, "prev_group": self.prev_group, "delta_ids": _encode_state_dict(self.delta_ids), @@ -165,6 +185,7 @@ class EventContext: event_type=input["event_type"], event_state_key=input["event_state_key"], state_group=input["state_group"], + state_group_before_event=input["state_group_before_event"], prev_group=input["prev_group"], delta_ids=_decode_state_dict(input["delta_ids"]), rejected=input["rejected"], diff --git a/synapse/handlers/federation.py b/synapse/handlers/federation.py index b7916de909..05dd8d2671 100644 --- a/synapse/handlers/federation.py +++ b/synapse/handlers/federation.py @@ -2280,6 +2280,7 @@ class FederationHandler(BaseHandler): return EventContext.with_state( state_group=state_group, + state_group_before_event=context.state_group_before_event, current_state_ids=current_state_ids, prev_state_ids=prev_state_ids, prev_group=prev_group, diff --git a/synapse/state/__init__.py b/synapse/state/__init__.py index 2c04ab1854..139beef8ed 100644 --- a/synapse/state/__init__.py +++ b/synapse/state/__init__.py @@ -16,6 +16,7 @@ import logging from collections import namedtuple +from typing import Iterable, Optional from six import iteritems, itervalues @@ -27,6 +28,7 @@ from twisted.internet import defer from synapse.api.constants import EventTypes from synapse.api.room_versions import KNOWN_ROOM_VERSIONS, StateResolutionVersions +from synapse.events import EventBase from synapse.events.snapshot import EventContext from synapse.logging.utils import log_function from synapse.state import v1, v2 @@ -212,15 +214,17 @@ class StateHandler(object): return joined_hosts @defer.inlineCallbacks - def compute_event_context(self, event, old_state=None): + def compute_event_context( + self, event: EventBase, old_state: Optional[Iterable[EventBase]] = None + ): """Build an EventContext structure for the event. This works out what the current state should be for the event, and generates a new state group if necessary. Args: - event (synapse.events.EventBase): - old_state (dict|None): The state at the event if it can't be + event: + old_state: The state at the event if it can't be calculated from existing events. This is normally only specified when receiving an event from federation where we don't have the prev events for, e.g. when backfilling. @@ -251,113 +255,103 @@ class StateHandler(object): # group for it. context = EventContext.with_state( state_group=None, + state_group_before_event=None, current_state_ids=current_state_ids, prev_state_ids=prev_state_ids, ) return context + # + # first of all, figure out the state before the event + # + if old_state: - # We already have the state, so we don't need to calculate it. - # Let's just correctly fill out the context and create a - # new state group for it. - - prev_state_ids = {(s.type, s.state_key): s.event_id for s in old_state} - - if event.is_state(): - key = (event.type, event.state_key) - if key in prev_state_ids: - replaces = prev_state_ids[key] - if replaces != event.event_id: # Paranoia check - event.unsigned["replaces_state"] = replaces - current_state_ids = dict(prev_state_ids) - current_state_ids[key] = event.event_id - else: - current_state_ids = prev_state_ids + # if we're given the state before the event, then we use that + state_ids_before_event = { + (s.type, s.state_key): s.event_id for s in old_state + } + state_group_before_event = None + state_group_before_event_prev_group = None + deltas_to_state_group_before_event = None - state_group = yield self.state_store.store_state_group( - event.event_id, - event.room_id, - prev_group=None, - delta_ids=None, - current_state_ids=current_state_ids, - ) + else: + # otherwise, we'll need to resolve the state across the prev_events. + logger.debug("calling resolve_state_groups from compute_event_context") - context = EventContext.with_state( - state_group=state_group, - current_state_ids=current_state_ids, - prev_state_ids=prev_state_ids, + entry = yield self.resolve_state_groups_for_events( + event.room_id, event.prev_event_ids() ) - return context + state_ids_before_event = entry.state + state_group_before_event = entry.state_group + state_group_before_event_prev_group = entry.prev_group + deltas_to_state_group_before_event = entry.delta_ids - logger.debug("calling resolve_state_groups from compute_event_context") + # + # make sure that we have a state group at that point. If it's not a state event, + # that will be the state group for the new event. If it *is* a state event, + # it might get rejected (in which case we'll need to persist it with the + # previous state group) + # - entry = yield self.resolve_state_groups_for_events( - event.room_id, event.prev_event_ids() - ) + if not state_group_before_event: + state_group_before_event = yield self.state_store.store_state_group( + event.event_id, + event.room_id, + prev_group=state_group_before_event_prev_group, + delta_ids=deltas_to_state_group_before_event, + current_state_ids=state_ids_before_event, + ) - prev_state_ids = entry.state - prev_group = None - delta_ids = None + # XXX: can we update the state cache entry for the new state group? or + # could we set a flag on resolve_state_groups_for_events to tell it to + # always make a state group? + + # + # now if it's not a state event, we're done + # + + if not event.is_state(): + return EventContext.with_state( + state_group_before_event=state_group_before_event, + state_group=state_group_before_event, + current_state_ids=state_ids_before_event, + prev_state_ids=state_ids_before_event, + prev_group=state_group_before_event_prev_group, + delta_ids=deltas_to_state_group_before_event, + ) - if event.is_state(): - # If this is a state event then we need to create a new state - # group for the state after this event. + # + # otherwise, we'll need to create a new state group for after the event + # - key = (event.type, event.state_key) - if key in prev_state_ids: - replaces = prev_state_ids[key] + key = (event.type, event.state_key) + if key in state_ids_before_event: + replaces = state_ids_before_event[key] + if replaces != event.event_id: event.unsigned["replaces_state"] = replaces - current_state_ids = dict(prev_state_ids) - current_state_ids[key] = event.event_id - - if entry.state_group: - # If the state at the event has a state group assigned then - # we can use that as the prev group - prev_group = entry.state_group - delta_ids = {key: event.event_id} - elif entry.prev_group: - # If the state at the event only has a prev group, then we can - # use that as a prev group too. - prev_group = entry.prev_group - delta_ids = dict(entry.delta_ids) - delta_ids[key] = event.event_id - - state_group = yield self.state_store.store_state_group( - event.event_id, - event.room_id, - prev_group=prev_group, - delta_ids=delta_ids, - current_state_ids=current_state_ids, - ) - else: - current_state_ids = prev_state_ids - prev_group = entry.prev_group - delta_ids = entry.delta_ids - - if entry.state_group is None: - entry.state_group = yield self.state_store.store_state_group( - event.event_id, - event.room_id, - prev_group=entry.prev_group, - delta_ids=entry.delta_ids, - current_state_ids=current_state_ids, - ) - entry.state_id = entry.state_group - - state_group = entry.state_group - - context = EventContext.with_state( - state_group=state_group, - current_state_ids=current_state_ids, - prev_state_ids=prev_state_ids, - prev_group=prev_group, + state_ids_after_event = dict(state_ids_before_event) + state_ids_after_event[key] = event.event_id + delta_ids = {key: event.event_id} + + state_group_after_event = yield self.state_store.store_state_group( + event.event_id, + event.room_id, + prev_group=state_group_before_event, delta_ids=delta_ids, + current_state_ids=state_ids_after_event, ) - return context + return EventContext.with_state( + state_group=state_group_after_event, + state_group_before_event=state_group_before_event, + current_state_ids=state_ids_after_event, + prev_state_ids=state_ids_before_event, + prev_group=state_group_before_event, + delta_ids=delta_ids, + ) @measure_func() @defer.inlineCallbacks diff --git a/synapse/storage/data_stores/main/state.py b/synapse/storage/data_stores/main/state.py index 3132848034..9e1541988e 100644 --- a/synapse/storage/data_stores/main/state.py +++ b/synapse/storage/data_stores/main/state.py @@ -1231,7 +1231,7 @@ class StateStore(StateGroupWorkerStore, StateBackgroundUpdateStore): # if the event was rejected, just give it the same state as its # predecessor. if context.rejected: - state_groups[event.event_id] = context.prev_group + state_groups[event.event_id] = context.state_group_before_event continue state_groups[event.event_id] = context.state_group diff --git a/tests/handlers/test_federation.py b/tests/handlers/test_federation.py index d56220f403..b4d92cf732 100644 --- a/tests/handlers/test_federation.py +++ b/tests/handlers/test_federation.py @@ -12,13 +12,19 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. +import logging + from synapse.api.constants import EventTypes from synapse.api.errors import AuthError, Codes +from synapse.federation.federation_base import event_from_pdu_json +from synapse.logging.context import LoggingContext, run_in_background from synapse.rest import admin from synapse.rest.client.v1 import login, room from tests import unittest +logger = logging.getLogger(__name__) + class FederationTestCase(unittest.HomeserverTestCase): servlets = [ @@ -79,3 +85,123 @@ class FederationTestCase(unittest.HomeserverTestCase): self.assertEqual(failure.code, 403, failure) self.assertEqual(failure.errcode, Codes.FORBIDDEN, failure) self.assertEqual(failure.msg, "You are not invited to this room.") + + def test_rejected_message_event_state(self): + """ + Check that we store the state group correctly for rejected non-state events. + + Regression test for #6289. + """ + OTHER_SERVER = "otherserver" + OTHER_USER = "@otheruser:" + OTHER_SERVER + + # create the room + user_id = self.register_user("kermit", "test") + tok = self.login("kermit", "test") + room_id = self.helper.create_room_as(room_creator=user_id, tok=tok) + + # pretend that another server has joined + join_event = self._build_and_send_join_event(OTHER_SERVER, OTHER_USER, room_id) + + # check the state group + sg = self.successResultOf( + self.store._get_state_group_for_event(join_event.event_id) + ) + + # build and send an event which will be rejected + ev = event_from_pdu_json( + { + "type": EventTypes.Message, + "content": {}, + "room_id": room_id, + "sender": "@yetanotheruser:" + OTHER_SERVER, + "depth": join_event["depth"] + 1, + "prev_events": [join_event.event_id], + "auth_events": [], + "origin_server_ts": self.clock.time_msec(), + }, + join_event.format_version, + ) + + with LoggingContext(request="send_rejected"): + d = run_in_background(self.handler.on_receive_pdu, OTHER_SERVER, ev) + self.get_success(d) + + # that should have been rejected + e = self.get_success(self.store.get_event(ev.event_id, allow_rejected=True)) + self.assertIsNotNone(e.rejected_reason) + + # ... and the state group should be the same as before + sg2 = self.successResultOf(self.store._get_state_group_for_event(ev.event_id)) + + self.assertEqual(sg, sg2) + + def test_rejected_state_event_state(self): + """ + Check that we store the state group correctly for rejected state events. + + Regression test for #6289. + """ + OTHER_SERVER = "otherserver" + OTHER_USER = "@otheruser:" + OTHER_SERVER + + # create the room + user_id = self.register_user("kermit", "test") + tok = self.login("kermit", "test") + room_id = self.helper.create_room_as(room_creator=user_id, tok=tok) + + # pretend that another server has joined + join_event = self._build_and_send_join_event(OTHER_SERVER, OTHER_USER, room_id) + + # check the state group + sg = self.successResultOf( + self.store._get_state_group_for_event(join_event.event_id) + ) + + # build and send an event which will be rejected + ev = event_from_pdu_json( + { + "type": "org.matrix.test", + "state_key": "test_key", + "content": {}, + "room_id": room_id, + "sender": "@yetanotheruser:" + OTHER_SERVER, + "depth": join_event["depth"] + 1, + "prev_events": [join_event.event_id], + "auth_events": [], + "origin_server_ts": self.clock.time_msec(), + }, + join_event.format_version, + ) + + with LoggingContext(request="send_rejected"): + d = run_in_background(self.handler.on_receive_pdu, OTHER_SERVER, ev) + self.get_success(d) + + # that should have been rejected + e = self.get_success(self.store.get_event(ev.event_id, allow_rejected=True)) + self.assertIsNotNone(e.rejected_reason) + + # ... and the state group should be the same as before + sg2 = self.successResultOf(self.store._get_state_group_for_event(ev.event_id)) + + self.assertEqual(sg, sg2) + + def _build_and_send_join_event(self, other_server, other_user, room_id): + join_event = self.get_success( + self.handler.on_make_join_request(other_server, room_id, other_user) + ) + # the auth code requires that a signature exists, but doesn't check that + # signature... go figure. + join_event.signatures[other_server] = {"x": "y"} + with LoggingContext(request="send_join"): + d = run_in_background( + self.handler.on_send_join_request, other_server, join_event + ) + self.get_success(d) + + # sanity-check: the room should show that the new user is a member + r = self.get_success(self.store.get_current_state_ids(room_id)) + self.assertEqual(r[(EventTypes.Member, other_user)], join_event.event_id) + + return join_event diff --git a/tests/test_state.py b/tests/test_state.py index 38246555bd..176535947a 100644 --- a/tests/test_state.py +++ b/tests/test_state.py @@ -21,6 +21,7 @@ from synapse.api.auth import Auth from synapse.api.constants import EventTypes, Membership from synapse.api.room_versions import RoomVersions from synapse.events import FrozenEvent +from synapse.events.snapshot import EventContext from synapse.state import StateHandler, StateResolutionHandler from tests import unittest @@ -198,16 +199,22 @@ class StateTestCase(unittest.TestCase): self.store.register_events(graph.walk()) - context_store = {} + context_store = {} # type: dict[str, EventContext] for event in graph.walk(): context = yield self.state.compute_event_context(event) self.store.register_event_context(event, context) context_store[event.event_id] = context - prev_state_ids = yield context_store["D"].get_prev_state_ids(self.store) + ctx_c = context_store["C"] + ctx_d = context_store["D"] + + prev_state_ids = yield ctx_d.get_prev_state_ids(self.store) self.assertEqual(2, len(prev_state_ids)) + self.assertEqual(ctx_c.state_group, ctx_d.state_group_before_event) + self.assertEqual(ctx_d.state_group_before_event, ctx_d.state_group) + @defer.inlineCallbacks def test_branch_basic_conflict(self): graph = Graph( @@ -241,12 +248,19 @@ class StateTestCase(unittest.TestCase): self.store.register_event_context(event, context) context_store[event.event_id] = context - prev_state_ids = yield context_store["D"].get_prev_state_ids(self.store) + # C ends up winning the resolution between B and C + + ctx_c = context_store["C"] + ctx_d = context_store["D"] + prev_state_ids = yield ctx_d.get_prev_state_ids(self.store) self.assertSetEqual( {"START", "A", "C"}, {e_id for e_id in prev_state_ids.values()} ) + self.assertEqual(ctx_c.state_group, ctx_d.state_group_before_event) + self.assertEqual(ctx_d.state_group_before_event, ctx_d.state_group) + @defer.inlineCallbacks def test_branch_have_banned_conflict(self): graph = Graph( @@ -292,11 +306,18 @@ class StateTestCase(unittest.TestCase): self.store.register_event_context(event, context) context_store[event.event_id] = context - prev_state_ids = yield context_store["E"].get_prev_state_ids(self.store) + # C ends up winning the resolution between C and D because bans win over other + # changes + + ctx_c = context_store["C"] + ctx_e = context_store["E"] + prev_state_ids = yield ctx_e.get_prev_state_ids(self.store) self.assertSetEqual( {"START", "A", "B", "C"}, {e for e in prev_state_ids.values()} ) + self.assertEqual(ctx_c.state_group, ctx_e.state_group_before_event) + self.assertEqual(ctx_e.state_group_before_event, ctx_e.state_group) @defer.inlineCallbacks def test_branch_have_perms_conflict(self): @@ -360,12 +381,20 @@ class StateTestCase(unittest.TestCase): self.store.register_event_context(event, context) context_store[event.event_id] = context - prev_state_ids = yield context_store["D"].get_prev_state_ids(self.store) + # B ends up winning the resolution between B and C because power levels + # win over other changes. + ctx_b = context_store["B"] + ctx_d = context_store["D"] + + prev_state_ids = yield ctx_d.get_prev_state_ids(self.store) self.assertSetEqual( {"A1", "A2", "A3", "A5", "B"}, {e for e in prev_state_ids.values()} ) + self.assertEqual(ctx_b.state_group, ctx_d.state_group_before_event) + self.assertEqual(ctx_d.state_group_before_event, ctx_d.state_group) + def _add_depths(self, nodes, edges): def _get_depth(ev): node = nodes[ev] @@ -390,13 +419,16 @@ class StateTestCase(unittest.TestCase): context = yield self.state.compute_event_context(event, old_state=old_state) - current_state_ids = yield context.get_current_state_ids(self.store) + prev_state_ids = yield context.get_prev_state_ids(self.store) + self.assertCountEqual((e.event_id for e in old_state), prev_state_ids.values()) - self.assertEqual( - set(e.event_id for e in old_state), set(current_state_ids.values()) + current_state_ids = yield context.get_current_state_ids(self.store) + self.assertCountEqual( + (e.event_id for e in old_state), current_state_ids.values() ) - self.assertIsNotNone(context.state_group) + self.assertIsNotNone(context.state_group_before_event) + self.assertEqual(context.state_group_before_event, context.state_group) @defer.inlineCallbacks def test_annotate_with_old_state(self): @@ -411,11 +443,18 @@ class StateTestCase(unittest.TestCase): context = yield self.state.compute_event_context(event, old_state=old_state) prev_state_ids = yield context.get_prev_state_ids(self.store) + self.assertCountEqual((e.event_id for e in old_state), prev_state_ids.values()) - self.assertEqual( - set(e.event_id for e in old_state), set(prev_state_ids.values()) + current_state_ids = yield context.get_current_state_ids(self.store) + self.assertCountEqual( + (e.event_id for e in old_state + [event]), current_state_ids.values() ) + self.assertIsNotNone(context.state_group_before_event) + self.assertNotEqual(context.state_group_before_event, context.state_group) + self.assertEqual(context.state_group_before_event, context.prev_group) + self.assertEqual({("state", ""): event.event_id}, context.delta_ids) + @defer.inlineCallbacks def test_trivial_annotate_message(self): prev_event_id = "prev_event_id" -- cgit 1.4.1 From 5c3363233cca7044a333b7e19ba239eaf5587ff8 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 6 Nov 2019 17:02:05 +0000 Subject: Fix deleting state groups during room purge. And fix the tests to actually test that things got deleted. --- synapse/storage/data_stores/main/events.py | 27 ++++++++++++++------------- tests/rest/admin/test_admin.py | 4 +++- 2 files changed, 17 insertions(+), 14 deletions(-) (limited to 'tests') diff --git a/synapse/storage/data_stores/main/events.py b/synapse/storage/data_stores/main/events.py index d69c59f5a1..946823876a 100644 --- a/synapse/storage/data_stores/main/events.py +++ b/synapse/storage/data_stores/main/events.py @@ -1633,7 +1633,20 @@ class EventsStore( return self.runInteraction("purge_room", self._purge_room_txn, room_id) def _purge_room_txn(self, txn, room_id): - # First delete tables which lack an index on room_id but have one on event_id + # First we fetch all the state groups that should be deleted, before + # we delete that information. + txn.execute( + """ + SELECT DISTINCT state_group FROM events + INNER JOIN event_to_state_groups USING(event_id) + WHERE events.room_id = ? + """, + (room_id,), + ) + + state_groups = [row[0] for row in txn] + + # Now we delete tables which lack an index on room_id but have one on event_id for table in ( "event_auth", "event_edges", @@ -1717,18 +1730,6 @@ class EventsStore( # index on them. In any case we should be clearing out 'stream' tables # periodically anyway (#5888) - # Now we fetch all the state groups that should be deleted. - txn.execute( - """ - SELECT DISTINCT state_group FROM events - INNER JOIN event_to_state_groups USING(event_id) - WHERE events.room_id = ? - """, - (room_id,), - ) - - state_groups = [row[0] for row in txn] - # TODO: we could probably usefully do a bunch of cache invalidation here logger.info("[purge] done") diff --git a/tests/rest/admin/test_admin.py b/tests/rest/admin/test_admin.py index 8e1ca8b738..d9f1b95cb0 100644 --- a/tests/rest/admin/test_admin.py +++ b/tests/rest/admin/test_admin.py @@ -628,10 +628,12 @@ class PurgeRoomTestCase(unittest.HomeserverTestCase): "local_invites", "room_account_data", "room_tags", + "state_groups", + "state_groups_state", ): count = self.get_success( self.store._simple_select_one_onecol( - table="events", + table=table, keyvalues={"room_id": room_id}, retcol="COUNT(*)", desc="test_purge_room", -- cgit 1.4.1 From f03c9d34442368c8f2c26d2ac16b770bc451c76d Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Wed, 6 Nov 2019 15:47:40 +0000 Subject: Don't apply retention policy based filtering on state events As per MSC1763, 'Retention is only considered for non-state events.', so don't filter out state events based on the room's retention policy. --- synapse/visibility.py | 15 +++++++++------ tests/rest/client/test_retention.py | 10 ++++++++++ 2 files changed, 19 insertions(+), 6 deletions(-) (limited to 'tests') diff --git a/synapse/visibility.py b/synapse/visibility.py index 4498c156bc..4d4141dacc 100644 --- a/synapse/visibility.py +++ b/synapse/visibility.py @@ -111,14 +111,17 @@ def filter_events_for_client( if not event.is_state() and event.sender in ignore_list: return None - retention_policy = retention_policies[event.room_id] - max_lifetime = retention_policy.get("max_lifetime") + # Don't try to apply the room's retention policy if the event is a state event, as + # MSC1763 states that retention is only considered for non-state events. + if not event.is_state(): + retention_policy = retention_policies[event.room_id] + max_lifetime = retention_policy.get("max_lifetime") - if max_lifetime is not None: - oldest_allowed_ts = storage.main.clock.time_msec() - max_lifetime + if max_lifetime is not None: + oldest_allowed_ts = storage.main.clock.time_msec() - max_lifetime - if event.origin_server_ts < oldest_allowed_ts: - return None + if event.origin_server_ts < oldest_allowed_ts: + return None if event.event_id in always_include_ids: return event diff --git a/tests/rest/client/test_retention.py b/tests/rest/client/test_retention.py index 41ea9db689..7b6f25a838 100644 --- a/tests/rest/client/test_retention.py +++ b/tests/rest/client/test_retention.py @@ -164,6 +164,12 @@ class RetentionTestCase(unittest.HomeserverTestCase): self.assertEqual(filtered_events[0].event_id, valid_event_id, filtered_events) def _test_retention_event_purged(self, room_id, increment): + # Get the create event to, later, check that we can still access it. + message_handler = self.hs.get_message_handler() + create_event = self.get_success( + message_handler.get_room_data(self.user_id, room_id, EventTypes.Create) + ) + # Send a first event to the room. This is the event we'll want to be purged at the # end of the test. resp = self.helper.send( @@ -202,6 +208,10 @@ class RetentionTestCase(unittest.HomeserverTestCase): valid_event = self.get_event(room_id, valid_event_id) self.assertEqual(valid_event.get("content", {}).get("body"), "2", valid_event) + # Check that we can still access state events that were sent before the event that + # has been purged. + self.get_event(room_id, create_event.event_id) + def get_event(self, room_id, event_id, expected_code=200): url = "/_matrix/client/r0/rooms/%s/event/%s" % (room_id, event_id) -- cgit 1.4.1 From c350bc2f92d87e46a40f917f65c9e10e0f4999fc Mon Sep 17 00:00:00 2001 From: Andrew Morgan <1342360+anoadragon453@users.noreply.github.com> Date: Wed, 13 Nov 2019 19:09:20 +0000 Subject: Blacklist PurgeRoomTestCase (#6361) --- changelog.d/6361.misc | 1 + tests/rest/admin/test_admin.py | 2 ++ 2 files changed, 3 insertions(+) create mode 100644 changelog.d/6361.misc (limited to 'tests') diff --git a/changelog.d/6361.misc b/changelog.d/6361.misc new file mode 100644 index 0000000000..324d74ebf9 --- /dev/null +++ b/changelog.d/6361.misc @@ -0,0 +1 @@ +Temporarily blacklist the failing unit test PurgeRoomTestCase.test_purge_room. diff --git a/tests/rest/admin/test_admin.py b/tests/rest/admin/test_admin.py index d9f1b95cb0..9575058252 100644 --- a/tests/rest/admin/test_admin.py +++ b/tests/rest/admin/test_admin.py @@ -641,3 +641,5 @@ class PurgeRoomTestCase(unittest.HomeserverTestCase): ) self.assertEqual(count, 0, msg="Rows not purged in {}".format(table)) + + test_purge_room.skip = "Disabled because it's currently broken" -- cgit 1.4.1 From 7c24d0f443724082376c89f9f75954d81f524a8e Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Tue, 19 Nov 2019 13:22:37 +0000 Subject: Lint --- synapse/config/server.py | 39 ++++++++++------- synapse/handlers/pagination.py | 17 +++----- synapse/storage/data_stores/main/room.py | 49 +++++++++++---------- tests/rest/client/test_retention.py | 73 ++++++++++---------------------- 4 files changed, 77 insertions(+), 101 deletions(-) (limited to 'tests') diff --git a/synapse/config/server.py b/synapse/config/server.py index aa93a416f1..8a55ffac4f 100644 --- a/synapse/config/server.py +++ b/synapse/config/server.py @@ -19,7 +19,7 @@ import logging import os.path import re from textwrap import indent -from typing import List +from typing import List, Dict, Optional import attr import yaml @@ -287,13 +287,17 @@ class ServerConfig(Config): self.retention_default_min_lifetime = None self.retention_default_max_lifetime = None - self.retention_allowed_lifetime_min = retention_config.get("allowed_lifetime_min") + self.retention_allowed_lifetime_min = retention_config.get( + "allowed_lifetime_min" + ) if self.retention_allowed_lifetime_min is not None: self.retention_allowed_lifetime_min = self.parse_duration( self.retention_allowed_lifetime_min ) - self.retention_allowed_lifetime_max = retention_config.get("allowed_lifetime_max") + self.retention_allowed_lifetime_max = retention_config.get( + "allowed_lifetime_max" + ) if self.retention_allowed_lifetime_max is not None: self.retention_allowed_lifetime_max = self.parse_duration( self.retention_allowed_lifetime_max @@ -302,14 +306,15 @@ class ServerConfig(Config): if ( self.retention_allowed_lifetime_min is not None and self.retention_allowed_lifetime_max is not None - and self.retention_allowed_lifetime_min > self.retention_allowed_lifetime_max + and self.retention_allowed_lifetime_min + > self.retention_allowed_lifetime_max ): raise ConfigError( "Invalid retention policy limits: 'allowed_lifetime_min' can not be" " greater than 'allowed_lifetime_max'" ) - self.retention_purge_jobs = [] + self.retention_purge_jobs = [] # type: List[Dict[str, Optional[int]]] for purge_job_config in retention_config.get("purge_jobs", []): interval_config = purge_job_config.get("interval") @@ -342,18 +347,22 @@ class ServerConfig(Config): " 'longest_max_lifetime' value." ) - self.retention_purge_jobs.append({ - "interval": interval, - "shortest_max_lifetime": shortest_max_lifetime, - "longest_max_lifetime": longest_max_lifetime, - }) + self.retention_purge_jobs.append( + { + "interval": interval, + "shortest_max_lifetime": shortest_max_lifetime, + "longest_max_lifetime": longest_max_lifetime, + } + ) if not self.retention_purge_jobs: - self.retention_purge_jobs = [{ - "interval": self.parse_duration("1d"), - "shortest_max_lifetime": None, - "longest_max_lifetime": None, - }] + self.retention_purge_jobs = [ + { + "interval": self.parse_duration("1d"), + "shortest_max_lifetime": None, + "longest_max_lifetime": None, + } + ] self.listeners = [] # type: List[dict] for listener in config.get("listeners", []): diff --git a/synapse/handlers/pagination.py b/synapse/handlers/pagination.py index e1800177fa..d122c11a4d 100644 --- a/synapse/handlers/pagination.py +++ b/synapse/handlers/pagination.py @@ -154,20 +154,17 @@ class PaginationHandler(object): # Figure out what token we should start purging at. ts = self.clock.time_msec() - max_lifetime - stream_ordering = ( - yield self.store.find_first_stream_ordering_after_ts(ts) - ) + stream_ordering = yield self.store.find_first_stream_ordering_after_ts(ts) - r = ( - yield self.store.get_room_event_after_stream_ordering( - room_id, stream_ordering, - ) + r = yield self.store.get_room_event_after_stream_ordering( + room_id, stream_ordering, ) if not r: logger.warning( "[purge] purging events not possible: No event found " "(ts %i => stream_ordering %i)", - ts, stream_ordering, + ts, + stream_ordering, ) continue @@ -186,9 +183,7 @@ class PaginationHandler(object): # the background so that it's not blocking any other operation apart from # other purges in the same room. run_as_background_process( - "_purge_history", - self._purge_history, - purge_id, room_id, token, True, + "_purge_history", self._purge_history, purge_id, room_id, token, True, ) def start_purge_history(self, room_id, token, delete_local_events=False): diff --git a/synapse/storage/data_stores/main/room.py b/synapse/storage/data_stores/main/room.py index 54a7d24c73..7fceae59ca 100644 --- a/synapse/storage/data_stores/main/room.py +++ b/synapse/storage/data_stores/main/room.py @@ -334,8 +334,9 @@ class RoomStore(RoomWorkerStore, SearchStore): WHERE state.room_id > ? AND state.type = '%s' ORDER BY state.room_id ASC LIMIT ?; - """ % EventTypes.Retention, - (last_room, batch_size) + """ + % EventTypes.Retention, + (last_room, batch_size), ) rows = self.cursor_to_dict(txn) @@ -358,15 +359,13 @@ class RoomStore(RoomWorkerStore, SearchStore): "event_id": row["event_id"], "min_lifetime": retention_policy.get("min_lifetime"), "max_lifetime": retention_policy.get("max_lifetime"), - } + }, ) logger.info("Inserted %d rows into room_retention", len(rows)) self._background_update_progress_txn( - txn, "insert_room_retention", { - "room_id": rows[-1]["room_id"], - } + txn, "insert_room_retention", {"room_id": rows[-1]["room_id"]} ) if batch_size > len(rows): @@ -375,8 +374,7 @@ class RoomStore(RoomWorkerStore, SearchStore): return False end = yield self.runInteraction( - "insert_room_retention", - _background_insert_retention_txn, + "insert_room_retention", _background_insert_retention_txn, ) if end: @@ -585,17 +583,15 @@ class RoomStore(RoomWorkerStore, SearchStore): ) def _store_retention_policy_for_room_txn(self, txn, event): - if ( - hasattr(event, "content") - and ("min_lifetime" in event.content or "max_lifetime" in event.content) + if hasattr(event, "content") and ( + "min_lifetime" in event.content or "max_lifetime" in event.content ): if ( - ("min_lifetime" in event.content and not isinstance( - event.content.get("min_lifetime"), integer_types - )) - or ("max_lifetime" in event.content and not isinstance( - event.content.get("max_lifetime"), integer_types - )) + "min_lifetime" in event.content + and not isinstance(event.content.get("min_lifetime"), integer_types) + ) or ( + "max_lifetime" in event.content + and not isinstance(event.content.get("max_lifetime"), integer_types) ): # Ignore the event if one of the value isn't an integer. return @@ -798,7 +794,9 @@ class RoomStore(RoomWorkerStore, SearchStore): return local_media_mxcs, remote_media_mxcs @defer.inlineCallbacks - def get_rooms_for_retention_period_in_range(self, min_ms, max_ms, include_null=False): + def get_rooms_for_retention_period_in_range( + self, min_ms, max_ms, include_null=False + ): """Retrieves all of the rooms within the given retention range. Optionally includes the rooms which don't have a retention policy. @@ -904,23 +902,24 @@ class RoomStore(RoomWorkerStore, SearchStore): INNER JOIN current_state_events USING (event_id, room_id) WHERE room_id = ?; """, - (room_id,) + (room_id,), ) return self.cursor_to_dict(txn) ret = yield self.runInteraction( - "get_retention_policy_for_room", - get_retention_policy_for_room_txn, + "get_retention_policy_for_room", get_retention_policy_for_room_txn, ) # If we don't know this room ID, ret will be None, in this case return the default # policy. if not ret: - defer.returnValue({ - "min_lifetime": self.config.retention_default_min_lifetime, - "max_lifetime": self.config.retention_default_max_lifetime, - }) + defer.returnValue( + { + "min_lifetime": self.config.retention_default_min_lifetime, + "max_lifetime": self.config.retention_default_max_lifetime, + } + ) row = ret[0] diff --git a/tests/rest/client/test_retention.py b/tests/rest/client/test_retention.py index 7b6f25a838..6bf485c239 100644 --- a/tests/rest/client/test_retention.py +++ b/tests/rest/client/test_retention.py @@ -61,9 +61,7 @@ class RetentionTestCase(unittest.HomeserverTestCase): self.helper.send_state( room_id=room_id, event_type=EventTypes.Retention, - body={ - "max_lifetime": one_day_ms * 4, - }, + body={"max_lifetime": one_day_ms * 4}, tok=self.token, expect_code=400, ) @@ -71,9 +69,7 @@ class RetentionTestCase(unittest.HomeserverTestCase): self.helper.send_state( room_id=room_id, event_type=EventTypes.Retention, - body={ - "max_lifetime": one_hour_ms, - }, + body={"max_lifetime": one_hour_ms}, tok=self.token, expect_code=400, ) @@ -89,9 +85,7 @@ class RetentionTestCase(unittest.HomeserverTestCase): self.helper.send_state( room_id=room_id, event_type=EventTypes.Retention, - body={ - "max_lifetime": lifetime, - }, + body={"max_lifetime": lifetime}, tok=self.token, ) @@ -115,20 +109,12 @@ class RetentionTestCase(unittest.HomeserverTestCase): events = [] # Send a first event, which should be filtered out at the end of the test. - resp = self.helper.send( - room_id=room_id, - body="1", - tok=self.token, - ) + resp = self.helper.send(room_id=room_id, body="1", tok=self.token) # Get the event from the store so that we end up with a FrozenEvent that we can # give to filter_events_for_client. We need to do this now because the event won't # be in the database anymore after it has expired. - events.append(self.get_success( - store.get_event( - resp.get("event_id") - ) - )) + events.append(self.get_success(store.get_event(resp.get("event_id")))) # Advance the time by 2 days. We're using the default retention policy, therefore # after this the first event will still be valid. @@ -143,20 +129,16 @@ class RetentionTestCase(unittest.HomeserverTestCase): valid_event_id = resp.get("event_id") - events.append(self.get_success( - store.get_event( - valid_event_id - ) - )) + events.append(self.get_success(store.get_event(valid_event_id))) # Advance the time by anothe 2 days. After this, the first event should be # outdated but not the second one. self.reactor.advance(one_day_ms * 2 / 1000) # Run filter_events_for_client with our list of FrozenEvents. - filtered_events = self.get_success(filter_events_for_client( - storage, self.user_id, events - )) + filtered_events = self.get_success( + filter_events_for_client(storage, self.user_id, events) + ) # We should only get one event back. self.assertEqual(len(filtered_events), 1, filtered_events) @@ -172,28 +154,22 @@ class RetentionTestCase(unittest.HomeserverTestCase): # Send a first event to the room. This is the event we'll want to be purged at the # end of the test. - resp = self.helper.send( - room_id=room_id, - body="1", - tok=self.token, - ) + resp = self.helper.send(room_id=room_id, body="1", tok=self.token) expired_event_id = resp.get("event_id") # Check that we can retrieve the event. expired_event = self.get_event(room_id, expired_event_id) - self.assertEqual(expired_event.get("content", {}).get("body"), "1", expired_event) + self.assertEqual( + expired_event.get("content", {}).get("body"), "1", expired_event + ) # Advance the time. self.reactor.advance(increment / 1000) # Send another event. We need this because the purge job won't purge the most # recent event in the room. - resp = self.helper.send( - room_id=room_id, - body="2", - tok=self.token, - ) + resp = self.helper.send(room_id=room_id, body="2", tok=self.token) valid_event_id = resp.get("event_id") @@ -240,8 +216,7 @@ class RetentionNoDefaultPolicyTestCase(unittest.HomeserverTestCase): mock_federation_client = Mock(spec=["backfill"]) self.hs = self.setup_test_homeserver( - config=config, - federation_client=mock_federation_client, + config=config, federation_client=mock_federation_client, ) return self.hs @@ -268,9 +243,7 @@ class RetentionNoDefaultPolicyTestCase(unittest.HomeserverTestCase): self.helper.send_state( room_id=room_id, event_type=EventTypes.Retention, - body={ - "max_lifetime": one_day_ms * 35, - }, + body={"max_lifetime": one_day_ms * 35}, tok=self.token, ) @@ -289,18 +262,16 @@ class RetentionNoDefaultPolicyTestCase(unittest.HomeserverTestCase): # Check that we can retrieve the event. expired_event = self.get_event(room_id, first_event_id) - self.assertEqual(expired_event.get("content", {}).get("body"), "1", expired_event) + self.assertEqual( + expired_event.get("content", {}).get("body"), "1", expired_event + ) # Advance the time by a month. self.reactor.advance(one_day_ms * 30 / 1000) # Send another event. We need this because the purge job won't purge the most # recent event in the room. - resp = self.helper.send( - room_id=room_id, - body="2", - tok=self.token, - ) + resp = self.helper.send(room_id=room_id, body="2", tok=self.token) second_event_id = resp.get("event_id") @@ -313,7 +284,9 @@ class RetentionNoDefaultPolicyTestCase(unittest.HomeserverTestCase): ) if expected_code_for_first_event == 200: - self.assertEqual(first_event.get("content", {}).get("body"), "1", first_event) + self.assertEqual( + first_event.get("content", {}).get("body"), "1", first_event + ) # Check that the event that hasn't been purged can still be retrieved. second_event = self.get_event(room_id, second_event_id) -- cgit 1.4.1 From bf9a11c54d3c9ca1e4fa9420b567efceb1e46d5b Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Tue, 19 Nov 2019 13:30:04 +0000 Subject: Lint again --- synapse/config/server.py | 2 +- tests/rest/client/test_retention.py | 12 ++---------- 2 files changed, 3 insertions(+), 11 deletions(-) (limited to 'tests') diff --git a/synapse/config/server.py b/synapse/config/server.py index 8a55ffac4f..ed916e1400 100644 --- a/synapse/config/server.py +++ b/synapse/config/server.py @@ -19,7 +19,7 @@ import logging import os.path import re from textwrap import indent -from typing import List, Dict, Optional +from typing import Dict, List, Optional import attr import yaml diff --git a/tests/rest/client/test_retention.py b/tests/rest/client/test_retention.py index 6bf485c239..9e549d8a91 100644 --- a/tests/rest/client/test_retention.py +++ b/tests/rest/client/test_retention.py @@ -121,11 +121,7 @@ class RetentionTestCase(unittest.HomeserverTestCase): self.reactor.advance(one_day_ms * 2 / 1000) # Send another event, which shouldn't get filtered out. - resp = self.helper.send( - room_id=room_id, - body="2", - tok=self.token, - ) + resp = self.helper.send(room_id=room_id, body="2", tok=self.token) valid_event_id = resp.get("event_id") @@ -252,11 +248,7 @@ class RetentionNoDefaultPolicyTestCase(unittest.HomeserverTestCase): def _test_retention(self, room_id, expected_code_for_first_event=200): # Send a first event to the room. This is the event we'll want to be purged at the # end of the test. - resp = self.helper.send( - room_id=room_id, - body="1", - tok=self.token, - ) + resp = self.helper.send(room_id=room_id, body="1", tok=self.token) first_event_id = resp.get("event_id") -- cgit 1.4.1 From 6356f2088f0adb681fe24a8435955b19883fa3b4 Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Wed, 20 Nov 2019 12:09:06 +0000 Subject: Test if a purge can make /messages return 500 responses --- tests/rest/client/v1/test_rooms.py | 72 ++++++++++++++++++++++++++++++++++++++ 1 file changed, 72 insertions(+) (limited to 'tests') diff --git a/tests/rest/client/v1/test_rooms.py b/tests/rest/client/v1/test_rooms.py index 5e38fd6ced..ebaa67e899 100644 --- a/tests/rest/client/v1/test_rooms.py +++ b/tests/rest/client/v1/test_rooms.py @@ -25,7 +25,9 @@ from twisted.internet import defer import synapse.rest.admin from synapse.api.constants import EventContentFields, EventTypes, Membership +from synapse.handlers.pagination import PurgeStatus from synapse.rest.client.v1 import login, profile, room +from synapse.util.stringutils import random_string from tests import unittest @@ -910,6 +912,76 @@ class RoomMessageListTestCase(RoomBase): return channel.json_body["chunk"] + def test_room_messages_purge(self): + store = self.hs.get_datastore() + pagination_handler = self.hs.get_pagination_handler() + + # Send a first message in the room, which will be removed by the purge. + first_event_id = self.helper.send(self.room_id, "message 1")["event_id"] + first_token = self.get_success( + store.get_topological_token_for_event(first_event_id) + ) + + # Send a second message in the room, which won't be removed, and which we'll + # use as the marker to purge events before. + second_event_id = self.helper.send(self.room_id, "message 2")["event_id"] + second_token = self.get_success( + store.get_topological_token_for_event(second_event_id) + ) + + # Send a third event in the room to ensure we don't fall under any edge case + # due to our marker being the latest forward extremity in the room. + self.helper.send(self.room_id, "message 3") + + # Check that we get the first and second message when querying /messages. + request, channel = self.make_request( + "GET", + "/rooms/%s/messages?access_token=x&from=%s&dir=b&filter=%s" + % (self.room_id, second_token, json.dumps({"types": [EventTypes.Message]})), + ) + self.render(request) + self.assertEqual(channel.code, 200, channel.json_body) + + chunk = channel.json_body["chunk"] + self.assertEqual(len(chunk), 2, [event["content"] for event in chunk]) + + # Purge every event before the second event. + purge_id = random_string(16) + pagination_handler._purges_by_id[purge_id] = PurgeStatus() + self.get_success(pagination_handler._purge_history( + purge_id=purge_id, + room_id=self.room_id, + token=second_token, + delete_local_events=True, + )) + + # Check that we only get the second message through /message now that the first + # has been purged. + request, channel = self.make_request( + "GET", + "/rooms/%s/messages?access_token=x&from=%s&dir=b&filter=%s" + % (self.room_id, second_token, json.dumps({"types": [EventTypes.Message]})), + ) + self.render(request) + self.assertEqual(channel.code, 200, channel.json_body) + + chunk = channel.json_body["chunk"] + self.assertEqual(len(chunk), 1, [event["content"] for event in chunk]) + + # Check that we get no event, but also no error, when querying /messages with + # the token that was pointing at the first event, because we don't have it + # anymore. + request, channel = self.make_request( + "GET", + "/rooms/%s/messages?access_token=x&from=%s&dir=b&filter=%s" + % (self.room_id, first_token, json.dumps({"types": [EventTypes.Message]})), + ) + self.render(request) + self.assertEqual(channel.code, 200, channel.json_body) + + chunk = channel.json_body["chunk"] + self.assertEqual(len(chunk), 0, [event["content"] for event in chunk]) + class RoomSearchTestCase(unittest.HomeserverTestCase): servlets = [ -- cgit 1.4.1 From e2a20326e8141fdf9304434901da38c64b917a78 Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Wed, 20 Nov 2019 15:08:47 +0000 Subject: Lint --- tests/rest/client/v1/test_rooms.py | 14 ++++++++------ 1 file changed, 8 insertions(+), 6 deletions(-) (limited to 'tests') diff --git a/tests/rest/client/v1/test_rooms.py b/tests/rest/client/v1/test_rooms.py index ebaa67e899..e84e578f99 100644 --- a/tests/rest/client/v1/test_rooms.py +++ b/tests/rest/client/v1/test_rooms.py @@ -948,12 +948,14 @@ class RoomMessageListTestCase(RoomBase): # Purge every event before the second event. purge_id = random_string(16) pagination_handler._purges_by_id[purge_id] = PurgeStatus() - self.get_success(pagination_handler._purge_history( - purge_id=purge_id, - room_id=self.room_id, - token=second_token, - delete_local_events=True, - )) + self.get_success( + pagination_handler._purge_history( + purge_id=purge_id, + room_id=self.room_id, + token=second_token, + delete_local_events=True, + ) + ) # Check that we only get the second message through /message now that the first # has been purged. -- cgit 1.4.1 From 9eebd46048d0b34767047b2156760a1467f19ae6 Mon Sep 17 00:00:00 2001 From: Amber Brown Date: Tue, 26 Nov 2019 03:45:50 +1100 Subject: Improve the performance of structured logging (#6322) --- changelog.d/6322.misc | 1 + synapse/logging/_structured.py | 14 +++++- synapse/logging/_terse_json.py | 106 ++++++++++++++++++++++++++++++----------- tests/server.py | 2 + 4 files changed, 93 insertions(+), 30 deletions(-) create mode 100644 changelog.d/6322.misc (limited to 'tests') diff --git a/changelog.d/6322.misc b/changelog.d/6322.misc new file mode 100644 index 0000000000..70ef36ca80 --- /dev/null +++ b/changelog.d/6322.misc @@ -0,0 +1 @@ +Improve the performance of outputting structured logging. diff --git a/synapse/logging/_structured.py b/synapse/logging/_structured.py index 334ddaf39a..ffa7b20ca8 100644 --- a/synapse/logging/_structured.py +++ b/synapse/logging/_structured.py @@ -261,6 +261,18 @@ def parse_drain_configs( ) +class StoppableLogPublisher(LogPublisher): + """ + A log publisher that can tell its observers to shut down any external + communications. + """ + + def stop(self): + for obs in self._observers: + if hasattr(obs, "stop"): + obs.stop() + + def setup_structured_logging( hs, config, @@ -336,7 +348,7 @@ def setup_structured_logging( # We should never get here, but, just in case, throw an error. raise ConfigError("%s drain type cannot be configured" % (observer.type,)) - publisher = LogPublisher(*observers) + publisher = StoppableLogPublisher(*observers) log_filter = LogLevelFilterPredicate() for namespace, namespace_config in log_config.get( diff --git a/synapse/logging/_terse_json.py b/synapse/logging/_terse_json.py index 76ce7d8808..05fc64f409 100644 --- a/synapse/logging/_terse_json.py +++ b/synapse/logging/_terse_json.py @@ -17,25 +17,29 @@ Log formatters that output terse JSON. """ +import json import sys +import traceback from collections import deque from ipaddress import IPv4Address, IPv6Address, ip_address from math import floor -from typing import IO +from typing import IO, Optional import attr -from simplejson import dumps from zope.interface import implementer from twisted.application.internet import ClientService +from twisted.internet.defer import Deferred from twisted.internet.endpoints import ( HostnameEndpoint, TCP4ClientEndpoint, TCP6ClientEndpoint, ) +from twisted.internet.interfaces import IPushProducer, ITransport from twisted.internet.protocol import Factory, Protocol from twisted.logger import FileLogObserver, ILogObserver, Logger -from twisted.python.failure import Failure + +_encoder = json.JSONEncoder(ensure_ascii=False, separators=(",", ":")) def flatten_event(event: dict, metadata: dict, include_time: bool = False): @@ -141,11 +145,49 @@ def TerseJSONToConsoleLogObserver(outFile: IO[str], metadata: dict) -> FileLogOb def formatEvent(_event: dict) -> str: flattened = flatten_event(_event, metadata) - return dumps(flattened, ensure_ascii=False, separators=(",", ":")) + "\n" + return _encoder.encode(flattened) + "\n" return FileLogObserver(outFile, formatEvent) +@attr.s +@implementer(IPushProducer) +class LogProducer(object): + """ + An IPushProducer that writes logs from its buffer to its transport when it + is resumed. + + Args: + buffer: Log buffer to read logs from. + transport: Transport to write to. + """ + + transport = attr.ib(type=ITransport) + _buffer = attr.ib(type=deque) + _paused = attr.ib(default=False, type=bool, init=False) + + def pauseProducing(self): + self._paused = True + + def stopProducing(self): + self._paused = True + self._buffer = None + + def resumeProducing(self): + self._paused = False + + while self._paused is False and (self._buffer and self.transport.connected): + try: + event = self._buffer.popleft() + self.transport.write(_encoder.encode(event).encode("utf8")) + self.transport.write(b"\n") + except Exception: + # Something has gone wrong writing to the transport -- log it + # and break out of the while. + traceback.print_exc(file=sys.__stderr__) + break + + @attr.s @implementer(ILogObserver) class TerseJSONToTCPLogObserver(object): @@ -165,8 +207,9 @@ class TerseJSONToTCPLogObserver(object): metadata = attr.ib(type=dict) maximum_buffer = attr.ib(type=int) _buffer = attr.ib(default=attr.Factory(deque), type=deque) - _writer = attr.ib(default=None) + _connection_waiter = attr.ib(default=None, type=Optional[Deferred]) _logger = attr.ib(default=attr.Factory(Logger)) + _producer = attr.ib(default=None, type=Optional[LogProducer]) def start(self) -> None: @@ -187,38 +230,43 @@ class TerseJSONToTCPLogObserver(object): factory = Factory.forProtocol(Protocol) self._service = ClientService(endpoint, factory, clock=self.hs.get_reactor()) self._service.startService() + self._connect() - def _write_loop(self) -> None: + def stop(self): + self._service.stopService() + + def _connect(self) -> None: """ - Implement the write loop. + Triggers an attempt to connect then write to the remote if not already writing. """ - if self._writer: + if self._connection_waiter: return - self._writer = self._service.whenConnected() + self._connection_waiter = self._service.whenConnected(failAfterFailures=1) + + @self._connection_waiter.addErrback + def fail(r): + r.printTraceback(file=sys.__stderr__) + self._connection_waiter = None + self._connect() - @self._writer.addBoth + @self._connection_waiter.addCallback def writer(r): - if isinstance(r, Failure): - r.printTraceback(file=sys.__stderr__) - self._writer = None - self.hs.get_reactor().callLater(1, self._write_loop) + # We have a connection. If we already have a producer, and its + # transport is the same, just trigger a resumeProducing. + if self._producer and r.transport is self._producer.transport: + self._producer.resumeProducing() return - try: - for event in self._buffer: - r.transport.write( - dumps(event, ensure_ascii=False, separators=(",", ":")).encode( - "utf8" - ) - ) - r.transport.write(b"\n") - self._buffer.clear() - except Exception as e: - sys.__stderr__.write("Failed writing out logs with %s\n" % (str(e),)) - - self._writer = False - self.hs.get_reactor().callLater(1, self._write_loop) + # If the producer is still producing, stop it. + if self._producer: + self._producer.stopProducing() + + # Make a new producer and start it. + self._producer = LogProducer(buffer=self._buffer, transport=r.transport) + r.transport.registerProducer(self._producer, True) + self._producer.resumeProducing() + self._connection_waiter = None def _handle_pressure(self) -> None: """ @@ -277,4 +325,4 @@ class TerseJSONToTCPLogObserver(object): self._logger.failure("Failed clearing backpressure") # Try and write immediately. - self._write_loop() + self._connect() diff --git a/tests/server.py b/tests/server.py index f878aeaada..2b7cf4242e 100644 --- a/tests/server.py +++ b/tests/server.py @@ -379,6 +379,7 @@ class FakeTransport(object): disconnecting = False disconnected = False + connected = True buffer = attr.ib(default=b"") producer = attr.ib(default=None) autoflush = attr.ib(default=True) @@ -402,6 +403,7 @@ class FakeTransport(object): "FakeTransport: Delaying disconnect until buffer is flushed" ) else: + self.connected = False self.disconnected = True def abortConnection(self): -- cgit 1.4.1 From f0ef9708241ec65fe6f32f1ad36f719ab4ab2b53 Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Tue, 26 Nov 2019 17:49:12 +0000 Subject: Don't restrict the tests to v1 rooms --- tests/rest/client/test_retention.py | 2 -- 1 file changed, 2 deletions(-) (limited to 'tests') diff --git a/tests/rest/client/test_retention.py b/tests/rest/client/test_retention.py index 9e549d8a91..95475bb651 100644 --- a/tests/rest/client/test_retention.py +++ b/tests/rest/client/test_retention.py @@ -34,7 +34,6 @@ class RetentionTestCase(unittest.HomeserverTestCase): def make_homeserver(self, reactor, clock): config = self.default_config() - config["default_room_version"] = "1" config["retention"] = { "enabled": True, "default_policy": { @@ -204,7 +203,6 @@ class RetentionNoDefaultPolicyTestCase(unittest.HomeserverTestCase): def make_homeserver(self, reactor, clock): config = self.default_config() - config["default_room_version"] = "1" config["retention"] = { "enabled": True, } -- cgit 1.4.1 From ce578031f4d0fe6f1eb26de4cb3d30a4175468db Mon Sep 17 00:00:00 2001 From: Andrew Morgan <1342360+anoadragon453@users.noreply.github.com> Date: Tue, 26 Nov 2019 18:42:27 +0000 Subject: Remove assertion and provide a clear warning on startup for missing public_baseurl (#6379) --- changelog.d/6379.misc | 1 + synapse/config/emailconfig.py | 2 ++ synapse/config/registration.py | 7 +++++++ tests/rest/client/v2_alpha/test_register.py | 1 + 4 files changed, 11 insertions(+) create mode 100644 changelog.d/6379.misc (limited to 'tests') diff --git a/changelog.d/6379.misc b/changelog.d/6379.misc new file mode 100644 index 0000000000..725c2e7d87 --- /dev/null +++ b/changelog.d/6379.misc @@ -0,0 +1 @@ +Complain on startup instead of 500'ing during runtime when `public_baseurl` isn't set when necessary. \ No newline at end of file diff --git a/synapse/config/emailconfig.py b/synapse/config/emailconfig.py index 43fad0bf8b..ac1724045f 100644 --- a/synapse/config/emailconfig.py +++ b/synapse/config/emailconfig.py @@ -146,6 +146,8 @@ class EmailConfig(Config): if k not in email_config: missing.append("email." + k) + # public_baseurl is required to build password reset and validation links that + # will be emailed to users if config.get("public_baseurl") is None: missing.append("public_baseurl") diff --git a/synapse/config/registration.py b/synapse/config/registration.py index 1f6dac69da..ee9614c5f7 100644 --- a/synapse/config/registration.py +++ b/synapse/config/registration.py @@ -106,6 +106,13 @@ class RegistrationConfig(Config): account_threepid_delegates = config.get("account_threepid_delegates") or {} self.account_threepid_delegate_email = account_threepid_delegates.get("email") self.account_threepid_delegate_msisdn = account_threepid_delegates.get("msisdn") + if self.account_threepid_delegate_msisdn and not self.public_baseurl: + raise ConfigError( + "The configuration option `public_baseurl` is required if " + "`account_threepid_delegate.msisdn` is set, such that " + "clients know where to submit validation tokens to. Please " + "configure `public_baseurl`." + ) self.default_identity_server = config.get("default_identity_server") self.allow_guest_access = config.get("allow_guest_access", False) diff --git a/tests/rest/client/v2_alpha/test_register.py b/tests/rest/client/v2_alpha/test_register.py index dab87e5edf..c0d0d2b44e 100644 --- a/tests/rest/client/v2_alpha/test_register.py +++ b/tests/rest/client/v2_alpha/test_register.py @@ -203,6 +203,7 @@ class RegisterRestServletTestCase(unittest.HomeserverTestCase): @unittest.override_config( { + "public_baseurl": "https://test_server", "enable_registration_captcha": True, "user_consent": { "version": "1", -- cgit 1.4.1 From 0d27aba900136514a8801b902f9a8ac69150e2c0 Mon Sep 17 00:00:00 2001 From: Hubert Chathi Date: Wed, 27 Nov 2019 16:14:44 -0500 Subject: add etag and count to key backup endpoints (#5858) --- changelog.d/5858.feature | 1 + synapse/handlers/e2e_room_keys.py | 130 +++++++----- synapse/rest/client/v2_alpha/room_keys.py | 8 +- synapse/storage/data_stores/main/e2e_room_keys.py | 226 +++++++++++++++------ .../main/schema/delta/56/room_key_etag.sql | 17 ++ tests/handlers/test_e2e_room_keys.py | 31 +++ tests/storage/test_e2e_room_keys.py | 8 +- 7 files changed, 297 insertions(+), 124 deletions(-) create mode 100644 changelog.d/5858.feature create mode 100644 synapse/storage/data_stores/main/schema/delta/56/room_key_etag.sql (limited to 'tests') diff --git a/changelog.d/5858.feature b/changelog.d/5858.feature new file mode 100644 index 0000000000..55ee93051e --- /dev/null +++ b/changelog.d/5858.feature @@ -0,0 +1 @@ +Add etag and count fields to key backup endpoints to help clients guess if there are new keys. diff --git a/synapse/handlers/e2e_room_keys.py b/synapse/handlers/e2e_room_keys.py index 0cea445f0d..f1b4424a02 100644 --- a/synapse/handlers/e2e_room_keys.py +++ b/synapse/handlers/e2e_room_keys.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2017, 2018 New Vector Ltd +# Copyright 2019 Matrix.org Foundation C.I.C. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -103,14 +104,35 @@ class E2eRoomKeysHandler(object): rooms session_id(string): session ID to delete keys for, for None to delete keys for all sessions + Raises: + NotFoundError: if the backup version does not exist Returns: - A deferred of the deletion transaction + A dict containing the count and etag for the backup version """ # lock for consistency with uploading with (yield self._upload_linearizer.queue(user_id)): + # make sure the backup version exists + try: + version_info = yield self.store.get_e2e_room_keys_version_info( + user_id, version + ) + except StoreError as e: + if e.code == 404: + raise NotFoundError("Unknown backup version") + else: + raise + yield self.store.delete_e2e_room_keys(user_id, version, room_id, session_id) + version_etag = version_info["etag"] + 1 + yield self.store.update_e2e_room_keys_version( + user_id, version, None, version_etag + ) + + count = yield self.store.count_e2e_room_keys(user_id, version) + return {"etag": str(version_etag), "count": count} + @trace @defer.inlineCallbacks def upload_room_keys(self, user_id, version, room_keys): @@ -138,6 +160,9 @@ class E2eRoomKeysHandler(object): } } + Returns: + A dict containing the count and etag for the backup version + Raises: NotFoundError: if there are no versions defined RoomKeysVersionError: if the uploaded version is not the current version @@ -171,59 +196,62 @@ class E2eRoomKeysHandler(object): else: raise - # go through the room_keys. - # XXX: this should/could be done concurrently, given we're in a lock. + # Fetch any existing room keys for the sessions that have been + # submitted. Then compare them with the submitted keys. If the + # key is new, insert it; if the key should be updated, then update + # it; otherwise, drop it. + existing_keys = yield self.store.get_e2e_room_keys_multi( + user_id, version, room_keys["rooms"] + ) + to_insert = [] # batch the inserts together + changed = False # if anything has changed, we need to update the etag for room_id, room in iteritems(room_keys["rooms"]): - for session_id, session in iteritems(room["sessions"]): - yield self._upload_room_key( - user_id, version, room_id, session_id, session + for session_id, room_key in iteritems(room["sessions"]): + log_kv( + { + "message": "Trying to upload room key", + "room_id": room_id, + "session_id": session_id, + "user_id": user_id, + } ) - - @defer.inlineCallbacks - def _upload_room_key(self, user_id, version, room_id, session_id, room_key): - """Upload a given room_key for a given room and session into a given - version of the backup. Merges the key with any which might already exist. - - Args: - user_id(str): the user whose backup we're setting - version(str): the version ID of the backup we're updating - room_id(str): the ID of the room whose keys we're setting - session_id(str): the session whose room_key we're setting - room_key(dict): the room_key being set - """ - log_kv( - { - "message": "Trying to upload room key", - "room_id": room_id, - "session_id": session_id, - "user_id": user_id, - } - ) - # get the room_key for this particular row - current_room_key = None - try: - current_room_key = yield self.store.get_e2e_room_key( - user_id, version, room_id, session_id - ) - except StoreError as e: - if e.code == 404: - log_kv( - { - "message": "Room key not found.", - "room_id": room_id, - "user_id": user_id, - } + current_room_key = existing_keys.get(room_id, {}).get(session_id) + if current_room_key: + if self._should_replace_room_key(current_room_key, room_key): + log_kv({"message": "Replacing room key."}) + # updates are done one at a time in the DB, so send + # updates right away rather than batching them up, + # like we do with the inserts + yield self.store.update_e2e_room_key( + user_id, version, room_id, session_id, room_key + ) + changed = True + else: + log_kv({"message": "Not replacing room_key."}) + else: + log_kv( + { + "message": "Room key not found.", + "room_id": room_id, + "user_id": user_id, + } + ) + log_kv({"message": "Replacing room key."}) + to_insert.append((room_id, session_id, room_key)) + changed = True + + if len(to_insert): + yield self.store.add_e2e_room_keys(user_id, version, to_insert) + + version_etag = version_info["etag"] + if changed: + version_etag = version_etag + 1 + yield self.store.update_e2e_room_keys_version( + user_id, version, None, version_etag ) - else: - raise - if self._should_replace_room_key(current_room_key, room_key): - log_kv({"message": "Replacing room key."}) - yield self.store.set_e2e_room_key( - user_id, version, room_id, session_id, room_key - ) - else: - log_kv({"message": "Not replacing room_key."}) + count = yield self.store.count_e2e_room_keys(user_id, version) + return {"etag": str(version_etag), "count": count} @staticmethod def _should_replace_room_key(current_room_key, room_key): @@ -314,6 +342,8 @@ class E2eRoomKeysHandler(object): raise NotFoundError("Unknown backup version") else: raise + + res["count"] = yield self.store.count_e2e_room_keys(user_id, res["version"]) return res @trace diff --git a/synapse/rest/client/v2_alpha/room_keys.py b/synapse/rest/client/v2_alpha/room_keys.py index d596786430..d83ac8e3c5 100644 --- a/synapse/rest/client/v2_alpha/room_keys.py +++ b/synapse/rest/client/v2_alpha/room_keys.py @@ -134,8 +134,8 @@ class RoomKeysServlet(RestServlet): if room_id: body = {"rooms": {room_id: body}} - yield self.e2e_room_keys_handler.upload_room_keys(user_id, version, body) - return 200, {} + ret = yield self.e2e_room_keys_handler.upload_room_keys(user_id, version, body) + return 200, ret @defer.inlineCallbacks def on_GET(self, request, room_id, session_id): @@ -239,10 +239,10 @@ class RoomKeysServlet(RestServlet): user_id = requester.user.to_string() version = parse_string(request, "version") - yield self.e2e_room_keys_handler.delete_room_keys( + ret = yield self.e2e_room_keys_handler.delete_room_keys( user_id, version, room_id, session_id ) - return 200, {} + return 200, ret class RoomKeysNewVersionServlet(RestServlet): diff --git a/synapse/storage/data_stores/main/e2e_room_keys.py b/synapse/storage/data_stores/main/e2e_room_keys.py index 1cbbae5b63..113224fd7c 100644 --- a/synapse/storage/data_stores/main/e2e_room_keys.py +++ b/synapse/storage/data_stores/main/e2e_room_keys.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2017 New Vector Ltd +# Copyright 2019 Matrix.org Foundation C.I.C. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -24,49 +25,8 @@ from synapse.storage._base import SQLBaseStore class EndToEndRoomKeyStore(SQLBaseStore): @defer.inlineCallbacks - def get_e2e_room_key(self, user_id, version, room_id, session_id): - """Get the encrypted E2E room key for a given session from a given - backup version of room_keys. We only store the 'best' room key for a given - session at a given time, as determined by the handler. - - Args: - user_id(str): the user whose backup we're querying - version(str): the version ID of the backup for the set of keys we're querying - room_id(str): the ID of the room whose keys we're querying. - This is a bit redundant as it's implied by the session_id, but - we include for consistency with the rest of the API. - session_id(str): the session whose room_key we're querying. - - Returns: - A deferred dict giving the session_data and message metadata for - this room key. - """ - - row = yield self._simple_select_one( - table="e2e_room_keys", - keyvalues={ - "user_id": user_id, - "version": version, - "room_id": room_id, - "session_id": session_id, - }, - retcols=( - "first_message_index", - "forwarded_count", - "is_verified", - "session_data", - ), - desc="get_e2e_room_key", - ) - - row["session_data"] = json.loads(row["session_data"]) - - return row - - @defer.inlineCallbacks - def set_e2e_room_key(self, user_id, version, room_id, session_id, room_key): - """Replaces or inserts the encrypted E2E room key for a given session in - a given backup + def update_e2e_room_key(self, user_id, version, room_id, session_id, room_key): + """Replaces the encrypted E2E room key for a given session in a given backup Args: user_id(str): the user whose backup we're setting @@ -78,7 +38,7 @@ class EndToEndRoomKeyStore(SQLBaseStore): StoreError """ - yield self._simple_upsert( + yield self._simple_update_one( table="e2e_room_keys", keyvalues={ "user_id": user_id, @@ -86,21 +46,51 @@ class EndToEndRoomKeyStore(SQLBaseStore): "room_id": room_id, "session_id": session_id, }, - values={ + updatevalues={ "first_message_index": room_key["first_message_index"], "forwarded_count": room_key["forwarded_count"], "is_verified": room_key["is_verified"], "session_data": json.dumps(room_key["session_data"]), }, - lock=False, + desc="update_e2e_room_key", ) - log_kv( - { - "message": "Set room key", - "room_id": room_id, - "session_id": session_id, - "room_key": room_key, - } + + @defer.inlineCallbacks + def add_e2e_room_keys(self, user_id, version, room_keys): + """Bulk add room keys to a given backup. + + Args: + user_id (str): the user whose backup we're adding to + version (str): the version ID of the backup for the set of keys we're adding to + room_keys (iterable[(str, str, dict)]): the keys to add, in the form + (roomID, sessionID, keyData) + """ + + values = [] + for (room_id, session_id, room_key) in room_keys: + values.append( + { + "user_id": user_id, + "version": version, + "room_id": room_id, + "session_id": session_id, + "first_message_index": room_key["first_message_index"], + "forwarded_count": room_key["forwarded_count"], + "is_verified": room_key["is_verified"], + "session_data": json.dumps(room_key["session_data"]), + } + ) + log_kv( + { + "message": "Set room key", + "room_id": room_id, + "session_id": session_id, + "room_key": room_key, + } + ) + + yield self._simple_insert_many( + table="e2e_room_keys", values=values, desc="add_e2e_room_keys" ) @trace @@ -110,11 +100,11 @@ class EndToEndRoomKeyStore(SQLBaseStore): room, or a given session. Args: - user_id(str): the user whose backup we're querying - version(str): the version ID of the backup for the set of keys we're querying - room_id(str): Optional. the ID of the room whose keys we're querying, if any. + user_id (str): the user whose backup we're querying + version (str): the version ID of the backup for the set of keys we're querying + room_id (str): Optional. the ID of the room whose keys we're querying, if any. If not specified, we return the keys for all the rooms in the backup. - session_id(str): Optional. the session whose room_key we're querying, if any. + session_id (str): Optional. the session whose room_key we're querying, if any. If specified, we also require the room_id to be specified. If not specified, we return all the keys in this version of the backup (or for the specified room) @@ -162,6 +152,95 @@ class EndToEndRoomKeyStore(SQLBaseStore): return sessions + def get_e2e_room_keys_multi(self, user_id, version, room_keys): + """Get multiple room keys at a time. The difference between this function and + get_e2e_room_keys is that this function can be used to retrieve + multiple specific keys at a time, whereas get_e2e_room_keys is used for + getting all the keys in a backup version, all the keys for a room, or a + specific key. + + Args: + user_id (str): the user whose backup we're querying + version (str): the version ID of the backup we're querying about + room_keys (dict[str, dict[str, iterable[str]]]): a map from + room ID -> {"session": [session ids]} indicating the session IDs + that we want to query + + Returns: + Deferred[dict[str, dict[str, dict]]]: a map of room IDs to session IDs to room key + """ + + return self.runInteraction( + "get_e2e_room_keys_multi", + self._get_e2e_room_keys_multi_txn, + user_id, + version, + room_keys, + ) + + @staticmethod + def _get_e2e_room_keys_multi_txn(txn, user_id, version, room_keys): + if not room_keys: + return {} + + where_clauses = [] + params = [user_id, version] + for room_id, room in room_keys.items(): + sessions = list(room["sessions"]) + if not sessions: + continue + params.append(room_id) + params.extend(sessions) + where_clauses.append( + "(room_id = ? AND session_id IN (%s))" + % (",".join(["?" for _ in sessions]),) + ) + + # check if we're actually querying something + if not where_clauses: + return {} + + sql = """ + SELECT room_id, session_id, first_message_index, forwarded_count, + is_verified, session_data + FROM e2e_room_keys + WHERE user_id = ? AND version = ? AND (%s) + """ % ( + " OR ".join(where_clauses) + ) + + txn.execute(sql, params) + + ret = {} + + for row in txn: + room_id = row[0] + session_id = row[1] + ret.setdefault(room_id, {}) + ret[room_id][session_id] = { + "first_message_index": row[2], + "forwarded_count": row[3], + "is_verified": row[4], + "session_data": json.loads(row[5]), + } + + return ret + + def count_e2e_room_keys(self, user_id, version): + """Get the number of keys in a backup version. + + Args: + user_id (str): the user whose backup we're querying + version (str): the version ID of the backup we're querying about + """ + + return self._simple_select_one_onecol( + table="e2e_room_keys", + keyvalues={"user_id": user_id, "version": version}, + retcol="COUNT(*)", + desc="count_e2e_room_keys", + ) + @trace @defer.inlineCallbacks def delete_e2e_room_keys(self, user_id, version, room_id=None, session_id=None): @@ -219,6 +298,7 @@ class EndToEndRoomKeyStore(SQLBaseStore): version(str) algorithm(str) auth_data(object): opaque dict supplied by the client + etag(int): tag of the keys in the backup """ def _get_e2e_room_keys_version_info_txn(txn): @@ -236,10 +316,12 @@ class EndToEndRoomKeyStore(SQLBaseStore): txn, table="e2e_room_keys_versions", keyvalues={"user_id": user_id, "version": this_version, "deleted": 0}, - retcols=("version", "algorithm", "auth_data"), + retcols=("version", "algorithm", "auth_data", "etag"), ) result["auth_data"] = json.loads(result["auth_data"]) result["version"] = str(result["version"]) + if result["etag"] is None: + result["etag"] = 0 return result return self.runInteraction( @@ -288,21 +370,33 @@ class EndToEndRoomKeyStore(SQLBaseStore): ) @trace - def update_e2e_room_keys_version(self, user_id, version, info): + def update_e2e_room_keys_version( + self, user_id, version, info=None, version_etag=None + ): """Update a given backup version Args: user_id(str): the user whose backup version we're updating version(str): the version ID of the backup version we're updating - info(dict): the new backup version info to store + info (dict): the new backup version info to store. If None, then + the backup version info is not updated + version_etag (Optional[int]): etag of the keys in the backup. If + None, then the etag is not updated """ + updatevalues = {} - return self._simple_update( - table="e2e_room_keys_versions", - keyvalues={"user_id": user_id, "version": version}, - updatevalues={"auth_data": json.dumps(info["auth_data"])}, - desc="update_e2e_room_keys_version", - ) + if info is not None and "auth_data" in info: + updatevalues["auth_data"] = json.dumps(info["auth_data"]) + if version_etag is not None: + updatevalues["etag"] = version_etag + + if updatevalues: + return self._simple_update( + table="e2e_room_keys_versions", + keyvalues={"user_id": user_id, "version": version}, + updatevalues=updatevalues, + desc="update_e2e_room_keys_version", + ) @trace def delete_e2e_room_keys_version(self, user_id, version=None): diff --git a/synapse/storage/data_stores/main/schema/delta/56/room_key_etag.sql b/synapse/storage/data_stores/main/schema/delta/56/room_key_etag.sql new file mode 100644 index 0000000000..7d70dd071e --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/56/room_key_etag.sql @@ -0,0 +1,17 @@ +/* Copyright 2019 Matrix.org Foundation C.I.C. + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +-- store the current etag of backup version +ALTER TABLE e2e_room_keys_versions ADD COLUMN etag BIGINT; diff --git a/tests/handlers/test_e2e_room_keys.py b/tests/handlers/test_e2e_room_keys.py index 0bb96674a2..70f172eb02 100644 --- a/tests/handlers/test_e2e_room_keys.py +++ b/tests/handlers/test_e2e_room_keys.py @@ -1,6 +1,7 @@ # -*- coding: utf-8 -*- # Copyright 2016 OpenMarket Ltd # Copyright 2017 New Vector Ltd +# Copyright 2019 Matrix.org Foundation C.I.C. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -94,23 +95,29 @@ class E2eRoomKeysHandlerTestCase(unittest.TestCase): # check we can retrieve it as the current version res = yield self.handler.get_version_info(self.local_user) + version_etag = res["etag"] + del res["etag"] self.assertDictEqual( res, { "version": "1", "algorithm": "m.megolm_backup.v1", "auth_data": "first_version_auth_data", + "count": 0, }, ) # check we can retrieve it as a specific version res = yield self.handler.get_version_info(self.local_user, "1") + self.assertEqual(res["etag"], version_etag) + del res["etag"] self.assertDictEqual( res, { "version": "1", "algorithm": "m.megolm_backup.v1", "auth_data": "first_version_auth_data", + "count": 0, }, ) @@ -126,12 +133,14 @@ class E2eRoomKeysHandlerTestCase(unittest.TestCase): # check we can retrieve it as the current version res = yield self.handler.get_version_info(self.local_user) + del res["etag"] self.assertDictEqual( res, { "version": "2", "algorithm": "m.megolm_backup.v1", "auth_data": "second_version_auth_data", + "count": 0, }, ) @@ -158,12 +167,14 @@ class E2eRoomKeysHandlerTestCase(unittest.TestCase): # check we can retrieve it as the current version res = yield self.handler.get_version_info(self.local_user) + del res["etag"] self.assertDictEqual( res, { "algorithm": "m.megolm_backup.v1", "auth_data": "revised_first_version_auth_data", "version": version, + "count": 0, }, ) @@ -207,12 +218,14 @@ class E2eRoomKeysHandlerTestCase(unittest.TestCase): # check we can retrieve it as the current version res = yield self.handler.get_version_info(self.local_user) + del res["etag"] # etag is opaque, so don't test its contents self.assertDictEqual( res, { "algorithm": "m.megolm_backup.v1", "auth_data": "revised_first_version_auth_data", "version": version, + "count": 0, }, ) @@ -409,6 +422,11 @@ class E2eRoomKeysHandlerTestCase(unittest.TestCase): yield self.handler.upload_room_keys(self.local_user, version, room_keys) + # get the etag to compare to future versions + res = yield self.handler.get_version_info(self.local_user) + backup_etag = res["etag"] + self.assertEqual(res["count"], 1) + new_room_keys = copy.deepcopy(room_keys) new_room_key = new_room_keys["rooms"]["!abc:matrix.org"]["sessions"]["c0ff33"] @@ -423,6 +441,10 @@ class E2eRoomKeysHandlerTestCase(unittest.TestCase): "SSBBTSBBIEZJU0gK", ) + # the etag should be the same since the session did not change + res = yield self.handler.get_version_info(self.local_user) + self.assertEqual(res["etag"], backup_etag) + # test that marking the session as verified however /does/ replace it new_room_key["is_verified"] = True yield self.handler.upload_room_keys(self.local_user, version, new_room_keys) @@ -432,6 +454,11 @@ class E2eRoomKeysHandlerTestCase(unittest.TestCase): res["rooms"]["!abc:matrix.org"]["sessions"]["c0ff33"]["session_data"], "new" ) + # the etag should NOT be equal now, since the key changed + res = yield self.handler.get_version_info(self.local_user) + self.assertNotEqual(res["etag"], backup_etag) + backup_etag = res["etag"] + # test that a session with a higher forwarded_count doesn't replace one # with a lower forwarding count new_room_key["forwarded_count"] = 2 @@ -443,6 +470,10 @@ class E2eRoomKeysHandlerTestCase(unittest.TestCase): res["rooms"]["!abc:matrix.org"]["sessions"]["c0ff33"]["session_data"], "new" ) + # the etag should be the same since the session did not change + res = yield self.handler.get_version_info(self.local_user) + self.assertEqual(res["etag"], backup_etag) + # TODO: check edge cases as well as the common variations here @defer.inlineCallbacks diff --git a/tests/storage/test_e2e_room_keys.py b/tests/storage/test_e2e_room_keys.py index d128fde441..35dafbb904 100644 --- a/tests/storage/test_e2e_room_keys.py +++ b/tests/storage/test_e2e_room_keys.py @@ -39,8 +39,8 @@ class E2eRoomKeysHandlerTestCase(unittest.HomeserverTestCase): ) self.get_success( - self.store.set_e2e_room_key( - "user_id", version1, "room", "session", room_key + self.store.add_e2e_room_keys( + "user_id", version1, [("room", "session", room_key)] ) ) @@ -51,8 +51,8 @@ class E2eRoomKeysHandlerTestCase(unittest.HomeserverTestCase): ) self.get_success( - self.store.set_e2e_room_key( - "user_id", version2, "room", "session", room_key + self.store.add_e2e_room_keys( + "user_id", version2, [("room", "session", room_key)] ) ) -- cgit 1.4.1 From 0f87b912aba7e678041632bc9a6d1f7c2d24342c Mon Sep 17 00:00:00 2001 From: Amber Brown Date: Thu, 28 Nov 2019 08:54:07 +1100 Subject: Implementation of MSC2314 (#6176) --- changelog.d/6176.feature | 1 + synapse/federation/federation_server.py | 26 ++++++++---- synapse/federation/transport/server.py | 6 +-- sytest-blacklist | 6 ++- tests/federation/test_complexity.py | 28 ++---------- tests/federation/test_federation_sender.py | 4 +- tests/federation/test_federation_server.py | 63 +++++++++++++++++++++++++++ tests/handlers/test_typing.py | 3 ++ tests/replication/slave/storage/_base.py | 3 ++ tests/replication/tcp/streams/_base.py | 4 ++ tests/storage/test_roommember.py | 26 +----------- tests/unittest.py | 68 +++++++++++++++++++++++++++++- tests/utils.py | 1 + 13 files changed, 174 insertions(+), 65 deletions(-) create mode 100644 changelog.d/6176.feature (limited to 'tests') diff --git a/changelog.d/6176.feature b/changelog.d/6176.feature new file mode 100644 index 0000000000..3c66d689d4 --- /dev/null +++ b/changelog.d/6176.feature @@ -0,0 +1 @@ +Implement the `/_matrix/federation/unstable/net.atleastfornow/state/` API as drafted in MSC2314. diff --git a/synapse/federation/federation_server.py b/synapse/federation/federation_server.py index d942d77a72..84d4eca041 100644 --- a/synapse/federation/federation_server.py +++ b/synapse/federation/federation_server.py @@ -1,6 +1,7 @@ # -*- coding: utf-8 -*- # Copyright 2015, 2016 OpenMarket Ltd # Copyright 2018 New Vector Ltd +# Copyright 2019 Matrix.org Federation C.I.C # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -73,6 +74,7 @@ class FederationServer(FederationBase): self.auth = hs.get_auth() self.handler = hs.get_handlers().federation_handler + self.state = hs.get_state_handler() self._server_linearizer = Linearizer("fed_server") self._transaction_linearizer = Linearizer("fed_txn_handler") @@ -264,9 +266,6 @@ class FederationServer(FederationBase): await self.registry.on_edu(edu_type, origin, content) async def on_context_state_request(self, origin, room_id, event_id): - if not event_id: - raise NotImplementedError("Specify an event") - origin_host, _ = parse_server_name(origin) await self.check_server_matches_acl(origin_host, room_id) @@ -280,13 +279,18 @@ class FederationServer(FederationBase): # - but that's non-trivial to get right, and anyway somewhat defeats # the point of the linearizer. with (await self._server_linearizer.queue((origin, room_id))): - resp = await self._state_resp_cache.wrap( - (room_id, event_id), - self._on_context_state_request_compute, - room_id, - event_id, + resp = dict( + await self._state_resp_cache.wrap( + (room_id, event_id), + self._on_context_state_request_compute, + room_id, + event_id, + ) ) + room_version = await self.store.get_room_version(room_id) + resp["room_version"] = room_version + return 200, resp async def on_state_ids_request(self, origin, room_id, event_id): @@ -306,7 +310,11 @@ class FederationServer(FederationBase): return 200, {"pdu_ids": state_ids, "auth_chain_ids": auth_chain_ids} async def _on_context_state_request_compute(self, room_id, event_id): - pdus = await self.handler.get_state_for_pdu(room_id, event_id) + if event_id: + pdus = await self.handler.get_state_for_pdu(room_id, event_id) + else: + pdus = (await self.state.get_current_state(room_id)).values() + auth_chain = await self.store.get_auth_chain([pdu.event_id for pdu in pdus]) return { diff --git a/synapse/federation/transport/server.py b/synapse/federation/transport/server.py index 09baa9c57d..fefc789c85 100644 --- a/synapse/federation/transport/server.py +++ b/synapse/federation/transport/server.py @@ -421,7 +421,7 @@ class FederationEventServlet(BaseFederationServlet): return await self.handler.on_pdu_request(origin, event_id) -class FederationStateServlet(BaseFederationServlet): +class FederationStateV1Servlet(BaseFederationServlet): PATH = "/state/(?P[^/]*)/?" # This is when someone asks for all data for a given context. @@ -429,7 +429,7 @@ class FederationStateServlet(BaseFederationServlet): return await self.handler.on_context_state_request( origin, context, - parse_string_from_args(query, "event_id", None, required=True), + parse_string_from_args(query, "event_id", None, required=False), ) @@ -1360,7 +1360,7 @@ class RoomComplexityServlet(BaseFederationServlet): FEDERATION_SERVLET_CLASSES = ( FederationSendServlet, FederationEventServlet, - FederationStateServlet, + FederationStateV1Servlet, FederationStateIdsServlet, FederationBackfillServlet, FederationQueryServlet, diff --git a/sytest-blacklist b/sytest-blacklist index 11785fd43f..411cce0692 100644 --- a/sytest-blacklist +++ b/sytest-blacklist @@ -1,6 +1,6 @@ # This file serves as a blacklist for SyTest tests that we expect will fail in # Synapse. -# +# # Each line of this file is scanned by sytest during a run and if the line # exactly matches the name of a test, it will be marked as "expected fail", # meaning the test will still run, but failure will not mark the entire test @@ -29,3 +29,7 @@ Enabling an unknown default rule fails with 404 # Blacklisted due to https://github.com/matrix-org/synapse/issues/1663 New federated private chats get full presence information (SYN-115) + +# Blacklisted due to https://github.com/matrix-org/matrix-doc/pull/2314 removing +# this requirement from the spec +Inbound federation of state requires event_id as a mandatory paramater diff --git a/tests/federation/test_complexity.py b/tests/federation/test_complexity.py index 51714a2b06..24fa8dbb45 100644 --- a/tests/federation/test_complexity.py +++ b/tests/federation/test_complexity.py @@ -18,17 +18,14 @@ from mock import Mock from twisted.internet import defer from synapse.api.errors import Codes, SynapseError -from synapse.config.ratelimiting import FederationRateLimitConfig -from synapse.federation.transport import server from synapse.rest import admin from synapse.rest.client.v1 import login, room from synapse.types import UserID -from synapse.util.ratelimitutils import FederationRateLimiter from tests import unittest -class RoomComplexityTests(unittest.HomeserverTestCase): +class RoomComplexityTests(unittest.FederatingHomeserverTestCase): servlets = [ admin.register_servlets, @@ -41,25 +38,6 @@ class RoomComplexityTests(unittest.HomeserverTestCase): config["limit_remote_rooms"] = {"enabled": True, "complexity": 0.05} return config - def prepare(self, reactor, clock, homeserver): - class Authenticator(object): - def authenticate_request(self, request, content): - return defer.succeed("otherserver.nottld") - - ratelimiter = FederationRateLimiter( - clock, - FederationRateLimitConfig( - window_size=1, - sleep_limit=1, - sleep_msec=1, - reject_limit=1000, - concurrent_requests=1000, - ), - ) - server.register_servlets( - homeserver, self.resource, Authenticator(), ratelimiter - ) - def test_complexity_simple(self): u1 = self.register_user("u1", "pass") @@ -105,7 +83,7 @@ class RoomComplexityTests(unittest.HomeserverTestCase): d = handler._remote_join( None, - ["otherserver.example"], + ["other.example.com"], "roomid", UserID.from_string(u1), {"membership": "join"}, @@ -146,7 +124,7 @@ class RoomComplexityTests(unittest.HomeserverTestCase): d = handler._remote_join( None, - ["otherserver.example"], + ["other.example.com"], room_1, UserID.from_string(u1), {"membership": "join"}, diff --git a/tests/federation/test_federation_sender.py b/tests/federation/test_federation_sender.py index cce8d8c6de..d456267b87 100644 --- a/tests/federation/test_federation_sender.py +++ b/tests/federation/test_federation_sender.py @@ -19,7 +19,7 @@ from twisted.internet import defer from synapse.types import ReadReceipt -from tests.unittest import HomeserverTestCase +from tests.unittest import HomeserverTestCase, override_config class FederationSenderTestCases(HomeserverTestCase): @@ -29,6 +29,7 @@ class FederationSenderTestCases(HomeserverTestCase): federation_transport_client=Mock(spec=["send_transaction"]), ) + @override_config({"send_federation": True}) def test_send_receipts(self): mock_state_handler = self.hs.get_state_handler() mock_state_handler.get_current_hosts_in_room.return_value = ["test", "host2"] @@ -69,6 +70,7 @@ class FederationSenderTestCases(HomeserverTestCase): ], ) + @override_config({"send_federation": True}) def test_send_receipts_with_backoff(self): """Send two receipts in quick succession; the second should be flushed, but only after 20ms""" diff --git a/tests/federation/test_federation_server.py b/tests/federation/test_federation_server.py index b08be451aa..1ec8c40901 100644 --- a/tests/federation/test_federation_server.py +++ b/tests/federation/test_federation_server.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2018 New Vector Ltd +# Copyright 2019 Matrix.org Federation C.I.C # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -16,6 +17,8 @@ import logging from synapse.events import FrozenEvent from synapse.federation.federation_server import server_matches_acl_event +from synapse.rest import admin +from synapse.rest.client.v1 import login, room from tests import unittest @@ -41,6 +44,66 @@ class ServerACLsTestCase(unittest.TestCase): self.assertTrue(server_matches_acl_event("1:2:3:4", e)) +class StateQueryTests(unittest.FederatingHomeserverTestCase): + + servlets = [ + admin.register_servlets, + room.register_servlets, + login.register_servlets, + ] + + def test_without_event_id(self): + """ + Querying v1/state/ without an event ID will return the current + known state. + """ + u1 = self.register_user("u1", "pass") + u1_token = self.login("u1", "pass") + + room_1 = self.helper.create_room_as(u1, tok=u1_token) + self.inject_room_member(room_1, "@user:other.example.com", "join") + + request, channel = self.make_request( + "GET", "/_matrix/federation/v1/state/%s" % (room_1,) + ) + self.render(request) + self.assertEquals(200, channel.code, channel.result) + + self.assertEqual( + channel.json_body["room_version"], + self.hs.config.default_room_version.identifier, + ) + + members = set( + map( + lambda x: x["state_key"], + filter( + lambda x: x["type"] == "m.room.member", channel.json_body["pdus"] + ), + ) + ) + + self.assertEqual(members, set(["@user:other.example.com", u1])) + self.assertEqual(len(channel.json_body["pdus"]), 6) + + def test_needs_to_be_in_room(self): + """ + Querying v1/state/ requires the server + be in the room to provide data. + """ + u1 = self.register_user("u1", "pass") + u1_token = self.login("u1", "pass") + + room_1 = self.helper.create_room_as(u1, tok=u1_token) + + request, channel = self.make_request( + "GET", "/_matrix/federation/v1/state/%s" % (room_1,) + ) + self.render(request) + self.assertEquals(403, channel.code, channel.result) + self.assertEqual(channel.json_body["errcode"], "M_FORBIDDEN") + + def _create_acl_event(content): return FrozenEvent( { diff --git a/tests/handlers/test_typing.py b/tests/handlers/test_typing.py index 5ec568f4e6..f6d8660285 100644 --- a/tests/handlers/test_typing.py +++ b/tests/handlers/test_typing.py @@ -24,6 +24,7 @@ from synapse.api.errors import AuthError from synapse.types import UserID from tests import unittest +from tests.unittest import override_config from tests.utils import register_federation_servlets # Some local users to test with @@ -174,6 +175,7 @@ class TypingNotificationsTestCase(unittest.HomeserverTestCase): ], ) + @override_config({"send_federation": True}) def test_started_typing_remote_send(self): self.room_members = [U_APPLE, U_ONION] @@ -237,6 +239,7 @@ class TypingNotificationsTestCase(unittest.HomeserverTestCase): ], ) + @override_config({"send_federation": True}) def test_stopped_typing(self): self.room_members = [U_APPLE, U_BANANA, U_ONION] diff --git a/tests/replication/slave/storage/_base.py b/tests/replication/slave/storage/_base.py index 4f924ce451..e7472e3a93 100644 --- a/tests/replication/slave/storage/_base.py +++ b/tests/replication/slave/storage/_base.py @@ -48,7 +48,10 @@ class BaseSlavedStoreTestCase(unittest.HomeserverTestCase): server_factory = ReplicationStreamProtocolFactory(self.hs) self.streamer = server_factory.streamer + handler_factory = Mock() self.replication_handler = ReplicationClientHandler(self.slaved_store) + self.replication_handler.factory = handler_factory + client_factory = ReplicationClientFactory( self.hs, "client_name", self.replication_handler ) diff --git a/tests/replication/tcp/streams/_base.py b/tests/replication/tcp/streams/_base.py index ce3835ae6a..1d14e77255 100644 --- a/tests/replication/tcp/streams/_base.py +++ b/tests/replication/tcp/streams/_base.py @@ -12,6 +12,8 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. +from mock import Mock + from synapse.replication.tcp.commands import ReplicateCommand from synapse.replication.tcp.protocol import ClientReplicationStreamProtocol from synapse.replication.tcp.resource import ReplicationStreamProtocolFactory @@ -30,7 +32,9 @@ class BaseStreamTestCase(unittest.HomeserverTestCase): server = server_factory.buildProtocol(None) # build a replication client, with a dummy handler + handler_factory = Mock() self.test_handler = TestReplicationClientHandler() + self.test_handler.factory = handler_factory self.client = ClientReplicationStreamProtocol( "client", "test", clock, self.test_handler ) diff --git a/tests/storage/test_roommember.py b/tests/storage/test_roommember.py index 9ddd17f73d..105a0c2b02 100644 --- a/tests/storage/test_roommember.py +++ b/tests/storage/test_roommember.py @@ -16,8 +16,7 @@ from unittest.mock import Mock -from synapse.api.constants import EventTypes, Membership -from synapse.api.room_versions import RoomVersions +from synapse.api.constants import Membership from synapse.rest.admin import register_servlets_for_client_rest_resource from synapse.rest.client.v1 import login, room from synapse.types import Requester, UserID @@ -44,9 +43,6 @@ class RoomMemberStoreTestCase(unittest.HomeserverTestCase): # We can't test the RoomMemberStore on its own without the other event # storage logic self.store = hs.get_datastore() - self.storage = hs.get_storage() - self.event_builder_factory = hs.get_event_builder_factory() - self.event_creation_handler = hs.get_event_creation_handler() self.u_alice = self.register_user("alice", "pass") self.t_alice = self.login("alice", "pass") @@ -55,26 +51,6 @@ class RoomMemberStoreTestCase(unittest.HomeserverTestCase): # User elsewhere on another host self.u_charlie = UserID.from_string("@charlie:elsewhere") - def inject_room_member(self, room, user, membership, replaces_state=None): - builder = self.event_builder_factory.for_room_version( - RoomVersions.V1, - { - "type": EventTypes.Member, - "sender": user, - "state_key": user, - "room_id": room, - "content": {"membership": membership}, - }, - ) - - event, context = self.get_success( - self.event_creation_handler.create_new_client_event(builder) - ) - - self.get_success(self.storage.persistence.persist_event(event, context)) - - return event - def test_one_member(self): # Alice creates the room, and is automatically joined diff --git a/tests/unittest.py b/tests/unittest.py index 561cebc223..31997a0f31 100644 --- a/tests/unittest.py +++ b/tests/unittest.py @@ -1,6 +1,7 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd # Copyright 2018 New Vector +# Copyright 2019 Matrix.org Federation C.I.C # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,6 +14,7 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. + import gc import hashlib import hmac @@ -27,13 +29,17 @@ from twisted.internet.defer import Deferred, succeed from twisted.python.threadpool import ThreadPool from twisted.trial import unittest -from synapse.api.constants import EventTypes +from synapse.api.constants import EventTypes, Membership +from synapse.api.room_versions import KNOWN_ROOM_VERSIONS from synapse.config.homeserver import HomeServerConfig +from synapse.config.ratelimiting import FederationRateLimitConfig +from synapse.federation.transport import server as federation_server from synapse.http.server import JsonResource from synapse.http.site import SynapseRequest from synapse.logging.context import LoggingContext from synapse.server import HomeServer from synapse.types import Requester, UserID, create_requester +from synapse.util.ratelimitutils import FederationRateLimiter from tests.server import get_clock, make_request, render, setup_test_homeserver from tests.test_utils.logging_setup import setup_logging @@ -559,6 +565,66 @@ class HomeserverTestCase(TestCase): self.render(request) self.assertEqual(channel.code, 403, channel.result) + def inject_room_member(self, room: str, user: str, membership: Membership) -> None: + """ + Inject a membership event into a room. + + Args: + room: Room ID to inject the event into. + user: MXID of the user to inject the membership for. + membership: The membership type. + """ + event_builder_factory = self.hs.get_event_builder_factory() + event_creation_handler = self.hs.get_event_creation_handler() + + room_version = self.get_success(self.hs.get_datastore().get_room_version(room)) + + builder = event_builder_factory.for_room_version( + KNOWN_ROOM_VERSIONS[room_version], + { + "type": EventTypes.Member, + "sender": user, + "state_key": user, + "room_id": room, + "content": {"membership": membership}, + }, + ) + + event, context = self.get_success( + event_creation_handler.create_new_client_event(builder) + ) + + self.get_success( + self.hs.get_storage().persistence.persist_event(event, context) + ) + + +class FederatingHomeserverTestCase(HomeserverTestCase): + """ + A federating homeserver that authenticates incoming requests as `other.example.com`. + """ + + def prepare(self, reactor, clock, homeserver): + class Authenticator(object): + def authenticate_request(self, request, content): + return succeed("other.example.com") + + ratelimiter = FederationRateLimiter( + clock, + FederationRateLimitConfig( + window_size=1, + sleep_limit=1, + sleep_msec=1, + reject_limit=1000, + concurrent_requests=1000, + ), + ) + federation_server.register_servlets( + homeserver, self.resource, Authenticator(), ratelimiter + ) + + return super().prepare(reactor, clock, homeserver) + def override_config(extra_config): """A decorator which can be applied to test functions to give additional HS config diff --git a/tests/utils.py b/tests/utils.py index 7dc9bdc505..de2ac1ed33 100644 --- a/tests/utils.py +++ b/tests/utils.py @@ -109,6 +109,7 @@ def default_config(name, parse=False): """ config_dict = { "server_name": name, + "send_federation": False, "media_store_path": "media", "uploads_path": "uploads", # the test signing key is just an arbitrary ed25519 key to keep the config -- cgit 1.4.1 From 8c9a713f8db1d6fcc1f876ac6fbd0e54b5e5819c Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Thu, 28 Nov 2019 11:32:06 +0000 Subject: Add tests --- tests/rest/client/v1/test_rooms.py | 140 +++++++++++++++++++++++++++++++++++++ 1 file changed, 140 insertions(+) (limited to 'tests') diff --git a/tests/rest/client/v1/test_rooms.py b/tests/rest/client/v1/test_rooms.py index e84e578f99..eda2fabc71 100644 --- a/tests/rest/client/v1/test_rooms.py +++ b/tests/rest/client/v1/test_rooms.py @@ -1180,3 +1180,143 @@ class PerRoomProfilesForbiddenTestCase(unittest.HomeserverTestCase): res_displayname = channel.json_body["content"]["displayname"] self.assertEqual(res_displayname, self.displayname, channel.result) + + +class RoomMembershipReasonTestCase(unittest.HomeserverTestCase): + """Tests that clients can add a "reason" field to membership events and + that they get correctly added to the generated events and propagated. + """ + + servlets = [ + synapse.rest.admin.register_servlets_for_client_rest_resource, + room.register_servlets, + login.register_servlets, + ] + + def prepare(self, reactor, clock, homeserver): + self.creator = self.register_user("creator", "test") + self.creator_tok = self.login("creator", "test") + + self.second_user_id = self.register_user("second", "test") + self.second_tok = self.login("second", "test") + + self.room_id = self.helper.create_room_as(self.creator, tok=self.creator_tok) + + def test_join_reason(self): + reason = "hello" + request, channel = self.make_request( + "POST", + "/_matrix/client/r0/rooms/{}/join".format(self.room_id), + content={"reason": reason}, + access_token=self.second_tok, + ) + self.render(request) + self.assertEqual(channel.code, 200, channel.result) + + self._check_for_reason(reason) + + def test_leave_reason(self): + self.helper.join(self.room_id, user=self.second_user_id, tok=self.second_tok) + + reason = "hello" + request, channel = self.make_request( + "POST", + "/_matrix/client/r0/rooms/{}/leave".format(self.room_id), + content={"reason": reason}, + access_token=self.second_tok, + ) + self.render(request) + self.assertEqual(channel.code, 200, channel.result) + + self._check_for_reason(reason) + + def test_kick_reason(self): + self.helper.join(self.room_id, user=self.second_user_id, tok=self.second_tok) + + reason = "hello" + request, channel = self.make_request( + "POST", + "/_matrix/client/r0/rooms/{}/kick".format(self.room_id), + content={"reason": reason, "user_id": self.second_user_id}, + access_token=self.second_tok, + ) + self.render(request) + self.assertEqual(channel.code, 200, channel.result) + + self._check_for_reason(reason) + + def test_ban_reason(self): + self.helper.join(self.room_id, user=self.second_user_id, tok=self.second_tok) + + reason = "hello" + request, channel = self.make_request( + "POST", + "/_matrix/client/r0/rooms/{}/ban".format(self.room_id), + content={"reason": reason, "user_id": self.second_user_id}, + access_token=self.creator_tok, + ) + self.render(request) + self.assertEqual(channel.code, 200, channel.result) + + self._check_for_reason(reason) + + def test_unban_reason(self): + reason = "hello" + request, channel = self.make_request( + "POST", + "/_matrix/client/r0/rooms/{}/unban".format(self.room_id), + content={"reason": reason, "user_id": self.second_user_id}, + access_token=self.creator_tok, + ) + self.render(request) + self.assertEqual(channel.code, 200, channel.result) + + self._check_for_reason(reason) + + def test_invite_reason(self): + reason = "hello" + request, channel = self.make_request( + "POST", + "/_matrix/client/r0/rooms/{}/invite".format(self.room_id), + content={"reason": reason, "user_id": self.second_user_id}, + access_token=self.creator_tok, + ) + self.render(request) + self.assertEqual(channel.code, 200, channel.result) + + self._check_for_reason(reason) + + def test_reject_invite_reason(self): + self.helper.invite( + self.room_id, + src=self.creator, + targ=self.second_user_id, + tok=self.creator_tok, + ) + + reason = "hello" + request, channel = self.make_request( + "POST", + "/_matrix/client/r0/rooms/{}/leave".format(self.room_id), + content={"reason": reason}, + access_token=self.second_tok, + ) + self.render(request) + self.assertEqual(channel.code, 200, channel.result) + + self._check_for_reason(reason) + + def _check_for_reason(self, reason): + request, channel = self.make_request( + "GET", + "/_matrix/client/r0/rooms/{}/state/m.room.member/{}".format( + self.room_id, self.second_user_id + ), + access_token=self.creator_tok, + ) + self.render(request) + self.assertEqual(channel.code, 200, channel.result) + + event_content = channel.json_body + + self.assertEqual(event_content.get("reason"), reason, channel.result) -- cgit 1.4.1 From 54dd5dc12b0ac5c48303144c4a73ce3822209488 Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Tue, 3 Dec 2019 19:19:45 +0000 Subject: Add ephemeral messages support (MSC2228) (#6409) Implement part [MSC2228](https://github.com/matrix-org/matrix-doc/pull/2228). The parts that differ are: * the feature is hidden behind a configuration flag (`enable_ephemeral_messages`) * self-destruction doesn't happen for state events * only implement support for the `m.self_destruct_after` field (not the `m.self_destruct` one) * doesn't send synthetic redactions to clients because for this specific case we consider the clients to be able to destroy an event themselves, instead we just censor it (by pruning its JSON) in the database --- changelog.d/6409.feature | 1 + synapse/api/constants.py | 4 + synapse/config/server.py | 2 + synapse/handlers/federation.py | 8 ++ synapse/handlers/message.py | 123 +++++++++++++++++++- synapse/storage/data_stores/main/events.py | 126 ++++++++++++++++++++- .../main/schema/delta/56/event_expiry.sql | 21 ++++ tests/rest/client/test_ephemeral_message.py | 101 +++++++++++++++++ 8 files changed, 379 insertions(+), 7 deletions(-) create mode 100644 changelog.d/6409.feature create mode 100644 synapse/storage/data_stores/main/schema/delta/56/event_expiry.sql create mode 100644 tests/rest/client/test_ephemeral_message.py (limited to 'tests') diff --git a/changelog.d/6409.feature b/changelog.d/6409.feature new file mode 100644 index 0000000000..653ff5a5ad --- /dev/null +++ b/changelog.d/6409.feature @@ -0,0 +1 @@ +Add ephemeral messages support by partially implementing [MSC2228](https://github.com/matrix-org/matrix-doc/pull/2228). diff --git a/synapse/api/constants.py b/synapse/api/constants.py index e3f086f1c3..69cef369a5 100644 --- a/synapse/api/constants.py +++ b/synapse/api/constants.py @@ -147,3 +147,7 @@ class EventContentFields(object): # Labels for the event, cf https://github.com/matrix-org/matrix-doc/pull/2326 LABELS = "org.matrix.labels" + + # Timestamp to delete the event after + # cf https://github.com/matrix-org/matrix-doc/pull/2228 + SELF_DESTRUCT_AFTER = "org.matrix.self_destruct_after" diff --git a/synapse/config/server.py b/synapse/config/server.py index 7a9d711669..837fbe1582 100644 --- a/synapse/config/server.py +++ b/synapse/config/server.py @@ -490,6 +490,8 @@ class ServerConfig(Config): "cleanup_extremities_with_dummy_events", True ) + self.enable_ephemeral_messages = config.get("enable_ephemeral_messages", False) + def has_tls_listener(self) -> bool: return any(l["tls"] for l in self.listeners) diff --git a/synapse/handlers/federation.py b/synapse/handlers/federation.py index d3267734f7..d9d0cd9eef 100644 --- a/synapse/handlers/federation.py +++ b/synapse/handlers/federation.py @@ -121,6 +121,7 @@ class FederationHandler(BaseHandler): self.pusher_pool = hs.get_pusherpool() self.spam_checker = hs.get_spam_checker() self.event_creation_handler = hs.get_event_creation_handler() + self._message_handler = hs.get_message_handler() self._server_notices_mxid = hs.config.server_notices_mxid self.config = hs.config self.http_client = hs.get_simple_http_client() @@ -141,6 +142,8 @@ class FederationHandler(BaseHandler): self.third_party_event_rules = hs.get_third_party_event_rules() + self._ephemeral_messages_enabled = hs.config.enable_ephemeral_messages + @defer.inlineCallbacks def on_receive_pdu(self, origin, pdu, sent_to_us_directly=False): """ Process a PDU received via a federation /send/ transaction, or @@ -2715,6 +2718,11 @@ class FederationHandler(BaseHandler): event_and_contexts, backfilled=backfilled ) + if self._ephemeral_messages_enabled: + for (event, context) in event_and_contexts: + # If there's an expiry timestamp on the event, schedule its expiry. + self._message_handler.maybe_schedule_expiry(event) + if not backfilled: # Never notify for backfilled events for event, _ in event_and_contexts: yield self._notify_persisted_event(event, max_stream_id) diff --git a/synapse/handlers/message.py b/synapse/handlers/message.py index 3b0156f516..4f53a5f5dc 100644 --- a/synapse/handlers/message.py +++ b/synapse/handlers/message.py @@ -15,6 +15,7 @@ # See the License for the specific language governing permissions and # limitations under the License. import logging +from typing import Optional from six import iteritems, itervalues, string_types @@ -22,9 +23,16 @@ from canonicaljson import encode_canonical_json, json from twisted.internet import defer from twisted.internet.defer import succeed +from twisted.internet.interfaces import IDelayedCall from synapse import event_auth -from synapse.api.constants import EventTypes, Membership, RelationTypes, UserTypes +from synapse.api.constants import ( + EventContentFields, + EventTypes, + Membership, + RelationTypes, + UserTypes, +) from synapse.api.errors import ( AuthError, Codes, @@ -62,6 +70,17 @@ class MessageHandler(object): self.storage = hs.get_storage() self.state_store = self.storage.state self._event_serializer = hs.get_event_client_serializer() + self._ephemeral_events_enabled = hs.config.enable_ephemeral_messages + self._is_worker_app = bool(hs.config.worker_app) + + # The scheduled call to self._expire_event. None if no call is currently + # scheduled. + self._scheduled_expiry = None # type: Optional[IDelayedCall] + + if not hs.config.worker_app: + run_as_background_process( + "_schedule_next_expiry", self._schedule_next_expiry + ) @defer.inlineCallbacks def get_room_data( @@ -225,6 +244,100 @@ class MessageHandler(object): for user_id, profile in iteritems(users_with_profile) } + def maybe_schedule_expiry(self, event): + """Schedule the expiry of an event if there's not already one scheduled, + or if the one running is for an event that will expire after the provided + timestamp. + + This function needs to invalidate the event cache, which is only possible on + the master process, and therefore needs to be run on there. + + Args: + event (EventBase): The event to schedule the expiry of. + """ + assert not self._is_worker_app + + expiry_ts = event.content.get(EventContentFields.SELF_DESTRUCT_AFTER) + if not isinstance(expiry_ts, int) or event.is_state(): + return + + # _schedule_expiry_for_event won't actually schedule anything if there's already + # a task scheduled for a timestamp that's sooner than the provided one. + self._schedule_expiry_for_event(event.event_id, expiry_ts) + + @defer.inlineCallbacks + def _schedule_next_expiry(self): + """Retrieve the ID and the expiry timestamp of the next event to be expired, + and schedule an expiry task for it. + + If there's no event left to expire, set _expiry_scheduled to None so that a + future call to save_expiry_ts can schedule a new expiry task. + """ + # Try to get the expiry timestamp of the next event to expire. + res = yield self.store.get_next_event_to_expire() + if res: + event_id, expiry_ts = res + self._schedule_expiry_for_event(event_id, expiry_ts) + + def _schedule_expiry_for_event(self, event_id, expiry_ts): + """Schedule an expiry task for the provided event if there's not already one + scheduled at a timestamp that's sooner than the provided one. + + Args: + event_id (str): The ID of the event to expire. + expiry_ts (int): The timestamp at which to expire the event. + """ + if self._scheduled_expiry: + # If the provided timestamp refers to a time before the scheduled time of the + # next expiry task, cancel that task and reschedule it for this timestamp. + next_scheduled_expiry_ts = self._scheduled_expiry.getTime() * 1000 + if expiry_ts < next_scheduled_expiry_ts: + self._scheduled_expiry.cancel() + else: + return + + # Figure out how many seconds we need to wait before expiring the event. + now_ms = self.clock.time_msec() + delay = (expiry_ts - now_ms) / 1000 + + # callLater doesn't support negative delays, so trim the delay to 0 if we're + # in that case. + if delay < 0: + delay = 0 + + logger.info("Scheduling expiry for event %s in %.3fs", event_id, delay) + + self._scheduled_expiry = self.clock.call_later( + delay, + run_as_background_process, + "_expire_event", + self._expire_event, + event_id, + ) + + @defer.inlineCallbacks + def _expire_event(self, event_id): + """Retrieve and expire an event that needs to be expired from the database. + + If the event doesn't exist in the database, log it and delete the expiry date + from the database (so that we don't try to expire it again). + """ + assert self._ephemeral_events_enabled + + self._scheduled_expiry = None + + logger.info("Expiring event %s", event_id) + + try: + # Expire the event if we know about it. This function also deletes the expiry + # date from the database in the same database transaction. + yield self.store.expire_event(event_id) + except Exception as e: + logger.error("Could not expire event %s: %r", event_id, e) + + # Schedule the expiry of the next event to expire. + yield self._schedule_next_expiry() + # The duration (in ms) after which rooms should be removed # `_rooms_to_exclude_from_dummy_event_insertion` (with the effect that we will try @@ -295,6 +408,10 @@ class EventCreationHandler(object): 5 * 60 * 1000, ) + self._message_handler = hs.get_message_handler() + + self._ephemeral_events_enabled = hs.config.enable_ephemeral_messages + @defer.inlineCallbacks def create_event( self, @@ -877,6 +994,10 @@ class EventCreationHandler(object): event, context=context ) + if self._ephemeral_events_enabled: + # If there's an expiry timestamp on the event, schedule its expiry. + self._message_handler.maybe_schedule_expiry(event) + yield self.pusher_pool.on_new_notifications(event_stream_id, max_stream_id) def _notify(): diff --git a/synapse/storage/data_stores/main/events.py b/synapse/storage/data_stores/main/events.py index 2737a1d3ae..79c91fe284 100644 --- a/synapse/storage/data_stores/main/events.py +++ b/synapse/storage/data_stores/main/events.py @@ -130,6 +130,8 @@ class EventsStore( if self.hs.config.redaction_retention_period is not None: hs.get_clock().looping_call(_censor_redactions, 5 * 60 * 1000) + self._ephemeral_messages_enabled = hs.config.enable_ephemeral_messages + @defer.inlineCallbacks def _read_forward_extremities(self): def fetch(txn): @@ -940,6 +942,12 @@ class EventsStore( txn, event.event_id, labels, event.room_id, event.depth ) + if self._ephemeral_messages_enabled: + # If there's an expiry timestamp on the event, store it. + expiry_ts = event.content.get(EventContentFields.SELF_DESTRUCT_AFTER) + if isinstance(expiry_ts, int) and not event.is_state(): + self._insert_event_expiry_txn(txn, event.event_id, expiry_ts) + # Insert into the room_memberships table. self._store_room_members_txn( txn, @@ -1101,12 +1109,7 @@ class EventsStore( def _update_censor_txn(txn): for redaction_id, event_id, pruned_json in updates: if pruned_json: - self._simple_update_one_txn( - txn, - table="event_json", - keyvalues={"event_id": event_id}, - updatevalues={"json": pruned_json}, - ) + self._censor_event_txn(txn, event_id, pruned_json) self._simple_update_one_txn( txn, @@ -1117,6 +1120,22 @@ class EventsStore( yield self.runInteraction("_update_censor_txn", _update_censor_txn) + def _censor_event_txn(self, txn, event_id, pruned_json): + """Censor an event by replacing its JSON in the event_json table with the + provided pruned JSON. + + Args: + txn (LoggingTransaction): The database transaction. + event_id (str): The ID of the event to censor. + pruned_json (str): The pruned JSON + """ + self._simple_update_one_txn( + txn, + table="event_json", + keyvalues={"event_id": event_id}, + updatevalues={"json": pruned_json}, + ) + @defer.inlineCallbacks def count_daily_messages(self): """ @@ -1957,6 +1976,101 @@ class EventsStore( ], ) + def _insert_event_expiry_txn(self, txn, event_id, expiry_ts): + """Save the expiry timestamp associated with a given event ID. + + Args: + txn (LoggingTransaction): The database transaction to use. + event_id (str): The event ID the expiry timestamp is associated with. + expiry_ts (int): The timestamp at which to expire (delete) the event. + """ + return self._simple_insert_txn( + txn=txn, + table="event_expiry", + values={"event_id": event_id, "expiry_ts": expiry_ts}, + ) + + @defer.inlineCallbacks + def expire_event(self, event_id): + """Retrieve and expire an event that has expired, and delete its associated + expiry timestamp. If the event can't be retrieved, delete its associated + timestamp so we don't try to expire it again in the future. + + Args: + event_id (str): The ID of the event to delete. + """ + # Try to retrieve the event's content from the database or the event cache. + event = yield self.get_event(event_id) + + def delete_expired_event_txn(txn): + # Delete the expiry timestamp associated with this event from the database. + self._delete_event_expiry_txn(txn, event_id) + + if not event: + # If we can't find the event, log a warning and delete the expiry date + # from the database so that we don't try to expire it again in the + # future. + logger.warning( + "Can't expire event %s because we don't have it.", event_id + ) + return + + # Prune the event's dict then convert it to JSON. + pruned_json = encode_json(prune_event_dict(event.get_dict())) + + # Update the event_json table to replace the event's JSON with the pruned + # JSON. + self._censor_event_txn(txn, event.event_id, pruned_json) + + # We need to invalidate the event cache entry for this event because we + # changed its content in the database. We can't call + # self._invalidate_cache_and_stream because self.get_event_cache isn't of the + # right type. + txn.call_after(self._get_event_cache.invalidate, (event.event_id,)) + # Send that invalidation to replication so that other workers also invalidate + # the event cache. + self._send_invalidation_to_replication( + txn, "_get_event_cache", (event.event_id,) + ) + + yield self.runInteraction("delete_expired_event", delete_expired_event_txn) + + def _delete_event_expiry_txn(self, txn, event_id): + """Delete the expiry timestamp associated with an event ID without deleting the + actual event. + + Args: + txn (LoggingTransaction): The transaction to use to perform the deletion. + event_id (str): The event ID to delete the associated expiry timestamp of. + """ + return self._simple_delete_txn( + txn=txn, table="event_expiry", keyvalues={"event_id": event_id} + ) + + def get_next_event_to_expire(self): + """Retrieve the entry with the lowest expiry timestamp in the event_expiry + table, or None if there's no more event to expire. + + Returns: Deferred[Optional[Tuple[str, int]]] + A tuple containing the event ID as its first element and an expiry timestamp + as its second one, if there's at least one row in the event_expiry table. + None otherwise. + """ + + def get_next_event_to_expire_txn(txn): + txn.execute( + """ + SELECT event_id, expiry_ts FROM event_expiry + ORDER BY expiry_ts ASC LIMIT 1 + """ + ) + + return txn.fetchone() + + return self.runInteraction( + desc="get_next_event_to_expire", func=get_next_event_to_expire_txn + ) + AllNewEventsResult = namedtuple( "AllNewEventsResult", diff --git a/synapse/storage/data_stores/main/schema/delta/56/event_expiry.sql b/synapse/storage/data_stores/main/schema/delta/56/event_expiry.sql new file mode 100644 index 0000000000..81a36a8b1d --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/56/event_expiry.sql @@ -0,0 +1,21 @@ +/* Copyright 2019 The Matrix.org Foundation C.I.C. + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE TABLE IF NOT EXISTS event_expiry ( + event_id TEXT PRIMARY KEY, + expiry_ts BIGINT NOT NULL +); + +CREATE INDEX event_expiry_expiry_ts_idx ON event_expiry(expiry_ts); diff --git a/tests/rest/client/test_ephemeral_message.py b/tests/rest/client/test_ephemeral_message.py new file mode 100644 index 0000000000..5e9c07ebf3 --- /dev/null +++ b/tests/rest/client/test_ephemeral_message.py @@ -0,0 +1,101 @@ +# -*- coding: utf-8 -*- +# Copyright 2019 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +from synapse.api.constants import EventContentFields, EventTypes +from synapse.rest import admin +from synapse.rest.client.v1 import room + +from tests import unittest + + +class EphemeralMessageTestCase(unittest.HomeserverTestCase): + + user_id = "@user:test" + + servlets = [ + admin.register_servlets, + room.register_servlets, + ] + + def make_homeserver(self, reactor, clock): + config = self.default_config() + + config["enable_ephemeral_messages"] = True + + self.hs = self.setup_test_homeserver(config=config) + return self.hs + + def prepare(self, reactor, clock, homeserver): + self.room_id = self.helper.create_room_as(self.user_id) + + def test_message_expiry_no_delay(self): + """Tests that sending a message sent with a m.self_destruct_after field set to the + past results in that event being deleted right away. + """ + # Send a message in the room that has expired. From here, the reactor clock is + # at 200ms, so 0 is in the past, and even if that wasn't the case and the clock + # is at 0ms the code path is the same if the event's expiry timestamp is the + # current timestamp. + res = self.helper.send_event( + room_id=self.room_id, + type=EventTypes.Message, + content={ + "msgtype": "m.text", + "body": "hello", + EventContentFields.SELF_DESTRUCT_AFTER: 0, + }, + ) + event_id = res["event_id"] + + # Check that we can't retrieve the content of the event. + event_content = self.get_event(self.room_id, event_id)["content"] + self.assertFalse(bool(event_content), event_content) + + def test_message_expiry_delay(self): + """Tests that sending a message with a m.self_destruct_after field set to the + future results in that event not being deleted right away, but advancing the + clock to after that expiry timestamp causes the event to be deleted. + """ + # Send a message in the room that'll expire in 1s. + res = self.helper.send_event( + room_id=self.room_id, + type=EventTypes.Message, + content={ + "msgtype": "m.text", + "body": "hello", + EventContentFields.SELF_DESTRUCT_AFTER: self.clock.time_msec() + 1000, + }, + ) + event_id = res["event_id"] + + # Check that we can retrieve the content of the event before it has expired. + event_content = self.get_event(self.room_id, event_id)["content"] + self.assertTrue(bool(event_content), event_content) + + # Advance the clock to after the deletion. + self.reactor.advance(1) + + # Check that we can't retrieve the content of the event anymore. + event_content = self.get_event(self.room_id, event_id)["content"] + self.assertFalse(bool(event_content), event_content) + + def get_event(self, room_id, event_id, expected_code=200): + url = "/_matrix/client/r0/rooms/%s/event/%s" % (room_id, event_id) + + request, channel = self.make_request("GET", url) + self.render(request) + + self.assertEqual(channel.code, expected_code, channel.result) + + return channel.json_body -- cgit 1.4.1 From cb0aeb147e3b3defc27866ad0e4982e63600a7ee Mon Sep 17 00:00:00 2001 From: Neil Johnson Date: Wed, 4 Dec 2019 09:46:16 +0000 Subject: privacy by default for room dir (#6355) Ensure that the the default settings for the room directory are that the it is hidden from public view by default. --- UPGRADE.rst | 17 ++++++++++ changelog.d/6354.feature | 1 + docs/sample_config.yaml | 13 ++++---- synapse/config/server.py | 26 +++++++++------- tests/federation/transport/test_server.py | 52 +++++++++++++++++++++++++++++++ 5 files changed, 91 insertions(+), 18 deletions(-) create mode 100644 changelog.d/6354.feature create mode 100644 tests/federation/transport/test_server.py (limited to 'tests') diff --git a/UPGRADE.rst b/UPGRADE.rst index 5ebf16a73e..d9020f2663 100644 --- a/UPGRADE.rst +++ b/UPGRADE.rst @@ -75,6 +75,23 @@ for example: wget https://packages.matrix.org/debian/pool/main/m/matrix-synapse-py3/matrix-synapse-py3_1.3.0+stretch1_amd64.deb dpkg -i matrix-synapse-py3_1.3.0+stretch1_amd64.deb +Upgrading to v1.7.0 +=================== + +In an attempt to configure Synapse in a privacy preserving way, the default +behaviours of ``allow_public_rooms_without_auth`` and +``allow_public_rooms_over_federation`` have been inverted. This means that by +default, only authenticated users querying the Client/Server API will be able +to query the room directory, and relatedly that the server will not share +room directory information with other servers over federation. + +If your installation does not explicitly set these settings one way or the other +and you want either setting to be ``true`` then it will necessary to update +your homeserver configuration file accordingly. + +For more details on the surrounding context see our `explainer +`_. + Upgrading to v1.5.0 =================== diff --git a/changelog.d/6354.feature b/changelog.d/6354.feature new file mode 100644 index 0000000000..fed9db884b --- /dev/null +++ b/changelog.d/6354.feature @@ -0,0 +1 @@ +Configure privacy preserving settings by default for the room directory. diff --git a/docs/sample_config.yaml b/docs/sample_config.yaml index c7391f0c48..10664ae8f7 100644 --- a/docs/sample_config.yaml +++ b/docs/sample_config.yaml @@ -54,15 +54,16 @@ pid_file: DATADIR/homeserver.pid # #require_auth_for_profile_requests: true -# If set to 'false', requires authentication to access the server's public rooms -# directory through the client API. Defaults to 'true'. +# If set to 'true', removes the need for authentication to access the server's +# public rooms directory through the client API, meaning that anyone can +# query the room directory. Defaults to 'false'. # -#allow_public_rooms_without_auth: false +#allow_public_rooms_without_auth: true -# If set to 'false', forbids any other homeserver to fetch the server's public -# rooms directory via federation. Defaults to 'true'. +# If set to 'true', allows any other homeserver to fetch the server's public +# rooms directory via federation. Defaults to 'false'. # -#allow_public_rooms_over_federation: false +#allow_public_rooms_over_federation: true # The default room version for newly created rooms. # diff --git a/synapse/config/server.py b/synapse/config/server.py index 837fbe1582..a4bef00936 100644 --- a/synapse/config/server.py +++ b/synapse/config/server.py @@ -118,15 +118,16 @@ class ServerConfig(Config): self.allow_public_rooms_without_auth = False self.allow_public_rooms_over_federation = False else: - # If set to 'False', requires authentication to access the server's public - # rooms directory through the client API. Defaults to 'True'. + # If set to 'true', removes the need for authentication to access the server's + # public rooms directory through the client API, meaning that anyone can + # query the room directory. Defaults to 'false'. self.allow_public_rooms_without_auth = config.get( - "allow_public_rooms_without_auth", True + "allow_public_rooms_without_auth", False ) - # If set to 'False', forbids any other homeserver to fetch the server's public - # rooms directory via federation. Defaults to 'True'. + # If set to 'true', allows any other homeserver to fetch the server's public + # rooms directory via federation. Defaults to 'false'. self.allow_public_rooms_over_federation = config.get( - "allow_public_rooms_over_federation", True + "allow_public_rooms_over_federation", False ) default_room_version = config.get("default_room_version", DEFAULT_ROOM_VERSION) @@ -620,15 +621,16 @@ class ServerConfig(Config): # #require_auth_for_profile_requests: true - # If set to 'false', requires authentication to access the server's public rooms - # directory through the client API. Defaults to 'true'. + # If set to 'true', removes the need for authentication to access the server's + # public rooms directory through the client API, meaning that anyone can + # query the room directory. Defaults to 'false'. # - #allow_public_rooms_without_auth: false + #allow_public_rooms_without_auth: true - # If set to 'false', forbids any other homeserver to fetch the server's public - # rooms directory via federation. Defaults to 'true'. + # If set to 'true', allows any other homeserver to fetch the server's public + # rooms directory via federation. Defaults to 'false'. # - #allow_public_rooms_over_federation: false + #allow_public_rooms_over_federation: true # The default room version for newly created rooms. # diff --git a/tests/federation/transport/test_server.py b/tests/federation/transport/test_server.py new file mode 100644 index 0000000000..27d83bb7d9 --- /dev/null +++ b/tests/federation/transport/test_server.py @@ -0,0 +1,52 @@ +# -*- coding: utf-8 -*- +# Copyright 2019 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + + +from twisted.internet import defer + +from synapse.config.ratelimiting import FederationRateLimitConfig +from synapse.federation.transport import server +from synapse.util.ratelimitutils import FederationRateLimiter + +from tests import unittest +from tests.unittest import override_config + + +class RoomDirectoryFederationTests(unittest.HomeserverTestCase): + def prepare(self, reactor, clock, homeserver): + class Authenticator(object): + def authenticate_request(self, request, content): + return defer.succeed("otherserver.nottld") + + ratelimiter = FederationRateLimiter(clock, FederationRateLimitConfig()) + server.register_servlets( + homeserver, self.resource, Authenticator(), ratelimiter + ) + + @override_config({"allow_public_rooms_over_federation": False}) + def test_blocked_public_room_list_over_federation(self): + request, channel = self.make_request( + "GET", "/_matrix/federation/v1/publicRooms" + ) + self.render(request) + self.assertEquals(403, channel.code) + + @override_config({"allow_public_rooms_over_federation": True}) + def test_open_public_room_list_over_federation(self): + request, channel = self.make_request( + "GET", "/_matrix/federation/v1/publicRooms" + ) + self.render(request) + self.assertEquals(200, channel.code) -- cgit 1.4.1 From 65c6aee621fecff1c6a863d6b910c973196ad6bc Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Wed, 4 Dec 2019 14:36:39 +0000 Subject: Un-remove room purge test --- tests/rest/client/v1/test_rooms.py | 72 ++++++++++++++++++++++++++++++++++++++ 1 file changed, 72 insertions(+) (limited to 'tests') diff --git a/tests/rest/client/v1/test_rooms.py b/tests/rest/client/v1/test_rooms.py index 4095e63aef..1ca7fa742f 100644 --- a/tests/rest/client/v1/test_rooms.py +++ b/tests/rest/client/v1/test_rooms.py @@ -815,6 +815,78 @@ class RoomMessageListTestCase(RoomBase): self.assertTrue("chunk" in channel.json_body) self.assertTrue("end" in channel.json_body) + def test_room_messages_purge(self): + store = self.hs.get_datastore() + pagination_handler = self.hs.get_pagination_handler() + + # Send a first message in the room, which will be removed by the purge. + first_event_id = self.helper.send(self.room_id, "message 1")["event_id"] + first_token = self.get_success( + store.get_topological_token_for_event(first_event_id) + ) + + # Send a second message in the room, which won't be removed, and which we'll + # use as the marker to purge events before. + second_event_id = self.helper.send(self.room_id, "message 2")["event_id"] + second_token = self.get_success( + store.get_topological_token_for_event(second_event_id) + ) + + # Send a third event in the room to ensure we don't fall under any edge case + # due to our marker being the latest forward extremity in the room. + self.helper.send(self.room_id, "message 3") + + # Check that we get the first and second message when querying /messages. + request, channel = self.make_request( + "GET", + "/rooms/%s/messages?access_token=x&from=%s&dir=b&filter=%s" + % (self.room_id, second_token, json.dumps({"types": [EventTypes.Message]})), + ) + self.render(request) + self.assertEqual(channel.code, 200, channel.json_body) + + chunk = channel.json_body["chunk"] + self.assertEqual(len(chunk), 2, [event["content"] for event in chunk]) + + # Purge every event before the second event. + purge_id = random_string(16) + pagination_handler._purges_by_id[purge_id] = PurgeStatus() + self.get_success( + pagination_handler._purge_history( + purge_id=purge_id, + room_id=self.room_id, + token=second_token, + delete_local_events=True, + ) + ) + + # Check that we only get the second message through /message now that the first + # has been purged. + request, channel = self.make_request( + "GET", + "/rooms/%s/messages?access_token=x&from=%s&dir=b&filter=%s" + % (self.room_id, second_token, json.dumps({"types": [EventTypes.Message]})), + ) + self.render(request) + self.assertEqual(channel.code, 200, channel.json_body) + + chunk = channel.json_body["chunk"] + self.assertEqual(len(chunk), 1, [event["content"] for event in chunk]) + + # Check that we get no event, but also no error, when querying /messages with + # the token that was pointing at the first event, because we don't have it + # anymore. + request, channel = self.make_request( + "GET", + "/rooms/%s/messages?access_token=x&from=%s&dir=b&filter=%s" + % (self.room_id, first_token, json.dumps({"types": [EventTypes.Message]})), + ) + self.render(request) + self.assertEqual(channel.code, 200, channel.json_body) + + chunk = channel.json_body["chunk"] + self.assertEqual(len(chunk), 0, [event["content"] for event in chunk]) + class RoomSearchTestCase(unittest.HomeserverTestCase): servlets = [ -- cgit 1.4.1 From ee86abb2d6e9c7d553858e814b4343bcf95af75a Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 4 Dec 2019 10:15:55 +0000 Subject: Remove underscore from SQLBaseStore functions --- scripts-dev/hash_history.py | 2 +- scripts/synapse_port_db | 18 +-- synapse/app/user_dir.py | 2 +- synapse/storage/_base.py | 142 ++++++++++----------- synapse/storage/background_updates.py | 14 +- synapse/storage/data_stores/main/__init__.py | 16 +-- synapse/storage/data_stores/main/account_data.py | 18 +-- synapse/storage/data_stores/main/appservice.py | 10 +- synapse/storage/data_stores/main/cache.py | 2 +- synapse/storage/data_stores/main/client_ips.py | 8 +- synapse/storage/data_stores/main/deviceinbox.py | 4 +- synapse/storage/data_stores/main/devices.py | 46 +++---- synapse/storage/data_stores/main/directory.py | 12 +- synapse/storage/data_stores/main/e2e_room_keys.py | 20 +-- .../storage/data_stores/main/end_to_end_keys.py | 20 +-- .../storage/data_stores/main/event_federation.py | 16 +-- .../storage/data_stores/main/event_push_actions.py | 12 +- synapse/storage/data_stores/main/events.py | 52 ++++---- .../storage/data_stores/main/events_bg_updates.py | 10 +- synapse/storage/data_stores/main/events_worker.py | 6 +- synapse/storage/data_stores/main/filtering.py | 2 +- synapse/storage/data_stores/main/group_server.py | 128 +++++++++---------- synapse/storage/data_stores/main/keys.py | 6 +- .../storage/data_stores/main/media_repository.py | 24 ++-- .../data_stores/main/monthly_active_users.py | 6 +- synapse/storage/data_stores/main/openid.py | 2 +- synapse/storage/data_stores/main/presence.py | 8 +- synapse/storage/data_stores/main/profile.py | 24 ++-- synapse/storage/data_stores/main/push_rule.py | 24 ++-- synapse/storage/data_stores/main/pusher.py | 26 ++-- synapse/storage/data_stores/main/receipts.py | 16 +-- synapse/storage/data_stores/main/registration.py | 92 ++++++------- synapse/storage/data_stores/main/rejections.py | 4 +- synapse/storage/data_stores/main/relations.py | 4 +- synapse/storage/data_stores/main/room.py | 34 ++--- synapse/storage/data_stores/main/roommember.py | 16 +-- synapse/storage/data_stores/main/search.py | 8 +- synapse/storage/data_stores/main/signatures.py | 2 +- synapse/storage/data_stores/main/state.py | 36 +++--- synapse/storage/data_stores/main/state_deltas.py | 2 +- synapse/storage/data_stores/main/stats.py | 28 ++-- synapse/storage/data_stores/main/stream.py | 14 +- synapse/storage/data_stores/main/tags.py | 6 +- synapse/storage/data_stores/main/transactions.py | 12 +- synapse/storage/data_stores/main/user_directory.py | 56 ++++---- .../storage/data_stores/main/user_erasure_store.py | 4 +- tests/handlers/test_stats.py | 30 ++--- tests/handlers/test_user_directory.py | 12 +- tests/rest/admin/test_admin.py | 2 +- tests/storage/test__base.py | 8 +- tests/storage/test_base.py | 18 +-- tests/storage/test_client_ips.py | 12 +- tests/storage/test_event_push_actions.py | 4 +- tests/storage/test_redaction.py | 4 +- tests/storage/test_roommember.py | 2 +- tests/unittest.py | 2 +- 56 files changed, 550 insertions(+), 558 deletions(-) (limited to 'tests') diff --git a/scripts-dev/hash_history.py b/scripts-dev/hash_history.py index d20f6db176..bf3862a386 100644 --- a/scripts-dev/hash_history.py +++ b/scripts-dev/hash_history.py @@ -27,7 +27,7 @@ class Store(object): "_store_pdu_reference_hash_txn" ] _store_prev_pdu_hash_txn = SignatureStore.__dict__["_store_prev_pdu_hash_txn"] - _simple_insert_txn = SQLBaseStore.__dict__["_simple_insert_txn"] + simple_insert_txn = SQLBaseStore.__dict__["simple_insert_txn"] store = Store() diff --git a/scripts/synapse_port_db b/scripts/synapse_port_db index f24b8ffe67..9dd1700ff0 100755 --- a/scripts/synapse_port_db +++ b/scripts/synapse_port_db @@ -221,7 +221,7 @@ class Porter(object): def setup_table(self, table): if table in APPEND_ONLY_TABLES: # It's safe to just carry on inserting. - row = yield self.postgres_store._simple_select_one( + row = yield self.postgres_store.simple_select_one( table="port_from_sqlite3", keyvalues={"table_name": table}, retcols=("forward_rowid", "backward_rowid"), @@ -236,7 +236,7 @@ class Porter(object): ) backward_chunk = 0 else: - yield self.postgres_store._simple_insert( + yield self.postgres_store.simple_insert( table="port_from_sqlite3", values={ "table_name": table, @@ -266,7 +266,7 @@ class Porter(object): yield self.postgres_store.execute(delete_all) - yield self.postgres_store._simple_insert( + yield self.postgres_store.simple_insert( table="port_from_sqlite3", values={"table_name": table, "forward_rowid": 1, "backward_rowid": 0}, ) @@ -320,7 +320,7 @@ class Porter(object): if table == "user_directory_stream_pos": # We need to make sure there is a single row, `(X, null), as that is # what synapse expects to be there. - yield self.postgres_store._simple_insert( + yield self.postgres_store.simple_insert( table=table, values={"stream_id": None} ) self.progress.update(table, table_size) # Mark table as done @@ -375,7 +375,7 @@ class Porter(object): def insert(txn): self.postgres_store.insert_many_txn(txn, table, headers[1:], rows) - self.postgres_store._simple_update_one_txn( + self.postgres_store.simple_update_one_txn( txn, table="port_from_sqlite3", keyvalues={"table_name": table}, @@ -452,7 +452,7 @@ class Porter(object): ], ) - self.postgres_store._simple_update_one_txn( + self.postgres_store.simple_update_one_txn( txn, table="port_from_sqlite3", keyvalues={"table_name": "event_search"}, @@ -591,11 +591,11 @@ class Porter(object): # Step 2. Get tables. self.progress.set_state("Fetching tables") - sqlite_tables = yield self.sqlite_store._simple_select_onecol( + sqlite_tables = yield self.sqlite_store.simple_select_onecol( table="sqlite_master", keyvalues={"type": "table"}, retcol="name" ) - postgres_tables = yield self.postgres_store._simple_select_onecol( + postgres_tables = yield self.postgres_store.simple_select_onecol( table="information_schema.tables", keyvalues={}, retcol="distinct table_name", @@ -722,7 +722,7 @@ class Porter(object): next_chunk = yield self.sqlite_store.execute(get_start_id) next_chunk = max(max_inserted_rowid + 1, next_chunk) - yield self.postgres_store._simple_insert( + yield self.postgres_store.simple_insert( table="port_from_sqlite3", values={ "table_name": "sent_transactions", diff --git a/synapse/app/user_dir.py b/synapse/app/user_dir.py index 6cb100319f..0fa2b50999 100644 --- a/synapse/app/user_dir.py +++ b/synapse/app/user_dir.py @@ -64,7 +64,7 @@ class UserDirectorySlaveStore( super(UserDirectorySlaveStore, self).__init__(db_conn, hs) events_max = self._stream_id_gen.get_current_token() - curr_state_delta_prefill, min_curr_state_delta_id = self._get_cache_dict( + curr_state_delta_prefill, min_curr_state_delta_id = self.get_cache_dict( db_conn, "current_state_delta_stream", entity_column="room_id", diff --git a/synapse/storage/_base.py b/synapse/storage/_base.py index 1ed89d9f2a..9205e550bb 100644 --- a/synapse/storage/_base.py +++ b/synapse/storage/_base.py @@ -262,7 +262,7 @@ class SQLBaseStore(object): If the background updates have not completed, wait 15 sec and check again. """ - updates = yield self._simple_select_list( + updates = yield self.simple_select_list( "background_updates", keyvalues=None, retcols=["update_name"], @@ -307,7 +307,7 @@ class SQLBaseStore(object): self._clock.looping_call(loop, 10000) - def _new_transaction( + def new_transaction( self, conn, desc, after_callbacks, exception_callbacks, func, *args, **kwargs ): start = monotonic_time() @@ -444,7 +444,7 @@ class SQLBaseStore(object): try: result = yield self.runWithConnection( - self._new_transaction, + self.new_transaction, desc, after_callbacks, exception_callbacks, @@ -516,7 +516,7 @@ class SQLBaseStore(object): results = list(dict(zip(col_headers, row)) for row in cursor) return results - def _execute(self, desc, decoder, query, *args): + def execute(self, desc, decoder, query, *args): """Runs a single query for a result set. Args: @@ -541,7 +541,7 @@ class SQLBaseStore(object): # no complex WHERE clauses, just a dict of values for columns. @defer.inlineCallbacks - def _simple_insert(self, table, values, or_ignore=False, desc="_simple_insert"): + def simple_insert(self, table, values, or_ignore=False, desc="simple_insert"): """Executes an INSERT query on the named table. Args: @@ -557,7 +557,7 @@ class SQLBaseStore(object): `or_ignore` is True """ try: - yield self.runInteraction(desc, self._simple_insert_txn, table, values) + yield self.runInteraction(desc, self.simple_insert_txn, table, values) except self.database_engine.module.IntegrityError: # We have to do or_ignore flag at this layer, since we can't reuse # a cursor after we receive an error from the db. @@ -567,7 +567,7 @@ class SQLBaseStore(object): return True @staticmethod - def _simple_insert_txn(txn, table, values): + def simple_insert_txn(txn, table, values): keys, vals = zip(*values.items()) sql = "INSERT INTO %s (%s) VALUES(%s)" % ( @@ -578,11 +578,11 @@ class SQLBaseStore(object): txn.execute(sql, vals) - def _simple_insert_many(self, table, values, desc): - return self.runInteraction(desc, self._simple_insert_many_txn, table, values) + def simple_insert_many(self, table, values, desc): + return self.runInteraction(desc, self.simple_insert_many_txn, table, values) @staticmethod - def _simple_insert_many_txn(txn, table, values): + def simple_insert_many_txn(txn, table, values): if not values: return @@ -611,13 +611,13 @@ class SQLBaseStore(object): txn.executemany(sql, vals) @defer.inlineCallbacks - def _simple_upsert( + def simple_upsert( self, table, keyvalues, values, insertion_values={}, - desc="_simple_upsert", + desc="simple_upsert", lock=True, ): """ @@ -649,7 +649,7 @@ class SQLBaseStore(object): try: result = yield self.runInteraction( desc, - self._simple_upsert_txn, + self.simple_upsert_txn, table, keyvalues, values, @@ -669,7 +669,7 @@ class SQLBaseStore(object): "IntegrityError when upserting into %s; retrying: %s", table, e ) - def _simple_upsert_txn( + def simple_upsert_txn( self, txn, table, keyvalues, values, insertion_values={}, lock=True ): """ @@ -693,11 +693,11 @@ class SQLBaseStore(object): self.database_engine.can_native_upsert and table not in self._unsafe_to_upsert_tables ): - return self._simple_upsert_txn_native_upsert( + return self.simple_upsert_txn_native_upsert( txn, table, keyvalues, values, insertion_values=insertion_values ) else: - return self._simple_upsert_txn_emulated( + return self.simple_upsert_txn_emulated( txn, table, keyvalues, @@ -706,7 +706,7 @@ class SQLBaseStore(object): lock=lock, ) - def _simple_upsert_txn_emulated( + def simple_upsert_txn_emulated( self, txn, table, keyvalues, values, insertion_values={}, lock=True ): """ @@ -775,7 +775,7 @@ class SQLBaseStore(object): # successfully inserted return True - def _simple_upsert_txn_native_upsert( + def simple_upsert_txn_native_upsert( self, txn, table, keyvalues, values, insertion_values={} ): """ @@ -809,7 +809,7 @@ class SQLBaseStore(object): ) txn.execute(sql, list(allvalues.values())) - def _simple_upsert_many_txn( + def simple_upsert_many_txn( self, txn, table, key_names, key_values, value_names, value_values ): """ @@ -829,15 +829,15 @@ class SQLBaseStore(object): self.database_engine.can_native_upsert and table not in self._unsafe_to_upsert_tables ): - return self._simple_upsert_many_txn_native_upsert( + return self.simple_upsert_many_txn_native_upsert( txn, table, key_names, key_values, value_names, value_values ) else: - return self._simple_upsert_many_txn_emulated( + return self.simple_upsert_many_txn_emulated( txn, table, key_names, key_values, value_names, value_values ) - def _simple_upsert_many_txn_emulated( + def simple_upsert_many_txn_emulated( self, txn, table, key_names, key_values, value_names, value_values ): """ @@ -862,9 +862,9 @@ class SQLBaseStore(object): _keys = {x: y for x, y in zip(key_names, keyv)} _vals = {x: y for x, y in zip(value_names, valv)} - self._simple_upsert_txn_emulated(txn, table, _keys, _vals) + self.simple_upsert_txn_emulated(txn, table, _keys, _vals) - def _simple_upsert_many_txn_native_upsert( + def simple_upsert_many_txn_native_upsert( self, txn, table, key_names, key_values, value_names, value_values ): """ @@ -909,8 +909,8 @@ class SQLBaseStore(object): return txn.execute_batch(sql, args) - def _simple_select_one( - self, table, keyvalues, retcols, allow_none=False, desc="_simple_select_one" + def simple_select_one( + self, table, keyvalues, retcols, allow_none=False, desc="simple_select_one" ): """Executes a SELECT query on the named table, which is expected to return a single row, returning multiple columns from it. @@ -924,16 +924,16 @@ class SQLBaseStore(object): statement returns no rows """ return self.runInteraction( - desc, self._simple_select_one_txn, table, keyvalues, retcols, allow_none + desc, self.simple_select_one_txn, table, keyvalues, retcols, allow_none ) - def _simple_select_one_onecol( + def simple_select_one_onecol( self, table, keyvalues, retcol, allow_none=False, - desc="_simple_select_one_onecol", + desc="simple_select_one_onecol", ): """Executes a SELECT query on the named table, which is expected to return a single row, returning a single column from it. @@ -945,7 +945,7 @@ class SQLBaseStore(object): """ return self.runInteraction( desc, - self._simple_select_one_onecol_txn, + self.simple_select_one_onecol_txn, table, keyvalues, retcol, @@ -953,10 +953,10 @@ class SQLBaseStore(object): ) @classmethod - def _simple_select_one_onecol_txn( + def simple_select_one_onecol_txn( cls, txn, table, keyvalues, retcol, allow_none=False ): - ret = cls._simple_select_onecol_txn( + ret = cls.simple_select_onecol_txn( txn, table=table, keyvalues=keyvalues, retcol=retcol ) @@ -969,7 +969,7 @@ class SQLBaseStore(object): raise StoreError(404, "No row found") @staticmethod - def _simple_select_onecol_txn(txn, table, keyvalues, retcol): + def simple_select_onecol_txn(txn, table, keyvalues, retcol): sql = ("SELECT %(retcol)s FROM %(table)s") % {"retcol": retcol, "table": table} if keyvalues: @@ -980,8 +980,8 @@ class SQLBaseStore(object): return [r[0] for r in txn] - def _simple_select_onecol( - self, table, keyvalues, retcol, desc="_simple_select_onecol" + def simple_select_onecol( + self, table, keyvalues, retcol, desc="simple_select_onecol" ): """Executes a SELECT query on the named table, which returns a list comprising of the values of the named column from the selected rows. @@ -995,12 +995,10 @@ class SQLBaseStore(object): Deferred: Results in a list """ return self.runInteraction( - desc, self._simple_select_onecol_txn, table, keyvalues, retcol + desc, self.simple_select_onecol_txn, table, keyvalues, retcol ) - def _simple_select_list( - self, table, keyvalues, retcols, desc="_simple_select_list" - ): + def simple_select_list(self, table, keyvalues, retcols, desc="simple_select_list"): """Executes a SELECT query on the named table, which may return zero or more rows, returning the result as a list of dicts. @@ -1014,11 +1012,11 @@ class SQLBaseStore(object): defer.Deferred: resolves to list[dict[str, Any]] """ return self.runInteraction( - desc, self._simple_select_list_txn, table, keyvalues, retcols + desc, self.simple_select_list_txn, table, keyvalues, retcols ) @classmethod - def _simple_select_list_txn(cls, txn, table, keyvalues, retcols): + def simple_select_list_txn(cls, txn, table, keyvalues, retcols): """Executes a SELECT query on the named table, which may return zero or more rows, returning the result as a list of dicts. @@ -1044,14 +1042,14 @@ class SQLBaseStore(object): return cls.cursor_to_dict(txn) @defer.inlineCallbacks - def _simple_select_many_batch( + def simple_select_many_batch( self, table, column, iterable, retcols, keyvalues={}, - desc="_simple_select_many_batch", + desc="simple_select_many_batch", batch_size=100, ): """Executes a SELECT query on the named table, which may return zero or @@ -1080,7 +1078,7 @@ class SQLBaseStore(object): for chunk in chunks: rows = yield self.runInteraction( desc, - self._simple_select_many_txn, + self.simple_select_many_txn, table, column, chunk, @@ -1093,7 +1091,7 @@ class SQLBaseStore(object): return results @classmethod - def _simple_select_many_txn(cls, txn, table, column, iterable, keyvalues, retcols): + def simple_select_many_txn(cls, txn, table, column, iterable, keyvalues, retcols): """Executes a SELECT query on the named table, which may return zero or more rows, returning the result as a list of dicts. @@ -1126,13 +1124,13 @@ class SQLBaseStore(object): txn.execute(sql, values) return cls.cursor_to_dict(txn) - def _simple_update(self, table, keyvalues, updatevalues, desc): + def simple_update(self, table, keyvalues, updatevalues, desc): return self.runInteraction( - desc, self._simple_update_txn, table, keyvalues, updatevalues + desc, self.simple_update_txn, table, keyvalues, updatevalues ) @staticmethod - def _simple_update_txn(txn, table, keyvalues, updatevalues): + def simple_update_txn(txn, table, keyvalues, updatevalues): if keyvalues: where = "WHERE %s" % " AND ".join("%s = ?" % k for k in iterkeys(keyvalues)) else: @@ -1148,8 +1146,8 @@ class SQLBaseStore(object): return txn.rowcount - def _simple_update_one( - self, table, keyvalues, updatevalues, desc="_simple_update_one" + def simple_update_one( + self, table, keyvalues, updatevalues, desc="simple_update_one" ): """Executes an UPDATE query on the named table, setting new values for columns in a row matching the key values. @@ -1169,12 +1167,12 @@ class SQLBaseStore(object): the update column in the 'keyvalues' dict as well. """ return self.runInteraction( - desc, self._simple_update_one_txn, table, keyvalues, updatevalues + desc, self.simple_update_one_txn, table, keyvalues, updatevalues ) @classmethod - def _simple_update_one_txn(cls, txn, table, keyvalues, updatevalues): - rowcount = cls._simple_update_txn(txn, table, keyvalues, updatevalues) + def simple_update_one_txn(cls, txn, table, keyvalues, updatevalues): + rowcount = cls.simple_update_txn(txn, table, keyvalues, updatevalues) if rowcount == 0: raise StoreError(404, "No row found (%s)" % (table,)) @@ -1182,7 +1180,7 @@ class SQLBaseStore(object): raise StoreError(500, "More than one row matched (%s)" % (table,)) @staticmethod - def _simple_select_one_txn(txn, table, keyvalues, retcols, allow_none=False): + def simple_select_one_txn(txn, table, keyvalues, retcols, allow_none=False): select_sql = "SELECT %s FROM %s WHERE %s" % ( ", ".join(retcols), table, @@ -1201,7 +1199,7 @@ class SQLBaseStore(object): return dict(zip(retcols, row)) - def _simple_delete_one(self, table, keyvalues, desc="_simple_delete_one"): + def simple_delete_one(self, table, keyvalues, desc="simple_delete_one"): """Executes a DELETE query on the named table, expecting to delete a single row. @@ -1209,10 +1207,10 @@ class SQLBaseStore(object): table : string giving the table name keyvalues : dict of column names and values to select the row with """ - return self.runInteraction(desc, self._simple_delete_one_txn, table, keyvalues) + return self.runInteraction(desc, self.simple_delete_one_txn, table, keyvalues) @staticmethod - def _simple_delete_one_txn(txn, table, keyvalues): + def simple_delete_one_txn(txn, table, keyvalues): """Executes a DELETE query on the named table, expecting to delete a single row. @@ -1231,11 +1229,11 @@ class SQLBaseStore(object): if txn.rowcount > 1: raise StoreError(500, "More than one row matched (%s)" % (table,)) - def _simple_delete(self, table, keyvalues, desc): - return self.runInteraction(desc, self._simple_delete_txn, table, keyvalues) + def simple_delete(self, table, keyvalues, desc): + return self.runInteraction(desc, self.simple_delete_txn, table, keyvalues) @staticmethod - def _simple_delete_txn(txn, table, keyvalues): + def simple_delete_txn(txn, table, keyvalues): sql = "DELETE FROM %s WHERE %s" % ( table, " AND ".join("%s = ?" % (k,) for k in keyvalues), @@ -1244,13 +1242,13 @@ class SQLBaseStore(object): txn.execute(sql, list(keyvalues.values())) return txn.rowcount - def _simple_delete_many(self, table, column, iterable, keyvalues, desc): + def simple_delete_many(self, table, column, iterable, keyvalues, desc): return self.runInteraction( - desc, self._simple_delete_many_txn, table, column, iterable, keyvalues + desc, self.simple_delete_many_txn, table, column, iterable, keyvalues ) @staticmethod - def _simple_delete_many_txn(txn, table, column, iterable, keyvalues): + def simple_delete_many_txn(txn, table, column, iterable, keyvalues): """Executes a DELETE query on the named table. Filters rows by if value of `column` is in `iterable`. @@ -1283,7 +1281,7 @@ class SQLBaseStore(object): return txn.rowcount - def _get_cache_dict( + def get_cache_dict( self, db_conn, table, entity_column, stream_column, max_value, limit=100000 ): # Fetch a mapping of room_id -> max stream position for "recent" rooms. @@ -1349,7 +1347,7 @@ class SQLBaseStore(object): # which is fine. pass - def _simple_select_list_paginate( + def simple_select_list_paginate( self, table, keyvalues, @@ -1358,7 +1356,7 @@ class SQLBaseStore(object): limit, retcols, order_direction="ASC", - desc="_simple_select_list_paginate", + desc="simple_select_list_paginate", ): """ Executes a SELECT query on the named table with start and limit, @@ -1380,7 +1378,7 @@ class SQLBaseStore(object): """ return self.runInteraction( desc, - self._simple_select_list_paginate_txn, + self.simple_select_list_paginate_txn, table, keyvalues, orderby, @@ -1391,7 +1389,7 @@ class SQLBaseStore(object): ) @classmethod - def _simple_select_list_paginate_txn( + def simple_select_list_paginate_txn( cls, txn, table, @@ -1452,9 +1450,7 @@ class SQLBaseStore(object): txn.execute(sql_count) return txn.fetchone()[0] - def _simple_search_list( - self, table, term, col, retcols, desc="_simple_search_list" - ): + def simple_search_list(self, table, term, col, retcols, desc="simple_search_list"): """Executes a SELECT query on the named table, which may return zero or more rows, returning the result as a list of dicts. @@ -1469,11 +1465,11 @@ class SQLBaseStore(object): """ return self.runInteraction( - desc, self._simple_search_list_txn, table, term, col, retcols + desc, self.simple_search_list_txn, table, term, col, retcols ) @classmethod - def _simple_search_list_txn(cls, txn, table, term, col, retcols): + def simple_search_list_txn(cls, txn, table, term, col, retcols): """Executes a SELECT query on the named table, which may return zero or more rows, returning the result as a list of dicts. diff --git a/synapse/storage/background_updates.py b/synapse/storage/background_updates.py index 37d469ffd7..06955a0537 100644 --- a/synapse/storage/background_updates.py +++ b/synapse/storage/background_updates.py @@ -139,7 +139,7 @@ class BackgroundUpdateStore(SQLBaseStore): # otherwise, check if there are updates to be run. This is important, # as we may be running on a worker which doesn't perform the bg updates # itself, but still wants to wait for them to happen. - updates = yield self._simple_select_onecol( + updates = yield self.simple_select_onecol( "background_updates", keyvalues=None, retcol="1", @@ -161,7 +161,7 @@ class BackgroundUpdateStore(SQLBaseStore): if update_name in self._background_update_queue: return False - update_exists = await self._simple_select_one_onecol( + update_exists = await self.simple_select_one_onecol( "background_updates", keyvalues={"update_name": update_name}, retcol="1", @@ -184,7 +184,7 @@ class BackgroundUpdateStore(SQLBaseStore): no more work to do. """ if not self._background_update_queue: - updates = yield self._simple_select_list( + updates = yield self.simple_select_list( "background_updates", keyvalues=None, retcols=("update_name", "depends_on"), @@ -226,7 +226,7 @@ class BackgroundUpdateStore(SQLBaseStore): else: batch_size = self.DEFAULT_BACKGROUND_BATCH_SIZE - progress_json = yield self._simple_select_one_onecol( + progress_json = yield self.simple_select_one_onecol( "background_updates", keyvalues={"update_name": update_name}, retcol="progress_json", @@ -413,7 +413,7 @@ class BackgroundUpdateStore(SQLBaseStore): self._background_update_queue = [] progress_json = json.dumps(progress) - return self._simple_insert( + return self.simple_insert( "background_updates", {"update_name": update_name, "progress_json": progress_json}, ) @@ -429,7 +429,7 @@ class BackgroundUpdateStore(SQLBaseStore): self._background_update_queue = [ name for name in self._background_update_queue if name != update_name ] - return self._simple_delete_one( + return self.simple_delete_one( "background_updates", keyvalues={"update_name": update_name} ) @@ -444,7 +444,7 @@ class BackgroundUpdateStore(SQLBaseStore): progress_json = json.dumps(progress) - self._simple_update_one_txn( + self.simple_update_one_txn( txn, "background_updates", keyvalues={"update_name": update_name}, diff --git a/synapse/storage/data_stores/main/__init__.py b/synapse/storage/data_stores/main/__init__.py index 474924c68f..2a5b33dda1 100644 --- a/synapse/storage/data_stores/main/__init__.py +++ b/synapse/storage/data_stores/main/__init__.py @@ -173,7 +173,7 @@ class DataStore( self._presence_on_startup = self._get_active_presence(db_conn) - presence_cache_prefill, min_presence_val = self._get_cache_dict( + presence_cache_prefill, min_presence_val = self.get_cache_dict( db_conn, "presence_stream", entity_column="user_id", @@ -187,7 +187,7 @@ class DataStore( ) max_device_inbox_id = self._device_inbox_id_gen.get_current_token() - device_inbox_prefill, min_device_inbox_id = self._get_cache_dict( + device_inbox_prefill, min_device_inbox_id = self.get_cache_dict( db_conn, "device_inbox", entity_column="user_id", @@ -202,7 +202,7 @@ class DataStore( ) # The federation outbox and the local device inbox uses the same # stream_id generator. - device_outbox_prefill, min_device_outbox_id = self._get_cache_dict( + device_outbox_prefill, min_device_outbox_id = self.get_cache_dict( db_conn, "device_federation_outbox", entity_column="destination", @@ -228,7 +228,7 @@ class DataStore( ) events_max = self._stream_id_gen.get_current_token() - curr_state_delta_prefill, min_curr_state_delta_id = self._get_cache_dict( + curr_state_delta_prefill, min_curr_state_delta_id = self.get_cache_dict( db_conn, "current_state_delta_stream", entity_column="room_id", @@ -242,7 +242,7 @@ class DataStore( prefilled_cache=curr_state_delta_prefill, ) - _group_updates_prefill, min_group_updates_id = self._get_cache_dict( + _group_updates_prefill, min_group_updates_id = self.get_cache_dict( db_conn, "local_group_updates", entity_column="user_id", @@ -482,7 +482,7 @@ class DataStore( Returns: defer.Deferred: resolves to list[dict[str, Any]] """ - return self._simple_select_list( + return self.simple_select_list( table="users", keyvalues={}, retcols=["name", "password_hash", "is_guest", "admin", "user_type"], @@ -504,7 +504,7 @@ class DataStore( """ users = yield self.runInteraction( "get_users_paginate", - self._simple_select_list_paginate_txn, + self.simple_select_list_paginate_txn, table="users", keyvalues={"is_guest": False}, orderby=order, @@ -526,7 +526,7 @@ class DataStore( Returns: defer.Deferred: resolves to list[dict[str, Any]] """ - return self._simple_search_list( + return self.simple_search_list( table="users", term=term, col="name", diff --git a/synapse/storage/data_stores/main/account_data.py b/synapse/storage/data_stores/main/account_data.py index 22093484ed..b0d22faf3f 100644 --- a/synapse/storage/data_stores/main/account_data.py +++ b/synapse/storage/data_stores/main/account_data.py @@ -67,7 +67,7 @@ class AccountDataWorkerStore(SQLBaseStore): """ def get_account_data_for_user_txn(txn): - rows = self._simple_select_list_txn( + rows = self.simple_select_list_txn( txn, "account_data", {"user_id": user_id}, @@ -78,7 +78,7 @@ class AccountDataWorkerStore(SQLBaseStore): row["account_data_type"]: json.loads(row["content"]) for row in rows } - rows = self._simple_select_list_txn( + rows = self.simple_select_list_txn( txn, "room_account_data", {"user_id": user_id}, @@ -102,7 +102,7 @@ class AccountDataWorkerStore(SQLBaseStore): Returns: Deferred: A dict """ - result = yield self._simple_select_one_onecol( + result = yield self.simple_select_one_onecol( table="account_data", keyvalues={"user_id": user_id, "account_data_type": data_type}, retcol="content", @@ -127,7 +127,7 @@ class AccountDataWorkerStore(SQLBaseStore): """ def get_account_data_for_room_txn(txn): - rows = self._simple_select_list_txn( + rows = self.simple_select_list_txn( txn, "room_account_data", {"user_id": user_id, "room_id": room_id}, @@ -156,7 +156,7 @@ class AccountDataWorkerStore(SQLBaseStore): """ def get_account_data_for_room_and_type_txn(txn): - content_json = self._simple_select_one_onecol_txn( + content_json = self.simple_select_one_onecol_txn( txn, table="room_account_data", keyvalues={ @@ -300,9 +300,9 @@ class AccountDataStore(AccountDataWorkerStore): with self._account_data_id_gen.get_next() as next_id: # no need to lock here as room_account_data has a unique constraint - # on (user_id, room_id, account_data_type) so _simple_upsert will + # on (user_id, room_id, account_data_type) so simple_upsert will # retry if there is a conflict. - yield self._simple_upsert( + yield self.simple_upsert( desc="add_room_account_data", table="room_account_data", keyvalues={ @@ -346,9 +346,9 @@ class AccountDataStore(AccountDataWorkerStore): with self._account_data_id_gen.get_next() as next_id: # no need to lock here as account_data has a unique constraint on - # (user_id, account_data_type) so _simple_upsert will retry if + # (user_id, account_data_type) so simple_upsert will retry if # there is a conflict. - yield self._simple_upsert( + yield self.simple_upsert( desc="add_user_account_data", table="account_data", keyvalues={"user_id": user_id, "account_data_type": account_data_type}, diff --git a/synapse/storage/data_stores/main/appservice.py b/synapse/storage/data_stores/main/appservice.py index 81babf2029..6b82fd392a 100644 --- a/synapse/storage/data_stores/main/appservice.py +++ b/synapse/storage/data_stores/main/appservice.py @@ -133,7 +133,7 @@ class ApplicationServiceTransactionWorkerStore( A Deferred which resolves to a list of ApplicationServices, which may be empty. """ - results = yield self._simple_select_list( + results = yield self.simple_select_list( "application_services_state", dict(state=state), ["as_id"] ) # NB: This assumes this class is linked with ApplicationServiceStore @@ -155,7 +155,7 @@ class ApplicationServiceTransactionWorkerStore( Returns: A Deferred which resolves to ApplicationServiceState. """ - result = yield self._simple_select_one( + result = yield self.simple_select_one( "application_services_state", dict(as_id=service.id), ["state"], @@ -175,7 +175,7 @@ class ApplicationServiceTransactionWorkerStore( Returns: A Deferred which resolves when the state was set successfully. """ - return self._simple_upsert( + return self.simple_upsert( "application_services_state", dict(as_id=service.id), dict(state=state) ) @@ -249,7 +249,7 @@ class ApplicationServiceTransactionWorkerStore( ) # Set current txn_id for AS to 'txn_id' - self._simple_upsert_txn( + self.simple_upsert_txn( txn, "application_services_state", dict(as_id=service.id), @@ -257,7 +257,7 @@ class ApplicationServiceTransactionWorkerStore( ) # Delete txn - self._simple_delete_txn( + self.simple_delete_txn( txn, "application_services_txns", dict(txn_id=txn_id, as_id=service.id) ) diff --git a/synapse/storage/data_stores/main/cache.py b/synapse/storage/data_stores/main/cache.py index 258c08722a..de3256049d 100644 --- a/synapse/storage/data_stores/main/cache.py +++ b/synapse/storage/data_stores/main/cache.py @@ -95,7 +95,7 @@ class CacheInvalidationStore(SQLBaseStore): txn.call_after(ctx.__exit__, None, None, None) txn.call_after(self.hs.get_notifier().on_new_replication_data) - self._simple_insert_txn( + self.simple_insert_txn( txn, table="cache_invalidation_stream", values={ diff --git a/synapse/storage/data_stores/main/client_ips.py b/synapse/storage/data_stores/main/client_ips.py index cae93b0e22..66522a04b7 100644 --- a/synapse/storage/data_stores/main/client_ips.py +++ b/synapse/storage/data_stores/main/client_ips.py @@ -431,7 +431,7 @@ class ClientIpStore(ClientIpBackgroundUpdateStore): (user_id, access_token, ip), (user_agent, device_id, last_seen) = entry try: - self._simple_upsert_txn( + self.simple_upsert_txn( txn, table="user_ips", keyvalues={ @@ -450,7 +450,7 @@ class ClientIpStore(ClientIpBackgroundUpdateStore): # Technically an access token might not be associated with # a device so we need to check. if device_id: - self._simple_upsert_txn( + self.simple_upsert_txn( txn, table="devices", keyvalues={"user_id": user_id, "device_id": device_id}, @@ -483,7 +483,7 @@ class ClientIpStore(ClientIpBackgroundUpdateStore): if device_id is not None: keyvalues["device_id"] = device_id - res = yield self._simple_select_list( + res = yield self.simple_select_list( table="devices", keyvalues=keyvalues, retcols=("user_id", "ip", "user_agent", "device_id", "last_seen"), @@ -516,7 +516,7 @@ class ClientIpStore(ClientIpBackgroundUpdateStore): user_agent, _, last_seen = self._batch_row_update[key] results[(access_token, ip)] = (user_agent, last_seen) - rows = yield self._simple_select_list( + rows = yield self.simple_select_list( table="user_ips", keyvalues={"user_id": user_id}, retcols=["access_token", "ip", "user_agent", "last_seen"], diff --git a/synapse/storage/data_stores/main/deviceinbox.py b/synapse/storage/data_stores/main/deviceinbox.py index a23744f11c..206d39134d 100644 --- a/synapse/storage/data_stores/main/deviceinbox.py +++ b/synapse/storage/data_stores/main/deviceinbox.py @@ -314,7 +314,7 @@ class DeviceInboxStore(DeviceInboxWorkerStore, DeviceInboxBackgroundUpdateStore) # Check if we've already inserted a matching message_id for that # origin. This can happen if the origin doesn't receive our # acknowledgement from the first time we received the message. - already_inserted = self._simple_select_one_txn( + already_inserted = self.simple_select_one_txn( txn, table="device_federation_inbox", keyvalues={"origin": origin, "message_id": message_id}, @@ -326,7 +326,7 @@ class DeviceInboxStore(DeviceInboxWorkerStore, DeviceInboxBackgroundUpdateStore) # Add an entry for this message_id so that we know we've processed # it. - self._simple_insert_txn( + self.simple_insert_txn( txn, table="device_federation_inbox", values={ diff --git a/synapse/storage/data_stores/main/devices.py b/synapse/storage/data_stores/main/devices.py index a3ad23e783..727c582121 100644 --- a/synapse/storage/data_stores/main/devices.py +++ b/synapse/storage/data_stores/main/devices.py @@ -61,7 +61,7 @@ class DeviceWorkerStore(SQLBaseStore): Raises: StoreError: if the device is not found """ - return self._simple_select_one( + return self.simple_select_one( table="devices", keyvalues={"user_id": user_id, "device_id": device_id, "hidden": False}, retcols=("user_id", "device_id", "display_name"), @@ -80,7 +80,7 @@ class DeviceWorkerStore(SQLBaseStore): containing "device_id", "user_id" and "display_name" for each device. """ - devices = yield self._simple_select_list( + devices = yield self.simple_select_list( table="devices", keyvalues={"user_id": user_id, "hidden": False}, retcols=("user_id", "device_id", "display_name"), @@ -414,7 +414,7 @@ class DeviceWorkerStore(SQLBaseStore): from_user_id, stream_id, ) - self._simple_insert_txn( + self.simple_insert_txn( txn, "user_signature_stream", values={ @@ -466,7 +466,7 @@ class DeviceWorkerStore(SQLBaseStore): @cachedInlineCallbacks(num_args=2, tree=True) def _get_cached_user_device(self, user_id, device_id): - content = yield self._simple_select_one_onecol( + content = yield self.simple_select_one_onecol( table="device_lists_remote_cache", keyvalues={"user_id": user_id, "device_id": device_id}, retcol="content", @@ -476,7 +476,7 @@ class DeviceWorkerStore(SQLBaseStore): @cachedInlineCallbacks() def _get_cached_devices_for_user(self, user_id): - devices = yield self._simple_select_list( + devices = yield self.simple_select_list( table="device_lists_remote_cache", keyvalues={"user_id": user_id}, retcols=("device_id", "content"), @@ -584,7 +584,7 @@ class DeviceWorkerStore(SQLBaseStore): SELECT DISTINCT user_ids FROM user_signature_stream WHERE from_user_id = ? AND stream_id > ? """ - rows = yield self._execute( + rows = yield self.execute( "get_users_whose_signatures_changed", None, sql, user_id, from_key ) return set(user for row in rows for user in json.loads(row[0])) @@ -605,7 +605,7 @@ class DeviceWorkerStore(SQLBaseStore): WHERE ? < stream_id AND stream_id <= ? GROUP BY user_id, destination """ - return self._execute( + return self.execute( "get_all_device_list_changes_for_remotes", None, sql, from_key, to_key ) @@ -614,7 +614,7 @@ class DeviceWorkerStore(SQLBaseStore): """Get the last stream_id we got for a user. May be None if we haven't got any information for them. """ - return self._simple_select_one_onecol( + return self.simple_select_one_onecol( table="device_lists_remote_extremeties", keyvalues={"user_id": user_id}, retcol="stream_id", @@ -628,7 +628,7 @@ class DeviceWorkerStore(SQLBaseStore): inlineCallbacks=True, ) def get_device_list_last_stream_id_for_remotes(self, user_ids): - rows = yield self._simple_select_many_batch( + rows = yield self.simple_select_many_batch( table="device_lists_remote_extremeties", column="user_id", iterable=user_ids, @@ -722,7 +722,7 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): return False try: - inserted = yield self._simple_insert( + inserted = yield self.simple_insert( "devices", values={ "user_id": user_id, @@ -736,7 +736,7 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): if not inserted: # if the device already exists, check if it's a real device, or # if the device ID is reserved by something else - hidden = yield self._simple_select_one_onecol( + hidden = yield self.simple_select_one_onecol( "devices", keyvalues={"user_id": user_id, "device_id": device_id}, retcol="hidden", @@ -771,7 +771,7 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): Returns: defer.Deferred """ - yield self._simple_delete_one( + yield self.simple_delete_one( table="devices", keyvalues={"user_id": user_id, "device_id": device_id, "hidden": False}, desc="delete_device", @@ -789,7 +789,7 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): Returns: defer.Deferred """ - yield self._simple_delete_many( + yield self.simple_delete_many( table="devices", column="device_id", iterable=device_ids, @@ -818,7 +818,7 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): updates["display_name"] = new_display_name if not updates: return defer.succeed(None) - return self._simple_update_one( + return self.simple_update_one( table="devices", keyvalues={"user_id": user_id, "device_id": device_id, "hidden": False}, updatevalues=updates, @@ -829,7 +829,7 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): def mark_remote_user_device_list_as_unsubscribed(self, user_id): """Mark that we no longer track device lists for remote user. """ - yield self._simple_delete( + yield self.simple_delete( table="device_lists_remote_extremeties", keyvalues={"user_id": user_id}, desc="mark_remote_user_device_list_as_unsubscribed", @@ -866,7 +866,7 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): self, txn, user_id, device_id, content, stream_id ): if content.get("deleted"): - self._simple_delete_txn( + self.simple_delete_txn( txn, table="device_lists_remote_cache", keyvalues={"user_id": user_id, "device_id": device_id}, @@ -874,7 +874,7 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): txn.call_after(self.device_id_exists_cache.invalidate, (user_id, device_id)) else: - self._simple_upsert_txn( + self.simple_upsert_txn( txn, table="device_lists_remote_cache", keyvalues={"user_id": user_id, "device_id": device_id}, @@ -890,7 +890,7 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): self.get_device_list_last_stream_id_for_remote.invalidate, (user_id,) ) - self._simple_upsert_txn( + self.simple_upsert_txn( txn, table="device_lists_remote_extremeties", keyvalues={"user_id": user_id}, @@ -923,11 +923,11 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): ) def _update_remote_device_list_cache_txn(self, txn, user_id, devices, stream_id): - self._simple_delete_txn( + self.simple_delete_txn( txn, table="device_lists_remote_cache", keyvalues={"user_id": user_id} ) - self._simple_insert_many_txn( + self.simple_insert_many_txn( txn, table="device_lists_remote_cache", values=[ @@ -946,7 +946,7 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): self.get_device_list_last_stream_id_for_remote.invalidate, (user_id,) ) - self._simple_upsert_txn( + self.simple_upsert_txn( txn, table="device_lists_remote_extremeties", keyvalues={"user_id": user_id}, @@ -995,7 +995,7 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): [(user_id, device_id, stream_id) for device_id in device_ids], ) - self._simple_insert_many_txn( + self.simple_insert_many_txn( txn, table="device_lists_stream", values=[ @@ -1006,7 +1006,7 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): context = get_active_span_text_map() - self._simple_insert_many_txn( + self.simple_insert_many_txn( txn, table="device_lists_outbound_pokes", values=[ diff --git a/synapse/storage/data_stores/main/directory.py b/synapse/storage/data_stores/main/directory.py index 297966d9f4..d332f8a409 100644 --- a/synapse/storage/data_stores/main/directory.py +++ b/synapse/storage/data_stores/main/directory.py @@ -36,7 +36,7 @@ class DirectoryWorkerStore(SQLBaseStore): Deferred: results in namedtuple with keys "room_id" and "servers" or None if no association can be found """ - room_id = yield self._simple_select_one_onecol( + room_id = yield self.simple_select_one_onecol( "room_aliases", {"room_alias": room_alias.to_string()}, "room_id", @@ -47,7 +47,7 @@ class DirectoryWorkerStore(SQLBaseStore): if not room_id: return None - servers = yield self._simple_select_onecol( + servers = yield self.simple_select_onecol( "room_alias_servers", {"room_alias": room_alias.to_string()}, "server", @@ -60,7 +60,7 @@ class DirectoryWorkerStore(SQLBaseStore): return RoomAliasMapping(room_id, room_alias.to_string(), servers) def get_room_alias_creator(self, room_alias): - return self._simple_select_one_onecol( + return self.simple_select_one_onecol( table="room_aliases", keyvalues={"room_alias": room_alias}, retcol="creator", @@ -69,7 +69,7 @@ class DirectoryWorkerStore(SQLBaseStore): @cached(max_entries=5000) def get_aliases_for_room(self, room_id): - return self._simple_select_onecol( + return self.simple_select_onecol( "room_aliases", {"room_id": room_id}, "room_alias", @@ -93,7 +93,7 @@ class DirectoryStore(DirectoryWorkerStore): """ def alias_txn(txn): - self._simple_insert_txn( + self.simple_insert_txn( txn, "room_aliases", { @@ -103,7 +103,7 @@ class DirectoryStore(DirectoryWorkerStore): }, ) - self._simple_insert_many_txn( + self.simple_insert_many_txn( txn, table="room_alias_servers", values=[ diff --git a/synapse/storage/data_stores/main/e2e_room_keys.py b/synapse/storage/data_stores/main/e2e_room_keys.py index 113224fd7c..df89eda337 100644 --- a/synapse/storage/data_stores/main/e2e_room_keys.py +++ b/synapse/storage/data_stores/main/e2e_room_keys.py @@ -38,7 +38,7 @@ class EndToEndRoomKeyStore(SQLBaseStore): StoreError """ - yield self._simple_update_one( + yield self.simple_update_one( table="e2e_room_keys", keyvalues={ "user_id": user_id, @@ -89,7 +89,7 @@ class EndToEndRoomKeyStore(SQLBaseStore): } ) - yield self._simple_insert_many( + yield self.simple_insert_many( table="e2e_room_keys", values=values, desc="add_e2e_room_keys" ) @@ -125,7 +125,7 @@ class EndToEndRoomKeyStore(SQLBaseStore): if session_id: keyvalues["session_id"] = session_id - rows = yield self._simple_select_list( + rows = yield self.simple_select_list( table="e2e_room_keys", keyvalues=keyvalues, retcols=( @@ -234,7 +234,7 @@ class EndToEndRoomKeyStore(SQLBaseStore): version (str): the version ID of the backup we're querying about """ - return self._simple_select_one_onecol( + return self.simple_select_one_onecol( table="e2e_room_keys", keyvalues={"user_id": user_id, "version": version}, retcol="COUNT(*)", @@ -267,7 +267,7 @@ class EndToEndRoomKeyStore(SQLBaseStore): if session_id: keyvalues["session_id"] = session_id - yield self._simple_delete( + yield self.simple_delete( table="e2e_room_keys", keyvalues=keyvalues, desc="delete_e2e_room_keys" ) @@ -312,7 +312,7 @@ class EndToEndRoomKeyStore(SQLBaseStore): # it isn't there. raise StoreError(404, "No row found") - result = self._simple_select_one_txn( + result = self.simple_select_one_txn( txn, table="e2e_room_keys_versions", keyvalues={"user_id": user_id, "version": this_version, "deleted": 0}, @@ -352,7 +352,7 @@ class EndToEndRoomKeyStore(SQLBaseStore): new_version = str(int(current_version) + 1) - self._simple_insert_txn( + self.simple_insert_txn( txn, table="e2e_room_keys_versions", values={ @@ -391,7 +391,7 @@ class EndToEndRoomKeyStore(SQLBaseStore): updatevalues["etag"] = version_etag if updatevalues: - return self._simple_update( + return self.simple_update( table="e2e_room_keys_versions", keyvalues={"user_id": user_id, "version": version}, updatevalues=updatevalues, @@ -420,13 +420,13 @@ class EndToEndRoomKeyStore(SQLBaseStore): else: this_version = version - self._simple_delete_txn( + self.simple_delete_txn( txn, table="e2e_room_keys", keyvalues={"user_id": user_id, "version": this_version}, ) - return self._simple_update_one_txn( + return self.simple_update_one_txn( txn, table="e2e_room_keys_versions", keyvalues={"user_id": user_id, "version": this_version}, diff --git a/synapse/storage/data_stores/main/end_to_end_keys.py b/synapse/storage/data_stores/main/end_to_end_keys.py index 643327b57b..08bcdc4725 100644 --- a/synapse/storage/data_stores/main/end_to_end_keys.py +++ b/synapse/storage/data_stores/main/end_to_end_keys.py @@ -186,7 +186,7 @@ class EndToEndKeyWorkerStore(SQLBaseStore): key_id) to json string for key """ - rows = yield self._simple_select_many_batch( + rows = yield self.simple_select_many_batch( table="e2e_one_time_keys_json", column="key_id", iterable=key_ids, @@ -219,7 +219,7 @@ class EndToEndKeyWorkerStore(SQLBaseStore): # a unique constraint. If there is a race of two calls to # `add_e2e_one_time_keys` then they'll conflict and we will only # insert one set. - self._simple_insert_many_txn( + self.simple_insert_many_txn( txn, table="e2e_one_time_keys_json", values=[ @@ -350,7 +350,7 @@ class EndToEndKeyWorkerStore(SQLBaseStore): WHERE ? < stream_id AND stream_id <= ? GROUP BY user_id """ - return self._execute( + return self.execute( "get_all_user_signature_changes_for_remotes", None, sql, from_key, to_key ) @@ -367,7 +367,7 @@ class EndToEndKeyStore(EndToEndKeyWorkerStore, SQLBaseStore): set_tag("time_now", time_now) set_tag("device_keys", device_keys) - old_key_json = self._simple_select_one_onecol_txn( + old_key_json = self.simple_select_one_onecol_txn( txn, table="e2e_device_keys_json", keyvalues={"user_id": user_id, "device_id": device_id}, @@ -383,7 +383,7 @@ class EndToEndKeyStore(EndToEndKeyWorkerStore, SQLBaseStore): log_kv({"Message": "Device key already stored."}) return False - self._simple_upsert_txn( + self.simple_upsert_txn( txn, table="e2e_device_keys_json", keyvalues={"user_id": user_id, "device_id": device_id}, @@ -442,12 +442,12 @@ class EndToEndKeyStore(EndToEndKeyWorkerStore, SQLBaseStore): "user_id": user_id, } ) - self._simple_delete_txn( + self.simple_delete_txn( txn, table="e2e_device_keys_json", keyvalues={"user_id": user_id, "device_id": device_id}, ) - self._simple_delete_txn( + self.simple_delete_txn( txn, table="e2e_one_time_keys_json", keyvalues={"user_id": user_id, "device_id": device_id}, @@ -492,7 +492,7 @@ class EndToEndKeyStore(EndToEndKeyWorkerStore, SQLBaseStore): # The "keys" property must only have one entry, which will be the public # key, so we just grab the first value in there pubkey = next(iter(key["keys"].values())) - self._simple_insert_txn( + self.simple_insert_txn( txn, "devices", values={ @@ -505,7 +505,7 @@ class EndToEndKeyStore(EndToEndKeyWorkerStore, SQLBaseStore): # and finally, store the key itself with self._cross_signing_id_gen.get_next() as stream_id: - self._simple_insert_txn( + self.simple_insert_txn( txn, "e2e_cross_signing_keys", values={ @@ -539,7 +539,7 @@ class EndToEndKeyStore(EndToEndKeyWorkerStore, SQLBaseStore): user_id (str): the user who made the signatures signatures (iterable[SignatureListItem]): signatures to add """ - return self._simple_insert_many( + return self.simple_insert_many( "e2e_cross_signing_signatures", [ { diff --git a/synapse/storage/data_stores/main/event_federation.py b/synapse/storage/data_stores/main/event_federation.py index 90bef0cd2c..051ac7a8cb 100644 --- a/synapse/storage/data_stores/main/event_federation.py +++ b/synapse/storage/data_stores/main/event_federation.py @@ -126,7 +126,7 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBas Returns Deferred[int] """ - rows = yield self._simple_select_many_batch( + rows = yield self.simple_select_many_batch( table="events", column="event_id", iterable=event_ids, @@ -140,7 +140,7 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBas return max(row["depth"] for row in rows) def _get_oldest_events_in_room_txn(self, txn, room_id): - return self._simple_select_onecol_txn( + return self.simple_select_onecol_txn( txn, table="event_backward_extremities", keyvalues={"room_id": room_id}, @@ -235,7 +235,7 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBas @cached(max_entries=5000, iterable=True) def get_latest_event_ids_in_room(self, room_id): - return self._simple_select_onecol( + return self.simple_select_onecol( table="event_forward_extremities", keyvalues={"room_id": room_id}, retcol="event_id", @@ -271,7 +271,7 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBas ) def _get_min_depth_interaction(self, txn, room_id): - min_depth = self._simple_select_one_onecol_txn( + min_depth = self.simple_select_one_onecol_txn( txn, table="room_depth", keyvalues={"room_id": room_id}, @@ -383,7 +383,7 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBas queue = PriorityQueue() for event_id in event_list: - depth = self._simple_select_one_onecol_txn( + depth = self.simple_select_one_onecol_txn( txn, table="events", keyvalues={"event_id": event_id, "room_id": room_id}, @@ -468,7 +468,7 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBas Returns: Deferred[list[str]] """ - rows = yield self._simple_select_many_batch( + rows = yield self.simple_select_many_batch( table="event_edges", column="prev_event_id", iterable=event_ids, @@ -508,7 +508,7 @@ class EventFederationStore(EventFederationWorkerStore): if min_depth and depth >= min_depth: return - self._simple_upsert_txn( + self.simple_upsert_txn( txn, table="room_depth", keyvalues={"room_id": room_id}, @@ -520,7 +520,7 @@ class EventFederationStore(EventFederationWorkerStore): For the given event, update the event edges table and forward and backward extremities tables. """ - self._simple_insert_many_txn( + self.simple_insert_many_txn( txn, table="event_edges", values=[ diff --git a/synapse/storage/data_stores/main/event_push_actions.py b/synapse/storage/data_stores/main/event_push_actions.py index 04ce21ac66..0a37847cfd 100644 --- a/synapse/storage/data_stores/main/event_push_actions.py +++ b/synapse/storage/data_stores/main/event_push_actions.py @@ -441,7 +441,7 @@ class EventPushActionsWorkerStore(SQLBaseStore): ) def _add_push_actions_to_staging_txn(txn): - # We don't use _simple_insert_many here to avoid the overhead + # We don't use simple_insert_many here to avoid the overhead # of generating lists of dicts. sql = """ @@ -472,7 +472,7 @@ class EventPushActionsWorkerStore(SQLBaseStore): """ try: - res = yield self._simple_delete( + res = yield self.simple_delete( table="event_push_actions_staging", keyvalues={"event_id": event_id}, desc="remove_push_actions_from_staging", @@ -677,7 +677,7 @@ class EventPushActionsStore(EventPushActionsWorkerStore): ) for event, _ in events_and_contexts: - user_ids = self._simple_select_onecol_txn( + user_ids = self.simple_select_onecol_txn( txn, table="event_push_actions_staging", keyvalues={"event_id": event.event_id}, @@ -844,7 +844,7 @@ class EventPushActionsStore(EventPushActionsWorkerStore): the archiving process has caught up or not. """ - old_rotate_stream_ordering = self._simple_select_one_onecol_txn( + old_rotate_stream_ordering = self.simple_select_one_onecol_txn( txn, table="event_push_summary_stream_ordering", keyvalues={}, @@ -880,7 +880,7 @@ class EventPushActionsStore(EventPushActionsWorkerStore): return caught_up def _rotate_notifs_before_txn(self, txn, rotate_to_stream_ordering): - old_rotate_stream_ordering = self._simple_select_one_onecol_txn( + old_rotate_stream_ordering = self.simple_select_one_onecol_txn( txn, table="event_push_summary_stream_ordering", keyvalues={}, @@ -912,7 +912,7 @@ class EventPushActionsStore(EventPushActionsWorkerStore): # If the `old.user_id` above is NULL then we know there isn't already an # entry in the table, so we simply insert it. Otherwise we update the # existing table. - self._simple_insert_many_txn( + self.simple_insert_many_txn( txn, table="event_push_summary", values=[ diff --git a/synapse/storage/data_stores/main/events.py b/synapse/storage/data_stores/main/events.py index 79c91fe284..98ae69e996 100644 --- a/synapse/storage/data_stores/main/events.py +++ b/synapse/storage/data_stores/main/events.py @@ -432,7 +432,7 @@ class EventsStore( # event's auth chain, but its easier for now just to store them (and # it doesn't take much storage compared to storing the entire event # anyway). - self._simple_insert_many_txn( + self.simple_insert_many_txn( txn, table="event_auth", values=[ @@ -580,12 +580,12 @@ class EventsStore( self, txn, new_forward_extremities, max_stream_order ): for room_id, new_extrem in iteritems(new_forward_extremities): - self._simple_delete_txn( + self.simple_delete_txn( txn, table="event_forward_extremities", keyvalues={"room_id": room_id} ) txn.call_after(self.get_latest_event_ids_in_room.invalidate, (room_id,)) - self._simple_insert_many_txn( + self.simple_insert_many_txn( txn, table="event_forward_extremities", values=[ @@ -598,7 +598,7 @@ class EventsStore( # new stream_ordering to new forward extremeties in the room. # This allows us to later efficiently look up the forward extremeties # for a room before a given stream_ordering - self._simple_insert_many_txn( + self.simple_insert_many_txn( txn, table="stream_ordering_to_exterm", values=[ @@ -722,7 +722,7 @@ class EventsStore( # change in outlier status to our workers. stream_order = event.internal_metadata.stream_ordering state_group_id = context.state_group - self._simple_insert_txn( + self.simple_insert_txn( txn, table="ex_outlier_stream", values={ @@ -794,7 +794,7 @@ class EventsStore( d.pop("redacted_because", None) return d - self._simple_insert_many_txn( + self.simple_insert_many_txn( txn, table="event_json", values=[ @@ -811,7 +811,7 @@ class EventsStore( ], ) - self._simple_insert_many_txn( + self.simple_insert_many_txn( txn, table="events", values=[ @@ -841,7 +841,7 @@ class EventsStore( # If we're persisting an unredacted event we go and ensure # that we mark any redactions that reference this event as # requiring censoring. - self._simple_update_txn( + self.simple_update_txn( txn, table="redactions", keyvalues={"redacts": event.event_id}, @@ -983,7 +983,7 @@ class EventsStore( state_values.append(vals) - self._simple_insert_many_txn(txn, table="state_events", values=state_values) + self.simple_insert_many_txn(txn, table="state_events", values=state_values) # Prefill the event cache self._add_to_cache(txn, events_and_contexts) @@ -1032,7 +1032,7 @@ class EventsStore( # invalidate the cache for the redacted event txn.call_after(self._invalidate_get_event_cache, event.redacts) - self._simple_insert_txn( + self.simple_insert_txn( txn, table="redactions", values={ @@ -1077,9 +1077,7 @@ class EventsStore( LIMIT ? """ - rows = yield self._execute( - "_censor_redactions_fetch", None, sql, before_ts, 100 - ) + rows = yield self.execute("_censor_redactions_fetch", None, sql, before_ts, 100) updates = [] @@ -1111,7 +1109,7 @@ class EventsStore( if pruned_json: self._censor_event_txn(txn, event_id, pruned_json) - self._simple_update_one_txn( + self.simple_update_one_txn( txn, table="redactions", keyvalues={"event_id": redaction_id}, @@ -1129,7 +1127,7 @@ class EventsStore( event_id (str): The ID of the event to censor. pruned_json (str): The pruned JSON """ - self._simple_update_one_txn( + self.simple_update_one_txn( txn, table="event_json", keyvalues={"event_id": event_id}, @@ -1780,7 +1778,7 @@ class EventsStore( "[purge] found %i state groups to delete", len(state_groups_to_delete) ) - rows = self._simple_select_many_txn( + rows = self.simple_select_many_txn( txn, table="state_group_edges", column="prev_state_group", @@ -1807,15 +1805,15 @@ class EventsStore( curr_state = self._get_state_groups_from_groups_txn(txn, [sg]) curr_state = curr_state[sg] - self._simple_delete_txn( + self.simple_delete_txn( txn, table="state_groups_state", keyvalues={"state_group": sg} ) - self._simple_delete_txn( + self.simple_delete_txn( txn, table="state_group_edges", keyvalues={"state_group": sg} ) - self._simple_insert_many_txn( + self.simple_insert_many_txn( txn, table="state_groups_state", values=[ @@ -1852,7 +1850,7 @@ class EventsStore( state group. """ - rows = yield self._simple_select_many_batch( + rows = yield self.simple_select_many_batch( table="state_group_edges", column="prev_state_group", iterable=state_groups, @@ -1882,7 +1880,7 @@ class EventsStore( # first we have to delete the state groups states logger.info("[purge] removing %s from state_groups_state", room_id) - self._simple_delete_many_txn( + self.simple_delete_many_txn( txn, table="state_groups_state", column="state_group", @@ -1893,7 +1891,7 @@ class EventsStore( # ... and the state group edges logger.info("[purge] removing %s from state_group_edges", room_id) - self._simple_delete_many_txn( + self.simple_delete_many_txn( txn, table="state_group_edges", column="state_group", @@ -1904,7 +1902,7 @@ class EventsStore( # ... and the state groups logger.info("[purge] removing %s from state_groups", room_id) - self._simple_delete_many_txn( + self.simple_delete_many_txn( txn, table="state_groups", column="id", @@ -1921,7 +1919,7 @@ class EventsStore( @cachedInlineCallbacks(max_entries=5000) def _get_event_ordering(self, event_id): - res = yield self._simple_select_one( + res = yield self.simple_select_one( table="events", retcols=["topological_ordering", "stream_ordering"], keyvalues={"event_id": event_id}, @@ -1962,7 +1960,7 @@ class EventsStore( room_id (str): The ID of the room the event was sent to. topological_ordering (int): The position of the event in the room's topology. """ - return self._simple_insert_many_txn( + return self.simple_insert_many_txn( txn=txn, table="event_labels", values=[ @@ -1984,7 +1982,7 @@ class EventsStore( event_id (str): The event ID the expiry timestamp is associated with. expiry_ts (int): The timestamp at which to expire (delete) the event. """ - return self._simple_insert_txn( + return self.simple_insert_txn( txn=txn, table="event_expiry", values={"event_id": event_id, "expiry_ts": expiry_ts}, @@ -2043,7 +2041,7 @@ class EventsStore( txn (LoggingTransaction): The transaction to use to perform the deletion. event_id (str): The event ID to delete the associated expiry timestamp of. """ - return self._simple_delete_txn( + return self.simple_delete_txn( txn=txn, table="event_expiry", keyvalues={"event_id": event_id} ) diff --git a/synapse/storage/data_stores/main/events_bg_updates.py b/synapse/storage/data_stores/main/events_bg_updates.py index aa87f9abc5..37dfc8c871 100644 --- a/synapse/storage/data_stores/main/events_bg_updates.py +++ b/synapse/storage/data_stores/main/events_bg_updates.py @@ -189,7 +189,7 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): chunks = [event_ids[i : i + 100] for i in range(0, len(event_ids), 100)] for chunk in chunks: - ev_rows = self._simple_select_many_txn( + ev_rows = self.simple_select_many_txn( txn, table="event_json", column="event_id", @@ -366,7 +366,7 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): to_delete.intersection_update(original_set) - deleted = self._simple_delete_many_txn( + deleted = self.simple_delete_many_txn( txn=txn, table="event_forward_extremities", column="event_id", @@ -382,7 +382,7 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): if deleted: # We now need to invalidate the caches of these rooms - rows = self._simple_select_many_txn( + rows = self.simple_select_many_txn( txn, table="events", column="event_id", @@ -396,7 +396,7 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): self.get_latest_event_ids_in_room.invalidate, (room_id,) ) - self._simple_delete_many_txn( + self.simple_delete_many_txn( txn=txn, table="_extremities_to_check", column="event_id", @@ -533,7 +533,7 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): try: event_json = json.loads(event_json_raw) - self._simple_insert_many_txn( + self.simple_insert_many_txn( txn=txn, table="event_labels", values=[ diff --git a/synapse/storage/data_stores/main/events_worker.py b/synapse/storage/data_stores/main/events_worker.py index e782e8f481..ec4af29299 100644 --- a/synapse/storage/data_stores/main/events_worker.py +++ b/synapse/storage/data_stores/main/events_worker.py @@ -78,7 +78,7 @@ class EventsWorkerStore(SQLBaseStore): Deferred[int|None]: Timestamp in milliseconds, or None for events that were persisted before received_ts was implemented. """ - return self._simple_select_one_onecol( + return self.simple_select_one_onecol( table="events", keyvalues={"event_id": event_id}, retcol="received_ts", @@ -452,7 +452,7 @@ class EventsWorkerStore(SQLBaseStore): event_id for events, _ in event_list for event_id in events ) - row_dict = self._new_transaction( + row_dict = self.new_transaction( conn, "do_fetch", [], [], self._fetch_event_rows, events_to_fetch ) @@ -745,7 +745,7 @@ class EventsWorkerStore(SQLBaseStore): """Given a list of event ids, check if we have already processed and stored them as non outliers. """ - rows = yield self._simple_select_many_batch( + rows = yield self.simple_select_many_batch( table="events", retcols=("event_id",), column="event_id", diff --git a/synapse/storage/data_stores/main/filtering.py b/synapse/storage/data_stores/main/filtering.py index f05ace299a..17ef7b9354 100644 --- a/synapse/storage/data_stores/main/filtering.py +++ b/synapse/storage/data_stores/main/filtering.py @@ -30,7 +30,7 @@ class FilteringStore(SQLBaseStore): except ValueError: raise SynapseError(400, "Invalid filter ID", Codes.INVALID_PARAM) - def_json = yield self._simple_select_one_onecol( + def_json = yield self.simple_select_one_onecol( table="user_filters", keyvalues={"user_id": user_localpart, "filter_id": filter_id}, retcol="filter_json", diff --git a/synapse/storage/data_stores/main/group_server.py b/synapse/storage/data_stores/main/group_server.py index 5ded539af8..9e1d12bcb7 100644 --- a/synapse/storage/data_stores/main/group_server.py +++ b/synapse/storage/data_stores/main/group_server.py @@ -35,7 +35,7 @@ class GroupServerStore(SQLBaseStore): * "invite" * "open" """ - return self._simple_update_one( + return self.simple_update_one( table="groups", keyvalues={"group_id": group_id}, updatevalues={"join_policy": join_policy}, @@ -43,7 +43,7 @@ class GroupServerStore(SQLBaseStore): ) def get_group(self, group_id): - return self._simple_select_one( + return self.simple_select_one( table="groups", keyvalues={"group_id": group_id}, retcols=( @@ -65,7 +65,7 @@ class GroupServerStore(SQLBaseStore): if not include_private: keyvalues["is_public"] = True - return self._simple_select_list( + return self.simple_select_list( table="group_users", keyvalues=keyvalues, retcols=("user_id", "is_public", "is_admin"), @@ -75,7 +75,7 @@ class GroupServerStore(SQLBaseStore): def get_invited_users_in_group(self, group_id): # TODO: Pagination - return self._simple_select_onecol( + return self.simple_select_onecol( table="group_invites", keyvalues={"group_id": group_id}, retcol="user_id", @@ -89,7 +89,7 @@ class GroupServerStore(SQLBaseStore): if not include_private: keyvalues["is_public"] = True - return self._simple_select_list( + return self.simple_select_list( table="group_rooms", keyvalues=keyvalues, retcols=("room_id", "is_public"), @@ -180,7 +180,7 @@ class GroupServerStore(SQLBaseStore): an order of 1 will put the room first. Otherwise, the room gets added to the end. """ - room_in_group = self._simple_select_one_onecol_txn( + room_in_group = self.simple_select_one_onecol_txn( txn, table="group_rooms", keyvalues={"group_id": group_id, "room_id": room_id}, @@ -193,7 +193,7 @@ class GroupServerStore(SQLBaseStore): if category_id is None: category_id = _DEFAULT_CATEGORY_ID else: - cat_exists = self._simple_select_one_onecol_txn( + cat_exists = self.simple_select_one_onecol_txn( txn, table="group_room_categories", keyvalues={"group_id": group_id, "category_id": category_id}, @@ -204,7 +204,7 @@ class GroupServerStore(SQLBaseStore): raise SynapseError(400, "Category doesn't exist") # TODO: Check category is part of summary already - cat_exists = self._simple_select_one_onecol_txn( + cat_exists = self.simple_select_one_onecol_txn( txn, table="group_summary_room_categories", keyvalues={"group_id": group_id, "category_id": category_id}, @@ -224,7 +224,7 @@ class GroupServerStore(SQLBaseStore): (group_id, category_id, group_id, category_id), ) - existing = self._simple_select_one_txn( + existing = self.simple_select_one_txn( txn, table="group_summary_rooms", keyvalues={ @@ -257,7 +257,7 @@ class GroupServerStore(SQLBaseStore): to_update["room_order"] = order if is_public is not None: to_update["is_public"] = is_public - self._simple_update_txn( + self.simple_update_txn( txn, table="group_summary_rooms", keyvalues={ @@ -271,7 +271,7 @@ class GroupServerStore(SQLBaseStore): if is_public is None: is_public = True - self._simple_insert_txn( + self.simple_insert_txn( txn, table="group_summary_rooms", values={ @@ -287,7 +287,7 @@ class GroupServerStore(SQLBaseStore): if category_id is None: category_id = _DEFAULT_CATEGORY_ID - return self._simple_delete( + return self.simple_delete( table="group_summary_rooms", keyvalues={ "group_id": group_id, @@ -299,7 +299,7 @@ class GroupServerStore(SQLBaseStore): @defer.inlineCallbacks def get_group_categories(self, group_id): - rows = yield self._simple_select_list( + rows = yield self.simple_select_list( table="group_room_categories", keyvalues={"group_id": group_id}, retcols=("category_id", "is_public", "profile"), @@ -316,7 +316,7 @@ class GroupServerStore(SQLBaseStore): @defer.inlineCallbacks def get_group_category(self, group_id, category_id): - category = yield self._simple_select_one( + category = yield self.simple_select_one( table="group_room_categories", keyvalues={"group_id": group_id, "category_id": category_id}, retcols=("is_public", "profile"), @@ -343,7 +343,7 @@ class GroupServerStore(SQLBaseStore): else: update_values["is_public"] = is_public - return self._simple_upsert( + return self.simple_upsert( table="group_room_categories", keyvalues={"group_id": group_id, "category_id": category_id}, values=update_values, @@ -352,7 +352,7 @@ class GroupServerStore(SQLBaseStore): ) def remove_group_category(self, group_id, category_id): - return self._simple_delete( + return self.simple_delete( table="group_room_categories", keyvalues={"group_id": group_id, "category_id": category_id}, desc="remove_group_category", @@ -360,7 +360,7 @@ class GroupServerStore(SQLBaseStore): @defer.inlineCallbacks def get_group_roles(self, group_id): - rows = yield self._simple_select_list( + rows = yield self.simple_select_list( table="group_roles", keyvalues={"group_id": group_id}, retcols=("role_id", "is_public", "profile"), @@ -377,7 +377,7 @@ class GroupServerStore(SQLBaseStore): @defer.inlineCallbacks def get_group_role(self, group_id, role_id): - role = yield self._simple_select_one( + role = yield self.simple_select_one( table="group_roles", keyvalues={"group_id": group_id, "role_id": role_id}, retcols=("is_public", "profile"), @@ -404,7 +404,7 @@ class GroupServerStore(SQLBaseStore): else: update_values["is_public"] = is_public - return self._simple_upsert( + return self.simple_upsert( table="group_roles", keyvalues={"group_id": group_id, "role_id": role_id}, values=update_values, @@ -413,7 +413,7 @@ class GroupServerStore(SQLBaseStore): ) def remove_group_role(self, group_id, role_id): - return self._simple_delete( + return self.simple_delete( table="group_roles", keyvalues={"group_id": group_id, "role_id": role_id}, desc="remove_group_role", @@ -444,7 +444,7 @@ class GroupServerStore(SQLBaseStore): an order of 1 will put the user first. Otherwise, the user gets added to the end. """ - user_in_group = self._simple_select_one_onecol_txn( + user_in_group = self.simple_select_one_onecol_txn( txn, table="group_users", keyvalues={"group_id": group_id, "user_id": user_id}, @@ -457,7 +457,7 @@ class GroupServerStore(SQLBaseStore): if role_id is None: role_id = _DEFAULT_ROLE_ID else: - role_exists = self._simple_select_one_onecol_txn( + role_exists = self.simple_select_one_onecol_txn( txn, table="group_roles", keyvalues={"group_id": group_id, "role_id": role_id}, @@ -468,7 +468,7 @@ class GroupServerStore(SQLBaseStore): raise SynapseError(400, "Role doesn't exist") # TODO: Check role is part of the summary already - role_exists = self._simple_select_one_onecol_txn( + role_exists = self.simple_select_one_onecol_txn( txn, table="group_summary_roles", keyvalues={"group_id": group_id, "role_id": role_id}, @@ -488,7 +488,7 @@ class GroupServerStore(SQLBaseStore): (group_id, role_id, group_id, role_id), ) - existing = self._simple_select_one_txn( + existing = self.simple_select_one_txn( txn, table="group_summary_users", keyvalues={"group_id": group_id, "user_id": user_id, "role_id": role_id}, @@ -517,7 +517,7 @@ class GroupServerStore(SQLBaseStore): to_update["user_order"] = order if is_public is not None: to_update["is_public"] = is_public - self._simple_update_txn( + self.simple_update_txn( txn, table="group_summary_users", keyvalues={ @@ -531,7 +531,7 @@ class GroupServerStore(SQLBaseStore): if is_public is None: is_public = True - self._simple_insert_txn( + self.simple_insert_txn( txn, table="group_summary_users", values={ @@ -547,7 +547,7 @@ class GroupServerStore(SQLBaseStore): if role_id is None: role_id = _DEFAULT_ROLE_ID - return self._simple_delete( + return self.simple_delete( table="group_summary_users", keyvalues={"group_id": group_id, "role_id": role_id, "user_id": user_id}, desc="remove_user_from_summary", @@ -561,7 +561,7 @@ class GroupServerStore(SQLBaseStore): Deferred[list[str]]: A twisted.Deferred containing a list of group ids containing this room """ - return self._simple_select_onecol( + return self.simple_select_onecol( table="group_rooms", keyvalues={"room_id": room_id}, retcol="group_id", @@ -630,7 +630,7 @@ class GroupServerStore(SQLBaseStore): ) def is_user_in_group(self, user_id, group_id): - return self._simple_select_one_onecol( + return self.simple_select_one_onecol( table="group_users", keyvalues={"group_id": group_id, "user_id": user_id}, retcol="user_id", @@ -639,7 +639,7 @@ class GroupServerStore(SQLBaseStore): ).addCallback(lambda r: bool(r)) def is_user_admin_in_group(self, group_id, user_id): - return self._simple_select_one_onecol( + return self.simple_select_one_onecol( table="group_users", keyvalues={"group_id": group_id, "user_id": user_id}, retcol="is_admin", @@ -650,7 +650,7 @@ class GroupServerStore(SQLBaseStore): def add_group_invite(self, group_id, user_id): """Record that the group server has invited a user """ - return self._simple_insert( + return self.simple_insert( table="group_invites", values={"group_id": group_id, "user_id": user_id}, desc="add_group_invite", @@ -659,7 +659,7 @@ class GroupServerStore(SQLBaseStore): def is_user_invited_to_local_group(self, group_id, user_id): """Has the group server invited a user? """ - return self._simple_select_one_onecol( + return self.simple_select_one_onecol( table="group_invites", keyvalues={"group_id": group_id, "user_id": user_id}, retcol="user_id", @@ -682,7 +682,7 @@ class GroupServerStore(SQLBaseStore): """ def _get_users_membership_in_group_txn(txn): - row = self._simple_select_one_txn( + row = self.simple_select_one_txn( txn, table="group_users", keyvalues={"group_id": group_id, "user_id": user_id}, @@ -697,7 +697,7 @@ class GroupServerStore(SQLBaseStore): "is_privileged": row["is_admin"], } - row = self._simple_select_one_onecol_txn( + row = self.simple_select_one_onecol_txn( txn, table="group_invites", keyvalues={"group_id": group_id, "user_id": user_id}, @@ -738,7 +738,7 @@ class GroupServerStore(SQLBaseStore): """ def _add_user_to_group_txn(txn): - self._simple_insert_txn( + self.simple_insert_txn( txn, table="group_users", values={ @@ -749,14 +749,14 @@ class GroupServerStore(SQLBaseStore): }, ) - self._simple_delete_txn( + self.simple_delete_txn( txn, table="group_invites", keyvalues={"group_id": group_id, "user_id": user_id}, ) if local_attestation: - self._simple_insert_txn( + self.simple_insert_txn( txn, table="group_attestations_renewals", values={ @@ -766,7 +766,7 @@ class GroupServerStore(SQLBaseStore): }, ) if remote_attestation: - self._simple_insert_txn( + self.simple_insert_txn( txn, table="group_attestations_remote", values={ @@ -781,27 +781,27 @@ class GroupServerStore(SQLBaseStore): def remove_user_from_group(self, group_id, user_id): def _remove_user_from_group_txn(txn): - self._simple_delete_txn( + self.simple_delete_txn( txn, table="group_users", keyvalues={"group_id": group_id, "user_id": user_id}, ) - self._simple_delete_txn( + self.simple_delete_txn( txn, table="group_invites", keyvalues={"group_id": group_id, "user_id": user_id}, ) - self._simple_delete_txn( + self.simple_delete_txn( txn, table="group_attestations_renewals", keyvalues={"group_id": group_id, "user_id": user_id}, ) - self._simple_delete_txn( + self.simple_delete_txn( txn, table="group_attestations_remote", keyvalues={"group_id": group_id, "user_id": user_id}, ) - self._simple_delete_txn( + self.simple_delete_txn( txn, table="group_summary_users", keyvalues={"group_id": group_id, "user_id": user_id}, @@ -812,14 +812,14 @@ class GroupServerStore(SQLBaseStore): ) def add_room_to_group(self, group_id, room_id, is_public): - return self._simple_insert( + return self.simple_insert( table="group_rooms", values={"group_id": group_id, "room_id": room_id, "is_public": is_public}, desc="add_room_to_group", ) def update_room_in_group_visibility(self, group_id, room_id, is_public): - return self._simple_update( + return self.simple_update( table="group_rooms", keyvalues={"group_id": group_id, "room_id": room_id}, updatevalues={"is_public": is_public}, @@ -828,13 +828,13 @@ class GroupServerStore(SQLBaseStore): def remove_room_from_group(self, group_id, room_id): def _remove_room_from_group_txn(txn): - self._simple_delete_txn( + self.simple_delete_txn( txn, table="group_rooms", keyvalues={"group_id": group_id, "room_id": room_id}, ) - self._simple_delete_txn( + self.simple_delete_txn( txn, table="group_summary_rooms", keyvalues={"group_id": group_id, "room_id": room_id}, @@ -847,7 +847,7 @@ class GroupServerStore(SQLBaseStore): def get_publicised_groups_for_user(self, user_id): """Get all groups a user is publicising """ - return self._simple_select_onecol( + return self.simple_select_onecol( table="local_group_membership", keyvalues={"user_id": user_id, "membership": "join", "is_publicised": True}, retcol="group_id", @@ -857,7 +857,7 @@ class GroupServerStore(SQLBaseStore): def update_group_publicity(self, group_id, user_id, publicise): """Update whether the user is publicising their membership of the group """ - return self._simple_update_one( + return self.simple_update_one( table="local_group_membership", keyvalues={"group_id": group_id, "user_id": user_id}, updatevalues={"is_publicised": publicise}, @@ -893,12 +893,12 @@ class GroupServerStore(SQLBaseStore): def _register_user_group_membership_txn(txn, next_id): # TODO: Upsert? - self._simple_delete_txn( + self.simple_delete_txn( txn, table="local_group_membership", keyvalues={"group_id": group_id, "user_id": user_id}, ) - self._simple_insert_txn( + self.simple_insert_txn( txn, table="local_group_membership", values={ @@ -911,7 +911,7 @@ class GroupServerStore(SQLBaseStore): }, ) - self._simple_insert_txn( + self.simple_insert_txn( txn, table="local_group_updates", values={ @@ -930,7 +930,7 @@ class GroupServerStore(SQLBaseStore): if membership == "join": if local_attestation: - self._simple_insert_txn( + self.simple_insert_txn( txn, table="group_attestations_renewals", values={ @@ -940,7 +940,7 @@ class GroupServerStore(SQLBaseStore): }, ) if remote_attestation: - self._simple_insert_txn( + self.simple_insert_txn( txn, table="group_attestations_remote", values={ @@ -951,12 +951,12 @@ class GroupServerStore(SQLBaseStore): }, ) else: - self._simple_delete_txn( + self.simple_delete_txn( txn, table="group_attestations_renewals", keyvalues={"group_id": group_id, "user_id": user_id}, ) - self._simple_delete_txn( + self.simple_delete_txn( txn, table="group_attestations_remote", keyvalues={"group_id": group_id, "user_id": user_id}, @@ -976,7 +976,7 @@ class GroupServerStore(SQLBaseStore): def create_group( self, group_id, user_id, name, avatar_url, short_description, long_description ): - yield self._simple_insert( + yield self.simple_insert( table="groups", values={ "group_id": group_id, @@ -991,7 +991,7 @@ class GroupServerStore(SQLBaseStore): @defer.inlineCallbacks def update_group_profile(self, group_id, profile): - yield self._simple_update_one( + yield self.simple_update_one( table="groups", keyvalues={"group_id": group_id}, updatevalues=profile, @@ -1017,7 +1017,7 @@ class GroupServerStore(SQLBaseStore): def update_attestation_renewal(self, group_id, user_id, attestation): """Update an attestation that we have renewed """ - return self._simple_update_one( + return self.simple_update_one( table="group_attestations_renewals", keyvalues={"group_id": group_id, "user_id": user_id}, updatevalues={"valid_until_ms": attestation["valid_until_ms"]}, @@ -1027,7 +1027,7 @@ class GroupServerStore(SQLBaseStore): def update_remote_attestion(self, group_id, user_id, attestation): """Update an attestation that a remote has renewed """ - return self._simple_update_one( + return self.simple_update_one( table="group_attestations_remote", keyvalues={"group_id": group_id, "user_id": user_id}, updatevalues={ @@ -1046,7 +1046,7 @@ class GroupServerStore(SQLBaseStore): group_id (str) user_id (str) """ - return self._simple_delete( + return self.simple_delete( table="group_attestations_renewals", keyvalues={"group_id": group_id, "user_id": user_id}, desc="remove_attestation_renewal", @@ -1057,7 +1057,7 @@ class GroupServerStore(SQLBaseStore): """Get the attestation that proves the remote agrees that the user is in the group. """ - row = yield self._simple_select_one( + row = yield self.simple_select_one( table="group_attestations_remote", keyvalues={"group_id": group_id, "user_id": user_id}, retcols=("valid_until_ms", "attestation_json"), @@ -1072,7 +1072,7 @@ class GroupServerStore(SQLBaseStore): return None def get_joined_groups(self, user_id): - return self._simple_select_onecol( + return self.simple_select_onecol( table="local_group_membership", keyvalues={"user_id": user_id, "membership": "join"}, retcol="group_id", @@ -1188,7 +1188,7 @@ class GroupServerStore(SQLBaseStore): ] for table in tables: - self._simple_delete_txn( + self.simple_delete_txn( txn, table=table, keyvalues={"group_id": group_id} ) diff --git a/synapse/storage/data_stores/main/keys.py b/synapse/storage/data_stores/main/keys.py index ebc7db3ed6..c7150432b3 100644 --- a/synapse/storage/data_stores/main/keys.py +++ b/synapse/storage/data_stores/main/keys.py @@ -129,7 +129,7 @@ class KeyStore(SQLBaseStore): return self.runInteraction( "store_server_verify_keys", - self._simple_upsert_many_txn, + self.simple_upsert_many_txn, table="server_signature_keys", key_names=("server_name", "key_id"), key_values=key_values, @@ -157,7 +157,7 @@ class KeyStore(SQLBaseStore): ts_valid_until_ms (int): The time when this json stops being valid. key_json (bytes): The encoded JSON. """ - return self._simple_upsert( + return self.simple_upsert( table="server_keys_json", keyvalues={ "server_name": server_name, @@ -196,7 +196,7 @@ class KeyStore(SQLBaseStore): keyvalues["key_id"] = key_id if from_server is not None: keyvalues["from_server"] = from_server - rows = self._simple_select_list_txn( + rows = self.simple_select_list_txn( txn, "server_keys_json", keyvalues=keyvalues, diff --git a/synapse/storage/data_stores/main/media_repository.py b/synapse/storage/data_stores/main/media_repository.py index 0f2887bdce..0cb9446f96 100644 --- a/synapse/storage/data_stores/main/media_repository.py +++ b/synapse/storage/data_stores/main/media_repository.py @@ -39,7 +39,7 @@ class MediaRepositoryStore(MediaRepositoryBackgroundUpdateStore): Returns: None if the media_id doesn't exist. """ - return self._simple_select_one( + return self.simple_select_one( "local_media_repository", {"media_id": media_id}, ( @@ -64,7 +64,7 @@ class MediaRepositoryStore(MediaRepositoryBackgroundUpdateStore): user_id, url_cache=None, ): - return self._simple_insert( + return self.simple_insert( "local_media_repository", { "media_id": media_id, @@ -129,7 +129,7 @@ class MediaRepositoryStore(MediaRepositoryBackgroundUpdateStore): def store_url_cache( self, url, response_code, etag, expires_ts, og, media_id, download_ts ): - return self._simple_insert( + return self.simple_insert( "local_media_repository_url_cache", { "url": url, @@ -144,7 +144,7 @@ class MediaRepositoryStore(MediaRepositoryBackgroundUpdateStore): ) def get_local_media_thumbnails(self, media_id): - return self._simple_select_list( + return self.simple_select_list( "local_media_repository_thumbnails", {"media_id": media_id}, ( @@ -166,7 +166,7 @@ class MediaRepositoryStore(MediaRepositoryBackgroundUpdateStore): thumbnail_method, thumbnail_length, ): - return self._simple_insert( + return self.simple_insert( "local_media_repository_thumbnails", { "media_id": media_id, @@ -180,7 +180,7 @@ class MediaRepositoryStore(MediaRepositoryBackgroundUpdateStore): ) def get_cached_remote_media(self, origin, media_id): - return self._simple_select_one( + return self.simple_select_one( "remote_media_cache", {"media_origin": origin, "media_id": media_id}, ( @@ -205,7 +205,7 @@ class MediaRepositoryStore(MediaRepositoryBackgroundUpdateStore): upload_name, filesystem_id, ): - return self._simple_insert( + return self.simple_insert( "remote_media_cache", { "media_origin": origin, @@ -253,7 +253,7 @@ class MediaRepositoryStore(MediaRepositoryBackgroundUpdateStore): return self.runInteraction("update_cached_last_access_time", update_cache_txn) def get_remote_media_thumbnails(self, origin, media_id): - return self._simple_select_list( + return self.simple_select_list( "remote_media_cache_thumbnails", {"media_origin": origin, "media_id": media_id}, ( @@ -278,7 +278,7 @@ class MediaRepositoryStore(MediaRepositoryBackgroundUpdateStore): thumbnail_method, thumbnail_length, ): - return self._simple_insert( + return self.simple_insert( "remote_media_cache_thumbnails", { "media_origin": origin, @@ -300,18 +300,18 @@ class MediaRepositoryStore(MediaRepositoryBackgroundUpdateStore): " WHERE last_access_ts < ?" ) - return self._execute( + return self.execute( "get_remote_media_before", self.cursor_to_dict, sql, before_ts ) def delete_remote_media(self, media_origin, media_id): def delete_remote_media_txn(txn): - self._simple_delete_txn( + self.simple_delete_txn( txn, "remote_media_cache", keyvalues={"media_origin": media_origin, "media_id": media_id}, ) - self._simple_delete_txn( + self.simple_delete_txn( txn, "remote_media_cache_thumbnails", keyvalues={"media_origin": media_origin, "media_id": media_id}, diff --git a/synapse/storage/data_stores/main/monthly_active_users.py b/synapse/storage/data_stores/main/monthly_active_users.py index b41c3d317a..b8fc28f97b 100644 --- a/synapse/storage/data_stores/main/monthly_active_users.py +++ b/synapse/storage/data_stores/main/monthly_active_users.py @@ -32,7 +32,7 @@ class MonthlyActiveUsersStore(SQLBaseStore): self._clock = hs.get_clock() self.hs = hs # Do not add more reserved users than the total allowable number - self._new_transaction( + self.new_transaction( dbconn, "initialise_mau_threepids", [], @@ -261,7 +261,7 @@ class MonthlyActiveUsersStore(SQLBaseStore): # never be a big table and alternative approaches (batching multiple # upserts into a single txn) introduced a lot of extra complexity. # See https://github.com/matrix-org/synapse/issues/3854 for more - is_insert = self._simple_upsert_txn( + is_insert = self.simple_upsert_txn( txn, table="monthly_active_users", keyvalues={"user_id": user_id}, @@ -281,7 +281,7 @@ class MonthlyActiveUsersStore(SQLBaseStore): """ - return self._simple_select_one_onecol( + return self.simple_select_one_onecol( table="monthly_active_users", keyvalues={"user_id": user_id}, retcol="timestamp", diff --git a/synapse/storage/data_stores/main/openid.py b/synapse/storage/data_stores/main/openid.py index 79b40044d9..650e49750e 100644 --- a/synapse/storage/data_stores/main/openid.py +++ b/synapse/storage/data_stores/main/openid.py @@ -3,7 +3,7 @@ from synapse.storage._base import SQLBaseStore class OpenIdStore(SQLBaseStore): def insert_open_id_token(self, token, ts_valid_until_ms, user_id): - return self._simple_insert( + return self.simple_insert( table="open_id_tokens", values={ "token": token, diff --git a/synapse/storage/data_stores/main/presence.py b/synapse/storage/data_stores/main/presence.py index 523ed6575e..a5e121efd1 100644 --- a/synapse/storage/data_stores/main/presence.py +++ b/synapse/storage/data_stores/main/presence.py @@ -46,7 +46,7 @@ class PresenceStore(SQLBaseStore): txn.call_after(self._get_presence_for_user.invalidate, (state.user_id,)) # Actually insert new rows - self._simple_insert_many_txn( + self.simple_insert_many_txn( txn, table="presence_stream", values=[ @@ -103,7 +103,7 @@ class PresenceStore(SQLBaseStore): inlineCallbacks=True, ) def get_presence_for_users(self, user_ids): - rows = yield self._simple_select_many_batch( + rows = yield self.simple_select_many_batch( table="presence_stream", column="user_id", iterable=user_ids, @@ -129,7 +129,7 @@ class PresenceStore(SQLBaseStore): return self._presence_id_gen.get_current_token() def allow_presence_visible(self, observed_localpart, observer_userid): - return self._simple_insert( + return self.simple_insert( table="presence_allow_inbound", values={ "observed_user_id": observed_localpart, @@ -140,7 +140,7 @@ class PresenceStore(SQLBaseStore): ) def disallow_presence_visible(self, observed_localpart, observer_userid): - return self._simple_delete_one( + return self.simple_delete_one( table="presence_allow_inbound", keyvalues={ "observed_user_id": observed_localpart, diff --git a/synapse/storage/data_stores/main/profile.py b/synapse/storage/data_stores/main/profile.py index e4e8a1c1d6..c8b5b60301 100644 --- a/synapse/storage/data_stores/main/profile.py +++ b/synapse/storage/data_stores/main/profile.py @@ -24,7 +24,7 @@ class ProfileWorkerStore(SQLBaseStore): @defer.inlineCallbacks def get_profileinfo(self, user_localpart): try: - profile = yield self._simple_select_one( + profile = yield self.simple_select_one( table="profiles", keyvalues={"user_id": user_localpart}, retcols=("displayname", "avatar_url"), @@ -42,7 +42,7 @@ class ProfileWorkerStore(SQLBaseStore): ) def get_profile_displayname(self, user_localpart): - return self._simple_select_one_onecol( + return self.simple_select_one_onecol( table="profiles", keyvalues={"user_id": user_localpart}, retcol="displayname", @@ -50,7 +50,7 @@ class ProfileWorkerStore(SQLBaseStore): ) def get_profile_avatar_url(self, user_localpart): - return self._simple_select_one_onecol( + return self.simple_select_one_onecol( table="profiles", keyvalues={"user_id": user_localpart}, retcol="avatar_url", @@ -58,7 +58,7 @@ class ProfileWorkerStore(SQLBaseStore): ) def get_from_remote_profile_cache(self, user_id): - return self._simple_select_one( + return self.simple_select_one( table="remote_profile_cache", keyvalues={"user_id": user_id}, retcols=("displayname", "avatar_url"), @@ -67,12 +67,12 @@ class ProfileWorkerStore(SQLBaseStore): ) def create_profile(self, user_localpart): - return self._simple_insert( + return self.simple_insert( table="profiles", values={"user_id": user_localpart}, desc="create_profile" ) def set_profile_displayname(self, user_localpart, new_displayname): - return self._simple_update_one( + return self.simple_update_one( table="profiles", keyvalues={"user_id": user_localpart}, updatevalues={"displayname": new_displayname}, @@ -80,7 +80,7 @@ class ProfileWorkerStore(SQLBaseStore): ) def set_profile_avatar_url(self, user_localpart, new_avatar_url): - return self._simple_update_one( + return self.simple_update_one( table="profiles", keyvalues={"user_id": user_localpart}, updatevalues={"avatar_url": new_avatar_url}, @@ -95,7 +95,7 @@ class ProfileStore(ProfileWorkerStore): This should only be called when `is_subscribed_remote_profile_for_user` would return true for the user. """ - return self._simple_upsert( + return self.simple_upsert( table="remote_profile_cache", keyvalues={"user_id": user_id}, values={ @@ -107,7 +107,7 @@ class ProfileStore(ProfileWorkerStore): ) def update_remote_profile_cache(self, user_id, displayname, avatar_url): - return self._simple_update( + return self.simple_update( table="remote_profile_cache", keyvalues={"user_id": user_id}, values={ @@ -125,7 +125,7 @@ class ProfileStore(ProfileWorkerStore): """ subscribed = yield self.is_subscribed_remote_profile_for_user(user_id) if not subscribed: - yield self._simple_delete( + yield self.simple_delete( table="remote_profile_cache", keyvalues={"user_id": user_id}, desc="delete_remote_profile_cache", @@ -155,7 +155,7 @@ class ProfileStore(ProfileWorkerStore): def is_subscribed_remote_profile_for_user(self, user_id): """Check whether we are interested in a remote user's profile. """ - res = yield self._simple_select_one_onecol( + res = yield self.simple_select_one_onecol( table="group_users", keyvalues={"user_id": user_id}, retcol="user_id", @@ -166,7 +166,7 @@ class ProfileStore(ProfileWorkerStore): if res: return True - res = yield self._simple_select_one_onecol( + res = yield self.simple_select_one_onecol( table="group_invites", keyvalues={"user_id": user_id}, retcol="user_id", diff --git a/synapse/storage/data_stores/main/push_rule.py b/synapse/storage/data_stores/main/push_rule.py index b520062d84..75bd499bcd 100644 --- a/synapse/storage/data_stores/main/push_rule.py +++ b/synapse/storage/data_stores/main/push_rule.py @@ -75,7 +75,7 @@ class PushRulesWorkerStore( def __init__(self, db_conn, hs): super(PushRulesWorkerStore, self).__init__(db_conn, hs) - push_rules_prefill, push_rules_id = self._get_cache_dict( + push_rules_prefill, push_rules_id = self.get_cache_dict( db_conn, "push_rules_stream", entity_column="user_id", @@ -100,7 +100,7 @@ class PushRulesWorkerStore( @cachedInlineCallbacks(max_entries=5000) def get_push_rules_for_user(self, user_id): - rows = yield self._simple_select_list( + rows = yield self.simple_select_list( table="push_rules", keyvalues={"user_name": user_id}, retcols=( @@ -124,7 +124,7 @@ class PushRulesWorkerStore( @cachedInlineCallbacks(max_entries=5000) def get_push_rules_enabled_for_user(self, user_id): - results = yield self._simple_select_list( + results = yield self.simple_select_list( table="push_rules_enable", keyvalues={"user_name": user_id}, retcols=("user_name", "rule_id", "enabled"), @@ -162,7 +162,7 @@ class PushRulesWorkerStore( results = {user_id: [] for user_id in user_ids} - rows = yield self._simple_select_many_batch( + rows = yield self.simple_select_many_batch( table="push_rules", column="user_name", iterable=user_ids, @@ -320,7 +320,7 @@ class PushRulesWorkerStore( results = {user_id: {} for user_id in user_ids} - rows = yield self._simple_select_many_batch( + rows = yield self.simple_select_many_batch( table="push_rules_enable", column="user_name", iterable=user_ids, @@ -395,7 +395,7 @@ class PushRuleStore(PushRulesWorkerStore): relative_to_rule = before or after - res = self._simple_select_one_txn( + res = self.simple_select_one_txn( txn, table="push_rules", keyvalues={"user_name": user_id, "rule_id": relative_to_rule}, @@ -499,7 +499,7 @@ class PushRuleStore(PushRulesWorkerStore): actions_json, update_stream=True, ): - """Specialised version of _simple_upsert_txn that picks a push_rule_id + """Specialised version of simple_upsert_txn that picks a push_rule_id using the _push_rule_id_gen if it needs to insert the rule. It assumes that the "push_rules" table is locked""" @@ -518,7 +518,7 @@ class PushRuleStore(PushRulesWorkerStore): # We didn't update a row with the given rule_id so insert one push_rule_id = self._push_rule_id_gen.get_next() - self._simple_insert_txn( + self.simple_insert_txn( txn, table="push_rules", values={ @@ -561,7 +561,7 @@ class PushRuleStore(PushRulesWorkerStore): """ def delete_push_rule_txn(txn, stream_id, event_stream_ordering): - self._simple_delete_one_txn( + self.simple_delete_one_txn( txn, "push_rules", {"user_name": user_id, "rule_id": rule_id} ) @@ -596,7 +596,7 @@ class PushRuleStore(PushRulesWorkerStore): self, txn, stream_id, event_stream_ordering, user_id, rule_id, enabled ): new_id = self._push_rules_enable_id_gen.get_next() - self._simple_upsert_txn( + self.simple_upsert_txn( txn, "push_rules_enable", {"user_name": user_id, "rule_id": rule_id}, @@ -636,7 +636,7 @@ class PushRuleStore(PushRulesWorkerStore): update_stream=False, ) else: - self._simple_update_one_txn( + self.simple_update_one_txn( txn, "push_rules", {"user_name": user_id, "rule_id": rule_id}, @@ -675,7 +675,7 @@ class PushRuleStore(PushRulesWorkerStore): if data is not None: values.update(data) - self._simple_insert_txn(txn, "push_rules_stream", values=values) + self.simple_insert_txn(txn, "push_rules_stream", values=values) txn.call_after(self.get_push_rules_for_user.invalidate, (user_id,)) txn.call_after(self.get_push_rules_enabled_for_user.invalidate, (user_id,)) diff --git a/synapse/storage/data_stores/main/pusher.py b/synapse/storage/data_stores/main/pusher.py index d76861cdc0..d5a169872b 100644 --- a/synapse/storage/data_stores/main/pusher.py +++ b/synapse/storage/data_stores/main/pusher.py @@ -59,7 +59,7 @@ class PusherWorkerStore(SQLBaseStore): @defer.inlineCallbacks def user_has_pusher(self, user_id): - ret = yield self._simple_select_one_onecol( + ret = yield self.simple_select_one_onecol( "pushers", {"user_name": user_id}, "id", allow_none=True ) return ret is not None @@ -72,7 +72,7 @@ class PusherWorkerStore(SQLBaseStore): @defer.inlineCallbacks def get_pushers_by(self, keyvalues): - ret = yield self._simple_select_list( + ret = yield self.simple_select_list( "pushers", keyvalues, [ @@ -193,7 +193,7 @@ class PusherWorkerStore(SQLBaseStore): inlineCallbacks=True, ) def get_if_users_have_pushers(self, user_ids): - rows = yield self._simple_select_many_batch( + rows = yield self.simple_select_many_batch( table="pushers", column="user_name", iterable=user_ids, @@ -229,8 +229,8 @@ class PusherStore(PusherWorkerStore): ): with self._pushers_id_gen.get_next() as stream_id: # no need to lock because `pushers` has a unique key on - # (app_id, pushkey, user_name) so _simple_upsert will retry - yield self._simple_upsert( + # (app_id, pushkey, user_name) so simple_upsert will retry + yield self.simple_upsert( table="pushers", keyvalues={"app_id": app_id, "pushkey": pushkey, "user_name": user_id}, values={ @@ -269,7 +269,7 @@ class PusherStore(PusherWorkerStore): txn, self.get_if_user_has_pusher, (user_id,) ) - self._simple_delete_one_txn( + self.simple_delete_one_txn( txn, "pushers", {"app_id": app_id, "pushkey": pushkey, "user_name": user_id}, @@ -278,7 +278,7 @@ class PusherStore(PusherWorkerStore): # it's possible for us to end up with duplicate rows for # (app_id, pushkey, user_id) at different stream_ids, but that # doesn't really matter. - self._simple_insert_txn( + self.simple_insert_txn( txn, table="deleted_pushers", values={ @@ -296,7 +296,7 @@ class PusherStore(PusherWorkerStore): def update_pusher_last_stream_ordering( self, app_id, pushkey, user_id, last_stream_ordering ): - yield self._simple_update_one( + yield self.simple_update_one( "pushers", {"app_id": app_id, "pushkey": pushkey, "user_name": user_id}, {"last_stream_ordering": last_stream_ordering}, @@ -319,7 +319,7 @@ class PusherStore(PusherWorkerStore): Returns: Deferred[bool]: True if the pusher still exists; False if it has been deleted. """ - updated = yield self._simple_update( + updated = yield self.simple_update( table="pushers", keyvalues={"app_id": app_id, "pushkey": pushkey, "user_name": user_id}, updatevalues={ @@ -333,7 +333,7 @@ class PusherStore(PusherWorkerStore): @defer.inlineCallbacks def update_pusher_failing_since(self, app_id, pushkey, user_id, failing_since): - yield self._simple_update( + yield self.simple_update( table="pushers", keyvalues={"app_id": app_id, "pushkey": pushkey, "user_name": user_id}, updatevalues={"failing_since": failing_since}, @@ -342,7 +342,7 @@ class PusherStore(PusherWorkerStore): @defer.inlineCallbacks def get_throttle_params_by_room(self, pusher_id): - res = yield self._simple_select_list( + res = yield self.simple_select_list( "pusher_throttle", {"pusher": pusher_id}, ["room_id", "last_sent_ts", "throttle_ms"], @@ -361,8 +361,8 @@ class PusherStore(PusherWorkerStore): @defer.inlineCallbacks def set_throttle_params(self, pusher_id, room_id, params): # no need to lock because `pusher_throttle` has a primary key on - # (pusher, room_id) so _simple_upsert will retry - yield self._simple_upsert( + # (pusher, room_id) so simple_upsert will retry + yield self.simple_upsert( "pusher_throttle", {"pusher": pusher_id, "room_id": room_id}, params, diff --git a/synapse/storage/data_stores/main/receipts.py b/synapse/storage/data_stores/main/receipts.py index 8b17334ff4..380f388e30 100644 --- a/synapse/storage/data_stores/main/receipts.py +++ b/synapse/storage/data_stores/main/receipts.py @@ -61,7 +61,7 @@ class ReceiptsWorkerStore(SQLBaseStore): @cached(num_args=2) def get_receipts_for_room(self, room_id, receipt_type): - return self._simple_select_list( + return self.simple_select_list( table="receipts_linearized", keyvalues={"room_id": room_id, "receipt_type": receipt_type}, retcols=("user_id", "event_id"), @@ -70,7 +70,7 @@ class ReceiptsWorkerStore(SQLBaseStore): @cached(num_args=3) def get_last_receipt_event_id_for_user(self, user_id, room_id, receipt_type): - return self._simple_select_one_onecol( + return self.simple_select_one_onecol( table="receipts_linearized", keyvalues={ "room_id": room_id, @@ -84,7 +84,7 @@ class ReceiptsWorkerStore(SQLBaseStore): @cachedInlineCallbacks(num_args=2) def get_receipts_for_user(self, user_id, receipt_type): - rows = yield self._simple_select_list( + rows = yield self.simple_select_list( table="receipts_linearized", keyvalues={"user_id": user_id, "receipt_type": receipt_type}, retcols=("room_id", "event_id"), @@ -335,7 +335,7 @@ class ReceiptsStore(ReceiptsWorkerStore): otherwise, the rx timestamp of the event that the RR corresponds to (or 0 if the event is unknown) """ - res = self._simple_select_one_txn( + res = self.simple_select_one_txn( txn, table="events", retcols=["stream_ordering", "received_ts"], @@ -388,7 +388,7 @@ class ReceiptsStore(ReceiptsWorkerStore): (user_id, room_id, receipt_type), ) - self._simple_delete_txn( + self.simple_delete_txn( txn, table="receipts_linearized", keyvalues={ @@ -398,7 +398,7 @@ class ReceiptsStore(ReceiptsWorkerStore): }, ) - self._simple_insert_txn( + self.simple_insert_txn( txn, table="receipts_linearized", values={ @@ -514,7 +514,7 @@ class ReceiptsStore(ReceiptsWorkerStore): self._get_linearized_receipts_for_room.invalidate_many, (room_id,) ) - self._simple_delete_txn( + self.simple_delete_txn( txn, table="receipts_graph", keyvalues={ @@ -523,7 +523,7 @@ class ReceiptsStore(ReceiptsWorkerStore): "user_id": user_id, }, ) - self._simple_insert_txn( + self.simple_insert_txn( txn, table="receipts_graph", values={ diff --git a/synapse/storage/data_stores/main/registration.py b/synapse/storage/data_stores/main/registration.py index 653c9318cb..debc6706f5 100644 --- a/synapse/storage/data_stores/main/registration.py +++ b/synapse/storage/data_stores/main/registration.py @@ -45,7 +45,7 @@ class RegistrationWorkerStore(SQLBaseStore): @cached() def get_user_by_id(self, user_id): - return self._simple_select_one( + return self.simple_select_one( table="users", keyvalues={"name": user_id}, retcols=[ @@ -109,7 +109,7 @@ class RegistrationWorkerStore(SQLBaseStore): otherwise int representation of the timestamp (as a number of milliseconds since epoch). """ - res = yield self._simple_select_one_onecol( + res = yield self.simple_select_one_onecol( table="account_validity", keyvalues={"user_id": user_id}, retcol="expiration_ts_ms", @@ -137,7 +137,7 @@ class RegistrationWorkerStore(SQLBaseStore): """ def set_account_validity_for_user_txn(txn): - self._simple_update_txn( + self.simple_update_txn( txn=txn, table="account_validity", keyvalues={"user_id": user_id}, @@ -167,7 +167,7 @@ class RegistrationWorkerStore(SQLBaseStore): Raises: StoreError: The provided token is already set for another user. """ - yield self._simple_update_one( + yield self.simple_update_one( table="account_validity", keyvalues={"user_id": user_id}, updatevalues={"renewal_token": renewal_token}, @@ -184,7 +184,7 @@ class RegistrationWorkerStore(SQLBaseStore): Returns: defer.Deferred[str]: The ID of the user to which the token belongs. """ - res = yield self._simple_select_one_onecol( + res = yield self.simple_select_one_onecol( table="account_validity", keyvalues={"renewal_token": renewal_token}, retcol="user_id", @@ -203,7 +203,7 @@ class RegistrationWorkerStore(SQLBaseStore): Returns: defer.Deferred[str]: The renewal token associated with this user ID. """ - res = yield self._simple_select_one_onecol( + res = yield self.simple_select_one_onecol( table="account_validity", keyvalues={"user_id": user_id}, retcol="renewal_token", @@ -250,7 +250,7 @@ class RegistrationWorkerStore(SQLBaseStore): email_sent (bool): Flag which indicates whether a renewal email has been sent to this user. """ - yield self._simple_update_one( + yield self.simple_update_one( table="account_validity", keyvalues={"user_id": user_id}, updatevalues={"email_sent": email_sent}, @@ -265,7 +265,7 @@ class RegistrationWorkerStore(SQLBaseStore): Args: user_id (str): ID of the user to remove from the account validity table. """ - yield self._simple_delete_one( + yield self.simple_delete_one( table="account_validity", keyvalues={"user_id": user_id}, desc="delete_account_validity_for_user", @@ -281,7 +281,7 @@ class RegistrationWorkerStore(SQLBaseStore): Returns (bool): true iff the user is a server admin, false otherwise. """ - res = yield self._simple_select_one_onecol( + res = yield self.simple_select_one_onecol( table="users", keyvalues={"name": user.to_string()}, retcol="admin", @@ -299,7 +299,7 @@ class RegistrationWorkerStore(SQLBaseStore): admin (bool): true iff the user is to be a server admin, false otherwise. """ - return self._simple_update_one( + return self.simple_update_one( table="users", keyvalues={"name": user.to_string()}, updatevalues={"admin": 1 if admin else 0}, @@ -351,7 +351,7 @@ class RegistrationWorkerStore(SQLBaseStore): return res def is_real_user_txn(self, txn, user_id): - res = self._simple_select_one_onecol_txn( + res = self.simple_select_one_onecol_txn( txn=txn, table="users", keyvalues={"name": user_id}, @@ -361,7 +361,7 @@ class RegistrationWorkerStore(SQLBaseStore): return res is None def is_support_user_txn(self, txn, user_id): - res = self._simple_select_one_onecol_txn( + res = self.simple_select_one_onecol_txn( txn=txn, table="users", keyvalues={"name": user_id}, @@ -394,7 +394,7 @@ class RegistrationWorkerStore(SQLBaseStore): Returns: str|None: the mxid of the user, or None if they are not known """ - return await self._simple_select_one_onecol( + return await self.simple_select_one_onecol( table="user_external_ids", keyvalues={"auth_provider": auth_provider, "external_id": external_id}, retcol="user_id", @@ -536,7 +536,7 @@ class RegistrationWorkerStore(SQLBaseStore): Returns: str|None: user id or None if no user id/threepid mapping exists """ - ret = self._simple_select_one_txn( + ret = self.simple_select_one_txn( txn, "user_threepids", {"medium": medium, "address": address}, @@ -549,7 +549,7 @@ class RegistrationWorkerStore(SQLBaseStore): @defer.inlineCallbacks def user_add_threepid(self, user_id, medium, address, validated_at, added_at): - yield self._simple_upsert( + yield self.simple_upsert( "user_threepids", {"medium": medium, "address": address}, {"user_id": user_id, "validated_at": validated_at, "added_at": added_at}, @@ -557,7 +557,7 @@ class RegistrationWorkerStore(SQLBaseStore): @defer.inlineCallbacks def user_get_threepids(self, user_id): - ret = yield self._simple_select_list( + ret = yield self.simple_select_list( "user_threepids", {"user_id": user_id}, ["medium", "address", "validated_at", "added_at"], @@ -566,7 +566,7 @@ class RegistrationWorkerStore(SQLBaseStore): return ret def user_delete_threepid(self, user_id, medium, address): - return self._simple_delete( + return self.simple_delete( "user_threepids", keyvalues={"user_id": user_id, "medium": medium, "address": address}, desc="user_delete_threepid", @@ -579,7 +579,7 @@ class RegistrationWorkerStore(SQLBaseStore): user_id: The user id to delete all threepids of """ - return self._simple_delete( + return self.simple_delete( "user_threepids", keyvalues={"user_id": user_id}, desc="user_delete_threepids", @@ -601,7 +601,7 @@ class RegistrationWorkerStore(SQLBaseStore): """ # We need to use an upsert, in case they user had already bound the # threepid - return self._simple_upsert( + return self.simple_upsert( table="user_threepid_id_server", keyvalues={ "user_id": user_id, @@ -627,7 +627,7 @@ class RegistrationWorkerStore(SQLBaseStore): medium (str): The medium of the threepid (e.g "email") address (str): The address of the threepid (e.g "bob@example.com") """ - return self._simple_select_list( + return self.simple_select_list( table="user_threepid_id_server", keyvalues={"user_id": user_id}, retcols=["medium", "address"], @@ -648,7 +648,7 @@ class RegistrationWorkerStore(SQLBaseStore): Returns: Deferred """ - return self._simple_delete( + return self.simple_delete( table="user_threepid_id_server", keyvalues={ "user_id": user_id, @@ -671,7 +671,7 @@ class RegistrationWorkerStore(SQLBaseStore): Returns: Deferred[list[str]]: Resolves to a list of identity servers """ - return self._simple_select_onecol( + return self.simple_select_onecol( table="user_threepid_id_server", keyvalues={"user_id": user_id, "medium": medium, "address": address}, retcol="id_server", @@ -689,7 +689,7 @@ class RegistrationWorkerStore(SQLBaseStore): defer.Deferred(bool): The requested value. """ - res = yield self._simple_select_one_onecol( + res = yield self.simple_select_one_onecol( table="users", keyvalues={"name": user_id}, retcol="deactivated", @@ -776,12 +776,12 @@ class RegistrationWorkerStore(SQLBaseStore): """ def delete_threepid_session_txn(txn): - self._simple_delete_txn( + self.simple_delete_txn( txn, table="threepid_validation_token", keyvalues={"session_id": session_id}, ) - self._simple_delete_txn( + self.simple_delete_txn( txn, table="threepid_validation_session", keyvalues={"session_id": session_id}, @@ -961,7 +961,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): """ next_id = self._access_tokens_id_gen.get_next() - yield self._simple_insert( + yield self.simple_insert( "access_tokens", { "id": next_id, @@ -1037,7 +1037,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): # Ensure that the guest user actually exists # ``allow_none=False`` makes this raise an exception # if the row isn't in the database. - self._simple_select_one_txn( + self.simple_select_one_txn( txn, "users", keyvalues={"name": user_id, "is_guest": 1}, @@ -1045,7 +1045,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): allow_none=False, ) - self._simple_update_one_txn( + self.simple_update_one_txn( txn, "users", keyvalues={"name": user_id, "is_guest": 1}, @@ -1059,7 +1059,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): }, ) else: - self._simple_insert_txn( + self.simple_insert_txn( txn, "users", values={ @@ -1114,7 +1114,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): external_id: id on that system user_id: complete mxid that it is mapped to """ - return self._simple_insert( + return self.simple_insert( table="user_external_ids", values={ "auth_provider": auth_provider, @@ -1132,7 +1132,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): """ def user_set_password_hash_txn(txn): - self._simple_update_one_txn( + self.simple_update_one_txn( txn, "users", {"name": user_id}, {"password_hash": password_hash} ) self._invalidate_cache_and_stream(txn, self.get_user_by_id, (user_id,)) @@ -1152,7 +1152,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): """ def f(txn): - self._simple_update_one_txn( + self.simple_update_one_txn( txn, table="users", keyvalues={"name": user_id}, @@ -1176,7 +1176,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): """ def f(txn): - self._simple_update_one_txn( + self.simple_update_one_txn( txn, table="users", keyvalues={"name": user_id}, @@ -1234,7 +1234,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): def delete_access_token(self, access_token): def f(txn): - self._simple_delete_one_txn( + self.simple_delete_one_txn( txn, table="access_tokens", keyvalues={"token": access_token} ) @@ -1246,7 +1246,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): @cachedInlineCallbacks() def is_guest(self, user_id): - res = yield self._simple_select_one_onecol( + res = yield self.simple_select_one_onecol( table="users", keyvalues={"name": user_id}, retcol="is_guest", @@ -1261,7 +1261,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): Adds a user to the table of users who need to be parted from all the rooms they're in """ - return self._simple_insert( + return self.simple_insert( "users_pending_deactivation", values={"user_id": user_id}, desc="add_user_pending_deactivation", @@ -1274,7 +1274,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): """ # XXX: This should be simple_delete_one but we failed to put a unique index on # the table, so somehow duplicate entries have ended up in it. - return self._simple_delete( + return self.simple_delete( "users_pending_deactivation", keyvalues={"user_id": user_id}, desc="del_user_pending_deactivation", @@ -1285,7 +1285,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): Gets one user from the table of users waiting to be parted from all the rooms they're in. """ - return self._simple_select_one_onecol( + return self.simple_select_one_onecol( "users_pending_deactivation", keyvalues={}, retcol="user_id", @@ -1315,7 +1315,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): # Insert everything into a transaction in order to run atomically def validate_threepid_session_txn(txn): - row = self._simple_select_one_txn( + row = self.simple_select_one_txn( txn, table="threepid_validation_session", keyvalues={"session_id": session_id}, @@ -1333,7 +1333,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): 400, "This client_secret does not match the provided session_id" ) - row = self._simple_select_one_txn( + row = self.simple_select_one_txn( txn, table="threepid_validation_token", keyvalues={"session_id": session_id, "token": token}, @@ -1358,7 +1358,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): ) # Looks good. Validate the session - self._simple_update_txn( + self.simple_update_txn( txn, table="threepid_validation_session", keyvalues={"session_id": session_id}, @@ -1401,7 +1401,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): if validated_at: insertion_values["validated_at"] = validated_at - return self._simple_upsert( + return self.simple_upsert( table="threepid_validation_session", keyvalues={"session_id": session_id}, values={"last_send_attempt": send_attempt}, @@ -1439,7 +1439,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): def start_or_continue_validation_session_txn(txn): # Create or update a validation session - self._simple_upsert_txn( + self.simple_upsert_txn( txn, table="threepid_validation_session", keyvalues={"session_id": session_id}, @@ -1452,7 +1452,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): ) # Create a new validation token with this session ID - self._simple_insert_txn( + self.simple_insert_txn( txn, table="threepid_validation_token", values={ @@ -1501,7 +1501,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): ) def set_user_deactivated_status_txn(self, txn, user_id, deactivated): - self._simple_update_one_txn( + self.simple_update_one_txn( txn=txn, table="users", keyvalues={"name": user_id}, @@ -1560,7 +1560,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): expiration_ts, ) - self._simple_upsert_txn( + self.simple_upsert_txn( txn, "account_validity", keyvalues={"user_id": user_id}, diff --git a/synapse/storage/data_stores/main/rejections.py b/synapse/storage/data_stores/main/rejections.py index 7d5de0ea2e..f81f9279a1 100644 --- a/synapse/storage/data_stores/main/rejections.py +++ b/synapse/storage/data_stores/main/rejections.py @@ -22,7 +22,7 @@ logger = logging.getLogger(__name__) class RejectionsStore(SQLBaseStore): def _store_rejections_txn(self, txn, event_id, reason): - self._simple_insert_txn( + self.simple_insert_txn( txn, table="rejections", values={ @@ -33,7 +33,7 @@ class RejectionsStore(SQLBaseStore): ) def get_rejection_reason(self, event_id): - return self._simple_select_one_onecol( + return self.simple_select_one_onecol( table="rejections", retcol="reason", keyvalues={"event_id": event_id}, diff --git a/synapse/storage/data_stores/main/relations.py b/synapse/storage/data_stores/main/relations.py index 858f65582b..aa5e10538b 100644 --- a/synapse/storage/data_stores/main/relations.py +++ b/synapse/storage/data_stores/main/relations.py @@ -352,7 +352,7 @@ class RelationsStore(RelationsWorkerStore): aggregation_key = relation.get("key") - self._simple_insert_txn( + self.simple_insert_txn( txn, table="event_relations", values={ @@ -380,6 +380,6 @@ class RelationsStore(RelationsWorkerStore): redacted_event_id (str): The event that was redacted. """ - self._simple_delete_txn( + self.simple_delete_txn( txn, table="event_relations", keyvalues={"event_id": redacted_event_id} ) diff --git a/synapse/storage/data_stores/main/room.py b/synapse/storage/data_stores/main/room.py index b7f9024811..8f9b6365c1 100644 --- a/synapse/storage/data_stores/main/room.py +++ b/synapse/storage/data_stores/main/room.py @@ -53,7 +53,7 @@ class RoomWorkerStore(SQLBaseStore): Returns: A dict containing the room information, or None if the room is unknown. """ - return self._simple_select_one( + return self.simple_select_one( table="rooms", keyvalues={"room_id": room_id}, retcols=("room_id", "is_public", "creator"), @@ -62,7 +62,7 @@ class RoomWorkerStore(SQLBaseStore): ) def get_public_room_ids(self): - return self._simple_select_onecol( + return self.simple_select_onecol( table="rooms", keyvalues={"is_public": True}, retcol="room_id", @@ -266,7 +266,7 @@ class RoomWorkerStore(SQLBaseStore): @cached(max_entries=10000) def is_room_blocked(self, room_id): - return self._simple_select_one_onecol( + return self.simple_select_one_onecol( table="blocked_rooms", keyvalues={"room_id": room_id}, retcol="1", @@ -287,7 +287,7 @@ class RoomWorkerStore(SQLBaseStore): of RatelimitOverride are None or 0 then ratelimitng has been disabled for that user entirely. """ - row = yield self._simple_select_one( + row = yield self.simple_select_one( table="ratelimit_override", keyvalues={"user_id": user_id}, retcols=("messages_per_second", "burst_count"), @@ -407,7 +407,7 @@ class RoomStore(RoomWorkerStore, SearchStore): ev = json.loads(row["json"]) retention_policy = json.dumps(ev["content"]) - self._simple_insert_txn( + self.simple_insert_txn( txn=txn, table="room_retention", values={ @@ -453,7 +453,7 @@ class RoomStore(RoomWorkerStore, SearchStore): try: def store_room_txn(txn, next_id): - self._simple_insert_txn( + self.simple_insert_txn( txn, "rooms", { @@ -463,7 +463,7 @@ class RoomStore(RoomWorkerStore, SearchStore): }, ) if is_public: - self._simple_insert_txn( + self.simple_insert_txn( txn, table="public_room_list_stream", values={ @@ -482,14 +482,14 @@ class RoomStore(RoomWorkerStore, SearchStore): @defer.inlineCallbacks def set_room_is_public(self, room_id, is_public): def set_room_is_public_txn(txn, next_id): - self._simple_update_one_txn( + self.simple_update_one_txn( txn, table="rooms", keyvalues={"room_id": room_id}, updatevalues={"is_public": is_public}, ) - entries = self._simple_select_list_txn( + entries = self.simple_select_list_txn( txn, table="public_room_list_stream", keyvalues={ @@ -507,7 +507,7 @@ class RoomStore(RoomWorkerStore, SearchStore): add_to_stream = bool(entries[-1]["visibility"]) != is_public if add_to_stream: - self._simple_insert_txn( + self.simple_insert_txn( txn, table="public_room_list_stream", values={ @@ -547,7 +547,7 @@ class RoomStore(RoomWorkerStore, SearchStore): def set_room_is_public_appservice_txn(txn, next_id): if is_public: try: - self._simple_insert_txn( + self.simple_insert_txn( txn, table="appservice_room_list", values={ @@ -560,7 +560,7 @@ class RoomStore(RoomWorkerStore, SearchStore): # We've already inserted, nothing to do. return else: - self._simple_delete_txn( + self.simple_delete_txn( txn, table="appservice_room_list", keyvalues={ @@ -570,7 +570,7 @@ class RoomStore(RoomWorkerStore, SearchStore): }, ) - entries = self._simple_select_list_txn( + entries = self.simple_select_list_txn( txn, table="public_room_list_stream", keyvalues={ @@ -588,7 +588,7 @@ class RoomStore(RoomWorkerStore, SearchStore): add_to_stream = bool(entries[-1]["visibility"]) != is_public if add_to_stream: - self._simple_insert_txn( + self.simple_insert_txn( txn, table="public_room_list_stream", values={ @@ -652,7 +652,7 @@ class RoomStore(RoomWorkerStore, SearchStore): # Ignore the event if one of the value isn't an integer. return - self._simple_insert_txn( + self.simple_insert_txn( txn=txn, table="room_retention", values={ @@ -671,7 +671,7 @@ class RoomStore(RoomWorkerStore, SearchStore): self, room_id, event_id, user_id, reason, content, received_ts ): next_id = self._event_reports_id_gen.get_next() - return self._simple_insert( + return self.simple_insert( table="event_reports", values={ "id": next_id, @@ -717,7 +717,7 @@ class RoomStore(RoomWorkerStore, SearchStore): Returns: Deferred """ - yield self._simple_upsert( + yield self.simple_upsert( table="blocked_rooms", keyvalues={"room_id": room_id}, values={}, diff --git a/synapse/storage/data_stores/main/roommember.py b/synapse/storage/data_stores/main/roommember.py index b314d75941..fe2428a281 100644 --- a/synapse/storage/data_stores/main/roommember.py +++ b/synapse/storage/data_stores/main/roommember.py @@ -128,7 +128,7 @@ class RoomMemberWorkerStore(EventsWorkerStore): membership column is up to date """ - pending_update = self._simple_select_one_txn( + pending_update = self.simple_select_one_txn( txn, table="background_updates", keyvalues={"update_name": _CURRENT_STATE_MEMBERSHIP_UPDATE_NAME}, @@ -603,7 +603,7 @@ class RoomMemberWorkerStore(EventsWorkerStore): to `user_id` and ProfileInfo (or None if not join event). """ - rows = yield self._simple_select_many_batch( + rows = yield self.simple_select_many_batch( table="room_memberships", column="event_id", iterable=event_ids, @@ -643,7 +643,7 @@ class RoomMemberWorkerStore(EventsWorkerStore): # the returned user actually has the correct domain. like_clause = "%:" + host - rows = yield self._execute("is_host_joined", None, sql, room_id, like_clause) + rows = yield self.execute("is_host_joined", None, sql, room_id, like_clause) if not rows: return False @@ -683,7 +683,7 @@ class RoomMemberWorkerStore(EventsWorkerStore): # the returned user actually has the correct domain. like_clause = "%:" + host - rows = yield self._execute("was_host_joined", None, sql, room_id, like_clause) + rows = yield self.execute("was_host_joined", None, sql, room_id, like_clause) if not rows: return False @@ -805,7 +805,7 @@ class RoomMemberWorkerStore(EventsWorkerStore): Deferred[set[str]]: Set of room IDs. """ - room_ids = yield self._simple_select_onecol( + room_ids = yield self.simple_select_onecol( table="room_memberships", keyvalues={"membership": Membership.JOIN, "user_id": user_id}, retcol="room_id", @@ -820,7 +820,7 @@ class RoomMemberWorkerStore(EventsWorkerStore): """Get user_id and membership of a set of event IDs. """ - return self._simple_select_many_batch( + return self.simple_select_many_batch( table="room_memberships", column="event_id", iterable=member_event_ids, @@ -990,7 +990,7 @@ class RoomMemberStore(RoomMemberWorkerStore, RoomMemberBackgroundUpdateStore): def _store_room_members_txn(self, txn, events, backfilled): """Store a room member in the database. """ - self._simple_insert_many_txn( + self.simple_insert_many_txn( txn, table="room_memberships", values=[ @@ -1028,7 +1028,7 @@ class RoomMemberStore(RoomMemberWorkerStore, RoomMemberBackgroundUpdateStore): is_mine = self.hs.is_mine_id(event.state_key) if is_new_state and is_mine: if event.membership == Membership.INVITE: - self._simple_insert_txn( + self.simple_insert_txn( txn, table="local_invites", values={ diff --git a/synapse/storage/data_stores/main/search.py b/synapse/storage/data_stores/main/search.py index d1d7c6863d..f735cf095c 100644 --- a/synapse/storage/data_stores/main/search.py +++ b/synapse/storage/data_stores/main/search.py @@ -441,7 +441,7 @@ class SearchStore(SearchBackgroundUpdateStore): # entire table from the database. sql += " ORDER BY rank DESC LIMIT 500" - results = yield self._execute("search_msgs", self.cursor_to_dict, sql, *args) + results = yield self.execute("search_msgs", self.cursor_to_dict, sql, *args) results = list(filter(lambda row: row["room_id"] in room_ids, results)) @@ -455,7 +455,7 @@ class SearchStore(SearchBackgroundUpdateStore): count_sql += " GROUP BY room_id" - count_results = yield self._execute( + count_results = yield self.execute( "search_rooms_count", self.cursor_to_dict, count_sql, *count_args ) @@ -586,7 +586,7 @@ class SearchStore(SearchBackgroundUpdateStore): args.append(limit) - results = yield self._execute("search_rooms", self.cursor_to_dict, sql, *args) + results = yield self.execute("search_rooms", self.cursor_to_dict, sql, *args) results = list(filter(lambda row: row["room_id"] in room_ids, results)) @@ -600,7 +600,7 @@ class SearchStore(SearchBackgroundUpdateStore): count_sql += " GROUP BY room_id" - count_results = yield self._execute( + count_results = yield self.execute( "search_rooms_count", self.cursor_to_dict, count_sql, *count_args ) diff --git a/synapse/storage/data_stores/main/signatures.py b/synapse/storage/data_stores/main/signatures.py index 556191b76f..f3da29ce14 100644 --- a/synapse/storage/data_stores/main/signatures.py +++ b/synapse/storage/data_stores/main/signatures.py @@ -98,4 +98,4 @@ class SignatureStore(SignatureWorkerStore): } ) - self._simple_insert_many_txn(txn, table="event_reference_hashes", values=vals) + self.simple_insert_many_txn(txn, table="event_reference_hashes", values=vals) diff --git a/synapse/storage/data_stores/main/state.py b/synapse/storage/data_stores/main/state.py index 6a90daea31..2b33ec1a35 100644 --- a/synapse/storage/data_stores/main/state.py +++ b/synapse/storage/data_stores/main/state.py @@ -89,7 +89,7 @@ class StateGroupBackgroundUpdateStore(SQLBaseStore): count = 0 while next_group: - next_group = self._simple_select_one_onecol_txn( + next_group = self.simple_select_one_onecol_txn( txn, table="state_group_edges", keyvalues={"state_group": next_group}, @@ -192,7 +192,7 @@ class StateGroupBackgroundUpdateStore(SQLBaseStore): ): break - next_group = self._simple_select_one_onecol_txn( + next_group = self.simple_select_one_onecol_txn( txn, table="state_group_edges", keyvalues={"state_group": next_group}, @@ -431,7 +431,7 @@ class StateGroupWorkerStore( """ def _get_state_group_delta_txn(txn): - prev_group = self._simple_select_one_onecol_txn( + prev_group = self.simple_select_one_onecol_txn( txn, table="state_group_edges", keyvalues={"state_group": state_group}, @@ -442,7 +442,7 @@ class StateGroupWorkerStore( if not prev_group: return _GetStateGroupDelta(None, None) - delta_ids = self._simple_select_list_txn( + delta_ids = self.simple_select_list_txn( txn, table="state_groups_state", keyvalues={"state_group": state_group}, @@ -644,7 +644,7 @@ class StateGroupWorkerStore( @cached(max_entries=50000) def _get_state_group_for_event(self, event_id): - return self._simple_select_one_onecol( + return self.simple_select_one_onecol( table="event_to_state_groups", keyvalues={"event_id": event_id}, retcol="state_group", @@ -661,7 +661,7 @@ class StateGroupWorkerStore( def _get_state_group_for_events(self, event_ids): """Returns mapping event_id -> state_group """ - rows = yield self._simple_select_many_batch( + rows = yield self.simple_select_many_batch( table="event_to_state_groups", column="event_id", iterable=event_ids, @@ -902,7 +902,7 @@ class StateGroupWorkerStore( state_group = self.database_engine.get_next_state_group_id(txn) - self._simple_insert_txn( + self.simple_insert_txn( txn, table="state_groups", values={"id": state_group, "room_id": room_id, "event_id": event_id}, @@ -911,7 +911,7 @@ class StateGroupWorkerStore( # We persist as a delta if we can, while also ensuring the chain # of deltas isn't tooo long, as otherwise read performance degrades. if prev_group: - is_in_db = self._simple_select_one_onecol_txn( + is_in_db = self.simple_select_one_onecol_txn( txn, table="state_groups", keyvalues={"id": prev_group}, @@ -926,13 +926,13 @@ class StateGroupWorkerStore( potential_hops = self._count_state_group_hops_txn(txn, prev_group) if prev_group and potential_hops < MAX_STATE_DELTA_HOPS: - self._simple_insert_txn( + self.simple_insert_txn( txn, table="state_group_edges", values={"state_group": state_group, "prev_state_group": prev_group}, ) - self._simple_insert_many_txn( + self.simple_insert_many_txn( txn, table="state_groups_state", values=[ @@ -947,7 +947,7 @@ class StateGroupWorkerStore( ], ) else: - self._simple_insert_many_txn( + self.simple_insert_many_txn( txn, table="state_groups_state", values=[ @@ -1007,7 +1007,7 @@ class StateGroupWorkerStore( referenced. """ - rows = yield self._simple_select_many_batch( + rows = yield self.simple_select_many_batch( table="event_to_state_groups", column="state_group", iterable=state_groups, @@ -1065,7 +1065,7 @@ class StateBackgroundUpdateStore( batch_size = max(1, int(batch_size / BATCH_SIZE_SCALE_FACTOR)) if max_group is None: - rows = yield self._execute( + rows = yield self.execute( "_background_deduplicate_state", None, "SELECT coalesce(max(id), 0) FROM state_groups", @@ -1135,13 +1135,13 @@ class StateBackgroundUpdateStore( if prev_state.get(key, None) != value } - self._simple_delete_txn( + self.simple_delete_txn( txn, table="state_group_edges", keyvalues={"state_group": state_group}, ) - self._simple_insert_txn( + self.simple_insert_txn( txn, table="state_group_edges", values={ @@ -1150,13 +1150,13 @@ class StateBackgroundUpdateStore( }, ) - self._simple_delete_txn( + self.simple_delete_txn( txn, table="state_groups_state", keyvalues={"state_group": state_group}, ) - self._simple_insert_many_txn( + self.simple_insert_many_txn( txn, table="state_groups_state", values=[ @@ -1263,7 +1263,7 @@ class StateStore(StateGroupWorkerStore, StateBackgroundUpdateStore): state_groups[event.event_id] = context.state_group - self._simple_insert_many_txn( + self.simple_insert_many_txn( txn, table="event_to_state_groups", values=[ diff --git a/synapse/storage/data_stores/main/state_deltas.py b/synapse/storage/data_stores/main/state_deltas.py index 28f33ec18f..03b908026b 100644 --- a/synapse/storage/data_stores/main/state_deltas.py +++ b/synapse/storage/data_stores/main/state_deltas.py @@ -105,7 +105,7 @@ class StateDeltasStore(SQLBaseStore): ) def _get_max_stream_id_in_current_state_deltas_txn(self, txn): - return self._simple_select_one_onecol_txn( + return self.simple_select_one_onecol_txn( txn, table="current_state_delta_stream", keyvalues={}, diff --git a/synapse/storage/data_stores/main/stats.py b/synapse/storage/data_stores/main/stats.py index 45b3de7d56..3aeba859fd 100644 --- a/synapse/storage/data_stores/main/stats.py +++ b/synapse/storage/data_stores/main/stats.py @@ -186,7 +186,7 @@ class StatsStore(StateDeltasStore): """ Returns the stats processor positions. """ - return self._simple_select_one_onecol( + return self.simple_select_one_onecol( table="stats_incremental_position", keyvalues={}, retcol="stream_id", @@ -215,7 +215,7 @@ class StatsStore(StateDeltasStore): if field and "\0" in field: fields[col] = None - return self._simple_upsert( + return self.simple_upsert( table="room_stats_state", keyvalues={"room_id": room_id}, values=fields, @@ -257,7 +257,7 @@ class StatsStore(StateDeltasStore): ABSOLUTE_STATS_FIELDS[stats_type] + PER_SLICE_FIELDS[stats_type] ) - slice_list = self._simple_select_list_paginate_txn( + slice_list = self.simple_select_list_paginate_txn( txn, table + "_historical", {id_col: stats_id}, @@ -282,7 +282,7 @@ class StatsStore(StateDeltasStore): "name", "topic", "canonical_alias", "avatar", "join_rules", "history_visibility" """ - return self._simple_select_one( + return self.simple_select_one( "room_stats_state", {"room_id": room_id}, retcols=( @@ -308,7 +308,7 @@ class StatsStore(StateDeltasStore): """ table, id_col = TYPE_TO_TABLE[stats_type] - return self._simple_select_one_onecol( + return self.simple_select_one_onecol( "%s_current" % (table,), keyvalues={id_col: id}, retcol="completed_delta_stream_id", @@ -344,7 +344,7 @@ class StatsStore(StateDeltasStore): complete_with_stream_id=stream_id, ) - self._simple_update_one_txn( + self.simple_update_one_txn( txn, table="stats_incremental_position", keyvalues={}, @@ -517,17 +517,17 @@ class StatsStore(StateDeltasStore): else: self.database_engine.lock_table(txn, table) retcols = list(chain(absolutes.keys(), additive_relatives.keys())) - current_row = self._simple_select_one_txn( + current_row = self.simple_select_one_txn( txn, table, keyvalues, retcols, allow_none=True ) if current_row is None: merged_dict = {**keyvalues, **absolutes, **additive_relatives} - self._simple_insert_txn(txn, table, merged_dict) + self.simple_insert_txn(txn, table, merged_dict) else: for (key, val) in additive_relatives.items(): current_row[key] += val current_row.update(absolutes) - self._simple_update_one_txn(txn, table, keyvalues, current_row) + self.simple_update_one_txn(txn, table, keyvalues, current_row) def _upsert_copy_from_table_with_additive_relatives_txn( self, @@ -614,11 +614,11 @@ class StatsStore(StateDeltasStore): txn.execute(sql, qargs) else: self.database_engine.lock_table(txn, into_table) - src_row = self._simple_select_one_txn( + src_row = self.simple_select_one_txn( txn, src_table, keyvalues, copy_columns ) all_dest_keyvalues = {**keyvalues, **extra_dst_keyvalues} - dest_current_row = self._simple_select_one_txn( + dest_current_row = self.simple_select_one_txn( txn, into_table, keyvalues=all_dest_keyvalues, @@ -634,11 +634,11 @@ class StatsStore(StateDeltasStore): **src_row, **additive_relatives, } - self._simple_insert_txn(txn, into_table, merged_dict) + self.simple_insert_txn(txn, into_table, merged_dict) else: for (key, val) in additive_relatives.items(): src_row[key] = dest_current_row[key] + val - self._simple_update_txn(txn, into_table, all_dest_keyvalues, src_row) + self.simple_update_txn(txn, into_table, all_dest_keyvalues, src_row) def get_changes_room_total_events_and_bytes(self, min_pos, max_pos): """Fetches the counts of events in the given range of stream IDs. @@ -735,7 +735,7 @@ class StatsStore(StateDeltasStore): def _fetch_current_state_stats(txn): pos = self.get_room_max_stream_ordering() - rows = self._simple_select_many_txn( + rows = self.simple_select_many_txn( txn, table="current_state_events", column="type", diff --git a/synapse/storage/data_stores/main/stream.py b/synapse/storage/data_stores/main/stream.py index 21a410afd0..60487c4559 100644 --- a/synapse/storage/data_stores/main/stream.py +++ b/synapse/storage/data_stores/main/stream.py @@ -255,7 +255,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): super(StreamWorkerStore, self).__init__(db_conn, hs) events_max = self.get_room_max_stream_ordering() - event_cache_prefill, min_event_val = self._get_cache_dict( + event_cache_prefill, min_event_val = self.get_cache_dict( db_conn, "events", entity_column="room_id", @@ -576,7 +576,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): Returns: A deferred "s%d" stream token. """ - return self._simple_select_one_onecol( + return self.simple_select_one_onecol( table="events", keyvalues={"event_id": event_id}, retcol="stream_ordering" ).addCallback(lambda row: "s%d" % (row,)) @@ -589,7 +589,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): Returns: A deferred "t%d-%d" topological token. """ - return self._simple_select_one( + return self.simple_select_one( table="events", keyvalues={"event_id": event_id}, retcols=("stream_ordering", "topological_ordering"), @@ -613,7 +613,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): "SELECT coalesce(max(topological_ordering), 0) FROM events" " WHERE room_id = ? AND stream_ordering < ?" ) - return self._execute( + return self.execute( "get_max_topological_token", None, sql, room_id, stream_key ).addCallback(lambda r: r[0][0] if r else 0) @@ -709,7 +709,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): dict """ - results = self._simple_select_one_txn( + results = self.simple_select_one_txn( txn, "events", keyvalues={"event_id": event_id, "room_id": room_id}, @@ -797,7 +797,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): return upper_bound, events def get_federation_out_pos(self, typ): - return self._simple_select_one_onecol( + return self.simple_select_one_onecol( table="federation_stream_position", retcol="stream_id", keyvalues={"type": typ}, @@ -805,7 +805,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): ) def update_federation_out_pos(self, typ, stream_id): - return self._simple_update_one( + return self.simple_update_one( table="federation_stream_position", keyvalues={"type": typ}, updatevalues={"stream_id": stream_id}, diff --git a/synapse/storage/data_stores/main/tags.py b/synapse/storage/data_stores/main/tags.py index aa24339717..85012403be 100644 --- a/synapse/storage/data_stores/main/tags.py +++ b/synapse/storage/data_stores/main/tags.py @@ -41,7 +41,7 @@ class TagsWorkerStore(AccountDataWorkerStore): tag strings to tag content. """ - deferred = self._simple_select_list( + deferred = self.simple_select_list( "room_tags", {"user_id": user_id}, ["room_id", "tag", "content"] ) @@ -153,7 +153,7 @@ class TagsWorkerStore(AccountDataWorkerStore): Returns: A deferred list of string tags. """ - return self._simple_select_list( + return self.simple_select_list( table="room_tags", keyvalues={"user_id": user_id, "room_id": room_id}, retcols=("tag", "content"), @@ -178,7 +178,7 @@ class TagsStore(TagsWorkerStore): content_json = json.dumps(content) def add_tag_txn(txn, next_id): - self._simple_upsert_txn( + self.simple_upsert_txn( txn, table="room_tags", keyvalues={"user_id": user_id, "room_id": room_id, "tag": tag}, diff --git a/synapse/storage/data_stores/main/transactions.py b/synapse/storage/data_stores/main/transactions.py index 01b1be5e14..c162f3ea16 100644 --- a/synapse/storage/data_stores/main/transactions.py +++ b/synapse/storage/data_stores/main/transactions.py @@ -85,7 +85,7 @@ class TransactionStore(SQLBaseStore): ) def _get_received_txn_response(self, txn, transaction_id, origin): - result = self._simple_select_one_txn( + result = self.simple_select_one_txn( txn, table="received_transactions", keyvalues={"transaction_id": transaction_id, "origin": origin}, @@ -119,7 +119,7 @@ class TransactionStore(SQLBaseStore): response_json (str) """ - return self._simple_insert( + return self.simple_insert( table="received_transactions", values={ "transaction_id": transaction_id, @@ -160,7 +160,7 @@ class TransactionStore(SQLBaseStore): return result def _get_destination_retry_timings(self, txn, destination): - result = self._simple_select_one_txn( + result = self.simple_select_one_txn( txn, table="destinations", keyvalues={"destination": destination}, @@ -227,7 +227,7 @@ class TransactionStore(SQLBaseStore): # We need to be careful here as the data may have changed from under us # due to a worker setting the timings. - prev_row = self._simple_select_one_txn( + prev_row = self.simple_select_one_txn( txn, table="destinations", keyvalues={"destination": destination}, @@ -236,7 +236,7 @@ class TransactionStore(SQLBaseStore): ) if not prev_row: - self._simple_insert_txn( + self.simple_insert_txn( txn, table="destinations", values={ @@ -247,7 +247,7 @@ class TransactionStore(SQLBaseStore): }, ) elif retry_interval == 0 or prev_row["retry_interval"] < retry_interval: - self._simple_update_one_txn( + self.simple_update_one_txn( txn, "destinations", keyvalues={"destination": destination}, diff --git a/synapse/storage/data_stores/main/user_directory.py b/synapse/storage/data_stores/main/user_directory.py index 652abe0e6a..1a85aabbfb 100644 --- a/synapse/storage/data_stores/main/user_directory.py +++ b/synapse/storage/data_stores/main/user_directory.py @@ -85,7 +85,7 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore """ txn.execute(sql) rooms = [{"room_id": x[0], "events": x[1]} for x in txn.fetchall()] - self._simple_insert_many_txn(txn, TEMP_TABLE + "_rooms", rooms) + self.simple_insert_many_txn(txn, TEMP_TABLE + "_rooms", rooms) del rooms # If search all users is on, get all the users we want to add. @@ -100,13 +100,13 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore txn.execute("SELECT name FROM users") users = [{"user_id": x[0]} for x in txn.fetchall()] - self._simple_insert_many_txn(txn, TEMP_TABLE + "_users", users) + self.simple_insert_many_txn(txn, TEMP_TABLE + "_users", users) new_pos = yield self.get_max_stream_id_in_current_state_deltas() yield self.runInteraction( "populate_user_directory_temp_build", _make_staging_area ) - yield self._simple_insert(TEMP_TABLE + "_position", {"position": new_pos}) + yield self.simple_insert(TEMP_TABLE + "_position", {"position": new_pos}) yield self._end_background_update("populate_user_directory_createtables") return 1 @@ -116,7 +116,7 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore """ Update the user directory stream position, then clean up the old tables. """ - position = yield self._simple_select_one_onecol( + position = yield self.simple_select_one_onecol( TEMP_TABLE + "_position", None, "position" ) yield self.update_user_directory_stream_pos(position) @@ -243,7 +243,7 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore to_insert.clear() # We've finished a room. Delete it from the table. - yield self._simple_delete_one(TEMP_TABLE + "_rooms", {"room_id": room_id}) + yield self.simple_delete_one(TEMP_TABLE + "_rooms", {"room_id": room_id}) # Update the remaining counter. progress["remaining"] -= 1 yield self.runInteraction( @@ -312,7 +312,7 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore ) # We've finished processing a user. Delete it from the table. - yield self._simple_delete_one(TEMP_TABLE + "_users", {"user_id": user_id}) + yield self.simple_delete_one(TEMP_TABLE + "_users", {"user_id": user_id}) # Update the remaining counter. progress["remaining"] -= 1 yield self.runInteraction( @@ -361,7 +361,7 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore """ def _update_profile_in_user_dir_txn(txn): - new_entry = self._simple_upsert_txn( + new_entry = self.simple_upsert_txn( txn, table="user_directory", keyvalues={"user_id": user_id}, @@ -435,7 +435,7 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore ) elif isinstance(self.database_engine, Sqlite3Engine): value = "%s %s" % (user_id, display_name) if display_name else user_id - self._simple_upsert_txn( + self.simple_upsert_txn( txn, table="user_directory_search", keyvalues={"user_id": user_id}, @@ -462,7 +462,7 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore """ def _add_users_who_share_room_txn(txn): - self._simple_upsert_many_txn( + self.simple_upsert_many_txn( txn, table="users_who_share_private_rooms", key_names=["user_id", "other_user_id", "room_id"], @@ -489,7 +489,7 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore def _add_users_in_public_rooms_txn(txn): - self._simple_upsert_many_txn( + self.simple_upsert_many_txn( txn, table="users_in_public_rooms", key_names=["user_id", "room_id"], @@ -519,7 +519,7 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore @cached() def get_user_in_directory(self, user_id): - return self._simple_select_one( + return self.simple_select_one( table="user_directory", keyvalues={"user_id": user_id}, retcols=("display_name", "avatar_url"), @@ -528,7 +528,7 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore ) def update_user_directory_stream_pos(self, stream_id): - return self._simple_update_one( + return self.simple_update_one( table="user_directory_stream_pos", keyvalues={}, updatevalues={"stream_id": stream_id}, @@ -547,21 +547,21 @@ class UserDirectoryStore(UserDirectoryBackgroundUpdateStore): def remove_from_user_dir(self, user_id): def _remove_from_user_dir_txn(txn): - self._simple_delete_txn( + self.simple_delete_txn( txn, table="user_directory", keyvalues={"user_id": user_id} ) - self._simple_delete_txn( + self.simple_delete_txn( txn, table="user_directory_search", keyvalues={"user_id": user_id} ) - self._simple_delete_txn( + self.simple_delete_txn( txn, table="users_in_public_rooms", keyvalues={"user_id": user_id} ) - self._simple_delete_txn( + self.simple_delete_txn( txn, table="users_who_share_private_rooms", keyvalues={"user_id": user_id}, ) - self._simple_delete_txn( + self.simple_delete_txn( txn, table="users_who_share_private_rooms", keyvalues={"other_user_id": user_id}, @@ -575,14 +575,14 @@ class UserDirectoryStore(UserDirectoryBackgroundUpdateStore): """Get all user_ids that are in the room directory because they're in the given room_id """ - user_ids_share_pub = yield self._simple_select_onecol( + user_ids_share_pub = yield self.simple_select_onecol( table="users_in_public_rooms", keyvalues={"room_id": room_id}, retcol="user_id", desc="get_users_in_dir_due_to_room", ) - user_ids_share_priv = yield self._simple_select_onecol( + user_ids_share_priv = yield self.simple_select_onecol( table="users_who_share_private_rooms", keyvalues={"room_id": room_id}, retcol="other_user_id", @@ -605,17 +605,17 @@ class UserDirectoryStore(UserDirectoryBackgroundUpdateStore): """ def _remove_user_who_share_room_txn(txn): - self._simple_delete_txn( + self.simple_delete_txn( txn, table="users_who_share_private_rooms", keyvalues={"user_id": user_id, "room_id": room_id}, ) - self._simple_delete_txn( + self.simple_delete_txn( txn, table="users_who_share_private_rooms", keyvalues={"other_user_id": user_id, "room_id": room_id}, ) - self._simple_delete_txn( + self.simple_delete_txn( txn, table="users_in_public_rooms", keyvalues={"user_id": user_id, "room_id": room_id}, @@ -636,14 +636,14 @@ class UserDirectoryStore(UserDirectoryBackgroundUpdateStore): Returns: list: user_id """ - rows = yield self._simple_select_onecol( + rows = yield self.simple_select_onecol( table="users_who_share_private_rooms", keyvalues={"user_id": user_id}, retcol="room_id", desc="get_rooms_user_is_in", ) - pub_rows = yield self._simple_select_onecol( + pub_rows = yield self.simple_select_onecol( table="users_in_public_rooms", keyvalues={"user_id": user_id}, retcol="room_id", @@ -674,14 +674,14 @@ class UserDirectoryStore(UserDirectoryBackgroundUpdateStore): ) f2 USING (room_id) """ - rows = yield self._execute( + rows = yield self.execute( "get_rooms_in_common_for_users", None, sql, user_id, other_user_id ) return [room_id for room_id, in rows] def get_user_directory_stream_pos(self): - return self._simple_select_one_onecol( + return self.simple_select_one_onecol( table="user_directory_stream_pos", keyvalues={}, retcol="stream_id", @@ -786,9 +786,7 @@ class UserDirectoryStore(UserDirectoryBackgroundUpdateStore): # This should be unreachable. raise Exception("Unrecognized database engine") - results = yield self._execute( - "search_user_dir", self.cursor_to_dict, sql, *args - ) + results = yield self.execute("search_user_dir", self.cursor_to_dict, sql, *args) limited = len(results) > limit diff --git a/synapse/storage/data_stores/main/user_erasure_store.py b/synapse/storage/data_stores/main/user_erasure_store.py index aa4f0da5f0..37860af070 100644 --- a/synapse/storage/data_stores/main/user_erasure_store.py +++ b/synapse/storage/data_stores/main/user_erasure_store.py @@ -31,7 +31,7 @@ class UserErasureWorkerStore(SQLBaseStore): Returns: Deferred[bool]: True if the user has requested erasure """ - return self._simple_select_onecol( + return self.simple_select_onecol( table="erased_users", keyvalues={"user_id": user_id}, retcol="1", @@ -56,7 +56,7 @@ class UserErasureWorkerStore(SQLBaseStore): # iterate it multiple times, and (b) avoiding duplicates. user_ids = tuple(set(user_ids)) - rows = yield self._simple_select_many_batch( + rows = yield self.simple_select_many_batch( table="erased_users", column="user_id", iterable=user_ids, diff --git a/tests/handlers/test_stats.py b/tests/handlers/test_stats.py index e0075ccd32..380fd0d107 100644 --- a/tests/handlers/test_stats.py +++ b/tests/handlers/test_stats.py @@ -45,13 +45,13 @@ class StatsRoomTests(unittest.HomeserverTestCase): self.store._all_done = False self.get_success( - self.store._simple_insert( + self.store.simple_insert( "background_updates", {"update_name": "populate_stats_prepare", "progress_json": "{}"}, ) ) self.get_success( - self.store._simple_insert( + self.store.simple_insert( "background_updates", { "update_name": "populate_stats_process_rooms", @@ -61,7 +61,7 @@ class StatsRoomTests(unittest.HomeserverTestCase): ) ) self.get_success( - self.store._simple_insert( + self.store.simple_insert( "background_updates", { "update_name": "populate_stats_process_users", @@ -71,7 +71,7 @@ class StatsRoomTests(unittest.HomeserverTestCase): ) ) self.get_success( - self.store._simple_insert( + self.store.simple_insert( "background_updates", { "update_name": "populate_stats_cleanup", @@ -82,7 +82,7 @@ class StatsRoomTests(unittest.HomeserverTestCase): ) def get_all_room_state(self): - return self.store._simple_select_list( + return self.store.simple_select_list( "room_stats_state", None, retcols=("name", "topic", "canonical_alias") ) @@ -96,7 +96,7 @@ class StatsRoomTests(unittest.HomeserverTestCase): end_ts = self.store.quantise_stats_time(self.reactor.seconds() * 1000) return self.get_success( - self.store._simple_select_one( + self.store.simple_select_one( table + "_historical", {id_col: stat_id, end_ts: end_ts}, cols, @@ -180,7 +180,7 @@ class StatsRoomTests(unittest.HomeserverTestCase): self.handler.stats_enabled = True self.store._all_done = False self.get_success( - self.store._simple_update_one( + self.store.simple_update_one( table="stats_incremental_position", keyvalues={}, updatevalues={"stream_id": 0}, @@ -188,7 +188,7 @@ class StatsRoomTests(unittest.HomeserverTestCase): ) self.get_success( - self.store._simple_insert( + self.store.simple_insert( "background_updates", {"update_name": "populate_stats_prepare", "progress_json": "{}"}, ) @@ -205,13 +205,13 @@ class StatsRoomTests(unittest.HomeserverTestCase): # Now do the initial ingestion. self.get_success( - self.store._simple_insert( + self.store.simple_insert( "background_updates", {"update_name": "populate_stats_process_rooms", "progress_json": "{}"}, ) ) self.get_success( - self.store._simple_insert( + self.store.simple_insert( "background_updates", { "update_name": "populate_stats_cleanup", @@ -656,12 +656,12 @@ class StatsRoomTests(unittest.HomeserverTestCase): self.store._all_done = False self.get_success( - self.store._simple_delete( + self.store.simple_delete( "room_stats_current", {"1": 1}, "test_delete_stats" ) ) self.get_success( - self.store._simple_delete( + self.store.simple_delete( "user_stats_current", {"1": 1}, "test_delete_stats" ) ) @@ -675,7 +675,7 @@ class StatsRoomTests(unittest.HomeserverTestCase): self.store._all_done = False self.get_success( - self.store._simple_insert( + self.store.simple_insert( "background_updates", { "update_name": "populate_stats_process_rooms", @@ -685,7 +685,7 @@ class StatsRoomTests(unittest.HomeserverTestCase): ) ) self.get_success( - self.store._simple_insert( + self.store.simple_insert( "background_updates", { "update_name": "populate_stats_process_users", @@ -695,7 +695,7 @@ class StatsRoomTests(unittest.HomeserverTestCase): ) ) self.get_success( - self.store._simple_insert( + self.store.simple_insert( "background_updates", { "update_name": "populate_stats_cleanup", diff --git a/tests/handlers/test_user_directory.py b/tests/handlers/test_user_directory.py index c5e91a8c41..d5b1c5b4ac 100644 --- a/tests/handlers/test_user_directory.py +++ b/tests/handlers/test_user_directory.py @@ -158,7 +158,7 @@ class UserDirectoryTestCase(unittest.HomeserverTestCase): def get_users_in_public_rooms(self): r = self.get_success( - self.store._simple_select_list( + self.store.simple_select_list( "users_in_public_rooms", None, ("user_id", "room_id") ) ) @@ -169,7 +169,7 @@ class UserDirectoryTestCase(unittest.HomeserverTestCase): def get_users_who_share_private_rooms(self): return self.get_success( - self.store._simple_select_list( + self.store.simple_select_list( "users_who_share_private_rooms", None, ["user_id", "other_user_id", "room_id"], @@ -184,7 +184,7 @@ class UserDirectoryTestCase(unittest.HomeserverTestCase): self.store._all_done = False self.get_success( - self.store._simple_insert( + self.store.simple_insert( "background_updates", { "update_name": "populate_user_directory_createtables", @@ -193,7 +193,7 @@ class UserDirectoryTestCase(unittest.HomeserverTestCase): ) ) self.get_success( - self.store._simple_insert( + self.store.simple_insert( "background_updates", { "update_name": "populate_user_directory_process_rooms", @@ -203,7 +203,7 @@ class UserDirectoryTestCase(unittest.HomeserverTestCase): ) ) self.get_success( - self.store._simple_insert( + self.store.simple_insert( "background_updates", { "update_name": "populate_user_directory_process_users", @@ -213,7 +213,7 @@ class UserDirectoryTestCase(unittest.HomeserverTestCase): ) ) self.get_success( - self.store._simple_insert( + self.store.simple_insert( "background_updates", { "update_name": "populate_user_directory_cleanup", diff --git a/tests/rest/admin/test_admin.py b/tests/rest/admin/test_admin.py index 9575058252..124ce0768a 100644 --- a/tests/rest/admin/test_admin.py +++ b/tests/rest/admin/test_admin.py @@ -632,7 +632,7 @@ class PurgeRoomTestCase(unittest.HomeserverTestCase): "state_groups_state", ): count = self.get_success( - self.store._simple_select_one_onecol( + self.store.simple_select_one_onecol( table=table, keyvalues={"room_id": room_id}, retcol="COUNT(*)", diff --git a/tests/storage/test__base.py b/tests/storage/test__base.py index 9b81b536f5..7b7434a468 100644 --- a/tests/storage/test__base.py +++ b/tests/storage/test__base.py @@ -356,7 +356,7 @@ class UpsertManyTests(unittest.HomeserverTestCase): self.get_success( self.storage.runInteraction( "test", - self.storage._simple_upsert_many_txn, + self.storage.simple_upsert_many_txn, self.table_name, key_names, key_values, @@ -367,7 +367,7 @@ class UpsertManyTests(unittest.HomeserverTestCase): # Check results are what we expect res = self.get_success( - self.storage._simple_select_list( + self.storage.simple_select_list( self.table_name, None, ["id, username, value"] ) ) @@ -383,7 +383,7 @@ class UpsertManyTests(unittest.HomeserverTestCase): self.get_success( self.storage.runInteraction( "test", - self.storage._simple_upsert_many_txn, + self.storage.simple_upsert_many_txn, self.table_name, key_names, key_values, @@ -394,7 +394,7 @@ class UpsertManyTests(unittest.HomeserverTestCase): # Check results are what we expect res = self.get_success( - self.storage._simple_select_list( + self.storage.simple_select_list( self.table_name, None, ["id, username, value"] ) ) diff --git a/tests/storage/test_base.py b/tests/storage/test_base.py index c778de1f0c..de5e4a5fce 100644 --- a/tests/storage/test_base.py +++ b/tests/storage/test_base.py @@ -65,7 +65,7 @@ class SQLBaseStoreTestCase(unittest.TestCase): def test_insert_1col(self): self.mock_txn.rowcount = 1 - yield self.datastore._simple_insert( + yield self.datastore.simple_insert( table="tablename", values={"columname": "Value"} ) @@ -77,7 +77,7 @@ class SQLBaseStoreTestCase(unittest.TestCase): def test_insert_3cols(self): self.mock_txn.rowcount = 1 - yield self.datastore._simple_insert( + yield self.datastore.simple_insert( table="tablename", # Use OrderedDict() so we can assert on the SQL generated values=OrderedDict([("colA", 1), ("colB", 2), ("colC", 3)]), @@ -92,7 +92,7 @@ class SQLBaseStoreTestCase(unittest.TestCase): self.mock_txn.rowcount = 1 self.mock_txn.__iter__ = Mock(return_value=iter([("Value",)])) - value = yield self.datastore._simple_select_one_onecol( + value = yield self.datastore.simple_select_one_onecol( table="tablename", keyvalues={"keycol": "TheKey"}, retcol="retcol" ) @@ -106,7 +106,7 @@ class SQLBaseStoreTestCase(unittest.TestCase): self.mock_txn.rowcount = 1 self.mock_txn.fetchone.return_value = (1, 2, 3) - ret = yield self.datastore._simple_select_one( + ret = yield self.datastore.simple_select_one( table="tablename", keyvalues={"keycol": "TheKey"}, retcols=["colA", "colB", "colC"], @@ -122,7 +122,7 @@ class SQLBaseStoreTestCase(unittest.TestCase): self.mock_txn.rowcount = 0 self.mock_txn.fetchone.return_value = None - ret = yield self.datastore._simple_select_one( + ret = yield self.datastore.simple_select_one( table="tablename", keyvalues={"keycol": "Not here"}, retcols=["colA"], @@ -137,7 +137,7 @@ class SQLBaseStoreTestCase(unittest.TestCase): self.mock_txn.__iter__ = Mock(return_value=iter([(1,), (2,), (3,)])) self.mock_txn.description = (("colA", None, None, None, None, None, None),) - ret = yield self.datastore._simple_select_list( + ret = yield self.datastore.simple_select_list( table="tablename", keyvalues={"keycol": "A set"}, retcols=["colA"] ) @@ -150,7 +150,7 @@ class SQLBaseStoreTestCase(unittest.TestCase): def test_update_one_1col(self): self.mock_txn.rowcount = 1 - yield self.datastore._simple_update_one( + yield self.datastore.simple_update_one( table="tablename", keyvalues={"keycol": "TheKey"}, updatevalues={"columnname": "New Value"}, @@ -165,7 +165,7 @@ class SQLBaseStoreTestCase(unittest.TestCase): def test_update_one_4cols(self): self.mock_txn.rowcount = 1 - yield self.datastore._simple_update_one( + yield self.datastore.simple_update_one( table="tablename", keyvalues=OrderedDict([("colA", 1), ("colB", 2)]), updatevalues=OrderedDict([("colC", 3), ("colD", 4)]), @@ -180,7 +180,7 @@ class SQLBaseStoreTestCase(unittest.TestCase): def test_delete_one(self): self.mock_txn.rowcount = 1 - yield self.datastore._simple_delete_one( + yield self.datastore.simple_delete_one( table="tablename", keyvalues={"keycol": "Go away"} ) diff --git a/tests/storage/test_client_ips.py b/tests/storage/test_client_ips.py index afac5dec7f..25bdd2c163 100644 --- a/tests/storage/test_client_ips.py +++ b/tests/storage/test_client_ips.py @@ -81,7 +81,7 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): self.pump(0) result = self.get_success( - self.store._simple_select_list( + self.store.simple_select_list( table="user_ips", keyvalues={"user_id": user_id}, retcols=["access_token", "ip", "user_agent", "device_id", "last_seen"], @@ -112,7 +112,7 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): self.pump(0) result = self.get_success( - self.store._simple_select_list( + self.store.simple_select_list( table="user_ips", keyvalues={"user_id": user_id}, retcols=["access_token", "ip", "user_agent", "device_id", "last_seen"], @@ -218,7 +218,7 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): # But clear the associated entry in devices table self.get_success( - self.store._simple_update( + self.store.simple_update( table="devices", keyvalues={"user_id": user_id, "device_id": "device_id"}, updatevalues={"last_seen": None, "ip": None, "user_agent": None}, @@ -245,7 +245,7 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): # Register the background update to run again. self.get_success( - self.store._simple_insert( + self.store.simple_insert( table="background_updates", values={ "update_name": "devices_last_seen", @@ -297,7 +297,7 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): # We should see that in the DB result = self.get_success( - self.store._simple_select_list( + self.store.simple_select_list( table="user_ips", keyvalues={"user_id": user_id}, retcols=["access_token", "ip", "user_agent", "device_id", "last_seen"], @@ -323,7 +323,7 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): # We should get no results. result = self.get_success( - self.store._simple_select_list( + self.store.simple_select_list( table="user_ips", keyvalues={"user_id": user_id}, retcols=["access_token", "ip", "user_agent", "device_id", "last_seen"], diff --git a/tests/storage/test_event_push_actions.py b/tests/storage/test_event_push_actions.py index b114c6fb1d..2337a1ae46 100644 --- a/tests/storage/test_event_push_actions.py +++ b/tests/storage/test_event_push_actions.py @@ -116,7 +116,7 @@ class EventPushActionsStoreTestCase(tests.unittest.TestCase): yield _inject_actions(6, PlAIN_NOTIF) yield _rotate(7) - yield self.store._simple_delete( + yield self.store.simple_delete( table="event_push_actions", keyvalues={"1": 1}, desc="" ) @@ -135,7 +135,7 @@ class EventPushActionsStoreTestCase(tests.unittest.TestCase): @defer.inlineCallbacks def test_find_first_stream_ordering_after_ts(self): def add_event(so, ts): - return self.store._simple_insert( + return self.store.simple_insert( "events", { "stream_ordering": so, diff --git a/tests/storage/test_redaction.py b/tests/storage/test_redaction.py index 4561c3e383..4930b6777e 100644 --- a/tests/storage/test_redaction.py +++ b/tests/storage/test_redaction.py @@ -338,7 +338,7 @@ class RedactionTestCase(unittest.HomeserverTestCase): ) event_json = self.get_success( - self.store._simple_select_one_onecol( + self.store.simple_select_one_onecol( table="event_json", keyvalues={"event_id": msg_event.event_id}, retcol="json", @@ -356,7 +356,7 @@ class RedactionTestCase(unittest.HomeserverTestCase): self.reactor.advance(60 * 60 * 2) event_json = self.get_success( - self.store._simple_select_one_onecol( + self.store.simple_select_one_onecol( table="event_json", keyvalues={"event_id": msg_event.event_id}, retcol="json", diff --git a/tests/storage/test_roommember.py b/tests/storage/test_roommember.py index 105a0c2b02..d389cf578f 100644 --- a/tests/storage/test_roommember.py +++ b/tests/storage/test_roommember.py @@ -132,7 +132,7 @@ class CurrentStateMembershipUpdateTestCase(unittest.HomeserverTestCase): # Register the background update to run again. self.get_success( - self.store._simple_insert( + self.store.simple_insert( table="background_updates", values={ "update_name": "current_state_events_membership", diff --git a/tests/unittest.py b/tests/unittest.py index 31997a0f31..295573bc46 100644 --- a/tests/unittest.py +++ b/tests/unittest.py @@ -544,7 +544,7 @@ class HomeserverTestCase(TestCase): Add the given event as an extremity to the room. """ self.get_success( - self.hs.get_datastore()._simple_insert( + self.hs.get_datastore().simple_insert( table="event_forward_extremities", values={"room_id": room_id, "event_id": event_id}, desc="test_add_extremity", -- cgit 1.4.1 From 756d4942f5707922f29fe1fdfd945d73a19d7ac3 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 4 Dec 2019 13:52:46 +0000 Subject: Move DB pool and helper functions into dedicated Database class --- scripts/synapse_port_db | 73 +- synapse/app/_base.py | 2 +- synapse/app/user_dir.py | 2 +- synapse/module_api/__init__.py | 2 +- synapse/storage/_base.py | 1468 +------------------ synapse/storage/background_updates.py | 16 +- synapse/storage/data_stores/main/__init__.py | 36 +- synapse/storage/data_stores/main/account_data.py | 26 +- synapse/storage/data_stores/main/appservice.py | 24 +- synapse/storage/data_stores/main/cache.py | 6 +- synapse/storage/data_stores/main/client_ips.py | 24 +- synapse/storage/data_stores/main/deviceinbox.py | 20 +- synapse/storage/data_stores/main/devices.py | 70 +- synapse/storage/data_stores/main/directory.py | 20 +- synapse/storage/data_stores/main/e2e_room_keys.py | 28 +- .../storage/data_stores/main/end_to_end_keys.py | 44 +- .../storage/data_stores/main/event_federation.py | 40 +- .../storage/data_stores/main/event_push_actions.py | 42 +- synapse/storage/data_stores/main/events.py | 90 +- .../storage/data_stores/main/events_bg_updates.py | 24 +- synapse/storage/data_stores/main/events_worker.py | 18 +- synapse/storage/data_stores/main/filtering.py | 4 +- synapse/storage/data_stores/main/group_server.py | 160 +-- synapse/storage/data_stores/main/keys.py | 12 +- .../storage/data_stores/main/media_repository.py | 44 +- .../data_stores/main/monthly_active_users.py | 12 +- synapse/storage/data_stores/main/openid.py | 6 +- synapse/storage/data_stores/main/presence.py | 12 +- synapse/storage/data_stores/main/profile.py | 28 +- synapse/storage/data_stores/main/push_rule.py | 36 +- synapse/storage/data_stores/main/pusher.py | 34 +- synapse/storage/data_stores/main/receipts.py | 36 +- synapse/storage/data_stores/main/registration.py | 164 +-- synapse/storage/data_stores/main/rejections.py | 4 +- synapse/storage/data_stores/main/relations.py | 12 +- synapse/storage/data_stores/main/room.py | 74 +- synapse/storage/data_stores/main/roommember.py | 44 +- synapse/storage/data_stores/main/search.py | 30 +- synapse/storage/data_stores/main/signatures.py | 4 +- synapse/storage/data_stores/main/state.py | 54 +- synapse/storage/data_stores/main/state_deltas.py | 8 +- synapse/storage/data_stores/main/stats.py | 48 +- synapse/storage/data_stores/main/stream.py | 30 +- synapse/storage/data_stores/main/tags.py | 18 +- synapse/storage/data_stores/main/transactions.py | 22 +- synapse/storage/data_stores/main/user_directory.py | 80 +- .../storage/data_stores/main/user_erasure_store.py | 6 +- synapse/storage/database.py | 1485 ++++++++++++++++++++ tests/handlers/test_stats.py | 30 +- tests/handlers/test_user_directory.py | 12 +- tests/rest/admin/test_admin.py | 2 +- tests/storage/test__base.py | 16 +- tests/storage/test_background_update.py | 2 +- tests/storage/test_base.py | 18 +- tests/storage/test_cleanup_extrems.py | 4 +- tests/storage/test_client_ips.py | 12 +- tests/storage/test_event_federation.py | 8 +- tests/storage/test_event_push_actions.py | 12 +- tests/storage/test_monthly_active_users.py | 6 +- tests/storage/test_redaction.py | 4 +- tests/storage/test_roommember.py | 2 +- tests/unittest.py | 2 +- 62 files changed, 2377 insertions(+), 2295 deletions(-) create mode 100644 synapse/storage/database.py (limited to 'tests') diff --git a/scripts/synapse_port_db b/scripts/synapse_port_db index c4cf11d19a..7a2e177d3d 100755 --- a/scripts/synapse_port_db +++ b/scripts/synapse_port_db @@ -173,14 +173,14 @@ class Store( return (yield self.db_pool.runWithConnection(r)) def execute(self, f, *args, **kwargs): - return self.runInteraction(f.__name__, f, *args, **kwargs) + return self.db.runInteraction(f.__name__, f, *args, **kwargs) def execute_sql(self, sql, *args): def r(txn): txn.execute(sql, args) return txn.fetchall() - return self.runInteraction("execute_sql", r) + return self.db.runInteraction("execute_sql", r) def insert_many_txn(self, txn, table, headers, rows): sql = "INSERT INTO %s (%s) VALUES (%s)" % ( @@ -223,7 +223,7 @@ class Porter(object): def setup_table(self, table): if table in APPEND_ONLY_TABLES: # It's safe to just carry on inserting. - row = yield self.postgres_store.simple_select_one( + row = yield self.postgres_store.db.simple_select_one( table="port_from_sqlite3", keyvalues={"table_name": table}, retcols=("forward_rowid", "backward_rowid"), @@ -233,12 +233,14 @@ class Porter(object): total_to_port = None if row is None: if table == "sent_transactions": - forward_chunk, already_ported, total_to_port = ( - yield self._setup_sent_transactions() - ) + ( + forward_chunk, + already_ported, + total_to_port, + ) = yield self._setup_sent_transactions() backward_chunk = 0 else: - yield self.postgres_store.simple_insert( + yield self.postgres_store.db.simple_insert( table="port_from_sqlite3", values={ "table_name": table, @@ -268,7 +270,7 @@ class Porter(object): yield self.postgres_store.execute(delete_all) - yield self.postgres_store.simple_insert( + yield self.postgres_store.db.simple_insert( table="port_from_sqlite3", values={"table_name": table, "forward_rowid": 1, "backward_rowid": 0}, ) @@ -322,7 +324,7 @@ class Porter(object): if table == "user_directory_stream_pos": # We need to make sure there is a single row, `(X, null), as that is # what synapse expects to be there. - yield self.postgres_store.simple_insert( + yield self.postgres_store.db.simple_insert( table=table, values={"stream_id": None} ) self.progress.update(table, table_size) # Mark table as done @@ -363,7 +365,7 @@ class Porter(object): return headers, forward_rows, backward_rows - headers, frows, brows = yield self.sqlite_store.runInteraction("select", r) + headers, frows, brows = yield self.sqlite_store.db.runInteraction("select", r) if frows or brows: if frows: @@ -377,7 +379,7 @@ class Porter(object): def insert(txn): self.postgres_store.insert_many_txn(txn, table, headers[1:], rows) - self.postgres_store.simple_update_one_txn( + self.postgres_store.db.simple_update_one_txn( txn, table="port_from_sqlite3", keyvalues={"table_name": table}, @@ -416,7 +418,7 @@ class Porter(object): return headers, rows - headers, rows = yield self.sqlite_store.runInteraction("select", r) + headers, rows = yield self.sqlite_store.db.runInteraction("select", r) if rows: forward_chunk = rows[-1][0] + 1 @@ -433,8 +435,8 @@ class Porter(object): rows_dict = [] for row in rows: d = dict(zip(headers, row)) - if "\0" in d['value']: - logger.warning('dropping search row %s', d) + if "\0" in d["value"]: + logger.warning("dropping search row %s", d) else: rows_dict.append(d) @@ -454,7 +456,7 @@ class Porter(object): ], ) - self.postgres_store.simple_update_one_txn( + self.postgres_store.db.simple_update_one_txn( txn, table="port_from_sqlite3", keyvalues={"table_name": "event_search"}, @@ -504,17 +506,14 @@ class Porter(object): self.progress.set_state("Preparing %s" % config["name"]) conn = self.setup_db(config, engine) - db_pool = adbapi.ConnectionPool( - config["name"], **config["args"] - ) + db_pool = adbapi.ConnectionPool(config["name"], **config["args"]) hs = MockHomeserver(self.hs_config, engine, conn, db_pool) store = Store(conn, hs) - yield store.runInteraction( - "%s_engine.check_database" % config["name"], - engine.check_database, + yield store.db.runInteraction( + "%s_engine.check_database" % config["name"], engine.check_database, ) return store @@ -541,7 +540,9 @@ class Porter(object): self.sqlite_store = yield self.build_db_store(self.sqlite_config) # Check if all background updates are done, abort if not. - updates_complete = yield self.sqlite_store.has_completed_background_updates() + updates_complete = ( + yield self.sqlite_store.has_completed_background_updates() + ) if not updates_complete: sys.stderr.write( "Pending background updates exist in the SQLite3 database." @@ -582,22 +583,22 @@ class Porter(object): ) try: - yield self.postgres_store.runInteraction("alter_table", alter_table) + yield self.postgres_store.db.runInteraction("alter_table", alter_table) except Exception: # On Error Resume Next pass - yield self.postgres_store.runInteraction( + yield self.postgres_store.db.runInteraction( "create_port_table", create_port_table ) # Step 2. Get tables. self.progress.set_state("Fetching tables") - sqlite_tables = yield self.sqlite_store.simple_select_onecol( + sqlite_tables = yield self.sqlite_store.db.simple_select_onecol( table="sqlite_master", keyvalues={"type": "table"}, retcol="name" ) - postgres_tables = yield self.postgres_store.simple_select_onecol( + postgres_tables = yield self.postgres_store.db.simple_select_onecol( table="information_schema.tables", keyvalues={}, retcol="distinct table_name", @@ -687,11 +688,11 @@ class Porter(object): rows = txn.fetchall() headers = [column[0] for column in txn.description] - ts_ind = headers.index('ts') + ts_ind = headers.index("ts") return headers, [r for r in rows if r[ts_ind] < yesterday] - headers, rows = yield self.sqlite_store.runInteraction("select", r) + headers, rows = yield self.sqlite_store.db.runInteraction("select", r) rows = self._convert_rows("sent_transactions", headers, rows) @@ -724,7 +725,7 @@ class Porter(object): next_chunk = yield self.sqlite_store.execute(get_start_id) next_chunk = max(max_inserted_rowid + 1, next_chunk) - yield self.postgres_store.simple_insert( + yield self.postgres_store.db.simple_insert( table="port_from_sqlite3", values={ "table_name": "sent_transactions", @@ -737,7 +738,7 @@ class Porter(object): txn.execute( "SELECT count(*) FROM sent_transactions" " WHERE ts >= ?", (yesterday,) ) - size, = txn.fetchone() + (size,) = txn.fetchone() return int(size) remaining_count = yield self.sqlite_store.execute(get_sent_table_size) @@ -790,7 +791,7 @@ class Porter(object): next_id = curr_id + 1 txn.execute("ALTER SEQUENCE state_group_id_seq RESTART WITH %s", (next_id,)) - return self.postgres_store.runInteraction("setup_state_group_id_seq", r) + return self.postgres_store.db.runInteraction("setup_state_group_id_seq", r) ############################################## @@ -871,7 +872,7 @@ class CursesProgress(Progress): duration = int(now) - int(self.start_time) minutes, seconds = divmod(duration, 60) - duration_str = '%02dm %02ds' % (minutes, seconds) + duration_str = "%02dm %02ds" % (minutes, seconds) if self.finished: status = "Time spent: %s (Done!)" % (duration_str,) @@ -881,7 +882,7 @@ class CursesProgress(Progress): left = float(self.total_remaining) / self.total_processed est_remaining = (int(now) - self.start_time) * left - est_remaining_str = '%02dm %02ds remaining' % divmod(est_remaining, 60) + est_remaining_str = "%02dm %02ds remaining" % divmod(est_remaining, 60) else: est_remaining_str = "Unknown" status = "Time spent: %s (est. remaining: %s)" % ( @@ -967,7 +968,7 @@ if __name__ == "__main__": description="A script to port an existing synapse SQLite database to" " a new PostgreSQL database." ) - parser.add_argument("-v", action='store_true') + parser.add_argument("-v", action="store_true") parser.add_argument( "--sqlite-database", required=True, @@ -976,12 +977,12 @@ if __name__ == "__main__": ) parser.add_argument( "--postgres-config", - type=argparse.FileType('r'), + type=argparse.FileType("r"), required=True, help="The database config file for the PostgreSQL database", ) parser.add_argument( - "--curses", action='store_true', help="display a curses based progress UI" + "--curses", action="store_true", help="display a curses based progress UI" ) parser.add_argument( diff --git a/synapse/app/_base.py b/synapse/app/_base.py index 2ac7d5c064..9c96816096 100644 --- a/synapse/app/_base.py +++ b/synapse/app/_base.py @@ -269,7 +269,7 @@ def start(hs, listeners=None): # It is now safe to start your Synapse. hs.start_listening(listeners) - hs.get_datastore().start_profiling() + hs.get_datastore().db.start_profiling() setup_sentry(hs) setup_sdnotify(hs) diff --git a/synapse/app/user_dir.py b/synapse/app/user_dir.py index 0fa2b50999..b6d4481725 100644 --- a/synapse/app/user_dir.py +++ b/synapse/app/user_dir.py @@ -64,7 +64,7 @@ class UserDirectorySlaveStore( super(UserDirectorySlaveStore, self).__init__(db_conn, hs) events_max = self._stream_id_gen.get_current_token() - curr_state_delta_prefill, min_curr_state_delta_id = self.get_cache_dict( + curr_state_delta_prefill, min_curr_state_delta_id = self.db.get_cache_dict( db_conn, "current_state_delta_stream", entity_column="room_id", diff --git a/synapse/module_api/__init__.py b/synapse/module_api/__init__.py index 735b882363..305b9b0178 100644 --- a/synapse/module_api/__init__.py +++ b/synapse/module_api/__init__.py @@ -175,4 +175,4 @@ class ModuleApi(object): Returns: Deferred[object]: result of func """ - return self._store.runInteraction(desc, func, *args, **kwargs) + return self._store.db.runInteraction(desc, func, *args, **kwargs) diff --git a/synapse/storage/_base.py b/synapse/storage/_base.py index 9205e550bb..fd5bb3e1de 100644 --- a/synapse/storage/_base.py +++ b/synapse/storage/_base.py @@ -16,1304 +16,28 @@ # limitations under the License. import logging import random -import sys -import time -from typing import Iterable, Tuple -from six import PY2, iteritems, iterkeys, itervalues -from six.moves import builtins, intern, range +from six import PY2 +from six.moves import builtins from canonicaljson import json -from prometheus_client import Histogram -from twisted.internet import defer - -from synapse.api.errors import StoreError -from synapse.logging.context import LoggingContext, make_deferred_yieldable -from synapse.metrics.background_process_metrics import run_as_background_process -from synapse.storage.engines import PostgresEngine, Sqlite3Engine +from synapse.storage.database import LoggingTransaction # noqa: F401 +from synapse.storage.database import make_in_list_sql_clause # noqa: F401 +from synapse.storage.database import Database from synapse.types import get_domain_from_id -from synapse.util.stringutils import exception_to_unicode - -# import a function which will return a monotonic time, in seconds -try: - # on python 3, use time.monotonic, since time.clock can go backwards - from time import monotonic as monotonic_time -except ImportError: - # ... but python 2 doesn't have it - from time import clock as monotonic_time logger = logging.getLogger(__name__) -try: - MAX_TXN_ID = sys.maxint - 1 -except AttributeError: - # python 3 does not have a maximum int value - MAX_TXN_ID = 2 ** 63 - 1 - -sql_logger = logging.getLogger("synapse.storage.SQL") -transaction_logger = logging.getLogger("synapse.storage.txn") -perf_logger = logging.getLogger("synapse.storage.TIME") - -sql_scheduling_timer = Histogram("synapse_storage_schedule_time", "sec") - -sql_query_timer = Histogram("synapse_storage_query_time", "sec", ["verb"]) -sql_txn_timer = Histogram("synapse_storage_transaction_time", "sec", ["desc"]) - - -# Unique indexes which have been added in background updates. Maps from table name -# to the name of the background update which added the unique index to that table. -# -# This is used by the upsert logic to figure out which tables are safe to do a proper -# UPSERT on: until the relevant background update has completed, we -# have to emulate an upsert by locking the table. -# -UNIQUE_INDEX_BACKGROUND_UPDATES = { - "user_ips": "user_ips_device_unique_index", - "device_lists_remote_extremeties": "device_lists_remote_extremeties_unique_idx", - "device_lists_remote_cache": "device_lists_remote_cache_unique_idx", - "event_search": "event_search_event_id_idx", -} - - -class LoggingTransaction(object): - """An object that almost-transparently proxies for the 'txn' object - passed to the constructor. Adds logging and metrics to the .execute() - method. - - Args: - txn: The database transcation object to wrap. - name (str): The name of this transactions for logging. - database_engine (Sqlite3Engine|PostgresEngine) - after_callbacks(list|None): A list that callbacks will be appended to - that have been added by `call_after` which should be run on - successful completion of the transaction. None indicates that no - callbacks should be allowed to be scheduled to run. - exception_callbacks(list|None): A list that callbacks will be appended - to that have been added by `call_on_exception` which should be run - if transaction ends with an error. None indicates that no callbacks - should be allowed to be scheduled to run. - """ - - __slots__ = [ - "txn", - "name", - "database_engine", - "after_callbacks", - "exception_callbacks", - ] - - def __init__( - self, txn, name, database_engine, after_callbacks=None, exception_callbacks=None - ): - object.__setattr__(self, "txn", txn) - object.__setattr__(self, "name", name) - object.__setattr__(self, "database_engine", database_engine) - object.__setattr__(self, "after_callbacks", after_callbacks) - object.__setattr__(self, "exception_callbacks", exception_callbacks) - - def call_after(self, callback, *args, **kwargs): - """Call the given callback on the main twisted thread after the - transaction has finished. Used to invalidate the caches on the - correct thread. - """ - self.after_callbacks.append((callback, args, kwargs)) - - def call_on_exception(self, callback, *args, **kwargs): - self.exception_callbacks.append((callback, args, kwargs)) - - def __getattr__(self, name): - return getattr(self.txn, name) - - def __setattr__(self, name, value): - setattr(self.txn, name, value) - - def __iter__(self): - return self.txn.__iter__() - - def execute_batch(self, sql, args): - if isinstance(self.database_engine, PostgresEngine): - from psycopg2.extras import execute_batch - - self._do_execute(lambda *x: execute_batch(self.txn, *x), sql, args) - else: - for val in args: - self.execute(sql, val) - - def execute(self, sql, *args): - self._do_execute(self.txn.execute, sql, *args) - - def executemany(self, sql, *args): - self._do_execute(self.txn.executemany, sql, *args) - - def _make_sql_one_line(self, sql): - "Strip newlines out of SQL so that the loggers in the DB are on one line" - return " ".join(l.strip() for l in sql.splitlines() if l.strip()) - - def _do_execute(self, func, sql, *args): - sql = self._make_sql_one_line(sql) - - # TODO(paul): Maybe use 'info' and 'debug' for values? - sql_logger.debug("[SQL] {%s} %s", self.name, sql) - - sql = self.database_engine.convert_param_style(sql) - if args: - try: - sql_logger.debug("[SQL values] {%s} %r", self.name, args[0]) - except Exception: - # Don't let logging failures stop SQL from working - pass - - start = time.time() - - try: - return func(sql, *args) - except Exception as e: - logger.debug("[SQL FAIL] {%s} %s", self.name, e) - raise - finally: - secs = time.time() - start - sql_logger.debug("[SQL time] {%s} %f sec", self.name, secs) - sql_query_timer.labels(sql.split()[0]).observe(secs) - - -class PerformanceCounters(object): - def __init__(self): - self.current_counters = {} - self.previous_counters = {} - - def update(self, key, duration_secs): - count, cum_time = self.current_counters.get(key, (0, 0)) - count += 1 - cum_time += duration_secs - self.current_counters[key] = (count, cum_time) - - def interval(self, interval_duration_secs, limit=3): - counters = [] - for name, (count, cum_time) in iteritems(self.current_counters): - prev_count, prev_time = self.previous_counters.get(name, (0, 0)) - counters.append( - ( - (cum_time - prev_time) / interval_duration_secs, - count - prev_count, - name, - ) - ) - - self.previous_counters = dict(self.current_counters) - - counters.sort(reverse=True) - - top_n_counters = ", ".join( - "%s(%d): %.3f%%" % (name, count, 100 * ratio) - for ratio, count, name in counters[:limit] - ) - - return top_n_counters - class SQLBaseStore(object): - _TXN_ID = 0 - def __init__(self, db_conn, hs): self.hs = hs self._clock = hs.get_clock() - self._db_pool = hs.get_db_pool() - - self._previous_txn_total_time = 0 - self._current_txn_total_time = 0 - self._previous_loop_ts = 0 - - # TODO(paul): These can eventually be removed once the metrics code - # is running in mainline, and we have some nice monitoring frontends - # to watch it - self._txn_perf_counters = PerformanceCounters() - self.database_engine = hs.database_engine - - # A set of tables that are not safe to use native upserts in. - self._unsafe_to_upsert_tables = set(UNIQUE_INDEX_BACKGROUND_UPDATES.keys()) - - # We add the user_directory_search table to the blacklist on SQLite - # because the existing search table does not have an index, making it - # unsafe to use native upserts. - if isinstance(self.database_engine, Sqlite3Engine): - self._unsafe_to_upsert_tables.add("user_directory_search") - - if self.database_engine.can_native_upsert: - # Check ASAP (and then later, every 1s) to see if we have finished - # background updates of tables that aren't safe to update. - self._clock.call_later( - 0.0, - run_as_background_process, - "upsert_safety_check", - self._check_safe_to_upsert, - ) - + self.db = Database(hs) self.rand = random.SystemRandom() - @defer.inlineCallbacks - def _check_safe_to_upsert(self): - """ - Is it safe to use native UPSERT? - - If there are background updates, we will need to wait, as they may be - the addition of indexes that set the UNIQUE constraint that we require. - - If the background updates have not completed, wait 15 sec and check again. - """ - updates = yield self.simple_select_list( - "background_updates", - keyvalues=None, - retcols=["update_name"], - desc="check_background_updates", - ) - updates = [x["update_name"] for x in updates] - - for table, update_name in UNIQUE_INDEX_BACKGROUND_UPDATES.items(): - if update_name not in updates: - logger.debug("Now safe to upsert in %s", table) - self._unsafe_to_upsert_tables.discard(table) - - # If there's any updates still running, reschedule to run. - if updates: - self._clock.call_later( - 15.0, - run_as_background_process, - "upsert_safety_check", - self._check_safe_to_upsert, - ) - - def start_profiling(self): - self._previous_loop_ts = monotonic_time() - - def loop(): - curr = self._current_txn_total_time - prev = self._previous_txn_total_time - self._previous_txn_total_time = curr - - time_now = monotonic_time() - time_then = self._previous_loop_ts - self._previous_loop_ts = time_now - - duration = time_now - time_then - ratio = (curr - prev) / duration - - top_three_counters = self._txn_perf_counters.interval(duration, limit=3) - - perf_logger.info( - "Total database time: %.3f%% {%s}", ratio * 100, top_three_counters - ) - - self._clock.looping_call(loop, 10000) - - def new_transaction( - self, conn, desc, after_callbacks, exception_callbacks, func, *args, **kwargs - ): - start = monotonic_time() - txn_id = self._TXN_ID - - # We don't really need these to be unique, so lets stop it from - # growing really large. - self._TXN_ID = (self._TXN_ID + 1) % (MAX_TXN_ID) - - name = "%s-%x" % (desc, txn_id) - - transaction_logger.debug("[TXN START] {%s}", name) - - try: - i = 0 - N = 5 - while True: - cursor = LoggingTransaction( - conn.cursor(), - name, - self.database_engine, - after_callbacks, - exception_callbacks, - ) - try: - r = func(cursor, *args, **kwargs) - conn.commit() - return r - except self.database_engine.module.OperationalError as e: - # This can happen if the database disappears mid - # transaction. - logger.warning( - "[TXN OPERROR] {%s} %s %d/%d", - name, - exception_to_unicode(e), - i, - N, - ) - if i < N: - i += 1 - try: - conn.rollback() - except self.database_engine.module.Error as e1: - logger.warning( - "[TXN EROLL] {%s} %s", name, exception_to_unicode(e1) - ) - continue - raise - except self.database_engine.module.DatabaseError as e: - if self.database_engine.is_deadlock(e): - logger.warning("[TXN DEADLOCK] {%s} %d/%d", name, i, N) - if i < N: - i += 1 - try: - conn.rollback() - except self.database_engine.module.Error as e1: - logger.warning( - "[TXN EROLL] {%s} %s", - name, - exception_to_unicode(e1), - ) - continue - raise - finally: - # we're either about to retry with a new cursor, or we're about to - # release the connection. Once we release the connection, it could - # get used for another query, which might do a conn.rollback(). - # - # In the latter case, even though that probably wouldn't affect the - # results of this transaction, python's sqlite will reset all - # statements on the connection [1], which will make our cursor - # invalid [2]. - # - # In any case, continuing to read rows after commit()ing seems - # dubious from the PoV of ACID transactional semantics - # (sqlite explicitly says that once you commit, you may see rows - # from subsequent updates.) - # - # In psycopg2, cursors are essentially a client-side fabrication - - # all the data is transferred to the client side when the statement - # finishes executing - so in theory we could go on streaming results - # from the cursor, but attempting to do so would make us - # incompatible with sqlite, so let's make sure we're not doing that - # by closing the cursor. - # - # (*named* cursors in psycopg2 are different and are proper server- - # side things, but (a) we don't use them and (b) they are implicitly - # closed by ending the transaction anyway.) - # - # In short, if we haven't finished with the cursor yet, that's a - # problem waiting to bite us. - # - # TL;DR: we're done with the cursor, so we can close it. - # - # [1]: https://github.com/python/cpython/blob/v3.8.0/Modules/_sqlite/connection.c#L465 - # [2]: https://github.com/python/cpython/blob/v3.8.0/Modules/_sqlite/cursor.c#L236 - cursor.close() - except Exception as e: - logger.debug("[TXN FAIL] {%s} %s", name, e) - raise - finally: - end = monotonic_time() - duration = end - start - - LoggingContext.current_context().add_database_transaction(duration) - - transaction_logger.debug("[TXN END] {%s} %f sec", name, duration) - - self._current_txn_total_time += duration - self._txn_perf_counters.update(desc, duration) - sql_txn_timer.labels(desc).observe(duration) - - @defer.inlineCallbacks - def runInteraction(self, desc, func, *args, **kwargs): - """Starts a transaction on the database and runs a given function - - Arguments: - desc (str): description of the transaction, for logging and metrics - func (func): callback function, which will be called with a - database transaction (twisted.enterprise.adbapi.Transaction) as - its first argument, followed by `args` and `kwargs`. - - args (list): positional args to pass to `func` - kwargs (dict): named args to pass to `func` - - Returns: - Deferred: The result of func - """ - after_callbacks = [] - exception_callbacks = [] - - if LoggingContext.current_context() == LoggingContext.sentinel: - logger.warning("Starting db txn '%s' from sentinel context", desc) - - try: - result = yield self.runWithConnection( - self.new_transaction, - desc, - after_callbacks, - exception_callbacks, - func, - *args, - **kwargs - ) - - for after_callback, after_args, after_kwargs in after_callbacks: - after_callback(*after_args, **after_kwargs) - except: # noqa: E722, as we reraise the exception this is fine. - for after_callback, after_args, after_kwargs in exception_callbacks: - after_callback(*after_args, **after_kwargs) - raise - - return result - - @defer.inlineCallbacks - def runWithConnection(self, func, *args, **kwargs): - """Wraps the .runWithConnection() method on the underlying db_pool. - - Arguments: - func (func): callback function, which will be called with a - database connection (twisted.enterprise.adbapi.Connection) as - its first argument, followed by `args` and `kwargs`. - args (list): positional args to pass to `func` - kwargs (dict): named args to pass to `func` - - Returns: - Deferred: The result of func - """ - parent_context = LoggingContext.current_context() - if parent_context == LoggingContext.sentinel: - logger.warning( - "Starting db connection from sentinel context: metrics will be lost" - ) - parent_context = None - - start_time = monotonic_time() - - def inner_func(conn, *args, **kwargs): - with LoggingContext("runWithConnection", parent_context) as context: - sched_duration_sec = monotonic_time() - start_time - sql_scheduling_timer.observe(sched_duration_sec) - context.add_database_scheduled(sched_duration_sec) - - if self.database_engine.is_connection_closed(conn): - logger.debug("Reconnecting closed database connection") - conn.reconnect() - - return func(conn, *args, **kwargs) - - result = yield make_deferred_yieldable( - self._db_pool.runWithConnection(inner_func, *args, **kwargs) - ) - - return result - - @staticmethod - def cursor_to_dict(cursor): - """Converts a SQL cursor into an list of dicts. - - Args: - cursor : The DBAPI cursor which has executed a query. - Returns: - A list of dicts where the key is the column header. - """ - col_headers = list(intern(str(column[0])) for column in cursor.description) - results = list(dict(zip(col_headers, row)) for row in cursor) - return results - - def execute(self, desc, decoder, query, *args): - """Runs a single query for a result set. - - Args: - decoder - The function which can resolve the cursor results to - something meaningful. - query - The query string to execute - *args - Query args. - Returns: - The result of decoder(results) - """ - - def interaction(txn): - txn.execute(query, args) - if decoder: - return decoder(txn) - else: - return txn.fetchall() - - return self.runInteraction(desc, interaction) - - # "Simple" SQL API methods that operate on a single table with no JOINs, - # no complex WHERE clauses, just a dict of values for columns. - - @defer.inlineCallbacks - def simple_insert(self, table, values, or_ignore=False, desc="simple_insert"): - """Executes an INSERT query on the named table. - - Args: - table : string giving the table name - values : dict of new column names and values for them - or_ignore : bool stating whether an exception should be raised - when a conflicting row already exists. If True, False will be - returned by the function instead - desc : string giving a description of the transaction - - Returns: - bool: Whether the row was inserted or not. Only useful when - `or_ignore` is True - """ - try: - yield self.runInteraction(desc, self.simple_insert_txn, table, values) - except self.database_engine.module.IntegrityError: - # We have to do or_ignore flag at this layer, since we can't reuse - # a cursor after we receive an error from the db. - if not or_ignore: - raise - return False - return True - - @staticmethod - def simple_insert_txn(txn, table, values): - keys, vals = zip(*values.items()) - - sql = "INSERT INTO %s (%s) VALUES(%s)" % ( - table, - ", ".join(k for k in keys), - ", ".join("?" for _ in keys), - ) - - txn.execute(sql, vals) - - def simple_insert_many(self, table, values, desc): - return self.runInteraction(desc, self.simple_insert_many_txn, table, values) - - @staticmethod - def simple_insert_many_txn(txn, table, values): - if not values: - return - - # This is a *slight* abomination to get a list of tuples of key names - # and a list of tuples of value names. - # - # i.e. [{"a": 1, "b": 2}, {"c": 3, "d": 4}] - # => [("a", "b",), ("c", "d",)] and [(1, 2,), (3, 4,)] - # - # The sort is to ensure that we don't rely on dictionary iteration - # order. - keys, vals = zip( - *[zip(*(sorted(i.items(), key=lambda kv: kv[0]))) for i in values if i] - ) - - for k in keys: - if k != keys[0]: - raise RuntimeError("All items must have the same keys") - - sql = "INSERT INTO %s (%s) VALUES(%s)" % ( - table, - ", ".join(k for k in keys[0]), - ", ".join("?" for _ in keys[0]), - ) - - txn.executemany(sql, vals) - - @defer.inlineCallbacks - def simple_upsert( - self, - table, - keyvalues, - values, - insertion_values={}, - desc="simple_upsert", - lock=True, - ): - """ - - `lock` should generally be set to True (the default), but can be set - to False if either of the following are true: - - * there is a UNIQUE INDEX on the key columns. In this case a conflict - will cause an IntegrityError in which case this function will retry - the update. - - * we somehow know that we are the only thread which will be updating - this table. - - Args: - table (str): The table to upsert into - keyvalues (dict): The unique key columns and their new values - values (dict): The nonunique columns and their new values - insertion_values (dict): additional key/values to use only when - inserting - lock (bool): True to lock the table when doing the upsert. - Returns: - Deferred(None or bool): Native upserts always return None. Emulated - upserts return True if a new entry was created, False if an existing - one was updated. - """ - attempts = 0 - while True: - try: - result = yield self.runInteraction( - desc, - self.simple_upsert_txn, - table, - keyvalues, - values, - insertion_values, - lock=lock, - ) - return result - except self.database_engine.module.IntegrityError as e: - attempts += 1 - if attempts >= 5: - # don't retry forever, because things other than races - # can cause IntegrityErrors - raise - - # presumably we raced with another transaction: let's retry. - logger.warning( - "IntegrityError when upserting into %s; retrying: %s", table, e - ) - - def simple_upsert_txn( - self, txn, table, keyvalues, values, insertion_values={}, lock=True - ): - """ - Pick the UPSERT method which works best on the platform. Either the - native one (Pg9.5+, recent SQLites), or fall back to an emulated method. - - Args: - txn: The transaction to use. - table (str): The table to upsert into - keyvalues (dict): The unique key tables and their new values - values (dict): The nonunique columns and their new values - insertion_values (dict): additional key/values to use only when - inserting - lock (bool): True to lock the table when doing the upsert. - Returns: - None or bool: Native upserts always return None. Emulated - upserts return True if a new entry was created, False if an existing - one was updated. - """ - if ( - self.database_engine.can_native_upsert - and table not in self._unsafe_to_upsert_tables - ): - return self.simple_upsert_txn_native_upsert( - txn, table, keyvalues, values, insertion_values=insertion_values - ) - else: - return self.simple_upsert_txn_emulated( - txn, - table, - keyvalues, - values, - insertion_values=insertion_values, - lock=lock, - ) - - def simple_upsert_txn_emulated( - self, txn, table, keyvalues, values, insertion_values={}, lock=True - ): - """ - Args: - table (str): The table to upsert into - keyvalues (dict): The unique key tables and their new values - values (dict): The nonunique columns and their new values - insertion_values (dict): additional key/values to use only when - inserting - lock (bool): True to lock the table when doing the upsert. - Returns: - bool: Return True if a new entry was created, False if an existing - one was updated. - """ - # We need to lock the table :(, unless we're *really* careful - if lock: - self.database_engine.lock_table(txn, table) - - def _getwhere(key): - # If the value we're passing in is None (aka NULL), we need to use - # IS, not =, as NULL = NULL equals NULL (False). - if keyvalues[key] is None: - return "%s IS ?" % (key,) - else: - return "%s = ?" % (key,) - - if not values: - # If `values` is empty, then all of the values we care about are in - # the unique key, so there is nothing to UPDATE. We can just do a - # SELECT instead to see if it exists. - sql = "SELECT 1 FROM %s WHERE %s" % ( - table, - " AND ".join(_getwhere(k) for k in keyvalues), - ) - sqlargs = list(keyvalues.values()) - txn.execute(sql, sqlargs) - if txn.fetchall(): - # We have an existing record. - return False - else: - # First try to update. - sql = "UPDATE %s SET %s WHERE %s" % ( - table, - ", ".join("%s = ?" % (k,) for k in values), - " AND ".join(_getwhere(k) for k in keyvalues), - ) - sqlargs = list(values.values()) + list(keyvalues.values()) - - txn.execute(sql, sqlargs) - if txn.rowcount > 0: - # successfully updated at least one row. - return False - - # We didn't find any existing rows, so insert a new one - allvalues = {} - allvalues.update(keyvalues) - allvalues.update(values) - allvalues.update(insertion_values) - - sql = "INSERT INTO %s (%s) VALUES (%s)" % ( - table, - ", ".join(k for k in allvalues), - ", ".join("?" for _ in allvalues), - ) - txn.execute(sql, list(allvalues.values())) - # successfully inserted - return True - - def simple_upsert_txn_native_upsert( - self, txn, table, keyvalues, values, insertion_values={} - ): - """ - Use the native UPSERT functionality in recent PostgreSQL versions. - - Args: - table (str): The table to upsert into - keyvalues (dict): The unique key tables and their new values - values (dict): The nonunique columns and their new values - insertion_values (dict): additional key/values to use only when - inserting - Returns: - None - """ - allvalues = {} - allvalues.update(keyvalues) - allvalues.update(insertion_values) - - if not values: - latter = "NOTHING" - else: - allvalues.update(values) - latter = "UPDATE SET " + ", ".join(k + "=EXCLUDED." + k for k in values) - - sql = ("INSERT INTO %s (%s) VALUES (%s) ON CONFLICT (%s) DO %s") % ( - table, - ", ".join(k for k in allvalues), - ", ".join("?" for _ in allvalues), - ", ".join(k for k in keyvalues), - latter, - ) - txn.execute(sql, list(allvalues.values())) - - def simple_upsert_many_txn( - self, txn, table, key_names, key_values, value_names, value_values - ): - """ - Upsert, many times. - - Args: - table (str): The table to upsert into - key_names (list[str]): The key column names. - key_values (list[list]): A list of each row's key column values. - value_names (list[str]): The value column names. If empty, no - values will be used, even if value_values is provided. - value_values (list[list]): A list of each row's value column values. - Returns: - None - """ - if ( - self.database_engine.can_native_upsert - and table not in self._unsafe_to_upsert_tables - ): - return self.simple_upsert_many_txn_native_upsert( - txn, table, key_names, key_values, value_names, value_values - ) - else: - return self.simple_upsert_many_txn_emulated( - txn, table, key_names, key_values, value_names, value_values - ) - - def simple_upsert_many_txn_emulated( - self, txn, table, key_names, key_values, value_names, value_values - ): - """ - Upsert, many times, but without native UPSERT support or batching. - - Args: - table (str): The table to upsert into - key_names (list[str]): The key column names. - key_values (list[list]): A list of each row's key column values. - value_names (list[str]): The value column names. If empty, no - values will be used, even if value_values is provided. - value_values (list[list]): A list of each row's value column values. - Returns: - None - """ - # No value columns, therefore make a blank list so that the following - # zip() works correctly. - if not value_names: - value_values = [() for x in range(len(key_values))] - - for keyv, valv in zip(key_values, value_values): - _keys = {x: y for x, y in zip(key_names, keyv)} - _vals = {x: y for x, y in zip(value_names, valv)} - - self.simple_upsert_txn_emulated(txn, table, _keys, _vals) - - def simple_upsert_many_txn_native_upsert( - self, txn, table, key_names, key_values, value_names, value_values - ): - """ - Upsert, many times, using batching where possible. - - Args: - table (str): The table to upsert into - key_names (list[str]): The key column names. - key_values (list[list]): A list of each row's key column values. - value_names (list[str]): The value column names. If empty, no - values will be used, even if value_values is provided. - value_values (list[list]): A list of each row's value column values. - Returns: - None - """ - allnames = [] - allnames.extend(key_names) - allnames.extend(value_names) - - if not value_names: - # No value columns, therefore make a blank list so that the - # following zip() works correctly. - latter = "NOTHING" - value_values = [() for x in range(len(key_values))] - else: - latter = "UPDATE SET " + ", ".join( - k + "=EXCLUDED." + k for k in value_names - ) - - sql = "INSERT INTO %s (%s) VALUES (%s) ON CONFLICT (%s) DO %s" % ( - table, - ", ".join(k for k in allnames), - ", ".join("?" for _ in allnames), - ", ".join(key_names), - latter, - ) - - args = [] - - for x, y in zip(key_values, value_values): - args.append(tuple(x) + tuple(y)) - - return txn.execute_batch(sql, args) - - def simple_select_one( - self, table, keyvalues, retcols, allow_none=False, desc="simple_select_one" - ): - """Executes a SELECT query on the named table, which is expected to - return a single row, returning multiple columns from it. - - Args: - table : string giving the table name - keyvalues : dict of column names and values to select the row with - retcols : list of strings giving the names of the columns to return - - allow_none : If true, return None instead of failing if the SELECT - statement returns no rows - """ - return self.runInteraction( - desc, self.simple_select_one_txn, table, keyvalues, retcols, allow_none - ) - - def simple_select_one_onecol( - self, - table, - keyvalues, - retcol, - allow_none=False, - desc="simple_select_one_onecol", - ): - """Executes a SELECT query on the named table, which is expected to - return a single row, returning a single column from it. - - Args: - table : string giving the table name - keyvalues : dict of column names and values to select the row with - retcol : string giving the name of the column to return - """ - return self.runInteraction( - desc, - self.simple_select_one_onecol_txn, - table, - keyvalues, - retcol, - allow_none=allow_none, - ) - - @classmethod - def simple_select_one_onecol_txn( - cls, txn, table, keyvalues, retcol, allow_none=False - ): - ret = cls.simple_select_onecol_txn( - txn, table=table, keyvalues=keyvalues, retcol=retcol - ) - - if ret: - return ret[0] - else: - if allow_none: - return None - else: - raise StoreError(404, "No row found") - - @staticmethod - def simple_select_onecol_txn(txn, table, keyvalues, retcol): - sql = ("SELECT %(retcol)s FROM %(table)s") % {"retcol": retcol, "table": table} - - if keyvalues: - sql += " WHERE %s" % " AND ".join("%s = ?" % k for k in iterkeys(keyvalues)) - txn.execute(sql, list(keyvalues.values())) - else: - txn.execute(sql) - - return [r[0] for r in txn] - - def simple_select_onecol( - self, table, keyvalues, retcol, desc="simple_select_onecol" - ): - """Executes a SELECT query on the named table, which returns a list - comprising of the values of the named column from the selected rows. - - Args: - table (str): table name - keyvalues (dict|None): column names and values to select the rows with - retcol (str): column whos value we wish to retrieve. - - Returns: - Deferred: Results in a list - """ - return self.runInteraction( - desc, self.simple_select_onecol_txn, table, keyvalues, retcol - ) - - def simple_select_list(self, table, keyvalues, retcols, desc="simple_select_list"): - """Executes a SELECT query on the named table, which may return zero or - more rows, returning the result as a list of dicts. - - Args: - table (str): the table name - keyvalues (dict[str, Any] | None): - column names and values to select the rows with, or None to not - apply a WHERE clause. - retcols (iterable[str]): the names of the columns to return - Returns: - defer.Deferred: resolves to list[dict[str, Any]] - """ - return self.runInteraction( - desc, self.simple_select_list_txn, table, keyvalues, retcols - ) - - @classmethod - def simple_select_list_txn(cls, txn, table, keyvalues, retcols): - """Executes a SELECT query on the named table, which may return zero or - more rows, returning the result as a list of dicts. - - Args: - txn : Transaction object - table (str): the table name - keyvalues (dict[str, T] | None): - column names and values to select the rows with, or None to not - apply a WHERE clause. - retcols (iterable[str]): the names of the columns to return - """ - if keyvalues: - sql = "SELECT %s FROM %s WHERE %s" % ( - ", ".join(retcols), - table, - " AND ".join("%s = ?" % (k,) for k in keyvalues), - ) - txn.execute(sql, list(keyvalues.values())) - else: - sql = "SELECT %s FROM %s" % (", ".join(retcols), table) - txn.execute(sql) - - return cls.cursor_to_dict(txn) - - @defer.inlineCallbacks - def simple_select_many_batch( - self, - table, - column, - iterable, - retcols, - keyvalues={}, - desc="simple_select_many_batch", - batch_size=100, - ): - """Executes a SELECT query on the named table, which may return zero or - more rows, returning the result as a list of dicts. - - Filters rows by if value of `column` is in `iterable`. - - Args: - table : string giving the table name - column : column name to test for inclusion against `iterable` - iterable : list - keyvalues : dict of column names and values to select the rows with - retcols : list of strings giving the names of the columns to return - """ - results = [] - - if not iterable: - return results - - # iterables can not be sliced, so convert it to a list first - it_list = list(iterable) - - chunks = [ - it_list[i : i + batch_size] for i in range(0, len(it_list), batch_size) - ] - for chunk in chunks: - rows = yield self.runInteraction( - desc, - self.simple_select_many_txn, - table, - column, - chunk, - keyvalues, - retcols, - ) - - results.extend(rows) - - return results - - @classmethod - def simple_select_many_txn(cls, txn, table, column, iterable, keyvalues, retcols): - """Executes a SELECT query on the named table, which may return zero or - more rows, returning the result as a list of dicts. - - Filters rows by if value of `column` is in `iterable`. - - Args: - txn : Transaction object - table : string giving the table name - column : column name to test for inclusion against `iterable` - iterable : list - keyvalues : dict of column names and values to select the rows with - retcols : list of strings giving the names of the columns to return - """ - if not iterable: - return [] - - clause, values = make_in_list_sql_clause(txn.database_engine, column, iterable) - clauses = [clause] - - for key, value in iteritems(keyvalues): - clauses.append("%s = ?" % (key,)) - values.append(value) - - sql = "SELECT %s FROM %s WHERE %s" % ( - ", ".join(retcols), - table, - " AND ".join(clauses), - ) - - txn.execute(sql, values) - return cls.cursor_to_dict(txn) - - def simple_update(self, table, keyvalues, updatevalues, desc): - return self.runInteraction( - desc, self.simple_update_txn, table, keyvalues, updatevalues - ) - - @staticmethod - def simple_update_txn(txn, table, keyvalues, updatevalues): - if keyvalues: - where = "WHERE %s" % " AND ".join("%s = ?" % k for k in iterkeys(keyvalues)) - else: - where = "" - - update_sql = "UPDATE %s SET %s %s" % ( - table, - ", ".join("%s = ?" % (k,) for k in updatevalues), - where, - ) - - txn.execute(update_sql, list(updatevalues.values()) + list(keyvalues.values())) - - return txn.rowcount - - def simple_update_one( - self, table, keyvalues, updatevalues, desc="simple_update_one" - ): - """Executes an UPDATE query on the named table, setting new values for - columns in a row matching the key values. - - Args: - table : string giving the table name - keyvalues : dict of column names and values to select the row with - updatevalues : dict giving column names and values to update - retcols : optional list of column names to return - - If present, retcols gives a list of column names on which to perform - a SELECT statement *before* performing the UPDATE statement. The values - of these will be returned in a dict. - - These are performed within the same transaction, allowing an atomic - get-and-set. This can be used to implement compare-and-set by putting - the update column in the 'keyvalues' dict as well. - """ - return self.runInteraction( - desc, self.simple_update_one_txn, table, keyvalues, updatevalues - ) - - @classmethod - def simple_update_one_txn(cls, txn, table, keyvalues, updatevalues): - rowcount = cls.simple_update_txn(txn, table, keyvalues, updatevalues) - - if rowcount == 0: - raise StoreError(404, "No row found (%s)" % (table,)) - if rowcount > 1: - raise StoreError(500, "More than one row matched (%s)" % (table,)) - - @staticmethod - def simple_select_one_txn(txn, table, keyvalues, retcols, allow_none=False): - select_sql = "SELECT %s FROM %s WHERE %s" % ( - ", ".join(retcols), - table, - " AND ".join("%s = ?" % (k,) for k in keyvalues), - ) - - txn.execute(select_sql, list(keyvalues.values())) - row = txn.fetchone() - - if not row: - if allow_none: - return None - raise StoreError(404, "No row found (%s)" % (table,)) - if txn.rowcount > 1: - raise StoreError(500, "More than one row matched (%s)" % (table,)) - - return dict(zip(retcols, row)) - - def simple_delete_one(self, table, keyvalues, desc="simple_delete_one"): - """Executes a DELETE query on the named table, expecting to delete a - single row. - - Args: - table : string giving the table name - keyvalues : dict of column names and values to select the row with - """ - return self.runInteraction(desc, self.simple_delete_one_txn, table, keyvalues) - - @staticmethod - def simple_delete_one_txn(txn, table, keyvalues): - """Executes a DELETE query on the named table, expecting to delete a - single row. - - Args: - table : string giving the table name - keyvalues : dict of column names and values to select the row with - """ - sql = "DELETE FROM %s WHERE %s" % ( - table, - " AND ".join("%s = ?" % (k,) for k in keyvalues), - ) - - txn.execute(sql, list(keyvalues.values())) - if txn.rowcount == 0: - raise StoreError(404, "No row found (%s)" % (table,)) - if txn.rowcount > 1: - raise StoreError(500, "More than one row matched (%s)" % (table,)) - - def simple_delete(self, table, keyvalues, desc): - return self.runInteraction(desc, self.simple_delete_txn, table, keyvalues) - - @staticmethod - def simple_delete_txn(txn, table, keyvalues): - sql = "DELETE FROM %s WHERE %s" % ( - table, - " AND ".join("%s = ?" % (k,) for k in keyvalues), - ) - - txn.execute(sql, list(keyvalues.values())) - return txn.rowcount - - def simple_delete_many(self, table, column, iterable, keyvalues, desc): - return self.runInteraction( - desc, self.simple_delete_many_txn, table, column, iterable, keyvalues - ) - - @staticmethod - def simple_delete_many_txn(txn, table, column, iterable, keyvalues): - """Executes a DELETE query on the named table. - - Filters rows by if value of `column` is in `iterable`. - - Args: - txn : Transaction object - table : string giving the table name - column : column name to test for inclusion against `iterable` - iterable : list - keyvalues : dict of column names and values to select the rows with - - Returns: - int: Number rows deleted - """ - if not iterable: - return 0 - - sql = "DELETE FROM %s" % table - - clause, values = make_in_list_sql_clause(txn.database_engine, column, iterable) - clauses = [clause] - - for key, value in iteritems(keyvalues): - clauses.append("%s = ?" % (key,)) - values.append(value) - - if clauses: - sql = "%s WHERE %s" % (sql, " AND ".join(clauses)) - txn.execute(sql, values) - - return txn.rowcount - - def get_cache_dict( - self, db_conn, table, entity_column, stream_column, max_value, limit=100000 - ): - # Fetch a mapping of room_id -> max stream position for "recent" rooms. - # It doesn't really matter how many we get, the StreamChangeCache will - # do the right thing to ensure it respects the max size of cache. - sql = ( - "SELECT %(entity)s, MAX(%(stream)s) FROM %(table)s" - " WHERE %(stream)s > ? - %(limit)s" - " GROUP BY %(entity)s" - ) % { - "table": table, - "entity": entity_column, - "stream": stream_column, - "limit": limit, - } - - sql = self.database_engine.convert_param_style(sql) - - txn = db_conn.cursor() - txn.execute(sql, (int(max_value),)) - - cache = {row[0]: int(row[1]) for row in txn} - - txn.close() - - if cache: - min_val = min(itervalues(cache)) - else: - min_val = max_value - - return cache, min_val - def _invalidate_state_caches(self, room_id, members_changed): """Invalidates caches that are based on the current state, but does not stream invalidations down replication. @@ -1347,159 +71,6 @@ class SQLBaseStore(object): # which is fine. pass - def simple_select_list_paginate( - self, - table, - keyvalues, - orderby, - start, - limit, - retcols, - order_direction="ASC", - desc="simple_select_list_paginate", - ): - """ - Executes a SELECT query on the named table with start and limit, - of row numbers, which may return zero or number of rows from start to limit, - returning the result as a list of dicts. - - Args: - table (str): the table name - keyvalues (dict[str, T] | None): - column names and values to select the rows with, or None to not - apply a WHERE clause. - orderby (str): Column to order the results by. - start (int): Index to begin the query at. - limit (int): Number of results to return. - retcols (iterable[str]): the names of the columns to return - order_direction (str): Whether the results should be ordered "ASC" or "DESC". - Returns: - defer.Deferred: resolves to list[dict[str, Any]] - """ - return self.runInteraction( - desc, - self.simple_select_list_paginate_txn, - table, - keyvalues, - orderby, - start, - limit, - retcols, - order_direction=order_direction, - ) - - @classmethod - def simple_select_list_paginate_txn( - cls, - txn, - table, - keyvalues, - orderby, - start, - limit, - retcols, - order_direction="ASC", - ): - """ - Executes a SELECT query on the named table with start and limit, - of row numbers, which may return zero or number of rows from start to limit, - returning the result as a list of dicts. - - Args: - txn : Transaction object - table (str): the table name - keyvalues (dict[str, T] | None): - column names and values to select the rows with, or None to not - apply a WHERE clause. - orderby (str): Column to order the results by. - start (int): Index to begin the query at. - limit (int): Number of results to return. - retcols (iterable[str]): the names of the columns to return - order_direction (str): Whether the results should be ordered "ASC" or "DESC". - Returns: - defer.Deferred: resolves to list[dict[str, Any]] - """ - if order_direction not in ["ASC", "DESC"]: - raise ValueError("order_direction must be one of 'ASC' or 'DESC'.") - - if keyvalues: - where_clause = "WHERE " + " AND ".join("%s = ?" % (k,) for k in keyvalues) - else: - where_clause = "" - - sql = "SELECT %s FROM %s %s ORDER BY %s %s LIMIT ? OFFSET ?" % ( - ", ".join(retcols), - table, - where_clause, - orderby, - order_direction, - ) - txn.execute(sql, list(keyvalues.values()) + [limit, start]) - - return cls.cursor_to_dict(txn) - - def get_user_count_txn(self, txn): - """Get a total number of registered users in the users list. - - Args: - txn : Transaction object - Returns: - int : number of users - """ - sql_count = "SELECT COUNT(*) FROM users WHERE is_guest = 0;" - txn.execute(sql_count) - return txn.fetchone()[0] - - def simple_search_list(self, table, term, col, retcols, desc="simple_search_list"): - """Executes a SELECT query on the named table, which may return zero or - more rows, returning the result as a list of dicts. - - Args: - table (str): the table name - term (str | None): - term for searching the table matched to a column. - col (str): column to query term should be matched to - retcols (iterable[str]): the names of the columns to return - Returns: - defer.Deferred: resolves to list[dict[str, Any]] or None - """ - - return self.runInteraction( - desc, self.simple_search_list_txn, table, term, col, retcols - ) - - @classmethod - def simple_search_list_txn(cls, txn, table, term, col, retcols): - """Executes a SELECT query on the named table, which may return zero or - more rows, returning the result as a list of dicts. - - Args: - txn : Transaction object - table (str): the table name - term (str | None): - term for searching the table matched to a column. - col (str): column to query term should be matched to - retcols (iterable[str]): the names of the columns to return - Returns: - defer.Deferred: resolves to list[dict[str, Any]] or None - """ - if term: - sql = "SELECT %s FROM %s WHERE %s LIKE ?" % (", ".join(retcols), table, col) - termvalues = ["%%" + term + "%%"] - txn.execute(sql, termvalues) - else: - return 0 - - return cls.cursor_to_dict(txn) - - -class _RollbackButIsFineException(Exception): - """ This exception is used to rollback a transaction without implying - something went wrong. - """ - - pass - def db_to_json(db_content): """ @@ -1528,30 +99,3 @@ def db_to_json(db_content): except Exception: logging.warning("Tried to decode '%r' as JSON and failed", db_content) raise - - -def make_in_list_sql_clause( - database_engine, column: str, iterable: Iterable -) -> Tuple[str, Iterable]: - """Returns an SQL clause that checks the given column is in the iterable. - - On SQLite this expands to `column IN (?, ?, ...)`, whereas on Postgres - it expands to `column = ANY(?)`. While both DBs support the `IN` form, - using the `ANY` form on postgres means that it views queries with - different length iterables as the same, helping the query stats. - - Args: - database_engine - column: Name of the column - iterable: The values to check the column against. - - Returns: - A tuple of SQL query and the args - """ - - if database_engine.supports_using_any_list: - # This should hopefully be faster, but also makes postgres query - # stats easier to understand. - return "%s = ANY(?)" % (column,), [list(iterable)] - else: - return "%s IN (%s)" % (column, ",".join("?" for _ in iterable)), list(iterable) diff --git a/synapse/storage/background_updates.py b/synapse/storage/background_updates.py index 06955a0537..dfca94b0e0 100644 --- a/synapse/storage/background_updates.py +++ b/synapse/storage/background_updates.py @@ -139,7 +139,7 @@ class BackgroundUpdateStore(SQLBaseStore): # otherwise, check if there are updates to be run. This is important, # as we may be running on a worker which doesn't perform the bg updates # itself, but still wants to wait for them to happen. - updates = yield self.simple_select_onecol( + updates = yield self.db.simple_select_onecol( "background_updates", keyvalues=None, retcol="1", @@ -161,7 +161,7 @@ class BackgroundUpdateStore(SQLBaseStore): if update_name in self._background_update_queue: return False - update_exists = await self.simple_select_one_onecol( + update_exists = await self.db.simple_select_one_onecol( "background_updates", keyvalues={"update_name": update_name}, retcol="1", @@ -184,7 +184,7 @@ class BackgroundUpdateStore(SQLBaseStore): no more work to do. """ if not self._background_update_queue: - updates = yield self.simple_select_list( + updates = yield self.db.simple_select_list( "background_updates", keyvalues=None, retcols=("update_name", "depends_on"), @@ -226,7 +226,7 @@ class BackgroundUpdateStore(SQLBaseStore): else: batch_size = self.DEFAULT_BACKGROUND_BATCH_SIZE - progress_json = yield self.simple_select_one_onecol( + progress_json = yield self.db.simple_select_one_onecol( "background_updates", keyvalues={"update_name": update_name}, retcol="progress_json", @@ -391,7 +391,7 @@ class BackgroundUpdateStore(SQLBaseStore): def updater(progress, batch_size): if runner is not None: logger.info("Adding index %s to %s", index_name, table) - yield self.runWithConnection(runner) + yield self.db.runWithConnection(runner) yield self._end_background_update(update_name) return 1 @@ -413,7 +413,7 @@ class BackgroundUpdateStore(SQLBaseStore): self._background_update_queue = [] progress_json = json.dumps(progress) - return self.simple_insert( + return self.db.simple_insert( "background_updates", {"update_name": update_name, "progress_json": progress_json}, ) @@ -429,7 +429,7 @@ class BackgroundUpdateStore(SQLBaseStore): self._background_update_queue = [ name for name in self._background_update_queue if name != update_name ] - return self.simple_delete_one( + return self.db.simple_delete_one( "background_updates", keyvalues={"update_name": update_name} ) @@ -444,7 +444,7 @@ class BackgroundUpdateStore(SQLBaseStore): progress_json = json.dumps(progress) - self.simple_update_one_txn( + self.db.simple_update_one_txn( txn, "background_updates", keyvalues={"update_name": update_name}, diff --git a/synapse/storage/data_stores/main/__init__.py b/synapse/storage/data_stores/main/__init__.py index 2a5b33dda1..46f0f26af6 100644 --- a/synapse/storage/data_stores/main/__init__.py +++ b/synapse/storage/data_stores/main/__init__.py @@ -171,9 +171,11 @@ class DataStore( else: self._cache_id_gen = None + super(DataStore, self).__init__(db_conn, hs) + self._presence_on_startup = self._get_active_presence(db_conn) - presence_cache_prefill, min_presence_val = self.get_cache_dict( + presence_cache_prefill, min_presence_val = self.db.get_cache_dict( db_conn, "presence_stream", entity_column="user_id", @@ -187,7 +189,7 @@ class DataStore( ) max_device_inbox_id = self._device_inbox_id_gen.get_current_token() - device_inbox_prefill, min_device_inbox_id = self.get_cache_dict( + device_inbox_prefill, min_device_inbox_id = self.db.get_cache_dict( db_conn, "device_inbox", entity_column="user_id", @@ -202,7 +204,7 @@ class DataStore( ) # The federation outbox and the local device inbox uses the same # stream_id generator. - device_outbox_prefill, min_device_outbox_id = self.get_cache_dict( + device_outbox_prefill, min_device_outbox_id = self.db.get_cache_dict( db_conn, "device_federation_outbox", entity_column="destination", @@ -228,7 +230,7 @@ class DataStore( ) events_max = self._stream_id_gen.get_current_token() - curr_state_delta_prefill, min_curr_state_delta_id = self.get_cache_dict( + curr_state_delta_prefill, min_curr_state_delta_id = self.db.get_cache_dict( db_conn, "current_state_delta_stream", entity_column="room_id", @@ -242,7 +244,7 @@ class DataStore( prefilled_cache=curr_state_delta_prefill, ) - _group_updates_prefill, min_group_updates_id = self.get_cache_dict( + _group_updates_prefill, min_group_updates_id = self.db.get_cache_dict( db_conn, "local_group_updates", entity_column="user_id", @@ -262,8 +264,6 @@ class DataStore( # Used in _generate_user_daily_visits to keep track of progress self._last_user_visit_update = self._get_start_of_day() - super(DataStore, self).__init__(db_conn, hs) - def take_presence_startup_info(self): active_on_startup = self._presence_on_startup self._presence_on_startup = None @@ -283,7 +283,7 @@ class DataStore( txn = db_conn.cursor() txn.execute(sql, (PresenceState.OFFLINE,)) - rows = self.cursor_to_dict(txn) + rows = self.db.cursor_to_dict(txn) txn.close() for row in rows: @@ -296,7 +296,7 @@ class DataStore( Counts the number of users who used this homeserver in the last 24 hours. """ yesterday = int(self._clock.time_msec()) - (1000 * 60 * 60 * 24) - return self.runInteraction("count_daily_users", self._count_users, yesterday) + return self.db.runInteraction("count_daily_users", self._count_users, yesterday) def count_monthly_users(self): """ @@ -306,7 +306,7 @@ class DataStore( amongst other things, includes a 3 day grace period before a user counts. """ thirty_days_ago = int(self._clock.time_msec()) - (1000 * 60 * 60 * 24 * 30) - return self.runInteraction( + return self.db.runInteraction( "count_monthly_users", self._count_users, thirty_days_ago ) @@ -406,7 +406,7 @@ class DataStore( return results - return self.runInteraction("count_r30_users", _count_r30_users) + return self.db.runInteraction("count_r30_users", _count_r30_users) def _get_start_of_day(self): """ @@ -471,7 +471,7 @@ class DataStore( # frequently self._last_user_visit_update = now - return self.runInteraction( + return self.db.runInteraction( "generate_user_daily_visits", _generate_user_daily_visits ) @@ -482,7 +482,7 @@ class DataStore( Returns: defer.Deferred: resolves to list[dict[str, Any]] """ - return self.simple_select_list( + return self.db.simple_select_list( table="users", keyvalues={}, retcols=["name", "password_hash", "is_guest", "admin", "user_type"], @@ -502,9 +502,9 @@ class DataStore( Returns: defer.Deferred: resolves to json object {list[dict[str, Any]], count} """ - users = yield self.runInteraction( + users = yield self.db.runInteraction( "get_users_paginate", - self.simple_select_list_paginate_txn, + self.db.simple_select_list_paginate_txn, table="users", keyvalues={"is_guest": False}, orderby=order, @@ -512,7 +512,9 @@ class DataStore( limit=limit, retcols=["name", "password_hash", "is_guest", "admin", "user_type"], ) - count = yield self.runInteraction("get_users_paginate", self.get_user_count_txn) + count = yield self.db.runInteraction( + "get_users_paginate", self.get_user_count_txn + ) retval = {"users": users, "total": count} return retval @@ -526,7 +528,7 @@ class DataStore( Returns: defer.Deferred: resolves to list[dict[str, Any]] """ - return self.simple_search_list( + return self.db.simple_search_list( table="users", term=term, col="name", diff --git a/synapse/storage/data_stores/main/account_data.py b/synapse/storage/data_stores/main/account_data.py index b0d22faf3f..a96fe9485c 100644 --- a/synapse/storage/data_stores/main/account_data.py +++ b/synapse/storage/data_stores/main/account_data.py @@ -67,7 +67,7 @@ class AccountDataWorkerStore(SQLBaseStore): """ def get_account_data_for_user_txn(txn): - rows = self.simple_select_list_txn( + rows = self.db.simple_select_list_txn( txn, "account_data", {"user_id": user_id}, @@ -78,7 +78,7 @@ class AccountDataWorkerStore(SQLBaseStore): row["account_data_type"]: json.loads(row["content"]) for row in rows } - rows = self.simple_select_list_txn( + rows = self.db.simple_select_list_txn( txn, "room_account_data", {"user_id": user_id}, @@ -92,7 +92,7 @@ class AccountDataWorkerStore(SQLBaseStore): return global_account_data, by_room - return self.runInteraction( + return self.db.runInteraction( "get_account_data_for_user", get_account_data_for_user_txn ) @@ -102,7 +102,7 @@ class AccountDataWorkerStore(SQLBaseStore): Returns: Deferred: A dict """ - result = yield self.simple_select_one_onecol( + result = yield self.db.simple_select_one_onecol( table="account_data", keyvalues={"user_id": user_id, "account_data_type": data_type}, retcol="content", @@ -127,7 +127,7 @@ class AccountDataWorkerStore(SQLBaseStore): """ def get_account_data_for_room_txn(txn): - rows = self.simple_select_list_txn( + rows = self.db.simple_select_list_txn( txn, "room_account_data", {"user_id": user_id, "room_id": room_id}, @@ -138,7 +138,7 @@ class AccountDataWorkerStore(SQLBaseStore): row["account_data_type"]: json.loads(row["content"]) for row in rows } - return self.runInteraction( + return self.db.runInteraction( "get_account_data_for_room", get_account_data_for_room_txn ) @@ -156,7 +156,7 @@ class AccountDataWorkerStore(SQLBaseStore): """ def get_account_data_for_room_and_type_txn(txn): - content_json = self.simple_select_one_onecol_txn( + content_json = self.db.simple_select_one_onecol_txn( txn, table="room_account_data", keyvalues={ @@ -170,7 +170,7 @@ class AccountDataWorkerStore(SQLBaseStore): return json.loads(content_json) if content_json else None - return self.runInteraction( + return self.db.runInteraction( "get_account_data_for_room_and_type", get_account_data_for_room_and_type_txn ) @@ -207,7 +207,7 @@ class AccountDataWorkerStore(SQLBaseStore): room_results = txn.fetchall() return global_results, room_results - return self.runInteraction( + return self.db.runInteraction( "get_all_updated_account_data_txn", get_updated_account_data_txn ) @@ -252,7 +252,7 @@ class AccountDataWorkerStore(SQLBaseStore): if not changed: return {}, {} - return self.runInteraction( + return self.db.runInteraction( "get_updated_account_data_for_user", get_updated_account_data_for_user_txn ) @@ -302,7 +302,7 @@ class AccountDataStore(AccountDataWorkerStore): # no need to lock here as room_account_data has a unique constraint # on (user_id, room_id, account_data_type) so simple_upsert will # retry if there is a conflict. - yield self.simple_upsert( + yield self.db.simple_upsert( desc="add_room_account_data", table="room_account_data", keyvalues={ @@ -348,7 +348,7 @@ class AccountDataStore(AccountDataWorkerStore): # no need to lock here as account_data has a unique constraint on # (user_id, account_data_type) so simple_upsert will retry if # there is a conflict. - yield self.simple_upsert( + yield self.db.simple_upsert( desc="add_user_account_data", table="account_data", keyvalues={"user_id": user_id, "account_data_type": account_data_type}, @@ -388,4 +388,4 @@ class AccountDataStore(AccountDataWorkerStore): ) txn.execute(update_max_id_sql, (next_id, next_id)) - return self.runInteraction("update_account_data_max_stream_id", _update) + return self.db.runInteraction("update_account_data_max_stream_id", _update) diff --git a/synapse/storage/data_stores/main/appservice.py b/synapse/storage/data_stores/main/appservice.py index 6b82fd392a..6b2e12719c 100644 --- a/synapse/storage/data_stores/main/appservice.py +++ b/synapse/storage/data_stores/main/appservice.py @@ -133,7 +133,7 @@ class ApplicationServiceTransactionWorkerStore( A Deferred which resolves to a list of ApplicationServices, which may be empty. """ - results = yield self.simple_select_list( + results = yield self.db.simple_select_list( "application_services_state", dict(state=state), ["as_id"] ) # NB: This assumes this class is linked with ApplicationServiceStore @@ -155,7 +155,7 @@ class ApplicationServiceTransactionWorkerStore( Returns: A Deferred which resolves to ApplicationServiceState. """ - result = yield self.simple_select_one( + result = yield self.db.simple_select_one( "application_services_state", dict(as_id=service.id), ["state"], @@ -175,7 +175,7 @@ class ApplicationServiceTransactionWorkerStore( Returns: A Deferred which resolves when the state was set successfully. """ - return self.simple_upsert( + return self.db.simple_upsert( "application_services_state", dict(as_id=service.id), dict(state=state) ) @@ -216,7 +216,7 @@ class ApplicationServiceTransactionWorkerStore( ) return AppServiceTransaction(service=service, id=new_txn_id, events=events) - return self.runInteraction("create_appservice_txn", _create_appservice_txn) + return self.db.runInteraction("create_appservice_txn", _create_appservice_txn) def complete_appservice_txn(self, txn_id, service): """Completes an application service transaction. @@ -249,7 +249,7 @@ class ApplicationServiceTransactionWorkerStore( ) # Set current txn_id for AS to 'txn_id' - self.simple_upsert_txn( + self.db.simple_upsert_txn( txn, "application_services_state", dict(as_id=service.id), @@ -257,11 +257,13 @@ class ApplicationServiceTransactionWorkerStore( ) # Delete txn - self.simple_delete_txn( + self.db.simple_delete_txn( txn, "application_services_txns", dict(txn_id=txn_id, as_id=service.id) ) - return self.runInteraction("complete_appservice_txn", _complete_appservice_txn) + return self.db.runInteraction( + "complete_appservice_txn", _complete_appservice_txn + ) @defer.inlineCallbacks def get_oldest_unsent_txn(self, service): @@ -283,7 +285,7 @@ class ApplicationServiceTransactionWorkerStore( " ORDER BY txn_id ASC LIMIT 1", (service.id,), ) - rows = self.cursor_to_dict(txn) + rows = self.db.cursor_to_dict(txn) if not rows: return None @@ -291,7 +293,7 @@ class ApplicationServiceTransactionWorkerStore( return entry - entry = yield self.runInteraction( + entry = yield self.db.runInteraction( "get_oldest_unsent_appservice_txn", _get_oldest_unsent_txn ) @@ -321,7 +323,7 @@ class ApplicationServiceTransactionWorkerStore( "UPDATE appservice_stream_position SET stream_ordering = ?", (pos,) ) - return self.runInteraction( + return self.db.runInteraction( "set_appservice_last_pos", set_appservice_last_pos_txn ) @@ -350,7 +352,7 @@ class ApplicationServiceTransactionWorkerStore( return upper_bound, [row[1] for row in rows] - upper_bound, event_ids = yield self.runInteraction( + upper_bound, event_ids = yield self.db.runInteraction( "get_new_events_for_appservice", get_new_events_for_appservice_txn ) diff --git a/synapse/storage/data_stores/main/cache.py b/synapse/storage/data_stores/main/cache.py index de3256049d..54ed8574c4 100644 --- a/synapse/storage/data_stores/main/cache.py +++ b/synapse/storage/data_stores/main/cache.py @@ -95,7 +95,7 @@ class CacheInvalidationStore(SQLBaseStore): txn.call_after(ctx.__exit__, None, None, None) txn.call_after(self.hs.get_notifier().on_new_replication_data) - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="cache_invalidation_stream", values={ @@ -122,7 +122,9 @@ class CacheInvalidationStore(SQLBaseStore): txn.execute(sql, (last_id, limit)) return txn.fetchall() - return self.runInteraction("get_all_updated_caches", get_all_updated_caches_txn) + return self.db.runInteraction( + "get_all_updated_caches", get_all_updated_caches_txn + ) def get_cache_stream_token(self): if self._cache_id_gen: diff --git a/synapse/storage/data_stores/main/client_ips.py b/synapse/storage/data_stores/main/client_ips.py index 66522a04b7..6f2a720b97 100644 --- a/synapse/storage/data_stores/main/client_ips.py +++ b/synapse/storage/data_stores/main/client_ips.py @@ -91,7 +91,7 @@ class ClientIpBackgroundUpdateStore(background_updates.BackgroundUpdateStore): txn.execute("DROP INDEX IF EXISTS user_ips_user_ip") txn.close() - yield self.runWithConnection(f) + yield self.db.runWithConnection(f) yield self._end_background_update("user_ips_drop_nonunique_index") return 1 @@ -106,7 +106,7 @@ class ClientIpBackgroundUpdateStore(background_updates.BackgroundUpdateStore): def user_ips_analyze(txn): txn.execute("ANALYZE user_ips") - yield self.runInteraction("user_ips_analyze", user_ips_analyze) + yield self.db.runInteraction("user_ips_analyze", user_ips_analyze) yield self._end_background_update("user_ips_analyze") @@ -140,7 +140,7 @@ class ClientIpBackgroundUpdateStore(background_updates.BackgroundUpdateStore): return None # Get a last seen that has roughly `batch_size` since `begin_last_seen` - end_last_seen = yield self.runInteraction( + end_last_seen = yield self.db.runInteraction( "user_ips_dups_get_last_seen", get_last_seen ) @@ -275,7 +275,7 @@ class ClientIpBackgroundUpdateStore(background_updates.BackgroundUpdateStore): txn, "user_ips_remove_dupes", {"last_seen": end_last_seen} ) - yield self.runInteraction("user_ips_dups_remove", remove) + yield self.db.runInteraction("user_ips_dups_remove", remove) if last: yield self._end_background_update("user_ips_remove_dupes") @@ -352,7 +352,7 @@ class ClientIpBackgroundUpdateStore(background_updates.BackgroundUpdateStore): return len(rows) - updated = yield self.runInteraction( + updated = yield self.db.runInteraction( "_devices_last_seen_update", _devices_last_seen_update_txn ) @@ -417,12 +417,12 @@ class ClientIpStore(ClientIpBackgroundUpdateStore): to_update = self._batch_row_update self._batch_row_update = {} - return self.runInteraction( + return self.db.runInteraction( "_update_client_ips_batch", self._update_client_ips_batch_txn, to_update ) def _update_client_ips_batch_txn(self, txn, to_update): - if "user_ips" in self._unsafe_to_upsert_tables or ( + if "user_ips" in self.db._unsafe_to_upsert_tables or ( not self.database_engine.can_native_upsert ): self.database_engine.lock_table(txn, "user_ips") @@ -431,7 +431,7 @@ class ClientIpStore(ClientIpBackgroundUpdateStore): (user_id, access_token, ip), (user_agent, device_id, last_seen) = entry try: - self.simple_upsert_txn( + self.db.simple_upsert_txn( txn, table="user_ips", keyvalues={ @@ -450,7 +450,7 @@ class ClientIpStore(ClientIpBackgroundUpdateStore): # Technically an access token might not be associated with # a device so we need to check. if device_id: - self.simple_upsert_txn( + self.db.simple_upsert_txn( txn, table="devices", keyvalues={"user_id": user_id, "device_id": device_id}, @@ -483,7 +483,7 @@ class ClientIpStore(ClientIpBackgroundUpdateStore): if device_id is not None: keyvalues["device_id"] = device_id - res = yield self.simple_select_list( + res = yield self.db.simple_select_list( table="devices", keyvalues=keyvalues, retcols=("user_id", "ip", "user_agent", "device_id", "last_seen"), @@ -516,7 +516,7 @@ class ClientIpStore(ClientIpBackgroundUpdateStore): user_agent, _, last_seen = self._batch_row_update[key] results[(access_token, ip)] = (user_agent, last_seen) - rows = yield self.simple_select_list( + rows = yield self.db.simple_select_list( table="user_ips", keyvalues={"user_id": user_id}, retcols=["access_token", "ip", "user_agent", "last_seen"], @@ -577,4 +577,4 @@ class ClientIpStore(ClientIpBackgroundUpdateStore): def _prune_old_user_ips_txn(txn): txn.execute(sql, (timestamp,)) - await self.runInteraction("_prune_old_user_ips", _prune_old_user_ips_txn) + await self.db.runInteraction("_prune_old_user_ips", _prune_old_user_ips_txn) diff --git a/synapse/storage/data_stores/main/deviceinbox.py b/synapse/storage/data_stores/main/deviceinbox.py index 206d39134d..440793ad49 100644 --- a/synapse/storage/data_stores/main/deviceinbox.py +++ b/synapse/storage/data_stores/main/deviceinbox.py @@ -69,7 +69,7 @@ class DeviceInboxWorkerStore(SQLBaseStore): stream_pos = current_stream_id return messages, stream_pos - return self.runInteraction( + return self.db.runInteraction( "get_new_messages_for_device", get_new_messages_for_device_txn ) @@ -109,7 +109,7 @@ class DeviceInboxWorkerStore(SQLBaseStore): txn.execute(sql, (user_id, device_id, up_to_stream_id)) return txn.rowcount - count = yield self.runInteraction( + count = yield self.db.runInteraction( "delete_messages_for_device", delete_messages_for_device_txn ) @@ -178,7 +178,7 @@ class DeviceInboxWorkerStore(SQLBaseStore): stream_pos = current_stream_id return messages, stream_pos - return self.runInteraction( + return self.db.runInteraction( "get_new_device_msgs_for_remote", get_new_messages_for_remote_destination_txn, ) @@ -203,7 +203,7 @@ class DeviceInboxWorkerStore(SQLBaseStore): ) txn.execute(sql, (destination, up_to_stream_id)) - return self.runInteraction( + return self.db.runInteraction( "delete_device_msgs_for_remote", delete_messages_for_remote_destination_txn ) @@ -232,7 +232,7 @@ class DeviceInboxBackgroundUpdateStore(BackgroundUpdateStore): txn.execute("DROP INDEX IF EXISTS device_inbox_stream_id") txn.close() - yield self.runWithConnection(reindex_txn) + yield self.db.runWithConnection(reindex_txn) yield self._end_background_update(self.DEVICE_INBOX_STREAM_ID) @@ -294,7 +294,7 @@ class DeviceInboxStore(DeviceInboxWorkerStore, DeviceInboxBackgroundUpdateStore) with self._device_inbox_id_gen.get_next() as stream_id: now_ms = self.clock.time_msec() - yield self.runInteraction( + yield self.db.runInteraction( "add_messages_to_device_inbox", add_messages_txn, now_ms, stream_id ) for user_id in local_messages_by_user_then_device.keys(): @@ -314,7 +314,7 @@ class DeviceInboxStore(DeviceInboxWorkerStore, DeviceInboxBackgroundUpdateStore) # Check if we've already inserted a matching message_id for that # origin. This can happen if the origin doesn't receive our # acknowledgement from the first time we received the message. - already_inserted = self.simple_select_one_txn( + already_inserted = self.db.simple_select_one_txn( txn, table="device_federation_inbox", keyvalues={"origin": origin, "message_id": message_id}, @@ -326,7 +326,7 @@ class DeviceInboxStore(DeviceInboxWorkerStore, DeviceInboxBackgroundUpdateStore) # Add an entry for this message_id so that we know we've processed # it. - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="device_federation_inbox", values={ @@ -344,7 +344,7 @@ class DeviceInboxStore(DeviceInboxWorkerStore, DeviceInboxBackgroundUpdateStore) with self._device_inbox_id_gen.get_next() as stream_id: now_ms = self.clock.time_msec() - yield self.runInteraction( + yield self.db.runInteraction( "add_messages_from_remote_to_device_inbox", add_messages_txn, now_ms, @@ -465,6 +465,6 @@ class DeviceInboxStore(DeviceInboxWorkerStore, DeviceInboxBackgroundUpdateStore) return rows - return self.runInteraction( + return self.db.runInteraction( "get_all_new_device_messages", get_all_new_device_messages_txn ) diff --git a/synapse/storage/data_stores/main/devices.py b/synapse/storage/data_stores/main/devices.py index 727c582121..d98511ddd4 100644 --- a/synapse/storage/data_stores/main/devices.py +++ b/synapse/storage/data_stores/main/devices.py @@ -61,7 +61,7 @@ class DeviceWorkerStore(SQLBaseStore): Raises: StoreError: if the device is not found """ - return self.simple_select_one( + return self.db.simple_select_one( table="devices", keyvalues={"user_id": user_id, "device_id": device_id, "hidden": False}, retcols=("user_id", "device_id", "display_name"), @@ -80,7 +80,7 @@ class DeviceWorkerStore(SQLBaseStore): containing "device_id", "user_id" and "display_name" for each device. """ - devices = yield self.simple_select_list( + devices = yield self.db.simple_select_list( table="devices", keyvalues={"user_id": user_id, "hidden": False}, retcols=("user_id", "device_id", "display_name"), @@ -122,7 +122,7 @@ class DeviceWorkerStore(SQLBaseStore): # consider the device update to be too large, and simply skip the # stream_id; the rationale being that such a large device list update # is likely an error. - updates = yield self.runInteraction( + updates = yield self.db.runInteraction( "get_device_updates_by_remote", self._get_device_updates_by_remote_txn, destination, @@ -283,7 +283,7 @@ class DeviceWorkerStore(SQLBaseStore): """ devices = ( - yield self.runInteraction( + yield self.db.runInteraction( "_get_e2e_device_keys_txn", self._get_e2e_device_keys_txn, query_map.keys(), @@ -340,12 +340,12 @@ class DeviceWorkerStore(SQLBaseStore): rows = txn.fetchall() return rows[0][0] - return self.runInteraction("get_last_device_update_for_remote_user", f) + return self.db.runInteraction("get_last_device_update_for_remote_user", f) def mark_as_sent_devices_by_remote(self, destination, stream_id): """Mark that updates have successfully been sent to the destination. """ - return self.runInteraction( + return self.db.runInteraction( "mark_as_sent_devices_by_remote", self._mark_as_sent_devices_by_remote_txn, destination, @@ -399,7 +399,7 @@ class DeviceWorkerStore(SQLBaseStore): """ with self._device_list_id_gen.get_next() as stream_id: - yield self.runInteraction( + yield self.db.runInteraction( "add_user_sig_change_to_streams", self._add_user_signature_change_txn, from_user_id, @@ -414,7 +414,7 @@ class DeviceWorkerStore(SQLBaseStore): from_user_id, stream_id, ) - self.simple_insert_txn( + self.db.simple_insert_txn( txn, "user_signature_stream", values={ @@ -466,7 +466,7 @@ class DeviceWorkerStore(SQLBaseStore): @cachedInlineCallbacks(num_args=2, tree=True) def _get_cached_user_device(self, user_id, device_id): - content = yield self.simple_select_one_onecol( + content = yield self.db.simple_select_one_onecol( table="device_lists_remote_cache", keyvalues={"user_id": user_id, "device_id": device_id}, retcol="content", @@ -476,7 +476,7 @@ class DeviceWorkerStore(SQLBaseStore): @cachedInlineCallbacks() def _get_cached_devices_for_user(self, user_id): - devices = yield self.simple_select_list( + devices = yield self.db.simple_select_list( table="device_lists_remote_cache", keyvalues={"user_id": user_id}, retcols=("device_id", "content"), @@ -492,7 +492,7 @@ class DeviceWorkerStore(SQLBaseStore): Returns: (stream_id, devices) """ - return self.runInteraction( + return self.db.runInteraction( "get_devices_with_keys_by_user", self._get_devices_with_keys_by_user_txn, user_id, @@ -565,7 +565,7 @@ class DeviceWorkerStore(SQLBaseStore): return changes - return self.runInteraction( + return self.db.runInteraction( "get_users_whose_devices_changed", _get_users_whose_devices_changed_txn ) @@ -584,7 +584,7 @@ class DeviceWorkerStore(SQLBaseStore): SELECT DISTINCT user_ids FROM user_signature_stream WHERE from_user_id = ? AND stream_id > ? """ - rows = yield self.execute( + rows = yield self.db.execute( "get_users_whose_signatures_changed", None, sql, user_id, from_key ) return set(user for row in rows for user in json.loads(row[0])) @@ -605,7 +605,7 @@ class DeviceWorkerStore(SQLBaseStore): WHERE ? < stream_id AND stream_id <= ? GROUP BY user_id, destination """ - return self.execute( + return self.db.execute( "get_all_device_list_changes_for_remotes", None, sql, from_key, to_key ) @@ -614,7 +614,7 @@ class DeviceWorkerStore(SQLBaseStore): """Get the last stream_id we got for a user. May be None if we haven't got any information for them. """ - return self.simple_select_one_onecol( + return self.db.simple_select_one_onecol( table="device_lists_remote_extremeties", keyvalues={"user_id": user_id}, retcol="stream_id", @@ -628,7 +628,7 @@ class DeviceWorkerStore(SQLBaseStore): inlineCallbacks=True, ) def get_device_list_last_stream_id_for_remotes(self, user_ids): - rows = yield self.simple_select_many_batch( + rows = yield self.db.simple_select_many_batch( table="device_lists_remote_extremeties", column="user_id", iterable=user_ids, @@ -685,7 +685,7 @@ class DeviceBackgroundUpdateStore(BackgroundUpdateStore): txn.execute("DROP INDEX IF EXISTS device_lists_remote_extremeties_id") txn.close() - yield self.runWithConnection(f) + yield self.db.runWithConnection(f) yield self._end_background_update(DROP_DEVICE_LIST_STREAMS_NON_UNIQUE_INDEXES) return 1 @@ -722,7 +722,7 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): return False try: - inserted = yield self.simple_insert( + inserted = yield self.db.simple_insert( "devices", values={ "user_id": user_id, @@ -736,7 +736,7 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): if not inserted: # if the device already exists, check if it's a real device, or # if the device ID is reserved by something else - hidden = yield self.simple_select_one_onecol( + hidden = yield self.db.simple_select_one_onecol( "devices", keyvalues={"user_id": user_id, "device_id": device_id}, retcol="hidden", @@ -771,7 +771,7 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): Returns: defer.Deferred """ - yield self.simple_delete_one( + yield self.db.simple_delete_one( table="devices", keyvalues={"user_id": user_id, "device_id": device_id, "hidden": False}, desc="delete_device", @@ -789,7 +789,7 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): Returns: defer.Deferred """ - yield self.simple_delete_many( + yield self.db.simple_delete_many( table="devices", column="device_id", iterable=device_ids, @@ -818,7 +818,7 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): updates["display_name"] = new_display_name if not updates: return defer.succeed(None) - return self.simple_update_one( + return self.db.simple_update_one( table="devices", keyvalues={"user_id": user_id, "device_id": device_id, "hidden": False}, updatevalues=updates, @@ -829,7 +829,7 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): def mark_remote_user_device_list_as_unsubscribed(self, user_id): """Mark that we no longer track device lists for remote user. """ - yield self.simple_delete( + yield self.db.simple_delete( table="device_lists_remote_extremeties", keyvalues={"user_id": user_id}, desc="mark_remote_user_device_list_as_unsubscribed", @@ -853,7 +853,7 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): Returns: Deferred[None] """ - return self.runInteraction( + return self.db.runInteraction( "update_remote_device_list_cache_entry", self._update_remote_device_list_cache_entry_txn, user_id, @@ -866,7 +866,7 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): self, txn, user_id, device_id, content, stream_id ): if content.get("deleted"): - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="device_lists_remote_cache", keyvalues={"user_id": user_id, "device_id": device_id}, @@ -874,7 +874,7 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): txn.call_after(self.device_id_exists_cache.invalidate, (user_id, device_id)) else: - self.simple_upsert_txn( + self.db.simple_upsert_txn( txn, table="device_lists_remote_cache", keyvalues={"user_id": user_id, "device_id": device_id}, @@ -890,7 +890,7 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): self.get_device_list_last_stream_id_for_remote.invalidate, (user_id,) ) - self.simple_upsert_txn( + self.db.simple_upsert_txn( txn, table="device_lists_remote_extremeties", keyvalues={"user_id": user_id}, @@ -914,7 +914,7 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): Returns: Deferred[None] """ - return self.runInteraction( + return self.db.runInteraction( "update_remote_device_list_cache", self._update_remote_device_list_cache_txn, user_id, @@ -923,11 +923,11 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): ) def _update_remote_device_list_cache_txn(self, txn, user_id, devices, stream_id): - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="device_lists_remote_cache", keyvalues={"user_id": user_id} ) - self.simple_insert_many_txn( + self.db.simple_insert_many_txn( txn, table="device_lists_remote_cache", values=[ @@ -946,7 +946,7 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): self.get_device_list_last_stream_id_for_remote.invalidate, (user_id,) ) - self.simple_upsert_txn( + self.db.simple_upsert_txn( txn, table="device_lists_remote_extremeties", keyvalues={"user_id": user_id}, @@ -962,7 +962,7 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): (if any) should be poked. """ with self._device_list_id_gen.get_next() as stream_id: - yield self.runInteraction( + yield self.db.runInteraction( "add_device_change_to_streams", self._add_device_change_txn, user_id, @@ -995,7 +995,7 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): [(user_id, device_id, stream_id) for device_id in device_ids], ) - self.simple_insert_many_txn( + self.db.simple_insert_many_txn( txn, table="device_lists_stream", values=[ @@ -1006,7 +1006,7 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): context = get_active_span_text_map() - self.simple_insert_many_txn( + self.db.simple_insert_many_txn( txn, table="device_lists_outbound_pokes", values=[ @@ -1069,7 +1069,7 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): return run_as_background_process( "prune_old_outbound_device_pokes", - self.runInteraction, + self.db.runInteraction, "_prune_old_outbound_device_pokes", _prune_txn, ) diff --git a/synapse/storage/data_stores/main/directory.py b/synapse/storage/data_stores/main/directory.py index d332f8a409..c9e7de7d12 100644 --- a/synapse/storage/data_stores/main/directory.py +++ b/synapse/storage/data_stores/main/directory.py @@ -36,7 +36,7 @@ class DirectoryWorkerStore(SQLBaseStore): Deferred: results in namedtuple with keys "room_id" and "servers" or None if no association can be found """ - room_id = yield self.simple_select_one_onecol( + room_id = yield self.db.simple_select_one_onecol( "room_aliases", {"room_alias": room_alias.to_string()}, "room_id", @@ -47,7 +47,7 @@ class DirectoryWorkerStore(SQLBaseStore): if not room_id: return None - servers = yield self.simple_select_onecol( + servers = yield self.db.simple_select_onecol( "room_alias_servers", {"room_alias": room_alias.to_string()}, "server", @@ -60,7 +60,7 @@ class DirectoryWorkerStore(SQLBaseStore): return RoomAliasMapping(room_id, room_alias.to_string(), servers) def get_room_alias_creator(self, room_alias): - return self.simple_select_one_onecol( + return self.db.simple_select_one_onecol( table="room_aliases", keyvalues={"room_alias": room_alias}, retcol="creator", @@ -69,7 +69,7 @@ class DirectoryWorkerStore(SQLBaseStore): @cached(max_entries=5000) def get_aliases_for_room(self, room_id): - return self.simple_select_onecol( + return self.db.simple_select_onecol( "room_aliases", {"room_id": room_id}, "room_alias", @@ -93,7 +93,7 @@ class DirectoryStore(DirectoryWorkerStore): """ def alias_txn(txn): - self.simple_insert_txn( + self.db.simple_insert_txn( txn, "room_aliases", { @@ -103,7 +103,7 @@ class DirectoryStore(DirectoryWorkerStore): }, ) - self.simple_insert_many_txn( + self.db.simple_insert_many_txn( txn, table="room_alias_servers", values=[ @@ -117,7 +117,9 @@ class DirectoryStore(DirectoryWorkerStore): ) try: - ret = yield self.runInteraction("create_room_alias_association", alias_txn) + ret = yield self.db.runInteraction( + "create_room_alias_association", alias_txn + ) except self.database_engine.module.IntegrityError: raise SynapseError( 409, "Room alias %s already exists" % room_alias.to_string() @@ -126,7 +128,7 @@ class DirectoryStore(DirectoryWorkerStore): @defer.inlineCallbacks def delete_room_alias(self, room_alias): - room_id = yield self.runInteraction( + room_id = yield self.db.runInteraction( "delete_room_alias", self._delete_room_alias_txn, room_alias ) @@ -168,6 +170,6 @@ class DirectoryStore(DirectoryWorkerStore): txn, self.get_aliases_for_room, (new_room_id,) ) - return self.runInteraction( + return self.db.runInteraction( "_update_aliases_for_room_txn", _update_aliases_for_room_txn ) diff --git a/synapse/storage/data_stores/main/e2e_room_keys.py b/synapse/storage/data_stores/main/e2e_room_keys.py index df89eda337..84594cf0a9 100644 --- a/synapse/storage/data_stores/main/e2e_room_keys.py +++ b/synapse/storage/data_stores/main/e2e_room_keys.py @@ -38,7 +38,7 @@ class EndToEndRoomKeyStore(SQLBaseStore): StoreError """ - yield self.simple_update_one( + yield self.db.simple_update_one( table="e2e_room_keys", keyvalues={ "user_id": user_id, @@ -89,7 +89,7 @@ class EndToEndRoomKeyStore(SQLBaseStore): } ) - yield self.simple_insert_many( + yield self.db.simple_insert_many( table="e2e_room_keys", values=values, desc="add_e2e_room_keys" ) @@ -125,7 +125,7 @@ class EndToEndRoomKeyStore(SQLBaseStore): if session_id: keyvalues["session_id"] = session_id - rows = yield self.simple_select_list( + rows = yield self.db.simple_select_list( table="e2e_room_keys", keyvalues=keyvalues, retcols=( @@ -170,7 +170,7 @@ class EndToEndRoomKeyStore(SQLBaseStore): Deferred[dict[str, dict[str, dict]]]: a map of room IDs to session IDs to room key """ - return self.runInteraction( + return self.db.runInteraction( "get_e2e_room_keys_multi", self._get_e2e_room_keys_multi_txn, user_id, @@ -234,7 +234,7 @@ class EndToEndRoomKeyStore(SQLBaseStore): version (str): the version ID of the backup we're querying about """ - return self.simple_select_one_onecol( + return self.db.simple_select_one_onecol( table="e2e_room_keys", keyvalues={"user_id": user_id, "version": version}, retcol="COUNT(*)", @@ -267,7 +267,7 @@ class EndToEndRoomKeyStore(SQLBaseStore): if session_id: keyvalues["session_id"] = session_id - yield self.simple_delete( + yield self.db.simple_delete( table="e2e_room_keys", keyvalues=keyvalues, desc="delete_e2e_room_keys" ) @@ -312,7 +312,7 @@ class EndToEndRoomKeyStore(SQLBaseStore): # it isn't there. raise StoreError(404, "No row found") - result = self.simple_select_one_txn( + result = self.db.simple_select_one_txn( txn, table="e2e_room_keys_versions", keyvalues={"user_id": user_id, "version": this_version, "deleted": 0}, @@ -324,7 +324,7 @@ class EndToEndRoomKeyStore(SQLBaseStore): result["etag"] = 0 return result - return self.runInteraction( + return self.db.runInteraction( "get_e2e_room_keys_version_info", _get_e2e_room_keys_version_info_txn ) @@ -352,7 +352,7 @@ class EndToEndRoomKeyStore(SQLBaseStore): new_version = str(int(current_version) + 1) - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="e2e_room_keys_versions", values={ @@ -365,7 +365,7 @@ class EndToEndRoomKeyStore(SQLBaseStore): return new_version - return self.runInteraction( + return self.db.runInteraction( "create_e2e_room_keys_version_txn", _create_e2e_room_keys_version_txn ) @@ -391,7 +391,7 @@ class EndToEndRoomKeyStore(SQLBaseStore): updatevalues["etag"] = version_etag if updatevalues: - return self.simple_update( + return self.db.simple_update( table="e2e_room_keys_versions", keyvalues={"user_id": user_id, "version": version}, updatevalues=updatevalues, @@ -420,19 +420,19 @@ class EndToEndRoomKeyStore(SQLBaseStore): else: this_version = version - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="e2e_room_keys", keyvalues={"user_id": user_id, "version": this_version}, ) - return self.simple_update_one_txn( + return self.db.simple_update_one_txn( txn, table="e2e_room_keys_versions", keyvalues={"user_id": user_id, "version": this_version}, updatevalues={"deleted": 1}, ) - return self.runInteraction( + return self.db.runInteraction( "delete_e2e_room_keys_version", _delete_e2e_room_keys_version_txn ) diff --git a/synapse/storage/data_stores/main/end_to_end_keys.py b/synapse/storage/data_stores/main/end_to_end_keys.py index 08bcdc4725..38cd0ca9b8 100644 --- a/synapse/storage/data_stores/main/end_to_end_keys.py +++ b/synapse/storage/data_stores/main/end_to_end_keys.py @@ -48,7 +48,7 @@ class EndToEndKeyWorkerStore(SQLBaseStore): if not query_list: return {} - results = yield self.runInteraction( + results = yield self.db.runInteraction( "get_e2e_device_keys", self._get_e2e_device_keys_txn, query_list, @@ -125,7 +125,7 @@ class EndToEndKeyWorkerStore(SQLBaseStore): ) txn.execute(sql, query_params) - rows = self.cursor_to_dict(txn) + rows = self.db.cursor_to_dict(txn) result = {} for row in rows: @@ -143,7 +143,7 @@ class EndToEndKeyWorkerStore(SQLBaseStore): ) txn.execute(signature_sql, signature_query_params) - rows = self.cursor_to_dict(txn) + rows = self.db.cursor_to_dict(txn) # add each cross-signing signature to the correct device in the result dict. for row in rows: @@ -186,7 +186,7 @@ class EndToEndKeyWorkerStore(SQLBaseStore): key_id) to json string for key """ - rows = yield self.simple_select_many_batch( + rows = yield self.db.simple_select_many_batch( table="e2e_one_time_keys_json", column="key_id", iterable=key_ids, @@ -219,7 +219,7 @@ class EndToEndKeyWorkerStore(SQLBaseStore): # a unique constraint. If there is a race of two calls to # `add_e2e_one_time_keys` then they'll conflict and we will only # insert one set. - self.simple_insert_many_txn( + self.db.simple_insert_many_txn( txn, table="e2e_one_time_keys_json", values=[ @@ -238,7 +238,7 @@ class EndToEndKeyWorkerStore(SQLBaseStore): txn, self.count_e2e_one_time_keys, (user_id, device_id) ) - yield self.runInteraction( + yield self.db.runInteraction( "add_e2e_one_time_keys_insert", _add_e2e_one_time_keys ) @@ -261,7 +261,9 @@ class EndToEndKeyWorkerStore(SQLBaseStore): result[algorithm] = key_count return result - return self.runInteraction("count_e2e_one_time_keys", _count_e2e_one_time_keys) + return self.db.runInteraction( + "count_e2e_one_time_keys", _count_e2e_one_time_keys + ) def _get_e2e_cross_signing_key_txn(self, txn, user_id, key_type, from_user_id=None): """Returns a user's cross-signing key. @@ -322,7 +324,7 @@ class EndToEndKeyWorkerStore(SQLBaseStore): Returns: dict of the key data or None if not found """ - return self.runInteraction( + return self.db.runInteraction( "get_e2e_cross_signing_key", self._get_e2e_cross_signing_key_txn, user_id, @@ -350,7 +352,7 @@ class EndToEndKeyWorkerStore(SQLBaseStore): WHERE ? < stream_id AND stream_id <= ? GROUP BY user_id """ - return self.execute( + return self.db.execute( "get_all_user_signature_changes_for_remotes", None, sql, from_key, to_key ) @@ -367,7 +369,7 @@ class EndToEndKeyStore(EndToEndKeyWorkerStore, SQLBaseStore): set_tag("time_now", time_now) set_tag("device_keys", device_keys) - old_key_json = self.simple_select_one_onecol_txn( + old_key_json = self.db.simple_select_one_onecol_txn( txn, table="e2e_device_keys_json", keyvalues={"user_id": user_id, "device_id": device_id}, @@ -383,7 +385,7 @@ class EndToEndKeyStore(EndToEndKeyWorkerStore, SQLBaseStore): log_kv({"Message": "Device key already stored."}) return False - self.simple_upsert_txn( + self.db.simple_upsert_txn( txn, table="e2e_device_keys_json", keyvalues={"user_id": user_id, "device_id": device_id}, @@ -392,7 +394,7 @@ class EndToEndKeyStore(EndToEndKeyWorkerStore, SQLBaseStore): log_kv({"message": "Device keys stored."}) return True - return self.runInteraction("set_e2e_device_keys", _set_e2e_device_keys_txn) + return self.db.runInteraction("set_e2e_device_keys", _set_e2e_device_keys_txn) def claim_e2e_one_time_keys(self, query_list): """Take a list of one time keys out of the database""" @@ -431,7 +433,9 @@ class EndToEndKeyStore(EndToEndKeyWorkerStore, SQLBaseStore): ) return result - return self.runInteraction("claim_e2e_one_time_keys", _claim_e2e_one_time_keys) + return self.db.runInteraction( + "claim_e2e_one_time_keys", _claim_e2e_one_time_keys + ) def delete_e2e_keys_by_device(self, user_id, device_id): def delete_e2e_keys_by_device_txn(txn): @@ -442,12 +446,12 @@ class EndToEndKeyStore(EndToEndKeyWorkerStore, SQLBaseStore): "user_id": user_id, } ) - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="e2e_device_keys_json", keyvalues={"user_id": user_id, "device_id": device_id}, ) - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="e2e_one_time_keys_json", keyvalues={"user_id": user_id, "device_id": device_id}, @@ -456,7 +460,7 @@ class EndToEndKeyStore(EndToEndKeyWorkerStore, SQLBaseStore): txn, self.count_e2e_one_time_keys, (user_id, device_id) ) - return self.runInteraction( + return self.db.runInteraction( "delete_e2e_keys_by_device", delete_e2e_keys_by_device_txn ) @@ -492,7 +496,7 @@ class EndToEndKeyStore(EndToEndKeyWorkerStore, SQLBaseStore): # The "keys" property must only have one entry, which will be the public # key, so we just grab the first value in there pubkey = next(iter(key["keys"].values())) - self.simple_insert_txn( + self.db.simple_insert_txn( txn, "devices", values={ @@ -505,7 +509,7 @@ class EndToEndKeyStore(EndToEndKeyWorkerStore, SQLBaseStore): # and finally, store the key itself with self._cross_signing_id_gen.get_next() as stream_id: - self.simple_insert_txn( + self.db.simple_insert_txn( txn, "e2e_cross_signing_keys", values={ @@ -524,7 +528,7 @@ class EndToEndKeyStore(EndToEndKeyWorkerStore, SQLBaseStore): key_type (str): the type of cross-signing key to set key (dict): the key data """ - return self.runInteraction( + return self.db.runInteraction( "add_e2e_cross_signing_key", self._set_e2e_cross_signing_key_txn, user_id, @@ -539,7 +543,7 @@ class EndToEndKeyStore(EndToEndKeyWorkerStore, SQLBaseStore): user_id (str): the user who made the signatures signatures (iterable[SignatureListItem]): signatures to add """ - return self.simple_insert_many( + return self.db.simple_insert_many( "e2e_cross_signing_signatures", [ { diff --git a/synapse/storage/data_stores/main/event_federation.py b/synapse/storage/data_stores/main/event_federation.py index 051ac7a8cb..77e4353b59 100644 --- a/synapse/storage/data_stores/main/event_federation.py +++ b/synapse/storage/data_stores/main/event_federation.py @@ -58,7 +58,7 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBas Returns: list of event_ids """ - return self.runInteraction( + return self.db.runInteraction( "get_auth_chain_ids", self._get_auth_chain_ids_txn, event_ids, include_given ) @@ -90,12 +90,12 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBas return list(results) def get_oldest_events_in_room(self, room_id): - return self.runInteraction( + return self.db.runInteraction( "get_oldest_events_in_room", self._get_oldest_events_in_room_txn, room_id ) def get_oldest_events_with_depth_in_room(self, room_id): - return self.runInteraction( + return self.db.runInteraction( "get_oldest_events_with_depth_in_room", self.get_oldest_events_with_depth_in_room_txn, room_id, @@ -126,7 +126,7 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBas Returns Deferred[int] """ - rows = yield self.simple_select_many_batch( + rows = yield self.db.simple_select_many_batch( table="events", column="event_id", iterable=event_ids, @@ -140,7 +140,7 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBas return max(row["depth"] for row in rows) def _get_oldest_events_in_room_txn(self, txn, room_id): - return self.simple_select_onecol_txn( + return self.db.simple_select_onecol_txn( txn, table="event_backward_extremities", keyvalues={"room_id": room_id}, @@ -188,7 +188,7 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBas where *hashes* is a map from algorithm to hash. """ - return self.runInteraction( + return self.db.runInteraction( "get_latest_event_ids_and_hashes_in_room", self._get_latest_event_ids_and_hashes_in_room, room_id, @@ -229,13 +229,13 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBas txn.execute(sql, query_args) return [room_id for room_id, in txn] - return self.runInteraction( + return self.db.runInteraction( "get_rooms_with_many_extremities", _get_rooms_with_many_extremities_txn ) @cached(max_entries=5000, iterable=True) def get_latest_event_ids_in_room(self, room_id): - return self.simple_select_onecol( + return self.db.simple_select_onecol( table="event_forward_extremities", keyvalues={"room_id": room_id}, retcol="event_id", @@ -266,12 +266,12 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBas def get_min_depth(self, room_id): """ For hte given room, get the minimum depth we have seen for it. """ - return self.runInteraction( + return self.db.runInteraction( "get_min_depth", self._get_min_depth_interaction, room_id ) def _get_min_depth_interaction(self, txn, room_id): - min_depth = self.simple_select_one_onecol_txn( + min_depth = self.db.simple_select_one_onecol_txn( txn, table="room_depth", keyvalues={"room_id": room_id}, @@ -337,7 +337,7 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBas txn.execute(sql, (stream_ordering, room_id)) return [event_id for event_id, in txn] - return self.runInteraction( + return self.db.runInteraction( "get_forward_extremeties_for_room", get_forward_extremeties_for_room_txn ) @@ -352,7 +352,7 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBas limit (int) """ return ( - self.runInteraction( + self.db.runInteraction( "get_backfill_events", self._get_backfill_events, room_id, @@ -383,7 +383,7 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBas queue = PriorityQueue() for event_id in event_list: - depth = self.simple_select_one_onecol_txn( + depth = self.db.simple_select_one_onecol_txn( txn, table="events", keyvalues={"event_id": event_id, "room_id": room_id}, @@ -415,7 +415,7 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBas @defer.inlineCallbacks def get_missing_events(self, room_id, earliest_events, latest_events, limit): - ids = yield self.runInteraction( + ids = yield self.db.runInteraction( "get_missing_events", self._get_missing_events, room_id, @@ -468,7 +468,7 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBas Returns: Deferred[list[str]] """ - rows = yield self.simple_select_many_batch( + rows = yield self.db.simple_select_many_batch( table="event_edges", column="prev_event_id", iterable=event_ids, @@ -508,7 +508,7 @@ class EventFederationStore(EventFederationWorkerStore): if min_depth and depth >= min_depth: return - self.simple_upsert_txn( + self.db.simple_upsert_txn( txn, table="room_depth", keyvalues={"room_id": room_id}, @@ -520,7 +520,7 @@ class EventFederationStore(EventFederationWorkerStore): For the given event, update the event edges table and forward and backward extremities tables. """ - self.simple_insert_many_txn( + self.db.simple_insert_many_txn( txn, table="event_edges", values=[ @@ -604,13 +604,13 @@ class EventFederationStore(EventFederationWorkerStore): return run_as_background_process( "delete_old_forward_extrem_cache", - self.runInteraction, + self.db.runInteraction, "_delete_old_forward_extrem_cache", _delete_old_forward_extrem_cache_txn, ) def clean_room_for_join(self, room_id): - return self.runInteraction( + return self.db.runInteraction( "clean_room_for_join", self._clean_room_for_join_txn, room_id ) @@ -660,7 +660,7 @@ class EventFederationStore(EventFederationWorkerStore): return min_stream_id >= target_min_stream_id - result = yield self.runInteraction( + result = yield self.db.runInteraction( self.EVENT_AUTH_STATE_ONLY, delete_event_auth ) diff --git a/synapse/storage/data_stores/main/event_push_actions.py b/synapse/storage/data_stores/main/event_push_actions.py index 0a37847cfd..725d0881dc 100644 --- a/synapse/storage/data_stores/main/event_push_actions.py +++ b/synapse/storage/data_stores/main/event_push_actions.py @@ -93,7 +93,7 @@ class EventPushActionsWorkerStore(SQLBaseStore): def get_unread_event_push_actions_by_room_for_user( self, room_id, user_id, last_read_event_id ): - ret = yield self.runInteraction( + ret = yield self.db.runInteraction( "get_unread_event_push_actions_by_room", self._get_unread_counts_by_receipt_txn, room_id, @@ -177,7 +177,7 @@ class EventPushActionsWorkerStore(SQLBaseStore): txn.execute(sql, (min_stream_ordering, max_stream_ordering)) return [r[0] for r in txn] - ret = yield self.runInteraction("get_push_action_users_in_range", f) + ret = yield self.db.runInteraction("get_push_action_users_in_range", f) return ret @defer.inlineCallbacks @@ -229,7 +229,7 @@ class EventPushActionsWorkerStore(SQLBaseStore): txn.execute(sql, args) return txn.fetchall() - after_read_receipt = yield self.runInteraction( + after_read_receipt = yield self.db.runInteraction( "get_unread_push_actions_for_user_in_range_http_arr", get_after_receipt ) @@ -257,7 +257,7 @@ class EventPushActionsWorkerStore(SQLBaseStore): txn.execute(sql, args) return txn.fetchall() - no_read_receipt = yield self.runInteraction( + no_read_receipt = yield self.db.runInteraction( "get_unread_push_actions_for_user_in_range_http_nrr", get_no_receipt ) @@ -329,7 +329,7 @@ class EventPushActionsWorkerStore(SQLBaseStore): txn.execute(sql, args) return txn.fetchall() - after_read_receipt = yield self.runInteraction( + after_read_receipt = yield self.db.runInteraction( "get_unread_push_actions_for_user_in_range_email_arr", get_after_receipt ) @@ -357,7 +357,7 @@ class EventPushActionsWorkerStore(SQLBaseStore): txn.execute(sql, args) return txn.fetchall() - no_read_receipt = yield self.runInteraction( + no_read_receipt = yield self.db.runInteraction( "get_unread_push_actions_for_user_in_range_email_nrr", get_no_receipt ) @@ -407,7 +407,7 @@ class EventPushActionsWorkerStore(SQLBaseStore): txn.execute(sql, (user_id, min_stream_ordering)) return bool(txn.fetchone()) - return self.runInteraction( + return self.db.runInteraction( "get_if_maybe_push_in_range_for_user", _get_if_maybe_push_in_range_for_user_txn, ) @@ -458,7 +458,7 @@ class EventPushActionsWorkerStore(SQLBaseStore): ), ) - return self.runInteraction( + return self.db.runInteraction( "add_push_actions_to_staging", _add_push_actions_to_staging_txn ) @@ -472,7 +472,7 @@ class EventPushActionsWorkerStore(SQLBaseStore): """ try: - res = yield self.simple_delete( + res = yield self.db.simple_delete( table="event_push_actions_staging", keyvalues={"event_id": event_id}, desc="remove_push_actions_from_staging", @@ -489,7 +489,7 @@ class EventPushActionsWorkerStore(SQLBaseStore): def _find_stream_orderings_for_times(self): return run_as_background_process( "event_push_action_stream_orderings", - self.runInteraction, + self.db.runInteraction, "_find_stream_orderings_for_times", self._find_stream_orderings_for_times_txn, ) @@ -525,7 +525,7 @@ class EventPushActionsWorkerStore(SQLBaseStore): Deferred[int]: stream ordering of the first event received on/after the timestamp """ - return self.runInteraction( + return self.db.runInteraction( "_find_first_stream_ordering_after_ts_txn", self._find_first_stream_ordering_after_ts_txn, ts, @@ -677,7 +677,7 @@ class EventPushActionsStore(EventPushActionsWorkerStore): ) for event, _ in events_and_contexts: - user_ids = self.simple_select_onecol_txn( + user_ids = self.db.simple_select_onecol_txn( txn, table="event_push_actions_staging", keyvalues={"event_id": event.event_id}, @@ -727,9 +727,9 @@ class EventPushActionsStore(EventPushActionsWorkerStore): " LIMIT ?" % (before_clause,) ) txn.execute(sql, args) - return self.cursor_to_dict(txn) + return self.db.cursor_to_dict(txn) - push_actions = yield self.runInteraction("get_push_actions_for_user", f) + push_actions = yield self.db.runInteraction("get_push_actions_for_user", f) for pa in push_actions: pa["actions"] = _deserialize_action(pa["actions"], pa["highlight"]) return push_actions @@ -748,7 +748,7 @@ class EventPushActionsStore(EventPushActionsWorkerStore): txn.execute(sql, (stream_ordering,)) return txn.fetchone() - result = yield self.runInteraction("get_time_of_last_push_action_before", f) + result = yield self.db.runInteraction("get_time_of_last_push_action_before", f) return result[0] if result else None @defer.inlineCallbacks @@ -757,7 +757,9 @@ class EventPushActionsStore(EventPushActionsWorkerStore): txn.execute("SELECT MAX(stream_ordering) FROM event_push_actions") return txn.fetchone() - result = yield self.runInteraction("get_latest_push_action_stream_ordering", f) + result = yield self.db.runInteraction( + "get_latest_push_action_stream_ordering", f + ) return result[0] or 0 def _remove_push_actions_for_event_id_txn(self, txn, room_id, event_id): @@ -830,7 +832,7 @@ class EventPushActionsStore(EventPushActionsWorkerStore): while True: logger.info("Rotating notifications") - caught_up = yield self.runInteraction( + caught_up = yield self.db.runInteraction( "_rotate_notifs", self._rotate_notifs_txn ) if caught_up: @@ -844,7 +846,7 @@ class EventPushActionsStore(EventPushActionsWorkerStore): the archiving process has caught up or not. """ - old_rotate_stream_ordering = self.simple_select_one_onecol_txn( + old_rotate_stream_ordering = self.db.simple_select_one_onecol_txn( txn, table="event_push_summary_stream_ordering", keyvalues={}, @@ -880,7 +882,7 @@ class EventPushActionsStore(EventPushActionsWorkerStore): return caught_up def _rotate_notifs_before_txn(self, txn, rotate_to_stream_ordering): - old_rotate_stream_ordering = self.simple_select_one_onecol_txn( + old_rotate_stream_ordering = self.db.simple_select_one_onecol_txn( txn, table="event_push_summary_stream_ordering", keyvalues={}, @@ -912,7 +914,7 @@ class EventPushActionsStore(EventPushActionsWorkerStore): # If the `old.user_id` above is NULL then we know there isn't already an # entry in the table, so we simply insert it. Otherwise we update the # existing table. - self.simple_insert_many_txn( + self.db.simple_insert_many_txn( txn, table="event_push_summary", values=[ diff --git a/synapse/storage/data_stores/main/events.py b/synapse/storage/data_stores/main/events.py index 98ae69e996..01ec9ec397 100644 --- a/synapse/storage/data_stores/main/events.py +++ b/synapse/storage/data_stores/main/events.py @@ -143,7 +143,7 @@ class EventsStore( ) return txn.fetchall() - res = yield self.runInteraction("read_forward_extremities", fetch) + res = yield self.db.runInteraction("read_forward_extremities", fetch) self._current_forward_extremities_amount = c_counter(list(x[0] for x in res)) @_retry_on_integrity_error @@ -208,7 +208,7 @@ class EventsStore( for (event, context), stream in zip(events_and_contexts, stream_orderings): event.internal_metadata.stream_ordering = stream - yield self.runInteraction( + yield self.db.runInteraction( "persist_events", self._persist_events_txn, events_and_contexts=events_and_contexts, @@ -281,7 +281,7 @@ class EventsStore( results.extend(r[0] for r in txn if not json.loads(r[1]).get("soft_failed")) for chunk in batch_iter(event_ids, 100): - yield self.runInteraction( + yield self.db.runInteraction( "_get_events_which_are_prevs", _get_events_which_are_prevs_txn, chunk ) @@ -345,7 +345,7 @@ class EventsStore( existing_prevs.add(prev_event_id) for chunk in batch_iter(event_ids, 100): - yield self.runInteraction( + yield self.db.runInteraction( "_get_prevs_before_rejected", _get_prevs_before_rejected_txn, chunk ) @@ -432,7 +432,7 @@ class EventsStore( # event's auth chain, but its easier for now just to store them (and # it doesn't take much storage compared to storing the entire event # anyway). - self.simple_insert_many_txn( + self.db.simple_insert_many_txn( txn, table="event_auth", values=[ @@ -580,12 +580,12 @@ class EventsStore( self, txn, new_forward_extremities, max_stream_order ): for room_id, new_extrem in iteritems(new_forward_extremities): - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="event_forward_extremities", keyvalues={"room_id": room_id} ) txn.call_after(self.get_latest_event_ids_in_room.invalidate, (room_id,)) - self.simple_insert_many_txn( + self.db.simple_insert_many_txn( txn, table="event_forward_extremities", values=[ @@ -598,7 +598,7 @@ class EventsStore( # new stream_ordering to new forward extremeties in the room. # This allows us to later efficiently look up the forward extremeties # for a room before a given stream_ordering - self.simple_insert_many_txn( + self.db.simple_insert_many_txn( txn, table="stream_ordering_to_exterm", values=[ @@ -722,7 +722,7 @@ class EventsStore( # change in outlier status to our workers. stream_order = event.internal_metadata.stream_ordering state_group_id = context.state_group - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="ex_outlier_stream", values={ @@ -794,7 +794,7 @@ class EventsStore( d.pop("redacted_because", None) return d - self.simple_insert_many_txn( + self.db.simple_insert_many_txn( txn, table="event_json", values=[ @@ -811,7 +811,7 @@ class EventsStore( ], ) - self.simple_insert_many_txn( + self.db.simple_insert_many_txn( txn, table="events", values=[ @@ -841,7 +841,7 @@ class EventsStore( # If we're persisting an unredacted event we go and ensure # that we mark any redactions that reference this event as # requiring censoring. - self.simple_update_txn( + self.db.simple_update_txn( txn, table="redactions", keyvalues={"redacts": event.event_id}, @@ -983,7 +983,7 @@ class EventsStore( state_values.append(vals) - self.simple_insert_many_txn(txn, table="state_events", values=state_values) + self.db.simple_insert_many_txn(txn, table="state_events", values=state_values) # Prefill the event cache self._add_to_cache(txn, events_and_contexts) @@ -1014,7 +1014,7 @@ class EventsStore( ) txn.execute(sql + clause, args) - rows = self.cursor_to_dict(txn) + rows = self.db.cursor_to_dict(txn) for row in rows: event = ev_map[row["event_id"]] if not row["rejects"] and not row["redacts"]: @@ -1032,7 +1032,7 @@ class EventsStore( # invalidate the cache for the redacted event txn.call_after(self._invalidate_get_event_cache, event.redacts) - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="redactions", values={ @@ -1077,7 +1077,9 @@ class EventsStore( LIMIT ? """ - rows = yield self.execute("_censor_redactions_fetch", None, sql, before_ts, 100) + rows = yield self.db.execute( + "_censor_redactions_fetch", None, sql, before_ts, 100 + ) updates = [] @@ -1109,14 +1111,14 @@ class EventsStore( if pruned_json: self._censor_event_txn(txn, event_id, pruned_json) - self.simple_update_one_txn( + self.db.simple_update_one_txn( txn, table="redactions", keyvalues={"event_id": redaction_id}, updatevalues={"have_censored": True}, ) - yield self.runInteraction("_update_censor_txn", _update_censor_txn) + yield self.db.runInteraction("_update_censor_txn", _update_censor_txn) def _censor_event_txn(self, txn, event_id, pruned_json): """Censor an event by replacing its JSON in the event_json table with the @@ -1127,7 +1129,7 @@ class EventsStore( event_id (str): The ID of the event to censor. pruned_json (str): The pruned JSON """ - self.simple_update_one_txn( + self.db.simple_update_one_txn( txn, table="event_json", keyvalues={"event_id": event_id}, @@ -1153,7 +1155,7 @@ class EventsStore( (count,) = txn.fetchone() return count - ret = yield self.runInteraction("count_messages", _count_messages) + ret = yield self.db.runInteraction("count_messages", _count_messages) return ret @defer.inlineCallbacks @@ -1174,7 +1176,7 @@ class EventsStore( (count,) = txn.fetchone() return count - ret = yield self.runInteraction("count_daily_sent_messages", _count_messages) + ret = yield self.db.runInteraction("count_daily_sent_messages", _count_messages) return ret @defer.inlineCallbacks @@ -1189,7 +1191,7 @@ class EventsStore( (count,) = txn.fetchone() return count - ret = yield self.runInteraction("count_daily_active_rooms", _count) + ret = yield self.db.runInteraction("count_daily_active_rooms", _count) return ret def get_current_backfill_token(self): @@ -1241,7 +1243,7 @@ class EventsStore( return new_event_updates - return self.runInteraction( + return self.db.runInteraction( "get_all_new_forward_event_rows", get_all_new_forward_event_rows ) @@ -1286,7 +1288,7 @@ class EventsStore( return new_event_updates - return self.runInteraction( + return self.db.runInteraction( "get_all_new_backfill_event_rows", get_all_new_backfill_event_rows ) @@ -1379,7 +1381,7 @@ class EventsStore( backward_ex_outliers, ) - return self.runInteraction("get_all_new_events", get_all_new_events_txn) + return self.db.runInteraction("get_all_new_events", get_all_new_events_txn) def purge_history(self, room_id, token, delete_local_events): """Deletes room history before a certain point @@ -1399,7 +1401,7 @@ class EventsStore( deleted events. """ - return self.runInteraction( + return self.db.runInteraction( "purge_history", self._purge_history_txn, room_id, @@ -1647,7 +1649,7 @@ class EventsStore( Deferred[List[int]]: The list of state groups to delete. """ - return self.runInteraction("purge_room", self._purge_room_txn, room_id) + return self.db.runInteraction("purge_room", self._purge_room_txn, room_id) def _purge_room_txn(self, txn, room_id): # First we fetch all the state groups that should be deleted, before @@ -1766,7 +1768,7 @@ class EventsStore( to delete. """ - return self.runInteraction( + return self.db.runInteraction( "purge_unreferenced_state_groups", self._purge_unreferenced_state_groups, room_id, @@ -1778,7 +1780,7 @@ class EventsStore( "[purge] found %i state groups to delete", len(state_groups_to_delete) ) - rows = self.simple_select_many_txn( + rows = self.db.simple_select_many_txn( txn, table="state_group_edges", column="prev_state_group", @@ -1805,15 +1807,15 @@ class EventsStore( curr_state = self._get_state_groups_from_groups_txn(txn, [sg]) curr_state = curr_state[sg] - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="state_groups_state", keyvalues={"state_group": sg} ) - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="state_group_edges", keyvalues={"state_group": sg} ) - self.simple_insert_many_txn( + self.db.simple_insert_many_txn( txn, table="state_groups_state", values=[ @@ -1850,7 +1852,7 @@ class EventsStore( state group. """ - rows = yield self.simple_select_many_batch( + rows = yield self.db.simple_select_many_batch( table="state_group_edges", column="prev_state_group", iterable=state_groups, @@ -1869,7 +1871,7 @@ class EventsStore( state_groups_to_delete (list[int]): State groups to delete """ - return self.runInteraction( + return self.db.runInteraction( "purge_room_state", self._purge_room_state_txn, room_id, @@ -1880,7 +1882,7 @@ class EventsStore( # first we have to delete the state groups states logger.info("[purge] removing %s from state_groups_state", room_id) - self.simple_delete_many_txn( + self.db.simple_delete_many_txn( txn, table="state_groups_state", column="state_group", @@ -1891,7 +1893,7 @@ class EventsStore( # ... and the state group edges logger.info("[purge] removing %s from state_group_edges", room_id) - self.simple_delete_many_txn( + self.db.simple_delete_many_txn( txn, table="state_group_edges", column="state_group", @@ -1902,7 +1904,7 @@ class EventsStore( # ... and the state groups logger.info("[purge] removing %s from state_groups", room_id) - self.simple_delete_many_txn( + self.db.simple_delete_many_txn( txn, table="state_groups", column="id", @@ -1919,7 +1921,7 @@ class EventsStore( @cachedInlineCallbacks(max_entries=5000) def _get_event_ordering(self, event_id): - res = yield self.simple_select_one( + res = yield self.db.simple_select_one( table="events", retcols=["topological_ordering", "stream_ordering"], keyvalues={"event_id": event_id}, @@ -1942,7 +1944,7 @@ class EventsStore( txn.execute(sql, (from_token, to_token, limit)) return txn.fetchall() - return self.runInteraction( + return self.db.runInteraction( "get_all_updated_current_state_deltas", get_all_updated_current_state_deltas_txn, ) @@ -1960,7 +1962,7 @@ class EventsStore( room_id (str): The ID of the room the event was sent to. topological_ordering (int): The position of the event in the room's topology. """ - return self.simple_insert_many_txn( + return self.db.simple_insert_many_txn( txn=txn, table="event_labels", values=[ @@ -1982,7 +1984,7 @@ class EventsStore( event_id (str): The event ID the expiry timestamp is associated with. expiry_ts (int): The timestamp at which to expire (delete) the event. """ - return self.simple_insert_txn( + return self.db.simple_insert_txn( txn=txn, table="event_expiry", values={"event_id": event_id, "expiry_ts": expiry_ts}, @@ -2031,7 +2033,7 @@ class EventsStore( txn, "_get_event_cache", (event.event_id,) ) - yield self.runInteraction("delete_expired_event", delete_expired_event_txn) + yield self.db.runInteraction("delete_expired_event", delete_expired_event_txn) def _delete_event_expiry_txn(self, txn, event_id): """Delete the expiry timestamp associated with an event ID without deleting the @@ -2041,7 +2043,7 @@ class EventsStore( txn (LoggingTransaction): The transaction to use to perform the deletion. event_id (str): The event ID to delete the associated expiry timestamp of. """ - return self.simple_delete_txn( + return self.db.simple_delete_txn( txn=txn, table="event_expiry", keyvalues={"event_id": event_id} ) @@ -2065,7 +2067,7 @@ class EventsStore( return txn.fetchone() - return self.runInteraction( + return self.db.runInteraction( desc="get_next_event_to_expire", func=get_next_event_to_expire_txn ) diff --git a/synapse/storage/data_stores/main/events_bg_updates.py b/synapse/storage/data_stores/main/events_bg_updates.py index 37dfc8c871..365e966956 100644 --- a/synapse/storage/data_stores/main/events_bg_updates.py +++ b/synapse/storage/data_stores/main/events_bg_updates.py @@ -151,7 +151,7 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): return len(rows) - result = yield self.runInteraction( + result = yield self.db.runInteraction( self.EVENT_FIELDS_SENDER_URL_UPDATE_NAME, reindex_txn ) @@ -189,7 +189,7 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): chunks = [event_ids[i : i + 100] for i in range(0, len(event_ids), 100)] for chunk in chunks: - ev_rows = self.simple_select_many_txn( + ev_rows = self.db.simple_select_many_txn( txn, table="event_json", column="event_id", @@ -228,7 +228,7 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): return len(rows_to_update) - result = yield self.runInteraction( + result = yield self.db.runInteraction( self.EVENT_ORIGIN_SERVER_TS_NAME, reindex_search_txn ) @@ -366,7 +366,7 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): to_delete.intersection_update(original_set) - deleted = self.simple_delete_many_txn( + deleted = self.db.simple_delete_many_txn( txn=txn, table="event_forward_extremities", column="event_id", @@ -382,7 +382,7 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): if deleted: # We now need to invalidate the caches of these rooms - rows = self.simple_select_many_txn( + rows = self.db.simple_select_many_txn( txn, table="events", column="event_id", @@ -396,7 +396,7 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): self.get_latest_event_ids_in_room.invalidate, (room_id,) ) - self.simple_delete_many_txn( + self.db.simple_delete_many_txn( txn=txn, table="_extremities_to_check", column="event_id", @@ -406,7 +406,7 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): return len(original_set) - num_handled = yield self.runInteraction( + num_handled = yield self.db.runInteraction( "_cleanup_extremities_bg_update", _cleanup_extremities_bg_update_txn ) @@ -416,7 +416,7 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): def _drop_table_txn(txn): txn.execute("DROP TABLE _extremities_to_check") - yield self.runInteraction( + yield self.db.runInteraction( "_cleanup_extremities_bg_update_drop_table", _drop_table_txn ) @@ -470,7 +470,7 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): return len(rows) - count = yield self.runInteraction( + count = yield self.db.runInteraction( "_redactions_received_ts", _redactions_received_ts_txn ) @@ -501,7 +501,7 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): txn.execute("DROP INDEX redactions_censored_redacts") - yield self.runInteraction( + yield self.db.runInteraction( "_event_fix_redactions_bytes", _event_fix_redactions_bytes_txn ) @@ -533,7 +533,7 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): try: event_json = json.loads(event_json_raw) - self.simple_insert_many_txn( + self.db.simple_insert_many_txn( txn=txn, table="event_labels", values=[ @@ -565,7 +565,7 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): return nbrows - num_rows = yield self.runInteraction( + num_rows = yield self.db.runInteraction( desc="event_store_labels", func=_event_store_labels_txn ) diff --git a/synapse/storage/data_stores/main/events_worker.py b/synapse/storage/data_stores/main/events_worker.py index 6a08a746b6..e041fc5eac 100644 --- a/synapse/storage/data_stores/main/events_worker.py +++ b/synapse/storage/data_stores/main/events_worker.py @@ -78,7 +78,7 @@ class EventsWorkerStore(SQLBaseStore): Deferred[int|None]: Timestamp in milliseconds, or None for events that were persisted before received_ts was implemented. """ - return self.simple_select_one_onecol( + return self.db.simple_select_one_onecol( table="events", keyvalues={"event_id": event_id}, retcol="received_ts", @@ -117,7 +117,7 @@ class EventsWorkerStore(SQLBaseStore): return ts - return self.runInteraction( + return self.db.runInteraction( "get_approximate_received_ts", _get_approximate_received_ts_txn ) @@ -452,7 +452,7 @@ class EventsWorkerStore(SQLBaseStore): event_id for events, _ in event_list for event_id in events ) - row_dict = self.new_transaction( + row_dict = self.db.new_transaction( conn, "do_fetch", [], [], self._fetch_event_rows, events_to_fetch ) @@ -584,7 +584,7 @@ class EventsWorkerStore(SQLBaseStore): if should_start: run_as_background_process( - "fetch_events", self.runWithConnection, self._do_fetch + "fetch_events", self.db.runWithConnection, self._do_fetch ) logger.debug("Loading %d events: %s", len(events), events) @@ -745,7 +745,7 @@ class EventsWorkerStore(SQLBaseStore): """Given a list of event ids, check if we have already processed and stored them as non outliers. """ - rows = yield self.simple_select_many_batch( + rows = yield self.db.simple_select_many_batch( table="events", retcols=("event_id",), column="event_id", @@ -780,7 +780,9 @@ class EventsWorkerStore(SQLBaseStore): # break the input up into chunks of 100 input_iterator = iter(event_ids) for chunk in iter(lambda: list(itertools.islice(input_iterator, 100)), []): - yield self.runInteraction("have_seen_events", have_seen_events_txn, chunk) + yield self.db.runInteraction( + "have_seen_events", have_seen_events_txn, chunk + ) return results def _get_total_state_event_counts_txn(self, txn, room_id): @@ -807,7 +809,7 @@ class EventsWorkerStore(SQLBaseStore): Returns: Deferred[int] """ - return self.runInteraction( + return self.db.runInteraction( "get_total_state_event_counts", self._get_total_state_event_counts_txn, room_id, @@ -832,7 +834,7 @@ class EventsWorkerStore(SQLBaseStore): Returns: Deferred[int] """ - return self.runInteraction( + return self.db.runInteraction( "get_current_state_event_counts", self._get_current_state_event_counts_txn, room_id, diff --git a/synapse/storage/data_stores/main/filtering.py b/synapse/storage/data_stores/main/filtering.py index 17ef7b9354..342d6622a4 100644 --- a/synapse/storage/data_stores/main/filtering.py +++ b/synapse/storage/data_stores/main/filtering.py @@ -30,7 +30,7 @@ class FilteringStore(SQLBaseStore): except ValueError: raise SynapseError(400, "Invalid filter ID", Codes.INVALID_PARAM) - def_json = yield self.simple_select_one_onecol( + def_json = yield self.db.simple_select_one_onecol( table="user_filters", keyvalues={"user_id": user_localpart, "filter_id": filter_id}, retcol="filter_json", @@ -71,4 +71,4 @@ class FilteringStore(SQLBaseStore): return filter_id - return self.runInteraction("add_user_filter", _do_txn) + return self.db.runInteraction("add_user_filter", _do_txn) diff --git a/synapse/storage/data_stores/main/group_server.py b/synapse/storage/data_stores/main/group_server.py index 9e1d12bcb7..7f5e8dce66 100644 --- a/synapse/storage/data_stores/main/group_server.py +++ b/synapse/storage/data_stores/main/group_server.py @@ -35,7 +35,7 @@ class GroupServerStore(SQLBaseStore): * "invite" * "open" """ - return self.simple_update_one( + return self.db.simple_update_one( table="groups", keyvalues={"group_id": group_id}, updatevalues={"join_policy": join_policy}, @@ -43,7 +43,7 @@ class GroupServerStore(SQLBaseStore): ) def get_group(self, group_id): - return self.simple_select_one( + return self.db.simple_select_one( table="groups", keyvalues={"group_id": group_id}, retcols=( @@ -65,7 +65,7 @@ class GroupServerStore(SQLBaseStore): if not include_private: keyvalues["is_public"] = True - return self.simple_select_list( + return self.db.simple_select_list( table="group_users", keyvalues=keyvalues, retcols=("user_id", "is_public", "is_admin"), @@ -75,7 +75,7 @@ class GroupServerStore(SQLBaseStore): def get_invited_users_in_group(self, group_id): # TODO: Pagination - return self.simple_select_onecol( + return self.db.simple_select_onecol( table="group_invites", keyvalues={"group_id": group_id}, retcol="user_id", @@ -89,7 +89,7 @@ class GroupServerStore(SQLBaseStore): if not include_private: keyvalues["is_public"] = True - return self.simple_select_list( + return self.db.simple_select_list( table="group_rooms", keyvalues=keyvalues, retcols=("room_id", "is_public"), @@ -153,10 +153,12 @@ class GroupServerStore(SQLBaseStore): return rooms, categories - return self.runInteraction("get_rooms_for_summary", _get_rooms_for_summary_txn) + return self.db.runInteraction( + "get_rooms_for_summary", _get_rooms_for_summary_txn + ) def add_room_to_summary(self, group_id, room_id, category_id, order, is_public): - return self.runInteraction( + return self.db.runInteraction( "add_room_to_summary", self._add_room_to_summary_txn, group_id, @@ -180,7 +182,7 @@ class GroupServerStore(SQLBaseStore): an order of 1 will put the room first. Otherwise, the room gets added to the end. """ - room_in_group = self.simple_select_one_onecol_txn( + room_in_group = self.db.simple_select_one_onecol_txn( txn, table="group_rooms", keyvalues={"group_id": group_id, "room_id": room_id}, @@ -193,7 +195,7 @@ class GroupServerStore(SQLBaseStore): if category_id is None: category_id = _DEFAULT_CATEGORY_ID else: - cat_exists = self.simple_select_one_onecol_txn( + cat_exists = self.db.simple_select_one_onecol_txn( txn, table="group_room_categories", keyvalues={"group_id": group_id, "category_id": category_id}, @@ -204,7 +206,7 @@ class GroupServerStore(SQLBaseStore): raise SynapseError(400, "Category doesn't exist") # TODO: Check category is part of summary already - cat_exists = self.simple_select_one_onecol_txn( + cat_exists = self.db.simple_select_one_onecol_txn( txn, table="group_summary_room_categories", keyvalues={"group_id": group_id, "category_id": category_id}, @@ -224,7 +226,7 @@ class GroupServerStore(SQLBaseStore): (group_id, category_id, group_id, category_id), ) - existing = self.simple_select_one_txn( + existing = self.db.simple_select_one_txn( txn, table="group_summary_rooms", keyvalues={ @@ -257,7 +259,7 @@ class GroupServerStore(SQLBaseStore): to_update["room_order"] = order if is_public is not None: to_update["is_public"] = is_public - self.simple_update_txn( + self.db.simple_update_txn( txn, table="group_summary_rooms", keyvalues={ @@ -271,7 +273,7 @@ class GroupServerStore(SQLBaseStore): if is_public is None: is_public = True - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="group_summary_rooms", values={ @@ -287,7 +289,7 @@ class GroupServerStore(SQLBaseStore): if category_id is None: category_id = _DEFAULT_CATEGORY_ID - return self.simple_delete( + return self.db.simple_delete( table="group_summary_rooms", keyvalues={ "group_id": group_id, @@ -299,7 +301,7 @@ class GroupServerStore(SQLBaseStore): @defer.inlineCallbacks def get_group_categories(self, group_id): - rows = yield self.simple_select_list( + rows = yield self.db.simple_select_list( table="group_room_categories", keyvalues={"group_id": group_id}, retcols=("category_id", "is_public", "profile"), @@ -316,7 +318,7 @@ class GroupServerStore(SQLBaseStore): @defer.inlineCallbacks def get_group_category(self, group_id, category_id): - category = yield self.simple_select_one( + category = yield self.db.simple_select_one( table="group_room_categories", keyvalues={"group_id": group_id, "category_id": category_id}, retcols=("is_public", "profile"), @@ -343,7 +345,7 @@ class GroupServerStore(SQLBaseStore): else: update_values["is_public"] = is_public - return self.simple_upsert( + return self.db.simple_upsert( table="group_room_categories", keyvalues={"group_id": group_id, "category_id": category_id}, values=update_values, @@ -352,7 +354,7 @@ class GroupServerStore(SQLBaseStore): ) def remove_group_category(self, group_id, category_id): - return self.simple_delete( + return self.db.simple_delete( table="group_room_categories", keyvalues={"group_id": group_id, "category_id": category_id}, desc="remove_group_category", @@ -360,7 +362,7 @@ class GroupServerStore(SQLBaseStore): @defer.inlineCallbacks def get_group_roles(self, group_id): - rows = yield self.simple_select_list( + rows = yield self.db.simple_select_list( table="group_roles", keyvalues={"group_id": group_id}, retcols=("role_id", "is_public", "profile"), @@ -377,7 +379,7 @@ class GroupServerStore(SQLBaseStore): @defer.inlineCallbacks def get_group_role(self, group_id, role_id): - role = yield self.simple_select_one( + role = yield self.db.simple_select_one( table="group_roles", keyvalues={"group_id": group_id, "role_id": role_id}, retcols=("is_public", "profile"), @@ -404,7 +406,7 @@ class GroupServerStore(SQLBaseStore): else: update_values["is_public"] = is_public - return self.simple_upsert( + return self.db.simple_upsert( table="group_roles", keyvalues={"group_id": group_id, "role_id": role_id}, values=update_values, @@ -413,14 +415,14 @@ class GroupServerStore(SQLBaseStore): ) def remove_group_role(self, group_id, role_id): - return self.simple_delete( + return self.db.simple_delete( table="group_roles", keyvalues={"group_id": group_id, "role_id": role_id}, desc="remove_group_role", ) def add_user_to_summary(self, group_id, user_id, role_id, order, is_public): - return self.runInteraction( + return self.db.runInteraction( "add_user_to_summary", self._add_user_to_summary_txn, group_id, @@ -444,7 +446,7 @@ class GroupServerStore(SQLBaseStore): an order of 1 will put the user first. Otherwise, the user gets added to the end. """ - user_in_group = self.simple_select_one_onecol_txn( + user_in_group = self.db.simple_select_one_onecol_txn( txn, table="group_users", keyvalues={"group_id": group_id, "user_id": user_id}, @@ -457,7 +459,7 @@ class GroupServerStore(SQLBaseStore): if role_id is None: role_id = _DEFAULT_ROLE_ID else: - role_exists = self.simple_select_one_onecol_txn( + role_exists = self.db.simple_select_one_onecol_txn( txn, table="group_roles", keyvalues={"group_id": group_id, "role_id": role_id}, @@ -468,7 +470,7 @@ class GroupServerStore(SQLBaseStore): raise SynapseError(400, "Role doesn't exist") # TODO: Check role is part of the summary already - role_exists = self.simple_select_one_onecol_txn( + role_exists = self.db.simple_select_one_onecol_txn( txn, table="group_summary_roles", keyvalues={"group_id": group_id, "role_id": role_id}, @@ -488,7 +490,7 @@ class GroupServerStore(SQLBaseStore): (group_id, role_id, group_id, role_id), ) - existing = self.simple_select_one_txn( + existing = self.db.simple_select_one_txn( txn, table="group_summary_users", keyvalues={"group_id": group_id, "user_id": user_id, "role_id": role_id}, @@ -517,7 +519,7 @@ class GroupServerStore(SQLBaseStore): to_update["user_order"] = order if is_public is not None: to_update["is_public"] = is_public - self.simple_update_txn( + self.db.simple_update_txn( txn, table="group_summary_users", keyvalues={ @@ -531,7 +533,7 @@ class GroupServerStore(SQLBaseStore): if is_public is None: is_public = True - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="group_summary_users", values={ @@ -547,7 +549,7 @@ class GroupServerStore(SQLBaseStore): if role_id is None: role_id = _DEFAULT_ROLE_ID - return self.simple_delete( + return self.db.simple_delete( table="group_summary_users", keyvalues={"group_id": group_id, "role_id": role_id, "user_id": user_id}, desc="remove_user_from_summary", @@ -561,7 +563,7 @@ class GroupServerStore(SQLBaseStore): Deferred[list[str]]: A twisted.Deferred containing a list of group ids containing this room """ - return self.simple_select_onecol( + return self.db.simple_select_onecol( table="group_rooms", keyvalues={"room_id": room_id}, retcol="group_id", @@ -625,12 +627,12 @@ class GroupServerStore(SQLBaseStore): return users, roles - return self.runInteraction( + return self.db.runInteraction( "get_users_for_summary_by_role", _get_users_for_summary_txn ) def is_user_in_group(self, user_id, group_id): - return self.simple_select_one_onecol( + return self.db.simple_select_one_onecol( table="group_users", keyvalues={"group_id": group_id, "user_id": user_id}, retcol="user_id", @@ -639,7 +641,7 @@ class GroupServerStore(SQLBaseStore): ).addCallback(lambda r: bool(r)) def is_user_admin_in_group(self, group_id, user_id): - return self.simple_select_one_onecol( + return self.db.simple_select_one_onecol( table="group_users", keyvalues={"group_id": group_id, "user_id": user_id}, retcol="is_admin", @@ -650,7 +652,7 @@ class GroupServerStore(SQLBaseStore): def add_group_invite(self, group_id, user_id): """Record that the group server has invited a user """ - return self.simple_insert( + return self.db.simple_insert( table="group_invites", values={"group_id": group_id, "user_id": user_id}, desc="add_group_invite", @@ -659,7 +661,7 @@ class GroupServerStore(SQLBaseStore): def is_user_invited_to_local_group(self, group_id, user_id): """Has the group server invited a user? """ - return self.simple_select_one_onecol( + return self.db.simple_select_one_onecol( table="group_invites", keyvalues={"group_id": group_id, "user_id": user_id}, retcol="user_id", @@ -682,7 +684,7 @@ class GroupServerStore(SQLBaseStore): """ def _get_users_membership_in_group_txn(txn): - row = self.simple_select_one_txn( + row = self.db.simple_select_one_txn( txn, table="group_users", keyvalues={"group_id": group_id, "user_id": user_id}, @@ -697,7 +699,7 @@ class GroupServerStore(SQLBaseStore): "is_privileged": row["is_admin"], } - row = self.simple_select_one_onecol_txn( + row = self.db.simple_select_one_onecol_txn( txn, table="group_invites", keyvalues={"group_id": group_id, "user_id": user_id}, @@ -710,7 +712,7 @@ class GroupServerStore(SQLBaseStore): return {} - return self.runInteraction( + return self.db.runInteraction( "get_users_membership_info_in_group", _get_users_membership_in_group_txn ) @@ -738,7 +740,7 @@ class GroupServerStore(SQLBaseStore): """ def _add_user_to_group_txn(txn): - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="group_users", values={ @@ -749,14 +751,14 @@ class GroupServerStore(SQLBaseStore): }, ) - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="group_invites", keyvalues={"group_id": group_id, "user_id": user_id}, ) if local_attestation: - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="group_attestations_renewals", values={ @@ -766,7 +768,7 @@ class GroupServerStore(SQLBaseStore): }, ) if remote_attestation: - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="group_attestations_remote", values={ @@ -777,49 +779,49 @@ class GroupServerStore(SQLBaseStore): }, ) - return self.runInteraction("add_user_to_group", _add_user_to_group_txn) + return self.db.runInteraction("add_user_to_group", _add_user_to_group_txn) def remove_user_from_group(self, group_id, user_id): def _remove_user_from_group_txn(txn): - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="group_users", keyvalues={"group_id": group_id, "user_id": user_id}, ) - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="group_invites", keyvalues={"group_id": group_id, "user_id": user_id}, ) - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="group_attestations_renewals", keyvalues={"group_id": group_id, "user_id": user_id}, ) - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="group_attestations_remote", keyvalues={"group_id": group_id, "user_id": user_id}, ) - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="group_summary_users", keyvalues={"group_id": group_id, "user_id": user_id}, ) - return self.runInteraction( + return self.db.runInteraction( "remove_user_from_group", _remove_user_from_group_txn ) def add_room_to_group(self, group_id, room_id, is_public): - return self.simple_insert( + return self.db.simple_insert( table="group_rooms", values={"group_id": group_id, "room_id": room_id, "is_public": is_public}, desc="add_room_to_group", ) def update_room_in_group_visibility(self, group_id, room_id, is_public): - return self.simple_update( + return self.db.simple_update( table="group_rooms", keyvalues={"group_id": group_id, "room_id": room_id}, updatevalues={"is_public": is_public}, @@ -828,26 +830,26 @@ class GroupServerStore(SQLBaseStore): def remove_room_from_group(self, group_id, room_id): def _remove_room_from_group_txn(txn): - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="group_rooms", keyvalues={"group_id": group_id, "room_id": room_id}, ) - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="group_summary_rooms", keyvalues={"group_id": group_id, "room_id": room_id}, ) - return self.runInteraction( + return self.db.runInteraction( "remove_room_from_group", _remove_room_from_group_txn ) def get_publicised_groups_for_user(self, user_id): """Get all groups a user is publicising """ - return self.simple_select_onecol( + return self.db.simple_select_onecol( table="local_group_membership", keyvalues={"user_id": user_id, "membership": "join", "is_publicised": True}, retcol="group_id", @@ -857,7 +859,7 @@ class GroupServerStore(SQLBaseStore): def update_group_publicity(self, group_id, user_id, publicise): """Update whether the user is publicising their membership of the group """ - return self.simple_update_one( + return self.db.simple_update_one( table="local_group_membership", keyvalues={"group_id": group_id, "user_id": user_id}, updatevalues={"is_publicised": publicise}, @@ -893,12 +895,12 @@ class GroupServerStore(SQLBaseStore): def _register_user_group_membership_txn(txn, next_id): # TODO: Upsert? - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="local_group_membership", keyvalues={"group_id": group_id, "user_id": user_id}, ) - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="local_group_membership", values={ @@ -911,7 +913,7 @@ class GroupServerStore(SQLBaseStore): }, ) - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="local_group_updates", values={ @@ -930,7 +932,7 @@ class GroupServerStore(SQLBaseStore): if membership == "join": if local_attestation: - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="group_attestations_renewals", values={ @@ -940,7 +942,7 @@ class GroupServerStore(SQLBaseStore): }, ) if remote_attestation: - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="group_attestations_remote", values={ @@ -951,12 +953,12 @@ class GroupServerStore(SQLBaseStore): }, ) else: - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="group_attestations_renewals", keyvalues={"group_id": group_id, "user_id": user_id}, ) - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="group_attestations_remote", keyvalues={"group_id": group_id, "user_id": user_id}, @@ -965,7 +967,7 @@ class GroupServerStore(SQLBaseStore): return next_id with self._group_updates_id_gen.get_next() as next_id: - res = yield self.runInteraction( + res = yield self.db.runInteraction( "register_user_group_membership", _register_user_group_membership_txn, next_id, @@ -976,7 +978,7 @@ class GroupServerStore(SQLBaseStore): def create_group( self, group_id, user_id, name, avatar_url, short_description, long_description ): - yield self.simple_insert( + yield self.db.simple_insert( table="groups", values={ "group_id": group_id, @@ -991,7 +993,7 @@ class GroupServerStore(SQLBaseStore): @defer.inlineCallbacks def update_group_profile(self, group_id, profile): - yield self.simple_update_one( + yield self.db.simple_update_one( table="groups", keyvalues={"group_id": group_id}, updatevalues=profile, @@ -1008,16 +1010,16 @@ class GroupServerStore(SQLBaseStore): WHERE valid_until_ms <= ? """ txn.execute(sql, (valid_until_ms,)) - return self.cursor_to_dict(txn) + return self.db.cursor_to_dict(txn) - return self.runInteraction( + return self.db.runInteraction( "get_attestations_need_renewals", _get_attestations_need_renewals_txn ) def update_attestation_renewal(self, group_id, user_id, attestation): """Update an attestation that we have renewed """ - return self.simple_update_one( + return self.db.simple_update_one( table="group_attestations_renewals", keyvalues={"group_id": group_id, "user_id": user_id}, updatevalues={"valid_until_ms": attestation["valid_until_ms"]}, @@ -1027,7 +1029,7 @@ class GroupServerStore(SQLBaseStore): def update_remote_attestion(self, group_id, user_id, attestation): """Update an attestation that a remote has renewed """ - return self.simple_update_one( + return self.db.simple_update_one( table="group_attestations_remote", keyvalues={"group_id": group_id, "user_id": user_id}, updatevalues={ @@ -1046,7 +1048,7 @@ class GroupServerStore(SQLBaseStore): group_id (str) user_id (str) """ - return self.simple_delete( + return self.db.simple_delete( table="group_attestations_renewals", keyvalues={"group_id": group_id, "user_id": user_id}, desc="remove_attestation_renewal", @@ -1057,7 +1059,7 @@ class GroupServerStore(SQLBaseStore): """Get the attestation that proves the remote agrees that the user is in the group. """ - row = yield self.simple_select_one( + row = yield self.db.simple_select_one( table="group_attestations_remote", keyvalues={"group_id": group_id, "user_id": user_id}, retcols=("valid_until_ms", "attestation_json"), @@ -1072,7 +1074,7 @@ class GroupServerStore(SQLBaseStore): return None def get_joined_groups(self, user_id): - return self.simple_select_onecol( + return self.db.simple_select_onecol( table="local_group_membership", keyvalues={"user_id": user_id, "membership": "join"}, retcol="group_id", @@ -1099,7 +1101,7 @@ class GroupServerStore(SQLBaseStore): for row in txn ] - return self.runInteraction( + return self.db.runInteraction( "get_all_groups_for_user", _get_all_groups_for_user_txn ) @@ -1129,7 +1131,7 @@ class GroupServerStore(SQLBaseStore): for group_id, membership, gtype, content_json in txn ] - return self.runInteraction( + return self.db.runInteraction( "get_groups_changes_for_user", _get_groups_changes_for_user_txn ) @@ -1154,7 +1156,7 @@ class GroupServerStore(SQLBaseStore): for stream_id, group_id, user_id, gtype, content_json in txn ] - return self.runInteraction( + return self.db.runInteraction( "get_all_groups_changes", _get_all_groups_changes_txn ) @@ -1188,8 +1190,8 @@ class GroupServerStore(SQLBaseStore): ] for table in tables: - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table=table, keyvalues={"group_id": group_id} ) - return self.runInteraction("delete_group", _delete_group_txn) + return self.db.runInteraction("delete_group", _delete_group_txn) diff --git a/synapse/storage/data_stores/main/keys.py b/synapse/storage/data_stores/main/keys.py index c7150432b3..6b12f5a75f 100644 --- a/synapse/storage/data_stores/main/keys.py +++ b/synapse/storage/data_stores/main/keys.py @@ -92,7 +92,7 @@ class KeyStore(SQLBaseStore): _get_keys(txn, batch) return keys - return self.runInteraction("get_server_verify_keys", _txn) + return self.db.runInteraction("get_server_verify_keys", _txn) def store_server_verify_keys(self, from_server, ts_added_ms, verify_keys): """Stores NACL verification keys for remote servers. @@ -127,9 +127,9 @@ class KeyStore(SQLBaseStore): f((i,)) return res - return self.runInteraction( + return self.db.runInteraction( "store_server_verify_keys", - self.simple_upsert_many_txn, + self.db.simple_upsert_many_txn, table="server_signature_keys", key_names=("server_name", "key_id"), key_values=key_values, @@ -157,7 +157,7 @@ class KeyStore(SQLBaseStore): ts_valid_until_ms (int): The time when this json stops being valid. key_json (bytes): The encoded JSON. """ - return self.simple_upsert( + return self.db.simple_upsert( table="server_keys_json", keyvalues={ "server_name": server_name, @@ -196,7 +196,7 @@ class KeyStore(SQLBaseStore): keyvalues["key_id"] = key_id if from_server is not None: keyvalues["from_server"] = from_server - rows = self.simple_select_list_txn( + rows = self.db.simple_select_list_txn( txn, "server_keys_json", keyvalues=keyvalues, @@ -211,4 +211,4 @@ class KeyStore(SQLBaseStore): results[(server_name, key_id, from_server)] = rows return results - return self.runInteraction("get_server_keys_json", _get_server_keys_json_txn) + return self.db.runInteraction("get_server_keys_json", _get_server_keys_json_txn) diff --git a/synapse/storage/data_stores/main/media_repository.py b/synapse/storage/data_stores/main/media_repository.py index 0cb9446f96..ea02497784 100644 --- a/synapse/storage/data_stores/main/media_repository.py +++ b/synapse/storage/data_stores/main/media_repository.py @@ -39,7 +39,7 @@ class MediaRepositoryStore(MediaRepositoryBackgroundUpdateStore): Returns: None if the media_id doesn't exist. """ - return self.simple_select_one( + return self.db.simple_select_one( "local_media_repository", {"media_id": media_id}, ( @@ -64,7 +64,7 @@ class MediaRepositoryStore(MediaRepositoryBackgroundUpdateStore): user_id, url_cache=None, ): - return self.simple_insert( + return self.db.simple_insert( "local_media_repository", { "media_id": media_id, @@ -124,12 +124,12 @@ class MediaRepositoryStore(MediaRepositoryBackgroundUpdateStore): ) ) - return self.runInteraction("get_url_cache", get_url_cache_txn) + return self.db.runInteraction("get_url_cache", get_url_cache_txn) def store_url_cache( self, url, response_code, etag, expires_ts, og, media_id, download_ts ): - return self.simple_insert( + return self.db.simple_insert( "local_media_repository_url_cache", { "url": url, @@ -144,7 +144,7 @@ class MediaRepositoryStore(MediaRepositoryBackgroundUpdateStore): ) def get_local_media_thumbnails(self, media_id): - return self.simple_select_list( + return self.db.simple_select_list( "local_media_repository_thumbnails", {"media_id": media_id}, ( @@ -166,7 +166,7 @@ class MediaRepositoryStore(MediaRepositoryBackgroundUpdateStore): thumbnail_method, thumbnail_length, ): - return self.simple_insert( + return self.db.simple_insert( "local_media_repository_thumbnails", { "media_id": media_id, @@ -180,7 +180,7 @@ class MediaRepositoryStore(MediaRepositoryBackgroundUpdateStore): ) def get_cached_remote_media(self, origin, media_id): - return self.simple_select_one( + return self.db.simple_select_one( "remote_media_cache", {"media_origin": origin, "media_id": media_id}, ( @@ -205,7 +205,7 @@ class MediaRepositoryStore(MediaRepositoryBackgroundUpdateStore): upload_name, filesystem_id, ): - return self.simple_insert( + return self.db.simple_insert( "remote_media_cache", { "media_origin": origin, @@ -250,10 +250,12 @@ class MediaRepositoryStore(MediaRepositoryBackgroundUpdateStore): txn.executemany(sql, ((time_ms, media_id) for media_id in local_media)) - return self.runInteraction("update_cached_last_access_time", update_cache_txn) + return self.db.runInteraction( + "update_cached_last_access_time", update_cache_txn + ) def get_remote_media_thumbnails(self, origin, media_id): - return self.simple_select_list( + return self.db.simple_select_list( "remote_media_cache_thumbnails", {"media_origin": origin, "media_id": media_id}, ( @@ -278,7 +280,7 @@ class MediaRepositoryStore(MediaRepositoryBackgroundUpdateStore): thumbnail_method, thumbnail_length, ): - return self.simple_insert( + return self.db.simple_insert( "remote_media_cache_thumbnails", { "media_origin": origin, @@ -300,24 +302,24 @@ class MediaRepositoryStore(MediaRepositoryBackgroundUpdateStore): " WHERE last_access_ts < ?" ) - return self.execute( - "get_remote_media_before", self.cursor_to_dict, sql, before_ts + return self.db.execute( + "get_remote_media_before", self.db.cursor_to_dict, sql, before_ts ) def delete_remote_media(self, media_origin, media_id): def delete_remote_media_txn(txn): - self.simple_delete_txn( + self.db.simple_delete_txn( txn, "remote_media_cache", keyvalues={"media_origin": media_origin, "media_id": media_id}, ) - self.simple_delete_txn( + self.db.simple_delete_txn( txn, "remote_media_cache_thumbnails", keyvalues={"media_origin": media_origin, "media_id": media_id}, ) - return self.runInteraction("delete_remote_media", delete_remote_media_txn) + return self.db.runInteraction("delete_remote_media", delete_remote_media_txn) def get_expired_url_cache(self, now_ts): sql = ( @@ -331,7 +333,9 @@ class MediaRepositoryStore(MediaRepositoryBackgroundUpdateStore): txn.execute(sql, (now_ts,)) return [row[0] for row in txn] - return self.runInteraction("get_expired_url_cache", _get_expired_url_cache_txn) + return self.db.runInteraction( + "get_expired_url_cache", _get_expired_url_cache_txn + ) def delete_url_cache(self, media_ids): if len(media_ids) == 0: @@ -342,7 +346,7 @@ class MediaRepositoryStore(MediaRepositoryBackgroundUpdateStore): def _delete_url_cache_txn(txn): txn.executemany(sql, [(media_id,) for media_id in media_ids]) - return self.runInteraction("delete_url_cache", _delete_url_cache_txn) + return self.db.runInteraction("delete_url_cache", _delete_url_cache_txn) def get_url_cache_media_before(self, before_ts): sql = ( @@ -356,7 +360,7 @@ class MediaRepositoryStore(MediaRepositoryBackgroundUpdateStore): txn.execute(sql, (before_ts,)) return [row[0] for row in txn] - return self.runInteraction( + return self.db.runInteraction( "get_url_cache_media_before", _get_url_cache_media_before_txn ) @@ -373,6 +377,6 @@ class MediaRepositoryStore(MediaRepositoryBackgroundUpdateStore): txn.executemany(sql, [(media_id,) for media_id in media_ids]) - return self.runInteraction( + return self.db.runInteraction( "delete_url_cache_media", _delete_url_cache_media_txn ) diff --git a/synapse/storage/data_stores/main/monthly_active_users.py b/synapse/storage/data_stores/main/monthly_active_users.py index b8fc28f97b..34bf3a1880 100644 --- a/synapse/storage/data_stores/main/monthly_active_users.py +++ b/synapse/storage/data_stores/main/monthly_active_users.py @@ -32,7 +32,7 @@ class MonthlyActiveUsersStore(SQLBaseStore): self._clock = hs.get_clock() self.hs = hs # Do not add more reserved users than the total allowable number - self.new_transaction( + self.db.new_transaction( dbconn, "initialise_mau_threepids", [], @@ -146,7 +146,7 @@ class MonthlyActiveUsersStore(SQLBaseStore): txn.execute(sql, query_args) reserved_users = yield self.get_registered_reserved_users() - yield self.runInteraction( + yield self.db.runInteraction( "reap_monthly_active_users", _reap_users, reserved_users ) # It seems poor to invalidate the whole cache, Postgres supports @@ -174,7 +174,7 @@ class MonthlyActiveUsersStore(SQLBaseStore): (count,) = txn.fetchone() return count - return self.runInteraction("count_users", _count_users) + return self.db.runInteraction("count_users", _count_users) @defer.inlineCallbacks def get_registered_reserved_users(self): @@ -217,7 +217,7 @@ class MonthlyActiveUsersStore(SQLBaseStore): if is_support: return - yield self.runInteraction( + yield self.db.runInteraction( "upsert_monthly_active_user", self.upsert_monthly_active_user_txn, user_id ) @@ -261,7 +261,7 @@ class MonthlyActiveUsersStore(SQLBaseStore): # never be a big table and alternative approaches (batching multiple # upserts into a single txn) introduced a lot of extra complexity. # See https://github.com/matrix-org/synapse/issues/3854 for more - is_insert = self.simple_upsert_txn( + is_insert = self.db.simple_upsert_txn( txn, table="monthly_active_users", keyvalues={"user_id": user_id}, @@ -281,7 +281,7 @@ class MonthlyActiveUsersStore(SQLBaseStore): """ - return self.simple_select_one_onecol( + return self.db.simple_select_one_onecol( table="monthly_active_users", keyvalues={"user_id": user_id}, retcol="timestamp", diff --git a/synapse/storage/data_stores/main/openid.py b/synapse/storage/data_stores/main/openid.py index 650e49750e..cc21437e92 100644 --- a/synapse/storage/data_stores/main/openid.py +++ b/synapse/storage/data_stores/main/openid.py @@ -3,7 +3,7 @@ from synapse.storage._base import SQLBaseStore class OpenIdStore(SQLBaseStore): def insert_open_id_token(self, token, ts_valid_until_ms, user_id): - return self.simple_insert( + return self.db.simple_insert( table="open_id_tokens", values={ "token": token, @@ -28,4 +28,6 @@ class OpenIdStore(SQLBaseStore): else: return rows[0][0] - return self.runInteraction("get_user_id_for_token", get_user_id_for_token_txn) + return self.db.runInteraction( + "get_user_id_for_token", get_user_id_for_token_txn + ) diff --git a/synapse/storage/data_stores/main/presence.py b/synapse/storage/data_stores/main/presence.py index a5e121efd1..a2c83e0867 100644 --- a/synapse/storage/data_stores/main/presence.py +++ b/synapse/storage/data_stores/main/presence.py @@ -29,7 +29,7 @@ class PresenceStore(SQLBaseStore): ) with stream_ordering_manager as stream_orderings: - yield self.runInteraction( + yield self.db.runInteraction( "update_presence", self._update_presence_txn, stream_orderings, @@ -46,7 +46,7 @@ class PresenceStore(SQLBaseStore): txn.call_after(self._get_presence_for_user.invalidate, (state.user_id,)) # Actually insert new rows - self.simple_insert_many_txn( + self.db.simple_insert_many_txn( txn, table="presence_stream", values=[ @@ -88,7 +88,7 @@ class PresenceStore(SQLBaseStore): txn.execute(sql, (last_id, current_id)) return txn.fetchall() - return self.runInteraction( + return self.db.runInteraction( "get_all_presence_updates", get_all_presence_updates_txn ) @@ -103,7 +103,7 @@ class PresenceStore(SQLBaseStore): inlineCallbacks=True, ) def get_presence_for_users(self, user_ids): - rows = yield self.simple_select_many_batch( + rows = yield self.db.simple_select_many_batch( table="presence_stream", column="user_id", iterable=user_ids, @@ -129,7 +129,7 @@ class PresenceStore(SQLBaseStore): return self._presence_id_gen.get_current_token() def allow_presence_visible(self, observed_localpart, observer_userid): - return self.simple_insert( + return self.db.simple_insert( table="presence_allow_inbound", values={ "observed_user_id": observed_localpart, @@ -140,7 +140,7 @@ class PresenceStore(SQLBaseStore): ) def disallow_presence_visible(self, observed_localpart, observer_userid): - return self.simple_delete_one( + return self.db.simple_delete_one( table="presence_allow_inbound", keyvalues={ "observed_user_id": observed_localpart, diff --git a/synapse/storage/data_stores/main/profile.py b/synapse/storage/data_stores/main/profile.py index c8b5b60301..2b52cf9c1a 100644 --- a/synapse/storage/data_stores/main/profile.py +++ b/synapse/storage/data_stores/main/profile.py @@ -24,7 +24,7 @@ class ProfileWorkerStore(SQLBaseStore): @defer.inlineCallbacks def get_profileinfo(self, user_localpart): try: - profile = yield self.simple_select_one( + profile = yield self.db.simple_select_one( table="profiles", keyvalues={"user_id": user_localpart}, retcols=("displayname", "avatar_url"), @@ -42,7 +42,7 @@ class ProfileWorkerStore(SQLBaseStore): ) def get_profile_displayname(self, user_localpart): - return self.simple_select_one_onecol( + return self.db.simple_select_one_onecol( table="profiles", keyvalues={"user_id": user_localpart}, retcol="displayname", @@ -50,7 +50,7 @@ class ProfileWorkerStore(SQLBaseStore): ) def get_profile_avatar_url(self, user_localpart): - return self.simple_select_one_onecol( + return self.db.simple_select_one_onecol( table="profiles", keyvalues={"user_id": user_localpart}, retcol="avatar_url", @@ -58,7 +58,7 @@ class ProfileWorkerStore(SQLBaseStore): ) def get_from_remote_profile_cache(self, user_id): - return self.simple_select_one( + return self.db.simple_select_one( table="remote_profile_cache", keyvalues={"user_id": user_id}, retcols=("displayname", "avatar_url"), @@ -67,12 +67,12 @@ class ProfileWorkerStore(SQLBaseStore): ) def create_profile(self, user_localpart): - return self.simple_insert( + return self.db.simple_insert( table="profiles", values={"user_id": user_localpart}, desc="create_profile" ) def set_profile_displayname(self, user_localpart, new_displayname): - return self.simple_update_one( + return self.db.simple_update_one( table="profiles", keyvalues={"user_id": user_localpart}, updatevalues={"displayname": new_displayname}, @@ -80,7 +80,7 @@ class ProfileWorkerStore(SQLBaseStore): ) def set_profile_avatar_url(self, user_localpart, new_avatar_url): - return self.simple_update_one( + return self.db.simple_update_one( table="profiles", keyvalues={"user_id": user_localpart}, updatevalues={"avatar_url": new_avatar_url}, @@ -95,7 +95,7 @@ class ProfileStore(ProfileWorkerStore): This should only be called when `is_subscribed_remote_profile_for_user` would return true for the user. """ - return self.simple_upsert( + return self.db.simple_upsert( table="remote_profile_cache", keyvalues={"user_id": user_id}, values={ @@ -107,7 +107,7 @@ class ProfileStore(ProfileWorkerStore): ) def update_remote_profile_cache(self, user_id, displayname, avatar_url): - return self.simple_update( + return self.db.simple_update( table="remote_profile_cache", keyvalues={"user_id": user_id}, values={ @@ -125,7 +125,7 @@ class ProfileStore(ProfileWorkerStore): """ subscribed = yield self.is_subscribed_remote_profile_for_user(user_id) if not subscribed: - yield self.simple_delete( + yield self.db.simple_delete( table="remote_profile_cache", keyvalues={"user_id": user_id}, desc="delete_remote_profile_cache", @@ -144,9 +144,9 @@ class ProfileStore(ProfileWorkerStore): txn.execute(sql, (last_checked,)) - return self.cursor_to_dict(txn) + return self.db.cursor_to_dict(txn) - return self.runInteraction( + return self.db.runInteraction( "get_remote_profile_cache_entries_that_expire", _get_remote_profile_cache_entries_that_expire_txn, ) @@ -155,7 +155,7 @@ class ProfileStore(ProfileWorkerStore): def is_subscribed_remote_profile_for_user(self, user_id): """Check whether we are interested in a remote user's profile. """ - res = yield self.simple_select_one_onecol( + res = yield self.db.simple_select_one_onecol( table="group_users", keyvalues={"user_id": user_id}, retcol="user_id", @@ -166,7 +166,7 @@ class ProfileStore(ProfileWorkerStore): if res: return True - res = yield self.simple_select_one_onecol( + res = yield self.db.simple_select_one_onecol( table="group_invites", keyvalues={"user_id": user_id}, retcol="user_id", diff --git a/synapse/storage/data_stores/main/push_rule.py b/synapse/storage/data_stores/main/push_rule.py index 75bd499bcd..de682cc63a 100644 --- a/synapse/storage/data_stores/main/push_rule.py +++ b/synapse/storage/data_stores/main/push_rule.py @@ -75,7 +75,7 @@ class PushRulesWorkerStore( def __init__(self, db_conn, hs): super(PushRulesWorkerStore, self).__init__(db_conn, hs) - push_rules_prefill, push_rules_id = self.get_cache_dict( + push_rules_prefill, push_rules_id = self.db.get_cache_dict( db_conn, "push_rules_stream", entity_column="user_id", @@ -100,7 +100,7 @@ class PushRulesWorkerStore( @cachedInlineCallbacks(max_entries=5000) def get_push_rules_for_user(self, user_id): - rows = yield self.simple_select_list( + rows = yield self.db.simple_select_list( table="push_rules", keyvalues={"user_name": user_id}, retcols=( @@ -124,7 +124,7 @@ class PushRulesWorkerStore( @cachedInlineCallbacks(max_entries=5000) def get_push_rules_enabled_for_user(self, user_id): - results = yield self.simple_select_list( + results = yield self.db.simple_select_list( table="push_rules_enable", keyvalues={"user_name": user_id}, retcols=("user_name", "rule_id", "enabled"), @@ -146,7 +146,7 @@ class PushRulesWorkerStore( (count,) = txn.fetchone() return bool(count) - return self.runInteraction( + return self.db.runInteraction( "have_push_rules_changed", have_push_rules_changed_txn ) @@ -162,7 +162,7 @@ class PushRulesWorkerStore( results = {user_id: [] for user_id in user_ids} - rows = yield self.simple_select_many_batch( + rows = yield self.db.simple_select_many_batch( table="push_rules", column="user_name", iterable=user_ids, @@ -320,7 +320,7 @@ class PushRulesWorkerStore( results = {user_id: {} for user_id in user_ids} - rows = yield self.simple_select_many_batch( + rows = yield self.db.simple_select_many_batch( table="push_rules_enable", column="user_name", iterable=user_ids, @@ -350,7 +350,7 @@ class PushRuleStore(PushRulesWorkerStore): with self._push_rules_stream_id_gen.get_next() as ids: stream_id, event_stream_ordering = ids if before or after: - yield self.runInteraction( + yield self.db.runInteraction( "_add_push_rule_relative_txn", self._add_push_rule_relative_txn, stream_id, @@ -364,7 +364,7 @@ class PushRuleStore(PushRulesWorkerStore): after, ) else: - yield self.runInteraction( + yield self.db.runInteraction( "_add_push_rule_highest_priority_txn", self._add_push_rule_highest_priority_txn, stream_id, @@ -395,7 +395,7 @@ class PushRuleStore(PushRulesWorkerStore): relative_to_rule = before or after - res = self.simple_select_one_txn( + res = self.db.simple_select_one_txn( txn, table="push_rules", keyvalues={"user_name": user_id, "rule_id": relative_to_rule}, @@ -518,7 +518,7 @@ class PushRuleStore(PushRulesWorkerStore): # We didn't update a row with the given rule_id so insert one push_rule_id = self._push_rule_id_gen.get_next() - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="push_rules", values={ @@ -561,7 +561,7 @@ class PushRuleStore(PushRulesWorkerStore): """ def delete_push_rule_txn(txn, stream_id, event_stream_ordering): - self.simple_delete_one_txn( + self.db.simple_delete_one_txn( txn, "push_rules", {"user_name": user_id, "rule_id": rule_id} ) @@ -571,7 +571,7 @@ class PushRuleStore(PushRulesWorkerStore): with self._push_rules_stream_id_gen.get_next() as ids: stream_id, event_stream_ordering = ids - yield self.runInteraction( + yield self.db.runInteraction( "delete_push_rule", delete_push_rule_txn, stream_id, @@ -582,7 +582,7 @@ class PushRuleStore(PushRulesWorkerStore): def set_push_rule_enabled(self, user_id, rule_id, enabled): with self._push_rules_stream_id_gen.get_next() as ids: stream_id, event_stream_ordering = ids - yield self.runInteraction( + yield self.db.runInteraction( "_set_push_rule_enabled_txn", self._set_push_rule_enabled_txn, stream_id, @@ -596,7 +596,7 @@ class PushRuleStore(PushRulesWorkerStore): self, txn, stream_id, event_stream_ordering, user_id, rule_id, enabled ): new_id = self._push_rules_enable_id_gen.get_next() - self.simple_upsert_txn( + self.db.simple_upsert_txn( txn, "push_rules_enable", {"user_name": user_id, "rule_id": rule_id}, @@ -636,7 +636,7 @@ class PushRuleStore(PushRulesWorkerStore): update_stream=False, ) else: - self.simple_update_one_txn( + self.db.simple_update_one_txn( txn, "push_rules", {"user_name": user_id, "rule_id": rule_id}, @@ -655,7 +655,7 @@ class PushRuleStore(PushRulesWorkerStore): with self._push_rules_stream_id_gen.get_next() as ids: stream_id, event_stream_ordering = ids - yield self.runInteraction( + yield self.db.runInteraction( "set_push_rule_actions", set_push_rule_actions_txn, stream_id, @@ -675,7 +675,7 @@ class PushRuleStore(PushRulesWorkerStore): if data is not None: values.update(data) - self.simple_insert_txn(txn, "push_rules_stream", values=values) + self.db.simple_insert_txn(txn, "push_rules_stream", values=values) txn.call_after(self.get_push_rules_for_user.invalidate, (user_id,)) txn.call_after(self.get_push_rules_enabled_for_user.invalidate, (user_id,)) @@ -699,7 +699,7 @@ class PushRuleStore(PushRulesWorkerStore): txn.execute(sql, (last_id, current_id, limit)) return txn.fetchall() - return self.runInteraction( + return self.db.runInteraction( "get_all_push_rule_updates", get_all_push_rule_updates_txn ) diff --git a/synapse/storage/data_stores/main/pusher.py b/synapse/storage/data_stores/main/pusher.py index d5a169872b..f07309ef09 100644 --- a/synapse/storage/data_stores/main/pusher.py +++ b/synapse/storage/data_stores/main/pusher.py @@ -59,7 +59,7 @@ class PusherWorkerStore(SQLBaseStore): @defer.inlineCallbacks def user_has_pusher(self, user_id): - ret = yield self.simple_select_one_onecol( + ret = yield self.db.simple_select_one_onecol( "pushers", {"user_name": user_id}, "id", allow_none=True ) return ret is not None @@ -72,7 +72,7 @@ class PusherWorkerStore(SQLBaseStore): @defer.inlineCallbacks def get_pushers_by(self, keyvalues): - ret = yield self.simple_select_list( + ret = yield self.db.simple_select_list( "pushers", keyvalues, [ @@ -100,11 +100,11 @@ class PusherWorkerStore(SQLBaseStore): def get_all_pushers(self): def get_pushers(txn): txn.execute("SELECT * FROM pushers") - rows = self.cursor_to_dict(txn) + rows = self.db.cursor_to_dict(txn) return self._decode_pushers_rows(rows) - rows = yield self.runInteraction("get_all_pushers", get_pushers) + rows = yield self.db.runInteraction("get_all_pushers", get_pushers) return rows def get_all_updated_pushers(self, last_id, current_id, limit): @@ -134,7 +134,7 @@ class PusherWorkerStore(SQLBaseStore): return updated, deleted - return self.runInteraction( + return self.db.runInteraction( "get_all_updated_pushers", get_all_updated_pushers_txn ) @@ -177,7 +177,7 @@ class PusherWorkerStore(SQLBaseStore): return results - return self.runInteraction( + return self.db.runInteraction( "get_all_updated_pushers_rows", get_all_updated_pushers_rows_txn ) @@ -193,7 +193,7 @@ class PusherWorkerStore(SQLBaseStore): inlineCallbacks=True, ) def get_if_users_have_pushers(self, user_ids): - rows = yield self.simple_select_many_batch( + rows = yield self.db.simple_select_many_batch( table="pushers", column="user_name", iterable=user_ids, @@ -230,7 +230,7 @@ class PusherStore(PusherWorkerStore): with self._pushers_id_gen.get_next() as stream_id: # no need to lock because `pushers` has a unique key on # (app_id, pushkey, user_name) so simple_upsert will retry - yield self.simple_upsert( + yield self.db.simple_upsert( table="pushers", keyvalues={"app_id": app_id, "pushkey": pushkey, "user_name": user_id}, values={ @@ -255,7 +255,7 @@ class PusherStore(PusherWorkerStore): if user_has_pusher is not True: # invalidate, since we the user might not have had a pusher before - yield self.runInteraction( + yield self.db.runInteraction( "add_pusher", self._invalidate_cache_and_stream, self.get_if_user_has_pusher, @@ -269,7 +269,7 @@ class PusherStore(PusherWorkerStore): txn, self.get_if_user_has_pusher, (user_id,) ) - self.simple_delete_one_txn( + self.db.simple_delete_one_txn( txn, "pushers", {"app_id": app_id, "pushkey": pushkey, "user_name": user_id}, @@ -278,7 +278,7 @@ class PusherStore(PusherWorkerStore): # it's possible for us to end up with duplicate rows for # (app_id, pushkey, user_id) at different stream_ids, but that # doesn't really matter. - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="deleted_pushers", values={ @@ -290,13 +290,13 @@ class PusherStore(PusherWorkerStore): ) with self._pushers_id_gen.get_next() as stream_id: - yield self.runInteraction("delete_pusher", delete_pusher_txn, stream_id) + yield self.db.runInteraction("delete_pusher", delete_pusher_txn, stream_id) @defer.inlineCallbacks def update_pusher_last_stream_ordering( self, app_id, pushkey, user_id, last_stream_ordering ): - yield self.simple_update_one( + yield self.db.simple_update_one( "pushers", {"app_id": app_id, "pushkey": pushkey, "user_name": user_id}, {"last_stream_ordering": last_stream_ordering}, @@ -319,7 +319,7 @@ class PusherStore(PusherWorkerStore): Returns: Deferred[bool]: True if the pusher still exists; False if it has been deleted. """ - updated = yield self.simple_update( + updated = yield self.db.simple_update( table="pushers", keyvalues={"app_id": app_id, "pushkey": pushkey, "user_name": user_id}, updatevalues={ @@ -333,7 +333,7 @@ class PusherStore(PusherWorkerStore): @defer.inlineCallbacks def update_pusher_failing_since(self, app_id, pushkey, user_id, failing_since): - yield self.simple_update( + yield self.db.simple_update( table="pushers", keyvalues={"app_id": app_id, "pushkey": pushkey, "user_name": user_id}, updatevalues={"failing_since": failing_since}, @@ -342,7 +342,7 @@ class PusherStore(PusherWorkerStore): @defer.inlineCallbacks def get_throttle_params_by_room(self, pusher_id): - res = yield self.simple_select_list( + res = yield self.db.simple_select_list( "pusher_throttle", {"pusher": pusher_id}, ["room_id", "last_sent_ts", "throttle_ms"], @@ -362,7 +362,7 @@ class PusherStore(PusherWorkerStore): def set_throttle_params(self, pusher_id, room_id, params): # no need to lock because `pusher_throttle` has a primary key on # (pusher, room_id) so simple_upsert will retry - yield self.simple_upsert( + yield self.db.simple_upsert( "pusher_throttle", {"pusher": pusher_id, "room_id": room_id}, params, diff --git a/synapse/storage/data_stores/main/receipts.py b/synapse/storage/data_stores/main/receipts.py index 380f388e30..ac2d45bd5c 100644 --- a/synapse/storage/data_stores/main/receipts.py +++ b/synapse/storage/data_stores/main/receipts.py @@ -61,7 +61,7 @@ class ReceiptsWorkerStore(SQLBaseStore): @cached(num_args=2) def get_receipts_for_room(self, room_id, receipt_type): - return self.simple_select_list( + return self.db.simple_select_list( table="receipts_linearized", keyvalues={"room_id": room_id, "receipt_type": receipt_type}, retcols=("user_id", "event_id"), @@ -70,7 +70,7 @@ class ReceiptsWorkerStore(SQLBaseStore): @cached(num_args=3) def get_last_receipt_event_id_for_user(self, user_id, room_id, receipt_type): - return self.simple_select_one_onecol( + return self.db.simple_select_one_onecol( table="receipts_linearized", keyvalues={ "room_id": room_id, @@ -84,7 +84,7 @@ class ReceiptsWorkerStore(SQLBaseStore): @cachedInlineCallbacks(num_args=2) def get_receipts_for_user(self, user_id, receipt_type): - rows = yield self.simple_select_list( + rows = yield self.db.simple_select_list( table="receipts_linearized", keyvalues={"user_id": user_id, "receipt_type": receipt_type}, retcols=("room_id", "event_id"), @@ -108,7 +108,7 @@ class ReceiptsWorkerStore(SQLBaseStore): txn.execute(sql, (user_id,)) return txn.fetchall() - rows = yield self.runInteraction("get_receipts_for_user_with_orderings", f) + rows = yield self.db.runInteraction("get_receipts_for_user_with_orderings", f) return { row[0]: { "event_id": row[1], @@ -187,11 +187,11 @@ class ReceiptsWorkerStore(SQLBaseStore): txn.execute(sql, (room_id, to_key)) - rows = self.cursor_to_dict(txn) + rows = self.db.cursor_to_dict(txn) return rows - rows = yield self.runInteraction("get_linearized_receipts_for_room", f) + rows = yield self.db.runInteraction("get_linearized_receipts_for_room", f) if not rows: return [] @@ -237,9 +237,11 @@ class ReceiptsWorkerStore(SQLBaseStore): txn.execute(sql + clause, [to_key] + list(args)) - return self.cursor_to_dict(txn) + return self.db.cursor_to_dict(txn) - txn_results = yield self.runInteraction("_get_linearized_receipts_for_rooms", f) + txn_results = yield self.db.runInteraction( + "_get_linearized_receipts_for_rooms", f + ) results = {} for row in txn_results: @@ -282,7 +284,7 @@ class ReceiptsWorkerStore(SQLBaseStore): return list(r[0:5] + (json.loads(r[5]),) for r in txn) - return self.runInteraction( + return self.db.runInteraction( "get_all_updated_receipts", get_all_updated_receipts_txn ) @@ -335,7 +337,7 @@ class ReceiptsStore(ReceiptsWorkerStore): otherwise, the rx timestamp of the event that the RR corresponds to (or 0 if the event is unknown) """ - res = self.simple_select_one_txn( + res = self.db.simple_select_one_txn( txn, table="events", retcols=["stream_ordering", "received_ts"], @@ -388,7 +390,7 @@ class ReceiptsStore(ReceiptsWorkerStore): (user_id, room_id, receipt_type), ) - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="receipts_linearized", keyvalues={ @@ -398,7 +400,7 @@ class ReceiptsStore(ReceiptsWorkerStore): }, ) - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="receipts_linearized", values={ @@ -453,13 +455,13 @@ class ReceiptsStore(ReceiptsWorkerStore): else: raise RuntimeError("Unrecognized event_ids: %r" % (event_ids,)) - linearized_event_id = yield self.runInteraction( + linearized_event_id = yield self.db.runInteraction( "insert_receipt_conv", graph_to_linear ) stream_id_manager = self._receipts_id_gen.get_next() with stream_id_manager as stream_id: - event_ts = yield self.runInteraction( + event_ts = yield self.db.runInteraction( "insert_linearized_receipt", self.insert_linearized_receipt_txn, room_id, @@ -488,7 +490,7 @@ class ReceiptsStore(ReceiptsWorkerStore): return stream_id, max_persisted_id def insert_graph_receipt(self, room_id, receipt_type, user_id, event_ids, data): - return self.runInteraction( + return self.db.runInteraction( "insert_graph_receipt", self.insert_graph_receipt_txn, room_id, @@ -514,7 +516,7 @@ class ReceiptsStore(ReceiptsWorkerStore): self._get_linearized_receipts_for_room.invalidate_many, (room_id,) ) - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="receipts_graph", keyvalues={ @@ -523,7 +525,7 @@ class ReceiptsStore(ReceiptsWorkerStore): "user_id": user_id, }, ) - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="receipts_graph", values={ diff --git a/synapse/storage/data_stores/main/registration.py b/synapse/storage/data_stores/main/registration.py index debc6706f5..8f9aa87ceb 100644 --- a/synapse/storage/data_stores/main/registration.py +++ b/synapse/storage/data_stores/main/registration.py @@ -45,7 +45,7 @@ class RegistrationWorkerStore(SQLBaseStore): @cached() def get_user_by_id(self, user_id): - return self.simple_select_one( + return self.db.simple_select_one( table="users", keyvalues={"name": user_id}, retcols=[ @@ -94,7 +94,7 @@ class RegistrationWorkerStore(SQLBaseStore): including the keys `name`, `is_guest`, `device_id`, `token_id`, `valid_until_ms`. """ - return self.runInteraction( + return self.db.runInteraction( "get_user_by_access_token", self._query_for_auth, token ) @@ -109,7 +109,7 @@ class RegistrationWorkerStore(SQLBaseStore): otherwise int representation of the timestamp (as a number of milliseconds since epoch). """ - res = yield self.simple_select_one_onecol( + res = yield self.db.simple_select_one_onecol( table="account_validity", keyvalues={"user_id": user_id}, retcol="expiration_ts_ms", @@ -137,7 +137,7 @@ class RegistrationWorkerStore(SQLBaseStore): """ def set_account_validity_for_user_txn(txn): - self.simple_update_txn( + self.db.simple_update_txn( txn=txn, table="account_validity", keyvalues={"user_id": user_id}, @@ -151,7 +151,7 @@ class RegistrationWorkerStore(SQLBaseStore): txn, self.get_expiration_ts_for_user, (user_id,) ) - yield self.runInteraction( + yield self.db.runInteraction( "set_account_validity_for_user", set_account_validity_for_user_txn ) @@ -167,7 +167,7 @@ class RegistrationWorkerStore(SQLBaseStore): Raises: StoreError: The provided token is already set for another user. """ - yield self.simple_update_one( + yield self.db.simple_update_one( table="account_validity", keyvalues={"user_id": user_id}, updatevalues={"renewal_token": renewal_token}, @@ -184,7 +184,7 @@ class RegistrationWorkerStore(SQLBaseStore): Returns: defer.Deferred[str]: The ID of the user to which the token belongs. """ - res = yield self.simple_select_one_onecol( + res = yield self.db.simple_select_one_onecol( table="account_validity", keyvalues={"renewal_token": renewal_token}, retcol="user_id", @@ -203,7 +203,7 @@ class RegistrationWorkerStore(SQLBaseStore): Returns: defer.Deferred[str]: The renewal token associated with this user ID. """ - res = yield self.simple_select_one_onecol( + res = yield self.db.simple_select_one_onecol( table="account_validity", keyvalues={"user_id": user_id}, retcol="renewal_token", @@ -229,9 +229,9 @@ class RegistrationWorkerStore(SQLBaseStore): ) values = [False, now_ms, renew_at] txn.execute(sql, values) - return self.cursor_to_dict(txn) + return self.db.cursor_to_dict(txn) - res = yield self.runInteraction( + res = yield self.db.runInteraction( "get_users_expiring_soon", select_users_txn, self.clock.time_msec(), @@ -250,7 +250,7 @@ class RegistrationWorkerStore(SQLBaseStore): email_sent (bool): Flag which indicates whether a renewal email has been sent to this user. """ - yield self.simple_update_one( + yield self.db.simple_update_one( table="account_validity", keyvalues={"user_id": user_id}, updatevalues={"email_sent": email_sent}, @@ -265,7 +265,7 @@ class RegistrationWorkerStore(SQLBaseStore): Args: user_id (str): ID of the user to remove from the account validity table. """ - yield self.simple_delete_one( + yield self.db.simple_delete_one( table="account_validity", keyvalues={"user_id": user_id}, desc="delete_account_validity_for_user", @@ -281,7 +281,7 @@ class RegistrationWorkerStore(SQLBaseStore): Returns (bool): true iff the user is a server admin, false otherwise. """ - res = yield self.simple_select_one_onecol( + res = yield self.db.simple_select_one_onecol( table="users", keyvalues={"name": user.to_string()}, retcol="admin", @@ -299,7 +299,7 @@ class RegistrationWorkerStore(SQLBaseStore): admin (bool): true iff the user is to be a server admin, false otherwise. """ - return self.simple_update_one( + return self.db.simple_update_one( table="users", keyvalues={"name": user.to_string()}, updatevalues={"admin": 1 if admin else 0}, @@ -316,7 +316,7 @@ class RegistrationWorkerStore(SQLBaseStore): ) txn.execute(sql, (token,)) - rows = self.cursor_to_dict(txn) + rows = self.db.cursor_to_dict(txn) if rows: return rows[0] @@ -332,7 +332,9 @@ class RegistrationWorkerStore(SQLBaseStore): Returns: Deferred[bool]: True if user 'user_type' is null or empty string """ - res = yield self.runInteraction("is_real_user", self.is_real_user_txn, user_id) + res = yield self.db.runInteraction( + "is_real_user", self.is_real_user_txn, user_id + ) return res @cachedInlineCallbacks() @@ -345,13 +347,13 @@ class RegistrationWorkerStore(SQLBaseStore): Returns: Deferred[bool]: True if user is of type UserTypes.SUPPORT """ - res = yield self.runInteraction( + res = yield self.db.runInteraction( "is_support_user", self.is_support_user_txn, user_id ) return res def is_real_user_txn(self, txn, user_id): - res = self.simple_select_one_onecol_txn( + res = self.db.simple_select_one_onecol_txn( txn=txn, table="users", keyvalues={"name": user_id}, @@ -361,7 +363,7 @@ class RegistrationWorkerStore(SQLBaseStore): return res is None def is_support_user_txn(self, txn, user_id): - res = self.simple_select_one_onecol_txn( + res = self.db.simple_select_one_onecol_txn( txn=txn, table="users", keyvalues={"name": user_id}, @@ -380,7 +382,7 @@ class RegistrationWorkerStore(SQLBaseStore): txn.execute(sql, (user_id,)) return dict(txn) - return self.runInteraction("get_users_by_id_case_insensitive", f) + return self.db.runInteraction("get_users_by_id_case_insensitive", f) async def get_user_by_external_id( self, auth_provider: str, external_id: str @@ -394,7 +396,7 @@ class RegistrationWorkerStore(SQLBaseStore): Returns: str|None: the mxid of the user, or None if they are not known """ - return await self.simple_select_one_onecol( + return await self.db.simple_select_one_onecol( table="user_external_ids", keyvalues={"auth_provider": auth_provider, "external_id": external_id}, retcol="user_id", @@ -408,12 +410,12 @@ class RegistrationWorkerStore(SQLBaseStore): def _count_users(txn): txn.execute("SELECT COUNT(*) AS users FROM users") - rows = self.cursor_to_dict(txn) + rows = self.db.cursor_to_dict(txn) if rows: return rows[0]["users"] return 0 - ret = yield self.runInteraction("count_users", _count_users) + ret = yield self.db.runInteraction("count_users", _count_users) return ret def count_daily_user_type(self): @@ -445,7 +447,7 @@ class RegistrationWorkerStore(SQLBaseStore): results[row[0]] = row[1] return results - return self.runInteraction("count_daily_user_type", _count_daily_user_type) + return self.db.runInteraction("count_daily_user_type", _count_daily_user_type) @defer.inlineCallbacks def count_nonbridged_users(self): @@ -459,7 +461,7 @@ class RegistrationWorkerStore(SQLBaseStore): (count,) = txn.fetchone() return count - ret = yield self.runInteraction("count_users", _count_users) + ret = yield self.db.runInteraction("count_users", _count_users) return ret @defer.inlineCallbacks @@ -468,12 +470,12 @@ class RegistrationWorkerStore(SQLBaseStore): def _count_users(txn): txn.execute("SELECT COUNT(*) AS users FROM users where user_type is null") - rows = self.cursor_to_dict(txn) + rows = self.db.cursor_to_dict(txn) if rows: return rows[0]["users"] return 0 - ret = yield self.runInteraction("count_real_users", _count_users) + ret = yield self.db.runInteraction("count_real_users", _count_users) return ret @defer.inlineCallbacks @@ -503,7 +505,7 @@ class RegistrationWorkerStore(SQLBaseStore): return ( ( - yield self.runInteraction( + yield self.db.runInteraction( "find_next_generated_user_id", _find_next_generated_user_id ) ) @@ -520,7 +522,7 @@ class RegistrationWorkerStore(SQLBaseStore): Returns: Deferred[str|None]: user id or None if no user id/threepid mapping exists """ - user_id = yield self.runInteraction( + user_id = yield self.db.runInteraction( "get_user_id_by_threepid", self.get_user_id_by_threepid_txn, medium, address ) return user_id @@ -536,7 +538,7 @@ class RegistrationWorkerStore(SQLBaseStore): Returns: str|None: user id or None if no user id/threepid mapping exists """ - ret = self.simple_select_one_txn( + ret = self.db.simple_select_one_txn( txn, "user_threepids", {"medium": medium, "address": address}, @@ -549,7 +551,7 @@ class RegistrationWorkerStore(SQLBaseStore): @defer.inlineCallbacks def user_add_threepid(self, user_id, medium, address, validated_at, added_at): - yield self.simple_upsert( + yield self.db.simple_upsert( "user_threepids", {"medium": medium, "address": address}, {"user_id": user_id, "validated_at": validated_at, "added_at": added_at}, @@ -557,7 +559,7 @@ class RegistrationWorkerStore(SQLBaseStore): @defer.inlineCallbacks def user_get_threepids(self, user_id): - ret = yield self.simple_select_list( + ret = yield self.db.simple_select_list( "user_threepids", {"user_id": user_id}, ["medium", "address", "validated_at", "added_at"], @@ -566,7 +568,7 @@ class RegistrationWorkerStore(SQLBaseStore): return ret def user_delete_threepid(self, user_id, medium, address): - return self.simple_delete( + return self.db.simple_delete( "user_threepids", keyvalues={"user_id": user_id, "medium": medium, "address": address}, desc="user_delete_threepid", @@ -579,7 +581,7 @@ class RegistrationWorkerStore(SQLBaseStore): user_id: The user id to delete all threepids of """ - return self.simple_delete( + return self.db.simple_delete( "user_threepids", keyvalues={"user_id": user_id}, desc="user_delete_threepids", @@ -601,7 +603,7 @@ class RegistrationWorkerStore(SQLBaseStore): """ # We need to use an upsert, in case they user had already bound the # threepid - return self.simple_upsert( + return self.db.simple_upsert( table="user_threepid_id_server", keyvalues={ "user_id": user_id, @@ -627,7 +629,7 @@ class RegistrationWorkerStore(SQLBaseStore): medium (str): The medium of the threepid (e.g "email") address (str): The address of the threepid (e.g "bob@example.com") """ - return self.simple_select_list( + return self.db.simple_select_list( table="user_threepid_id_server", keyvalues={"user_id": user_id}, retcols=["medium", "address"], @@ -648,7 +650,7 @@ class RegistrationWorkerStore(SQLBaseStore): Returns: Deferred """ - return self.simple_delete( + return self.db.simple_delete( table="user_threepid_id_server", keyvalues={ "user_id": user_id, @@ -671,7 +673,7 @@ class RegistrationWorkerStore(SQLBaseStore): Returns: Deferred[list[str]]: Resolves to a list of identity servers """ - return self.simple_select_onecol( + return self.db.simple_select_onecol( table="user_threepid_id_server", keyvalues={"user_id": user_id, "medium": medium, "address": address}, retcol="id_server", @@ -689,7 +691,7 @@ class RegistrationWorkerStore(SQLBaseStore): defer.Deferred(bool): The requested value. """ - res = yield self.simple_select_one_onecol( + res = yield self.db.simple_select_one_onecol( table="users", keyvalues={"name": user_id}, retcol="deactivated", @@ -756,13 +758,13 @@ class RegistrationWorkerStore(SQLBaseStore): sql += " LIMIT 1" txn.execute(sql, list(keyvalues.values())) - rows = self.cursor_to_dict(txn) + rows = self.db.cursor_to_dict(txn) if not rows: return None return rows[0] - return self.runInteraction( + return self.db.runInteraction( "get_threepid_validation_session", get_threepid_validation_session_txn ) @@ -776,18 +778,18 @@ class RegistrationWorkerStore(SQLBaseStore): """ def delete_threepid_session_txn(txn): - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="threepid_validation_token", keyvalues={"session_id": session_id}, ) - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="threepid_validation_session", keyvalues={"session_id": session_id}, ) - return self.runInteraction( + return self.db.runInteraction( "delete_threepid_session", delete_threepid_session_txn ) @@ -857,7 +859,7 @@ class RegistrationBackgroundUpdateStore( (last_user, batch_size), ) - rows = self.cursor_to_dict(txn) + rows = self.db.cursor_to_dict(txn) if not rows: return True, 0 @@ -880,7 +882,7 @@ class RegistrationBackgroundUpdateStore( else: return False, len(rows) - end, nb_processed = yield self.runInteraction( + end, nb_processed = yield self.db.runInteraction( "users_set_deactivated_flag", _background_update_set_deactivated_flag_txn ) @@ -911,7 +913,7 @@ class RegistrationBackgroundUpdateStore( txn.executemany(sql, [(id_server,) for id_server in id_servers]) if id_servers: - yield self.runInteraction( + yield self.db.runInteraction( "_bg_user_threepids_grandfather", _bg_user_threepids_grandfather_txn ) @@ -961,7 +963,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): """ next_id = self._access_tokens_id_gen.get_next() - yield self.simple_insert( + yield self.db.simple_insert( "access_tokens", { "id": next_id, @@ -1003,7 +1005,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): Raises: StoreError if the user_id could not be registered. """ - return self.runInteraction( + return self.db.runInteraction( "register_user", self._register_user, user_id, @@ -1037,7 +1039,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): # Ensure that the guest user actually exists # ``allow_none=False`` makes this raise an exception # if the row isn't in the database. - self.simple_select_one_txn( + self.db.simple_select_one_txn( txn, "users", keyvalues={"name": user_id, "is_guest": 1}, @@ -1045,7 +1047,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): allow_none=False, ) - self.simple_update_one_txn( + self.db.simple_update_one_txn( txn, "users", keyvalues={"name": user_id, "is_guest": 1}, @@ -1059,7 +1061,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): }, ) else: - self.simple_insert_txn( + self.db.simple_insert_txn( txn, "users", values={ @@ -1114,7 +1116,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): external_id: id on that system user_id: complete mxid that it is mapped to """ - return self.simple_insert( + return self.db.simple_insert( table="user_external_ids", values={ "auth_provider": auth_provider, @@ -1132,12 +1134,14 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): """ def user_set_password_hash_txn(txn): - self.simple_update_one_txn( + self.db.simple_update_one_txn( txn, "users", {"name": user_id}, {"password_hash": password_hash} ) self._invalidate_cache_and_stream(txn, self.get_user_by_id, (user_id,)) - return self.runInteraction("user_set_password_hash", user_set_password_hash_txn) + return self.db.runInteraction( + "user_set_password_hash", user_set_password_hash_txn + ) def user_set_consent_version(self, user_id, consent_version): """Updates the user table to record privacy policy consent @@ -1152,7 +1156,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): """ def f(txn): - self.simple_update_one_txn( + self.db.simple_update_one_txn( txn, table="users", keyvalues={"name": user_id}, @@ -1160,7 +1164,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): ) self._invalidate_cache_and_stream(txn, self.get_user_by_id, (user_id,)) - return self.runInteraction("user_set_consent_version", f) + return self.db.runInteraction("user_set_consent_version", f) def user_set_consent_server_notice_sent(self, user_id, consent_version): """Updates the user table to record that we have sent the user a server @@ -1176,7 +1180,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): """ def f(txn): - self.simple_update_one_txn( + self.db.simple_update_one_txn( txn, table="users", keyvalues={"name": user_id}, @@ -1184,7 +1188,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): ) self._invalidate_cache_and_stream(txn, self.get_user_by_id, (user_id,)) - return self.runInteraction("user_set_consent_server_notice_sent", f) + return self.db.runInteraction("user_set_consent_server_notice_sent", f) def user_delete_access_tokens(self, user_id, except_token_id=None, device_id=None): """ @@ -1230,11 +1234,11 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): return tokens_and_devices - return self.runInteraction("user_delete_access_tokens", f) + return self.db.runInteraction("user_delete_access_tokens", f) def delete_access_token(self, access_token): def f(txn): - self.simple_delete_one_txn( + self.db.simple_delete_one_txn( txn, table="access_tokens", keyvalues={"token": access_token} ) @@ -1242,11 +1246,11 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): txn, self.get_user_by_access_token, (access_token,) ) - return self.runInteraction("delete_access_token", f) + return self.db.runInteraction("delete_access_token", f) @cachedInlineCallbacks() def is_guest(self, user_id): - res = yield self.simple_select_one_onecol( + res = yield self.db.simple_select_one_onecol( table="users", keyvalues={"name": user_id}, retcol="is_guest", @@ -1261,7 +1265,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): Adds a user to the table of users who need to be parted from all the rooms they're in """ - return self.simple_insert( + return self.db.simple_insert( "users_pending_deactivation", values={"user_id": user_id}, desc="add_user_pending_deactivation", @@ -1274,7 +1278,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): """ # XXX: This should be simple_delete_one but we failed to put a unique index on # the table, so somehow duplicate entries have ended up in it. - return self.simple_delete( + return self.db.simple_delete( "users_pending_deactivation", keyvalues={"user_id": user_id}, desc="del_user_pending_deactivation", @@ -1285,7 +1289,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): Gets one user from the table of users waiting to be parted from all the rooms they're in. """ - return self.simple_select_one_onecol( + return self.db.simple_select_one_onecol( "users_pending_deactivation", keyvalues={}, retcol="user_id", @@ -1315,7 +1319,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): # Insert everything into a transaction in order to run atomically def validate_threepid_session_txn(txn): - row = self.simple_select_one_txn( + row = self.db.simple_select_one_txn( txn, table="threepid_validation_session", keyvalues={"session_id": session_id}, @@ -1333,7 +1337,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): 400, "This client_secret does not match the provided session_id" ) - row = self.simple_select_one_txn( + row = self.db.simple_select_one_txn( txn, table="threepid_validation_token", keyvalues={"session_id": session_id, "token": token}, @@ -1358,7 +1362,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): ) # Looks good. Validate the session - self.simple_update_txn( + self.db.simple_update_txn( txn, table="threepid_validation_session", keyvalues={"session_id": session_id}, @@ -1368,7 +1372,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): return next_link # Return next_link if it exists - return self.runInteraction( + return self.db.runInteraction( "validate_threepid_session_txn", validate_threepid_session_txn ) @@ -1401,7 +1405,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): if validated_at: insertion_values["validated_at"] = validated_at - return self.simple_upsert( + return self.db.simple_upsert( table="threepid_validation_session", keyvalues={"session_id": session_id}, values={"last_send_attempt": send_attempt}, @@ -1439,7 +1443,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): def start_or_continue_validation_session_txn(txn): # Create or update a validation session - self.simple_upsert_txn( + self.db.simple_upsert_txn( txn, table="threepid_validation_session", keyvalues={"session_id": session_id}, @@ -1452,7 +1456,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): ) # Create a new validation token with this session ID - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="threepid_validation_token", values={ @@ -1463,7 +1467,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): }, ) - return self.runInteraction( + return self.db.runInteraction( "start_or_continue_validation_session", start_or_continue_validation_session_txn, ) @@ -1478,7 +1482,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): """ return txn.execute(sql, (ts,)) - return self.runInteraction( + return self.db.runInteraction( "cull_expired_threepid_validation_tokens", cull_expired_threepid_validation_tokens_txn, self.clock.time_msec(), @@ -1493,7 +1497,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): deactivated (bool): The value to set for `deactivated`. """ - yield self.runInteraction( + yield self.db.runInteraction( "set_user_deactivated_status", self.set_user_deactivated_status_txn, user_id, @@ -1501,7 +1505,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): ) def set_user_deactivated_status_txn(self, txn, user_id, deactivated): - self.simple_update_one_txn( + self.db.simple_update_one_txn( txn=txn, table="users", keyvalues={"name": user_id}, @@ -1529,14 +1533,14 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): ) txn.execute(sql, []) - res = self.cursor_to_dict(txn) + res = self.db.cursor_to_dict(txn) if res: for user in res: self.set_expiration_date_for_user_txn( txn, user["name"], use_delta=True ) - yield self.runInteraction( + yield self.db.runInteraction( "get_users_with_no_expiration_date", select_users_with_no_expiration_date_txn, ) @@ -1560,7 +1564,7 @@ class RegistrationStore(RegistrationBackgroundUpdateStore): expiration_ts, ) - self.simple_upsert_txn( + self.db.simple_upsert_txn( txn, "account_validity", keyvalues={"user_id": user_id}, diff --git a/synapse/storage/data_stores/main/rejections.py b/synapse/storage/data_stores/main/rejections.py index f81f9279a1..1c07c7a425 100644 --- a/synapse/storage/data_stores/main/rejections.py +++ b/synapse/storage/data_stores/main/rejections.py @@ -22,7 +22,7 @@ logger = logging.getLogger(__name__) class RejectionsStore(SQLBaseStore): def _store_rejections_txn(self, txn, event_id, reason): - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="rejections", values={ @@ -33,7 +33,7 @@ class RejectionsStore(SQLBaseStore): ) def get_rejection_reason(self, event_id): - return self.simple_select_one_onecol( + return self.db.simple_select_one_onecol( table="rejections", retcol="reason", keyvalues={"event_id": event_id}, diff --git a/synapse/storage/data_stores/main/relations.py b/synapse/storage/data_stores/main/relations.py index aa5e10538b..046c2b4845 100644 --- a/synapse/storage/data_stores/main/relations.py +++ b/synapse/storage/data_stores/main/relations.py @@ -129,7 +129,7 @@ class RelationsWorkerStore(SQLBaseStore): chunk=list(events[:limit]), next_batch=next_batch, prev_batch=from_token ) - return self.runInteraction( + return self.db.runInteraction( "get_recent_references_for_event", _get_recent_references_for_event_txn ) @@ -223,7 +223,7 @@ class RelationsWorkerStore(SQLBaseStore): chunk=list(events[:limit]), next_batch=next_batch, prev_batch=from_token ) - return self.runInteraction( + return self.db.runInteraction( "get_aggregation_groups_for_event", _get_aggregation_groups_for_event_txn ) @@ -268,7 +268,7 @@ class RelationsWorkerStore(SQLBaseStore): if row: return row[0] - edit_id = yield self.runInteraction( + edit_id = yield self.db.runInteraction( "get_applicable_edit", _get_applicable_edit_txn ) @@ -318,7 +318,7 @@ class RelationsWorkerStore(SQLBaseStore): return bool(txn.fetchone()) - return self.runInteraction( + return self.db.runInteraction( "get_if_user_has_annotated_event", _get_if_user_has_annotated_event ) @@ -352,7 +352,7 @@ class RelationsStore(RelationsWorkerStore): aggregation_key = relation.get("key") - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="event_relations", values={ @@ -380,6 +380,6 @@ class RelationsStore(RelationsWorkerStore): redacted_event_id (str): The event that was redacted. """ - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="event_relations", keyvalues={"event_id": redacted_event_id} ) diff --git a/synapse/storage/data_stores/main/room.py b/synapse/storage/data_stores/main/room.py index f309e3640c..a26ed47afc 100644 --- a/synapse/storage/data_stores/main/room.py +++ b/synapse/storage/data_stores/main/room.py @@ -54,7 +54,7 @@ class RoomWorkerStore(SQLBaseStore): Returns: A dict containing the room information, or None if the room is unknown. """ - return self.simple_select_one( + return self.db.simple_select_one( table="rooms", keyvalues={"room_id": room_id}, retcols=("room_id", "is_public", "creator"), @@ -63,7 +63,7 @@ class RoomWorkerStore(SQLBaseStore): ) def get_public_room_ids(self): - return self.simple_select_onecol( + return self.db.simple_select_onecol( table="rooms", keyvalues={"is_public": True}, retcol="room_id", @@ -120,7 +120,7 @@ class RoomWorkerStore(SQLBaseStore): txn.execute(sql, query_args) return txn.fetchone()[0] - return self.runInteraction("count_public_rooms", _count_public_rooms_txn) + return self.db.runInteraction("count_public_rooms", _count_public_rooms_txn) @defer.inlineCallbacks def get_largest_public_rooms( @@ -253,21 +253,21 @@ class RoomWorkerStore(SQLBaseStore): def _get_largest_public_rooms_txn(txn): txn.execute(sql, query_args) - results = self.cursor_to_dict(txn) + results = self.db.cursor_to_dict(txn) if not forwards: results.reverse() return results - ret_val = yield self.runInteraction( + ret_val = yield self.db.runInteraction( "get_largest_public_rooms", _get_largest_public_rooms_txn ) defer.returnValue(ret_val) @cached(max_entries=10000) def is_room_blocked(self, room_id): - return self.simple_select_one_onecol( + return self.db.simple_select_one_onecol( table="blocked_rooms", keyvalues={"room_id": room_id}, retcol="1", @@ -288,7 +288,7 @@ class RoomWorkerStore(SQLBaseStore): of RatelimitOverride are None or 0 then ratelimitng has been disabled for that user entirely. """ - row = yield self.simple_select_one( + row = yield self.db.simple_select_one( table="ratelimit_override", keyvalues={"user_id": user_id}, retcols=("messages_per_second", "burst_count"), @@ -330,9 +330,9 @@ class RoomWorkerStore(SQLBaseStore): (room_id,), ) - return self.cursor_to_dict(txn) + return self.db.cursor_to_dict(txn) - ret = yield self.runInteraction( + ret = yield self.db.runInteraction( "get_retention_policy_for_room", get_retention_policy_for_room_txn, ) @@ -396,7 +396,7 @@ class RoomBackgroundUpdateStore(BackgroundUpdateStore): (last_room, batch_size), ) - rows = self.cursor_to_dict(txn) + rows = self.db.cursor_to_dict(txn) if not rows: return True @@ -408,7 +408,7 @@ class RoomBackgroundUpdateStore(BackgroundUpdateStore): ev = json.loads(row["json"]) retention_policy = json.dumps(ev["content"]) - self.simple_insert_txn( + self.db.simple_insert_txn( txn=txn, table="room_retention", values={ @@ -430,7 +430,7 @@ class RoomBackgroundUpdateStore(BackgroundUpdateStore): else: return False - end = yield self.runInteraction( + end = yield self.db.runInteraction( "insert_room_retention", _background_insert_retention_txn, ) @@ -461,7 +461,7 @@ class RoomStore(RoomBackgroundUpdateStore, RoomWorkerStore, SearchStore): try: def store_room_txn(txn, next_id): - self.simple_insert_txn( + self.db.simple_insert_txn( txn, "rooms", { @@ -471,7 +471,7 @@ class RoomStore(RoomBackgroundUpdateStore, RoomWorkerStore, SearchStore): }, ) if is_public: - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="public_room_list_stream", values={ @@ -482,7 +482,7 @@ class RoomStore(RoomBackgroundUpdateStore, RoomWorkerStore, SearchStore): ) with self._public_room_id_gen.get_next() as next_id: - yield self.runInteraction("store_room_txn", store_room_txn, next_id) + yield self.db.runInteraction("store_room_txn", store_room_txn, next_id) except Exception as e: logger.error("store_room with room_id=%s failed: %s", room_id, e) raise StoreError(500, "Problem creating room.") @@ -490,14 +490,14 @@ class RoomStore(RoomBackgroundUpdateStore, RoomWorkerStore, SearchStore): @defer.inlineCallbacks def set_room_is_public(self, room_id, is_public): def set_room_is_public_txn(txn, next_id): - self.simple_update_one_txn( + self.db.simple_update_one_txn( txn, table="rooms", keyvalues={"room_id": room_id}, updatevalues={"is_public": is_public}, ) - entries = self.simple_select_list_txn( + entries = self.db.simple_select_list_txn( txn, table="public_room_list_stream", keyvalues={ @@ -515,7 +515,7 @@ class RoomStore(RoomBackgroundUpdateStore, RoomWorkerStore, SearchStore): add_to_stream = bool(entries[-1]["visibility"]) != is_public if add_to_stream: - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="public_room_list_stream", values={ @@ -528,7 +528,7 @@ class RoomStore(RoomBackgroundUpdateStore, RoomWorkerStore, SearchStore): ) with self._public_room_id_gen.get_next() as next_id: - yield self.runInteraction( + yield self.db.runInteraction( "set_room_is_public", set_room_is_public_txn, next_id ) self.hs.get_notifier().on_new_replication_data() @@ -555,7 +555,7 @@ class RoomStore(RoomBackgroundUpdateStore, RoomWorkerStore, SearchStore): def set_room_is_public_appservice_txn(txn, next_id): if is_public: try: - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="appservice_room_list", values={ @@ -568,7 +568,7 @@ class RoomStore(RoomBackgroundUpdateStore, RoomWorkerStore, SearchStore): # We've already inserted, nothing to do. return else: - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="appservice_room_list", keyvalues={ @@ -578,7 +578,7 @@ class RoomStore(RoomBackgroundUpdateStore, RoomWorkerStore, SearchStore): }, ) - entries = self.simple_select_list_txn( + entries = self.db.simple_select_list_txn( txn, table="public_room_list_stream", keyvalues={ @@ -596,7 +596,7 @@ class RoomStore(RoomBackgroundUpdateStore, RoomWorkerStore, SearchStore): add_to_stream = bool(entries[-1]["visibility"]) != is_public if add_to_stream: - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="public_room_list_stream", values={ @@ -609,7 +609,7 @@ class RoomStore(RoomBackgroundUpdateStore, RoomWorkerStore, SearchStore): ) with self._public_room_id_gen.get_next() as next_id: - yield self.runInteraction( + yield self.db.runInteraction( "set_room_is_public_appservice", set_room_is_public_appservice_txn, next_id, @@ -626,7 +626,7 @@ class RoomStore(RoomBackgroundUpdateStore, RoomWorkerStore, SearchStore): row = txn.fetchone() return row[0] or 0 - return self.runInteraction("get_rooms", f) + return self.db.runInteraction("get_rooms", f) def _store_room_topic_txn(self, txn, event): if hasattr(event, "content") and "topic" in event.content: @@ -660,7 +660,7 @@ class RoomStore(RoomBackgroundUpdateStore, RoomWorkerStore, SearchStore): # Ignore the event if one of the value isn't an integer. return - self.simple_insert_txn( + self.db.simple_insert_txn( txn=txn, table="room_retention", values={ @@ -679,7 +679,7 @@ class RoomStore(RoomBackgroundUpdateStore, RoomWorkerStore, SearchStore): self, room_id, event_id, user_id, reason, content, received_ts ): next_id = self._event_reports_id_gen.get_next() - return self.simple_insert( + return self.db.simple_insert( table="event_reports", values={ "id": next_id, @@ -712,7 +712,9 @@ class RoomStore(RoomBackgroundUpdateStore, RoomWorkerStore, SearchStore): if prev_id == current_id: return defer.succeed([]) - return self.runInteraction("get_all_new_public_rooms", get_all_new_public_rooms) + return self.db.runInteraction( + "get_all_new_public_rooms", get_all_new_public_rooms + ) @defer.inlineCallbacks def block_room(self, room_id, user_id): @@ -725,14 +727,14 @@ class RoomStore(RoomBackgroundUpdateStore, RoomWorkerStore, SearchStore): Returns: Deferred """ - yield self.simple_upsert( + yield self.db.simple_upsert( table="blocked_rooms", keyvalues={"room_id": room_id}, values={}, insertion_values={"user_id": user_id}, desc="block_room", ) - yield self.runInteraction( + yield self.db.runInteraction( "block_room_invalidation", self._invalidate_cache_and_stream, self.is_room_blocked, @@ -763,7 +765,9 @@ class RoomStore(RoomBackgroundUpdateStore, RoomWorkerStore, SearchStore): return local_media_mxcs, remote_media_mxcs - return self.runInteraction("get_media_ids_in_room", _get_media_mxcs_in_room_txn) + return self.db.runInteraction( + "get_media_ids_in_room", _get_media_mxcs_in_room_txn + ) def quarantine_media_ids_in_room(self, room_id, quarantined_by): """For a room loops through all events with media and quarantines @@ -802,7 +806,7 @@ class RoomStore(RoomBackgroundUpdateStore, RoomWorkerStore, SearchStore): return total_media_quarantined - return self.runInteraction( + return self.db.runInteraction( "quarantine_media_in_room", _quarantine_media_in_room_txn ) @@ -907,7 +911,7 @@ class RoomStore(RoomBackgroundUpdateStore, RoomWorkerStore, SearchStore): txn.execute(sql, args) - rows = self.cursor_to_dict(txn) + rows = self.db.cursor_to_dict(txn) rooms_dict = {} for row in rows: @@ -923,7 +927,7 @@ class RoomStore(RoomBackgroundUpdateStore, RoomWorkerStore, SearchStore): txn.execute(sql) - rows = self.cursor_to_dict(txn) + rows = self.db.cursor_to_dict(txn) # If a room isn't already in the dict (i.e. it doesn't have a retention # policy in its state), add it with a null policy. @@ -936,7 +940,7 @@ class RoomStore(RoomBackgroundUpdateStore, RoomWorkerStore, SearchStore): return rooms_dict - rooms = yield self.runInteraction( + rooms = yield self.db.runInteraction( "get_rooms_for_retention_period_in_range", get_rooms_for_retention_period_in_range_txn, ) diff --git a/synapse/storage/data_stores/main/roommember.py b/synapse/storage/data_stores/main/roommember.py index fe2428a281..7f4d02b25b 100644 --- a/synapse/storage/data_stores/main/roommember.py +++ b/synapse/storage/data_stores/main/roommember.py @@ -116,7 +116,7 @@ class RoomMemberWorkerStore(EventsWorkerStore): txn.execute(query) return list(txn)[0][0] - count = yield self.runInteraction("get_known_servers", _transact) + count = yield self.db.runInteraction("get_known_servers", _transact) # We always know about ourselves, even if we have nothing in # room_memberships (for example, the server is new). @@ -128,7 +128,7 @@ class RoomMemberWorkerStore(EventsWorkerStore): membership column is up to date """ - pending_update = self.simple_select_one_txn( + pending_update = self.db.simple_select_one_txn( txn, table="background_updates", keyvalues={"update_name": _CURRENT_STATE_MEMBERSHIP_UPDATE_NAME}, @@ -144,7 +144,7 @@ class RoomMemberWorkerStore(EventsWorkerStore): 15.0, run_as_background_process, "_check_safe_current_state_events_membership_updated", - self.runInteraction, + self.db.runInteraction, "_check_safe_current_state_events_membership_updated", self._check_safe_current_state_events_membership_updated_txn, ) @@ -161,7 +161,7 @@ class RoomMemberWorkerStore(EventsWorkerStore): @cached(max_entries=100000, iterable=True) def get_users_in_room(self, room_id): - return self.runInteraction( + return self.db.runInteraction( "get_users_in_room", self.get_users_in_room_txn, room_id ) @@ -269,7 +269,7 @@ class RoomMemberWorkerStore(EventsWorkerStore): return res - return self.runInteraction("get_room_summary", _get_room_summary_txn) + return self.db.runInteraction("get_room_summary", _get_room_summary_txn) def _get_user_counts_in_room_txn(self, txn, room_id): """ @@ -339,7 +339,7 @@ class RoomMemberWorkerStore(EventsWorkerStore): if not membership_list: return defer.succeed(None) - rooms = yield self.runInteraction( + rooms = yield self.db.runInteraction( "get_rooms_for_user_where_membership_is", self._get_rooms_for_user_where_membership_is_txn, user_id, @@ -392,7 +392,7 @@ class RoomMemberWorkerStore(EventsWorkerStore): ) txn.execute(sql, (user_id, *args)) - results = [RoomsForUser(**r) for r in self.cursor_to_dict(txn)] + results = [RoomsForUser(**r) for r in self.db.cursor_to_dict(txn)] if do_invite: sql = ( @@ -412,7 +412,7 @@ class RoomMemberWorkerStore(EventsWorkerStore): stream_ordering=r["stream_ordering"], membership=Membership.INVITE, ) - for r in self.cursor_to_dict(txn) + for r in self.db.cursor_to_dict(txn) ) return results @@ -603,7 +603,7 @@ class RoomMemberWorkerStore(EventsWorkerStore): to `user_id` and ProfileInfo (or None if not join event). """ - rows = yield self.simple_select_many_batch( + rows = yield self.db.simple_select_many_batch( table="room_memberships", column="event_id", iterable=event_ids, @@ -643,7 +643,7 @@ class RoomMemberWorkerStore(EventsWorkerStore): # the returned user actually has the correct domain. like_clause = "%:" + host - rows = yield self.execute("is_host_joined", None, sql, room_id, like_clause) + rows = yield self.db.execute("is_host_joined", None, sql, room_id, like_clause) if not rows: return False @@ -683,7 +683,7 @@ class RoomMemberWorkerStore(EventsWorkerStore): # the returned user actually has the correct domain. like_clause = "%:" + host - rows = yield self.execute("was_host_joined", None, sql, room_id, like_clause) + rows = yield self.db.execute("was_host_joined", None, sql, room_id, like_clause) if not rows: return False @@ -753,7 +753,7 @@ class RoomMemberWorkerStore(EventsWorkerStore): rows = txn.fetchall() return rows[0][0] - count = yield self.runInteraction("did_forget_membership", f) + count = yield self.db.runInteraction("did_forget_membership", f) return count == 0 @cached() @@ -790,7 +790,7 @@ class RoomMemberWorkerStore(EventsWorkerStore): txn.execute(sql, (user_id,)) return set(row[0] for row in txn if row[1] == 0) - return self.runInteraction( + return self.db.runInteraction( "get_forgotten_rooms_for_user", _get_forgotten_rooms_for_user_txn ) @@ -805,7 +805,7 @@ class RoomMemberWorkerStore(EventsWorkerStore): Deferred[set[str]]: Set of room IDs. """ - room_ids = yield self.simple_select_onecol( + room_ids = yield self.db.simple_select_onecol( table="room_memberships", keyvalues={"membership": Membership.JOIN, "user_id": user_id}, retcol="room_id", @@ -820,7 +820,7 @@ class RoomMemberWorkerStore(EventsWorkerStore): """Get user_id and membership of a set of event IDs. """ - return self.simple_select_many_batch( + return self.db.simple_select_many_batch( table="room_memberships", column="event_id", iterable=member_event_ids, @@ -874,7 +874,7 @@ class RoomMemberBackgroundUpdateStore(BackgroundUpdateStore): txn.execute(sql, (target_min_stream_id, max_stream_id, batch_size)) - rows = self.cursor_to_dict(txn) + rows = self.db.cursor_to_dict(txn) if not rows: return 0 @@ -915,7 +915,7 @@ class RoomMemberBackgroundUpdateStore(BackgroundUpdateStore): return len(rows) - result = yield self.runInteraction( + result = yield self.db.runInteraction( _MEMBERSHIP_PROFILE_UPDATE_NAME, add_membership_profile_txn ) @@ -971,7 +971,7 @@ class RoomMemberBackgroundUpdateStore(BackgroundUpdateStore): # string, which will compare before all room IDs correctly. last_processed_room = progress.get("last_processed_room", "") - row_count, finished = yield self.runInteraction( + row_count, finished = yield self.db.runInteraction( "_background_current_state_membership_update", _background_current_state_membership_txn, last_processed_room, @@ -990,7 +990,7 @@ class RoomMemberStore(RoomMemberWorkerStore, RoomMemberBackgroundUpdateStore): def _store_room_members_txn(self, txn, events, backfilled): """Store a room member in the database. """ - self.simple_insert_many_txn( + self.db.simple_insert_many_txn( txn, table="room_memberships", values=[ @@ -1028,7 +1028,7 @@ class RoomMemberStore(RoomMemberWorkerStore, RoomMemberBackgroundUpdateStore): is_mine = self.hs.is_mine_id(event.state_key) if is_new_state and is_mine: if event.membership == Membership.INVITE: - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="local_invites", values={ @@ -1068,7 +1068,7 @@ class RoomMemberStore(RoomMemberWorkerStore, RoomMemberBackgroundUpdateStore): txn.execute(sql, (stream_ordering, True, room_id, user_id)) with self._stream_id_gen.get_next() as stream_ordering: - yield self.runInteraction("locally_reject_invite", f, stream_ordering) + yield self.db.runInteraction("locally_reject_invite", f, stream_ordering) def forget(self, user_id, room_id): """Indicate that user_id wishes to discard history for room_id.""" @@ -1091,7 +1091,7 @@ class RoomMemberStore(RoomMemberWorkerStore, RoomMemberBackgroundUpdateStore): txn, self.get_forgotten_rooms_for_user, (user_id,) ) - return self.runInteraction("forget_membership", f) + return self.db.runInteraction("forget_membership", f) class _JoinedHostsCache(object): diff --git a/synapse/storage/data_stores/main/search.py b/synapse/storage/data_stores/main/search.py index f735cf095c..55a604850e 100644 --- a/synapse/storage/data_stores/main/search.py +++ b/synapse/storage/data_stores/main/search.py @@ -93,7 +93,7 @@ class SearchBackgroundUpdateStore(BackgroundUpdateStore): # store_search_entries_txn with a generator function, but that # would mean having two cursors open on the database at once. # Instead we just build a list of results. - rows = self.cursor_to_dict(txn) + rows = self.db.cursor_to_dict(txn) if not rows: return 0 @@ -159,7 +159,7 @@ class SearchBackgroundUpdateStore(BackgroundUpdateStore): return len(event_search_rows) - result = yield self.runInteraction( + result = yield self.db.runInteraction( self.EVENT_SEARCH_UPDATE_NAME, reindex_search_txn ) @@ -206,7 +206,7 @@ class SearchBackgroundUpdateStore(BackgroundUpdateStore): conn.set_session(autocommit=False) if isinstance(self.database_engine, PostgresEngine): - yield self.runWithConnection(create_index) + yield self.db.runWithConnection(create_index) yield self._end_background_update(self.EVENT_SEARCH_USE_GIN_POSTGRES_NAME) return 1 @@ -237,12 +237,12 @@ class SearchBackgroundUpdateStore(BackgroundUpdateStore): ) conn.set_session(autocommit=False) - yield self.runWithConnection(create_index) + yield self.db.runWithConnection(create_index) pg = dict(progress) pg["have_added_indexes"] = True - yield self.runInteraction( + yield self.db.runInteraction( self.EVENT_SEARCH_ORDER_UPDATE_NAME, self._background_update_progress_txn, self.EVENT_SEARCH_ORDER_UPDATE_NAME, @@ -280,7 +280,7 @@ class SearchBackgroundUpdateStore(BackgroundUpdateStore): return len(rows), True - num_rows, finished = yield self.runInteraction( + num_rows, finished = yield self.db.runInteraction( self.EVENT_SEARCH_ORDER_UPDATE_NAME, reindex_search_txn ) @@ -441,7 +441,9 @@ class SearchStore(SearchBackgroundUpdateStore): # entire table from the database. sql += " ORDER BY rank DESC LIMIT 500" - results = yield self.execute("search_msgs", self.cursor_to_dict, sql, *args) + results = yield self.db.execute( + "search_msgs", self.db.cursor_to_dict, sql, *args + ) results = list(filter(lambda row: row["room_id"] in room_ids, results)) @@ -455,8 +457,8 @@ class SearchStore(SearchBackgroundUpdateStore): count_sql += " GROUP BY room_id" - count_results = yield self.execute( - "search_rooms_count", self.cursor_to_dict, count_sql, *count_args + count_results = yield self.db.execute( + "search_rooms_count", self.db.cursor_to_dict, count_sql, *count_args ) count = sum(row["count"] for row in count_results if row["room_id"] in room_ids) @@ -586,7 +588,9 @@ class SearchStore(SearchBackgroundUpdateStore): args.append(limit) - results = yield self.execute("search_rooms", self.cursor_to_dict, sql, *args) + results = yield self.db.execute( + "search_rooms", self.db.cursor_to_dict, sql, *args + ) results = list(filter(lambda row: row["room_id"] in room_ids, results)) @@ -600,8 +604,8 @@ class SearchStore(SearchBackgroundUpdateStore): count_sql += " GROUP BY room_id" - count_results = yield self.execute( - "search_rooms_count", self.cursor_to_dict, count_sql, *count_args + count_results = yield self.db.execute( + "search_rooms_count", self.db.cursor_to_dict, count_sql, *count_args ) count = sum(row["count"] for row in count_results if row["room_id"] in room_ids) @@ -686,7 +690,7 @@ class SearchStore(SearchBackgroundUpdateStore): return highlight_words - return self.runInteraction("_find_highlights", f) + return self.db.runInteraction("_find_highlights", f) def _to_postgres_options(options_dict): diff --git a/synapse/storage/data_stores/main/signatures.py b/synapse/storage/data_stores/main/signatures.py index f3da29ce14..563216b63c 100644 --- a/synapse/storage/data_stores/main/signatures.py +++ b/synapse/storage/data_stores/main/signatures.py @@ -48,7 +48,7 @@ class SignatureWorkerStore(SQLBaseStore): for event_id in event_ids } - return self.runInteraction("get_event_reference_hashes", f) + return self.db.runInteraction("get_event_reference_hashes", f) @defer.inlineCallbacks def add_event_hashes(self, event_ids): @@ -98,4 +98,4 @@ class SignatureStore(SignatureWorkerStore): } ) - self.simple_insert_many_txn(txn, table="event_reference_hashes", values=vals) + self.db.simple_insert_many_txn(txn, table="event_reference_hashes", values=vals) diff --git a/synapse/storage/data_stores/main/state.py b/synapse/storage/data_stores/main/state.py index 2b33ec1a35..851e81d6b3 100644 --- a/synapse/storage/data_stores/main/state.py +++ b/synapse/storage/data_stores/main/state.py @@ -89,7 +89,7 @@ class StateGroupBackgroundUpdateStore(SQLBaseStore): count = 0 while next_group: - next_group = self.simple_select_one_onecol_txn( + next_group = self.db.simple_select_one_onecol_txn( txn, table="state_group_edges", keyvalues={"state_group": next_group}, @@ -192,7 +192,7 @@ class StateGroupBackgroundUpdateStore(SQLBaseStore): ): break - next_group = self.simple_select_one_onecol_txn( + next_group = self.db.simple_select_one_onecol_txn( txn, table="state_group_edges", keyvalues={"state_group": next_group}, @@ -348,7 +348,9 @@ class StateGroupWorkerStore( (intern_string(r[0]), intern_string(r[1])): to_ascii(r[2]) for r in txn } - return self.runInteraction("get_current_state_ids", _get_current_state_ids_txn) + return self.db.runInteraction( + "get_current_state_ids", _get_current_state_ids_txn + ) # FIXME: how should this be cached? def get_filtered_current_state_ids(self, room_id, state_filter=StateFilter.all()): @@ -392,7 +394,7 @@ class StateGroupWorkerStore( return results - return self.runInteraction( + return self.db.runInteraction( "get_filtered_current_state_ids", _get_filtered_current_state_ids_txn ) @@ -431,7 +433,7 @@ class StateGroupWorkerStore( """ def _get_state_group_delta_txn(txn): - prev_group = self.simple_select_one_onecol_txn( + prev_group = self.db.simple_select_one_onecol_txn( txn, table="state_group_edges", keyvalues={"state_group": state_group}, @@ -442,7 +444,7 @@ class StateGroupWorkerStore( if not prev_group: return _GetStateGroupDelta(None, None) - delta_ids = self.simple_select_list_txn( + delta_ids = self.db.simple_select_list_txn( txn, table="state_groups_state", keyvalues={"state_group": state_group}, @@ -454,7 +456,9 @@ class StateGroupWorkerStore( {(row["type"], row["state_key"]): row["event_id"] for row in delta_ids}, ) - return self.runInteraction("get_state_group_delta", _get_state_group_delta_txn) + return self.db.runInteraction( + "get_state_group_delta", _get_state_group_delta_txn + ) @defer.inlineCallbacks def get_state_groups_ids(self, _room_id, event_ids): @@ -540,7 +544,7 @@ class StateGroupWorkerStore( chunks = [groups[i : i + 100] for i in range(0, len(groups), 100)] for chunk in chunks: - res = yield self.runInteraction( + res = yield self.db.runInteraction( "_get_state_groups_from_groups", self._get_state_groups_from_groups_txn, chunk, @@ -644,7 +648,7 @@ class StateGroupWorkerStore( @cached(max_entries=50000) def _get_state_group_for_event(self, event_id): - return self.simple_select_one_onecol( + return self.db.simple_select_one_onecol( table="event_to_state_groups", keyvalues={"event_id": event_id}, retcol="state_group", @@ -661,7 +665,7 @@ class StateGroupWorkerStore( def _get_state_group_for_events(self, event_ids): """Returns mapping event_id -> state_group """ - rows = yield self.simple_select_many_batch( + rows = yield self.db.simple_select_many_batch( table="event_to_state_groups", column="event_id", iterable=event_ids, @@ -902,7 +906,7 @@ class StateGroupWorkerStore( state_group = self.database_engine.get_next_state_group_id(txn) - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="state_groups", values={"id": state_group, "room_id": room_id, "event_id": event_id}, @@ -911,7 +915,7 @@ class StateGroupWorkerStore( # We persist as a delta if we can, while also ensuring the chain # of deltas isn't tooo long, as otherwise read performance degrades. if prev_group: - is_in_db = self.simple_select_one_onecol_txn( + is_in_db = self.db.simple_select_one_onecol_txn( txn, table="state_groups", keyvalues={"id": prev_group}, @@ -926,13 +930,13 @@ class StateGroupWorkerStore( potential_hops = self._count_state_group_hops_txn(txn, prev_group) if prev_group and potential_hops < MAX_STATE_DELTA_HOPS: - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="state_group_edges", values={"state_group": state_group, "prev_state_group": prev_group}, ) - self.simple_insert_many_txn( + self.db.simple_insert_many_txn( txn, table="state_groups_state", values=[ @@ -947,7 +951,7 @@ class StateGroupWorkerStore( ], ) else: - self.simple_insert_many_txn( + self.db.simple_insert_many_txn( txn, table="state_groups_state", values=[ @@ -993,7 +997,7 @@ class StateGroupWorkerStore( return state_group - return self.runInteraction("store_state_group", _store_state_group_txn) + return self.db.runInteraction("store_state_group", _store_state_group_txn) @defer.inlineCallbacks def get_referenced_state_groups(self, state_groups): @@ -1007,7 +1011,7 @@ class StateGroupWorkerStore( referenced. """ - rows = yield self.simple_select_many_batch( + rows = yield self.db.simple_select_many_batch( table="event_to_state_groups", column="state_group", iterable=state_groups, @@ -1065,7 +1069,7 @@ class StateBackgroundUpdateStore( batch_size = max(1, int(batch_size / BATCH_SIZE_SCALE_FACTOR)) if max_group is None: - rows = yield self.execute( + rows = yield self.db.execute( "_background_deduplicate_state", None, "SELECT coalesce(max(id), 0) FROM state_groups", @@ -1135,13 +1139,13 @@ class StateBackgroundUpdateStore( if prev_state.get(key, None) != value } - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="state_group_edges", keyvalues={"state_group": state_group}, ) - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="state_group_edges", values={ @@ -1150,13 +1154,13 @@ class StateBackgroundUpdateStore( }, ) - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="state_groups_state", keyvalues={"state_group": state_group}, ) - self.simple_insert_many_txn( + self.db.simple_insert_many_txn( txn, table="state_groups_state", values=[ @@ -1183,7 +1187,7 @@ class StateBackgroundUpdateStore( return False, batch_size - finished, result = yield self.runInteraction( + finished, result = yield self.db.runInteraction( self.STATE_GROUP_DEDUPLICATION_UPDATE_NAME, reindex_txn ) @@ -1218,7 +1222,7 @@ class StateBackgroundUpdateStore( ) txn.execute("DROP INDEX IF EXISTS state_groups_state_id") - yield self.runWithConnection(reindex_txn) + yield self.db.runWithConnection(reindex_txn) yield self._end_background_update(self.STATE_GROUP_INDEX_UPDATE_NAME) @@ -1263,7 +1267,7 @@ class StateStore(StateGroupWorkerStore, StateBackgroundUpdateStore): state_groups[event.event_id] = context.state_group - self.simple_insert_many_txn( + self.db.simple_insert_many_txn( txn, table="event_to_state_groups", values=[ diff --git a/synapse/storage/data_stores/main/state_deltas.py b/synapse/storage/data_stores/main/state_deltas.py index 03b908026b..12c982cb26 100644 --- a/synapse/storage/data_stores/main/state_deltas.py +++ b/synapse/storage/data_stores/main/state_deltas.py @@ -98,14 +98,14 @@ class StateDeltasStore(SQLBaseStore): ORDER BY stream_id ASC """ txn.execute(sql, (prev_stream_id, clipped_stream_id)) - return clipped_stream_id, self.cursor_to_dict(txn) + return clipped_stream_id, self.db.cursor_to_dict(txn) - return self.runInteraction( + return self.db.runInteraction( "get_current_state_deltas", get_current_state_deltas_txn ) def _get_max_stream_id_in_current_state_deltas_txn(self, txn): - return self.simple_select_one_onecol_txn( + return self.db.simple_select_one_onecol_txn( txn, table="current_state_delta_stream", keyvalues={}, @@ -113,7 +113,7 @@ class StateDeltasStore(SQLBaseStore): ) def get_max_stream_id_in_current_state_deltas(self): - return self.runInteraction( + return self.db.runInteraction( "get_max_stream_id_in_current_state_deltas", self._get_max_stream_id_in_current_state_deltas_txn, ) diff --git a/synapse/storage/data_stores/main/stats.py b/synapse/storage/data_stores/main/stats.py index 3aeba859fd..974ffc15bd 100644 --- a/synapse/storage/data_stores/main/stats.py +++ b/synapse/storage/data_stores/main/stats.py @@ -117,7 +117,7 @@ class StatsStore(StateDeltasStore): txn.execute(sql, (last_user_id, batch_size)) return [r for r, in txn] - users_to_work_on = yield self.runInteraction( + users_to_work_on = yield self.db.runInteraction( "_populate_stats_process_users", _get_next_batch ) @@ -130,7 +130,7 @@ class StatsStore(StateDeltasStore): yield self._calculate_and_set_initial_state_for_user(user_id) progress["last_user_id"] = user_id - yield self.runInteraction( + yield self.db.runInteraction( "populate_stats_process_users", self._background_update_progress_txn, "populate_stats_process_users", @@ -160,7 +160,7 @@ class StatsStore(StateDeltasStore): txn.execute(sql, (last_room_id, batch_size)) return [r for r, in txn] - rooms_to_work_on = yield self.runInteraction( + rooms_to_work_on = yield self.db.runInteraction( "populate_stats_rooms_get_batch", _get_next_batch ) @@ -173,7 +173,7 @@ class StatsStore(StateDeltasStore): yield self._calculate_and_set_initial_state_for_room(room_id) progress["last_room_id"] = room_id - yield self.runInteraction( + yield self.db.runInteraction( "_populate_stats_process_rooms", self._background_update_progress_txn, "populate_stats_process_rooms", @@ -186,7 +186,7 @@ class StatsStore(StateDeltasStore): """ Returns the stats processor positions. """ - return self.simple_select_one_onecol( + return self.db.simple_select_one_onecol( table="stats_incremental_position", keyvalues={}, retcol="stream_id", @@ -215,7 +215,7 @@ class StatsStore(StateDeltasStore): if field and "\0" in field: fields[col] = None - return self.simple_upsert( + return self.db.simple_upsert( table="room_stats_state", keyvalues={"room_id": room_id}, values=fields, @@ -236,7 +236,7 @@ class StatsStore(StateDeltasStore): Deferred[list[dict]], where the dict has the keys of ABSOLUTE_STATS_FIELDS[stats_type], and "bucket_size" and "end_ts". """ - return self.runInteraction( + return self.db.runInteraction( "get_statistics_for_subject", self._get_statistics_for_subject_txn, stats_type, @@ -257,7 +257,7 @@ class StatsStore(StateDeltasStore): ABSOLUTE_STATS_FIELDS[stats_type] + PER_SLICE_FIELDS[stats_type] ) - slice_list = self.simple_select_list_paginate_txn( + slice_list = self.db.simple_select_list_paginate_txn( txn, table + "_historical", {id_col: stats_id}, @@ -282,7 +282,7 @@ class StatsStore(StateDeltasStore): "name", "topic", "canonical_alias", "avatar", "join_rules", "history_visibility" """ - return self.simple_select_one( + return self.db.simple_select_one( "room_stats_state", {"room_id": room_id}, retcols=( @@ -308,7 +308,7 @@ class StatsStore(StateDeltasStore): """ table, id_col = TYPE_TO_TABLE[stats_type] - return self.simple_select_one_onecol( + return self.db.simple_select_one_onecol( "%s_current" % (table,), keyvalues={id_col: id}, retcol="completed_delta_stream_id", @@ -344,14 +344,14 @@ class StatsStore(StateDeltasStore): complete_with_stream_id=stream_id, ) - self.simple_update_one_txn( + self.db.simple_update_one_txn( txn, table="stats_incremental_position", keyvalues={}, updatevalues={"stream_id": stream_id}, ) - return self.runInteraction( + return self.db.runInteraction( "bulk_update_stats_delta", _bulk_update_stats_delta_txn ) @@ -382,7 +382,7 @@ class StatsStore(StateDeltasStore): Does not work with per-slice fields. """ - return self.runInteraction( + return self.db.runInteraction( "update_stats_delta", self._update_stats_delta_txn, ts, @@ -517,17 +517,17 @@ class StatsStore(StateDeltasStore): else: self.database_engine.lock_table(txn, table) retcols = list(chain(absolutes.keys(), additive_relatives.keys())) - current_row = self.simple_select_one_txn( + current_row = self.db.simple_select_one_txn( txn, table, keyvalues, retcols, allow_none=True ) if current_row is None: merged_dict = {**keyvalues, **absolutes, **additive_relatives} - self.simple_insert_txn(txn, table, merged_dict) + self.db.simple_insert_txn(txn, table, merged_dict) else: for (key, val) in additive_relatives.items(): current_row[key] += val current_row.update(absolutes) - self.simple_update_one_txn(txn, table, keyvalues, current_row) + self.db.simple_update_one_txn(txn, table, keyvalues, current_row) def _upsert_copy_from_table_with_additive_relatives_txn( self, @@ -614,11 +614,11 @@ class StatsStore(StateDeltasStore): txn.execute(sql, qargs) else: self.database_engine.lock_table(txn, into_table) - src_row = self.simple_select_one_txn( + src_row = self.db.simple_select_one_txn( txn, src_table, keyvalues, copy_columns ) all_dest_keyvalues = {**keyvalues, **extra_dst_keyvalues} - dest_current_row = self.simple_select_one_txn( + dest_current_row = self.db.simple_select_one_txn( txn, into_table, keyvalues=all_dest_keyvalues, @@ -634,11 +634,11 @@ class StatsStore(StateDeltasStore): **src_row, **additive_relatives, } - self.simple_insert_txn(txn, into_table, merged_dict) + self.db.simple_insert_txn(txn, into_table, merged_dict) else: for (key, val) in additive_relatives.items(): src_row[key] = dest_current_row[key] + val - self.simple_update_txn(txn, into_table, all_dest_keyvalues, src_row) + self.db.simple_update_txn(txn, into_table, all_dest_keyvalues, src_row) def get_changes_room_total_events_and_bytes(self, min_pos, max_pos): """Fetches the counts of events in the given range of stream IDs. @@ -652,7 +652,7 @@ class StatsStore(StateDeltasStore): changes. """ - return self.runInteraction( + return self.db.runInteraction( "stats_incremental_total_events_and_bytes", self.get_changes_room_total_events_and_bytes_txn, min_pos, @@ -735,7 +735,7 @@ class StatsStore(StateDeltasStore): def _fetch_current_state_stats(txn): pos = self.get_room_max_stream_ordering() - rows = self.simple_select_many_txn( + rows = self.db.simple_select_many_txn( txn, table="current_state_events", column="type", @@ -791,7 +791,7 @@ class StatsStore(StateDeltasStore): current_state_events_count, users_in_room, pos, - ) = yield self.runInteraction( + ) = yield self.db.runInteraction( "get_initial_state_for_room", _fetch_current_state_stats ) @@ -866,7 +866,7 @@ class StatsStore(StateDeltasStore): (count,) = txn.fetchone() return count, pos - joined_rooms, pos = yield self.runInteraction( + joined_rooms, pos = yield self.db.runInteraction( "calculate_and_set_initial_state_for_user", _calculate_and_set_initial_state_for_user_txn, ) diff --git a/synapse/storage/data_stores/main/stream.py b/synapse/storage/data_stores/main/stream.py index 60487c4559..2ff8c57109 100644 --- a/synapse/storage/data_stores/main/stream.py +++ b/synapse/storage/data_stores/main/stream.py @@ -255,7 +255,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): super(StreamWorkerStore, self).__init__(db_conn, hs) events_max = self.get_room_max_stream_ordering() - event_cache_prefill, min_event_val = self.get_cache_dict( + event_cache_prefill, min_event_val = self.db.get_cache_dict( db_conn, "events", entity_column="room_id", @@ -400,7 +400,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): rows = [_EventDictReturn(row[0], None, row[1]) for row in txn] return rows - rows = yield self.runInteraction("get_room_events_stream_for_room", f) + rows = yield self.db.runInteraction("get_room_events_stream_for_room", f) ret = yield self.get_events_as_list( [r.event_id for r in rows], get_prev_content=True @@ -450,7 +450,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): return rows - rows = yield self.runInteraction("get_membership_changes_for_user", f) + rows = yield self.db.runInteraction("get_membership_changes_for_user", f) ret = yield self.get_events_as_list( [r.event_id for r in rows], get_prev_content=True @@ -511,7 +511,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): end_token = RoomStreamToken.parse(end_token) - rows, token = yield self.runInteraction( + rows, token = yield self.db.runInteraction( "get_recent_event_ids_for_room", self._paginate_room_events_txn, room_id, @@ -548,7 +548,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): txn.execute(sql, (room_id, stream_ordering)) return txn.fetchone() - return self.runInteraction("get_room_event_after_stream_ordering", _f) + return self.db.runInteraction("get_room_event_after_stream_ordering", _f) @defer.inlineCallbacks def get_room_events_max_id(self, room_id=None): @@ -562,7 +562,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): if room_id is None: return "s%d" % (token,) else: - topo = yield self.runInteraction( + topo = yield self.db.runInteraction( "_get_max_topological_txn", self._get_max_topological_txn, room_id ) return "t%d-%d" % (topo, token) @@ -576,7 +576,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): Returns: A deferred "s%d" stream token. """ - return self.simple_select_one_onecol( + return self.db.simple_select_one_onecol( table="events", keyvalues={"event_id": event_id}, retcol="stream_ordering" ).addCallback(lambda row: "s%d" % (row,)) @@ -589,7 +589,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): Returns: A deferred "t%d-%d" topological token. """ - return self.simple_select_one( + return self.db.simple_select_one( table="events", keyvalues={"event_id": event_id}, retcols=("stream_ordering", "topological_ordering"), @@ -613,7 +613,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): "SELECT coalesce(max(topological_ordering), 0) FROM events" " WHERE room_id = ? AND stream_ordering < ?" ) - return self.execute( + return self.db.execute( "get_max_topological_token", None, sql, room_id, stream_key ).addCallback(lambda r: r[0][0] if r else 0) @@ -667,7 +667,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): dict """ - results = yield self.runInteraction( + results = yield self.db.runInteraction( "get_events_around", self._get_events_around_txn, room_id, @@ -709,7 +709,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): dict """ - results = self.simple_select_one_txn( + results = self.db.simple_select_one_txn( txn, "events", keyvalues={"event_id": event_id, "room_id": room_id}, @@ -788,7 +788,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): return upper_bound, [row[1] for row in rows] - upper_bound, event_ids = yield self.runInteraction( + upper_bound, event_ids = yield self.db.runInteraction( "get_all_new_events_stream", get_all_new_events_stream_txn ) @@ -797,7 +797,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): return upper_bound, events def get_federation_out_pos(self, typ): - return self.simple_select_one_onecol( + return self.db.simple_select_one_onecol( table="federation_stream_position", retcol="stream_id", keyvalues={"type": typ}, @@ -805,7 +805,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): ) def update_federation_out_pos(self, typ, stream_id): - return self.simple_update_one( + return self.db.simple_update_one( table="federation_stream_position", keyvalues={"type": typ}, updatevalues={"stream_id": stream_id}, @@ -956,7 +956,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): if to_key: to_key = RoomStreamToken.parse(to_key) - rows, token = yield self.runInteraction( + rows, token = yield self.db.runInteraction( "paginate_room_events", self._paginate_room_events_txn, room_id, diff --git a/synapse/storage/data_stores/main/tags.py b/synapse/storage/data_stores/main/tags.py index 85012403be..2aa1bafd48 100644 --- a/synapse/storage/data_stores/main/tags.py +++ b/synapse/storage/data_stores/main/tags.py @@ -41,7 +41,7 @@ class TagsWorkerStore(AccountDataWorkerStore): tag strings to tag content. """ - deferred = self.simple_select_list( + deferred = self.db.simple_select_list( "room_tags", {"user_id": user_id}, ["room_id", "tag", "content"] ) @@ -78,7 +78,7 @@ class TagsWorkerStore(AccountDataWorkerStore): txn.execute(sql, (last_id, current_id, limit)) return txn.fetchall() - tag_ids = yield self.runInteraction( + tag_ids = yield self.db.runInteraction( "get_all_updated_tags", get_all_updated_tags_txn ) @@ -98,7 +98,7 @@ class TagsWorkerStore(AccountDataWorkerStore): batch_size = 50 results = [] for i in range(0, len(tag_ids), batch_size): - tags = yield self.runInteraction( + tags = yield self.db.runInteraction( "get_all_updated_tag_content", get_tag_content, tag_ids[i : i + batch_size], @@ -135,7 +135,9 @@ class TagsWorkerStore(AccountDataWorkerStore): if not changed: return {} - room_ids = yield self.runInteraction("get_updated_tags", get_updated_tags_txn) + room_ids = yield self.db.runInteraction( + "get_updated_tags", get_updated_tags_txn + ) results = {} if room_ids: @@ -153,7 +155,7 @@ class TagsWorkerStore(AccountDataWorkerStore): Returns: A deferred list of string tags. """ - return self.simple_select_list( + return self.db.simple_select_list( table="room_tags", keyvalues={"user_id": user_id, "room_id": room_id}, retcols=("tag", "content"), @@ -178,7 +180,7 @@ class TagsStore(TagsWorkerStore): content_json = json.dumps(content) def add_tag_txn(txn, next_id): - self.simple_upsert_txn( + self.db.simple_upsert_txn( txn, table="room_tags", keyvalues={"user_id": user_id, "room_id": room_id, "tag": tag}, @@ -187,7 +189,7 @@ class TagsStore(TagsWorkerStore): self._update_revision_txn(txn, user_id, room_id, next_id) with self._account_data_id_gen.get_next() as next_id: - yield self.runInteraction("add_tag", add_tag_txn, next_id) + yield self.db.runInteraction("add_tag", add_tag_txn, next_id) self.get_tags_for_user.invalidate((user_id,)) @@ -210,7 +212,7 @@ class TagsStore(TagsWorkerStore): self._update_revision_txn(txn, user_id, room_id, next_id) with self._account_data_id_gen.get_next() as next_id: - yield self.runInteraction("remove_tag", remove_tag_txn, next_id) + yield self.db.runInteraction("remove_tag", remove_tag_txn, next_id) self.get_tags_for_user.invalidate((user_id,)) diff --git a/synapse/storage/data_stores/main/transactions.py b/synapse/storage/data_stores/main/transactions.py index c162f3ea16..c0d155a43c 100644 --- a/synapse/storage/data_stores/main/transactions.py +++ b/synapse/storage/data_stores/main/transactions.py @@ -77,7 +77,7 @@ class TransactionStore(SQLBaseStore): this transaction or a 2-tuple of (int, dict) """ - return self.runInteraction( + return self.db.runInteraction( "get_received_txn_response", self._get_received_txn_response, transaction_id, @@ -85,7 +85,7 @@ class TransactionStore(SQLBaseStore): ) def _get_received_txn_response(self, txn, transaction_id, origin): - result = self.simple_select_one_txn( + result = self.db.simple_select_one_txn( txn, table="received_transactions", keyvalues={"transaction_id": transaction_id, "origin": origin}, @@ -119,7 +119,7 @@ class TransactionStore(SQLBaseStore): response_json (str) """ - return self.simple_insert( + return self.db.simple_insert( table="received_transactions", values={ "transaction_id": transaction_id, @@ -148,7 +148,7 @@ class TransactionStore(SQLBaseStore): if result is not SENTINEL: return result - result = yield self.runInteraction( + result = yield self.db.runInteraction( "get_destination_retry_timings", self._get_destination_retry_timings, destination, @@ -160,7 +160,7 @@ class TransactionStore(SQLBaseStore): return result def _get_destination_retry_timings(self, txn, destination): - result = self.simple_select_one_txn( + result = self.db.simple_select_one_txn( txn, table="destinations", keyvalues={"destination": destination}, @@ -187,7 +187,7 @@ class TransactionStore(SQLBaseStore): """ self._destination_retry_cache.pop(destination, None) - return self.runInteraction( + return self.db.runInteraction( "set_destination_retry_timings", self._set_destination_retry_timings, destination, @@ -227,7 +227,7 @@ class TransactionStore(SQLBaseStore): # We need to be careful here as the data may have changed from under us # due to a worker setting the timings. - prev_row = self.simple_select_one_txn( + prev_row = self.db.simple_select_one_txn( txn, table="destinations", keyvalues={"destination": destination}, @@ -236,7 +236,7 @@ class TransactionStore(SQLBaseStore): ) if not prev_row: - self.simple_insert_txn( + self.db.simple_insert_txn( txn, table="destinations", values={ @@ -247,7 +247,7 @@ class TransactionStore(SQLBaseStore): }, ) elif retry_interval == 0 or prev_row["retry_interval"] < retry_interval: - self.simple_update_one_txn( + self.db.simple_update_one_txn( txn, "destinations", keyvalues={"destination": destination}, @@ -270,4 +270,6 @@ class TransactionStore(SQLBaseStore): def _cleanup_transactions_txn(txn): txn.execute("DELETE FROM received_transactions WHERE ts < ?", (month_ago,)) - return self.runInteraction("_cleanup_transactions", _cleanup_transactions_txn) + return self.db.runInteraction( + "_cleanup_transactions", _cleanup_transactions_txn + ) diff --git a/synapse/storage/data_stores/main/user_directory.py b/synapse/storage/data_stores/main/user_directory.py index 1a85aabbfb..7118bd62f3 100644 --- a/synapse/storage/data_stores/main/user_directory.py +++ b/synapse/storage/data_stores/main/user_directory.py @@ -85,7 +85,7 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore """ txn.execute(sql) rooms = [{"room_id": x[0], "events": x[1]} for x in txn.fetchall()] - self.simple_insert_many_txn(txn, TEMP_TABLE + "_rooms", rooms) + self.db.simple_insert_many_txn(txn, TEMP_TABLE + "_rooms", rooms) del rooms # If search all users is on, get all the users we want to add. @@ -100,13 +100,13 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore txn.execute("SELECT name FROM users") users = [{"user_id": x[0]} for x in txn.fetchall()] - self.simple_insert_many_txn(txn, TEMP_TABLE + "_users", users) + self.db.simple_insert_many_txn(txn, TEMP_TABLE + "_users", users) new_pos = yield self.get_max_stream_id_in_current_state_deltas() - yield self.runInteraction( + yield self.db.runInteraction( "populate_user_directory_temp_build", _make_staging_area ) - yield self.simple_insert(TEMP_TABLE + "_position", {"position": new_pos}) + yield self.db.simple_insert(TEMP_TABLE + "_position", {"position": new_pos}) yield self._end_background_update("populate_user_directory_createtables") return 1 @@ -116,7 +116,7 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore """ Update the user directory stream position, then clean up the old tables. """ - position = yield self.simple_select_one_onecol( + position = yield self.db.simple_select_one_onecol( TEMP_TABLE + "_position", None, "position" ) yield self.update_user_directory_stream_pos(position) @@ -126,7 +126,7 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore txn.execute("DROP TABLE IF EXISTS " + TEMP_TABLE + "_users") txn.execute("DROP TABLE IF EXISTS " + TEMP_TABLE + "_position") - yield self.runInteraction( + yield self.db.runInteraction( "populate_user_directory_cleanup", _delete_staging_area ) @@ -170,7 +170,7 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore return rooms_to_work_on - rooms_to_work_on = yield self.runInteraction( + rooms_to_work_on = yield self.db.runInteraction( "populate_user_directory_temp_read", _get_next_batch ) @@ -243,10 +243,10 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore to_insert.clear() # We've finished a room. Delete it from the table. - yield self.simple_delete_one(TEMP_TABLE + "_rooms", {"room_id": room_id}) + yield self.db.simple_delete_one(TEMP_TABLE + "_rooms", {"room_id": room_id}) # Update the remaining counter. progress["remaining"] -= 1 - yield self.runInteraction( + yield self.db.runInteraction( "populate_user_directory", self._background_update_progress_txn, "populate_user_directory_process_rooms", @@ -291,7 +291,7 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore return users_to_work_on - users_to_work_on = yield self.runInteraction( + users_to_work_on = yield self.db.runInteraction( "populate_user_directory_temp_read", _get_next_batch ) @@ -312,10 +312,10 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore ) # We've finished processing a user. Delete it from the table. - yield self.simple_delete_one(TEMP_TABLE + "_users", {"user_id": user_id}) + yield self.db.simple_delete_one(TEMP_TABLE + "_users", {"user_id": user_id}) # Update the remaining counter. progress["remaining"] -= 1 - yield self.runInteraction( + yield self.db.runInteraction( "populate_user_directory", self._background_update_progress_txn, "populate_user_directory_process_users", @@ -361,7 +361,7 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore """ def _update_profile_in_user_dir_txn(txn): - new_entry = self.simple_upsert_txn( + new_entry = self.db.simple_upsert_txn( txn, table="user_directory", keyvalues={"user_id": user_id}, @@ -435,7 +435,7 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore ) elif isinstance(self.database_engine, Sqlite3Engine): value = "%s %s" % (user_id, display_name) if display_name else user_id - self.simple_upsert_txn( + self.db.simple_upsert_txn( txn, table="user_directory_search", keyvalues={"user_id": user_id}, @@ -448,7 +448,7 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore txn.call_after(self.get_user_in_directory.invalidate, (user_id,)) - return self.runInteraction( + return self.db.runInteraction( "update_profile_in_user_dir", _update_profile_in_user_dir_txn ) @@ -462,7 +462,7 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore """ def _add_users_who_share_room_txn(txn): - self.simple_upsert_many_txn( + self.db.simple_upsert_many_txn( txn, table="users_who_share_private_rooms", key_names=["user_id", "other_user_id", "room_id"], @@ -474,7 +474,7 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore value_values=None, ) - return self.runInteraction( + return self.db.runInteraction( "add_users_who_share_room", _add_users_who_share_room_txn ) @@ -489,7 +489,7 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore def _add_users_in_public_rooms_txn(txn): - self.simple_upsert_many_txn( + self.db.simple_upsert_many_txn( txn, table="users_in_public_rooms", key_names=["user_id", "room_id"], @@ -498,7 +498,7 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore value_values=None, ) - return self.runInteraction( + return self.db.runInteraction( "add_users_in_public_rooms", _add_users_in_public_rooms_txn ) @@ -513,13 +513,13 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore txn.execute("DELETE FROM users_who_share_private_rooms") txn.call_after(self.get_user_in_directory.invalidate_all) - return self.runInteraction( + return self.db.runInteraction( "delete_all_from_user_dir", _delete_all_from_user_dir_txn ) @cached() def get_user_in_directory(self, user_id): - return self.simple_select_one( + return self.db.simple_select_one( table="user_directory", keyvalues={"user_id": user_id}, retcols=("display_name", "avatar_url"), @@ -528,7 +528,7 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore ) def update_user_directory_stream_pos(self, stream_id): - return self.simple_update_one( + return self.db.simple_update_one( table="user_directory_stream_pos", keyvalues={}, updatevalues={"stream_id": stream_id}, @@ -547,42 +547,42 @@ class UserDirectoryStore(UserDirectoryBackgroundUpdateStore): def remove_from_user_dir(self, user_id): def _remove_from_user_dir_txn(txn): - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="user_directory", keyvalues={"user_id": user_id} ) - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="user_directory_search", keyvalues={"user_id": user_id} ) - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="users_in_public_rooms", keyvalues={"user_id": user_id} ) - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="users_who_share_private_rooms", keyvalues={"user_id": user_id}, ) - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="users_who_share_private_rooms", keyvalues={"other_user_id": user_id}, ) txn.call_after(self.get_user_in_directory.invalidate, (user_id,)) - return self.runInteraction("remove_from_user_dir", _remove_from_user_dir_txn) + return self.db.runInteraction("remove_from_user_dir", _remove_from_user_dir_txn) @defer.inlineCallbacks def get_users_in_dir_due_to_room(self, room_id): """Get all user_ids that are in the room directory because they're in the given room_id """ - user_ids_share_pub = yield self.simple_select_onecol( + user_ids_share_pub = yield self.db.simple_select_onecol( table="users_in_public_rooms", keyvalues={"room_id": room_id}, retcol="user_id", desc="get_users_in_dir_due_to_room", ) - user_ids_share_priv = yield self.simple_select_onecol( + user_ids_share_priv = yield self.db.simple_select_onecol( table="users_who_share_private_rooms", keyvalues={"room_id": room_id}, retcol="other_user_id", @@ -605,23 +605,23 @@ class UserDirectoryStore(UserDirectoryBackgroundUpdateStore): """ def _remove_user_who_share_room_txn(txn): - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="users_who_share_private_rooms", keyvalues={"user_id": user_id, "room_id": room_id}, ) - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="users_who_share_private_rooms", keyvalues={"other_user_id": user_id, "room_id": room_id}, ) - self.simple_delete_txn( + self.db.simple_delete_txn( txn, table="users_in_public_rooms", keyvalues={"user_id": user_id, "room_id": room_id}, ) - return self.runInteraction( + return self.db.runInteraction( "remove_user_who_share_room", _remove_user_who_share_room_txn ) @@ -636,14 +636,14 @@ class UserDirectoryStore(UserDirectoryBackgroundUpdateStore): Returns: list: user_id """ - rows = yield self.simple_select_onecol( + rows = yield self.db.simple_select_onecol( table="users_who_share_private_rooms", keyvalues={"user_id": user_id}, retcol="room_id", desc="get_rooms_user_is_in", ) - pub_rows = yield self.simple_select_onecol( + pub_rows = yield self.db.simple_select_onecol( table="users_in_public_rooms", keyvalues={"user_id": user_id}, retcol="room_id", @@ -674,14 +674,14 @@ class UserDirectoryStore(UserDirectoryBackgroundUpdateStore): ) f2 USING (room_id) """ - rows = yield self.execute( + rows = yield self.db.execute( "get_rooms_in_common_for_users", None, sql, user_id, other_user_id ) return [room_id for room_id, in rows] def get_user_directory_stream_pos(self): - return self.simple_select_one_onecol( + return self.db.simple_select_one_onecol( table="user_directory_stream_pos", keyvalues={}, retcol="stream_id", @@ -786,7 +786,9 @@ class UserDirectoryStore(UserDirectoryBackgroundUpdateStore): # This should be unreachable. raise Exception("Unrecognized database engine") - results = yield self.execute("search_user_dir", self.cursor_to_dict, sql, *args) + results = yield self.db.execute( + "search_user_dir", self.db.cursor_to_dict, sql, *args + ) limited = len(results) > limit diff --git a/synapse/storage/data_stores/main/user_erasure_store.py b/synapse/storage/data_stores/main/user_erasure_store.py index 37860af070..af8025bc17 100644 --- a/synapse/storage/data_stores/main/user_erasure_store.py +++ b/synapse/storage/data_stores/main/user_erasure_store.py @@ -31,7 +31,7 @@ class UserErasureWorkerStore(SQLBaseStore): Returns: Deferred[bool]: True if the user has requested erasure """ - return self.simple_select_onecol( + return self.db.simple_select_onecol( table="erased_users", keyvalues={"user_id": user_id}, retcol="1", @@ -56,7 +56,7 @@ class UserErasureWorkerStore(SQLBaseStore): # iterate it multiple times, and (b) avoiding duplicates. user_ids = tuple(set(user_ids)) - rows = yield self.simple_select_many_batch( + rows = yield self.db.simple_select_many_batch( table="erased_users", column="user_id", iterable=user_ids, @@ -88,4 +88,4 @@ class UserErasureStore(UserErasureWorkerStore): self._invalidate_cache_and_stream(txn, self.is_user_erased, (user_id,)) - return self.runInteraction("mark_user_erased", f) + return self.db.runInteraction("mark_user_erased", f) diff --git a/synapse/storage/database.py b/synapse/storage/database.py new file mode 100644 index 0000000000..c2e121a001 --- /dev/null +++ b/synapse/storage/database.py @@ -0,0 +1,1485 @@ +# -*- coding: utf-8 -*- +# Copyright 2014-2016 OpenMarket Ltd +# Copyright 2017-2018 New Vector Ltd +# Copyright 2019 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import logging +import random +import sys +import time +from typing import Iterable, Tuple + +from six import iteritems, iterkeys, itervalues +from six.moves import intern, range + +from prometheus_client import Histogram + +from twisted.internet import defer + +from synapse.api.errors import StoreError +from synapse.logging.context import LoggingContext, make_deferred_yieldable +from synapse.metrics.background_process_metrics import run_as_background_process +from synapse.storage.engines import PostgresEngine, Sqlite3Engine +from synapse.util.stringutils import exception_to_unicode + +# import a function which will return a monotonic time, in seconds +try: + # on python 3, use time.monotonic, since time.clock can go backwards + from time import monotonic as monotonic_time +except ImportError: + # ... but python 2 doesn't have it + from time import clock as monotonic_time + +logger = logging.getLogger(__name__) + +try: + MAX_TXN_ID = sys.maxint - 1 +except AttributeError: + # python 3 does not have a maximum int value + MAX_TXN_ID = 2 ** 63 - 1 + +sql_logger = logging.getLogger("synapse.storage.SQL") +transaction_logger = logging.getLogger("synapse.storage.txn") +perf_logger = logging.getLogger("synapse.storage.TIME") + +sql_scheduling_timer = Histogram("synapse_storage_schedule_time", "sec") + +sql_query_timer = Histogram("synapse_storage_query_time", "sec", ["verb"]) +sql_txn_timer = Histogram("synapse_storage_transaction_time", "sec", ["desc"]) + + +# Unique indexes which have been added in background updates. Maps from table name +# to the name of the background update which added the unique index to that table. +# +# This is used by the upsert logic to figure out which tables are safe to do a proper +# UPSERT on: until the relevant background update has completed, we +# have to emulate an upsert by locking the table. +# +UNIQUE_INDEX_BACKGROUND_UPDATES = { + "user_ips": "user_ips_device_unique_index", + "device_lists_remote_extremeties": "device_lists_remote_extremeties_unique_idx", + "device_lists_remote_cache": "device_lists_remote_cache_unique_idx", + "event_search": "event_search_event_id_idx", +} + + +class LoggingTransaction(object): + """An object that almost-transparently proxies for the 'txn' object + passed to the constructor. Adds logging and metrics to the .execute() + method. + + Args: + txn: The database transcation object to wrap. + name (str): The name of this transactions for logging. + database_engine (Sqlite3Engine|PostgresEngine) + after_callbacks(list|None): A list that callbacks will be appended to + that have been added by `call_after` which should be run on + successful completion of the transaction. None indicates that no + callbacks should be allowed to be scheduled to run. + exception_callbacks(list|None): A list that callbacks will be appended + to that have been added by `call_on_exception` which should be run + if transaction ends with an error. None indicates that no callbacks + should be allowed to be scheduled to run. + """ + + __slots__ = [ + "txn", + "name", + "database_engine", + "after_callbacks", + "exception_callbacks", + ] + + def __init__( + self, txn, name, database_engine, after_callbacks=None, exception_callbacks=None + ): + object.__setattr__(self, "txn", txn) + object.__setattr__(self, "name", name) + object.__setattr__(self, "database_engine", database_engine) + object.__setattr__(self, "after_callbacks", after_callbacks) + object.__setattr__(self, "exception_callbacks", exception_callbacks) + + def call_after(self, callback, *args, **kwargs): + """Call the given callback on the main twisted thread after the + transaction has finished. Used to invalidate the caches on the + correct thread. + """ + self.after_callbacks.append((callback, args, kwargs)) + + def call_on_exception(self, callback, *args, **kwargs): + self.exception_callbacks.append((callback, args, kwargs)) + + def __getattr__(self, name): + return getattr(self.txn, name) + + def __setattr__(self, name, value): + setattr(self.txn, name, value) + + def __iter__(self): + return self.txn.__iter__() + + def execute_batch(self, sql, args): + if isinstance(self.database_engine, PostgresEngine): + from psycopg2.extras import execute_batch + + self._do_execute(lambda *x: execute_batch(self.txn, *x), sql, args) + else: + for val in args: + self.execute(sql, val) + + def execute(self, sql, *args): + self._do_execute(self.txn.execute, sql, *args) + + def executemany(self, sql, *args): + self._do_execute(self.txn.executemany, sql, *args) + + def _make_sql_one_line(self, sql): + "Strip newlines out of SQL so that the loggers in the DB are on one line" + return " ".join(l.strip() for l in sql.splitlines() if l.strip()) + + def _do_execute(self, func, sql, *args): + sql = self._make_sql_one_line(sql) + + # TODO(paul): Maybe use 'info' and 'debug' for values? + sql_logger.debug("[SQL] {%s} %s", self.name, sql) + + sql = self.database_engine.convert_param_style(sql) + if args: + try: + sql_logger.debug("[SQL values] {%s} %r", self.name, args[0]) + except Exception: + # Don't let logging failures stop SQL from working + pass + + start = time.time() + + try: + return func(sql, *args) + except Exception as e: + logger.debug("[SQL FAIL] {%s} %s", self.name, e) + raise + finally: + secs = time.time() - start + sql_logger.debug("[SQL time] {%s} %f sec", self.name, secs) + sql_query_timer.labels(sql.split()[0]).observe(secs) + + +class PerformanceCounters(object): + def __init__(self): + self.current_counters = {} + self.previous_counters = {} + + def update(self, key, duration_secs): + count, cum_time = self.current_counters.get(key, (0, 0)) + count += 1 + cum_time += duration_secs + self.current_counters[key] = (count, cum_time) + + def interval(self, interval_duration_secs, limit=3): + counters = [] + for name, (count, cum_time) in iteritems(self.current_counters): + prev_count, prev_time = self.previous_counters.get(name, (0, 0)) + counters.append( + ( + (cum_time - prev_time) / interval_duration_secs, + count - prev_count, + name, + ) + ) + + self.previous_counters = dict(self.current_counters) + + counters.sort(reverse=True) + + top_n_counters = ", ".join( + "%s(%d): %.3f%%" % (name, count, 100 * ratio) + for ratio, count, name in counters[:limit] + ) + + return top_n_counters + + +class Database(object): + _TXN_ID = 0 + + def __init__(self, hs): + self.hs = hs + self._clock = hs.get_clock() + self._db_pool = hs.get_db_pool() + + self._previous_txn_total_time = 0 + self._current_txn_total_time = 0 + self._previous_loop_ts = 0 + + # TODO(paul): These can eventually be removed once the metrics code + # is running in mainline, and we have some nice monitoring frontends + # to watch it + self._txn_perf_counters = PerformanceCounters() + + self.database_engine = hs.database_engine + + # A set of tables that are not safe to use native upserts in. + self._unsafe_to_upsert_tables = set(UNIQUE_INDEX_BACKGROUND_UPDATES.keys()) + + # We add the user_directory_search table to the blacklist on SQLite + # because the existing search table does not have an index, making it + # unsafe to use native upserts. + if isinstance(self.database_engine, Sqlite3Engine): + self._unsafe_to_upsert_tables.add("user_directory_search") + + if self.database_engine.can_native_upsert: + # Check ASAP (and then later, every 1s) to see if we have finished + # background updates of tables that aren't safe to update. + self._clock.call_later( + 0.0, + run_as_background_process, + "upsert_safety_check", + self._check_safe_to_upsert, + ) + + self.rand = random.SystemRandom() + + @defer.inlineCallbacks + def _check_safe_to_upsert(self): + """ + Is it safe to use native UPSERT? + + If there are background updates, we will need to wait, as they may be + the addition of indexes that set the UNIQUE constraint that we require. + + If the background updates have not completed, wait 15 sec and check again. + """ + updates = yield self.simple_select_list( + "background_updates", + keyvalues=None, + retcols=["update_name"], + desc="check_background_updates", + ) + updates = [x["update_name"] for x in updates] + + for table, update_name in UNIQUE_INDEX_BACKGROUND_UPDATES.items(): + if update_name not in updates: + logger.debug("Now safe to upsert in %s", table) + self._unsafe_to_upsert_tables.discard(table) + + # If there's any updates still running, reschedule to run. + if updates: + self._clock.call_later( + 15.0, + run_as_background_process, + "upsert_safety_check", + self._check_safe_to_upsert, + ) + + def start_profiling(self): + self._previous_loop_ts = monotonic_time() + + def loop(): + curr = self._current_txn_total_time + prev = self._previous_txn_total_time + self._previous_txn_total_time = curr + + time_now = monotonic_time() + time_then = self._previous_loop_ts + self._previous_loop_ts = time_now + + duration = time_now - time_then + ratio = (curr - prev) / duration + + top_three_counters = self._txn_perf_counters.interval(duration, limit=3) + + perf_logger.info( + "Total database time: %.3f%% {%s}", ratio * 100, top_three_counters + ) + + self._clock.looping_call(loop, 10000) + + def new_transaction( + self, conn, desc, after_callbacks, exception_callbacks, func, *args, **kwargs + ): + start = monotonic_time() + txn_id = self._TXN_ID + + # We don't really need these to be unique, so lets stop it from + # growing really large. + self._TXN_ID = (self._TXN_ID + 1) % (MAX_TXN_ID) + + name = "%s-%x" % (desc, txn_id) + + transaction_logger.debug("[TXN START] {%s}", name) + + try: + i = 0 + N = 5 + while True: + cursor = LoggingTransaction( + conn.cursor(), + name, + self.database_engine, + after_callbacks, + exception_callbacks, + ) + try: + r = func(cursor, *args, **kwargs) + conn.commit() + return r + except self.database_engine.module.OperationalError as e: + # This can happen if the database disappears mid + # transaction. + logger.warning( + "[TXN OPERROR] {%s} %s %d/%d", + name, + exception_to_unicode(e), + i, + N, + ) + if i < N: + i += 1 + try: + conn.rollback() + except self.database_engine.module.Error as e1: + logger.warning( + "[TXN EROLL] {%s} %s", name, exception_to_unicode(e1) + ) + continue + raise + except self.database_engine.module.DatabaseError as e: + if self.database_engine.is_deadlock(e): + logger.warning("[TXN DEADLOCK] {%s} %d/%d", name, i, N) + if i < N: + i += 1 + try: + conn.rollback() + except self.database_engine.module.Error as e1: + logger.warning( + "[TXN EROLL] {%s} %s", + name, + exception_to_unicode(e1), + ) + continue + raise + finally: + # we're either about to retry with a new cursor, or we're about to + # release the connection. Once we release the connection, it could + # get used for another query, which might do a conn.rollback(). + # + # In the latter case, even though that probably wouldn't affect the + # results of this transaction, python's sqlite will reset all + # statements on the connection [1], which will make our cursor + # invalid [2]. + # + # In any case, continuing to read rows after commit()ing seems + # dubious from the PoV of ACID transactional semantics + # (sqlite explicitly says that once you commit, you may see rows + # from subsequent updates.) + # + # In psycopg2, cursors are essentially a client-side fabrication - + # all the data is transferred to the client side when the statement + # finishes executing - so in theory we could go on streaming results + # from the cursor, but attempting to do so would make us + # incompatible with sqlite, so let's make sure we're not doing that + # by closing the cursor. + # + # (*named* cursors in psycopg2 are different and are proper server- + # side things, but (a) we don't use them and (b) they are implicitly + # closed by ending the transaction anyway.) + # + # In short, if we haven't finished with the cursor yet, that's a + # problem waiting to bite us. + # + # TL;DR: we're done with the cursor, so we can close it. + # + # [1]: https://github.com/python/cpython/blob/v3.8.0/Modules/_sqlite/connection.c#L465 + # [2]: https://github.com/python/cpython/blob/v3.8.0/Modules/_sqlite/cursor.c#L236 + cursor.close() + except Exception as e: + logger.debug("[TXN FAIL] {%s} %s", name, e) + raise + finally: + end = monotonic_time() + duration = end - start + + LoggingContext.current_context().add_database_transaction(duration) + + transaction_logger.debug("[TXN END] {%s} %f sec", name, duration) + + self._current_txn_total_time += duration + self._txn_perf_counters.update(desc, duration) + sql_txn_timer.labels(desc).observe(duration) + + @defer.inlineCallbacks + def runInteraction(self, desc, func, *args, **kwargs): + """Starts a transaction on the database and runs a given function + + Arguments: + desc (str): description of the transaction, for logging and metrics + func (func): callback function, which will be called with a + database transaction (twisted.enterprise.adbapi.Transaction) as + its first argument, followed by `args` and `kwargs`. + + args (list): positional args to pass to `func` + kwargs (dict): named args to pass to `func` + + Returns: + Deferred: The result of func + """ + after_callbacks = [] + exception_callbacks = [] + + if LoggingContext.current_context() == LoggingContext.sentinel: + logger.warning("Starting db txn '%s' from sentinel context", desc) + + try: + result = yield self.runWithConnection( + self.new_transaction, + desc, + after_callbacks, + exception_callbacks, + func, + *args, + **kwargs + ) + + for after_callback, after_args, after_kwargs in after_callbacks: + after_callback(*after_args, **after_kwargs) + except: # noqa: E722, as we reraise the exception this is fine. + for after_callback, after_args, after_kwargs in exception_callbacks: + after_callback(*after_args, **after_kwargs) + raise + + return result + + @defer.inlineCallbacks + def runWithConnection(self, func, *args, **kwargs): + """Wraps the .runWithConnection() method on the underlying db_pool. + + Arguments: + func (func): callback function, which will be called with a + database connection (twisted.enterprise.adbapi.Connection) as + its first argument, followed by `args` and `kwargs`. + args (list): positional args to pass to `func` + kwargs (dict): named args to pass to `func` + + Returns: + Deferred: The result of func + """ + parent_context = LoggingContext.current_context() + if parent_context == LoggingContext.sentinel: + logger.warning( + "Starting db connection from sentinel context: metrics will be lost" + ) + parent_context = None + + start_time = monotonic_time() + + def inner_func(conn, *args, **kwargs): + with LoggingContext("runWithConnection", parent_context) as context: + sched_duration_sec = monotonic_time() - start_time + sql_scheduling_timer.observe(sched_duration_sec) + context.add_database_scheduled(sched_duration_sec) + + if self.database_engine.is_connection_closed(conn): + logger.debug("Reconnecting closed database connection") + conn.reconnect() + + return func(conn, *args, **kwargs) + + result = yield make_deferred_yieldable( + self._db_pool.runWithConnection(inner_func, *args, **kwargs) + ) + + return result + + @staticmethod + def cursor_to_dict(cursor): + """Converts a SQL cursor into an list of dicts. + + Args: + cursor : The DBAPI cursor which has executed a query. + Returns: + A list of dicts where the key is the column header. + """ + col_headers = list(intern(str(column[0])) for column in cursor.description) + results = list(dict(zip(col_headers, row)) for row in cursor) + return results + + def execute(self, desc, decoder, query, *args): + """Runs a single query for a result set. + + Args: + decoder - The function which can resolve the cursor results to + something meaningful. + query - The query string to execute + *args - Query args. + Returns: + The result of decoder(results) + """ + + def interaction(txn): + txn.execute(query, args) + if decoder: + return decoder(txn) + else: + return txn.fetchall() + + return self.runInteraction(desc, interaction) + + # "Simple" SQL API methods that operate on a single table with no JOINs, + # no complex WHERE clauses, just a dict of values for columns. + + @defer.inlineCallbacks + def simple_insert(self, table, values, or_ignore=False, desc="simple_insert"): + """Executes an INSERT query on the named table. + + Args: + table : string giving the table name + values : dict of new column names and values for them + or_ignore : bool stating whether an exception should be raised + when a conflicting row already exists. If True, False will be + returned by the function instead + desc : string giving a description of the transaction + + Returns: + bool: Whether the row was inserted or not. Only useful when + `or_ignore` is True + """ + try: + yield self.runInteraction(desc, self.simple_insert_txn, table, values) + except self.database_engine.module.IntegrityError: + # We have to do or_ignore flag at this layer, since we can't reuse + # a cursor after we receive an error from the db. + if not or_ignore: + raise + return False + return True + + @staticmethod + def simple_insert_txn(txn, table, values): + keys, vals = zip(*values.items()) + + sql = "INSERT INTO %s (%s) VALUES(%s)" % ( + table, + ", ".join(k for k in keys), + ", ".join("?" for _ in keys), + ) + + txn.execute(sql, vals) + + def simple_insert_many(self, table, values, desc): + return self.runInteraction(desc, self.simple_insert_many_txn, table, values) + + @staticmethod + def simple_insert_many_txn(txn, table, values): + if not values: + return + + # This is a *slight* abomination to get a list of tuples of key names + # and a list of tuples of value names. + # + # i.e. [{"a": 1, "b": 2}, {"c": 3, "d": 4}] + # => [("a", "b",), ("c", "d",)] and [(1, 2,), (3, 4,)] + # + # The sort is to ensure that we don't rely on dictionary iteration + # order. + keys, vals = zip( + *[zip(*(sorted(i.items(), key=lambda kv: kv[0]))) for i in values if i] + ) + + for k in keys: + if k != keys[0]: + raise RuntimeError("All items must have the same keys") + + sql = "INSERT INTO %s (%s) VALUES(%s)" % ( + table, + ", ".join(k for k in keys[0]), + ", ".join("?" for _ in keys[0]), + ) + + txn.executemany(sql, vals) + + @defer.inlineCallbacks + def simple_upsert( + self, + table, + keyvalues, + values, + insertion_values={}, + desc="simple_upsert", + lock=True, + ): + """ + + `lock` should generally be set to True (the default), but can be set + to False if either of the following are true: + + * there is a UNIQUE INDEX on the key columns. In this case a conflict + will cause an IntegrityError in which case this function will retry + the update. + + * we somehow know that we are the only thread which will be updating + this table. + + Args: + table (str): The table to upsert into + keyvalues (dict): The unique key columns and their new values + values (dict): The nonunique columns and their new values + insertion_values (dict): additional key/values to use only when + inserting + lock (bool): True to lock the table when doing the upsert. + Returns: + Deferred(None or bool): Native upserts always return None. Emulated + upserts return True if a new entry was created, False if an existing + one was updated. + """ + attempts = 0 + while True: + try: + result = yield self.runInteraction( + desc, + self.simple_upsert_txn, + table, + keyvalues, + values, + insertion_values, + lock=lock, + ) + return result + except self.database_engine.module.IntegrityError as e: + attempts += 1 + if attempts >= 5: + # don't retry forever, because things other than races + # can cause IntegrityErrors + raise + + # presumably we raced with another transaction: let's retry. + logger.warning( + "IntegrityError when upserting into %s; retrying: %s", table, e + ) + + def simple_upsert_txn( + self, txn, table, keyvalues, values, insertion_values={}, lock=True + ): + """ + Pick the UPSERT method which works best on the platform. Either the + native one (Pg9.5+, recent SQLites), or fall back to an emulated method. + + Args: + txn: The transaction to use. + table (str): The table to upsert into + keyvalues (dict): The unique key tables and their new values + values (dict): The nonunique columns and their new values + insertion_values (dict): additional key/values to use only when + inserting + lock (bool): True to lock the table when doing the upsert. + Returns: + None or bool: Native upserts always return None. Emulated + upserts return True if a new entry was created, False if an existing + one was updated. + """ + if ( + self.database_engine.can_native_upsert + and table not in self._unsafe_to_upsert_tables + ): + return self.simple_upsert_txn_native_upsert( + txn, table, keyvalues, values, insertion_values=insertion_values + ) + else: + return self.simple_upsert_txn_emulated( + txn, + table, + keyvalues, + values, + insertion_values=insertion_values, + lock=lock, + ) + + def simple_upsert_txn_emulated( + self, txn, table, keyvalues, values, insertion_values={}, lock=True + ): + """ + Args: + table (str): The table to upsert into + keyvalues (dict): The unique key tables and their new values + values (dict): The nonunique columns and their new values + insertion_values (dict): additional key/values to use only when + inserting + lock (bool): True to lock the table when doing the upsert. + Returns: + bool: Return True if a new entry was created, False if an existing + one was updated. + """ + # We need to lock the table :(, unless we're *really* careful + if lock: + self.database_engine.lock_table(txn, table) + + def _getwhere(key): + # If the value we're passing in is None (aka NULL), we need to use + # IS, not =, as NULL = NULL equals NULL (False). + if keyvalues[key] is None: + return "%s IS ?" % (key,) + else: + return "%s = ?" % (key,) + + if not values: + # If `values` is empty, then all of the values we care about are in + # the unique key, so there is nothing to UPDATE. We can just do a + # SELECT instead to see if it exists. + sql = "SELECT 1 FROM %s WHERE %s" % ( + table, + " AND ".join(_getwhere(k) for k in keyvalues), + ) + sqlargs = list(keyvalues.values()) + txn.execute(sql, sqlargs) + if txn.fetchall(): + # We have an existing record. + return False + else: + # First try to update. + sql = "UPDATE %s SET %s WHERE %s" % ( + table, + ", ".join("%s = ?" % (k,) for k in values), + " AND ".join(_getwhere(k) for k in keyvalues), + ) + sqlargs = list(values.values()) + list(keyvalues.values()) + + txn.execute(sql, sqlargs) + if txn.rowcount > 0: + # successfully updated at least one row. + return False + + # We didn't find any existing rows, so insert a new one + allvalues = {} + allvalues.update(keyvalues) + allvalues.update(values) + allvalues.update(insertion_values) + + sql = "INSERT INTO %s (%s) VALUES (%s)" % ( + table, + ", ".join(k for k in allvalues), + ", ".join("?" for _ in allvalues), + ) + txn.execute(sql, list(allvalues.values())) + # successfully inserted + return True + + def simple_upsert_txn_native_upsert( + self, txn, table, keyvalues, values, insertion_values={} + ): + """ + Use the native UPSERT functionality in recent PostgreSQL versions. + + Args: + table (str): The table to upsert into + keyvalues (dict): The unique key tables and their new values + values (dict): The nonunique columns and their new values + insertion_values (dict): additional key/values to use only when + inserting + Returns: + None + """ + allvalues = {} + allvalues.update(keyvalues) + allvalues.update(insertion_values) + + if not values: + latter = "NOTHING" + else: + allvalues.update(values) + latter = "UPDATE SET " + ", ".join(k + "=EXCLUDED." + k for k in values) + + sql = ("INSERT INTO %s (%s) VALUES (%s) ON CONFLICT (%s) DO %s") % ( + table, + ", ".join(k for k in allvalues), + ", ".join("?" for _ in allvalues), + ", ".join(k for k in keyvalues), + latter, + ) + txn.execute(sql, list(allvalues.values())) + + def simple_upsert_many_txn( + self, txn, table, key_names, key_values, value_names, value_values + ): + """ + Upsert, many times. + + Args: + table (str): The table to upsert into + key_names (list[str]): The key column names. + key_values (list[list]): A list of each row's key column values. + value_names (list[str]): The value column names. If empty, no + values will be used, even if value_values is provided. + value_values (list[list]): A list of each row's value column values. + Returns: + None + """ + if ( + self.database_engine.can_native_upsert + and table not in self._unsafe_to_upsert_tables + ): + return self.simple_upsert_many_txn_native_upsert( + txn, table, key_names, key_values, value_names, value_values + ) + else: + return self.simple_upsert_many_txn_emulated( + txn, table, key_names, key_values, value_names, value_values + ) + + def simple_upsert_many_txn_emulated( + self, txn, table, key_names, key_values, value_names, value_values + ): + """ + Upsert, many times, but without native UPSERT support or batching. + + Args: + table (str): The table to upsert into + key_names (list[str]): The key column names. + key_values (list[list]): A list of each row's key column values. + value_names (list[str]): The value column names. If empty, no + values will be used, even if value_values is provided. + value_values (list[list]): A list of each row's value column values. + Returns: + None + """ + # No value columns, therefore make a blank list so that the following + # zip() works correctly. + if not value_names: + value_values = [() for x in range(len(key_values))] + + for keyv, valv in zip(key_values, value_values): + _keys = {x: y for x, y in zip(key_names, keyv)} + _vals = {x: y for x, y in zip(value_names, valv)} + + self.simple_upsert_txn_emulated(txn, table, _keys, _vals) + + def simple_upsert_many_txn_native_upsert( + self, txn, table, key_names, key_values, value_names, value_values + ): + """ + Upsert, many times, using batching where possible. + + Args: + table (str): The table to upsert into + key_names (list[str]): The key column names. + key_values (list[list]): A list of each row's key column values. + value_names (list[str]): The value column names. If empty, no + values will be used, even if value_values is provided. + value_values (list[list]): A list of each row's value column values. + Returns: + None + """ + allnames = [] + allnames.extend(key_names) + allnames.extend(value_names) + + if not value_names: + # No value columns, therefore make a blank list so that the + # following zip() works correctly. + latter = "NOTHING" + value_values = [() for x in range(len(key_values))] + else: + latter = "UPDATE SET " + ", ".join( + k + "=EXCLUDED." + k for k in value_names + ) + + sql = "INSERT INTO %s (%s) VALUES (%s) ON CONFLICT (%s) DO %s" % ( + table, + ", ".join(k for k in allnames), + ", ".join("?" for _ in allnames), + ", ".join(key_names), + latter, + ) + + args = [] + + for x, y in zip(key_values, value_values): + args.append(tuple(x) + tuple(y)) + + return txn.execute_batch(sql, args) + + def simple_select_one( + self, table, keyvalues, retcols, allow_none=False, desc="simple_select_one" + ): + """Executes a SELECT query on the named table, which is expected to + return a single row, returning multiple columns from it. + + Args: + table : string giving the table name + keyvalues : dict of column names and values to select the row with + retcols : list of strings giving the names of the columns to return + + allow_none : If true, return None instead of failing if the SELECT + statement returns no rows + """ + return self.runInteraction( + desc, self.simple_select_one_txn, table, keyvalues, retcols, allow_none + ) + + def simple_select_one_onecol( + self, + table, + keyvalues, + retcol, + allow_none=False, + desc="simple_select_one_onecol", + ): + """Executes a SELECT query on the named table, which is expected to + return a single row, returning a single column from it. + + Args: + table : string giving the table name + keyvalues : dict of column names and values to select the row with + retcol : string giving the name of the column to return + """ + return self.runInteraction( + desc, + self.simple_select_one_onecol_txn, + table, + keyvalues, + retcol, + allow_none=allow_none, + ) + + @classmethod + def simple_select_one_onecol_txn( + cls, txn, table, keyvalues, retcol, allow_none=False + ): + ret = cls.simple_select_onecol_txn( + txn, table=table, keyvalues=keyvalues, retcol=retcol + ) + + if ret: + return ret[0] + else: + if allow_none: + return None + else: + raise StoreError(404, "No row found") + + @staticmethod + def simple_select_onecol_txn(txn, table, keyvalues, retcol): + sql = ("SELECT %(retcol)s FROM %(table)s") % {"retcol": retcol, "table": table} + + if keyvalues: + sql += " WHERE %s" % " AND ".join("%s = ?" % k for k in iterkeys(keyvalues)) + txn.execute(sql, list(keyvalues.values())) + else: + txn.execute(sql) + + return [r[0] for r in txn] + + def simple_select_onecol( + self, table, keyvalues, retcol, desc="simple_select_onecol" + ): + """Executes a SELECT query on the named table, which returns a list + comprising of the values of the named column from the selected rows. + + Args: + table (str): table name + keyvalues (dict|None): column names and values to select the rows with + retcol (str): column whos value we wish to retrieve. + + Returns: + Deferred: Results in a list + """ + return self.runInteraction( + desc, self.simple_select_onecol_txn, table, keyvalues, retcol + ) + + def simple_select_list(self, table, keyvalues, retcols, desc="simple_select_list"): + """Executes a SELECT query on the named table, which may return zero or + more rows, returning the result as a list of dicts. + + Args: + table (str): the table name + keyvalues (dict[str, Any] | None): + column names and values to select the rows with, or None to not + apply a WHERE clause. + retcols (iterable[str]): the names of the columns to return + Returns: + defer.Deferred: resolves to list[dict[str, Any]] + """ + return self.runInteraction( + desc, self.simple_select_list_txn, table, keyvalues, retcols + ) + + @classmethod + def simple_select_list_txn(cls, txn, table, keyvalues, retcols): + """Executes a SELECT query on the named table, which may return zero or + more rows, returning the result as a list of dicts. + + Args: + txn : Transaction object + table (str): the table name + keyvalues (dict[str, T] | None): + column names and values to select the rows with, or None to not + apply a WHERE clause. + retcols (iterable[str]): the names of the columns to return + """ + if keyvalues: + sql = "SELECT %s FROM %s WHERE %s" % ( + ", ".join(retcols), + table, + " AND ".join("%s = ?" % (k,) for k in keyvalues), + ) + txn.execute(sql, list(keyvalues.values())) + else: + sql = "SELECT %s FROM %s" % (", ".join(retcols), table) + txn.execute(sql) + + return cls.cursor_to_dict(txn) + + @defer.inlineCallbacks + def simple_select_many_batch( + self, + table, + column, + iterable, + retcols, + keyvalues={}, + desc="simple_select_many_batch", + batch_size=100, + ): + """Executes a SELECT query on the named table, which may return zero or + more rows, returning the result as a list of dicts. + + Filters rows by if value of `column` is in `iterable`. + + Args: + table : string giving the table name + column : column name to test for inclusion against `iterable` + iterable : list + keyvalues : dict of column names and values to select the rows with + retcols : list of strings giving the names of the columns to return + """ + results = [] + + if not iterable: + return results + + # iterables can not be sliced, so convert it to a list first + it_list = list(iterable) + + chunks = [ + it_list[i : i + batch_size] for i in range(0, len(it_list), batch_size) + ] + for chunk in chunks: + rows = yield self.runInteraction( + desc, + self.simple_select_many_txn, + table, + column, + chunk, + keyvalues, + retcols, + ) + + results.extend(rows) + + return results + + @classmethod + def simple_select_many_txn(cls, txn, table, column, iterable, keyvalues, retcols): + """Executes a SELECT query on the named table, which may return zero or + more rows, returning the result as a list of dicts. + + Filters rows by if value of `column` is in `iterable`. + + Args: + txn : Transaction object + table : string giving the table name + column : column name to test for inclusion against `iterable` + iterable : list + keyvalues : dict of column names and values to select the rows with + retcols : list of strings giving the names of the columns to return + """ + if not iterable: + return [] + + clause, values = make_in_list_sql_clause(txn.database_engine, column, iterable) + clauses = [clause] + + for key, value in iteritems(keyvalues): + clauses.append("%s = ?" % (key,)) + values.append(value) + + sql = "SELECT %s FROM %s WHERE %s" % ( + ", ".join(retcols), + table, + " AND ".join(clauses), + ) + + txn.execute(sql, values) + return cls.cursor_to_dict(txn) + + def simple_update(self, table, keyvalues, updatevalues, desc): + return self.runInteraction( + desc, self.simple_update_txn, table, keyvalues, updatevalues + ) + + @staticmethod + def simple_update_txn(txn, table, keyvalues, updatevalues): + if keyvalues: + where = "WHERE %s" % " AND ".join("%s = ?" % k for k in iterkeys(keyvalues)) + else: + where = "" + + update_sql = "UPDATE %s SET %s %s" % ( + table, + ", ".join("%s = ?" % (k,) for k in updatevalues), + where, + ) + + txn.execute(update_sql, list(updatevalues.values()) + list(keyvalues.values())) + + return txn.rowcount + + def simple_update_one( + self, table, keyvalues, updatevalues, desc="simple_update_one" + ): + """Executes an UPDATE query on the named table, setting new values for + columns in a row matching the key values. + + Args: + table : string giving the table name + keyvalues : dict of column names and values to select the row with + updatevalues : dict giving column names and values to update + retcols : optional list of column names to return + + If present, retcols gives a list of column names on which to perform + a SELECT statement *before* performing the UPDATE statement. The values + of these will be returned in a dict. + + These are performed within the same transaction, allowing an atomic + get-and-set. This can be used to implement compare-and-set by putting + the update column in the 'keyvalues' dict as well. + """ + return self.runInteraction( + desc, self.simple_update_one_txn, table, keyvalues, updatevalues + ) + + @classmethod + def simple_update_one_txn(cls, txn, table, keyvalues, updatevalues): + rowcount = cls.simple_update_txn(txn, table, keyvalues, updatevalues) + + if rowcount == 0: + raise StoreError(404, "No row found (%s)" % (table,)) + if rowcount > 1: + raise StoreError(500, "More than one row matched (%s)" % (table,)) + + @staticmethod + def simple_select_one_txn(txn, table, keyvalues, retcols, allow_none=False): + select_sql = "SELECT %s FROM %s WHERE %s" % ( + ", ".join(retcols), + table, + " AND ".join("%s = ?" % (k,) for k in keyvalues), + ) + + txn.execute(select_sql, list(keyvalues.values())) + row = txn.fetchone() + + if not row: + if allow_none: + return None + raise StoreError(404, "No row found (%s)" % (table,)) + if txn.rowcount > 1: + raise StoreError(500, "More than one row matched (%s)" % (table,)) + + return dict(zip(retcols, row)) + + def simple_delete_one(self, table, keyvalues, desc="simple_delete_one"): + """Executes a DELETE query on the named table, expecting to delete a + single row. + + Args: + table : string giving the table name + keyvalues : dict of column names and values to select the row with + """ + return self.runInteraction(desc, self.simple_delete_one_txn, table, keyvalues) + + @staticmethod + def simple_delete_one_txn(txn, table, keyvalues): + """Executes a DELETE query on the named table, expecting to delete a + single row. + + Args: + table : string giving the table name + keyvalues : dict of column names and values to select the row with + """ + sql = "DELETE FROM %s WHERE %s" % ( + table, + " AND ".join("%s = ?" % (k,) for k in keyvalues), + ) + + txn.execute(sql, list(keyvalues.values())) + if txn.rowcount == 0: + raise StoreError(404, "No row found (%s)" % (table,)) + if txn.rowcount > 1: + raise StoreError(500, "More than one row matched (%s)" % (table,)) + + def simple_delete(self, table, keyvalues, desc): + return self.runInteraction(desc, self.simple_delete_txn, table, keyvalues) + + @staticmethod + def simple_delete_txn(txn, table, keyvalues): + sql = "DELETE FROM %s WHERE %s" % ( + table, + " AND ".join("%s = ?" % (k,) for k in keyvalues), + ) + + txn.execute(sql, list(keyvalues.values())) + return txn.rowcount + + def simple_delete_many(self, table, column, iterable, keyvalues, desc): + return self.runInteraction( + desc, self.simple_delete_many_txn, table, column, iterable, keyvalues + ) + + @staticmethod + def simple_delete_many_txn(txn, table, column, iterable, keyvalues): + """Executes a DELETE query on the named table. + + Filters rows by if value of `column` is in `iterable`. + + Args: + txn : Transaction object + table : string giving the table name + column : column name to test for inclusion against `iterable` + iterable : list + keyvalues : dict of column names and values to select the rows with + + Returns: + int: Number rows deleted + """ + if not iterable: + return 0 + + sql = "DELETE FROM %s" % table + + clause, values = make_in_list_sql_clause(txn.database_engine, column, iterable) + clauses = [clause] + + for key, value in iteritems(keyvalues): + clauses.append("%s = ?" % (key,)) + values.append(value) + + if clauses: + sql = "%s WHERE %s" % (sql, " AND ".join(clauses)) + txn.execute(sql, values) + + return txn.rowcount + + def get_cache_dict( + self, db_conn, table, entity_column, stream_column, max_value, limit=100000 + ): + # Fetch a mapping of room_id -> max stream position for "recent" rooms. + # It doesn't really matter how many we get, the StreamChangeCache will + # do the right thing to ensure it respects the max size of cache. + sql = ( + "SELECT %(entity)s, MAX(%(stream)s) FROM %(table)s" + " WHERE %(stream)s > ? - %(limit)s" + " GROUP BY %(entity)s" + ) % { + "table": table, + "entity": entity_column, + "stream": stream_column, + "limit": limit, + } + + sql = self.database_engine.convert_param_style(sql) + + txn = db_conn.cursor() + txn.execute(sql, (int(max_value),)) + + cache = {row[0]: int(row[1]) for row in txn} + + txn.close() + + if cache: + min_val = min(itervalues(cache)) + else: + min_val = max_value + + return cache, min_val + + def simple_select_list_paginate( + self, + table, + keyvalues, + orderby, + start, + limit, + retcols, + order_direction="ASC", + desc="simple_select_list_paginate", + ): + """ + Executes a SELECT query on the named table with start and limit, + of row numbers, which may return zero or number of rows from start to limit, + returning the result as a list of dicts. + + Args: + table (str): the table name + keyvalues (dict[str, T] | None): + column names and values to select the rows with, or None to not + apply a WHERE clause. + orderby (str): Column to order the results by. + start (int): Index to begin the query at. + limit (int): Number of results to return. + retcols (iterable[str]): the names of the columns to return + order_direction (str): Whether the results should be ordered "ASC" or "DESC". + Returns: + defer.Deferred: resolves to list[dict[str, Any]] + """ + return self.runInteraction( + desc, + self.simple_select_list_paginate_txn, + table, + keyvalues, + orderby, + start, + limit, + retcols, + order_direction=order_direction, + ) + + @classmethod + def simple_select_list_paginate_txn( + cls, + txn, + table, + keyvalues, + orderby, + start, + limit, + retcols, + order_direction="ASC", + ): + """ + Executes a SELECT query on the named table with start and limit, + of row numbers, which may return zero or number of rows from start to limit, + returning the result as a list of dicts. + + Args: + txn : Transaction object + table (str): the table name + keyvalues (dict[str, T] | None): + column names and values to select the rows with, or None to not + apply a WHERE clause. + orderby (str): Column to order the results by. + start (int): Index to begin the query at. + limit (int): Number of results to return. + retcols (iterable[str]): the names of the columns to return + order_direction (str): Whether the results should be ordered "ASC" or "DESC". + Returns: + defer.Deferred: resolves to list[dict[str, Any]] + """ + if order_direction not in ["ASC", "DESC"]: + raise ValueError("order_direction must be one of 'ASC' or 'DESC'.") + + if keyvalues: + where_clause = "WHERE " + " AND ".join("%s = ?" % (k,) for k in keyvalues) + else: + where_clause = "" + + sql = "SELECT %s FROM %s %s ORDER BY %s %s LIMIT ? OFFSET ?" % ( + ", ".join(retcols), + table, + where_clause, + orderby, + order_direction, + ) + txn.execute(sql, list(keyvalues.values()) + [limit, start]) + + return cls.cursor_to_dict(txn) + + def get_user_count_txn(self, txn): + """Get a total number of registered users in the users list. + + Args: + txn : Transaction object + Returns: + int : number of users + """ + sql_count = "SELECT COUNT(*) FROM users WHERE is_guest = 0;" + txn.execute(sql_count) + return txn.fetchone()[0] + + def simple_search_list(self, table, term, col, retcols, desc="simple_search_list"): + """Executes a SELECT query on the named table, which may return zero or + more rows, returning the result as a list of dicts. + + Args: + table (str): the table name + term (str | None): + term for searching the table matched to a column. + col (str): column to query term should be matched to + retcols (iterable[str]): the names of the columns to return + Returns: + defer.Deferred: resolves to list[dict[str, Any]] or None + """ + + return self.runInteraction( + desc, self.simple_search_list_txn, table, term, col, retcols + ) + + @classmethod + def simple_search_list_txn(cls, txn, table, term, col, retcols): + """Executes a SELECT query on the named table, which may return zero or + more rows, returning the result as a list of dicts. + + Args: + txn : Transaction object + table (str): the table name + term (str | None): + term for searching the table matched to a column. + col (str): column to query term should be matched to + retcols (iterable[str]): the names of the columns to return + Returns: + defer.Deferred: resolves to list[dict[str, Any]] or None + """ + if term: + sql = "SELECT %s FROM %s WHERE %s LIKE ?" % (", ".join(retcols), table, col) + termvalues = ["%%" + term + "%%"] + txn.execute(sql, termvalues) + else: + return 0 + + return cls.cursor_to_dict(txn) + + +def make_in_list_sql_clause( + database_engine, column: str, iterable: Iterable +) -> Tuple[str, Iterable]: + """Returns an SQL clause that checks the given column is in the iterable. + + On SQLite this expands to `column IN (?, ?, ...)`, whereas on Postgres + it expands to `column = ANY(?)`. While both DBs support the `IN` form, + using the `ANY` form on postgres means that it views queries with + different length iterables as the same, helping the query stats. + + Args: + database_engine + column: Name of the column + iterable: The values to check the column against. + + Returns: + A tuple of SQL query and the args + """ + + if database_engine.supports_using_any_list: + # This should hopefully be faster, but also makes postgres query + # stats easier to understand. + return "%s = ANY(?)" % (column,), [list(iterable)] + else: + return "%s IN (%s)" % (column, ",".join("?" for _ in iterable)), list(iterable) diff --git a/tests/handlers/test_stats.py b/tests/handlers/test_stats.py index 380fd0d107..7f7962c3dd 100644 --- a/tests/handlers/test_stats.py +++ b/tests/handlers/test_stats.py @@ -45,13 +45,13 @@ class StatsRoomTests(unittest.HomeserverTestCase): self.store._all_done = False self.get_success( - self.store.simple_insert( + self.store.db.simple_insert( "background_updates", {"update_name": "populate_stats_prepare", "progress_json": "{}"}, ) ) self.get_success( - self.store.simple_insert( + self.store.db.simple_insert( "background_updates", { "update_name": "populate_stats_process_rooms", @@ -61,7 +61,7 @@ class StatsRoomTests(unittest.HomeserverTestCase): ) ) self.get_success( - self.store.simple_insert( + self.store.db.simple_insert( "background_updates", { "update_name": "populate_stats_process_users", @@ -71,7 +71,7 @@ class StatsRoomTests(unittest.HomeserverTestCase): ) ) self.get_success( - self.store.simple_insert( + self.store.db.simple_insert( "background_updates", { "update_name": "populate_stats_cleanup", @@ -82,7 +82,7 @@ class StatsRoomTests(unittest.HomeserverTestCase): ) def get_all_room_state(self): - return self.store.simple_select_list( + return self.store.db.simple_select_list( "room_stats_state", None, retcols=("name", "topic", "canonical_alias") ) @@ -96,7 +96,7 @@ class StatsRoomTests(unittest.HomeserverTestCase): end_ts = self.store.quantise_stats_time(self.reactor.seconds() * 1000) return self.get_success( - self.store.simple_select_one( + self.store.db.simple_select_one( table + "_historical", {id_col: stat_id, end_ts: end_ts}, cols, @@ -180,7 +180,7 @@ class StatsRoomTests(unittest.HomeserverTestCase): self.handler.stats_enabled = True self.store._all_done = False self.get_success( - self.store.simple_update_one( + self.store.db.simple_update_one( table="stats_incremental_position", keyvalues={}, updatevalues={"stream_id": 0}, @@ -188,7 +188,7 @@ class StatsRoomTests(unittest.HomeserverTestCase): ) self.get_success( - self.store.simple_insert( + self.store.db.simple_insert( "background_updates", {"update_name": "populate_stats_prepare", "progress_json": "{}"}, ) @@ -205,13 +205,13 @@ class StatsRoomTests(unittest.HomeserverTestCase): # Now do the initial ingestion. self.get_success( - self.store.simple_insert( + self.store.db.simple_insert( "background_updates", {"update_name": "populate_stats_process_rooms", "progress_json": "{}"}, ) ) self.get_success( - self.store.simple_insert( + self.store.db.simple_insert( "background_updates", { "update_name": "populate_stats_cleanup", @@ -656,12 +656,12 @@ class StatsRoomTests(unittest.HomeserverTestCase): self.store._all_done = False self.get_success( - self.store.simple_delete( + self.store.db.simple_delete( "room_stats_current", {"1": 1}, "test_delete_stats" ) ) self.get_success( - self.store.simple_delete( + self.store.db.simple_delete( "user_stats_current", {"1": 1}, "test_delete_stats" ) ) @@ -675,7 +675,7 @@ class StatsRoomTests(unittest.HomeserverTestCase): self.store._all_done = False self.get_success( - self.store.simple_insert( + self.store.db.simple_insert( "background_updates", { "update_name": "populate_stats_process_rooms", @@ -685,7 +685,7 @@ class StatsRoomTests(unittest.HomeserverTestCase): ) ) self.get_success( - self.store.simple_insert( + self.store.db.simple_insert( "background_updates", { "update_name": "populate_stats_process_users", @@ -695,7 +695,7 @@ class StatsRoomTests(unittest.HomeserverTestCase): ) ) self.get_success( - self.store.simple_insert( + self.store.db.simple_insert( "background_updates", { "update_name": "populate_stats_cleanup", diff --git a/tests/handlers/test_user_directory.py b/tests/handlers/test_user_directory.py index d5b1c5b4ac..bc9d441541 100644 --- a/tests/handlers/test_user_directory.py +++ b/tests/handlers/test_user_directory.py @@ -158,7 +158,7 @@ class UserDirectoryTestCase(unittest.HomeserverTestCase): def get_users_in_public_rooms(self): r = self.get_success( - self.store.simple_select_list( + self.store.db.simple_select_list( "users_in_public_rooms", None, ("user_id", "room_id") ) ) @@ -169,7 +169,7 @@ class UserDirectoryTestCase(unittest.HomeserverTestCase): def get_users_who_share_private_rooms(self): return self.get_success( - self.store.simple_select_list( + self.store.db.simple_select_list( "users_who_share_private_rooms", None, ["user_id", "other_user_id", "room_id"], @@ -184,7 +184,7 @@ class UserDirectoryTestCase(unittest.HomeserverTestCase): self.store._all_done = False self.get_success( - self.store.simple_insert( + self.store.db.simple_insert( "background_updates", { "update_name": "populate_user_directory_createtables", @@ -193,7 +193,7 @@ class UserDirectoryTestCase(unittest.HomeserverTestCase): ) ) self.get_success( - self.store.simple_insert( + self.store.db.simple_insert( "background_updates", { "update_name": "populate_user_directory_process_rooms", @@ -203,7 +203,7 @@ class UserDirectoryTestCase(unittest.HomeserverTestCase): ) ) self.get_success( - self.store.simple_insert( + self.store.db.simple_insert( "background_updates", { "update_name": "populate_user_directory_process_users", @@ -213,7 +213,7 @@ class UserDirectoryTestCase(unittest.HomeserverTestCase): ) ) self.get_success( - self.store.simple_insert( + self.store.db.simple_insert( "background_updates", { "update_name": "populate_user_directory_cleanup", diff --git a/tests/rest/admin/test_admin.py b/tests/rest/admin/test_admin.py index 124ce0768a..0ed2594381 100644 --- a/tests/rest/admin/test_admin.py +++ b/tests/rest/admin/test_admin.py @@ -632,7 +632,7 @@ class PurgeRoomTestCase(unittest.HomeserverTestCase): "state_groups_state", ): count = self.get_success( - self.store.simple_select_one_onecol( + self.store.db.simple_select_one_onecol( table=table, keyvalues={"room_id": room_id}, retcol="COUNT(*)", diff --git a/tests/storage/test__base.py b/tests/storage/test__base.py index 7b7434a468..d491ea2924 100644 --- a/tests/storage/test__base.py +++ b/tests/storage/test__base.py @@ -323,7 +323,7 @@ class UpsertManyTests(unittest.HomeserverTestCase): self.table_name = "table_" + hs.get_secrets().token_hex(6) self.get_success( - self.storage.runInteraction( + self.storage.db.runInteraction( "create", lambda x, *a: x.execute(*a), "CREATE TABLE %s (id INTEGER, username TEXT, value TEXT)" @@ -331,7 +331,7 @@ class UpsertManyTests(unittest.HomeserverTestCase): ) ) self.get_success( - self.storage.runInteraction( + self.storage.db.runInteraction( "index", lambda x, *a: x.execute(*a), "CREATE UNIQUE INDEX %sindex ON %s(id, username)" @@ -354,9 +354,9 @@ class UpsertManyTests(unittest.HomeserverTestCase): value_values = [["hello"], ["there"]] self.get_success( - self.storage.runInteraction( + self.storage.db.runInteraction( "test", - self.storage.simple_upsert_many_txn, + self.storage.db.simple_upsert_many_txn, self.table_name, key_names, key_values, @@ -367,7 +367,7 @@ class UpsertManyTests(unittest.HomeserverTestCase): # Check results are what we expect res = self.get_success( - self.storage.simple_select_list( + self.storage.db.simple_select_list( self.table_name, None, ["id, username, value"] ) ) @@ -381,9 +381,9 @@ class UpsertManyTests(unittest.HomeserverTestCase): value_values = [["bleb"]] self.get_success( - self.storage.runInteraction( + self.storage.db.runInteraction( "test", - self.storage.simple_upsert_many_txn, + self.storage.db.simple_upsert_many_txn, self.table_name, key_names, key_values, @@ -394,7 +394,7 @@ class UpsertManyTests(unittest.HomeserverTestCase): # Check results are what we expect res = self.get_success( - self.storage.simple_select_list( + self.storage.db.simple_select_list( self.table_name, None, ["id, username, value"] ) ) diff --git a/tests/storage/test_background_update.py b/tests/storage/test_background_update.py index 9fabe3fbc0..e360297df9 100644 --- a/tests/storage/test_background_update.py +++ b/tests/storage/test_background_update.py @@ -37,7 +37,7 @@ class BackgroundUpdateTestCase(unittest.TestCase): def update(progress, count): self.clock.advance_time_msec(count * duration_ms) progress = {"my_key": progress["my_key"] + 1} - yield self.store.runInteraction( + yield self.store.db.runInteraction( "update_progress", self.store._background_update_progress_txn, "test_update", diff --git a/tests/storage/test_base.py b/tests/storage/test_base.py index de5e4a5fce..7915d48a9e 100644 --- a/tests/storage/test_base.py +++ b/tests/storage/test_base.py @@ -65,7 +65,7 @@ class SQLBaseStoreTestCase(unittest.TestCase): def test_insert_1col(self): self.mock_txn.rowcount = 1 - yield self.datastore.simple_insert( + yield self.datastore.db.simple_insert( table="tablename", values={"columname": "Value"} ) @@ -77,7 +77,7 @@ class SQLBaseStoreTestCase(unittest.TestCase): def test_insert_3cols(self): self.mock_txn.rowcount = 1 - yield self.datastore.simple_insert( + yield self.datastore.db.simple_insert( table="tablename", # Use OrderedDict() so we can assert on the SQL generated values=OrderedDict([("colA", 1), ("colB", 2), ("colC", 3)]), @@ -92,7 +92,7 @@ class SQLBaseStoreTestCase(unittest.TestCase): self.mock_txn.rowcount = 1 self.mock_txn.__iter__ = Mock(return_value=iter([("Value",)])) - value = yield self.datastore.simple_select_one_onecol( + value = yield self.datastore.db.simple_select_one_onecol( table="tablename", keyvalues={"keycol": "TheKey"}, retcol="retcol" ) @@ -106,7 +106,7 @@ class SQLBaseStoreTestCase(unittest.TestCase): self.mock_txn.rowcount = 1 self.mock_txn.fetchone.return_value = (1, 2, 3) - ret = yield self.datastore.simple_select_one( + ret = yield self.datastore.db.simple_select_one( table="tablename", keyvalues={"keycol": "TheKey"}, retcols=["colA", "colB", "colC"], @@ -122,7 +122,7 @@ class SQLBaseStoreTestCase(unittest.TestCase): self.mock_txn.rowcount = 0 self.mock_txn.fetchone.return_value = None - ret = yield self.datastore.simple_select_one( + ret = yield self.datastore.db.simple_select_one( table="tablename", keyvalues={"keycol": "Not here"}, retcols=["colA"], @@ -137,7 +137,7 @@ class SQLBaseStoreTestCase(unittest.TestCase): self.mock_txn.__iter__ = Mock(return_value=iter([(1,), (2,), (3,)])) self.mock_txn.description = (("colA", None, None, None, None, None, None),) - ret = yield self.datastore.simple_select_list( + ret = yield self.datastore.db.simple_select_list( table="tablename", keyvalues={"keycol": "A set"}, retcols=["colA"] ) @@ -150,7 +150,7 @@ class SQLBaseStoreTestCase(unittest.TestCase): def test_update_one_1col(self): self.mock_txn.rowcount = 1 - yield self.datastore.simple_update_one( + yield self.datastore.db.simple_update_one( table="tablename", keyvalues={"keycol": "TheKey"}, updatevalues={"columnname": "New Value"}, @@ -165,7 +165,7 @@ class SQLBaseStoreTestCase(unittest.TestCase): def test_update_one_4cols(self): self.mock_txn.rowcount = 1 - yield self.datastore.simple_update_one( + yield self.datastore.db.simple_update_one( table="tablename", keyvalues=OrderedDict([("colA", 1), ("colB", 2)]), updatevalues=OrderedDict([("colC", 3), ("colD", 4)]), @@ -180,7 +180,7 @@ class SQLBaseStoreTestCase(unittest.TestCase): def test_delete_one(self): self.mock_txn.rowcount = 1 - yield self.datastore.simple_delete_one( + yield self.datastore.db.simple_delete_one( table="tablename", keyvalues={"keycol": "Go away"} ) diff --git a/tests/storage/test_cleanup_extrems.py b/tests/storage/test_cleanup_extrems.py index 69dcaa63d5..e454bbff29 100644 --- a/tests/storage/test_cleanup_extrems.py +++ b/tests/storage/test_cleanup_extrems.py @@ -62,7 +62,9 @@ class CleanupExtremBackgroundUpdateStoreTestCase(HomeserverTestCase): prepare_database.executescript(txn, schema_path) self.get_success( - self.store.runInteraction("test_delete_forward_extremities", run_delta_file) + self.store.db.runInteraction( + "test_delete_forward_extremities", run_delta_file + ) ) # Ugh, have to reset this flag diff --git a/tests/storage/test_client_ips.py b/tests/storage/test_client_ips.py index 25bdd2c163..c4f838907c 100644 --- a/tests/storage/test_client_ips.py +++ b/tests/storage/test_client_ips.py @@ -81,7 +81,7 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): self.pump(0) result = self.get_success( - self.store.simple_select_list( + self.store.db.simple_select_list( table="user_ips", keyvalues={"user_id": user_id}, retcols=["access_token", "ip", "user_agent", "device_id", "last_seen"], @@ -112,7 +112,7 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): self.pump(0) result = self.get_success( - self.store.simple_select_list( + self.store.db.simple_select_list( table="user_ips", keyvalues={"user_id": user_id}, retcols=["access_token", "ip", "user_agent", "device_id", "last_seen"], @@ -218,7 +218,7 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): # But clear the associated entry in devices table self.get_success( - self.store.simple_update( + self.store.db.simple_update( table="devices", keyvalues={"user_id": user_id, "device_id": "device_id"}, updatevalues={"last_seen": None, "ip": None, "user_agent": None}, @@ -245,7 +245,7 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): # Register the background update to run again. self.get_success( - self.store.simple_insert( + self.store.db.simple_insert( table="background_updates", values={ "update_name": "devices_last_seen", @@ -297,7 +297,7 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): # We should see that in the DB result = self.get_success( - self.store.simple_select_list( + self.store.db.simple_select_list( table="user_ips", keyvalues={"user_id": user_id}, retcols=["access_token", "ip", "user_agent", "device_id", "last_seen"], @@ -323,7 +323,7 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): # We should get no results. result = self.get_success( - self.store.simple_select_list( + self.store.db.simple_select_list( table="user_ips", keyvalues={"user_id": user_id}, retcols=["access_token", "ip", "user_agent", "device_id", "last_seen"], diff --git a/tests/storage/test_event_federation.py b/tests/storage/test_event_federation.py index 2fe50377f8..eadfb90a22 100644 --- a/tests/storage/test_event_federation.py +++ b/tests/storage/test_event_federation.py @@ -61,7 +61,7 @@ class EventFederationWorkerStoreTestCase(tests.unittest.TestCase): ) for i in range(0, 11): - yield self.store.runInteraction("insert", insert_event, i) + yield self.store.db.runInteraction("insert", insert_event, i) # this should get the last five and five others r = yield self.store.get_prev_events_for_room(room_id) @@ -93,9 +93,9 @@ class EventFederationWorkerStoreTestCase(tests.unittest.TestCase): ) for i in range(0, 20): - yield self.store.runInteraction("insert", insert_event, i, room1) - yield self.store.runInteraction("insert", insert_event, i, room2) - yield self.store.runInteraction("insert", insert_event, i, room3) + yield self.store.db.runInteraction("insert", insert_event, i, room1) + yield self.store.db.runInteraction("insert", insert_event, i, room2) + yield self.store.db.runInteraction("insert", insert_event, i, room3) # Test simple case r = yield self.store.get_rooms_with_many_extremities(5, 5, []) diff --git a/tests/storage/test_event_push_actions.py b/tests/storage/test_event_push_actions.py index 2337a1ae46..d4bcf1821e 100644 --- a/tests/storage/test_event_push_actions.py +++ b/tests/storage/test_event_push_actions.py @@ -55,7 +55,7 @@ class EventPushActionsStoreTestCase(tests.unittest.TestCase): @defer.inlineCallbacks def _assert_counts(noitf_count, highlight_count): - counts = yield self.store.runInteraction( + counts = yield self.store.db.runInteraction( "", self.store._get_unread_counts_by_pos_txn, room_id, user_id, 0 ) self.assertEquals( @@ -74,7 +74,7 @@ class EventPushActionsStoreTestCase(tests.unittest.TestCase): yield self.store.add_push_actions_to_staging( event.event_id, {user_id: action} ) - yield self.store.runInteraction( + yield self.store.db.runInteraction( "", self.store._set_push_actions_for_event_and_users_txn, [(event, None)], @@ -82,12 +82,12 @@ class EventPushActionsStoreTestCase(tests.unittest.TestCase): ) def _rotate(stream): - return self.store.runInteraction( + return self.store.db.runInteraction( "", self.store._rotate_notifs_before_txn, stream ) def _mark_read(stream, depth): - return self.store.runInteraction( + return self.store.db.runInteraction( "", self.store._remove_old_push_actions_before_txn, room_id, @@ -116,7 +116,7 @@ class EventPushActionsStoreTestCase(tests.unittest.TestCase): yield _inject_actions(6, PlAIN_NOTIF) yield _rotate(7) - yield self.store.simple_delete( + yield self.store.db.simple_delete( table="event_push_actions", keyvalues={"1": 1}, desc="" ) @@ -135,7 +135,7 @@ class EventPushActionsStoreTestCase(tests.unittest.TestCase): @defer.inlineCallbacks def test_find_first_stream_ordering_after_ts(self): def add_event(so, ts): - return self.store.simple_insert( + return self.store.db.simple_insert( "events", { "stream_ordering": so, diff --git a/tests/storage/test_monthly_active_users.py b/tests/storage/test_monthly_active_users.py index 90a63dc477..3c78faab45 100644 --- a/tests/storage/test_monthly_active_users.py +++ b/tests/storage/test_monthly_active_users.py @@ -65,7 +65,7 @@ class MonthlyActiveUsersTestCase(unittest.HomeserverTestCase): self.store.user_add_threepid(user1, "email", user1_email, now, now) self.store.user_add_threepid(user2, "email", user2_email, now, now) - self.store.runInteraction( + self.store.db.runInteraction( "initialise", self.store._initialise_reserved_users, threepids ) self.pump() @@ -183,7 +183,7 @@ class MonthlyActiveUsersTestCase(unittest.HomeserverTestCase): ) self.hs.config.mau_limits_reserved_threepids = threepids - self.store.runInteraction( + self.store.db.runInteraction( "initialise", self.store._initialise_reserved_users, threepids ) count = self.store.get_monthly_active_count() @@ -244,7 +244,7 @@ class MonthlyActiveUsersTestCase(unittest.HomeserverTestCase): {"medium": "email", "address": user2_email}, ] self.hs.config.mau_limits_reserved_threepids = threepids - self.store.runInteraction( + self.store.db.runInteraction( "initialise", self.store._initialise_reserved_users, threepids ) diff --git a/tests/storage/test_redaction.py b/tests/storage/test_redaction.py index 4930b6777e..dc45173355 100644 --- a/tests/storage/test_redaction.py +++ b/tests/storage/test_redaction.py @@ -338,7 +338,7 @@ class RedactionTestCase(unittest.HomeserverTestCase): ) event_json = self.get_success( - self.store.simple_select_one_onecol( + self.store.db.simple_select_one_onecol( table="event_json", keyvalues={"event_id": msg_event.event_id}, retcol="json", @@ -356,7 +356,7 @@ class RedactionTestCase(unittest.HomeserverTestCase): self.reactor.advance(60 * 60 * 2) event_json = self.get_success( - self.store.simple_select_one_onecol( + self.store.db.simple_select_one_onecol( table="event_json", keyvalues={"event_id": msg_event.event_id}, retcol="json", diff --git a/tests/storage/test_roommember.py b/tests/storage/test_roommember.py index d389cf578f..5f957680a2 100644 --- a/tests/storage/test_roommember.py +++ b/tests/storage/test_roommember.py @@ -132,7 +132,7 @@ class CurrentStateMembershipUpdateTestCase(unittest.HomeserverTestCase): # Register the background update to run again. self.get_success( - self.store.simple_insert( + self.store.db.simple_insert( table="background_updates", values={ "update_name": "current_state_events_membership", diff --git a/tests/unittest.py b/tests/unittest.py index 295573bc46..fc856a574a 100644 --- a/tests/unittest.py +++ b/tests/unittest.py @@ -544,7 +544,7 @@ class HomeserverTestCase(TestCase): Add the given event as an extremity to the room. """ self.get_success( - self.hs.get_datastore().simple_insert( + self.hs.get_datastore().db.simple_insert( table="event_forward_extremities", values={"room_id": room_id, "event_id": event_id}, desc="test_add_extremity", -- cgit 1.4.1 From 4a33a6dd19590b8e6626a5af5a69507dc11236f8 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 4 Dec 2019 15:09:36 +0000 Subject: Move background update handling out of store --- synapse/app/homeserver.py | 2 +- synapse/rest/media/v1/preview_url_resource.py | 2 +- synapse/storage/background_updates.py | 15 +++---- synapse/storage/data_stores/main/client_ips.py | 36 ++++++++-------- synapse/storage/data_stores/main/deviceinbox.py | 9 ++-- synapse/storage/data_stores/main/devices.py | 15 ++++--- .../storage/data_stores/main/event_federation.py | 6 +-- .../storage/data_stores/main/event_push_actions.py | 4 +- synapse/storage/data_stores/main/events.py | 6 +-- .../storage/data_stores/main/events_bg_updates.py | 49 +++++++++++---------- .../storage/data_stores/main/media_repository.py | 6 +-- synapse/storage/data_stores/main/registration.py | 21 ++++----- synapse/storage/data_stores/main/room.py | 9 ++-- synapse/storage/data_stores/main/roommember.py | 27 +++++++----- synapse/storage/data_stores/main/search.py | 31 ++++++++------ synapse/storage/data_stores/main/state.py | 19 ++++---- synapse/storage/data_stores/main/stats.py | 20 ++++----- synapse/storage/data_stores/main/user_directory.py | 33 ++++++++------ synapse/storage/database.py | 3 ++ synmark/__init__.py | 6 +-- tests/handlers/test_stats.py | 50 +++++++++++++++------- tests/handlers/test_user_directory.py | 18 +++++--- tests/storage/test_background_update.py | 26 +++++++---- tests/storage/test_cleanup_extrems.py | 14 ++++-- tests/storage/test_client_ips.py | 26 ++++++++--- tests/storage/test_roommember.py | 18 +++++--- tests/unittest.py | 10 +++-- 27 files changed, 281 insertions(+), 200 deletions(-) (limited to 'tests') diff --git a/synapse/app/homeserver.py b/synapse/app/homeserver.py index 267aebaae9..9f81a857ab 100644 --- a/synapse/app/homeserver.py +++ b/synapse/app/homeserver.py @@ -436,7 +436,7 @@ def setup(config_options): _base.start(hs, config.listeners) hs.get_pusherpool().start() - hs.get_datastore().start_doing_background_updates() + hs.get_datastore().db.updates.start_doing_background_updates() except Exception: # Print the exception and bail out. print("Error during startup:", file=sys.stderr) diff --git a/synapse/rest/media/v1/preview_url_resource.py b/synapse/rest/media/v1/preview_url_resource.py index fb0d02aa83..6b978be876 100644 --- a/synapse/rest/media/v1/preview_url_resource.py +++ b/synapse/rest/media/v1/preview_url_resource.py @@ -402,7 +402,7 @@ class PreviewUrlResource(DirectServeResource): logger.info("Running url preview cache expiry") - if not (yield self.store.has_completed_background_updates()): + if not (yield self.store.db.updates.has_completed_background_updates()): logger.info("Still running DB updates; skipping expiry") return diff --git a/synapse/storage/background_updates.py b/synapse/storage/background_updates.py index dfca94b0e0..a9a13a2658 100644 --- a/synapse/storage/background_updates.py +++ b/synapse/storage/background_updates.py @@ -22,7 +22,6 @@ from twisted.internet import defer from synapse.metrics.background_process_metrics import run_as_background_process from . import engines -from ._base import SQLBaseStore logger = logging.getLogger(__name__) @@ -74,7 +73,7 @@ class BackgroundUpdatePerformance(object): return float(self.total_item_count) / float(self.total_duration_ms) -class BackgroundUpdateStore(SQLBaseStore): +class BackgroundUpdater(object): """ Background updates are updates to the database that run in the background. Each update processes a batch of data at once. We attempt to limit the impact of each update by monitoring how long each batch takes to @@ -86,8 +85,10 @@ class BackgroundUpdateStore(SQLBaseStore): BACKGROUND_UPDATE_INTERVAL_MS = 1000 BACKGROUND_UPDATE_DURATION_MS = 100 - def __init__(self, db_conn, hs): - super(BackgroundUpdateStore, self).__init__(db_conn, hs) + def __init__(self, hs, database): + self._clock = hs.get_clock() + self.db = database + self._background_update_performance = {} self._background_update_queue = [] self._background_update_handlers = {} @@ -101,9 +102,7 @@ class BackgroundUpdateStore(SQLBaseStore): logger.info("Starting background schema updates") while True: if sleep: - yield self.hs.get_clock().sleep( - self.BACKGROUND_UPDATE_INTERVAL_MS / 1000.0 - ) + yield self._clock.sleep(self.BACKGROUND_UPDATE_INTERVAL_MS / 1000.0) try: result = yield self.do_next_background_update( @@ -380,7 +379,7 @@ class BackgroundUpdateStore(SQLBaseStore): logger.debug("[SQL] %s", sql) c.execute(sql) - if isinstance(self.database_engine, engines.PostgresEngine): + if isinstance(self.db.database_engine, engines.PostgresEngine): runner = create_index_psql elif psql_only: runner = None diff --git a/synapse/storage/data_stores/main/client_ips.py b/synapse/storage/data_stores/main/client_ips.py index 6f2a720b97..7b470a58f1 100644 --- a/synapse/storage/data_stores/main/client_ips.py +++ b/synapse/storage/data_stores/main/client_ips.py @@ -20,7 +20,7 @@ from six import iteritems from twisted.internet import defer from synapse.metrics.background_process_metrics import wrap_as_background_process -from synapse.storage import background_updates +from synapse.storage._base import SQLBaseStore from synapse.util.caches import CACHE_SIZE_FACTOR from synapse.util.caches.descriptors import Cache @@ -32,41 +32,41 @@ logger = logging.getLogger(__name__) LAST_SEEN_GRANULARITY = 120 * 1000 -class ClientIpBackgroundUpdateStore(background_updates.BackgroundUpdateStore): +class ClientIpBackgroundUpdateStore(SQLBaseStore): def __init__(self, db_conn, hs): super(ClientIpBackgroundUpdateStore, self).__init__(db_conn, hs) - self.register_background_index_update( + self.db.updates.register_background_index_update( "user_ips_device_index", index_name="user_ips_device_id", table="user_ips", columns=["user_id", "device_id", "last_seen"], ) - self.register_background_index_update( + self.db.updates.register_background_index_update( "user_ips_last_seen_index", index_name="user_ips_last_seen", table="user_ips", columns=["user_id", "last_seen"], ) - self.register_background_index_update( + self.db.updates.register_background_index_update( "user_ips_last_seen_only_index", index_name="user_ips_last_seen_only", table="user_ips", columns=["last_seen"], ) - self.register_background_update_handler( + self.db.updates.register_background_update_handler( "user_ips_analyze", self._analyze_user_ip ) - self.register_background_update_handler( + self.db.updates.register_background_update_handler( "user_ips_remove_dupes", self._remove_user_ip_dupes ) # Register a unique index - self.register_background_index_update( + self.db.updates.register_background_index_update( "user_ips_device_unique_index", index_name="user_ips_user_token_ip_unique_index", table="user_ips", @@ -75,12 +75,12 @@ class ClientIpBackgroundUpdateStore(background_updates.BackgroundUpdateStore): ) # Drop the old non-unique index - self.register_background_update_handler( + self.db.updates.register_background_update_handler( "user_ips_drop_nonunique_index", self._remove_user_ip_nonunique ) # Update the last seen info in devices. - self.register_background_update_handler( + self.db.updates.register_background_update_handler( "devices_last_seen", self._devices_last_seen_update ) @@ -92,7 +92,7 @@ class ClientIpBackgroundUpdateStore(background_updates.BackgroundUpdateStore): txn.close() yield self.db.runWithConnection(f) - yield self._end_background_update("user_ips_drop_nonunique_index") + yield self.db.updates._end_background_update("user_ips_drop_nonunique_index") return 1 @defer.inlineCallbacks @@ -108,7 +108,7 @@ class ClientIpBackgroundUpdateStore(background_updates.BackgroundUpdateStore): yield self.db.runInteraction("user_ips_analyze", user_ips_analyze) - yield self._end_background_update("user_ips_analyze") + yield self.db.updates._end_background_update("user_ips_analyze") return 1 @@ -271,14 +271,14 @@ class ClientIpBackgroundUpdateStore(background_updates.BackgroundUpdateStore): (user_id, access_token, ip, device_id, user_agent, last_seen), ) - self._background_update_progress_txn( + self.db.updates._background_update_progress_txn( txn, "user_ips_remove_dupes", {"last_seen": end_last_seen} ) yield self.db.runInteraction("user_ips_dups_remove", remove) if last: - yield self._end_background_update("user_ips_remove_dupes") + yield self.db.updates._end_background_update("user_ips_remove_dupes") return batch_size @@ -344,7 +344,7 @@ class ClientIpBackgroundUpdateStore(background_updates.BackgroundUpdateStore): txn.execute_batch(sql, rows) _, _, _, user_id, device_id = rows[-1] - self._background_update_progress_txn( + self.db.updates._background_update_progress_txn( txn, "devices_last_seen", {"last_user_id": user_id, "last_device_id": device_id}, @@ -357,7 +357,7 @@ class ClientIpBackgroundUpdateStore(background_updates.BackgroundUpdateStore): ) if not updated: - yield self._end_background_update("devices_last_seen") + yield self.db.updates._end_background_update("devices_last_seen") return updated @@ -546,7 +546,9 @@ class ClientIpStore(ClientIpBackgroundUpdateStore): # Nothing to do return - if not await self.has_completed_background_update("devices_last_seen"): + if not await self.db.updates.has_completed_background_update( + "devices_last_seen" + ): # Only start pruning if we have finished populating the devices # last seen info. return diff --git a/synapse/storage/data_stores/main/deviceinbox.py b/synapse/storage/data_stores/main/deviceinbox.py index 440793ad49..3c9f09301a 100644 --- a/synapse/storage/data_stores/main/deviceinbox.py +++ b/synapse/storage/data_stores/main/deviceinbox.py @@ -21,7 +21,6 @@ from twisted.internet import defer from synapse.logging.opentracing import log_kv, set_tag, trace from synapse.storage._base import SQLBaseStore, make_in_list_sql_clause -from synapse.storage.background_updates import BackgroundUpdateStore from synapse.util.caches.expiringcache import ExpiringCache logger = logging.getLogger(__name__) @@ -208,20 +207,20 @@ class DeviceInboxWorkerStore(SQLBaseStore): ) -class DeviceInboxBackgroundUpdateStore(BackgroundUpdateStore): +class DeviceInboxBackgroundUpdateStore(SQLBaseStore): DEVICE_INBOX_STREAM_ID = "device_inbox_stream_drop" def __init__(self, db_conn, hs): super(DeviceInboxBackgroundUpdateStore, self).__init__(db_conn, hs) - self.register_background_index_update( + self.db.updates.register_background_index_update( "device_inbox_stream_index", index_name="device_inbox_stream_id_user_id", table="device_inbox", columns=["stream_id", "user_id"], ) - self.register_background_update_handler( + self.db.updates.register_background_update_handler( self.DEVICE_INBOX_STREAM_ID, self._background_drop_index_device_inbox ) @@ -234,7 +233,7 @@ class DeviceInboxBackgroundUpdateStore(BackgroundUpdateStore): yield self.db.runWithConnection(reindex_txn) - yield self._end_background_update(self.DEVICE_INBOX_STREAM_ID) + yield self.db.updates._end_background_update(self.DEVICE_INBOX_STREAM_ID) return 1 diff --git a/synapse/storage/data_stores/main/devices.py b/synapse/storage/data_stores/main/devices.py index d98511ddd4..91ddaf137e 100644 --- a/synapse/storage/data_stores/main/devices.py +++ b/synapse/storage/data_stores/main/devices.py @@ -31,7 +31,6 @@ from synapse.logging.opentracing import ( ) from synapse.metrics.background_process_metrics import run_as_background_process from synapse.storage._base import SQLBaseStore, db_to_json, make_in_list_sql_clause -from synapse.storage.background_updates import BackgroundUpdateStore from synapse.types import get_verify_key_from_cross_signing_key from synapse.util import batch_iter from synapse.util.caches.descriptors import ( @@ -642,11 +641,11 @@ class DeviceWorkerStore(SQLBaseStore): return results -class DeviceBackgroundUpdateStore(BackgroundUpdateStore): +class DeviceBackgroundUpdateStore(SQLBaseStore): def __init__(self, db_conn, hs): super(DeviceBackgroundUpdateStore, self).__init__(db_conn, hs) - self.register_background_index_update( + self.db.updates.register_background_index_update( "device_lists_stream_idx", index_name="device_lists_stream_user_id", table="device_lists_stream", @@ -654,7 +653,7 @@ class DeviceBackgroundUpdateStore(BackgroundUpdateStore): ) # create a unique index on device_lists_remote_cache - self.register_background_index_update( + self.db.updates.register_background_index_update( "device_lists_remote_cache_unique_idx", index_name="device_lists_remote_cache_unique_id", table="device_lists_remote_cache", @@ -663,7 +662,7 @@ class DeviceBackgroundUpdateStore(BackgroundUpdateStore): ) # And one on device_lists_remote_extremeties - self.register_background_index_update( + self.db.updates.register_background_index_update( "device_lists_remote_extremeties_unique_idx", index_name="device_lists_remote_extremeties_unique_idx", table="device_lists_remote_extremeties", @@ -672,7 +671,7 @@ class DeviceBackgroundUpdateStore(BackgroundUpdateStore): ) # once they complete, we can remove the old non-unique indexes. - self.register_background_update_handler( + self.db.updates.register_background_update_handler( DROP_DEVICE_LIST_STREAMS_NON_UNIQUE_INDEXES, self._drop_device_list_streams_non_unique_indexes, ) @@ -686,7 +685,9 @@ class DeviceBackgroundUpdateStore(BackgroundUpdateStore): txn.close() yield self.db.runWithConnection(f) - yield self._end_background_update(DROP_DEVICE_LIST_STREAMS_NON_UNIQUE_INDEXES) + yield self.db.updates._end_background_update( + DROP_DEVICE_LIST_STREAMS_NON_UNIQUE_INDEXES + ) return 1 diff --git a/synapse/storage/data_stores/main/event_federation.py b/synapse/storage/data_stores/main/event_federation.py index 77e4353b59..31d2e8eb28 100644 --- a/synapse/storage/data_stores/main/event_federation.py +++ b/synapse/storage/data_stores/main/event_federation.py @@ -494,7 +494,7 @@ class EventFederationStore(EventFederationWorkerStore): def __init__(self, db_conn, hs): super(EventFederationStore, self).__init__(db_conn, hs) - self.register_background_update_handler( + self.db.updates.register_background_update_handler( self.EVENT_AUTH_STATE_ONLY, self._background_delete_non_state_event_auth ) @@ -654,7 +654,7 @@ class EventFederationStore(EventFederationWorkerStore): "max_stream_id_exclusive": min_stream_id, } - self._background_update_progress_txn( + self.db.updates._background_update_progress_txn( txn, self.EVENT_AUTH_STATE_ONLY, new_progress ) @@ -665,6 +665,6 @@ class EventFederationStore(EventFederationWorkerStore): ) if not result: - yield self._end_background_update(self.EVENT_AUTH_STATE_ONLY) + yield self.db.updates._end_background_update(self.EVENT_AUTH_STATE_ONLY) return batch_size diff --git a/synapse/storage/data_stores/main/event_push_actions.py b/synapse/storage/data_stores/main/event_push_actions.py index 725d0881dc..eec054cd48 100644 --- a/synapse/storage/data_stores/main/event_push_actions.py +++ b/synapse/storage/data_stores/main/event_push_actions.py @@ -614,14 +614,14 @@ class EventPushActionsStore(EventPushActionsWorkerStore): def __init__(self, db_conn, hs): super(EventPushActionsStore, self).__init__(db_conn, hs) - self.register_background_index_update( + self.db.updates.register_background_index_update( self.EPA_HIGHLIGHT_INDEX, index_name="event_push_actions_u_highlight", table="event_push_actions", columns=["user_id", "stream_ordering"], ) - self.register_background_index_update( + self.db.updates.register_background_index_update( "event_push_actions_highlights_index", index_name="event_push_actions_highlights_index", table="event_push_actions", diff --git a/synapse/storage/data_stores/main/events.py b/synapse/storage/data_stores/main/events.py index 01ec9ec397..d644c82784 100644 --- a/synapse/storage/data_stores/main/events.py +++ b/synapse/storage/data_stores/main/events.py @@ -38,7 +38,6 @@ from synapse.logging.utils import log_function from synapse.metrics import BucketCollector from synapse.metrics.background_process_metrics import run_as_background_process from synapse.storage._base import make_in_list_sql_clause -from synapse.storage.background_updates import BackgroundUpdateStore from synapse.storage.data_stores.main.event_federation import EventFederationStore from synapse.storage.data_stores.main.events_worker import EventsWorkerStore from synapse.storage.data_stores.main.state import StateGroupWorkerStore @@ -94,10 +93,7 @@ def _retry_on_integrity_error(func): # inherits from EventFederationStore so that we can call _update_backward_extremities # and _handle_mult_prev_events (though arguably those could both be moved in here) class EventsStore( - StateGroupWorkerStore, - EventFederationStore, - EventsWorkerStore, - BackgroundUpdateStore, + StateGroupWorkerStore, EventFederationStore, EventsWorkerStore, ): def __init__(self, db_conn, hs): super(EventsStore, self).__init__(db_conn, hs) diff --git a/synapse/storage/data_stores/main/events_bg_updates.py b/synapse/storage/data_stores/main/events_bg_updates.py index 365e966956..cb1fc30c31 100644 --- a/synapse/storage/data_stores/main/events_bg_updates.py +++ b/synapse/storage/data_stores/main/events_bg_updates.py @@ -22,13 +22,12 @@ from canonicaljson import json from twisted.internet import defer from synapse.api.constants import EventContentFields -from synapse.storage._base import make_in_list_sql_clause -from synapse.storage.background_updates import BackgroundUpdateStore +from synapse.storage._base import SQLBaseStore, make_in_list_sql_clause logger = logging.getLogger(__name__) -class EventsBackgroundUpdatesStore(BackgroundUpdateStore): +class EventsBackgroundUpdatesStore(SQLBaseStore): EVENT_ORIGIN_SERVER_TS_NAME = "event_origin_server_ts" EVENT_FIELDS_SENDER_URL_UPDATE_NAME = "event_fields_sender_url" @@ -37,15 +36,15 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): def __init__(self, db_conn, hs): super(EventsBackgroundUpdatesStore, self).__init__(db_conn, hs) - self.register_background_update_handler( + self.db.updates.register_background_update_handler( self.EVENT_ORIGIN_SERVER_TS_NAME, self._background_reindex_origin_server_ts ) - self.register_background_update_handler( + self.db.updates.register_background_update_handler( self.EVENT_FIELDS_SENDER_URL_UPDATE_NAME, self._background_reindex_fields_sender, ) - self.register_background_index_update( + self.db.updates.register_background_index_update( "event_contains_url_index", index_name="event_contains_url_index", table="events", @@ -56,7 +55,7 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): # an event_id index on event_search is useful for the purge_history # api. Plus it means we get to enforce some integrity with a UNIQUE # clause - self.register_background_index_update( + self.db.updates.register_background_index_update( "event_search_event_id_idx", index_name="event_search_event_id_idx", table="event_search", @@ -65,16 +64,16 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): psql_only=True, ) - self.register_background_update_handler( + self.db.updates.register_background_update_handler( self.DELETE_SOFT_FAILED_EXTREMITIES, self._cleanup_extremities_bg_update ) - self.register_background_update_handler( + self.db.updates.register_background_update_handler( "redactions_received_ts", self._redactions_received_ts ) # This index gets deleted in `event_fix_redactions_bytes` update - self.register_background_index_update( + self.db.updates.register_background_index_update( "event_fix_redactions_bytes_create_index", index_name="redactions_censored_redacts", table="redactions", @@ -82,11 +81,11 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): where_clause="have_censored", ) - self.register_background_update_handler( + self.db.updates.register_background_update_handler( "event_fix_redactions_bytes", self._event_fix_redactions_bytes ) - self.register_background_update_handler( + self.db.updates.register_background_update_handler( "event_store_labels", self._event_store_labels ) @@ -145,7 +144,7 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): "rows_inserted": rows_inserted + len(rows), } - self._background_update_progress_txn( + self.db.updates._background_update_progress_txn( txn, self.EVENT_FIELDS_SENDER_URL_UPDATE_NAME, progress ) @@ -156,7 +155,9 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): ) if not result: - yield self._end_background_update(self.EVENT_FIELDS_SENDER_URL_UPDATE_NAME) + yield self.db.updates._end_background_update( + self.EVENT_FIELDS_SENDER_URL_UPDATE_NAME + ) return result @@ -222,7 +223,7 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): "rows_inserted": rows_inserted + len(rows_to_update), } - self._background_update_progress_txn( + self.db.updates._background_update_progress_txn( txn, self.EVENT_ORIGIN_SERVER_TS_NAME, progress ) @@ -233,7 +234,9 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): ) if not result: - yield self._end_background_update(self.EVENT_ORIGIN_SERVER_TS_NAME) + yield self.db.updates._end_background_update( + self.EVENT_ORIGIN_SERVER_TS_NAME + ) return result @@ -411,7 +414,9 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): ) if not num_handled: - yield self._end_background_update(self.DELETE_SOFT_FAILED_EXTREMITIES) + yield self.db.updates._end_background_update( + self.DELETE_SOFT_FAILED_EXTREMITIES + ) def _drop_table_txn(txn): txn.execute("DROP TABLE _extremities_to_check") @@ -464,7 +469,7 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): txn.execute(sql, (self._clock.time_msec(), last_event_id, upper_event_id)) - self._background_update_progress_txn( + self.db.updates._background_update_progress_txn( txn, "redactions_received_ts", {"last_event_id": upper_event_id} ) @@ -475,7 +480,7 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): ) if not count: - yield self._end_background_update("redactions_received_ts") + yield self.db.updates._end_background_update("redactions_received_ts") return count @@ -505,7 +510,7 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): "_event_fix_redactions_bytes", _event_fix_redactions_bytes_txn ) - yield self._end_background_update("event_fix_redactions_bytes") + yield self.db.updates._end_background_update("event_fix_redactions_bytes") return 1 @@ -559,7 +564,7 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): nbrows += 1 last_row_event_id = event_id - self._background_update_progress_txn( + self.db.updates._background_update_progress_txn( txn, "event_store_labels", {"last_event_id": last_row_event_id} ) @@ -570,6 +575,6 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore): ) if not num_rows: - yield self._end_background_update("event_store_labels") + yield self.db.updates._end_background_update("event_store_labels") return num_rows diff --git a/synapse/storage/data_stores/main/media_repository.py b/synapse/storage/data_stores/main/media_repository.py index ea02497784..03c9c6f8ae 100644 --- a/synapse/storage/data_stores/main/media_repository.py +++ b/synapse/storage/data_stores/main/media_repository.py @@ -12,14 +12,14 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. -from synapse.storage.background_updates import BackgroundUpdateStore +from synapse.storage._base import SQLBaseStore -class MediaRepositoryBackgroundUpdateStore(BackgroundUpdateStore): +class MediaRepositoryBackgroundUpdateStore(SQLBaseStore): def __init__(self, db_conn, hs): super(MediaRepositoryBackgroundUpdateStore, self).__init__(db_conn, hs) - self.register_background_index_update( + self.db.updates.register_background_index_update( update_name="local_media_repository_url_idx", index_name="local_media_repository_url_idx", table="local_media_repository", diff --git a/synapse/storage/data_stores/main/registration.py b/synapse/storage/data_stores/main/registration.py index 8f9aa87ceb..1ef143c6d8 100644 --- a/synapse/storage/data_stores/main/registration.py +++ b/synapse/storage/data_stores/main/registration.py @@ -26,7 +26,6 @@ from twisted.internet.defer import Deferred from synapse.api.constants import UserTypes from synapse.api.errors import Codes, StoreError, SynapseError, ThreepidValidationError from synapse.metrics.background_process_metrics import run_as_background_process -from synapse.storage import background_updates from synapse.storage._base import SQLBaseStore from synapse.types import UserID from synapse.util.caches.descriptors import cached, cachedInlineCallbacks @@ -794,23 +793,21 @@ class RegistrationWorkerStore(SQLBaseStore): ) -class RegistrationBackgroundUpdateStore( - RegistrationWorkerStore, background_updates.BackgroundUpdateStore -): +class RegistrationBackgroundUpdateStore(RegistrationWorkerStore): def __init__(self, db_conn, hs): super(RegistrationBackgroundUpdateStore, self).__init__(db_conn, hs) self.clock = hs.get_clock() self.config = hs.config - self.register_background_index_update( + self.db.updates.register_background_index_update( "access_tokens_device_index", index_name="access_tokens_device_id", table="access_tokens", columns=["user_id", "device_id"], ) - self.register_background_index_update( + self.db.updates.register_background_index_update( "users_creation_ts", index_name="users_creation_ts", table="users", @@ -820,13 +817,13 @@ class RegistrationBackgroundUpdateStore( # we no longer use refresh tokens, but it's possible that some people # might have a background update queued to build this index. Just # clear the background update. - self.register_noop_background_update("refresh_tokens_device_index") + self.db.updates.register_noop_background_update("refresh_tokens_device_index") - self.register_background_update_handler( + self.db.updates.register_background_update_handler( "user_threepids_grandfather", self._bg_user_threepids_grandfather ) - self.register_background_update_handler( + self.db.updates.register_background_update_handler( "users_set_deactivated_flag", self._background_update_set_deactivated_flag ) @@ -873,7 +870,7 @@ class RegistrationBackgroundUpdateStore( logger.info("Marked %d rows as deactivated", rows_processed_nb) - self._background_update_progress_txn( + self.db.updates._background_update_progress_txn( txn, "users_set_deactivated_flag", {"user_id": rows[-1]["name"]} ) @@ -887,7 +884,7 @@ class RegistrationBackgroundUpdateStore( ) if end: - yield self._end_background_update("users_set_deactivated_flag") + yield self.db.updates._end_background_update("users_set_deactivated_flag") return nb_processed @@ -917,7 +914,7 @@ class RegistrationBackgroundUpdateStore( "_bg_user_threepids_grandfather", _bg_user_threepids_grandfather_txn ) - yield self._end_background_update("user_threepids_grandfather") + yield self.db.updates._end_background_update("user_threepids_grandfather") return 1 diff --git a/synapse/storage/data_stores/main/room.py b/synapse/storage/data_stores/main/room.py index a26ed47afc..da42dae243 100644 --- a/synapse/storage/data_stores/main/room.py +++ b/synapse/storage/data_stores/main/room.py @@ -28,7 +28,6 @@ from twisted.internet import defer from synapse.api.constants import EventTypes from synapse.api.errors import StoreError from synapse.storage._base import SQLBaseStore -from synapse.storage.background_updates import BackgroundUpdateStore from synapse.storage.data_stores.main.search import SearchStore from synapse.types import ThirdPartyInstanceID from synapse.util.caches.descriptors import cached, cachedInlineCallbacks @@ -361,13 +360,13 @@ class RoomWorkerStore(SQLBaseStore): defer.returnValue(row) -class RoomBackgroundUpdateStore(BackgroundUpdateStore): +class RoomBackgroundUpdateStore(SQLBaseStore): def __init__(self, db_conn, hs): super(RoomBackgroundUpdateStore, self).__init__(db_conn, hs) self.config = hs.config - self.register_background_update_handler( + self.db.updates.register_background_update_handler( "insert_room_retention", self._background_insert_retention, ) @@ -421,7 +420,7 @@ class RoomBackgroundUpdateStore(BackgroundUpdateStore): logger.info("Inserted %d rows into room_retention", len(rows)) - self._background_update_progress_txn( + self.db.updates._background_update_progress_txn( txn, "insert_room_retention", {"room_id": rows[-1]["room_id"]} ) @@ -435,7 +434,7 @@ class RoomBackgroundUpdateStore(BackgroundUpdateStore): ) if end: - yield self._end_background_update("insert_room_retention") + yield self.db.updates._end_background_update("insert_room_retention") defer.returnValue(batch_size) diff --git a/synapse/storage/data_stores/main/roommember.py b/synapse/storage/data_stores/main/roommember.py index 7f4d02b25b..929f6b0d39 100644 --- a/synapse/storage/data_stores/main/roommember.py +++ b/synapse/storage/data_stores/main/roommember.py @@ -26,8 +26,11 @@ from twisted.internet import defer from synapse.api.constants import EventTypes, Membership from synapse.metrics import LaterGauge from synapse.metrics.background_process_metrics import run_as_background_process -from synapse.storage._base import LoggingTransaction, make_in_list_sql_clause -from synapse.storage.background_updates import BackgroundUpdateStore +from synapse.storage._base import ( + LoggingTransaction, + SQLBaseStore, + make_in_list_sql_clause, +) from synapse.storage.data_stores.main.events_worker import EventsWorkerStore from synapse.storage.engines import Sqlite3Engine from synapse.storage.roommember import ( @@ -831,17 +834,17 @@ class RoomMemberWorkerStore(EventsWorkerStore): ) -class RoomMemberBackgroundUpdateStore(BackgroundUpdateStore): +class RoomMemberBackgroundUpdateStore(SQLBaseStore): def __init__(self, db_conn, hs): super(RoomMemberBackgroundUpdateStore, self).__init__(db_conn, hs) - self.register_background_update_handler( + self.db.updates.register_background_update_handler( _MEMBERSHIP_PROFILE_UPDATE_NAME, self._background_add_membership_profile ) - self.register_background_update_handler( + self.db.updates.register_background_update_handler( _CURRENT_STATE_MEMBERSHIP_UPDATE_NAME, self._background_current_state_membership, ) - self.register_background_index_update( + self.db.updates.register_background_index_update( "room_membership_forgotten_idx", index_name="room_memberships_user_room_forgotten", table="room_memberships", @@ -909,7 +912,7 @@ class RoomMemberBackgroundUpdateStore(BackgroundUpdateStore): "max_stream_id_exclusive": min_stream_id, } - self._background_update_progress_txn( + self.db.updates._background_update_progress_txn( txn, _MEMBERSHIP_PROFILE_UPDATE_NAME, progress ) @@ -920,7 +923,9 @@ class RoomMemberBackgroundUpdateStore(BackgroundUpdateStore): ) if not result: - yield self._end_background_update(_MEMBERSHIP_PROFILE_UPDATE_NAME) + yield self.db.updates._end_background_update( + _MEMBERSHIP_PROFILE_UPDATE_NAME + ) return result @@ -959,7 +964,7 @@ class RoomMemberBackgroundUpdateStore(BackgroundUpdateStore): last_processed_room = next_room - self._background_update_progress_txn( + self.db.updates._background_update_progress_txn( txn, _CURRENT_STATE_MEMBERSHIP_UPDATE_NAME, {"last_processed_room": last_processed_room}, @@ -978,7 +983,9 @@ class RoomMemberBackgroundUpdateStore(BackgroundUpdateStore): ) if finished: - yield self._end_background_update(_CURRENT_STATE_MEMBERSHIP_UPDATE_NAME) + yield self.db.updates._end_background_update( + _CURRENT_STATE_MEMBERSHIP_UPDATE_NAME + ) return row_count diff --git a/synapse/storage/data_stores/main/search.py b/synapse/storage/data_stores/main/search.py index 55a604850e..ffa1817e64 100644 --- a/synapse/storage/data_stores/main/search.py +++ b/synapse/storage/data_stores/main/search.py @@ -24,8 +24,7 @@ from canonicaljson import json from twisted.internet import defer from synapse.api.errors import SynapseError -from synapse.storage._base import make_in_list_sql_clause -from synapse.storage.background_updates import BackgroundUpdateStore +from synapse.storage._base import SQLBaseStore, make_in_list_sql_clause from synapse.storage.engines import PostgresEngine, Sqlite3Engine logger = logging.getLogger(__name__) @@ -36,7 +35,7 @@ SearchEntry = namedtuple( ) -class SearchBackgroundUpdateStore(BackgroundUpdateStore): +class SearchBackgroundUpdateStore(SQLBaseStore): EVENT_SEARCH_UPDATE_NAME = "event_search" EVENT_SEARCH_ORDER_UPDATE_NAME = "event_search_order" @@ -49,10 +48,10 @@ class SearchBackgroundUpdateStore(BackgroundUpdateStore): if not hs.config.enable_search: return - self.register_background_update_handler( + self.db.updates.register_background_update_handler( self.EVENT_SEARCH_UPDATE_NAME, self._background_reindex_search ) - self.register_background_update_handler( + self.db.updates.register_background_update_handler( self.EVENT_SEARCH_ORDER_UPDATE_NAME, self._background_reindex_search_order ) @@ -61,9 +60,11 @@ class SearchBackgroundUpdateStore(BackgroundUpdateStore): # a GIN index. However, it's possible that some people might still have # the background update queued, so we register a handler to clear the # background update. - self.register_noop_background_update(self.EVENT_SEARCH_USE_GIST_POSTGRES_NAME) + self.db.updates.register_noop_background_update( + self.EVENT_SEARCH_USE_GIST_POSTGRES_NAME + ) - self.register_background_update_handler( + self.db.updates.register_background_update_handler( self.EVENT_SEARCH_USE_GIN_POSTGRES_NAME, self._background_reindex_gin_search ) @@ -153,7 +154,7 @@ class SearchBackgroundUpdateStore(BackgroundUpdateStore): "rows_inserted": rows_inserted + len(event_search_rows), } - self._background_update_progress_txn( + self.db.updates._background_update_progress_txn( txn, self.EVENT_SEARCH_UPDATE_NAME, progress ) @@ -164,7 +165,7 @@ class SearchBackgroundUpdateStore(BackgroundUpdateStore): ) if not result: - yield self._end_background_update(self.EVENT_SEARCH_UPDATE_NAME) + yield self.db.updates._end_background_update(self.EVENT_SEARCH_UPDATE_NAME) return result @@ -208,7 +209,9 @@ class SearchBackgroundUpdateStore(BackgroundUpdateStore): if isinstance(self.database_engine, PostgresEngine): yield self.db.runWithConnection(create_index) - yield self._end_background_update(self.EVENT_SEARCH_USE_GIN_POSTGRES_NAME) + yield self.db.updates._end_background_update( + self.EVENT_SEARCH_USE_GIN_POSTGRES_NAME + ) return 1 @defer.inlineCallbacks @@ -244,7 +247,7 @@ class SearchBackgroundUpdateStore(BackgroundUpdateStore): yield self.db.runInteraction( self.EVENT_SEARCH_ORDER_UPDATE_NAME, - self._background_update_progress_txn, + self.db.updates._background_update_progress_txn, self.EVENT_SEARCH_ORDER_UPDATE_NAME, pg, ) @@ -274,7 +277,7 @@ class SearchBackgroundUpdateStore(BackgroundUpdateStore): "have_added_indexes": True, } - self._background_update_progress_txn( + self.db.updates._background_update_progress_txn( txn, self.EVENT_SEARCH_ORDER_UPDATE_NAME, progress ) @@ -285,7 +288,9 @@ class SearchBackgroundUpdateStore(BackgroundUpdateStore): ) if not finished: - yield self._end_background_update(self.EVENT_SEARCH_ORDER_UPDATE_NAME) + yield self.db.updates._end_background_update( + self.EVENT_SEARCH_ORDER_UPDATE_NAME + ) return num_rows diff --git a/synapse/storage/data_stores/main/state.py b/synapse/storage/data_stores/main/state.py index 851e81d6b3..7d5a9f8128 100644 --- a/synapse/storage/data_stores/main/state.py +++ b/synapse/storage/data_stores/main/state.py @@ -27,7 +27,6 @@ from synapse.api.errors import NotFoundError from synapse.events import EventBase from synapse.events.snapshot import EventContext from synapse.storage._base import SQLBaseStore -from synapse.storage.background_updates import BackgroundUpdateStore from synapse.storage.data_stores.main.events_worker import EventsWorkerStore from synapse.storage.engines import PostgresEngine from synapse.storage.state import StateFilter @@ -1023,9 +1022,7 @@ class StateGroupWorkerStore( return set(row["state_group"] for row in rows) -class StateBackgroundUpdateStore( - StateGroupBackgroundUpdateStore, BackgroundUpdateStore -): +class StateBackgroundUpdateStore(StateGroupBackgroundUpdateStore): STATE_GROUP_DEDUPLICATION_UPDATE_NAME = "state_group_state_deduplication" STATE_GROUP_INDEX_UPDATE_NAME = "state_group_state_type_index" @@ -1034,21 +1031,21 @@ class StateBackgroundUpdateStore( def __init__(self, db_conn, hs): super(StateBackgroundUpdateStore, self).__init__(db_conn, hs) - self.register_background_update_handler( + self.db.updates.register_background_update_handler( self.STATE_GROUP_DEDUPLICATION_UPDATE_NAME, self._background_deduplicate_state, ) - self.register_background_update_handler( + self.db.updates.register_background_update_handler( self.STATE_GROUP_INDEX_UPDATE_NAME, self._background_index_state ) - self.register_background_index_update( + self.db.updates.register_background_index_update( self.CURRENT_STATE_INDEX_UPDATE_NAME, index_name="current_state_events_member_index", table="current_state_events", columns=["state_key"], where_clause="type='m.room.member'", ) - self.register_background_index_update( + self.db.updates.register_background_index_update( self.EVENT_STATE_GROUP_INDEX_UPDATE_NAME, index_name="event_to_state_groups_sg_index", table="event_to_state_groups", @@ -1181,7 +1178,7 @@ class StateBackgroundUpdateStore( "max_group": max_group, } - self._background_update_progress_txn( + self.db.updates._background_update_progress_txn( txn, self.STATE_GROUP_DEDUPLICATION_UPDATE_NAME, progress ) @@ -1192,7 +1189,7 @@ class StateBackgroundUpdateStore( ) if finished: - yield self._end_background_update( + yield self.db.updates._end_background_update( self.STATE_GROUP_DEDUPLICATION_UPDATE_NAME ) @@ -1224,7 +1221,7 @@ class StateBackgroundUpdateStore( yield self.db.runWithConnection(reindex_txn) - yield self._end_background_update(self.STATE_GROUP_INDEX_UPDATE_NAME) + yield self.db.updates._end_background_update(self.STATE_GROUP_INDEX_UPDATE_NAME) return 1 diff --git a/synapse/storage/data_stores/main/stats.py b/synapse/storage/data_stores/main/stats.py index 974ffc15bd..6b91988c2a 100644 --- a/synapse/storage/data_stores/main/stats.py +++ b/synapse/storage/data_stores/main/stats.py @@ -68,17 +68,17 @@ class StatsStore(StateDeltasStore): self.stats_delta_processing_lock = DeferredLock() - self.register_background_update_handler( + self.db.updates.register_background_update_handler( "populate_stats_process_rooms", self._populate_stats_process_rooms ) - self.register_background_update_handler( + self.db.updates.register_background_update_handler( "populate_stats_process_users", self._populate_stats_process_users ) # we no longer need to perform clean-up, but we will give ourselves # the potential to reintroduce it in the future – so documentation # will still encourage the use of this no-op handler. - self.register_noop_background_update("populate_stats_cleanup") - self.register_noop_background_update("populate_stats_prepare") + self.db.updates.register_noop_background_update("populate_stats_cleanup") + self.db.updates.register_noop_background_update("populate_stats_prepare") def quantise_stats_time(self, ts): """ @@ -102,7 +102,7 @@ class StatsStore(StateDeltasStore): This is a background update which regenerates statistics for users. """ if not self.stats_enabled: - yield self._end_background_update("populate_stats_process_users") + yield self.db.updates._end_background_update("populate_stats_process_users") return 1 last_user_id = progress.get("last_user_id", "") @@ -123,7 +123,7 @@ class StatsStore(StateDeltasStore): # No more rooms -- complete the transaction. if not users_to_work_on: - yield self._end_background_update("populate_stats_process_users") + yield self.db.updates._end_background_update("populate_stats_process_users") return 1 for user_id in users_to_work_on: @@ -132,7 +132,7 @@ class StatsStore(StateDeltasStore): yield self.db.runInteraction( "populate_stats_process_users", - self._background_update_progress_txn, + self.db.updates._background_update_progress_txn, "populate_stats_process_users", progress, ) @@ -145,7 +145,7 @@ class StatsStore(StateDeltasStore): This is a background update which regenerates statistics for rooms. """ if not self.stats_enabled: - yield self._end_background_update("populate_stats_process_rooms") + yield self.db.updates._end_background_update("populate_stats_process_rooms") return 1 last_room_id = progress.get("last_room_id", "") @@ -166,7 +166,7 @@ class StatsStore(StateDeltasStore): # No more rooms -- complete the transaction. if not rooms_to_work_on: - yield self._end_background_update("populate_stats_process_rooms") + yield self.db.updates._end_background_update("populate_stats_process_rooms") return 1 for room_id in rooms_to_work_on: @@ -175,7 +175,7 @@ class StatsStore(StateDeltasStore): yield self.db.runInteraction( "_populate_stats_process_rooms", - self._background_update_progress_txn, + self.db.updates._background_update_progress_txn, "populate_stats_process_rooms", progress, ) diff --git a/synapse/storage/data_stores/main/user_directory.py b/synapse/storage/data_stores/main/user_directory.py index 7118bd62f3..62ffb34b29 100644 --- a/synapse/storage/data_stores/main/user_directory.py +++ b/synapse/storage/data_stores/main/user_directory.py @@ -19,7 +19,6 @@ import re from twisted.internet import defer from synapse.api.constants import EventTypes, JoinRules -from synapse.storage.background_updates import BackgroundUpdateStore from synapse.storage.data_stores.main.state import StateFilter from synapse.storage.data_stores.main.state_deltas import StateDeltasStore from synapse.storage.engines import PostgresEngine, Sqlite3Engine @@ -32,7 +31,7 @@ logger = logging.getLogger(__name__) TEMP_TABLE = "_temp_populate_user_directory" -class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore): +class UserDirectoryBackgroundUpdateStore(StateDeltasStore): # How many records do we calculate before sending it to # add_users_who_share_private_rooms? @@ -43,19 +42,19 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore self.server_name = hs.hostname - self.register_background_update_handler( + self.db.updates.register_background_update_handler( "populate_user_directory_createtables", self._populate_user_directory_createtables, ) - self.register_background_update_handler( + self.db.updates.register_background_update_handler( "populate_user_directory_process_rooms", self._populate_user_directory_process_rooms, ) - self.register_background_update_handler( + self.db.updates.register_background_update_handler( "populate_user_directory_process_users", self._populate_user_directory_process_users, ) - self.register_background_update_handler( + self.db.updates.register_background_update_handler( "populate_user_directory_cleanup", self._populate_user_directory_cleanup ) @@ -108,7 +107,9 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore ) yield self.db.simple_insert(TEMP_TABLE + "_position", {"position": new_pos}) - yield self._end_background_update("populate_user_directory_createtables") + yield self.db.updates._end_background_update( + "populate_user_directory_createtables" + ) return 1 @defer.inlineCallbacks @@ -130,7 +131,7 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore "populate_user_directory_cleanup", _delete_staging_area ) - yield self._end_background_update("populate_user_directory_cleanup") + yield self.db.updates._end_background_update("populate_user_directory_cleanup") return 1 @defer.inlineCallbacks @@ -176,7 +177,9 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore # No more rooms -- complete the transaction. if not rooms_to_work_on: - yield self._end_background_update("populate_user_directory_process_rooms") + yield self.db.updates._end_background_update( + "populate_user_directory_process_rooms" + ) return 1 logger.info( @@ -248,7 +251,7 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore progress["remaining"] -= 1 yield self.db.runInteraction( "populate_user_directory", - self._background_update_progress_txn, + self.db.updates._background_update_progress_txn, "populate_user_directory_process_rooms", progress, ) @@ -267,7 +270,9 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore If search_all_users is enabled, add all of the users to the user directory. """ if not self.hs.config.user_directory_search_all_users: - yield self._end_background_update("populate_user_directory_process_users") + yield self.db.updates._end_background_update( + "populate_user_directory_process_users" + ) return 1 def _get_next_batch(txn): @@ -297,7 +302,9 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore # No more users -- complete the transaction. if not users_to_work_on: - yield self._end_background_update("populate_user_directory_process_users") + yield self.db.updates._end_background_update( + "populate_user_directory_process_users" + ) return 1 logger.info( @@ -317,7 +324,7 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore, BackgroundUpdateStore progress["remaining"] -= 1 yield self.db.runInteraction( "populate_user_directory", - self._background_update_progress_txn, + self.db.updates._background_update_progress_txn, "populate_user_directory_process_users", progress, ) diff --git a/synapse/storage/database.py b/synapse/storage/database.py index ac64d80806..be36c1b829 100644 --- a/synapse/storage/database.py +++ b/synapse/storage/database.py @@ -30,6 +30,7 @@ from twisted.internet import defer from synapse.api.errors import StoreError from synapse.logging.context import LoggingContext, make_deferred_yieldable from synapse.metrics.background_process_metrics import run_as_background_process +from synapse.storage.background_updates import BackgroundUpdater from synapse.storage.engines import PostgresEngine, Sqlite3Engine from synapse.util.stringutils import exception_to_unicode @@ -223,6 +224,8 @@ class Database(object): self._clock = hs.get_clock() self._db_pool = hs.get_db_pool() + self.updates = BackgroundUpdater(hs, self) + self._previous_txn_total_time = 0 self._current_txn_total_time = 0 self._previous_loop_ts = 0 diff --git a/synmark/__init__.py b/synmark/__init__.py index 570eb818d9..afe4fad8cb 100644 --- a/synmark/__init__.py +++ b/synmark/__init__.py @@ -47,9 +47,9 @@ async def make_homeserver(reactor, config=None): stor = hs.get_datastore() # Run the database background updates. - if hasattr(stor, "do_next_background_update"): - while not await stor.has_completed_background_updates(): - await stor.do_next_background_update(1) + if hasattr(stor.db.updates, "do_next_background_update"): + while not await stor.db.updates.has_completed_background_updates(): + await stor.db.updates.do_next_background_update(1) def cleanup(): for i in cleanup_tasks: diff --git a/tests/handlers/test_stats.py b/tests/handlers/test_stats.py index 7f7962c3dd..d9d312f0fb 100644 --- a/tests/handlers/test_stats.py +++ b/tests/handlers/test_stats.py @@ -42,7 +42,7 @@ class StatsRoomTests(unittest.HomeserverTestCase): Add the background updates we need to run. """ # Ugh, have to reset this flag - self.store._all_done = False + self.store.db.updates._all_done = False self.get_success( self.store.db.simple_insert( @@ -108,8 +108,12 @@ class StatsRoomTests(unittest.HomeserverTestCase): # Do the initial population of the stats via the background update self._add_background_updates() - while not self.get_success(self.store.has_completed_background_updates()): - self.get_success(self.store.do_next_background_update(100), by=0.1) + while not self.get_success( + self.store.db.updates.has_completed_background_updates() + ): + self.get_success( + self.store.db.updates.do_next_background_update(100), by=0.1 + ) def test_initial_room(self): """ @@ -141,8 +145,12 @@ class StatsRoomTests(unittest.HomeserverTestCase): # Do the initial population of the user directory via the background update self._add_background_updates() - while not self.get_success(self.store.has_completed_background_updates()): - self.get_success(self.store.do_next_background_update(100), by=0.1) + while not self.get_success( + self.store.db.updates.has_completed_background_updates() + ): + self.get_success( + self.store.db.updates.do_next_background_update(100), by=0.1 + ) r = self.get_success(self.get_all_room_state()) @@ -178,7 +186,7 @@ class StatsRoomTests(unittest.HomeserverTestCase): # the position that the deltas should begin at, once they take over. self.hs.config.stats_enabled = True self.handler.stats_enabled = True - self.store._all_done = False + self.store.db.updates._all_done = False self.get_success( self.store.db.simple_update_one( table="stats_incremental_position", @@ -194,8 +202,12 @@ class StatsRoomTests(unittest.HomeserverTestCase): ) ) - while not self.get_success(self.store.has_completed_background_updates()): - self.get_success(self.store.do_next_background_update(100), by=0.1) + while not self.get_success( + self.store.db.updates.has_completed_background_updates() + ): + self.get_success( + self.store.db.updates.do_next_background_update(100), by=0.1 + ) # Now, before the table is actually ingested, add some more events. self.helper.invite(room=room_1, src=u1, targ=u2, tok=u1_token) @@ -221,9 +233,13 @@ class StatsRoomTests(unittest.HomeserverTestCase): ) ) - self.store._all_done = False - while not self.get_success(self.store.has_completed_background_updates()): - self.get_success(self.store.do_next_background_update(100), by=0.1) + self.store.db.updates._all_done = False + while not self.get_success( + self.store.db.updates.has_completed_background_updates() + ): + self.get_success( + self.store.db.updates.do_next_background_update(100), by=0.1 + ) self.reactor.advance(86401) @@ -653,7 +669,7 @@ class StatsRoomTests(unittest.HomeserverTestCase): # preparation stage of the initial background update # Ugh, have to reset this flag - self.store._all_done = False + self.store.db.updates._all_done = False self.get_success( self.store.db.simple_delete( @@ -673,7 +689,7 @@ class StatsRoomTests(unittest.HomeserverTestCase): # now do the background updates - self.store._all_done = False + self.store.db.updates._all_done = False self.get_success( self.store.db.simple_insert( "background_updates", @@ -705,8 +721,12 @@ class StatsRoomTests(unittest.HomeserverTestCase): ) ) - while not self.get_success(self.store.has_completed_background_updates()): - self.get_success(self.store.do_next_background_update(100), by=0.1) + while not self.get_success( + self.store.db.updates.has_completed_background_updates() + ): + self.get_success( + self.store.db.updates.do_next_background_update(100), by=0.1 + ) r1stats_complete = self._get_current_stats("room", r1) u1stats_complete = self._get_current_stats("user", u1) diff --git a/tests/handlers/test_user_directory.py b/tests/handlers/test_user_directory.py index bc9d441541..26071059d2 100644 --- a/tests/handlers/test_user_directory.py +++ b/tests/handlers/test_user_directory.py @@ -181,7 +181,7 @@ class UserDirectoryTestCase(unittest.HomeserverTestCase): Add the background updates we need to run. """ # Ugh, have to reset this flag - self.store._all_done = False + self.store.db.updates._all_done = False self.get_success( self.store.db.simple_insert( @@ -255,8 +255,12 @@ class UserDirectoryTestCase(unittest.HomeserverTestCase): # Do the initial population of the user directory via the background update self._add_background_updates() - while not self.get_success(self.store.has_completed_background_updates()): - self.get_success(self.store.do_next_background_update(100), by=0.1) + while not self.get_success( + self.store.db.updates.has_completed_background_updates() + ): + self.get_success( + self.store.db.updates.do_next_background_update(100), by=0.1 + ) shares_private = self.get_users_who_share_private_rooms() public_users = self.get_users_in_public_rooms() @@ -290,8 +294,12 @@ class UserDirectoryTestCase(unittest.HomeserverTestCase): # Do the initial population of the user directory via the background update self._add_background_updates() - while not self.get_success(self.store.has_completed_background_updates()): - self.get_success(self.store.do_next_background_update(100), by=0.1) + while not self.get_success( + self.store.db.updates.has_completed_background_updates() + ): + self.get_success( + self.store.db.updates.do_next_background_update(100), by=0.1 + ) shares_private = self.get_users_who_share_private_rooms() public_users = self.get_users_in_public_rooms() diff --git a/tests/storage/test_background_update.py b/tests/storage/test_background_update.py index e360297df9..aec76f4ab1 100644 --- a/tests/storage/test_background_update.py +++ b/tests/storage/test_background_update.py @@ -15,7 +15,7 @@ class BackgroundUpdateTestCase(unittest.TestCase): self.update_handler = Mock() - yield self.store.register_background_update_handler( + yield self.store.db.updates.register_background_update_handler( "test_update", self.update_handler ) @@ -23,7 +23,7 @@ class BackgroundUpdateTestCase(unittest.TestCase): # (perhaps we should run them as part of the test HS setup, since we # run all of the other schema setup stuff there?) while True: - res = yield self.store.do_next_background_update(1000) + res = yield self.store.db.updates.do_next_background_update(1000) if res is None: break @@ -39,7 +39,7 @@ class BackgroundUpdateTestCase(unittest.TestCase): progress = {"my_key": progress["my_key"] + 1} yield self.store.db.runInteraction( "update_progress", - self.store._background_update_progress_txn, + self.store.db.updates._background_update_progress_txn, "test_update", progress, ) @@ -47,29 +47,37 @@ class BackgroundUpdateTestCase(unittest.TestCase): self.update_handler.side_effect = update - yield self.store.start_background_update("test_update", {"my_key": 1}) + yield self.store.db.updates.start_background_update( + "test_update", {"my_key": 1} + ) self.update_handler.reset_mock() - result = yield self.store.do_next_background_update(duration_ms * desired_count) + result = yield self.store.db.updates.do_next_background_update( + duration_ms * desired_count + ) self.assertIsNotNone(result) self.update_handler.assert_called_once_with( - {"my_key": 1}, self.store.DEFAULT_BACKGROUND_BATCH_SIZE + {"my_key": 1}, self.store.db.updates.DEFAULT_BACKGROUND_BATCH_SIZE ) # second step: complete the update @defer.inlineCallbacks def update(progress, count): - yield self.store._end_background_update("test_update") + yield self.store.db.updates._end_background_update("test_update") return count self.update_handler.side_effect = update self.update_handler.reset_mock() - result = yield self.store.do_next_background_update(duration_ms * desired_count) + result = yield self.store.db.updates.do_next_background_update( + duration_ms * desired_count + ) self.assertIsNotNone(result) self.update_handler.assert_called_once_with({"my_key": 2}, desired_count) # third step: we don't expect to be called any more self.update_handler.reset_mock() - result = yield self.store.do_next_background_update(duration_ms * desired_count) + result = yield self.store.db.updates.do_next_background_update( + duration_ms * desired_count + ) self.assertIsNone(result) self.assertFalse(self.update_handler.called) diff --git a/tests/storage/test_cleanup_extrems.py b/tests/storage/test_cleanup_extrems.py index e454bbff29..029ac26454 100644 --- a/tests/storage/test_cleanup_extrems.py +++ b/tests/storage/test_cleanup_extrems.py @@ -46,7 +46,9 @@ class CleanupExtremBackgroundUpdateStoreTestCase(HomeserverTestCase): """Re run the background update to clean up the extremities. """ # Make sure we don't clash with in progress updates. - self.assertTrue(self.store._all_done, "Background updates are still ongoing") + self.assertTrue( + self.store.db.updates._all_done, "Background updates are still ongoing" + ) schema_path = os.path.join( prepare_database.dir_path, @@ -68,10 +70,14 @@ class CleanupExtremBackgroundUpdateStoreTestCase(HomeserverTestCase): ) # Ugh, have to reset this flag - self.store._all_done = False + self.store.db.updates._all_done = False - while not self.get_success(self.store.has_completed_background_updates()): - self.get_success(self.store.do_next_background_update(100), by=0.1) + while not self.get_success( + self.store.db.updates.has_completed_background_updates() + ): + self.get_success( + self.store.db.updates.do_next_background_update(100), by=0.1 + ) def test_soft_failed_extremities_handled_correctly(self): """Test that extremities are correctly calculated in the presence of diff --git a/tests/storage/test_client_ips.py b/tests/storage/test_client_ips.py index c4f838907c..fc279340d4 100644 --- a/tests/storage/test_client_ips.py +++ b/tests/storage/test_client_ips.py @@ -202,8 +202,12 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): def test_devices_last_seen_bg_update(self): # First make sure we have completed all updates. - while not self.get_success(self.store.has_completed_background_updates()): - self.get_success(self.store.do_next_background_update(100), by=0.1) + while not self.get_success( + self.store.db.updates.has_completed_background_updates() + ): + self.get_success( + self.store.db.updates.do_next_background_update(100), by=0.1 + ) # Insert a user IP user_id = "@user:id" @@ -256,11 +260,15 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): ) # ... and tell the DataStore that it hasn't finished all updates yet - self.store._all_done = False + self.store.db.updates._all_done = False # Now let's actually drive the updates to completion - while not self.get_success(self.store.has_completed_background_updates()): - self.get_success(self.store.do_next_background_update(100), by=0.1) + while not self.get_success( + self.store.db.updates.has_completed_background_updates() + ): + self.get_success( + self.store.db.updates.do_next_background_update(100), by=0.1 + ) # We should now get the correct result again result = self.get_success( @@ -281,8 +289,12 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): def test_old_user_ips_pruned(self): # First make sure we have completed all updates. - while not self.get_success(self.store.has_completed_background_updates()): - self.get_success(self.store.do_next_background_update(100), by=0.1) + while not self.get_success( + self.store.db.updates.has_completed_background_updates() + ): + self.get_success( + self.store.db.updates.do_next_background_update(100), by=0.1 + ) # Insert a user IP user_id = "@user:id" diff --git a/tests/storage/test_roommember.py b/tests/storage/test_roommember.py index 5f957680a2..7840f63fe3 100644 --- a/tests/storage/test_roommember.py +++ b/tests/storage/test_roommember.py @@ -122,8 +122,12 @@ class CurrentStateMembershipUpdateTestCase(unittest.HomeserverTestCase): def test_can_rerun_update(self): # First make sure we have completed all updates. - while not self.get_success(self.store.has_completed_background_updates()): - self.get_success(self.store.do_next_background_update(100), by=0.1) + while not self.get_success( + self.store.db.updates.has_completed_background_updates() + ): + self.get_success( + self.store.db.updates.do_next_background_update(100), by=0.1 + ) # Now let's create a room, which will insert a membership user = UserID("alice", "test") @@ -143,8 +147,12 @@ class CurrentStateMembershipUpdateTestCase(unittest.HomeserverTestCase): ) # ... and tell the DataStore that it hasn't finished all updates yet - self.store._all_done = False + self.store.db.updates._all_done = False # Now let's actually drive the updates to completion - while not self.get_success(self.store.has_completed_background_updates()): - self.get_success(self.store.do_next_background_update(100), by=0.1) + while not self.get_success( + self.store.db.updates.has_completed_background_updates() + ): + self.get_success( + self.store.db.updates.do_next_background_update(100), by=0.1 + ) diff --git a/tests/unittest.py b/tests/unittest.py index fc856a574a..68d245ec9f 100644 --- a/tests/unittest.py +++ b/tests/unittest.py @@ -401,10 +401,12 @@ class HomeserverTestCase(TestCase): hs = setup_test_homeserver(self.addCleanup, *args, **kwargs) stor = hs.get_datastore() - # Run the database background updates. - if hasattr(stor, "do_next_background_update"): - while not self.get_success(stor.has_completed_background_updates()): - self.get_success(stor.do_next_background_update(1)) + # Run the database background updates, when running against "master". + if hs.__class__.__name__ == "TestHomeServer": + while not self.get_success( + stor.db.updates.has_completed_background_updates() + ): + self.get_success(stor.db.updates.do_next_background_update(1)) return hs -- cgit 1.4.1 From 4ca3ef10b9a8d15cf351d67d574088d944c2e3b1 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Thu, 5 Dec 2019 15:53:10 +0000 Subject: Fixup tests --- tests/rest/client/v1/test_presence.py | 3 +++ tests/rest/client/v1/test_profile.py | 10 +++++++++- tests/utils.py | 4 +++- 3 files changed, 15 insertions(+), 2 deletions(-) (limited to 'tests') diff --git a/tests/rest/client/v1/test_presence.py b/tests/rest/client/v1/test_presence.py index 66c2b68707..0fdff79aa7 100644 --- a/tests/rest/client/v1/test_presence.py +++ b/tests/rest/client/v1/test_presence.py @@ -15,6 +15,8 @@ from mock import Mock +from twisted.internet import defer + from synapse.rest.client.v1 import presence from synapse.types import UserID @@ -36,6 +38,7 @@ class PresenceTestCase(unittest.HomeserverTestCase): ) hs.presence_handler = Mock() + hs.presence_handler.set_state.return_value = defer.succeed(None) return hs diff --git a/tests/rest/client/v1/test_profile.py b/tests/rest/client/v1/test_profile.py index 140d8b3772..12c5e95cb5 100644 --- a/tests/rest/client/v1/test_profile.py +++ b/tests/rest/client/v1/test_profile.py @@ -52,6 +52,14 @@ class MockHandlerProfileTestCase(unittest.TestCase): ] ) + self.mock_handler.get_displayname.return_value = defer.succeed(Mock()) + self.mock_handler.set_displayname.return_value = defer.succeed(Mock()) + self.mock_handler.get_avatar_url.return_value = defer.succeed(Mock()) + self.mock_handler.set_avatar_url.return_value = defer.succeed(Mock()) + self.mock_handler.check_profile_query_allowed.return_value = defer.succeed( + Mock() + ) + hs = yield setup_test_homeserver( self.addCleanup, "test", @@ -63,7 +71,7 @@ class MockHandlerProfileTestCase(unittest.TestCase): ) def _get_user_by_req(request=None, allow_guest=False): - return synapse.types.create_requester(myid) + return defer.succeed(synapse.types.create_requester(myid)) hs.get_auth().get_user_by_req = _get_user_by_req diff --git a/tests/utils.py b/tests/utils.py index de2ac1ed33..c57da59191 100644 --- a/tests/utils.py +++ b/tests/utils.py @@ -461,7 +461,9 @@ class MockHttpResource(HttpServer): try: args = [urlparse.unquote(u) for u in matcher.groups()] - (code, response) = yield func(mock_request, *args) + (code, response) = yield defer.ensureDeferred( + func(mock_request, *args) + ) return code, response except CodeMessageException as e: return (e.code, cs_error(e.msg, code=e.errcode)) -- cgit 1.4.1 From 8437e2383ed2dffacca5395851023eeacb33d7ba Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Thu, 5 Dec 2019 17:58:25 +0000 Subject: Port SyncHandler to async/await --- synapse/handlers/events.py | 30 +++--- synapse/handlers/sync.py | 251 +++++++++++++++++++++----------------------- synapse/notifier.py | 29 +++-- synapse/util/metrics.py | 23 ++-- tests/handlers/test_sync.py | 33 +++--- tests/unittest.py | 7 +- 6 files changed, 182 insertions(+), 191 deletions(-) (limited to 'tests') diff --git a/synapse/handlers/events.py b/synapse/handlers/events.py index 45fe13c62f..ec18a42a68 100644 --- a/synapse/handlers/events.py +++ b/synapse/handlers/events.py @@ -16,8 +16,6 @@ import logging import random -from twisted.internet import defer - from synapse.api.constants import EventTypes, Membership from synapse.api.errors import AuthError, SynapseError from synapse.events import EventBase @@ -50,9 +48,8 @@ class EventStreamHandler(BaseHandler): self._server_notices_sender = hs.get_server_notices_sender() self._event_serializer = hs.get_event_client_serializer() - @defer.inlineCallbacks @log_function - def get_stream( + async def get_stream( self, auth_user_id, pagin_config, @@ -69,17 +66,17 @@ class EventStreamHandler(BaseHandler): """ if room_id: - blocked = yield self.store.is_room_blocked(room_id) + blocked = await self.store.is_room_blocked(room_id) if blocked: raise SynapseError(403, "This room has been blocked on this server") # send any outstanding server notices to the user. - yield self._server_notices_sender.on_user_syncing(auth_user_id) + await self._server_notices_sender.on_user_syncing(auth_user_id) auth_user = UserID.from_string(auth_user_id) presence_handler = self.hs.get_presence_handler() - context = yield presence_handler.user_syncing( + context = await presence_handler.user_syncing( auth_user_id, affect_presence=affect_presence ) with context: @@ -91,7 +88,7 @@ class EventStreamHandler(BaseHandler): # thundering herds on restart. timeout = random.randint(int(timeout * 0.9), int(timeout * 1.1)) - events, tokens = yield self.notifier.get_events_for( + events, tokens = await self.notifier.get_events_for( auth_user, pagin_config, timeout, @@ -112,14 +109,14 @@ class EventStreamHandler(BaseHandler): # Send down presence. if event.state_key == auth_user_id: # Send down presence for everyone in the room. - users = yield self.state.get_current_users_in_room( + users = await self.state.get_current_users_in_room( event.room_id ) - states = yield presence_handler.get_states(users, as_event=True) + states = await presence_handler.get_states(users, as_event=True) to_add.extend(states) else: - ev = yield presence_handler.get_state( + ev = await presence_handler.get_state( UserID.from_string(event.state_key), as_event=True ) to_add.append(ev) @@ -128,7 +125,7 @@ class EventStreamHandler(BaseHandler): time_now = self.clock.time_msec() - chunks = yield self._event_serializer.serialize_events( + chunks = await self._event_serializer.serialize_events( events, time_now, as_client_event=as_client_event, @@ -151,8 +148,7 @@ class EventHandler(BaseHandler): super(EventHandler, self).__init__(hs) self.storage = hs.get_storage() - @defer.inlineCallbacks - def get_event(self, user, room_id, event_id): + async def get_event(self, user, room_id, event_id): """Retrieve a single specified event. Args: @@ -167,15 +163,15 @@ class EventHandler(BaseHandler): AuthError if the user does not have the rights to inspect this event. """ - event = yield self.store.get_event(event_id, check_room_id=room_id) + event = await self.store.get_event(event_id, check_room_id=room_id) if not event: return None - users = yield self.store.get_users_in_room(event.room_id) + users = await self.store.get_users_in_room(event.room_id) is_peeking = user.to_string() not in users - filtered = yield filter_events_for_client( + filtered = await filter_events_for_client( self.storage, user.to_string(), [event], is_peeking=is_peeking ) diff --git a/synapse/handlers/sync.py b/synapse/handlers/sync.py index b536d410e5..12751fd8c0 100644 --- a/synapse/handlers/sync.py +++ b/synapse/handlers/sync.py @@ -22,8 +22,6 @@ from six import iteritems, itervalues from prometheus_client import Counter -from twisted.internet import defer - from synapse.api.constants import EventTypes, Membership from synapse.logging.context import LoggingContext from synapse.push.clientformat import format_push_rules_for_user @@ -241,8 +239,7 @@ class SyncHandler(object): expiry_ms=LAZY_LOADED_MEMBERS_CACHE_MAX_AGE, ) - @defer.inlineCallbacks - def wait_for_sync_for_user( + async def wait_for_sync_for_user( self, sync_config, since_token=None, timeout=0, full_state=False ): """Get the sync for a client if we have new data for it now. Otherwise @@ -255,9 +252,9 @@ class SyncHandler(object): # not been exceeded (if not part of the group by this point, almost certain # auth_blocking will occur) user_id = sync_config.user.to_string() - yield self.auth.check_auth_blocking(user_id) + await self.auth.check_auth_blocking(user_id) - res = yield self.response_cache.wrap( + res = await self.response_cache.wrap( sync_config.request_key, self._wait_for_sync_for_user, sync_config, @@ -267,8 +264,9 @@ class SyncHandler(object): ) return res - @defer.inlineCallbacks - def _wait_for_sync_for_user(self, sync_config, since_token, timeout, full_state): + async def _wait_for_sync_for_user( + self, sync_config, since_token, timeout, full_state + ): if since_token is None: sync_type = "initial_sync" elif full_state: @@ -283,7 +281,7 @@ class SyncHandler(object): if timeout == 0 or since_token is None or full_state: # we are going to return immediately, so don't bother calling # notifier.wait_for_events. - result = yield self.current_sync_for_user( + result = await self.current_sync_for_user( sync_config, since_token, full_state=full_state ) else: @@ -291,7 +289,7 @@ class SyncHandler(object): def current_sync_callback(before_token, after_token): return self.current_sync_for_user(sync_config, since_token) - result = yield self.notifier.wait_for_events( + result = await self.notifier.wait_for_events( sync_config.user.to_string(), timeout, current_sync_callback, @@ -314,15 +312,13 @@ class SyncHandler(object): """ return self.generate_sync_result(sync_config, since_token, full_state) - @defer.inlineCallbacks - def push_rules_for_user(self, user): + async def push_rules_for_user(self, user): user_id = user.to_string() - rules = yield self.store.get_push_rules_for_user(user_id) + rules = await self.store.get_push_rules_for_user(user_id) rules = format_push_rules_for_user(user, rules) return rules - @defer.inlineCallbacks - def ephemeral_by_room(self, sync_result_builder, now_token, since_token=None): + async def ephemeral_by_room(self, sync_result_builder, now_token, since_token=None): """Get the ephemeral events for each room the user is in Args: sync_result_builder(SyncResultBuilder) @@ -343,7 +339,7 @@ class SyncHandler(object): room_ids = sync_result_builder.joined_room_ids typing_source = self.event_sources.sources["typing"] - typing, typing_key = yield typing_source.get_new_events( + typing, typing_key = typing_source.get_new_events( user=sync_config.user, from_key=typing_key, limit=sync_config.filter_collection.ephemeral_limit(), @@ -365,7 +361,7 @@ class SyncHandler(object): receipt_key = since_token.receipt_key if since_token else "0" receipt_source = self.event_sources.sources["receipt"] - receipts, receipt_key = yield receipt_source.get_new_events( + receipts, receipt_key = await receipt_source.get_new_events( user=sync_config.user, from_key=receipt_key, limit=sync_config.filter_collection.ephemeral_limit(), @@ -382,8 +378,7 @@ class SyncHandler(object): return now_token, ephemeral_by_room - @defer.inlineCallbacks - def _load_filtered_recents( + async def _load_filtered_recents( self, room_id, sync_config, @@ -415,10 +410,10 @@ class SyncHandler(object): # ensure that we always include current state in the timeline current_state_ids = frozenset() if any(e.is_state() for e in recents): - current_state_ids = yield self.state.get_current_state_ids(room_id) + current_state_ids = await self.state.get_current_state_ids(room_id) current_state_ids = frozenset(itervalues(current_state_ids)) - recents = yield filter_events_for_client( + recents = await filter_events_for_client( self.storage, sync_config.user.to_string(), recents, @@ -449,14 +444,14 @@ class SyncHandler(object): # Otherwise, we want to return the last N events in the room # in toplogical ordering. if since_key: - events, end_key = yield self.store.get_room_events_stream_for_room( + events, end_key = await self.store.get_room_events_stream_for_room( room_id, limit=load_limit + 1, from_key=since_key, to_key=end_key, ) else: - events, end_key = yield self.store.get_recent_events_for_room( + events, end_key = await self.store.get_recent_events_for_room( room_id, limit=load_limit + 1, end_token=end_key ) loaded_recents = sync_config.filter_collection.filter_room_timeline( @@ -468,10 +463,10 @@ class SyncHandler(object): # ensure that we always include current state in the timeline current_state_ids = frozenset() if any(e.is_state() for e in loaded_recents): - current_state_ids = yield self.state.get_current_state_ids(room_id) + current_state_ids = await self.state.get_current_state_ids(room_id) current_state_ids = frozenset(itervalues(current_state_ids)) - loaded_recents = yield filter_events_for_client( + loaded_recents = await filter_events_for_client( self.storage, sync_config.user.to_string(), loaded_recents, @@ -498,8 +493,7 @@ class SyncHandler(object): limited=limited or newly_joined_room, ) - @defer.inlineCallbacks - def get_state_after_event(self, event, state_filter=StateFilter.all()): + async def get_state_after_event(self, event, state_filter=StateFilter.all()): """ Get the room state after the given event @@ -511,7 +505,7 @@ class SyncHandler(object): Returns: A Deferred map from ((type, state_key)->Event) """ - state_ids = yield self.state_store.get_state_ids_for_event( + state_ids = await self.state_store.get_state_ids_for_event( event.event_id, state_filter=state_filter ) if event.is_state(): @@ -519,8 +513,9 @@ class SyncHandler(object): state_ids[(event.type, event.state_key)] = event.event_id return state_ids - @defer.inlineCallbacks - def get_state_at(self, room_id, stream_position, state_filter=StateFilter.all()): + async def get_state_at( + self, room_id, stream_position, state_filter=StateFilter.all() + ): """ Get the room state at a particular stream position Args: @@ -536,13 +531,13 @@ class SyncHandler(object): # get_recent_events_for_room operates by topo ordering. This therefore # does not reliably give you the state at the given stream position. # (https://github.com/matrix-org/synapse/issues/3305) - last_events, _ = yield self.store.get_recent_events_for_room( + last_events, _ = await self.store.get_recent_events_for_room( room_id, end_token=stream_position.room_key, limit=1 ) if last_events: last_event = last_events[-1] - state = yield self.get_state_after_event( + state = await self.get_state_after_event( last_event, state_filter=state_filter ) @@ -551,8 +546,7 @@ class SyncHandler(object): state = {} return state - @defer.inlineCallbacks - def compute_summary(self, room_id, sync_config, batch, state, now_token): + async def compute_summary(self, room_id, sync_config, batch, state, now_token): """ Works out a room summary block for this room, summarising the number of joined members in the room, and providing the 'hero' members if the room has no name so clients can consistently name rooms. Also adds @@ -574,7 +568,7 @@ class SyncHandler(object): # FIXME: we could/should get this from room_stats when matthew/stats lands # FIXME: this promulgates https://github.com/matrix-org/synapse/issues/3305 - last_events, _ = yield self.store.get_recent_event_ids_for_room( + last_events, _ = await self.store.get_recent_event_ids_for_room( room_id, end_token=now_token.room_key, limit=1 ) @@ -582,7 +576,7 @@ class SyncHandler(object): return None last_event = last_events[-1] - state_ids = yield self.state_store.get_state_ids_for_event( + state_ids = await self.state_store.get_state_ids_for_event( last_event.event_id, state_filter=StateFilter.from_types( [(EventTypes.Name, ""), (EventTypes.CanonicalAlias, "")] @@ -590,7 +584,7 @@ class SyncHandler(object): ) # this is heavily cached, thus: fast. - details = yield self.store.get_room_summary(room_id) + details = await self.store.get_room_summary(room_id) name_id = state_ids.get((EventTypes.Name, "")) canonical_alias_id = state_ids.get((EventTypes.CanonicalAlias, "")) @@ -608,12 +602,12 @@ class SyncHandler(object): # calculating heroes. Empty strings are falsey, so we check # for the "name" value and default to an empty string. if name_id: - name = yield self.store.get_event(name_id, allow_none=True) + name = await self.store.get_event(name_id, allow_none=True) if name and name.content.get("name"): return summary if canonical_alias_id: - canonical_alias = yield self.store.get_event( + canonical_alias = await self.store.get_event( canonical_alias_id, allow_none=True ) if canonical_alias and canonical_alias.content.get("alias"): @@ -678,7 +672,7 @@ class SyncHandler(object): ) ] - missing_hero_state = yield self.store.get_events(missing_hero_event_ids) + missing_hero_state = await self.store.get_events(missing_hero_event_ids) missing_hero_state = missing_hero_state.values() for s in missing_hero_state: @@ -697,8 +691,7 @@ class SyncHandler(object): logger.debug("found LruCache for %r", cache_key) return cache - @defer.inlineCallbacks - def compute_state_delta( + async def compute_state_delta( self, room_id, batch, sync_config, since_token, now_token, full_state ): """ Works out the difference in state between the start of the timeline @@ -759,16 +752,16 @@ class SyncHandler(object): if full_state: if batch: - current_state_ids = yield self.state_store.get_state_ids_for_event( + current_state_ids = await self.state_store.get_state_ids_for_event( batch.events[-1].event_id, state_filter=state_filter ) - state_ids = yield self.state_store.get_state_ids_for_event( + state_ids = await self.state_store.get_state_ids_for_event( batch.events[0].event_id, state_filter=state_filter ) else: - current_state_ids = yield self.get_state_at( + current_state_ids = await self.get_state_at( room_id, stream_position=now_token, state_filter=state_filter ) @@ -783,13 +776,13 @@ class SyncHandler(object): ) elif batch.limited: if batch: - state_at_timeline_start = yield self.state_store.get_state_ids_for_event( + state_at_timeline_start = await self.state_store.get_state_ids_for_event( batch.events[0].event_id, state_filter=state_filter ) else: # We can get here if the user has ignored the senders of all # the recent events. - state_at_timeline_start = yield self.get_state_at( + state_at_timeline_start = await self.get_state_at( room_id, stream_position=now_token, state_filter=state_filter ) @@ -807,19 +800,19 @@ class SyncHandler(object): # about them). state_filter = StateFilter.all() - state_at_previous_sync = yield self.get_state_at( + state_at_previous_sync = await self.get_state_at( room_id, stream_position=since_token, state_filter=state_filter ) if batch: - current_state_ids = yield self.state_store.get_state_ids_for_event( + current_state_ids = await self.state_store.get_state_ids_for_event( batch.events[-1].event_id, state_filter=state_filter ) else: # Its not clear how we get here, but empirically we do # (#5407). Logging has been added elsewhere to try and # figure out where this state comes from. - current_state_ids = yield self.get_state_at( + current_state_ids = await self.get_state_at( room_id, stream_position=now_token, state_filter=state_filter ) @@ -843,7 +836,7 @@ class SyncHandler(object): # So we fish out all the member events corresponding to the # timeline here, and then dedupe any redundant ones below. - state_ids = yield self.state_store.get_state_ids_for_event( + state_ids = await self.state_store.get_state_ids_for_event( batch.events[0].event_id, # we only want members! state_filter=StateFilter.from_types( @@ -883,7 +876,7 @@ class SyncHandler(object): state = {} if state_ids: - state = yield self.store.get_events(list(state_ids.values())) + state = await self.store.get_events(list(state_ids.values())) return { (e.type, e.state_key): e @@ -892,10 +885,9 @@ class SyncHandler(object): ) } - @defer.inlineCallbacks - def unread_notifs_for_room_id(self, room_id, sync_config): + async def unread_notifs_for_room_id(self, room_id, sync_config): with Measure(self.clock, "unread_notifs_for_room_id"): - last_unread_event_id = yield self.store.get_last_receipt_event_id_for_user( + last_unread_event_id = await self.store.get_last_receipt_event_id_for_user( user_id=sync_config.user.to_string(), room_id=room_id, receipt_type="m.read", @@ -903,7 +895,7 @@ class SyncHandler(object): notifs = [] if last_unread_event_id: - notifs = yield self.store.get_unread_event_push_actions_by_room_for_user( + notifs = await self.store.get_unread_event_push_actions_by_room_for_user( room_id, sync_config.user.to_string(), last_unread_event_id ) return notifs @@ -912,8 +904,9 @@ class SyncHandler(object): # count is whatever it was last time. return None - @defer.inlineCallbacks - def generate_sync_result(self, sync_config, since_token=None, full_state=False): + async def generate_sync_result( + self, sync_config, since_token=None, full_state=False + ): """Generates a sync result. Args: @@ -928,7 +921,7 @@ class SyncHandler(object): # this is due to some of the underlying streams not supporting the ability # to query up to a given point. # Always use the `now_token` in `SyncResultBuilder` - now_token = yield self.event_sources.get_current_token() + now_token = await self.event_sources.get_current_token() logger.info( "Calculating sync response for %r between %s and %s", @@ -944,10 +937,9 @@ class SyncHandler(object): # See https://github.com/matrix-org/matrix-doc/issues/1144 raise NotImplementedError() else: - joined_room_ids = yield self.get_rooms_for_user_at( + joined_room_ids = await self.get_rooms_for_user_at( user_id, now_token.room_stream_id ) - sync_result_builder = SyncResultBuilder( sync_config, full_state, @@ -956,11 +948,11 @@ class SyncHandler(object): joined_room_ids=joined_room_ids, ) - account_data_by_room = yield self._generate_sync_entry_for_account_data( + account_data_by_room = await self._generate_sync_entry_for_account_data( sync_result_builder ) - res = yield self._generate_sync_entry_for_rooms( + res = await self._generate_sync_entry_for_rooms( sync_result_builder, account_data_by_room ) newly_joined_rooms, newly_joined_or_invited_users, _, _ = res @@ -970,13 +962,13 @@ class SyncHandler(object): since_token is None and sync_config.filter_collection.blocks_all_presence() ) if self.hs_config.use_presence and not block_all_presence_data: - yield self._generate_sync_entry_for_presence( + await self._generate_sync_entry_for_presence( sync_result_builder, newly_joined_rooms, newly_joined_or_invited_users ) - yield self._generate_sync_entry_for_to_device(sync_result_builder) + await self._generate_sync_entry_for_to_device(sync_result_builder) - device_lists = yield self._generate_sync_entry_for_device_list( + device_lists = await self._generate_sync_entry_for_device_list( sync_result_builder, newly_joined_rooms=newly_joined_rooms, newly_joined_or_invited_users=newly_joined_or_invited_users, @@ -987,11 +979,11 @@ class SyncHandler(object): device_id = sync_config.device_id one_time_key_counts = {} if device_id: - one_time_key_counts = yield self.store.count_e2e_one_time_keys( + one_time_key_counts = await self.store.count_e2e_one_time_keys( user_id, device_id ) - yield self._generate_sync_entry_for_groups(sync_result_builder) + await self._generate_sync_entry_for_groups(sync_result_builder) # debug for https://github.com/matrix-org/synapse/issues/4422 for joined_room in sync_result_builder.joined: @@ -1015,18 +1007,17 @@ class SyncHandler(object): ) @measure_func("_generate_sync_entry_for_groups") - @defer.inlineCallbacks - def _generate_sync_entry_for_groups(self, sync_result_builder): + async def _generate_sync_entry_for_groups(self, sync_result_builder): user_id = sync_result_builder.sync_config.user.to_string() since_token = sync_result_builder.since_token now_token = sync_result_builder.now_token if since_token and since_token.groups_key: - results = yield self.store.get_groups_changes_for_user( + results = self.store.get_groups_changes_for_user( user_id, since_token.groups_key, now_token.groups_key ) else: - results = yield self.store.get_all_groups_for_user( + results = await self.store.get_all_groups_for_user( user_id, now_token.groups_key ) @@ -1059,8 +1050,7 @@ class SyncHandler(object): ) @measure_func("_generate_sync_entry_for_device_list") - @defer.inlineCallbacks - def _generate_sync_entry_for_device_list( + async def _generate_sync_entry_for_device_list( self, sync_result_builder, newly_joined_rooms, @@ -1108,32 +1098,32 @@ class SyncHandler(object): # room with by looking at all users that have left a room plus users # that were in a room we've left. - users_who_share_room = yield self.store.get_users_who_share_room_with_user( + users_who_share_room = await self.store.get_users_who_share_room_with_user( user_id ) # Step 1a, check for changes in devices of users we share a room with - users_that_have_changed = yield self.store.get_users_whose_devices_changed( + users_that_have_changed = await self.store.get_users_whose_devices_changed( since_token.device_list_key, users_who_share_room ) # Step 1b, check for newly joined rooms for room_id in newly_joined_rooms: - joined_users = yield self.state.get_current_users_in_room(room_id) + joined_users = await self.state.get_current_users_in_room(room_id) newly_joined_or_invited_users.update(joined_users) # TODO: Check that these users are actually new, i.e. either they # weren't in the previous sync *or* they left and rejoined. users_that_have_changed.update(newly_joined_or_invited_users) - user_signatures_changed = yield self.store.get_users_whose_signatures_changed( + user_signatures_changed = await self.store.get_users_whose_signatures_changed( user_id, since_token.device_list_key ) users_that_have_changed.update(user_signatures_changed) # Now find users that we no longer track for room_id in newly_left_rooms: - left_users = yield self.state.get_current_users_in_room(room_id) + left_users = await self.state.get_current_users_in_room(room_id) newly_left_users.update(left_users) # Remove any users that we still share a room with. @@ -1143,8 +1133,7 @@ class SyncHandler(object): else: return DeviceLists(changed=[], left=[]) - @defer.inlineCallbacks - def _generate_sync_entry_for_to_device(self, sync_result_builder): + async def _generate_sync_entry_for_to_device(self, sync_result_builder): """Generates the portion of the sync response. Populates `sync_result_builder` with the result. @@ -1165,14 +1154,14 @@ class SyncHandler(object): # We only delete messages when a new message comes in, but that's # fine so long as we delete them at some point. - deleted = yield self.store.delete_messages_for_device( + deleted = await self.store.delete_messages_for_device( user_id, device_id, since_stream_id ) logger.debug( "Deleted %d to-device messages up to %d", deleted, since_stream_id ) - messages, stream_id = yield self.store.get_new_messages_for_device( + messages, stream_id = await self.store.get_new_messages_for_device( user_id, device_id, since_stream_id, now_token.to_device_key ) @@ -1190,8 +1179,7 @@ class SyncHandler(object): else: sync_result_builder.to_device = [] - @defer.inlineCallbacks - def _generate_sync_entry_for_account_data(self, sync_result_builder): + async def _generate_sync_entry_for_account_data(self, sync_result_builder): """Generates the account data portion of the sync response. Populates `sync_result_builder` with the result. @@ -1209,25 +1197,25 @@ class SyncHandler(object): ( account_data, account_data_by_room, - ) = yield self.store.get_updated_account_data_for_user( + ) = self.store.get_updated_account_data_for_user( user_id, since_token.account_data_key ) - push_rules_changed = yield self.store.have_push_rules_changed_for_user( + push_rules_changed = await self.store.have_push_rules_changed_for_user( user_id, int(since_token.push_rules_key) ) if push_rules_changed: - account_data["m.push_rules"] = yield self.push_rules_for_user( + account_data["m.push_rules"] = await self.push_rules_for_user( sync_config.user ) else: ( account_data, account_data_by_room, - ) = yield self.store.get_account_data_for_user(sync_config.user.to_string()) + ) = await self.store.get_account_data_for_user(sync_config.user.to_string()) - account_data["m.push_rules"] = yield self.push_rules_for_user( + account_data["m.push_rules"] = await self.push_rules_for_user( sync_config.user ) @@ -1242,8 +1230,7 @@ class SyncHandler(object): return account_data_by_room - @defer.inlineCallbacks - def _generate_sync_entry_for_presence( + async def _generate_sync_entry_for_presence( self, sync_result_builder, newly_joined_rooms, newly_joined_or_invited_users ): """Generates the presence portion of the sync response. Populates the @@ -1271,7 +1258,7 @@ class SyncHandler(object): presence_key = None include_offline = False - presence, presence_key = yield presence_source.get_new_events( + presence, presence_key = await presence_source.get_new_events( user=user, from_key=presence_key, is_guest=sync_config.is_guest, @@ -1283,12 +1270,12 @@ class SyncHandler(object): extra_users_ids = set(newly_joined_or_invited_users) for room_id in newly_joined_rooms: - users = yield self.state.get_current_users_in_room(room_id) + users = await self.state.get_current_users_in_room(room_id) extra_users_ids.update(users) extra_users_ids.discard(user.to_string()) if extra_users_ids: - states = yield self.presence_handler.get_states(extra_users_ids) + states = await self.presence_handler.get_states(extra_users_ids) presence.extend(states) # Deduplicate the presence entries so that there's at most one per user @@ -1298,8 +1285,9 @@ class SyncHandler(object): sync_result_builder.presence = presence - @defer.inlineCallbacks - def _generate_sync_entry_for_rooms(self, sync_result_builder, account_data_by_room): + async def _generate_sync_entry_for_rooms( + self, sync_result_builder, account_data_by_room + ): """Generates the rooms portion of the sync response. Populates the `sync_result_builder` with the result. @@ -1321,7 +1309,7 @@ class SyncHandler(object): if block_all_room_ephemeral: ephemeral_by_room = {} else: - now_token, ephemeral_by_room = yield self.ephemeral_by_room( + now_token, ephemeral_by_room = await self.ephemeral_by_room( sync_result_builder, now_token=sync_result_builder.now_token, since_token=sync_result_builder.since_token, @@ -1333,16 +1321,16 @@ class SyncHandler(object): since_token = sync_result_builder.since_token if not sync_result_builder.full_state: if since_token and not ephemeral_by_room and not account_data_by_room: - have_changed = yield self._have_rooms_changed(sync_result_builder) + have_changed = await self._have_rooms_changed(sync_result_builder) if not have_changed: - tags_by_room = yield self.store.get_updated_tags( + tags_by_room = await self.store.get_updated_tags( user_id, since_token.account_data_key ) if not tags_by_room: logger.debug("no-oping sync") return [], [], [], [] - ignored_account_data = yield self.store.get_global_account_data_by_type_for_user( + ignored_account_data = await self.store.get_global_account_data_by_type_for_user( "m.ignored_user_list", user_id=user_id ) @@ -1352,18 +1340,18 @@ class SyncHandler(object): ignored_users = frozenset() if since_token: - res = yield self._get_rooms_changed(sync_result_builder, ignored_users) + res = await self._get_rooms_changed(sync_result_builder, ignored_users) room_entries, invited, newly_joined_rooms, newly_left_rooms = res - tags_by_room = yield self.store.get_updated_tags( + tags_by_room = await self.store.get_updated_tags( user_id, since_token.account_data_key ) else: - res = yield self._get_all_rooms(sync_result_builder, ignored_users) + res = await self._get_all_rooms(sync_result_builder, ignored_users) room_entries, invited, newly_joined_rooms = res newly_left_rooms = [] - tags_by_room = yield self.store.get_tags_for_user(user_id) + tags_by_room = await self.store.get_tags_for_user(user_id) def handle_room_entries(room_entry): return self._generate_room_entry( @@ -1376,7 +1364,7 @@ class SyncHandler(object): always_include=sync_result_builder.full_state, ) - yield concurrently_execute(handle_room_entries, room_entries, 10) + await concurrently_execute(handle_room_entries, room_entries, 10) sync_result_builder.invited.extend(invited) @@ -1410,8 +1398,7 @@ class SyncHandler(object): newly_left_users, ) - @defer.inlineCallbacks - def _have_rooms_changed(self, sync_result_builder): + async def _have_rooms_changed(self, sync_result_builder): """Returns whether there may be any new events that should be sent down the sync. Returns True if there are. """ @@ -1422,7 +1409,7 @@ class SyncHandler(object): assert since_token # Get a list of membership change events that have happened. - rooms_changed = yield self.store.get_membership_changes_for_user( + rooms_changed = await self.store.get_membership_changes_for_user( user_id, since_token.room_key, now_token.room_key ) @@ -1435,8 +1422,7 @@ class SyncHandler(object): return True return False - @defer.inlineCallbacks - def _get_rooms_changed(self, sync_result_builder, ignored_users): + async def _get_rooms_changed(self, sync_result_builder, ignored_users): """Gets the the changes that have happened since the last sync. Args: @@ -1461,7 +1447,7 @@ class SyncHandler(object): assert since_token # Get a list of membership change events that have happened. - rooms_changed = yield self.store.get_membership_changes_for_user( + rooms_changed = await self.store.get_membership_changes_for_user( user_id, since_token.room_key, now_token.room_key ) @@ -1499,11 +1485,11 @@ class SyncHandler(object): continue if room_id in sync_result_builder.joined_room_ids or has_join: - old_state_ids = yield self.get_state_at(room_id, since_token) + old_state_ids = await self.get_state_at(room_id, since_token) old_mem_ev_id = old_state_ids.get((EventTypes.Member, user_id), None) old_mem_ev = None if old_mem_ev_id: - old_mem_ev = yield self.store.get_event( + old_mem_ev = await self.store.get_event( old_mem_ev_id, allow_none=True ) @@ -1536,13 +1522,13 @@ class SyncHandler(object): newly_left_rooms.append(room_id) else: if not old_state_ids: - old_state_ids = yield self.get_state_at(room_id, since_token) + old_state_ids = await self.get_state_at(room_id, since_token) old_mem_ev_id = old_state_ids.get( (EventTypes.Member, user_id), None ) old_mem_ev = None if old_mem_ev_id: - old_mem_ev = yield self.store.get_event( + old_mem_ev = await self.store.get_event( old_mem_ev_id, allow_none=True ) if old_mem_ev and old_mem_ev.membership == Membership.JOIN: @@ -1566,7 +1552,7 @@ class SyncHandler(object): if leave_events: leave_event = leave_events[-1] - leave_stream_token = yield self.store.get_stream_token_for_event( + leave_stream_token = await self.store.get_stream_token_for_event( leave_event.event_id ) leave_token = since_token.copy_and_replace( @@ -1603,7 +1589,7 @@ class SyncHandler(object): timeline_limit = sync_config.filter_collection.timeline_limit() # Get all events for rooms we're currently joined to. - room_to_events = yield self.store.get_room_events_stream_for_rooms( + room_to_events = await self.store.get_room_events_stream_for_rooms( room_ids=sync_result_builder.joined_room_ids, from_key=since_token.room_key, to_key=now_token.room_key, @@ -1652,8 +1638,7 @@ class SyncHandler(object): return room_entries, invited, newly_joined_rooms, newly_left_rooms - @defer.inlineCallbacks - def _get_all_rooms(self, sync_result_builder, ignored_users): + async def _get_all_rooms(self, sync_result_builder, ignored_users): """Returns entries for all rooms for the user. Args: @@ -1677,7 +1662,7 @@ class SyncHandler(object): Membership.BAN, ) - room_list = yield self.store.get_rooms_for_user_where_membership_is( + room_list = await self.store.get_rooms_for_user_where_membership_is( user_id=user_id, membership_list=membership_list ) @@ -1700,7 +1685,7 @@ class SyncHandler(object): elif event.membership == Membership.INVITE: if event.sender in ignored_users: continue - invite = yield self.store.get_event(event.event_id) + invite = await self.store.get_event(event.event_id) invited.append(InvitedSyncResult(room_id=event.room_id, invite=invite)) elif event.membership in (Membership.LEAVE, Membership.BAN): # Always send down rooms we were banned or kicked from. @@ -1726,8 +1711,7 @@ class SyncHandler(object): return room_entries, invited, [] - @defer.inlineCallbacks - def _generate_room_entry( + async def _generate_room_entry( self, sync_result_builder, ignored_users, @@ -1769,7 +1753,7 @@ class SyncHandler(object): since_token = room_builder.since_token upto_token = room_builder.upto_token - batch = yield self._load_filtered_recents( + batch = await self._load_filtered_recents( room_id, sync_config, now_token=upto_token, @@ -1796,7 +1780,7 @@ class SyncHandler(object): # tag was added by synapse e.g. for server notice rooms. if full_state: user_id = sync_result_builder.sync_config.user.to_string() - tags = yield self.store.get_tags_for_room(user_id, room_id) + tags = await self.store.get_tags_for_room(user_id, room_id) # If there aren't any tags, don't send the empty tags list down # sync @@ -1821,7 +1805,7 @@ class SyncHandler(object): ): return - state = yield self.compute_state_delta( + state = await self.compute_state_delta( room_id, batch, sync_config, since_token, now_token, full_state=full_state ) @@ -1844,7 +1828,7 @@ class SyncHandler(object): ) or since_token is None ): - summary = yield self.compute_summary( + summary = await self.compute_summary( room_id, sync_config, batch, state, now_token ) @@ -1861,7 +1845,7 @@ class SyncHandler(object): ) if room_sync or always_include: - notifs = yield self.unread_notifs_for_room_id(room_id, sync_config) + notifs = await self.unread_notifs_for_room_id(room_id, sync_config) if notifs is not None: unread_notifications["notification_count"] = notifs["notify_count"] @@ -1887,8 +1871,7 @@ class SyncHandler(object): else: raise Exception("Unrecognized rtype: %r", room_builder.rtype) - @defer.inlineCallbacks - def get_rooms_for_user_at(self, user_id, stream_ordering): + async def get_rooms_for_user_at(self, user_id, stream_ordering): """Get set of joined rooms for a user at the given stream ordering. The stream ordering *must* be recent, otherwise this may throw an @@ -1903,7 +1886,7 @@ class SyncHandler(object): Deferred[frozenset[str]]: Set of room_ids the user is in at given stream_ordering. """ - joined_rooms = yield self.store.get_rooms_for_user_with_stream_ordering(user_id) + joined_rooms = await self.store.get_rooms_for_user_with_stream_ordering(user_id) joined_room_ids = set() @@ -1921,10 +1904,10 @@ class SyncHandler(object): logger.info("User joined room after current token: %s", room_id) - extrems = yield self.store.get_forward_extremeties_for_room( + extrems = await self.store.get_forward_extremeties_for_room( room_id, stream_ordering ) - users_in_room = yield self.state.get_current_users_in_room(room_id, extrems) + users_in_room = await self.state.get_current_users_in_room(room_id, extrems) if user_id in users_in_room: joined_room_ids.add(room_id) diff --git a/synapse/notifier.py b/synapse/notifier.py index af161a81d7..5f5f765bea 100644 --- a/synapse/notifier.py +++ b/synapse/notifier.py @@ -304,8 +304,7 @@ class Notifier(object): without waking up any of the normal user event streams""" self.notify_replication() - @defer.inlineCallbacks - def wait_for_events( + async def wait_for_events( self, user_id, timeout, callback, room_ids=None, from_token=StreamToken.START ): """Wait until the callback returns a non empty response or the @@ -313,9 +312,9 @@ class Notifier(object): """ user_stream = self.user_to_user_stream.get(user_id) if user_stream is None: - current_token = yield self.event_sources.get_current_token() + current_token = await self.event_sources.get_current_token() if room_ids is None: - room_ids = yield self.store.get_rooms_for_user(user_id) + room_ids = await self.store.get_rooms_for_user(user_id) user_stream = _NotifierUserStream( user_id=user_id, rooms=room_ids, @@ -344,11 +343,11 @@ class Notifier(object): self.hs.get_reactor(), ) with PreserveLoggingContext(): - yield listener.deferred + await listener.deferred current_token = user_stream.current_token - result = yield callback(prev_token, current_token) + result = await callback(prev_token, current_token) if result: break @@ -364,12 +363,11 @@ class Notifier(object): # This happened if there was no timeout or if the timeout had # already expired. current_token = user_stream.current_token - result = yield callback(prev_token, current_token) + result = await callback(prev_token, current_token) return result - @defer.inlineCallbacks - def get_events_for( + async def get_events_for( self, user, pagination_config, @@ -391,15 +389,14 @@ class Notifier(object): """ from_token = pagination_config.from_token if not from_token: - from_token = yield self.event_sources.get_current_token() + from_token = await self.event_sources.get_current_token() limit = pagination_config.limit - room_ids, is_joined = yield self._get_room_ids(user, explicit_room_id) + room_ids, is_joined = await self._get_room_ids(user, explicit_room_id) is_peeking = not is_joined - @defer.inlineCallbacks - def check_for_updates(before_token, after_token): + async def check_for_updates(before_token, after_token): if not after_token.is_after(before_token): return EventStreamResult([], (from_token, from_token)) @@ -415,7 +412,7 @@ class Notifier(object): if only_keys and name not in only_keys: continue - new_events, new_key = yield source.get_new_events( + new_events, new_key = await source.get_new_events( user=user, from_key=getattr(from_token, keyname), limit=limit, @@ -425,7 +422,7 @@ class Notifier(object): ) if name == "room": - new_events = yield filter_events_for_client( + new_events = await filter_events_for_client( self.storage, user.to_string(), new_events, @@ -461,7 +458,7 @@ class Notifier(object): user_id_for_stream, ) - result = yield self.wait_for_events( + result = await self.wait_for_events( user_id_for_stream, timeout, check_for_updates, diff --git a/synapse/util/metrics.py b/synapse/util/metrics.py index 3286804322..63ddaaba87 100644 --- a/synapse/util/metrics.py +++ b/synapse/util/metrics.py @@ -13,6 +13,7 @@ # See the License for the specific language governing permissions and # limitations under the License. +import inspect import logging from functools import wraps @@ -64,12 +65,22 @@ def measure_func(name=None): def wrapper(func): block_name = func.__name__ if name is None else name - @wraps(func) - @defer.inlineCallbacks - def measured_func(self, *args, **kwargs): - with Measure(self.clock, block_name): - r = yield func(self, *args, **kwargs) - return r + if inspect.iscoroutinefunction(func): + + @wraps(func) + async def measured_func(self, *args, **kwargs): + with Measure(self.clock, block_name): + r = await func(self, *args, **kwargs) + return r + + else: + + @wraps(func) + @defer.inlineCallbacks + def measured_func(self, *args, **kwargs): + with Measure(self.clock, block_name): + r = yield func(self, *args, **kwargs) + return r return measured_func diff --git a/tests/handlers/test_sync.py b/tests/handlers/test_sync.py index 31f54bbd7d..758ee071a5 100644 --- a/tests/handlers/test_sync.py +++ b/tests/handlers/test_sync.py @@ -12,54 +12,53 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. -from twisted.internet import defer from synapse.api.errors import Codes, ResourceLimitError from synapse.api.filtering import DEFAULT_FILTER_COLLECTION -from synapse.handlers.sync import SyncConfig, SyncHandler +from synapse.handlers.sync import SyncConfig from synapse.types import UserID import tests.unittest import tests.utils -from tests.utils import setup_test_homeserver -class SyncTestCase(tests.unittest.TestCase): +class SyncTestCase(tests.unittest.HomeserverTestCase): """ Tests Sync Handler. """ - @defer.inlineCallbacks - def setUp(self): - self.hs = yield setup_test_homeserver(self.addCleanup) - self.sync_handler = SyncHandler(self.hs) + def prepare(self, reactor, clock, hs): + self.hs = hs + self.sync_handler = self.hs.get_sync_handler() self.store = self.hs.get_datastore() - @defer.inlineCallbacks def test_wait_for_sync_for_user_auth_blocking(self): user_id1 = "@user1:server" user_id2 = "@user2:server" sync_config = self._generate_sync_config(user_id1) + self.reactor.advance(100) # So we get not 0 time self.hs.config.limit_usage_by_mau = True self.hs.config.max_mau_value = 1 # Check that the happy case does not throw errors - yield self.store.upsert_monthly_active_user(user_id1) - yield self.sync_handler.wait_for_sync_for_user(sync_config) + self.get_success(self.store.upsert_monthly_active_user(user_id1)) + self.get_success(self.sync_handler.wait_for_sync_for_user(sync_config)) # Test that global lock works self.hs.config.hs_disabled = True - with self.assertRaises(ResourceLimitError) as e: - yield self.sync_handler.wait_for_sync_for_user(sync_config) - self.assertEquals(e.exception.errcode, Codes.RESOURCE_LIMIT_EXCEEDED) + e = self.get_failure( + self.sync_handler.wait_for_sync_for_user(sync_config), ResourceLimitError + ) + self.assertEquals(e.value.errcode, Codes.RESOURCE_LIMIT_EXCEEDED) self.hs.config.hs_disabled = False sync_config = self._generate_sync_config(user_id2) - with self.assertRaises(ResourceLimitError) as e: - yield self.sync_handler.wait_for_sync_for_user(sync_config) - self.assertEquals(e.exception.errcode, Codes.RESOURCE_LIMIT_EXCEEDED) + e = self.get_failure( + self.sync_handler.wait_for_sync_for_user(sync_config), ResourceLimitError + ) + self.assertEquals(e.value.errcode, Codes.RESOURCE_LIMIT_EXCEEDED) def _generate_sync_config(self, user_id): return SyncConfig( diff --git a/tests/unittest.py b/tests/unittest.py index 295573bc46..a1bdd963e6 100644 --- a/tests/unittest.py +++ b/tests/unittest.py @@ -18,6 +18,7 @@ import gc import hashlib import hmac +import inspect import logging import time @@ -25,7 +26,7 @@ from mock import Mock from canonicaljson import json -from twisted.internet.defer import Deferred, succeed +from twisted.internet.defer import Deferred, ensureDeferred, succeed from twisted.python.threadpool import ThreadPool from twisted.trial import unittest @@ -415,6 +416,8 @@ class HomeserverTestCase(TestCase): self.reactor.pump([by] * 100) def get_success(self, d, by=0.0): + if inspect.isawaitable(d): + d = ensureDeferred(d) if not isinstance(d, Deferred): return d self.pump(by=by) @@ -424,6 +427,8 @@ class HomeserverTestCase(TestCase): """ Run a Deferred and get a Failure from it. The failure must be of the type `exc`. """ + if inspect.isawaitable(d): + d = ensureDeferred(d) if not isinstance(d, Deferred): return d self.pump() -- cgit 1.4.1 From b3a4e35ca84a29fe4ccdfb1125ed098c68405d6c Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Fri, 6 Dec 2019 10:14:59 +0000 Subject: Fixup functions to consistently return deferreds --- synapse/handlers/sync.py | 6 +++--- synapse/handlers/typing.py | 2 +- synapse/storage/data_stores/main/account_data.py | 2 +- synapse/storage/data_stores/main/group_server.py | 4 ++-- tests/handlers/test_typing.py | 24 ++++++++++++++++++------ tests/rest/client/v1/test_typing.py | 4 +++- 6 files changed, 28 insertions(+), 14 deletions(-) (limited to 'tests') diff --git a/synapse/handlers/sync.py b/synapse/handlers/sync.py index 12751fd8c0..2d3b8ba73c 100644 --- a/synapse/handlers/sync.py +++ b/synapse/handlers/sync.py @@ -339,7 +339,7 @@ class SyncHandler(object): room_ids = sync_result_builder.joined_room_ids typing_source = self.event_sources.sources["typing"] - typing, typing_key = typing_source.get_new_events( + typing, typing_key = await typing_source.get_new_events( user=sync_config.user, from_key=typing_key, limit=sync_config.filter_collection.ephemeral_limit(), @@ -1013,7 +1013,7 @@ class SyncHandler(object): now_token = sync_result_builder.now_token if since_token and since_token.groups_key: - results = self.store.get_groups_changes_for_user( + results = await self.store.get_groups_changes_for_user( user_id, since_token.groups_key, now_token.groups_key ) else: @@ -1197,7 +1197,7 @@ class SyncHandler(object): ( account_data, account_data_by_room, - ) = self.store.get_updated_account_data_for_user( + ) = await self.store.get_updated_account_data_for_user( user_id, since_token.account_data_key ) diff --git a/synapse/handlers/typing.py b/synapse/handlers/typing.py index 856337b7e2..6f78454322 100644 --- a/synapse/handlers/typing.py +++ b/synapse/handlers/typing.py @@ -313,7 +313,7 @@ class TypingNotificationEventSource(object): events.append(self._make_event_for(room_id)) - return events, handler._latest_room_serial + return defer.succeed((events, handler._latest_room_serial)) def get_current_key(self): return self.get_typing_handler()._latest_room_serial diff --git a/synapse/storage/data_stores/main/account_data.py b/synapse/storage/data_stores/main/account_data.py index b0d22faf3f..ed97b3ffe5 100644 --- a/synapse/storage/data_stores/main/account_data.py +++ b/synapse/storage/data_stores/main/account_data.py @@ -250,7 +250,7 @@ class AccountDataWorkerStore(SQLBaseStore): user_id, int(stream_id) ) if not changed: - return {}, {} + return defer.succeed(({}, {})) return self.runInteraction( "get_updated_account_data_for_user", get_updated_account_data_for_user_txn diff --git a/synapse/storage/data_stores/main/group_server.py b/synapse/storage/data_stores/main/group_server.py index 9e1d12bcb7..d29155a3b5 100644 --- a/synapse/storage/data_stores/main/group_server.py +++ b/synapse/storage/data_stores/main/group_server.py @@ -1109,7 +1109,7 @@ class GroupServerStore(SQLBaseStore): user_id, from_token ) if not has_changed: - return [] + return defer.succeed([]) def _get_groups_changes_for_user_txn(txn): sql = """ @@ -1139,7 +1139,7 @@ class GroupServerStore(SQLBaseStore): from_token ) if not has_changed: - return [] + return defer.succeed([]) def _get_all_groups_changes_txn(txn): sql = """ diff --git a/tests/handlers/test_typing.py b/tests/handlers/test_typing.py index f6d8660285..92b8726093 100644 --- a/tests/handlers/test_typing.py +++ b/tests/handlers/test_typing.py @@ -163,7 +163,9 @@ class TypingNotificationsTestCase(unittest.HomeserverTestCase): self.on_new_event.assert_has_calls([call("typing_key", 1, rooms=[ROOM_ID])]) self.assertEquals(self.event_source.get_current_key(), 1) - events = self.event_source.get_new_events(room_ids=[ROOM_ID], from_key=0) + events = self.get_success( + self.event_source.get_new_events(room_ids=[ROOM_ID], from_key=0) + ) self.assertEquals( events[0], [ @@ -227,7 +229,9 @@ class TypingNotificationsTestCase(unittest.HomeserverTestCase): self.on_new_event.assert_has_calls([call("typing_key", 1, rooms=[ROOM_ID])]) self.assertEquals(self.event_source.get_current_key(), 1) - events = self.event_source.get_new_events(room_ids=[ROOM_ID], from_key=0) + events = self.get_success( + self.event_source.get_new_events(room_ids=[ROOM_ID], from_key=0) + ) self.assertEquals( events[0], [ @@ -279,7 +283,9 @@ class TypingNotificationsTestCase(unittest.HomeserverTestCase): ) self.assertEquals(self.event_source.get_current_key(), 1) - events = self.event_source.get_new_events(room_ids=[ROOM_ID], from_key=0) + events = self.get_success( + self.event_source.get_new_events(room_ids=[ROOM_ID], from_key=0) + ) self.assertEquals( events[0], [{"type": "m.typing", "room_id": ROOM_ID, "content": {"user_ids": []}}], @@ -300,7 +306,9 @@ class TypingNotificationsTestCase(unittest.HomeserverTestCase): self.on_new_event.reset_mock() self.assertEquals(self.event_source.get_current_key(), 1) - events = self.event_source.get_new_events(room_ids=[ROOM_ID], from_key=0) + events = self.get_success( + self.event_source.get_new_events(room_ids=[ROOM_ID], from_key=0) + ) self.assertEquals( events[0], [ @@ -317,7 +325,9 @@ class TypingNotificationsTestCase(unittest.HomeserverTestCase): self.on_new_event.assert_has_calls([call("typing_key", 2, rooms=[ROOM_ID])]) self.assertEquals(self.event_source.get_current_key(), 2) - events = self.event_source.get_new_events(room_ids=[ROOM_ID], from_key=1) + events = self.get_success( + self.event_source.get_new_events(room_ids=[ROOM_ID], from_key=1) + ) self.assertEquals( events[0], [{"type": "m.typing", "room_id": ROOM_ID, "content": {"user_ids": []}}], @@ -335,7 +345,9 @@ class TypingNotificationsTestCase(unittest.HomeserverTestCase): self.on_new_event.reset_mock() self.assertEquals(self.event_source.get_current_key(), 3) - events = self.event_source.get_new_events(room_ids=[ROOM_ID], from_key=0) + events = self.get_success( + self.event_source.get_new_events(room_ids=[ROOM_ID], from_key=0) + ) self.assertEquals( events[0], [ diff --git a/tests/rest/client/v1/test_typing.py b/tests/rest/client/v1/test_typing.py index 30fb77bac8..4bc3aaf02d 100644 --- a/tests/rest/client/v1/test_typing.py +++ b/tests/rest/client/v1/test_typing.py @@ -109,7 +109,9 @@ class RoomTypingTestCase(unittest.HomeserverTestCase): self.assertEquals(200, channel.code) self.assertEquals(self.event_source.get_current_key(), 1) - events = self.event_source.get_new_events(from_key=0, room_ids=[self.room_id]) + events = self.get_success( + self.event_source.get_new_events(from_key=0, room_ids=[self.room_id]) + ) self.assertEquals( events[0], [ -- cgit 1.4.1 From 9a4fb457cf5918c85068ea249cd2d58b3e2e3cfc Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Fri, 6 Dec 2019 13:08:40 +0000 Subject: Change DataStores to accept 'database' param. --- synapse/app/federation_sender.py | 5 +++-- synapse/app/user_dir.py | 5 +++-- synapse/replication/slave/storage/_base.py | 5 +++-- synapse/replication/slave/storage/account_data.py | 5 +++-- synapse/replication/slave/storage/client_ips.py | 5 +++-- synapse/replication/slave/storage/deviceinbox.py | 5 +++-- synapse/replication/slave/storage/devices.py | 5 +++-- synapse/replication/slave/storage/events.py | 5 +++-- synapse/replication/slave/storage/filtering.py | 5 +++-- synapse/replication/slave/storage/groups.py | 5 +++-- synapse/replication/slave/storage/presence.py | 5 +++-- synapse/replication/slave/storage/push_rule.py | 5 +++-- synapse/replication/slave/storage/pushers.py | 5 +++-- synapse/replication/slave/storage/receipts.py | 5 +++-- synapse/replication/slave/storage/room.py | 5 +++-- synapse/storage/_base.py | 2 +- synapse/storage/data_stores/main/__init__.py | 5 +++-- synapse/storage/data_stores/main/account_data.py | 9 +++++---- synapse/storage/data_stores/main/appservice.py | 5 +++-- synapse/storage/data_stores/main/client_ips.py | 9 +++++---- synapse/storage/data_stores/main/deviceinbox.py | 9 +++++---- synapse/storage/data_stores/main/devices.py | 9 +++++---- synapse/storage/data_stores/main/event_federation.py | 5 +++-- synapse/storage/data_stores/main/event_push_actions.py | 9 +++++---- synapse/storage/data_stores/main/events.py | 5 +++-- synapse/storage/data_stores/main/events_bg_updates.py | 5 +++-- synapse/storage/data_stores/main/events_worker.py | 5 +++-- synapse/storage/data_stores/main/media_repository.py | 11 +++++++---- synapse/storage/data_stores/main/monthly_active_users.py | 7 ++++--- synapse/storage/data_stores/main/push_rule.py | 5 +++-- synapse/storage/data_stores/main/receipts.py | 9 +++++---- synapse/storage/data_stores/main/registration.py | 13 +++++++------ synapse/storage/data_stores/main/room.py | 9 +++++---- synapse/storage/data_stores/main/roommember.py | 13 +++++++------ synapse/storage/data_stores/main/search.py | 9 +++++---- synapse/storage/data_stores/main/state.py | 13 +++++++------ synapse/storage/data_stores/main/stats.py | 5 +++-- synapse/storage/data_stores/main/stream.py | 5 +++-- synapse/storage/data_stores/main/transactions.py | 5 +++-- synapse/storage/data_stores/main/user_directory.py | 9 +++++---- tests/storage/test_appservice.py | 5 +++-- 41 files changed, 156 insertions(+), 114 deletions(-) (limited to 'tests') diff --git a/synapse/app/federation_sender.py b/synapse/app/federation_sender.py index 448e45e00f..f24920a7d6 100644 --- a/synapse/app/federation_sender.py +++ b/synapse/app/federation_sender.py @@ -40,6 +40,7 @@ from synapse.replication.slave.storage.transactions import SlavedTransactionStor from synapse.replication.tcp.client import ReplicationClientHandler from synapse.replication.tcp.streams._base import ReceiptsStream from synapse.server import HomeServer +from synapse.storage.database import Database from synapse.storage.engines import create_engine from synapse.types import ReadReceipt from synapse.util.async_helpers import Linearizer @@ -59,8 +60,8 @@ class FederationSenderSlaveStore( SlavedDeviceStore, SlavedPresenceStore, ): - def __init__(self, db_conn, hs): - super(FederationSenderSlaveStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(FederationSenderSlaveStore, self).__init__(database, db_conn, hs) # We pull out the current federation stream position now so that we # always have a known value for the federation position in memory so diff --git a/synapse/app/user_dir.py b/synapse/app/user_dir.py index b6d4481725..c01fb34a9b 100644 --- a/synapse/app/user_dir.py +++ b/synapse/app/user_dir.py @@ -43,6 +43,7 @@ from synapse.replication.tcp.streams.events import ( from synapse.rest.client.v2_alpha import user_directory from synapse.server import HomeServer from synapse.storage.data_stores.main.user_directory import UserDirectoryStore +from synapse.storage.database import Database from synapse.storage.engines import create_engine from synapse.util.caches.stream_change_cache import StreamChangeCache from synapse.util.httpresourcetree import create_resource_tree @@ -60,8 +61,8 @@ class UserDirectorySlaveStore( UserDirectoryStore, BaseSlavedStore, ): - def __init__(self, db_conn, hs): - super(UserDirectorySlaveStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(UserDirectorySlaveStore, self).__init__(database, db_conn, hs) events_max = self._stream_id_gen.get_current_token() curr_state_delta_prefill, min_curr_state_delta_id = self.db.get_cache_dict( diff --git a/synapse/replication/slave/storage/_base.py b/synapse/replication/slave/storage/_base.py index 6ece1d6745..b91a528245 100644 --- a/synapse/replication/slave/storage/_base.py +++ b/synapse/replication/slave/storage/_base.py @@ -20,6 +20,7 @@ import six from synapse.storage._base import SQLBaseStore from synapse.storage.data_stores.main.cache import CURRENT_STATE_CACHE_NAME +from synapse.storage.database import Database from synapse.storage.engines import PostgresEngine from ._slaved_id_tracker import SlavedIdTracker @@ -35,8 +36,8 @@ def __func__(inp): class BaseSlavedStore(SQLBaseStore): - def __init__(self, db_conn, hs): - super(BaseSlavedStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(BaseSlavedStore, self).__init__(database, db_conn, hs) if isinstance(self.database_engine, PostgresEngine): self._cache_id_gen = SlavedIdTracker( db_conn, "cache_invalidation_stream", "stream_id" diff --git a/synapse/replication/slave/storage/account_data.py b/synapse/replication/slave/storage/account_data.py index bc2f6a12ae..ebe94909cb 100644 --- a/synapse/replication/slave/storage/account_data.py +++ b/synapse/replication/slave/storage/account_data.py @@ -18,15 +18,16 @@ from synapse.replication.slave.storage._base import BaseSlavedStore from synapse.replication.slave.storage._slaved_id_tracker import SlavedIdTracker from synapse.storage.data_stores.main.account_data import AccountDataWorkerStore from synapse.storage.data_stores.main.tags import TagsWorkerStore +from synapse.storage.database import Database class SlavedAccountDataStore(TagsWorkerStore, AccountDataWorkerStore, BaseSlavedStore): - def __init__(self, db_conn, hs): + def __init__(self, database: Database, db_conn, hs): self._account_data_id_gen = SlavedIdTracker( db_conn, "account_data_max_stream_id", "stream_id" ) - super(SlavedAccountDataStore, self).__init__(db_conn, hs) + super(SlavedAccountDataStore, self).__init__(database, db_conn, hs) def get_max_account_data_stream_id(self): return self._account_data_id_gen.get_current_token() diff --git a/synapse/replication/slave/storage/client_ips.py b/synapse/replication/slave/storage/client_ips.py index b4f58cea19..fbf996e33a 100644 --- a/synapse/replication/slave/storage/client_ips.py +++ b/synapse/replication/slave/storage/client_ips.py @@ -14,6 +14,7 @@ # limitations under the License. from synapse.storage.data_stores.main.client_ips import LAST_SEEN_GRANULARITY +from synapse.storage.database import Database from synapse.util.caches import CACHE_SIZE_FACTOR from synapse.util.caches.descriptors import Cache @@ -21,8 +22,8 @@ from ._base import BaseSlavedStore class SlavedClientIpStore(BaseSlavedStore): - def __init__(self, db_conn, hs): - super(SlavedClientIpStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(SlavedClientIpStore, self).__init__(database, db_conn, hs) self.client_ip_last_seen = Cache( name="client_ip_last_seen", keylen=4, max_entries=50000 * CACHE_SIZE_FACTOR diff --git a/synapse/replication/slave/storage/deviceinbox.py b/synapse/replication/slave/storage/deviceinbox.py index 9fb6c5c6ff..0c237c6e0f 100644 --- a/synapse/replication/slave/storage/deviceinbox.py +++ b/synapse/replication/slave/storage/deviceinbox.py @@ -16,13 +16,14 @@ from synapse.replication.slave.storage._base import BaseSlavedStore from synapse.replication.slave.storage._slaved_id_tracker import SlavedIdTracker from synapse.storage.data_stores.main.deviceinbox import DeviceInboxWorkerStore +from synapse.storage.database import Database from synapse.util.caches.expiringcache import ExpiringCache from synapse.util.caches.stream_change_cache import StreamChangeCache class SlavedDeviceInboxStore(DeviceInboxWorkerStore, BaseSlavedStore): - def __init__(self, db_conn, hs): - super(SlavedDeviceInboxStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(SlavedDeviceInboxStore, self).__init__(database, db_conn, hs) self._device_inbox_id_gen = SlavedIdTracker( db_conn, "device_max_stream_id", "stream_id" ) diff --git a/synapse/replication/slave/storage/devices.py b/synapse/replication/slave/storage/devices.py index de50748c30..dc625e0d7a 100644 --- a/synapse/replication/slave/storage/devices.py +++ b/synapse/replication/slave/storage/devices.py @@ -18,12 +18,13 @@ from synapse.replication.slave.storage._slaved_id_tracker import SlavedIdTracker from synapse.replication.tcp.streams._base import DeviceListsStream, UserSignatureStream from synapse.storage.data_stores.main.devices import DeviceWorkerStore from synapse.storage.data_stores.main.end_to_end_keys import EndToEndKeyWorkerStore +from synapse.storage.database import Database from synapse.util.caches.stream_change_cache import StreamChangeCache class SlavedDeviceStore(EndToEndKeyWorkerStore, DeviceWorkerStore, BaseSlavedStore): - def __init__(self, db_conn, hs): - super(SlavedDeviceStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(SlavedDeviceStore, self).__init__(database, db_conn, hs) self.hs = hs diff --git a/synapse/replication/slave/storage/events.py b/synapse/replication/slave/storage/events.py index d0a0eaf75b..29f35b9915 100644 --- a/synapse/replication/slave/storage/events.py +++ b/synapse/replication/slave/storage/events.py @@ -31,6 +31,7 @@ from synapse.storage.data_stores.main.signatures import SignatureWorkerStore from synapse.storage.data_stores.main.state import StateGroupWorkerStore from synapse.storage.data_stores.main.stream import StreamWorkerStore from synapse.storage.data_stores.main.user_erasure_store import UserErasureWorkerStore +from synapse.storage.database import Database from ._base import BaseSlavedStore from ._slaved_id_tracker import SlavedIdTracker @@ -59,13 +60,13 @@ class SlavedEventStore( RelationsWorkerStore, BaseSlavedStore, ): - def __init__(self, db_conn, hs): + def __init__(self, database: Database, db_conn, hs): self._stream_id_gen = SlavedIdTracker(db_conn, "events", "stream_ordering") self._backfill_id_gen = SlavedIdTracker( db_conn, "events", "stream_ordering", step=-1 ) - super(SlavedEventStore, self).__init__(db_conn, hs) + super(SlavedEventStore, self).__init__(database, db_conn, hs) # Cached functions can't be accessed through a class instance so we need # to reach inside the __dict__ to extract them. diff --git a/synapse/replication/slave/storage/filtering.py b/synapse/replication/slave/storage/filtering.py index 5c84ebd125..bcb0688954 100644 --- a/synapse/replication/slave/storage/filtering.py +++ b/synapse/replication/slave/storage/filtering.py @@ -14,13 +14,14 @@ # limitations under the License. from synapse.storage.data_stores.main.filtering import FilteringStore +from synapse.storage.database import Database from ._base import BaseSlavedStore class SlavedFilteringStore(BaseSlavedStore): - def __init__(self, db_conn, hs): - super(SlavedFilteringStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(SlavedFilteringStore, self).__init__(database, db_conn, hs) # Filters are immutable so this cache doesn't need to be expired get_user_filter = FilteringStore.__dict__["get_user_filter"] diff --git a/synapse/replication/slave/storage/groups.py b/synapse/replication/slave/storage/groups.py index 28a46edd28..69a4ae42f9 100644 --- a/synapse/replication/slave/storage/groups.py +++ b/synapse/replication/slave/storage/groups.py @@ -14,6 +14,7 @@ # limitations under the License. from synapse.storage import DataStore +from synapse.storage.database import Database from synapse.util.caches.stream_change_cache import StreamChangeCache from ._base import BaseSlavedStore, __func__ @@ -21,8 +22,8 @@ from ._slaved_id_tracker import SlavedIdTracker class SlavedGroupServerStore(BaseSlavedStore): - def __init__(self, db_conn, hs): - super(SlavedGroupServerStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(SlavedGroupServerStore, self).__init__(database, db_conn, hs) self.hs = hs diff --git a/synapse/replication/slave/storage/presence.py b/synapse/replication/slave/storage/presence.py index 747ced0c84..f552e7c972 100644 --- a/synapse/replication/slave/storage/presence.py +++ b/synapse/replication/slave/storage/presence.py @@ -15,6 +15,7 @@ from synapse.storage import DataStore from synapse.storage.data_stores.main.presence import PresenceStore +from synapse.storage.database import Database from synapse.util.caches.stream_change_cache import StreamChangeCache from ._base import BaseSlavedStore, __func__ @@ -22,8 +23,8 @@ from ._slaved_id_tracker import SlavedIdTracker class SlavedPresenceStore(BaseSlavedStore): - def __init__(self, db_conn, hs): - super(SlavedPresenceStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(SlavedPresenceStore, self).__init__(database, db_conn, hs) self._presence_id_gen = SlavedIdTracker(db_conn, "presence_stream", "stream_id") self._presence_on_startup = self._get_active_presence(db_conn) diff --git a/synapse/replication/slave/storage/push_rule.py b/synapse/replication/slave/storage/push_rule.py index 3655f05e54..eebd5a1fb6 100644 --- a/synapse/replication/slave/storage/push_rule.py +++ b/synapse/replication/slave/storage/push_rule.py @@ -15,17 +15,18 @@ # limitations under the License. from synapse.storage.data_stores.main.push_rule import PushRulesWorkerStore +from synapse.storage.database import Database from ._slaved_id_tracker import SlavedIdTracker from .events import SlavedEventStore class SlavedPushRuleStore(SlavedEventStore, PushRulesWorkerStore): - def __init__(self, db_conn, hs): + def __init__(self, database: Database, db_conn, hs): self._push_rules_stream_id_gen = SlavedIdTracker( db_conn, "push_rules_stream", "stream_id" ) - super(SlavedPushRuleStore, self).__init__(db_conn, hs) + super(SlavedPushRuleStore, self).__init__(database, db_conn, hs) def get_push_rules_stream_token(self): return ( diff --git a/synapse/replication/slave/storage/pushers.py b/synapse/replication/slave/storage/pushers.py index b4331d0799..f22c2d44a3 100644 --- a/synapse/replication/slave/storage/pushers.py +++ b/synapse/replication/slave/storage/pushers.py @@ -15,14 +15,15 @@ # limitations under the License. from synapse.storage.data_stores.main.pusher import PusherWorkerStore +from synapse.storage.database import Database from ._base import BaseSlavedStore from ._slaved_id_tracker import SlavedIdTracker class SlavedPusherStore(PusherWorkerStore, BaseSlavedStore): - def __init__(self, db_conn, hs): - super(SlavedPusherStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(SlavedPusherStore, self).__init__(database, db_conn, hs) self._pushers_id_gen = SlavedIdTracker( db_conn, "pushers", "id", extra_tables=[("deleted_pushers", "stream_id")] ) diff --git a/synapse/replication/slave/storage/receipts.py b/synapse/replication/slave/storage/receipts.py index 43d823c601..d40dc6e1f5 100644 --- a/synapse/replication/slave/storage/receipts.py +++ b/synapse/replication/slave/storage/receipts.py @@ -15,6 +15,7 @@ # limitations under the License. from synapse.storage.data_stores.main.receipts import ReceiptsWorkerStore +from synapse.storage.database import Database from ._base import BaseSlavedStore from ._slaved_id_tracker import SlavedIdTracker @@ -29,14 +30,14 @@ from ._slaved_id_tracker import SlavedIdTracker class SlavedReceiptsStore(ReceiptsWorkerStore, BaseSlavedStore): - def __init__(self, db_conn, hs): + def __init__(self, database: Database, db_conn, hs): # We instantiate this first as the ReceiptsWorkerStore constructor # needs to be able to call get_max_receipt_stream_id self._receipts_id_gen = SlavedIdTracker( db_conn, "receipts_linearized", "stream_id" ) - super(SlavedReceiptsStore, self).__init__(db_conn, hs) + super(SlavedReceiptsStore, self).__init__(database, db_conn, hs) def get_max_receipt_stream_id(self): return self._receipts_id_gen.get_current_token() diff --git a/synapse/replication/slave/storage/room.py b/synapse/replication/slave/storage/room.py index d9ad386b28..3a20f45316 100644 --- a/synapse/replication/slave/storage/room.py +++ b/synapse/replication/slave/storage/room.py @@ -14,14 +14,15 @@ # limitations under the License. from synapse.storage.data_stores.main.room import RoomWorkerStore +from synapse.storage.database import Database from ._base import BaseSlavedStore from ._slaved_id_tracker import SlavedIdTracker class RoomStore(RoomWorkerStore, BaseSlavedStore): - def __init__(self, db_conn, hs): - super(RoomStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(RoomStore, self).__init__(database, db_conn, hs) self._public_room_id_gen = SlavedIdTracker( db_conn, "public_room_list_stream", "stream_id" ) diff --git a/synapse/storage/_base.py b/synapse/storage/_base.py index b7e27d4e97..f9e7f9a71e 100644 --- a/synapse/storage/_base.py +++ b/synapse/storage/_base.py @@ -37,7 +37,7 @@ class SQLBaseStore(object): per data store (and not one per physical database). """ - def __init__(self, db_conn, hs): + def __init__(self, database: Database, db_conn, hs): self.hs = hs self._clock = hs.get_clock() self.database_engine = hs.database_engine diff --git a/synapse/storage/data_stores/main/__init__.py b/synapse/storage/data_stores/main/__init__.py index 6adb8adb04..7f5fd81bcf 100644 --- a/synapse/storage/data_stores/main/__init__.py +++ b/synapse/storage/data_stores/main/__init__.py @@ -20,6 +20,7 @@ import logging import time from synapse.api.constants import PresenceState +from synapse.storage.database import Database from synapse.storage.engines import PostgresEngine from synapse.storage.util.id_generators import ( ChainedIdGenerator, @@ -111,7 +112,7 @@ class DataStore( RelationsStore, CacheInvalidationStore, ): - def __init__(self, db_conn, hs): + def __init__(self, database: Database, db_conn, hs): self.hs = hs self._clock = hs.get_clock() self.database_engine = hs.database_engine @@ -169,7 +170,7 @@ class DataStore( else: self._cache_id_gen = None - super(DataStore, self).__init__(db_conn, hs) + super(DataStore, self).__init__(database, db_conn, hs) self._presence_on_startup = self._get_active_presence(db_conn) diff --git a/synapse/storage/data_stores/main/account_data.py b/synapse/storage/data_stores/main/account_data.py index a96fe9485c..44d20c19bf 100644 --- a/synapse/storage/data_stores/main/account_data.py +++ b/synapse/storage/data_stores/main/account_data.py @@ -22,6 +22,7 @@ from canonicaljson import json from twisted.internet import defer from synapse.storage._base import SQLBaseStore +from synapse.storage.database import Database from synapse.storage.util.id_generators import StreamIdGenerator from synapse.util.caches.descriptors import cached, cachedInlineCallbacks from synapse.util.caches.stream_change_cache import StreamChangeCache @@ -38,13 +39,13 @@ class AccountDataWorkerStore(SQLBaseStore): # the abstract methods being implemented. __metaclass__ = abc.ABCMeta - def __init__(self, db_conn, hs): + def __init__(self, database: Database, db_conn, hs): account_max = self.get_max_account_data_stream_id() self._account_data_stream_cache = StreamChangeCache( "AccountDataAndTagsChangeCache", account_max ) - super(AccountDataWorkerStore, self).__init__(db_conn, hs) + super(AccountDataWorkerStore, self).__init__(database, db_conn, hs) @abc.abstractmethod def get_max_account_data_stream_id(self): @@ -270,12 +271,12 @@ class AccountDataWorkerStore(SQLBaseStore): class AccountDataStore(AccountDataWorkerStore): - def __init__(self, db_conn, hs): + def __init__(self, database: Database, db_conn, hs): self._account_data_id_gen = StreamIdGenerator( db_conn, "account_data_max_stream_id", "stream_id" ) - super(AccountDataStore, self).__init__(db_conn, hs) + super(AccountDataStore, self).__init__(database, db_conn, hs) def get_max_account_data_stream_id(self): """Get the current max stream id for the private user data stream diff --git a/synapse/storage/data_stores/main/appservice.py b/synapse/storage/data_stores/main/appservice.py index 6b2e12719c..b2f39649fd 100644 --- a/synapse/storage/data_stores/main/appservice.py +++ b/synapse/storage/data_stores/main/appservice.py @@ -24,6 +24,7 @@ from synapse.appservice import AppServiceTransaction from synapse.config.appservice import load_appservices from synapse.storage._base import SQLBaseStore from synapse.storage.data_stores.main.events_worker import EventsWorkerStore +from synapse.storage.database import Database logger = logging.getLogger(__name__) @@ -48,13 +49,13 @@ def _make_exclusive_regex(services_cache): class ApplicationServiceWorkerStore(SQLBaseStore): - def __init__(self, db_conn, hs): + def __init__(self, database: Database, db_conn, hs): self.services_cache = load_appservices( hs.hostname, hs.config.app_service_config_files ) self.exclusive_user_regex = _make_exclusive_regex(self.services_cache) - super(ApplicationServiceWorkerStore, self).__init__(db_conn, hs) + super(ApplicationServiceWorkerStore, self).__init__(database, db_conn, hs) def get_app_services(self): return self.services_cache diff --git a/synapse/storage/data_stores/main/client_ips.py b/synapse/storage/data_stores/main/client_ips.py index 7b470a58f1..320c5b0f07 100644 --- a/synapse/storage/data_stores/main/client_ips.py +++ b/synapse/storage/data_stores/main/client_ips.py @@ -21,6 +21,7 @@ from twisted.internet import defer from synapse.metrics.background_process_metrics import wrap_as_background_process from synapse.storage._base import SQLBaseStore +from synapse.storage.database import Database from synapse.util.caches import CACHE_SIZE_FACTOR from synapse.util.caches.descriptors import Cache @@ -33,8 +34,8 @@ LAST_SEEN_GRANULARITY = 120 * 1000 class ClientIpBackgroundUpdateStore(SQLBaseStore): - def __init__(self, db_conn, hs): - super(ClientIpBackgroundUpdateStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(ClientIpBackgroundUpdateStore, self).__init__(database, db_conn, hs) self.db.updates.register_background_index_update( "user_ips_device_index", @@ -363,13 +364,13 @@ class ClientIpBackgroundUpdateStore(SQLBaseStore): class ClientIpStore(ClientIpBackgroundUpdateStore): - def __init__(self, db_conn, hs): + def __init__(self, database: Database, db_conn, hs): self.client_ip_last_seen = Cache( name="client_ip_last_seen", keylen=4, max_entries=50000 * CACHE_SIZE_FACTOR ) - super(ClientIpStore, self).__init__(db_conn, hs) + super(ClientIpStore, self).__init__(database, db_conn, hs) self.user_ips_max_age = hs.config.user_ips_max_age diff --git a/synapse/storage/data_stores/main/deviceinbox.py b/synapse/storage/data_stores/main/deviceinbox.py index 3c9f09301a..85cfa16850 100644 --- a/synapse/storage/data_stores/main/deviceinbox.py +++ b/synapse/storage/data_stores/main/deviceinbox.py @@ -21,6 +21,7 @@ from twisted.internet import defer from synapse.logging.opentracing import log_kv, set_tag, trace from synapse.storage._base import SQLBaseStore, make_in_list_sql_clause +from synapse.storage.database import Database from synapse.util.caches.expiringcache import ExpiringCache logger = logging.getLogger(__name__) @@ -210,8 +211,8 @@ class DeviceInboxWorkerStore(SQLBaseStore): class DeviceInboxBackgroundUpdateStore(SQLBaseStore): DEVICE_INBOX_STREAM_ID = "device_inbox_stream_drop" - def __init__(self, db_conn, hs): - super(DeviceInboxBackgroundUpdateStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(DeviceInboxBackgroundUpdateStore, self).__init__(database, db_conn, hs) self.db.updates.register_background_index_update( "device_inbox_stream_index", @@ -241,8 +242,8 @@ class DeviceInboxBackgroundUpdateStore(SQLBaseStore): class DeviceInboxStore(DeviceInboxWorkerStore, DeviceInboxBackgroundUpdateStore): DEVICE_INBOX_STREAM_ID = "device_inbox_stream_drop" - def __init__(self, db_conn, hs): - super(DeviceInboxStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(DeviceInboxStore, self).__init__(database, db_conn, hs) # Map of (user_id, device_id) to the last stream_id that has been # deleted up to. This is so that we can no op deletions. diff --git a/synapse/storage/data_stores/main/devices.py b/synapse/storage/data_stores/main/devices.py index 91ddaf137e..9a828231c4 100644 --- a/synapse/storage/data_stores/main/devices.py +++ b/synapse/storage/data_stores/main/devices.py @@ -31,6 +31,7 @@ from synapse.logging.opentracing import ( ) from synapse.metrics.background_process_metrics import run_as_background_process from synapse.storage._base import SQLBaseStore, db_to_json, make_in_list_sql_clause +from synapse.storage.database import Database from synapse.types import get_verify_key_from_cross_signing_key from synapse.util import batch_iter from synapse.util.caches.descriptors import ( @@ -642,8 +643,8 @@ class DeviceWorkerStore(SQLBaseStore): class DeviceBackgroundUpdateStore(SQLBaseStore): - def __init__(self, db_conn, hs): - super(DeviceBackgroundUpdateStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(DeviceBackgroundUpdateStore, self).__init__(database, db_conn, hs) self.db.updates.register_background_index_update( "device_lists_stream_idx", @@ -692,8 +693,8 @@ class DeviceBackgroundUpdateStore(SQLBaseStore): class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): - def __init__(self, db_conn, hs): - super(DeviceStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(DeviceStore, self).__init__(database, db_conn, hs) # Map of (user_id, device_id) -> bool. If there is an entry that implies # the device exists. diff --git a/synapse/storage/data_stores/main/event_federation.py b/synapse/storage/data_stores/main/event_federation.py index 31d2e8eb28..1f517e8fad 100644 --- a/synapse/storage/data_stores/main/event_federation.py +++ b/synapse/storage/data_stores/main/event_federation.py @@ -28,6 +28,7 @@ from synapse.metrics.background_process_metrics import run_as_background_process from synapse.storage._base import SQLBaseStore, make_in_list_sql_clause from synapse.storage.data_stores.main.events_worker import EventsWorkerStore from synapse.storage.data_stores.main.signatures import SignatureWorkerStore +from synapse.storage.database import Database from synapse.util.caches.descriptors import cached logger = logging.getLogger(__name__) @@ -491,8 +492,8 @@ class EventFederationStore(EventFederationWorkerStore): EVENT_AUTH_STATE_ONLY = "event_auth_state_only" - def __init__(self, db_conn, hs): - super(EventFederationStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(EventFederationStore, self).__init__(database, db_conn, hs) self.db.updates.register_background_update_handler( self.EVENT_AUTH_STATE_ONLY, self._background_delete_non_state_event_auth diff --git a/synapse/storage/data_stores/main/event_push_actions.py b/synapse/storage/data_stores/main/event_push_actions.py index eec054cd48..9988a6d3fc 100644 --- a/synapse/storage/data_stores/main/event_push_actions.py +++ b/synapse/storage/data_stores/main/event_push_actions.py @@ -24,6 +24,7 @@ from twisted.internet import defer from synapse.metrics.background_process_metrics import run_as_background_process from synapse.storage._base import LoggingTransaction, SQLBaseStore +from synapse.storage.database import Database from synapse.util.caches.descriptors import cachedInlineCallbacks logger = logging.getLogger(__name__) @@ -68,8 +69,8 @@ def _deserialize_action(actions, is_highlight): class EventPushActionsWorkerStore(SQLBaseStore): - def __init__(self, db_conn, hs): - super(EventPushActionsWorkerStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(EventPushActionsWorkerStore, self).__init__(database, db_conn, hs) # These get correctly set by _find_stream_orderings_for_times_txn self.stream_ordering_month_ago = None @@ -611,8 +612,8 @@ class EventPushActionsWorkerStore(SQLBaseStore): class EventPushActionsStore(EventPushActionsWorkerStore): EPA_HIGHLIGHT_INDEX = "epa_highlight_index" - def __init__(self, db_conn, hs): - super(EventPushActionsStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(EventPushActionsStore, self).__init__(database, db_conn, hs) self.db.updates.register_background_index_update( self.EPA_HIGHLIGHT_INDEX, diff --git a/synapse/storage/data_stores/main/events.py b/synapse/storage/data_stores/main/events.py index d644c82784..da1529f6ea 100644 --- a/synapse/storage/data_stores/main/events.py +++ b/synapse/storage/data_stores/main/events.py @@ -41,6 +41,7 @@ from synapse.storage._base import make_in_list_sql_clause from synapse.storage.data_stores.main.event_federation import EventFederationStore from synapse.storage.data_stores.main.events_worker import EventsWorkerStore from synapse.storage.data_stores.main.state import StateGroupWorkerStore +from synapse.storage.database import Database from synapse.types import RoomStreamToken, get_domain_from_id from synapse.util import batch_iter from synapse.util.caches.descriptors import cached, cachedInlineCallbacks @@ -95,8 +96,8 @@ def _retry_on_integrity_error(func): class EventsStore( StateGroupWorkerStore, EventFederationStore, EventsWorkerStore, ): - def __init__(self, db_conn, hs): - super(EventsStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(EventsStore, self).__init__(database, db_conn, hs) # Collect metrics on the number of forward extremities that exist. # Counter of number of extremities to count diff --git a/synapse/storage/data_stores/main/events_bg_updates.py b/synapse/storage/data_stores/main/events_bg_updates.py index cb1fc30c31..efee17b929 100644 --- a/synapse/storage/data_stores/main/events_bg_updates.py +++ b/synapse/storage/data_stores/main/events_bg_updates.py @@ -23,6 +23,7 @@ from twisted.internet import defer from synapse.api.constants import EventContentFields from synapse.storage._base import SQLBaseStore, make_in_list_sql_clause +from synapse.storage.database import Database logger = logging.getLogger(__name__) @@ -33,8 +34,8 @@ class EventsBackgroundUpdatesStore(SQLBaseStore): EVENT_FIELDS_SENDER_URL_UPDATE_NAME = "event_fields_sender_url" DELETE_SOFT_FAILED_EXTREMITIES = "delete_soft_failed_extremities" - def __init__(self, db_conn, hs): - super(EventsBackgroundUpdatesStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(EventsBackgroundUpdatesStore, self).__init__(database, db_conn, hs) self.db.updates.register_background_update_handler( self.EVENT_ORIGIN_SERVER_TS_NAME, self._background_reindex_origin_server_ts diff --git a/synapse/storage/data_stores/main/events_worker.py b/synapse/storage/data_stores/main/events_worker.py index e041fc5eac..9ee117ce0f 100644 --- a/synapse/storage/data_stores/main/events_worker.py +++ b/synapse/storage/data_stores/main/events_worker.py @@ -33,6 +33,7 @@ from synapse.events.utils import prune_event from synapse.logging.context import LoggingContext, PreserveLoggingContext from synapse.metrics.background_process_metrics import run_as_background_process from synapse.storage._base import SQLBaseStore, make_in_list_sql_clause +from synapse.storage.database import Database from synapse.types import get_domain_from_id from synapse.util import batch_iter from synapse.util.caches.descriptors import Cache @@ -55,8 +56,8 @@ _EventCacheEntry = namedtuple("_EventCacheEntry", ("event", "redacted_event")) class EventsWorkerStore(SQLBaseStore): - def __init__(self, db_conn, hs): - super(EventsWorkerStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(EventsWorkerStore, self).__init__(database, db_conn, hs) self._get_event_cache = Cache( "*getEvent*", keylen=3, max_entries=hs.config.event_cache_size diff --git a/synapse/storage/data_stores/main/media_repository.py b/synapse/storage/data_stores/main/media_repository.py index 03c9c6f8ae..80ca36dedf 100644 --- a/synapse/storage/data_stores/main/media_repository.py +++ b/synapse/storage/data_stores/main/media_repository.py @@ -13,11 +13,14 @@ # See the License for the specific language governing permissions and # limitations under the License. from synapse.storage._base import SQLBaseStore +from synapse.storage.database import Database class MediaRepositoryBackgroundUpdateStore(SQLBaseStore): - def __init__(self, db_conn, hs): - super(MediaRepositoryBackgroundUpdateStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(MediaRepositoryBackgroundUpdateStore, self).__init__( + database, db_conn, hs + ) self.db.updates.register_background_index_update( update_name="local_media_repository_url_idx", @@ -31,8 +34,8 @@ class MediaRepositoryBackgroundUpdateStore(SQLBaseStore): class MediaRepositoryStore(MediaRepositoryBackgroundUpdateStore): """Persistence for attachments and avatars""" - def __init__(self, db_conn, hs): - super(MediaRepositoryStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(MediaRepositoryStore, self).__init__(database, db_conn, hs) def get_local_media(self, media_id): """Get the metadata for a local piece of media diff --git a/synapse/storage/data_stores/main/monthly_active_users.py b/synapse/storage/data_stores/main/monthly_active_users.py index 34bf3a1880..27158534cb 100644 --- a/synapse/storage/data_stores/main/monthly_active_users.py +++ b/synapse/storage/data_stores/main/monthly_active_users.py @@ -17,6 +17,7 @@ import logging from twisted.internet import defer from synapse.storage._base import SQLBaseStore +from synapse.storage.database import Database from synapse.util.caches.descriptors import cached logger = logging.getLogger(__name__) @@ -27,13 +28,13 @@ LAST_SEEN_GRANULARITY = 60 * 60 * 1000 class MonthlyActiveUsersStore(SQLBaseStore): - def __init__(self, dbconn, hs): - super(MonthlyActiveUsersStore, self).__init__(None, hs) + def __init__(self, database: Database, db_conn, hs): + super(MonthlyActiveUsersStore, self).__init__(database, db_conn, hs) self._clock = hs.get_clock() self.hs = hs # Do not add more reserved users than the total allowable number self.db.new_transaction( - dbconn, + db_conn, "initialise_mau_threepids", [], [], diff --git a/synapse/storage/data_stores/main/push_rule.py b/synapse/storage/data_stores/main/push_rule.py index de682cc63a..5ba13aa973 100644 --- a/synapse/storage/data_stores/main/push_rule.py +++ b/synapse/storage/data_stores/main/push_rule.py @@ -27,6 +27,7 @@ from synapse.storage.data_stores.main.appservice import ApplicationServiceWorker from synapse.storage.data_stores.main.pusher import PusherWorkerStore from synapse.storage.data_stores.main.receipts import ReceiptsWorkerStore from synapse.storage.data_stores.main.roommember import RoomMemberWorkerStore +from synapse.storage.database import Database from synapse.storage.push_rule import InconsistentRuleException, RuleNotFoundException from synapse.util.caches.descriptors import cachedInlineCallbacks, cachedList from synapse.util.caches.stream_change_cache import StreamChangeCache @@ -72,8 +73,8 @@ class PushRulesWorkerStore( # the abstract methods being implemented. __metaclass__ = abc.ABCMeta - def __init__(self, db_conn, hs): - super(PushRulesWorkerStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(PushRulesWorkerStore, self).__init__(database, db_conn, hs) push_rules_prefill, push_rules_id = self.db.get_cache_dict( db_conn, diff --git a/synapse/storage/data_stores/main/receipts.py b/synapse/storage/data_stores/main/receipts.py index ac2d45bd5c..96e54d145e 100644 --- a/synapse/storage/data_stores/main/receipts.py +++ b/synapse/storage/data_stores/main/receipts.py @@ -22,6 +22,7 @@ from canonicaljson import json from twisted.internet import defer from synapse.storage._base import SQLBaseStore, make_in_list_sql_clause +from synapse.storage.database import Database from synapse.storage.util.id_generators import StreamIdGenerator from synapse.util.caches.descriptors import cached, cachedInlineCallbacks, cachedList from synapse.util.caches.stream_change_cache import StreamChangeCache @@ -38,8 +39,8 @@ class ReceiptsWorkerStore(SQLBaseStore): # the abstract methods being implemented. __metaclass__ = abc.ABCMeta - def __init__(self, db_conn, hs): - super(ReceiptsWorkerStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(ReceiptsWorkerStore, self).__init__(database, db_conn, hs) self._receipts_stream_cache = StreamChangeCache( "ReceiptsRoomChangeCache", self.get_max_receipt_stream_id() @@ -315,14 +316,14 @@ class ReceiptsWorkerStore(SQLBaseStore): class ReceiptsStore(ReceiptsWorkerStore): - def __init__(self, db_conn, hs): + def __init__(self, database: Database, db_conn, hs): # We instantiate this first as the ReceiptsWorkerStore constructor # needs to be able to call get_max_receipt_stream_id self._receipts_id_gen = StreamIdGenerator( db_conn, "receipts_linearized", "stream_id" ) - super(ReceiptsStore, self).__init__(db_conn, hs) + super(ReceiptsStore, self).__init__(database, db_conn, hs) def get_max_receipt_stream_id(self): return self._receipts_id_gen.get_current_token() diff --git a/synapse/storage/data_stores/main/registration.py b/synapse/storage/data_stores/main/registration.py index 1ef143c6d8..5e8ecac0ea 100644 --- a/synapse/storage/data_stores/main/registration.py +++ b/synapse/storage/data_stores/main/registration.py @@ -27,6 +27,7 @@ from synapse.api.constants import UserTypes from synapse.api.errors import Codes, StoreError, SynapseError, ThreepidValidationError from synapse.metrics.background_process_metrics import run_as_background_process from synapse.storage._base import SQLBaseStore +from synapse.storage.database import Database from synapse.types import UserID from synapse.util.caches.descriptors import cached, cachedInlineCallbacks @@ -36,8 +37,8 @@ logger = logging.getLogger(__name__) class RegistrationWorkerStore(SQLBaseStore): - def __init__(self, db_conn, hs): - super(RegistrationWorkerStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(RegistrationWorkerStore, self).__init__(database, db_conn, hs) self.config = hs.config self.clock = hs.get_clock() @@ -794,8 +795,8 @@ class RegistrationWorkerStore(SQLBaseStore): class RegistrationBackgroundUpdateStore(RegistrationWorkerStore): - def __init__(self, db_conn, hs): - super(RegistrationBackgroundUpdateStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(RegistrationBackgroundUpdateStore, self).__init__(database, db_conn, hs) self.clock = hs.get_clock() self.config = hs.config @@ -920,8 +921,8 @@ class RegistrationBackgroundUpdateStore(RegistrationWorkerStore): class RegistrationStore(RegistrationBackgroundUpdateStore): - def __init__(self, db_conn, hs): - super(RegistrationStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(RegistrationStore, self).__init__(database, db_conn, hs) self._account_validity = hs.config.account_validity diff --git a/synapse/storage/data_stores/main/room.py b/synapse/storage/data_stores/main/room.py index da42dae243..0148be20d3 100644 --- a/synapse/storage/data_stores/main/room.py +++ b/synapse/storage/data_stores/main/room.py @@ -29,6 +29,7 @@ from synapse.api.constants import EventTypes from synapse.api.errors import StoreError from synapse.storage._base import SQLBaseStore from synapse.storage.data_stores.main.search import SearchStore +from synapse.storage.database import Database from synapse.types import ThirdPartyInstanceID from synapse.util.caches.descriptors import cached, cachedInlineCallbacks @@ -361,8 +362,8 @@ class RoomWorkerStore(SQLBaseStore): class RoomBackgroundUpdateStore(SQLBaseStore): - def __init__(self, db_conn, hs): - super(RoomBackgroundUpdateStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(RoomBackgroundUpdateStore, self).__init__(database, db_conn, hs) self.config = hs.config @@ -440,8 +441,8 @@ class RoomBackgroundUpdateStore(SQLBaseStore): class RoomStore(RoomBackgroundUpdateStore, RoomWorkerStore, SearchStore): - def __init__(self, db_conn, hs): - super(RoomStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(RoomStore, self).__init__(database, db_conn, hs) self.config = hs.config diff --git a/synapse/storage/data_stores/main/roommember.py b/synapse/storage/data_stores/main/roommember.py index 929f6b0d39..92e3b9c512 100644 --- a/synapse/storage/data_stores/main/roommember.py +++ b/synapse/storage/data_stores/main/roommember.py @@ -32,6 +32,7 @@ from synapse.storage._base import ( make_in_list_sql_clause, ) from synapse.storage.data_stores.main.events_worker import EventsWorkerStore +from synapse.storage.database import Database from synapse.storage.engines import Sqlite3Engine from synapse.storage.roommember import ( GetRoomsForUserWithStreamOrdering, @@ -54,8 +55,8 @@ _CURRENT_STATE_MEMBERSHIP_UPDATE_NAME = "current_state_events_membership" class RoomMemberWorkerStore(EventsWorkerStore): - def __init__(self, db_conn, hs): - super(RoomMemberWorkerStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(RoomMemberWorkerStore, self).__init__(database, db_conn, hs) # Is the current_state_events.membership up to date? Or is the # background update still running? @@ -835,8 +836,8 @@ class RoomMemberWorkerStore(EventsWorkerStore): class RoomMemberBackgroundUpdateStore(SQLBaseStore): - def __init__(self, db_conn, hs): - super(RoomMemberBackgroundUpdateStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(RoomMemberBackgroundUpdateStore, self).__init__(database, db_conn, hs) self.db.updates.register_background_update_handler( _MEMBERSHIP_PROFILE_UPDATE_NAME, self._background_add_membership_profile ) @@ -991,8 +992,8 @@ class RoomMemberBackgroundUpdateStore(SQLBaseStore): class RoomMemberStore(RoomMemberWorkerStore, RoomMemberBackgroundUpdateStore): - def __init__(self, db_conn, hs): - super(RoomMemberStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(RoomMemberStore, self).__init__(database, db_conn, hs) def _store_room_members_txn(self, txn, events, backfilled): """Store a room member in the database. diff --git a/synapse/storage/data_stores/main/search.py b/synapse/storage/data_stores/main/search.py index ffa1817e64..4eec2fae5e 100644 --- a/synapse/storage/data_stores/main/search.py +++ b/synapse/storage/data_stores/main/search.py @@ -25,6 +25,7 @@ from twisted.internet import defer from synapse.api.errors import SynapseError from synapse.storage._base import SQLBaseStore, make_in_list_sql_clause +from synapse.storage.database import Database from synapse.storage.engines import PostgresEngine, Sqlite3Engine logger = logging.getLogger(__name__) @@ -42,8 +43,8 @@ class SearchBackgroundUpdateStore(SQLBaseStore): EVENT_SEARCH_USE_GIST_POSTGRES_NAME = "event_search_postgres_gist" EVENT_SEARCH_USE_GIN_POSTGRES_NAME = "event_search_postgres_gin" - def __init__(self, db_conn, hs): - super(SearchBackgroundUpdateStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(SearchBackgroundUpdateStore, self).__init__(database, db_conn, hs) if not hs.config.enable_search: return @@ -342,8 +343,8 @@ class SearchBackgroundUpdateStore(SQLBaseStore): class SearchStore(SearchBackgroundUpdateStore): - def __init__(self, db_conn, hs): - super(SearchStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(SearchStore, self).__init__(database, db_conn, hs) def store_event_search_txn(self, txn, event, key, value): """Add event to the search table diff --git a/synapse/storage/data_stores/main/state.py b/synapse/storage/data_stores/main/state.py index 7d5a9f8128..9ef7b48c74 100644 --- a/synapse/storage/data_stores/main/state.py +++ b/synapse/storage/data_stores/main/state.py @@ -28,6 +28,7 @@ from synapse.events import EventBase from synapse.events.snapshot import EventContext from synapse.storage._base import SQLBaseStore from synapse.storage.data_stores.main.events_worker import EventsWorkerStore +from synapse.storage.database import Database from synapse.storage.engines import PostgresEngine from synapse.storage.state import StateFilter from synapse.util.caches import get_cache_factor_for, intern_string @@ -213,8 +214,8 @@ class StateGroupWorkerStore( STATE_GROUP_INDEX_UPDATE_NAME = "state_group_state_type_index" CURRENT_STATE_INDEX_UPDATE_NAME = "current_state_members_idx" - def __init__(self, db_conn, hs): - super(StateGroupWorkerStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(StateGroupWorkerStore, self).__init__(database, db_conn, hs) # Originally the state store used a single DictionaryCache to cache the # event IDs for the state types in a given state group to avoid hammering @@ -1029,8 +1030,8 @@ class StateBackgroundUpdateStore(StateGroupBackgroundUpdateStore): CURRENT_STATE_INDEX_UPDATE_NAME = "current_state_members_idx" EVENT_STATE_GROUP_INDEX_UPDATE_NAME = "event_to_state_groups_sg_index" - def __init__(self, db_conn, hs): - super(StateBackgroundUpdateStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(StateBackgroundUpdateStore, self).__init__(database, db_conn, hs) self.db.updates.register_background_update_handler( self.STATE_GROUP_DEDUPLICATION_UPDATE_NAME, self._background_deduplicate_state, @@ -1245,8 +1246,8 @@ class StateStore(StateGroupWorkerStore, StateBackgroundUpdateStore): * `state_groups_state`: Maps state group to state events. """ - def __init__(self, db_conn, hs): - super(StateStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(StateStore, self).__init__(database, db_conn, hs) def _store_event_state_mappings_txn( self, txn, events_and_contexts: Iterable[Tuple[EventBase, EventContext]] diff --git a/synapse/storage/data_stores/main/stats.py b/synapse/storage/data_stores/main/stats.py index 40579bf965..7bc186e9a1 100644 --- a/synapse/storage/data_stores/main/stats.py +++ b/synapse/storage/data_stores/main/stats.py @@ -22,6 +22,7 @@ from twisted.internet.defer import DeferredLock from synapse.api.constants import EventTypes, Membership from synapse.storage.data_stores.main.state_deltas import StateDeltasStore +from synapse.storage.database import Database from synapse.storage.engines import PostgresEngine from synapse.util.caches.descriptors import cached @@ -58,8 +59,8 @@ TYPE_TO_ORIGIN_TABLE = {"room": ("rooms", "room_id"), "user": ("users", "name")} class StatsStore(StateDeltasStore): - def __init__(self, db_conn, hs): - super(StatsStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(StatsStore, self).__init__(database, db_conn, hs) self.server_name = hs.hostname self.clock = self.hs.get_clock() diff --git a/synapse/storage/data_stores/main/stream.py b/synapse/storage/data_stores/main/stream.py index 2ff8c57109..140da8dad6 100644 --- a/synapse/storage/data_stores/main/stream.py +++ b/synapse/storage/data_stores/main/stream.py @@ -47,6 +47,7 @@ from twisted.internet import defer from synapse.logging.context import make_deferred_yieldable, run_in_background from synapse.storage._base import SQLBaseStore from synapse.storage.data_stores.main.events_worker import EventsWorkerStore +from synapse.storage.database import Database from synapse.storage.engines import PostgresEngine from synapse.types import RoomStreamToken from synapse.util.caches.stream_change_cache import StreamChangeCache @@ -251,8 +252,8 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): __metaclass__ = abc.ABCMeta - def __init__(self, db_conn, hs): - super(StreamWorkerStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(StreamWorkerStore, self).__init__(database, db_conn, hs) events_max = self.get_room_max_stream_ordering() event_cache_prefill, min_event_val = self.db.get_cache_dict( diff --git a/synapse/storage/data_stores/main/transactions.py b/synapse/storage/data_stores/main/transactions.py index c0d155a43c..5b07c2fbc0 100644 --- a/synapse/storage/data_stores/main/transactions.py +++ b/synapse/storage/data_stores/main/transactions.py @@ -24,6 +24,7 @@ from twisted.internet import defer from synapse.metrics.background_process_metrics import run_as_background_process from synapse.storage._base import SQLBaseStore, db_to_json +from synapse.storage.database import Database from synapse.util.caches.expiringcache import ExpiringCache # py2 sqlite has buffer hardcoded as only binary type, so we must use it, @@ -52,8 +53,8 @@ class TransactionStore(SQLBaseStore): """A collection of queries for handling PDUs. """ - def __init__(self, db_conn, hs): - super(TransactionStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(TransactionStore, self).__init__(database, db_conn, hs) self._clock.looping_call(self._start_cleanup_transactions, 30 * 60 * 1000) diff --git a/synapse/storage/data_stores/main/user_directory.py b/synapse/storage/data_stores/main/user_directory.py index 62ffb34b29..90c180ec6d 100644 --- a/synapse/storage/data_stores/main/user_directory.py +++ b/synapse/storage/data_stores/main/user_directory.py @@ -21,6 +21,7 @@ from twisted.internet import defer from synapse.api.constants import EventTypes, JoinRules from synapse.storage.data_stores.main.state import StateFilter from synapse.storage.data_stores.main.state_deltas import StateDeltasStore +from synapse.storage.database import Database from synapse.storage.engines import PostgresEngine, Sqlite3Engine from synapse.types import get_domain_from_id, get_localpart_from_id from synapse.util.caches.descriptors import cached @@ -37,8 +38,8 @@ class UserDirectoryBackgroundUpdateStore(StateDeltasStore): # add_users_who_share_private_rooms? SHARE_PRIVATE_WORKING_SET = 500 - def __init__(self, db_conn, hs): - super(UserDirectoryBackgroundUpdateStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(UserDirectoryBackgroundUpdateStore, self).__init__(database, db_conn, hs) self.server_name = hs.hostname @@ -549,8 +550,8 @@ class UserDirectoryStore(UserDirectoryBackgroundUpdateStore): # add_users_who_share_private_rooms? SHARE_PRIVATE_WORKING_SET = 500 - def __init__(self, db_conn, hs): - super(UserDirectoryStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(UserDirectoryStore, self).__init__(database, db_conn, hs) def remove_from_user_dir(self, user_id): def _remove_from_user_dir_txn(txn): diff --git a/tests/storage/test_appservice.py b/tests/storage/test_appservice.py index dfeea24599..1679112d82 100644 --- a/tests/storage/test_appservice.py +++ b/tests/storage/test_appservice.py @@ -28,6 +28,7 @@ from synapse.storage.data_stores.main.appservice import ( ApplicationServiceStore, ApplicationServiceTransactionStore, ) +from synapse.storage.database import Database from tests import unittest from tests.utils import setup_test_homeserver @@ -382,8 +383,8 @@ class ApplicationServiceTransactionStoreTestCase(unittest.TestCase): # required for ApplicationServiceTransactionStoreTestCase tests class TestTransactionStore(ApplicationServiceTransactionStore, ApplicationServiceStore): - def __init__(self, db_conn, hs): - super(TestTransactionStore, self).__init__(db_conn, hs) + def __init__(self, database: Database, db_conn, hs): + super(TestTransactionStore, self).__init__(database, db_conn, hs) class ApplicationServiceStoreConfigTestCase(unittest.TestCase): -- cgit 1.4.1 From 852f80d8a697c9d556b1b2641a2e4c1797cbbb46 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Fri, 6 Dec 2019 16:02:50 +0000 Subject: Fixup tests --- tests/replication/slave/storage/_base.py | 5 ++++- tests/storage/test_appservice.py | 12 +++++++----- tests/storage/test_base.py | 3 ++- tests/storage/test_profile.py | 3 +-- tests/storage/test_user_directory.py | 4 +--- tests/test_federation.py | 16 +++++----------- 6 files changed, 20 insertions(+), 23 deletions(-) (limited to 'tests') diff --git a/tests/replication/slave/storage/_base.py b/tests/replication/slave/storage/_base.py index e7472e3a93..3dae83c543 100644 --- a/tests/replication/slave/storage/_base.py +++ b/tests/replication/slave/storage/_base.py @@ -20,6 +20,7 @@ from synapse.replication.tcp.client import ( ReplicationClientHandler, ) from synapse.replication.tcp.resource import ReplicationStreamProtocolFactory +from synapse.storage.database import Database from tests import unittest from tests.server import FakeTransport @@ -42,7 +43,9 @@ class BaseSlavedStoreTestCase(unittest.HomeserverTestCase): self.master_store = self.hs.get_datastore() self.storage = hs.get_storage() - self.slaved_store = self.STORE_TYPE(self.hs.get_db_conn(), self.hs) + self.slaved_store = self.STORE_TYPE( + Database(hs), self.hs.get_db_conn(), self.hs + ) self.event_id = 0 server_factory = ReplicationStreamProtocolFactory(self.hs) diff --git a/tests/storage/test_appservice.py b/tests/storage/test_appservice.py index 1679112d82..2e521e9ab7 100644 --- a/tests/storage/test_appservice.py +++ b/tests/storage/test_appservice.py @@ -55,7 +55,8 @@ class ApplicationServiceStoreTestCase(unittest.TestCase): self._add_appservice("token2", "as2", "some_url", "some_hs_token", "bob") self._add_appservice("token3", "as3", "some_url", "some_hs_token", "bob") # must be done after inserts - self.store = ApplicationServiceStore(hs.get_db_conn(), hs) + database = Database(hs) + self.store = ApplicationServiceStore(database, hs.get_db_conn(), hs) def tearDown(self): # TODO: suboptimal that we need to create files for tests! @@ -124,7 +125,8 @@ class ApplicationServiceTransactionStoreTestCase(unittest.TestCase): self.as_yaml_files = [] - self.store = TestTransactionStore(hs.get_db_conn(), hs) + database = Database(hs) + self.store = TestTransactionStore(database, hs.get_db_conn(), hs) def _add_service(self, url, as_token, id): as_yaml = dict( @@ -417,7 +419,7 @@ class ApplicationServiceStoreConfigTestCase(unittest.TestCase): hs.config.event_cache_size = 1 hs.config.password_providers = [] - ApplicationServiceStore(hs.get_db_conn(), hs) + ApplicationServiceStore(Database(hs), hs.get_db_conn(), hs) @defer.inlineCallbacks def test_duplicate_ids(self): @@ -433,7 +435,7 @@ class ApplicationServiceStoreConfigTestCase(unittest.TestCase): hs.config.password_providers = [] with self.assertRaises(ConfigError) as cm: - ApplicationServiceStore(hs.get_db_conn(), hs) + ApplicationServiceStore(Database(hs), hs.get_db_conn(), hs) e = cm.exception self.assertIn(f1, str(e)) @@ -454,7 +456,7 @@ class ApplicationServiceStoreConfigTestCase(unittest.TestCase): hs.config.password_providers = [] with self.assertRaises(ConfigError) as cm: - ApplicationServiceStore(hs.get_db_conn(), hs) + ApplicationServiceStore(Database(hs), hs.get_db_conn(), hs) e = cm.exception self.assertIn(f1, str(e)) diff --git a/tests/storage/test_base.py b/tests/storage/test_base.py index 7915d48a9e..537cfe9f64 100644 --- a/tests/storage/test_base.py +++ b/tests/storage/test_base.py @@ -21,6 +21,7 @@ from mock import Mock from twisted.internet import defer from synapse.storage._base import SQLBaseStore +from synapse.storage.database import Database from synapse.storage.engines import create_engine from tests import unittest @@ -59,7 +60,7 @@ class SQLBaseStoreTestCase(unittest.TestCase): "test", db_pool=self.db_pool, config=config, database_engine=fake_engine ) - self.datastore = SQLBaseStore(None, hs) + self.datastore = SQLBaseStore(Database(hs), None, hs) @defer.inlineCallbacks def test_insert_1col(self): diff --git a/tests/storage/test_profile.py b/tests/storage/test_profile.py index 24c7fe16c3..9b6f7211ae 100644 --- a/tests/storage/test_profile.py +++ b/tests/storage/test_profile.py @@ -16,7 +16,6 @@ from twisted.internet import defer -from synapse.storage.data_stores.main.profile import ProfileStore from synapse.types import UserID from tests import unittest @@ -28,7 +27,7 @@ class ProfileStoreTestCase(unittest.TestCase): def setUp(self): hs = yield setup_test_homeserver(self.addCleanup) - self.store = ProfileStore(hs.get_db_conn(), hs) + self.store = hs.get_datastore() self.u_frank = UserID.from_string("@frank:test") diff --git a/tests/storage/test_user_directory.py b/tests/storage/test_user_directory.py index 7eea57c0e2..6a545d2eb0 100644 --- a/tests/storage/test_user_directory.py +++ b/tests/storage/test_user_directory.py @@ -15,8 +15,6 @@ from twisted.internet import defer -from synapse.storage.data_stores.main.user_directory import UserDirectoryStore - from tests import unittest from tests.utils import setup_test_homeserver @@ -29,7 +27,7 @@ class UserDirectoryStoreTestCase(unittest.TestCase): @defer.inlineCallbacks def setUp(self): self.hs = yield setup_test_homeserver(self.addCleanup) - self.store = UserDirectoryStore(self.hs.get_db_conn(), self.hs) + self.store = self.hs.get_datastore() # alice and bob are both in !room_id. bobby is not but shares # a homeserver with alice. diff --git a/tests/test_federation.py b/tests/test_federation.py index 7d82b58466..ad165d7295 100644 --- a/tests/test_federation.py +++ b/tests/test_federation.py @@ -33,6 +33,8 @@ class MessageAcceptTests(unittest.TestCase): self.reactor.advance(0.1) self.room_id = self.successResultOf(room)["room_id"] + self.store = self.homeserver.get_datastore() + # Figure out what the most recent event is most_recent = self.successResultOf( maybeDeferred( @@ -77,10 +79,7 @@ class MessageAcceptTests(unittest.TestCase): # Make sure we actually joined the room self.assertEqual( self.successResultOf( - maybeDeferred( - self.homeserver.get_datastore().get_latest_event_ids_in_room, - self.room_id, - ) + maybeDeferred(self.store.get_latest_event_ids_in_room, self.room_id) )[0], "$join:test.serv", ) @@ -100,10 +99,7 @@ class MessageAcceptTests(unittest.TestCase): # Figure out what the most recent event is most_recent = self.successResultOf( - maybeDeferred( - self.homeserver.get_datastore().get_latest_event_ids_in_room, - self.room_id, - ) + maybeDeferred(self.store.get_latest_event_ids_in_room, self.room_id) )[0] # Now lie about an event @@ -141,7 +137,5 @@ class MessageAcceptTests(unittest.TestCase): ) # Make sure the invalid event isn't there - extrem = maybeDeferred( - self.homeserver.get_datastore().get_latest_event_ids_in_room, self.room_id - ) + extrem = maybeDeferred(self.store.get_latest_event_ids_in_room, self.room_id) self.assertEqual(self.successResultOf(extrem)[0], "$join:test.serv") -- cgit 1.4.1 From f166a8d1f52f1fda1a64d8cd50f17275d63c8843 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Mon, 9 Dec 2019 13:42:49 +0000 Subject: Remove SnapshotCache in favour of ResponseCache --- synapse/handlers/initial_sync.py | 19 +++---- synapse/util/caches/snapshot_cache.py | 94 ----------------------------------- tests/util/test_snapshot_cache.py | 63 ----------------------- 3 files changed, 8 insertions(+), 168 deletions(-) delete mode 100644 synapse/util/caches/snapshot_cache.py delete mode 100644 tests/util/test_snapshot_cache.py (limited to 'tests') diff --git a/synapse/handlers/initial_sync.py b/synapse/handlers/initial_sync.py index 81dce96f4b..73c110a92b 100644 --- a/synapse/handlers/initial_sync.py +++ b/synapse/handlers/initial_sync.py @@ -26,7 +26,7 @@ from synapse.streams.config import PaginationConfig from synapse.types import StreamToken, UserID from synapse.util import unwrapFirstError from synapse.util.async_helpers import concurrently_execute -from synapse.util.caches.snapshot_cache import SnapshotCache +from synapse.util.caches.response_cache import ResponseCache from synapse.visibility import filter_events_for_client from ._base import BaseHandler @@ -41,7 +41,7 @@ class InitialSyncHandler(BaseHandler): self.state = hs.get_state_handler() self.clock = hs.get_clock() self.validator = EventValidator() - self.snapshot_cache = SnapshotCache() + self.snapshot_cache = ResponseCache(hs, "initial_sync_cache") self._event_serializer = hs.get_event_client_serializer() self.storage = hs.get_storage() self.state_store = self.storage.state @@ -79,17 +79,14 @@ class InitialSyncHandler(BaseHandler): as_client_event, include_archived, ) - now_ms = self.clock.time_msec() - result = self.snapshot_cache.get(now_ms, key) - if result is not None: - return result - return self.snapshot_cache.set( - now_ms, + return self.snapshot_cache.wrap( key, - self._snapshot_all_rooms( - user_id, pagin_config, as_client_event, include_archived - ), + self._snapshot_all_rooms, + user_id, + pagin_config, + as_client_event, + include_archived, ) @defer.inlineCallbacks diff --git a/synapse/util/caches/snapshot_cache.py b/synapse/util/caches/snapshot_cache.py deleted file mode 100644 index 8318db8d2c..0000000000 --- a/synapse/util/caches/snapshot_cache.py +++ /dev/null @@ -1,94 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2015, 2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -from synapse.util.async_helpers import ObservableDeferred - - -class SnapshotCache(object): - """Cache for snapshots like the response of /initialSync. - The response of initialSync only has to be a recent snapshot of the - server state. It shouldn't matter to clients if it is a few minutes out - of date. - - This caches a deferred response. Until the deferred completes it will be - returned from the cache. This means that if the client retries the request - while the response is still being computed, that original response will be - used rather than trying to compute a new response. - - Once the deferred completes it will removed from the cache after 5 minutes. - We delay removing it from the cache because a client retrying its request - could race with us finishing computing the response. - - Rather than tracking precisely how long something has been in the cache we - keep two generations of completed responses. Every 5 minutes discard the - old generation, move the new generation to the old generation, and set the - new generation to be empty. This means that a result will be in the cache - somewhere between 5 and 10 minutes. - """ - - DURATION_MS = 5 * 60 * 1000 # Cache results for 5 minutes. - - def __init__(self): - self.pending_result_cache = {} # Request that haven't finished yet. - self.prev_result_cache = {} # The older requests that have finished. - self.next_result_cache = {} # The newer requests that have finished. - self.time_last_rotated_ms = 0 - - def rotate(self, time_now_ms): - # Rotate once if the cache duration has passed since the last rotation. - if time_now_ms - self.time_last_rotated_ms >= self.DURATION_MS: - self.prev_result_cache = self.next_result_cache - self.next_result_cache = {} - self.time_last_rotated_ms += self.DURATION_MS - - # Rotate again if the cache duration has passed twice since the last - # rotation. - if time_now_ms - self.time_last_rotated_ms >= self.DURATION_MS: - self.prev_result_cache = self.next_result_cache - self.next_result_cache = {} - self.time_last_rotated_ms = time_now_ms - - def get(self, time_now_ms, key): - self.rotate(time_now_ms) - # This cache is intended to deduplicate requests, so we expect it to be - # missed most of the time. So we just lookup the key in all of the - # dictionaries rather than trying to short circuit the lookup if the - # key is found. - result = self.prev_result_cache.get(key) - result = self.next_result_cache.get(key, result) - result = self.pending_result_cache.get(key, result) - if result is not None: - return result.observe() - else: - return None - - def set(self, time_now_ms, key, deferred): - self.rotate(time_now_ms) - - result = ObservableDeferred(deferred) - - self.pending_result_cache[key] = result - - def shuffle_along(r): - # When the deferred completes we shuffle it along to the first - # generation of the result cache. So that it will eventually - # expire from the rotation of that cache. - self.next_result_cache[key] = result - self.pending_result_cache.pop(key, None) - return r - - result.addBoth(shuffle_along) - - return result.observe() diff --git a/tests/util/test_snapshot_cache.py b/tests/util/test_snapshot_cache.py deleted file mode 100644 index 1a44f72425..0000000000 --- a/tests/util/test_snapshot_cache.py +++ /dev/null @@ -1,63 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2015, 2016 OpenMarket Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - - -from twisted.internet.defer import Deferred - -from synapse.util.caches.snapshot_cache import SnapshotCache - -from .. import unittest - - -class SnapshotCacheTestCase(unittest.TestCase): - def setUp(self): - self.cache = SnapshotCache() - self.cache.DURATION_MS = 1 - - def test_get_set(self): - # Check that getting a missing key returns None - self.assertEquals(self.cache.get(0, "key"), None) - - # Check that setting a key with a deferred returns - # a deferred that resolves when the initial deferred does - d = Deferred() - set_result = self.cache.set(0, "key", d) - self.assertIsNotNone(set_result) - self.assertFalse(set_result.called) - - # Check that getting the key before the deferred has resolved - # returns a deferred that resolves when the initial deferred does. - get_result_at_10 = self.cache.get(10, "key") - self.assertIsNotNone(get_result_at_10) - self.assertFalse(get_result_at_10.called) - - # Check that the returned deferreds resolve when the initial deferred - # does. - d.callback("v") - self.assertTrue(set_result.called) - self.assertTrue(get_result_at_10.called) - - # Check that getting the key after the deferred has resolved - # before the cache expires returns a resolved deferred. - get_result_at_11 = self.cache.get(11, "key") - self.assertIsNotNone(get_result_at_11) - if isinstance(get_result_at_11, Deferred): - # The cache may return the actual result rather than a deferred - self.assertTrue(get_result_at_11.called) - - # Check that getting the key after the deferred has resolved - # after the cache expires returns None - get_result_at_12 = self.cache.get(12, "key") - self.assertIsNone(get_result_at_12) -- cgit 1.4.1 From adfdd82b21ae296ed77453b2f51d55414890f162 Mon Sep 17 00:00:00 2001 From: Neil Johnson Date: Mon, 9 Dec 2019 13:59:27 +0000 Subject: Back out perf regression from get_cross_signing_keys_from_cache. (#6494) Back out cross-signing code added in Synapse 1.5.0, which caused a performance regression. --- changelog.d/6494.bugfix | 1 + synapse/handlers/e2e_keys.py | 38 ++++++++------------------------------ sytest-blacklist | 3 +++ tests/handlers/test_e2e_keys.py | 8 ++++++++ 4 files changed, 20 insertions(+), 30 deletions(-) create mode 100644 changelog.d/6494.bugfix (limited to 'tests') diff --git a/changelog.d/6494.bugfix b/changelog.d/6494.bugfix new file mode 100644 index 0000000000..78726d5d7f --- /dev/null +++ b/changelog.d/6494.bugfix @@ -0,0 +1 @@ +Back out cross-signing code added in Synapse 1.5.0, which caused a performance regression. diff --git a/synapse/handlers/e2e_keys.py b/synapse/handlers/e2e_keys.py index 28c12753c1..57a10daefd 100644 --- a/synapse/handlers/e2e_keys.py +++ b/synapse/handlers/e2e_keys.py @@ -264,7 +264,6 @@ class E2eKeysHandler(object): return ret - @defer.inlineCallbacks def get_cross_signing_keys_from_cache(self, query, from_user_id): """Get cross-signing keys for users from the database @@ -284,35 +283,14 @@ class E2eKeysHandler(object): self_signing_keys = {} user_signing_keys = {} - for user_id in query: - # XXX: consider changing the store functions to allow querying - # multiple users simultaneously. - key = yield self.store.get_e2e_cross_signing_key( - user_id, "master", from_user_id - ) - if key: - master_keys[user_id] = key - - key = yield self.store.get_e2e_cross_signing_key( - user_id, "self_signing", from_user_id - ) - if key: - self_signing_keys[user_id] = key - - # users can see other users' master and self-signing keys, but can - # only see their own user-signing keys - if from_user_id == user_id: - key = yield self.store.get_e2e_cross_signing_key( - user_id, "user_signing", from_user_id - ) - if key: - user_signing_keys[user_id] = key - - return { - "master_keys": master_keys, - "self_signing_keys": self_signing_keys, - "user_signing_keys": user_signing_keys, - } + # Currently a stub, implementation coming in https://github.com/matrix-org/synapse/pull/6486 + return defer.succeed( + { + "master_keys": master_keys, + "self_signing_keys": self_signing_keys, + "user_signing_keys": user_signing_keys, + } + ) @trace @defer.inlineCallbacks diff --git a/sytest-blacklist b/sytest-blacklist index 411cce0692..79b2d4402a 100644 --- a/sytest-blacklist +++ b/sytest-blacklist @@ -33,3 +33,6 @@ New federated private chats get full presence information (SYN-115) # Blacklisted due to https://github.com/matrix-org/matrix-doc/pull/2314 removing # this requirement from the spec Inbound federation of state requires event_id as a mandatory paramater + +# Blacklisted until https://github.com/matrix-org/synapse/pull/6486 lands +Can upload self-signing keys diff --git a/tests/handlers/test_e2e_keys.py b/tests/handlers/test_e2e_keys.py index 854eb6c024..fdfa2cbbc4 100644 --- a/tests/handlers/test_e2e_keys.py +++ b/tests/handlers/test_e2e_keys.py @@ -183,6 +183,10 @@ class E2eKeysHandlerTestCase(unittest.TestCase): ) self.assertDictEqual(devices["master_keys"], {local_user: keys2["master_key"]}) + test_replace_master_key.skip = ( + "Disabled waiting on #https://github.com/matrix-org/synapse/pull/6486" + ) + @defer.inlineCallbacks def test_reupload_signatures(self): """re-uploading a signature should not fail""" @@ -503,3 +507,7 @@ class E2eKeysHandlerTestCase(unittest.TestCase): ], other_master_key["signatures"][local_user]["ed25519:" + usersigning_pubkey], ) + + test_upload_signatures.skip = ( + "Disabled waiting on #https://github.com/matrix-org/synapse/pull/6486" + ) -- cgit 1.4.1 From 9a2223d4c8c2a46f7ab072597bdba2614bc3e6ea Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Tue, 10 Dec 2019 11:22:12 +0000 Subject: Fix make_deferred_yieldable to work with coroutines --- synapse/logging/context.py | 9 ++++++++- tests/util/test_logcontext.py | 24 ++++++++++++++++++++++++ 2 files changed, 32 insertions(+), 1 deletion(-) (limited to 'tests') diff --git a/synapse/logging/context.py b/synapse/logging/context.py index 2c1fb9ddac..9f484ce59e 100644 --- a/synapse/logging/context.py +++ b/synapse/logging/context.py @@ -23,6 +23,7 @@ them. See doc/log_contexts.rst for details on how this works. """ +import inspect import logging import threading import types @@ -612,7 +613,8 @@ def run_in_background(f, *args, **kwargs): def make_deferred_yieldable(deferred): - """Given a deferred, make it follow the Synapse logcontext rules: + """Given a deferred (or coroutine), make it follow the Synapse logcontext + rules: If the deferred has completed (or is not actually a Deferred), essentially does nothing (just returns another completed deferred with the @@ -624,6 +626,11 @@ def make_deferred_yieldable(deferred): (This is more-or-less the opposite operation to run_in_background.) """ + if inspect.isawaitable(deferred): + # If we're given a coroutine we need to convert it to a deferred so that + # we can attach callbacks (and not immediately return). + deferred = defer.ensureDeferred(deferred) + if not isinstance(deferred, defer.Deferred): return deferred diff --git a/tests/util/test_logcontext.py b/tests/util/test_logcontext.py index 8b8455c8b7..281b32c4b8 100644 --- a/tests/util/test_logcontext.py +++ b/tests/util/test_logcontext.py @@ -179,6 +179,30 @@ class LoggingContextTestCase(unittest.TestCase): nested_context = nested_logging_context(suffix="bar") self.assertEqual(nested_context.request, "foo-bar") + @defer.inlineCallbacks + def test_make_deferred_yieldable_with_await(self): + # an async function which retuns an incomplete coroutine, but doesn't + # follow the synapse rules. + + async def blocking_function(): + d = defer.Deferred() + reactor.callLater(0, d.callback, None) + await d + + sentinel_context = LoggingContext.current_context() + + with LoggingContext() as context_one: + context_one.request = "one" + + d1 = make_deferred_yieldable(blocking_function()) + # make sure that the context was reset by make_deferred_yieldable + self.assertIs(LoggingContext.current_context(), sentinel_context) + + yield d1 + + # now it should be restored + self._check_test_key("one") + # a function which returns a deferred which has been "called", but # which had a function which returned another incomplete deferred on -- cgit 1.4.1 From bc5cb8bfe8c5b0b551de9a97912cd00caeaf6b48 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Fri, 6 Dec 2019 17:42:18 +0000 Subject: Remove database config parsing from apps. --- synapse/app/admin_cmd.py | 5 ----- synapse/app/appservice.py | 5 ----- synapse/app/client_reader.py | 5 ----- synapse/app/event_creator.py | 5 ----- synapse/app/federation_reader.py | 5 ----- synapse/app/federation_sender.py | 5 ----- synapse/app/frontend_proxy.py | 5 ----- synapse/app/homeserver.py | 7 +------ synapse/app/media_repository.py | 5 ----- synapse/app/pusher.py | 5 ----- synapse/app/synchrotron.py | 5 ----- synapse/app/user_dir.py | 5 ----- synapse/server.py | 10 +++++++++- tests/test_types.py | 19 ++++++++----------- tests/utils.py | 2 -- 15 files changed, 18 insertions(+), 75 deletions(-) (limited to 'tests') diff --git a/synapse/app/admin_cmd.py b/synapse/app/admin_cmd.py index 04751a6a5e..51a909419f 100644 --- a/synapse/app/admin_cmd.py +++ b/synapse/app/admin_cmd.py @@ -45,7 +45,6 @@ from synapse.replication.slave.storage.registration import SlavedRegistrationSto from synapse.replication.slave.storage.room import RoomStore from synapse.replication.tcp.client import ReplicationClientHandler from synapse.server import HomeServer -from synapse.storage.engines import create_engine from synapse.util.logcontext import LoggingContext from synapse.util.versionstring import get_version_string @@ -229,14 +228,10 @@ def start(config_options): synapse.events.USE_FROZEN_DICTS = config.use_frozen_dicts - database_engine = create_engine(config.database_config) - ss = AdminCmdServer( config.server_name, - db_config=config.database_config, config=config, version_string="Synapse/" + get_version_string(synapse), - database_engine=database_engine, ) setup_logging(ss, config, use_worker_options=True) diff --git a/synapse/app/appservice.py b/synapse/app/appservice.py index 02b900f382..e82e0f11e3 100644 --- a/synapse/app/appservice.py +++ b/synapse/app/appservice.py @@ -34,7 +34,6 @@ from synapse.replication.slave.storage.events import SlavedEventStore from synapse.replication.slave.storage.registration import SlavedRegistrationStore from synapse.replication.tcp.client import ReplicationClientHandler from synapse.server import HomeServer -from synapse.storage.engines import create_engine from synapse.util.httpresourcetree import create_resource_tree from synapse.util.manhole import manhole from synapse.util.versionstring import get_version_string @@ -143,8 +142,6 @@ def start(config_options): events.USE_FROZEN_DICTS = config.use_frozen_dicts - database_engine = create_engine(config.database_config) - if config.notify_appservices: sys.stderr.write( "\nThe appservices must be disabled in the main synapse process" @@ -159,10 +156,8 @@ def start(config_options): ps = AppserviceServer( config.server_name, - db_config=config.database_config, config=config, version_string="Synapse/" + get_version_string(synapse), - database_engine=database_engine, ) setup_logging(ps, config, use_worker_options=True) diff --git a/synapse/app/client_reader.py b/synapse/app/client_reader.py index dadb487d5f..3edfe19567 100644 --- a/synapse/app/client_reader.py +++ b/synapse/app/client_reader.py @@ -62,7 +62,6 @@ from synapse.rest.client.v2_alpha.keys import KeyChangesServlet, KeyQueryServlet from synapse.rest.client.v2_alpha.register import RegisterRestServlet from synapse.rest.client.versions import VersionsRestServlet from synapse.server import HomeServer -from synapse.storage.engines import create_engine from synapse.util.httpresourcetree import create_resource_tree from synapse.util.manhole import manhole from synapse.util.versionstring import get_version_string @@ -181,14 +180,10 @@ def start(config_options): events.USE_FROZEN_DICTS = config.use_frozen_dicts - database_engine = create_engine(config.database_config) - ss = ClientReaderServer( config.server_name, - db_config=config.database_config, config=config, version_string="Synapse/" + get_version_string(synapse), - database_engine=database_engine, ) setup_logging(ss, config, use_worker_options=True) diff --git a/synapse/app/event_creator.py b/synapse/app/event_creator.py index d110599a35..d0ddbe38fc 100644 --- a/synapse/app/event_creator.py +++ b/synapse/app/event_creator.py @@ -57,7 +57,6 @@ from synapse.rest.client.v1.room import ( ) from synapse.server import HomeServer from synapse.storage.data_stores.main.user_directory import UserDirectoryStore -from synapse.storage.engines import create_engine from synapse.util.httpresourcetree import create_resource_tree from synapse.util.manhole import manhole from synapse.util.versionstring import get_version_string @@ -180,14 +179,10 @@ def start(config_options): events.USE_FROZEN_DICTS = config.use_frozen_dicts - database_engine = create_engine(config.database_config) - ss = EventCreatorServer( config.server_name, - db_config=config.database_config, config=config, version_string="Synapse/" + get_version_string(synapse), - database_engine=database_engine, ) setup_logging(ss, config, use_worker_options=True) diff --git a/synapse/app/federation_reader.py b/synapse/app/federation_reader.py index 418c086254..311523e0ed 100644 --- a/synapse/app/federation_reader.py +++ b/synapse/app/federation_reader.py @@ -46,7 +46,6 @@ from synapse.replication.slave.storage.transactions import SlavedTransactionStor from synapse.replication.tcp.client import ReplicationClientHandler from synapse.rest.key.v2 import KeyApiV2Resource from synapse.server import HomeServer -from synapse.storage.engines import create_engine from synapse.util.httpresourcetree import create_resource_tree from synapse.util.manhole import manhole from synapse.util.versionstring import get_version_string @@ -162,14 +161,10 @@ def start(config_options): events.USE_FROZEN_DICTS = config.use_frozen_dicts - database_engine = create_engine(config.database_config) - ss = FederationReaderServer( config.server_name, - db_config=config.database_config, config=config, version_string="Synapse/" + get_version_string(synapse), - database_engine=database_engine, ) setup_logging(ss, config, use_worker_options=True) diff --git a/synapse/app/federation_sender.py b/synapse/app/federation_sender.py index f24920a7d6..83c436229c 100644 --- a/synapse/app/federation_sender.py +++ b/synapse/app/federation_sender.py @@ -41,7 +41,6 @@ from synapse.replication.tcp.client import ReplicationClientHandler from synapse.replication.tcp.streams._base import ReceiptsStream from synapse.server import HomeServer from synapse.storage.database import Database -from synapse.storage.engines import create_engine from synapse.types import ReadReceipt from synapse.util.async_helpers import Linearizer from synapse.util.httpresourcetree import create_resource_tree @@ -174,8 +173,6 @@ def start(config_options): events.USE_FROZEN_DICTS = config.use_frozen_dicts - database_engine = create_engine(config.database_config) - if config.send_federation: sys.stderr.write( "\nThe send_federation must be disabled in the main synapse process" @@ -190,10 +187,8 @@ def start(config_options): ss = FederationSenderServer( config.server_name, - db_config=config.database_config, config=config, version_string="Synapse/" + get_version_string(synapse), - database_engine=database_engine, ) setup_logging(ss, config, use_worker_options=True) diff --git a/synapse/app/frontend_proxy.py b/synapse/app/frontend_proxy.py index e647459d0e..30e435eead 100644 --- a/synapse/app/frontend_proxy.py +++ b/synapse/app/frontend_proxy.py @@ -39,7 +39,6 @@ from synapse.replication.slave.storage.registration import SlavedRegistrationSto from synapse.replication.tcp.client import ReplicationClientHandler from synapse.rest.client.v2_alpha._base import client_patterns from synapse.server import HomeServer -from synapse.storage.engines import create_engine from synapse.util.httpresourcetree import create_resource_tree from synapse.util.manhole import manhole from synapse.util.versionstring import get_version_string @@ -234,14 +233,10 @@ def start(config_options): events.USE_FROZEN_DICTS = config.use_frozen_dicts - database_engine = create_engine(config.database_config) - ss = FrontendProxyServer( config.server_name, - db_config=config.database_config, config=config, version_string="Synapse/" + get_version_string(synapse), - database_engine=database_engine, ) setup_logging(ss, config, use_worker_options=True) diff --git a/synapse/app/homeserver.py b/synapse/app/homeserver.py index df65d0a989..79341784c5 100644 --- a/synapse/app/homeserver.py +++ b/synapse/app/homeserver.py @@ -69,7 +69,7 @@ from synapse.rest.media.v0.content_repository import ContentRepoResource from synapse.rest.well_known import WellKnownResource from synapse.server import HomeServer from synapse.storage import DataStore -from synapse.storage.engines import IncorrectDatabaseSetup, create_engine +from synapse.storage.engines import IncorrectDatabaseSetup from synapse.storage.prepare_database import UpgradeDatabaseException from synapse.util.caches import CACHE_SIZE_FACTOR from synapse.util.httpresourcetree import create_resource_tree @@ -328,15 +328,10 @@ def setup(config_options): events.USE_FROZEN_DICTS = config.use_frozen_dicts - database_engine = create_engine(config.database_config) - config.database_config["args"]["cp_openfun"] = database_engine.on_new_connection - hs = SynapseHomeServer( config.server_name, - db_config=config.database_config, config=config, version_string="Synapse/" + get_version_string(synapse), - database_engine=database_engine, ) synapse.config.logger.setup_logging(hs, config, use_worker_options=False) diff --git a/synapse/app/media_repository.py b/synapse/app/media_repository.py index 2c6dd3ef02..4c80f257e2 100644 --- a/synapse/app/media_repository.py +++ b/synapse/app/media_repository.py @@ -40,7 +40,6 @@ from synapse.rest.admin import register_servlets_for_media_repo from synapse.rest.media.v0.content_repository import ContentRepoResource from synapse.server import HomeServer from synapse.storage.data_stores.main.media_repository import MediaRepositoryStore -from synapse.storage.engines import create_engine from synapse.util.httpresourcetree import create_resource_tree from synapse.util.manhole import manhole from synapse.util.versionstring import get_version_string @@ -157,14 +156,10 @@ def start(config_options): events.USE_FROZEN_DICTS = config.use_frozen_dicts - database_engine = create_engine(config.database_config) - ss = MediaRepositoryServer( config.server_name, - db_config=config.database_config, config=config, version_string="Synapse/" + get_version_string(synapse), - database_engine=database_engine, ) setup_logging(ss, config, use_worker_options=True) diff --git a/synapse/app/pusher.py b/synapse/app/pusher.py index 01a5ffc363..e157cbf64b 100644 --- a/synapse/app/pusher.py +++ b/synapse/app/pusher.py @@ -36,7 +36,6 @@ from synapse.replication.slave.storage.receipts import SlavedReceiptsStore from synapse.replication.tcp.client import ReplicationClientHandler from synapse.server import HomeServer from synapse.storage import DataStore -from synapse.storage.engines import create_engine from synapse.util.httpresourcetree import create_resource_tree from synapse.util.manhole import manhole from synapse.util.versionstring import get_version_string @@ -198,14 +197,10 @@ def start(config_options): # Force the pushers to start since they will be disabled in the main config config.start_pushers = True - database_engine = create_engine(config.database_config) - ps = PusherServer( config.server_name, - db_config=config.database_config, config=config, version_string="Synapse/" + get_version_string(synapse), - database_engine=database_engine, ) setup_logging(ps, config, use_worker_options=True) diff --git a/synapse/app/synchrotron.py b/synapse/app/synchrotron.py index 288ee64b42..dd2132e608 100644 --- a/synapse/app/synchrotron.py +++ b/synapse/app/synchrotron.py @@ -55,7 +55,6 @@ from synapse.rest.client.v1.room import RoomInitialSyncRestServlet from synapse.rest.client.v2_alpha import sync from synapse.server import HomeServer from synapse.storage.data_stores.main.presence import UserPresenceState -from synapse.storage.engines import create_engine from synapse.util.httpresourcetree import create_resource_tree from synapse.util.manhole import manhole from synapse.util.stringutils import random_string @@ -437,14 +436,10 @@ def start(config_options): synapse.events.USE_FROZEN_DICTS = config.use_frozen_dicts - database_engine = create_engine(config.database_config) - ss = SynchrotronServer( config.server_name, - db_config=config.database_config, config=config, version_string="Synapse/" + get_version_string(synapse), - database_engine=database_engine, application_service_handler=SynchrotronApplicationService(), ) diff --git a/synapse/app/user_dir.py b/synapse/app/user_dir.py index c01fb34a9b..1257098f92 100644 --- a/synapse/app/user_dir.py +++ b/synapse/app/user_dir.py @@ -44,7 +44,6 @@ from synapse.rest.client.v2_alpha import user_directory from synapse.server import HomeServer from synapse.storage.data_stores.main.user_directory import UserDirectoryStore from synapse.storage.database import Database -from synapse.storage.engines import create_engine from synapse.util.caches.stream_change_cache import StreamChangeCache from synapse.util.httpresourcetree import create_resource_tree from synapse.util.manhole import manhole @@ -200,8 +199,6 @@ def start(config_options): events.USE_FROZEN_DICTS = config.use_frozen_dicts - database_engine = create_engine(config.database_config) - if config.update_user_directory: sys.stderr.write( "\nThe update_user_directory must be disabled in the main synapse process" @@ -216,10 +213,8 @@ def start(config_options): ss = UserDirectoryServer( config.server_name, - db_config=config.database_config, config=config, version_string="Synapse/" + get_version_string(synapse), - database_engine=database_engine, ) setup_logging(ss, config, use_worker_options=True) diff --git a/synapse/server.py b/synapse/server.py index 2db3dab221..a8b5459ff3 100644 --- a/synapse/server.py +++ b/synapse/server.py @@ -97,6 +97,7 @@ from synapse.server_notices.worker_server_notices_sender import ( ) from synapse.state import StateHandler, StateResolutionHandler from synapse.storage import DataStores, Storage +from synapse.storage.engines import create_engine from synapse.streams.events import EventSources from synapse.util import Clock from synapse.util.distributor import Distributor @@ -209,7 +210,7 @@ class HomeServer(object): # instantiated during setup() for future return by get_datastore() DATASTORE_CLASS = abc.abstractproperty() - def __init__(self, hostname, reactor=None, **kwargs): + def __init__(self, hostname, config, reactor=None, **kwargs): """ Args: hostname : The hostname for the server. @@ -219,6 +220,7 @@ class HomeServer(object): self._reactor = reactor self.hostname = hostname + self.config = config self._building = {} self._listening_services = [] self.start_time = None @@ -229,6 +231,12 @@ class HomeServer(object): self.admin_redaction_ratelimiter = Ratelimiter() self.registration_ratelimiter = Ratelimiter() + self.database_engine = create_engine(config.database_config) + config.database_config.setdefault("args", {})[ + "cp_openfun" + ] = self.database_engine.on_new_connection + self.db_config = config.database_config + self.datastores = None # Other kwargs are explicit dependencies diff --git a/tests/test_types.py b/tests/test_types.py index 9ab5f829b0..8d97c751ea 100644 --- a/tests/test_types.py +++ b/tests/test_types.py @@ -17,18 +17,15 @@ from synapse.api.errors import SynapseError from synapse.types import GroupID, RoomAlias, UserID, map_username_to_mxid_localpart from tests import unittest -from tests.utils import TestHomeServer -mock_homeserver = TestHomeServer(hostname="my.domain") - -class UserIDTestCase(unittest.TestCase): +class UserIDTestCase(unittest.HomeserverTestCase): def test_parse(self): - user = UserID.from_string("@1234abcd:my.domain") + user = UserID.from_string("@1234abcd:test") self.assertEquals("1234abcd", user.localpart) - self.assertEquals("my.domain", user.domain) - self.assertEquals(True, mock_homeserver.is_mine(user)) + self.assertEquals("test", user.domain) + self.assertEquals(True, self.hs.is_mine(user)) def test_pase_empty(self): with self.assertRaises(SynapseError): @@ -48,13 +45,13 @@ class UserIDTestCase(unittest.TestCase): self.assertTrue(userA != userB) -class RoomAliasTestCase(unittest.TestCase): +class RoomAliasTestCase(unittest.HomeserverTestCase): def test_parse(self): - room = RoomAlias.from_string("#channel:my.domain") + room = RoomAlias.from_string("#channel:test") self.assertEquals("channel", room.localpart) - self.assertEquals("my.domain", room.domain) - self.assertEquals(True, mock_homeserver.is_mine(room)) + self.assertEquals("test", room.domain) + self.assertEquals(True, self.hs.is_mine(room)) def test_build(self): room = RoomAlias("channel", "my.domain") diff --git a/tests/utils.py b/tests/utils.py index c57da59191..585f305b9a 100644 --- a/tests/utils.py +++ b/tests/utils.py @@ -260,9 +260,7 @@ def setup_test_homeserver( hs = homeserverToUse( name, config=config, - db_config=config.database_config, version_string="Synapse/tests", - database_engine=db_engine, tls_server_context_factory=Mock(), tls_client_options_factory=Mock(), reactor=reactor, -- cgit 1.4.1 From 257ef2c727abbb289c460b930e8bac18ab80e053 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Tue, 10 Dec 2019 11:13:15 +0000 Subject: Port handlers.account_validity to async/await. --- synapse/handlers/account_data.py | 2 +- synapse/handlers/account_validity.py | 86 ++++++++++++++--------------- tests/rest/client/v2_alpha/test_register.py | 3 +- 3 files changed, 42 insertions(+), 49 deletions(-) (limited to 'tests') diff --git a/synapse/handlers/account_data.py b/synapse/handlers/account_data.py index fe44d62a1c..20ec1ca01b 100644 --- a/synapse/handlers/account_data.py +++ b/synapse/handlers/account_data.py @@ -25,7 +25,7 @@ class AccountDataEventSource(object): user_id = user.to_string() last_stream_id = from_key - current_stream_id = await self.store.get_max_account_data_stream_id() + current_stream_id = self.store.get_max_account_data_stream_id() results = [] tags = await self.store.get_updated_tags(user_id, last_stream_id) diff --git a/synapse/handlers/account_validity.py b/synapse/handlers/account_validity.py index d04e0fe576..829f52eca1 100644 --- a/synapse/handlers/account_validity.py +++ b/synapse/handlers/account_validity.py @@ -18,8 +18,7 @@ import email.utils import logging from email.mime.multipart import MIMEMultipart from email.mime.text import MIMEText - -from twisted.internet import defer +from typing import List from synapse.api.errors import StoreError from synapse.logging.context import make_deferred_yieldable @@ -78,42 +77,39 @@ class AccountValidityHandler(object): # run as a background process to make sure that the database transactions # have a logcontext to report to return run_as_background_process( - "send_renewals", self.send_renewal_emails + "send_renewals", self._send_renewal_emails ) self.clock.looping_call(send_emails, 30 * 60 * 1000) - @defer.inlineCallbacks - def send_renewal_emails(self): + async def _send_renewal_emails(self): """Gets the list of users whose account is expiring in the amount of time configured in the ``renew_at`` parameter from the ``account_validity`` configuration, and sends renewal emails to all of these users as long as they have an email 3PID attached to their account. """ - expiring_users = yield self.store.get_users_expiring_soon() + expiring_users = await self.store.get_users_expiring_soon() if expiring_users: for user in expiring_users: - yield self._send_renewal_email( + await self._send_renewal_email( user_id=user["user_id"], expiration_ts=user["expiration_ts_ms"] ) - @defer.inlineCallbacks - def send_renewal_email_to_user(self, user_id): - expiration_ts = yield self.store.get_expiration_ts_for_user(user_id) - yield self._send_renewal_email(user_id, expiration_ts) + async def send_renewal_email_to_user(self, user_id: str): + expiration_ts = await self.store.get_expiration_ts_for_user(user_id) + await self._send_renewal_email(user_id, expiration_ts) - @defer.inlineCallbacks - def _send_renewal_email(self, user_id, expiration_ts): + async def _send_renewal_email(self, user_id: str, expiration_ts: int): """Sends out a renewal email to every email address attached to the given user with a unique link allowing them to renew their account. Args: - user_id (str): ID of the user to send email(s) to. - expiration_ts (int): Timestamp in milliseconds for the expiration date of + user_id: ID of the user to send email(s) to. + expiration_ts: Timestamp in milliseconds for the expiration date of this user's account (used in the email templates). """ - addresses = yield self._get_email_addresses_for_user(user_id) + addresses = await self._get_email_addresses_for_user(user_id) # Stop right here if the user doesn't have at least one email address. # In this case, they will have to ask their server admin to renew their @@ -125,7 +121,7 @@ class AccountValidityHandler(object): return try: - user_display_name = yield self.store.get_profile_displayname( + user_display_name = await self.store.get_profile_displayname( UserID.from_string(user_id).localpart ) if user_display_name is None: @@ -133,7 +129,7 @@ class AccountValidityHandler(object): except StoreError: user_display_name = user_id - renewal_token = yield self._get_renewal_token(user_id) + renewal_token = await self._get_renewal_token(user_id) url = "%s_matrix/client/unstable/account_validity/renew?token=%s" % ( self.hs.config.public_baseurl, renewal_token, @@ -165,7 +161,7 @@ class AccountValidityHandler(object): logger.info("Sending renewal email to %s", address) - yield make_deferred_yieldable( + await make_deferred_yieldable( self.sendmail( self.hs.config.email_smtp_host, self._raw_from, @@ -180,19 +176,18 @@ class AccountValidityHandler(object): ) ) - yield self.store.set_renewal_mail_status(user_id=user_id, email_sent=True) + await self.store.set_renewal_mail_status(user_id=user_id, email_sent=True) - @defer.inlineCallbacks - def _get_email_addresses_for_user(self, user_id): + async def _get_email_addresses_for_user(self, user_id: str) -> List[str]: """Retrieve the list of email addresses attached to a user's account. Args: - user_id (str): ID of the user to lookup email addresses for. + user_id: ID of the user to lookup email addresses for. Returns: - defer.Deferred[list[str]]: Email addresses for this account. + Email addresses for this account. """ - threepids = yield self.store.user_get_threepids(user_id) + threepids = await self.store.user_get_threepids(user_id) addresses = [] for threepid in threepids: @@ -201,16 +196,15 @@ class AccountValidityHandler(object): return addresses - @defer.inlineCallbacks - def _get_renewal_token(self, user_id): + async def _get_renewal_token(self, user_id: str) -> str: """Generates a 32-byte long random string that will be inserted into the user's renewal email's unique link, then saves it into the database. Args: - user_id (str): ID of the user to generate a string for. + user_id: ID of the user to generate a string for. Returns: - defer.Deferred[str]: The generated string. + The generated string. Raises: StoreError(500): Couldn't generate a unique string after 5 attempts. @@ -219,52 +213,52 @@ class AccountValidityHandler(object): while attempts < 5: try: renewal_token = stringutils.random_string(32) - yield self.store.set_renewal_token_for_user(user_id, renewal_token) + await self.store.set_renewal_token_for_user(user_id, renewal_token) return renewal_token except StoreError: attempts += 1 raise StoreError(500, "Couldn't generate a unique string as refresh string.") - @defer.inlineCallbacks - def renew_account(self, renewal_token): + async def renew_account(self, renewal_token: str) -> bool: """Renews the account attached to a given renewal token by pushing back the expiration date by the current validity period in the server's configuration. Args: - renewal_token (str): Token sent with the renewal request. + renewal_token: Token sent with the renewal request. Returns: - bool: Whether the provided token is valid. + Whether the provided token is valid. """ try: - user_id = yield self.store.get_user_from_renewal_token(renewal_token) + user_id = await self.store.get_user_from_renewal_token(renewal_token) except StoreError: - defer.returnValue(False) + return False logger.debug("Renewing an account for user %s", user_id) - yield self.renew_account_for_user(user_id) + await self.renew_account_for_user(user_id) - defer.returnValue(True) + return True - @defer.inlineCallbacks - def renew_account_for_user(self, user_id, expiration_ts=None, email_sent=False): + async def renew_account_for_user( + self, user_id: str, expiration_ts: int = None, email_sent: bool = False + ) -> int: """Renews the account attached to a given user by pushing back the expiration date by the current validity period in the server's configuration. Args: - renewal_token (str): Token sent with the renewal request. - expiration_ts (int): New expiration date. Defaults to now + validity period. - email_sent (bool): Whether an email has been sent for this validity period. + renewal_token: Token sent with the renewal request. + expiration_ts: New expiration date. Defaults to now + validity period. + email_sen: Whether an email has been sent for this validity period. Defaults to False. Returns: - defer.Deferred[int]: New expiration date for this account, as a timestamp - in milliseconds since epoch. + New expiration date for this account, as a timestamp in + milliseconds since epoch. """ if expiration_ts is None: expiration_ts = self.clock.time_msec() + self._account_validity.period - yield self.store.set_account_validity_for_user( + await self.store.set_account_validity_for_user( user_id=user_id, expiration_ts=expiration_ts, email_sent=email_sent ) diff --git a/tests/rest/client/v2_alpha/test_register.py b/tests/rest/client/v2_alpha/test_register.py index c0d0d2b44e..d0c997e385 100644 --- a/tests/rest/client/v2_alpha/test_register.py +++ b/tests/rest/client/v2_alpha/test_register.py @@ -391,9 +391,8 @@ class AccountValidityRenewalByEmailTestCase(unittest.HomeserverTestCase): # Email config. self.email_attempts = [] - def sendmail(*args, **kwargs): + async def sendmail(*args, **kwargs): self.email_attempts.append((args, kwargs)) - return config["email"] = { "enable_notifs": True, -- cgit 1.4.1 From 40eda849338b6e47a5804b4cf7000e9d2417c4d8 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Tue, 10 Dec 2019 16:22:29 +0000 Subject: Fix race which caused deleted devices to reappear (#6514) Stop the `update_client_ips` background job from recreating deleted devices. --- changelog.d/6514.bugfix | 1 + synapse/storage/data_stores/main/client_ips.py | 8 +++-- tests/storage/test_client_ips.py | 49 +++++++++++++++----------- 3 files changed, 35 insertions(+), 23 deletions(-) create mode 100644 changelog.d/6514.bugfix (limited to 'tests') diff --git a/changelog.d/6514.bugfix b/changelog.d/6514.bugfix new file mode 100644 index 0000000000..6dc1985c24 --- /dev/null +++ b/changelog.d/6514.bugfix @@ -0,0 +1 @@ +Fix race which occasionally caused deleted devices to reappear. diff --git a/synapse/storage/data_stores/main/client_ips.py b/synapse/storage/data_stores/main/client_ips.py index 320c5b0f07..add3037b69 100644 --- a/synapse/storage/data_stores/main/client_ips.py +++ b/synapse/storage/data_stores/main/client_ips.py @@ -451,16 +451,18 @@ class ClientIpStore(ClientIpBackgroundUpdateStore): # Technically an access token might not be associated with # a device so we need to check. if device_id: - self.db.simple_upsert_txn( + # this is always an update rather than an upsert: the row should + # already exist, and if it doesn't, that may be because it has been + # deleted, and we don't want to re-create it. + self.db.simple_update_txn( txn, table="devices", keyvalues={"user_id": user_id, "device_id": device_id}, - values={ + updatevalues={ "user_agent": user_agent, "last_seen": last_seen, "ip": ip, }, - lock=False, ) except Exception as e: # Failed to upsert, log and continue diff --git a/tests/storage/test_client_ips.py b/tests/storage/test_client_ips.py index fc279340d4..bf674dd184 100644 --- a/tests/storage/test_client_ips.py +++ b/tests/storage/test_client_ips.py @@ -37,9 +37,13 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): self.reactor.advance(12345678) user_id = "@user:id" + device_id = "MY_DEVICE" + + # Insert a user IP + self.get_success(self.store.store_device(user_id, device_id, "display name",)) self.get_success( self.store.insert_client_ip( - user_id, "access_token", "ip", "user_agent", "device_id" + user_id, "access_token", "ip", "user_agent", device_id ) ) @@ -47,14 +51,14 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): self.reactor.advance(10) result = self.get_success( - self.store.get_last_client_ip_by_device(user_id, "device_id") + self.store.get_last_client_ip_by_device(user_id, device_id) ) - r = result[(user_id, "device_id")] + r = result[(user_id, device_id)] self.assertDictContainsSubset( { "user_id": user_id, - "device_id": "device_id", + "device_id": device_id, "ip": "ip", "user_agent": "user_agent", "last_seen": 12345678000, @@ -209,14 +213,16 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): self.store.db.updates.do_next_background_update(100), by=0.1 ) - # Insert a user IP user_id = "@user:id" + device_id = "MY_DEVICE" + + # Insert a user IP + self.get_success(self.store.store_device(user_id, device_id, "display name",)) self.get_success( self.store.insert_client_ip( - user_id, "access_token", "ip", "user_agent", "device_id" + user_id, "access_token", "ip", "user_agent", device_id ) ) - # Force persisting to disk self.reactor.advance(200) @@ -224,7 +230,7 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): self.get_success( self.store.db.simple_update( table="devices", - keyvalues={"user_id": user_id, "device_id": "device_id"}, + keyvalues={"user_id": user_id, "device_id": device_id}, updatevalues={"last_seen": None, "ip": None, "user_agent": None}, desc="test_devices_last_seen_bg_update", ) @@ -232,14 +238,14 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): # We should now get nulls when querying result = self.get_success( - self.store.get_last_client_ip_by_device(user_id, "device_id") + self.store.get_last_client_ip_by_device(user_id, device_id) ) - r = result[(user_id, "device_id")] + r = result[(user_id, device_id)] self.assertDictContainsSubset( { "user_id": user_id, - "device_id": "device_id", + "device_id": device_id, "ip": None, "user_agent": None, "last_seen": None, @@ -272,14 +278,14 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): # We should now get the correct result again result = self.get_success( - self.store.get_last_client_ip_by_device(user_id, "device_id") + self.store.get_last_client_ip_by_device(user_id, device_id) ) - r = result[(user_id, "device_id")] + r = result[(user_id, device_id)] self.assertDictContainsSubset( { "user_id": user_id, - "device_id": "device_id", + "device_id": device_id, "ip": "ip", "user_agent": "user_agent", "last_seen": 0, @@ -296,11 +302,14 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): self.store.db.updates.do_next_background_update(100), by=0.1 ) - # Insert a user IP user_id = "@user:id" + device_id = "MY_DEVICE" + + # Insert a user IP + self.get_success(self.store.store_device(user_id, device_id, "display name",)) self.get_success( self.store.insert_client_ip( - user_id, "access_token", "ip", "user_agent", "device_id" + user_id, "access_token", "ip", "user_agent", device_id ) ) @@ -324,7 +333,7 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): "access_token": "access_token", "ip": "ip", "user_agent": "user_agent", - "device_id": "device_id", + "device_id": device_id, "last_seen": 0, } ], @@ -347,14 +356,14 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): # But we should still get the correct values for the device result = self.get_success( - self.store.get_last_client_ip_by_device(user_id, "device_id") + self.store.get_last_client_ip_by_device(user_id, device_id) ) - r = result[(user_id, "device_id")] + r = result[(user_id, device_id)] self.assertDictContainsSubset( { "user_id": user_id, - "device_id": "device_id", + "device_id": device_id, "ip": "ip", "user_agent": "user_agent", "last_seen": 0, -- cgit 1.4.1 From e77237b9350a7054657b1a641883a09c6b5d44f3 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Tue, 10 Dec 2019 17:01:37 +0000 Subject: convert to async: FederationHandler.on_receive_pdu and associated functions: * on_receive_pdu * handle_queued_pdus * get_missing_events_for_pdu --- synapse/handlers/federation.py | 49 +++++++++++++++++++----------------------- tests/test_federation.py | 14 +++++++----- 2 files changed, 31 insertions(+), 32 deletions(-) (limited to 'tests') diff --git a/synapse/handlers/federation.py b/synapse/handlers/federation.py index e54d509b62..01d9c5120e 100644 --- a/synapse/handlers/federation.py +++ b/synapse/handlers/federation.py @@ -165,8 +165,7 @@ class FederationHandler(BaseHandler): self._ephemeral_messages_enabled = hs.config.enable_ephemeral_messages - @defer.inlineCallbacks - def on_receive_pdu(self, origin, pdu, sent_to_us_directly=False): + async def on_receive_pdu(self, origin, pdu, sent_to_us_directly=False) -> None: """ Process a PDU received via a federation /send/ transaction, or via backfill of missing prev_events @@ -176,8 +175,6 @@ class FederationHandler(BaseHandler): pdu (FrozenEvent): received PDU sent_to_us_directly (bool): True if this event was pushed to us; False if we pulled it as the result of a missing prev_event. - - Returns (Deferred): completes with None """ room_id = pdu.room_id @@ -186,7 +183,7 @@ class FederationHandler(BaseHandler): logger.info("handling received PDU: %s", pdu) # We reprocess pdus when we have seen them only as outliers - existing = yield self.store.get_event( + existing = await self.store.get_event( event_id, allow_none=True, allow_rejected=True ) @@ -230,7 +227,7 @@ class FederationHandler(BaseHandler): # # Note that if we were never in the room then we would have already # dropped the event, since we wouldn't know the room version. - is_in_room = yield self.auth.check_host_in_room(room_id, self.server_name) + is_in_room = await self.auth.check_host_in_room(room_id, self.server_name) if not is_in_room: logger.info( "[%s %s] Ignoring PDU from %s as we're not in the room", @@ -246,12 +243,12 @@ class FederationHandler(BaseHandler): # Get missing pdus if necessary. if not pdu.internal_metadata.is_outlier(): # We only backfill backwards to the min depth. - min_depth = yield self.get_min_depth_for_context(pdu.room_id) + min_depth = await self.get_min_depth_for_context(pdu.room_id) logger.debug("[%s %s] min_depth: %d", room_id, event_id, min_depth) prevs = set(pdu.prev_event_ids()) - seen = yield self.store.have_seen_events(prevs) + seen = await self.store.have_seen_events(prevs) if min_depth and pdu.depth < min_depth: # This is so that we don't notify the user about this @@ -271,7 +268,7 @@ class FederationHandler(BaseHandler): len(missing_prevs), shortstr(missing_prevs), ) - with (yield self._room_pdu_linearizer.queue(pdu.room_id)): + with (await self._room_pdu_linearizer.queue(pdu.room_id)): logger.info( "[%s %s] Acquired room lock to fetch %d missing prev_events", room_id, @@ -280,7 +277,7 @@ class FederationHandler(BaseHandler): ) try: - yield self._get_missing_events_for_pdu( + await self._get_missing_events_for_pdu( origin, pdu, prevs, min_depth ) except Exception as e: @@ -291,7 +288,7 @@ class FederationHandler(BaseHandler): # Update the set of things we've seen after trying to # fetch the missing stuff - seen = yield self.store.have_seen_events(prevs) + seen = await self.store.have_seen_events(prevs) if not prevs - seen: logger.info( @@ -355,7 +352,7 @@ class FederationHandler(BaseHandler): event_map = {event_id: pdu} try: # Get the state of the events we know about - ours = yield self.state_store.get_state_groups_ids(room_id, seen) + ours = await self.state_store.get_state_groups_ids(room_id, seen) # state_maps is a list of mappings from (type, state_key) to event_id state_maps = list( @@ -372,7 +369,7 @@ class FederationHandler(BaseHandler): "Requesting state at missing prev_event %s", event_id, ) - room_version = yield self.store.get_room_version(room_id) + room_version = await self.store.get_room_version(room_id) with nested_logging_context(p): # note that if any of the missing prevs share missing state or @@ -381,11 +378,11 @@ class FederationHandler(BaseHandler): ( remote_state, got_auth_chain, - ) = yield self._get_state_for_room(origin, room_id, p) + ) = await self._get_state_for_room(origin, room_id, p) # we want the state *after* p; _get_state_for_room returns the # state *before* p. - remote_event = yield self.federation_client.get_pdu( + remote_event = await self.federation_client.get_pdu( [origin], p, room_version, outlier=True ) @@ -410,7 +407,7 @@ class FederationHandler(BaseHandler): for x in remote_state: event_map[x.event_id] = x - state_map = yield resolve_events_with_store( + state_map = await resolve_events_with_store( room_version, state_maps, event_map, @@ -422,7 +419,7 @@ class FederationHandler(BaseHandler): # First though we need to fetch all the events that are in # state_map, so we can build up the state below. - evs = yield self.store.get_events( + evs = await self.store.get_events( list(state_map.values()), get_prev_content=False, redact_behaviour=EventRedactBehaviour.AS_IS, @@ -446,12 +443,11 @@ class FederationHandler(BaseHandler): affected=event_id, ) - yield self._process_received_pdu( + await self._process_received_pdu( origin, pdu, state=state, auth_chain=auth_chain ) - @defer.inlineCallbacks - def _get_missing_events_for_pdu(self, origin, pdu, prevs, min_depth): + async def _get_missing_events_for_pdu(self, origin, pdu, prevs, min_depth): """ Args: origin (str): Origin of the pdu. Will be called to get the missing events @@ -463,12 +459,12 @@ class FederationHandler(BaseHandler): room_id = pdu.room_id event_id = pdu.event_id - seen = yield self.store.have_seen_events(prevs) + seen = await self.store.have_seen_events(prevs) if not prevs - seen: return - latest = yield self.store.get_latest_event_ids_in_room(room_id) + latest = await self.store.get_latest_event_ids_in_room(room_id) # We add the prev events that we have seen to the latest # list to ensure the remote server doesn't give them to us @@ -532,7 +528,7 @@ class FederationHandler(BaseHandler): # All that said: Let's try increasing the timout to 60s and see what happens. try: - missing_events = yield self.federation_client.get_missing_events( + missing_events = await self.federation_client.get_missing_events( origin, room_id, earliest_events_ids=list(latest), @@ -571,7 +567,7 @@ class FederationHandler(BaseHandler): ) with nested_logging_context(ev.event_id): try: - yield self.on_receive_pdu(origin, ev, sent_to_us_directly=False) + await self.on_receive_pdu(origin, ev, sent_to_us_directly=False) except FederationError as e: if e.code == 403: logger.warning( @@ -1328,8 +1324,7 @@ class FederationHandler(BaseHandler): return True - @defer.inlineCallbacks - def _handle_queued_pdus(self, room_queue): + async def _handle_queued_pdus(self, room_queue): """Process PDUs which got queued up while we were busy send_joining. Args: @@ -1345,7 +1340,7 @@ class FederationHandler(BaseHandler): p.room_id, ) with nested_logging_context(p.event_id): - yield self.on_receive_pdu(origin, p, sent_to_us_directly=True) + await self.on_receive_pdu(origin, p, sent_to_us_directly=True) except Exception as e: logger.warning( "Error handling queued PDU %s from %s: %s", p.event_id, origin, e diff --git a/tests/test_federation.py b/tests/test_federation.py index ad165d7295..68684460c6 100644 --- a/tests/test_federation.py +++ b/tests/test_federation.py @@ -1,6 +1,6 @@ from mock import Mock -from twisted.internet.defer import maybeDeferred, succeed +from twisted.internet.defer import ensureDeferred, maybeDeferred, succeed from synapse.events import FrozenEvent from synapse.logging.context import LoggingContext @@ -70,8 +70,10 @@ class MessageAcceptTests(unittest.TestCase): ) # Send the join, it should return None (which is not an error) - d = self.handler.on_receive_pdu( - "test.serv", join_event, sent_to_us_directly=True + d = ensureDeferred( + self.handler.on_receive_pdu( + "test.serv", join_event, sent_to_us_directly=True + ) ) self.reactor.advance(1) self.assertEqual(self.successResultOf(d), None) @@ -119,8 +121,10 @@ class MessageAcceptTests(unittest.TestCase): ) with LoggingContext(request="lying_event"): - d = self.handler.on_receive_pdu( - "test.serv", lying_event, sent_to_us_directly=True + d = ensureDeferred( + self.handler.on_receive_pdu( + "test.serv", lying_event, sent_to_us_directly=True + ) ) # Step the reactor, so the database fetches come back -- cgit 1.4.1 From cb2db179945f567410b565f29725dff28449f013 Mon Sep 17 00:00:00 2001 From: Hubert Chathi Date: Thu, 12 Dec 2019 12:03:28 -0500 Subject: look up cross-signing keys from the DB in bulk (#6486) --- changelog.d/6486.bugfix | 1 + synapse/handlers/e2e_keys.py | 35 +++- .../storage/data_stores/main/end_to_end_keys.py | 217 ++++++++++++++++++++- synapse/util/caches/descriptors.py | 2 +- tests/handlers/test_e2e_keys.py | 8 - 5 files changed, 242 insertions(+), 21 deletions(-) create mode 100644 changelog.d/6486.bugfix (limited to 'tests') diff --git a/changelog.d/6486.bugfix b/changelog.d/6486.bugfix new file mode 100644 index 0000000000..b98c5a9ae5 --- /dev/null +++ b/changelog.d/6486.bugfix @@ -0,0 +1 @@ +Improve performance of looking up cross-signing keys. diff --git a/synapse/handlers/e2e_keys.py b/synapse/handlers/e2e_keys.py index 57a10daefd..2d889364d4 100644 --- a/synapse/handlers/e2e_keys.py +++ b/synapse/handlers/e2e_keys.py @@ -264,6 +264,7 @@ class E2eKeysHandler(object): return ret + @defer.inlineCallbacks def get_cross_signing_keys_from_cache(self, query, from_user_id): """Get cross-signing keys for users from the database @@ -283,14 +284,32 @@ class E2eKeysHandler(object): self_signing_keys = {} user_signing_keys = {} - # Currently a stub, implementation coming in https://github.com/matrix-org/synapse/pull/6486 - return defer.succeed( - { - "master_keys": master_keys, - "self_signing_keys": self_signing_keys, - "user_signing_keys": user_signing_keys, - } - ) + user_ids = list(query) + + keys = yield self.store.get_e2e_cross_signing_keys_bulk(user_ids, from_user_id) + + for user_id, user_info in keys.items(): + if user_info is None: + continue + if "master" in user_info: + master_keys[user_id] = user_info["master"] + if "self_signing" in user_info: + self_signing_keys[user_id] = user_info["self_signing"] + + if ( + from_user_id in keys + and keys[from_user_id] is not None + and "user_signing" in keys[from_user_id] + ): + # users can see other users' master and self-signing keys, but can + # only see their own user-signing keys + user_signing_keys[from_user_id] = keys[from_user_id]["user_signing"] + + return { + "master_keys": master_keys, + "self_signing_keys": self_signing_keys, + "user_signing_keys": user_signing_keys, + } @trace @defer.inlineCallbacks diff --git a/synapse/storage/data_stores/main/end_to_end_keys.py b/synapse/storage/data_stores/main/end_to_end_keys.py index 38cd0ca9b8..e551606f9d 100644 --- a/synapse/storage/data_stores/main/end_to_end_keys.py +++ b/synapse/storage/data_stores/main/end_to_end_keys.py @@ -14,15 +14,18 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. +from typing import Dict, List + from six import iteritems from canonicaljson import encode_canonical_json, json +from twisted.enterprise.adbapi import Connection from twisted.internet import defer from synapse.logging.opentracing import log_kv, set_tag, trace from synapse.storage._base import SQLBaseStore, db_to_json -from synapse.util.caches.descriptors import cached +from synapse.util.caches.descriptors import cached, cachedList class EndToEndKeyWorkerStore(SQLBaseStore): @@ -271,7 +274,7 @@ class EndToEndKeyWorkerStore(SQLBaseStore): Args: txn (twisted.enterprise.adbapi.Connection): db connection user_id (str): the user whose key is being requested - key_type (str): the type of key that is being set: either 'master' + key_type (str): the type of key that is being requested: either 'master' for a master key, 'self_signing' for a self-signing key, or 'user_signing' for a user-signing key from_user_id (str): if specified, signatures made by this user on @@ -316,8 +319,10 @@ class EndToEndKeyWorkerStore(SQLBaseStore): """Returns a user's cross-signing key. Args: - user_id (str): the user whose self-signing key is being requested - key_type (str): the type of cross-signing key to get + user_id (str): the user whose key is being requested + key_type (str): the type of key that is being requested: either 'master' + for a master key, 'self_signing' for a self-signing key, or + 'user_signing' for a user-signing key from_user_id (str): if specified, signatures made by this user on the self-signing key will be included in the result @@ -332,6 +337,206 @@ class EndToEndKeyWorkerStore(SQLBaseStore): from_user_id, ) + @cached(num_args=1) + def _get_bare_e2e_cross_signing_keys(self, user_id): + """Dummy function. Only used to make a cache for + _get_bare_e2e_cross_signing_keys_bulk. + """ + raise NotImplementedError() + + @cachedList( + cached_method_name="_get_bare_e2e_cross_signing_keys", + list_name="user_ids", + num_args=1, + ) + def _get_bare_e2e_cross_signing_keys_bulk( + self, user_ids: List[str] + ) -> Dict[str, Dict[str, dict]]: + """Returns the cross-signing keys for a set of users. The output of this + function should be passed to _get_e2e_cross_signing_signatures_txn if + the signatures for the calling user need to be fetched. + + Args: + user_ids (list[str]): the users whose keys are being requested + + Returns: + dict[str, dict[str, dict]]: mapping from user ID to key type to key + data. If a user's cross-signing keys were not found, either + their user ID will not be in the dict, or their user ID will map + to None. + + """ + return self.db.runInteraction( + "get_bare_e2e_cross_signing_keys_bulk", + self._get_bare_e2e_cross_signing_keys_bulk_txn, + user_ids, + ) + + def _get_bare_e2e_cross_signing_keys_bulk_txn( + self, txn: Connection, user_ids: List[str], + ) -> Dict[str, Dict[str, dict]]: + """Returns the cross-signing keys for a set of users. The output of this + function should be passed to _get_e2e_cross_signing_signatures_txn if + the signatures for the calling user need to be fetched. + + Args: + txn (twisted.enterprise.adbapi.Connection): db connection + user_ids (list[str]): the users whose keys are being requested + + Returns: + dict[str, dict[str, dict]]: mapping from user ID to key type to key + data. If a user's cross-signing keys were not found, their user + ID will not be in the dict. + + """ + result = {} + + batch_size = 100 + chunks = [ + user_ids[i : i + batch_size] for i in range(0, len(user_ids), batch_size) + ] + for user_chunk in chunks: + sql = """ + SELECT k.user_id, k.keytype, k.keydata, k.stream_id + FROM e2e_cross_signing_keys k + INNER JOIN (SELECT user_id, keytype, MAX(stream_id) AS stream_id + FROM e2e_cross_signing_keys + GROUP BY user_id, keytype) s + USING (user_id, stream_id, keytype) + WHERE k.user_id IN (%s) + """ % ( + ",".join("?" for u in user_chunk), + ) + query_params = [] + query_params.extend(user_chunk) + + txn.execute(sql, query_params) + rows = self.db.cursor_to_dict(txn) + + for row in rows: + user_id = row["user_id"] + key_type = row["keytype"] + key = json.loads(row["keydata"]) + user_info = result.setdefault(user_id, {}) + user_info[key_type] = key + + return result + + def _get_e2e_cross_signing_signatures_txn( + self, txn: Connection, keys: Dict[str, Dict[str, dict]], from_user_id: str, + ) -> Dict[str, Dict[str, dict]]: + """Returns the cross-signing signatures made by a user on a set of keys. + + Args: + txn (twisted.enterprise.adbapi.Connection): db connection + keys (dict[str, dict[str, dict]]): a map of user ID to key type to + key data. This dict will be modified to add signatures. + from_user_id (str): fetch the signatures made by this user + + Returns: + dict[str, dict[str, dict]]: mapping from user ID to key type to key + data. The return value will be the same as the keys argument, + with the modifications included. + """ + + # find out what cross-signing keys (a.k.a. devices) we need to get + # signatures for. This is a map of (user_id, device_id) to key type + # (device_id is the key's public part). + devices = {} + + for user_id, user_info in keys.items(): + if user_info is None: + continue + for key_type, key in user_info.items(): + device_id = None + for k in key["keys"].values(): + device_id = k + devices[(user_id, device_id)] = key_type + + device_list = list(devices) + + # split into batches + batch_size = 100 + chunks = [ + device_list[i : i + batch_size] + for i in range(0, len(device_list), batch_size) + ] + for user_chunk in chunks: + sql = """ + SELECT target_user_id, target_device_id, key_id, signature + FROM e2e_cross_signing_signatures + WHERE user_id = ? + AND (%s) + """ % ( + " OR ".join( + "(target_user_id = ? AND target_device_id = ?)" for d in devices + ) + ) + query_params = [from_user_id] + for item in devices: + # item is a (user_id, device_id) tuple + query_params.extend(item) + + txn.execute(sql, query_params) + rows = self.db.cursor_to_dict(txn) + + # and add the signatures to the appropriate keys + for row in rows: + key_id = row["key_id"] + target_user_id = row["target_user_id"] + target_device_id = row["target_device_id"] + key_type = devices[(target_user_id, target_device_id)] + # We need to copy everything, because the result may have come + # from the cache. dict.copy only does a shallow copy, so we + # need to recursively copy the dicts that will be modified. + user_info = keys[target_user_id] = keys[target_user_id].copy() + target_user_key = user_info[key_type] = user_info[key_type].copy() + if "signatures" in target_user_key: + signatures = target_user_key["signatures"] = target_user_key[ + "signatures" + ].copy() + if from_user_id in signatures: + user_sigs = signatures[from_user_id] = signatures[from_user_id] + user_sigs[key_id] = row["signature"] + else: + signatures[from_user_id] = {key_id: row["signature"]} + else: + target_user_key["signatures"] = { + from_user_id: {key_id: row["signature"]} + } + + return keys + + @defer.inlineCallbacks + def get_e2e_cross_signing_keys_bulk( + self, user_ids: List[str], from_user_id: str = None + ) -> defer.Deferred: + """Returns the cross-signing keys for a set of users. + + Args: + user_ids (list[str]): the users whose keys are being requested + from_user_id (str): if specified, signatures made by this user on + the self-signing keys will be included in the result + + Returns: + Deferred[dict[str, dict[str, dict]]]: map of user ID to key type to + key data. If a user's cross-signing keys were not found, either + their user ID will not be in the dict, or their user ID will map + to None. + """ + + result = yield self._get_bare_e2e_cross_signing_keys_bulk(user_ids) + + if from_user_id: + result = yield self.db.runInteraction( + "get_e2e_cross_signing_signatures", + self._get_e2e_cross_signing_signatures_txn, + result, + from_user_id, + ) + + return result + def get_all_user_signature_changes_for_remotes(self, from_key, to_key): """Return a list of changes from the user signature stream to notify remotes. Note that the user signature stream represents when a user signs their @@ -520,6 +725,10 @@ class EndToEndKeyStore(EndToEndKeyWorkerStore, SQLBaseStore): }, ) + self._invalidate_cache_and_stream( + txn, self._get_bare_e2e_cross_signing_keys, (user_id,) + ) + def set_e2e_cross_signing_key(self, user_id, key_type, key): """Set a user's cross-signing key. diff --git a/synapse/util/caches/descriptors.py b/synapse/util/caches/descriptors.py index 84f5ae22c3..2e8f6543e5 100644 --- a/synapse/util/caches/descriptors.py +++ b/synapse/util/caches/descriptors.py @@ -271,7 +271,7 @@ class _CacheDescriptorBase(object): else: self.function_to_call = orig - arg_spec = inspect.getargspec(orig) + arg_spec = inspect.getfullargspec(orig) all_args = arg_spec.args if "cache_context" in all_args: diff --git a/tests/handlers/test_e2e_keys.py b/tests/handlers/test_e2e_keys.py index fdfa2cbbc4..854eb6c024 100644 --- a/tests/handlers/test_e2e_keys.py +++ b/tests/handlers/test_e2e_keys.py @@ -183,10 +183,6 @@ class E2eKeysHandlerTestCase(unittest.TestCase): ) self.assertDictEqual(devices["master_keys"], {local_user: keys2["master_key"]}) - test_replace_master_key.skip = ( - "Disabled waiting on #https://github.com/matrix-org/synapse/pull/6486" - ) - @defer.inlineCallbacks def test_reupload_signatures(self): """re-uploading a signature should not fail""" @@ -507,7 +503,3 @@ class E2eKeysHandlerTestCase(unittest.TestCase): ], other_master_key["signatures"][local_user]["ed25519:" + usersigning_pubkey], ) - - test_upload_signatures.skip = ( - "Disabled waiting on #https://github.com/matrix-org/synapse/pull/6486" - ) -- cgit 1.4.1 From 1da15f05f5c9c1e47c9fd1323caff869c2e55aa3 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Fri, 13 Dec 2019 12:55:32 +0000 Subject: sanity-checking for events used in state res (#6531) When we perform state resolution, check that all of the events involved are in the right room. --- changelog.d/6531.misc | 1 + synapse/handlers/federation.py | 1 + synapse/state/__init__.py | 32 ++++++++----- synapse/state/v1.py | 34 +++++++++++--- synapse/state/v2.py | 100 ++++++++++++++++++++++++++++++----------- tests/state/test_v2.py | 3 ++ 6 files changed, 128 insertions(+), 43 deletions(-) create mode 100644 changelog.d/6531.misc (limited to 'tests') diff --git a/changelog.d/6531.misc b/changelog.d/6531.misc new file mode 100644 index 0000000000..598efb79fc --- /dev/null +++ b/changelog.d/6531.misc @@ -0,0 +1 @@ +Improve sanity-checking when receiving events over federation. diff --git a/synapse/handlers/federation.py b/synapse/handlers/federation.py index 2ea69c5468..1d39a9a4f5 100644 --- a/synapse/handlers/federation.py +++ b/synapse/handlers/federation.py @@ -396,6 +396,7 @@ class FederationHandler(BaseHandler): event_map[x.event_id] = x state_map = await resolve_events_with_store( + room_id, room_version, state_maps, event_map, diff --git a/synapse/state/__init__.py b/synapse/state/__init__.py index 3e6d62eef1..5accc071ab 100644 --- a/synapse/state/__init__.py +++ b/synapse/state/__init__.py @@ -16,7 +16,7 @@ import logging from collections import namedtuple -from typing import Iterable, Optional +from typing import Dict, Iterable, List, Optional, Tuple from six import iteritems, itervalues @@ -417,6 +417,7 @@ class StateHandler(object): with Measure(self.clock, "state._resolve_events"): new_state = yield resolve_events_with_store( + event.room_id, room_version, state_set_ids, event_map=state_map, @@ -462,7 +463,7 @@ class StateResolutionHandler(object): not be called for a single state group Args: - room_id (str): room we are resolving for (used for logging) + room_id (str): room we are resolving for (used for logging and sanity checks) room_version (str): version of the room state_groups_ids (dict[int, dict[(str, str), str]]): map from state group id to the state in that state group @@ -518,6 +519,7 @@ class StateResolutionHandler(object): logger.info("Resolving conflicted state for %r", room_id) with Measure(self.clock, "state._resolve_events"): new_state = yield resolve_events_with_store( + room_id, room_version, list(itervalues(state_groups_ids)), event_map=event_map, @@ -589,36 +591,44 @@ def _make_state_cache_entry(new_state, state_groups_ids): ) -def resolve_events_with_store(room_version, state_sets, event_map, state_res_store): +def resolve_events_with_store( + room_id: str, + room_version: str, + state_sets: List[Dict[Tuple[str, str], str]], + event_map: Optional[Dict[str, EventBase]], + state_res_store: "StateResolutionStore", +): """ Args: - room_version(str): Version of the room + room_id: the room we are working in + + room_version: Version of the room - state_sets(list): List of dicts of (type, state_key) -> event_id, + state_sets: List of dicts of (type, state_key) -> event_id, which are the different state groups to resolve. - event_map(dict[str,FrozenEvent]|None): + event_map: a dict from event_id to event, for any events that we happen to have in flight (eg, those currently being persisted). This will be used as a starting point fof finding the state we need; any missing events will be requested via state_map_factory. - If None, all events will be fetched via state_map_factory. + If None, all events will be fetched via state_res_store. - state_res_store (StateResolutionStore) + state_res_store: a place to fetch events from - Returns + Returns: Deferred[dict[(str, str), str]]: a map from (type, state_key) to event_id. """ v = KNOWN_ROOM_VERSIONS[room_version] if v.state_res == StateResolutionVersions.V1: return v1.resolve_events_with_store( - state_sets, event_map, state_res_store.get_events + room_id, state_sets, event_map, state_res_store.get_events ) else: return v2.resolve_events_with_store( - room_version, state_sets, event_map, state_res_store + room_id, room_version, state_sets, event_map, state_res_store ) diff --git a/synapse/state/v1.py b/synapse/state/v1.py index a2f92d9ff9..b2f9865f39 100644 --- a/synapse/state/v1.py +++ b/synapse/state/v1.py @@ -15,6 +15,7 @@ import hashlib import logging +from typing import Callable, Dict, List, Optional, Tuple from six import iteritems, iterkeys, itervalues @@ -24,6 +25,7 @@ from synapse import event_auth from synapse.api.constants import EventTypes from synapse.api.errors import AuthError from synapse.api.room_versions import RoomVersions +from synapse.events import EventBase logger = logging.getLogger(__name__) @@ -32,13 +34,20 @@ POWER_KEY = (EventTypes.PowerLevels, "") @defer.inlineCallbacks -def resolve_events_with_store(state_sets, event_map, state_map_factory): +def resolve_events_with_store( + room_id: str, + state_sets: List[Dict[Tuple[str, str], str]], + event_map: Optional[Dict[str, EventBase]], + state_map_factory: Callable, +): """ Args: - state_sets(list): List of dicts of (type, state_key) -> event_id, + room_id: the room we are working in + + state_sets: List of dicts of (type, state_key) -> event_id, which are the different state groups to resolve. - event_map(dict[str,FrozenEvent]|None): + event_map: a dict from event_id to event, for any events that we happen to have in flight (eg, those currently being persisted). This will be used as a starting point fof finding the state we need; any missing @@ -46,11 +55,11 @@ def resolve_events_with_store(state_sets, event_map, state_map_factory): If None, all events will be fetched via state_map_factory. - state_map_factory(func): will be called + state_map_factory: will be called with a list of event_ids that are needed, and should return with a Deferred of dict of event_id to event. - Returns + Returns: Deferred[dict[(str, str), str]]: a map from (type, state_key) to event_id. """ @@ -76,6 +85,14 @@ def resolve_events_with_store(state_sets, event_map, state_map_factory): if event_map is not None: state_map.update(event_map) + # everything in the state map should be in the right room + for event in state_map.values(): + if event.room_id != room_id: + raise Exception( + "Attempting to state-resolve for room %s with event %s which is in %s" + % (room_id, event.event_id, event.room_id,) + ) + # get the ids of the auth events which allow us to authenticate the # conflicted state, picking only from the unconflicting state. # @@ -95,6 +112,13 @@ def resolve_events_with_store(state_sets, event_map, state_map_factory): ) state_map_new = yield state_map_factory(new_needed_events) + for event in state_map_new.values(): + if event.room_id != room_id: + raise Exception( + "Attempting to state-resolve for room %s with event %s which is in %s" + % (room_id, event.event_id, event.room_id,) + ) + state_map.update(state_map_new) return _resolve_with_state( diff --git a/synapse/state/v2.py b/synapse/state/v2.py index b327c86f40..cb77ed5b78 100644 --- a/synapse/state/v2.py +++ b/synapse/state/v2.py @@ -16,29 +16,40 @@ import heapq import itertools import logging +from typing import Dict, List, Optional, Tuple from six import iteritems, itervalues from twisted.internet import defer +import synapse.state from synapse import event_auth from synapse.api.constants import EventTypes from synapse.api.errors import AuthError +from synapse.events import EventBase logger = logging.getLogger(__name__) @defer.inlineCallbacks -def resolve_events_with_store(room_version, state_sets, event_map, state_res_store): +def resolve_events_with_store( + room_id: str, + room_version: str, + state_sets: List[Dict[Tuple[str, str], str]], + event_map: Optional[Dict[str, EventBase]], + state_res_store: "synapse.state.StateResolutionStore", +): """Resolves the state using the v2 state resolution algorithm Args: - room_version (str): The room version + room_id: the room we are working in + + room_version: The room version - state_sets(list): List of dicts of (type, state_key) -> event_id, + state_sets: List of dicts of (type, state_key) -> event_id, which are the different state groups to resolve. - event_map(dict[str,FrozenEvent]|None): + event_map: a dict from event_id to event, for any events that we happen to have in flight (eg, those currently being persisted). This will be used as a starting point fof finding the state we need; any missing @@ -46,9 +57,9 @@ def resolve_events_with_store(room_version, state_sets, event_map, state_res_sto If None, all events will be fetched via state_res_store. - state_res_store (StateResolutionStore) + state_res_store: - Returns + Returns: Deferred[dict[(str, str), str]]: a map from (type, state_key) to event_id. """ @@ -84,6 +95,14 @@ def resolve_events_with_store(room_version, state_sets, event_map, state_res_sto ) event_map.update(events) + # everything in the event map should be in the right room + for event in event_map.values(): + if event.room_id != room_id: + raise Exception( + "Attempting to state-resolve for room %s with event %s which is in %s" + % (room_id, event.event_id, event.room_id,) + ) + full_conflicted_set = set(eid for eid in full_conflicted_set if eid in event_map) logger.debug("%d full_conflicted_set entries", len(full_conflicted_set)) @@ -94,13 +113,14 @@ def resolve_events_with_store(room_version, state_sets, event_map, state_res_sto ) sorted_power_events = yield _reverse_topological_power_sort( - power_events, event_map, state_res_store, full_conflicted_set + room_id, power_events, event_map, state_res_store, full_conflicted_set ) logger.debug("sorted %d power events", len(sorted_power_events)) # Now sequentially auth each one resolved_state = yield _iterative_auth_checks( + room_id, room_version, sorted_power_events, unconflicted_state, @@ -121,13 +141,18 @@ def resolve_events_with_store(room_version, state_sets, event_map, state_res_sto pl = resolved_state.get((EventTypes.PowerLevels, ""), None) leftover_events = yield _mainline_sort( - leftover_events, pl, event_map, state_res_store + room_id, leftover_events, pl, event_map, state_res_store ) logger.debug("resolving remaining events") resolved_state = yield _iterative_auth_checks( - room_version, leftover_events, resolved_state, event_map, state_res_store + room_id, + room_version, + leftover_events, + resolved_state, + event_map, + state_res_store, ) logger.debug("resolved") @@ -141,11 +166,12 @@ def resolve_events_with_store(room_version, state_sets, event_map, state_res_sto @defer.inlineCallbacks -def _get_power_level_for_sender(event_id, event_map, state_res_store): +def _get_power_level_for_sender(room_id, event_id, event_map, state_res_store): """Return the power level of the sender of the given event according to their auth events. Args: + room_id (str) event_id (str) event_map (dict[str,FrozenEvent]) state_res_store (StateResolutionStore) @@ -153,11 +179,11 @@ def _get_power_level_for_sender(event_id, event_map, state_res_store): Returns: Deferred[int] """ - event = yield _get_event(event_id, event_map, state_res_store) + event = yield _get_event(room_id, event_id, event_map, state_res_store) pl = None for aid in event.auth_event_ids(): - aev = yield _get_event(aid, event_map, state_res_store) + aev = yield _get_event(room_id, aid, event_map, state_res_store) if (aev.type, aev.state_key) == (EventTypes.PowerLevels, ""): pl = aev break @@ -165,7 +191,7 @@ def _get_power_level_for_sender(event_id, event_map, state_res_store): if pl is None: # Couldn't find power level. Check if they're the creator of the room for aid in event.auth_event_ids(): - aev = yield _get_event(aid, event_map, state_res_store) + aev = yield _get_event(room_id, aid, event_map, state_res_store) if (aev.type, aev.state_key) == (EventTypes.Create, ""): if aev.content.get("creator") == event.sender: return 100 @@ -279,7 +305,7 @@ def _is_power_event(event): @defer.inlineCallbacks def _add_event_and_auth_chain_to_graph( - graph, event_id, event_map, state_res_store, auth_diff + graph, room_id, event_id, event_map, state_res_store, auth_diff ): """Helper function for _reverse_topological_power_sort that add the event and its auth chain (that is in the auth diff) to the graph @@ -287,6 +313,7 @@ def _add_event_and_auth_chain_to_graph( Args: graph (dict[str, set[str]]): A map from event ID to the events auth event IDs + room_id (str): the room we are working in event_id (str): Event to add to the graph event_map (dict[str,FrozenEvent]) state_res_store (StateResolutionStore) @@ -298,7 +325,7 @@ def _add_event_and_auth_chain_to_graph( eid = state.pop() graph.setdefault(eid, set()) - event = yield _get_event(eid, event_map, state_res_store) + event = yield _get_event(room_id, eid, event_map, state_res_store) for aid in event.auth_event_ids(): if aid in auth_diff: if aid not in graph: @@ -308,11 +335,14 @@ def _add_event_and_auth_chain_to_graph( @defer.inlineCallbacks -def _reverse_topological_power_sort(event_ids, event_map, state_res_store, auth_diff): +def _reverse_topological_power_sort( + room_id, event_ids, event_map, state_res_store, auth_diff +): """Returns a list of the event_ids sorted by reverse topological ordering, and then by power level and origin_server_ts Args: + room_id (str): the room we are working in event_ids (list[str]): The events to sort event_map (dict[str,FrozenEvent]) state_res_store (StateResolutionStore) @@ -325,12 +355,14 @@ def _reverse_topological_power_sort(event_ids, event_map, state_res_store, auth_ graph = {} for event_id in event_ids: yield _add_event_and_auth_chain_to_graph( - graph, event_id, event_map, state_res_store, auth_diff + graph, room_id, event_id, event_map, state_res_store, auth_diff ) event_to_pl = {} for event_id in graph: - pl = yield _get_power_level_for_sender(event_id, event_map, state_res_store) + pl = yield _get_power_level_for_sender( + room_id, event_id, event_map, state_res_store + ) event_to_pl[event_id] = pl def _get_power_order(event_id): @@ -348,12 +380,13 @@ def _reverse_topological_power_sort(event_ids, event_map, state_res_store, auth_ @defer.inlineCallbacks def _iterative_auth_checks( - room_version, event_ids, base_state, event_map, state_res_store + room_id, room_version, event_ids, base_state, event_map, state_res_store ): """Sequentially apply auth checks to each event in given list, updating the state as it goes along. Args: + room_id (str) room_version (str) event_ids (list[str]): Ordered list of events to apply auth checks to base_state (dict[tuple[str, str], str]): The set of state to start with @@ -370,7 +403,7 @@ def _iterative_auth_checks( auth_events = {} for aid in event.auth_event_ids(): - ev = yield _get_event(aid, event_map, state_res_store) + ev = yield _get_event(room_id, aid, event_map, state_res_store) if ev.rejected_reason is None: auth_events[(ev.type, ev.state_key)] = ev @@ -378,7 +411,7 @@ def _iterative_auth_checks( for key in event_auth.auth_types_for_event(event): if key in resolved_state: ev_id = resolved_state[key] - ev = yield _get_event(ev_id, event_map, state_res_store) + ev = yield _get_event(room_id, ev_id, event_map, state_res_store) if ev.rejected_reason is None: auth_events[key] = event_map[ev_id] @@ -400,11 +433,14 @@ def _iterative_auth_checks( @defer.inlineCallbacks -def _mainline_sort(event_ids, resolved_power_event_id, event_map, state_res_store): +def _mainline_sort( + room_id, event_ids, resolved_power_event_id, event_map, state_res_store +): """Returns a sorted list of event_ids sorted by mainline ordering based on the given event resolved_power_event_id Args: + room_id (str): room we're working in event_ids (list[str]): Events to sort resolved_power_event_id (str): The final resolved power level event ID event_map (dict[str,FrozenEvent]) @@ -417,11 +453,11 @@ def _mainline_sort(event_ids, resolved_power_event_id, event_map, state_res_stor pl = resolved_power_event_id while pl: mainline.append(pl) - pl_ev = yield _get_event(pl, event_map, state_res_store) + pl_ev = yield _get_event(room_id, pl, event_map, state_res_store) auth_events = pl_ev.auth_event_ids() pl = None for aid in auth_events: - ev = yield _get_event(aid, event_map, state_res_store) + ev = yield _get_event(room_id, aid, event_map, state_res_store) if (ev.type, ev.state_key) == (EventTypes.PowerLevels, ""): pl = aid break @@ -457,6 +493,8 @@ def _get_mainline_depth_for_event(event, mainline_map, event_map, state_res_stor Deferred[int] """ + room_id = event.room_id + # We do an iterative search, replacing `event with the power level in its # auth events (if any) while event: @@ -468,7 +506,7 @@ def _get_mainline_depth_for_event(event, mainline_map, event_map, state_res_stor event = None for aid in auth_events: - aev = yield _get_event(aid, event_map, state_res_store) + aev = yield _get_event(room_id, aid, event_map, state_res_store) if (aev.type, aev.state_key) == (EventTypes.PowerLevels, ""): event = aev break @@ -478,11 +516,12 @@ def _get_mainline_depth_for_event(event, mainline_map, event_map, state_res_stor @defer.inlineCallbacks -def _get_event(event_id, event_map, state_res_store): +def _get_event(room_id, event_id, event_map, state_res_store): """Helper function to look up event in event_map, falling back to looking it up in the store Args: + room_id (str) event_id (str) event_map (dict[str,FrozenEvent]) state_res_store (StateResolutionStore) @@ -493,7 +532,14 @@ def _get_event(event_id, event_map, state_res_store): if event_id not in event_map: events = yield state_res_store.get_events([event_id], allow_rejected=True) event_map.update(events) - return event_map[event_id] + event = event_map[event_id] + assert event is not None + if event.room_id != room_id: + raise Exception( + "In state res for room %s, event %s is in %s" + % (room_id, event_id, event.room_id) + ) + return event def lexicographical_topological_sort(graph, key): diff --git a/tests/state/test_v2.py b/tests/state/test_v2.py index 8d3845c870..0f341d3ac3 100644 --- a/tests/state/test_v2.py +++ b/tests/state/test_v2.py @@ -58,6 +58,7 @@ class FakeEvent(object): self.type = type self.state_key = state_key self.content = content + self.room_id = ROOM_ID def to_event(self, auth_events, prev_events): """Given the auth_events and prev_events, convert to a Frozen Event @@ -418,6 +419,7 @@ class StateTestCase(unittest.TestCase): state_before = dict(state_at_event[prev_events[0]]) else: state_d = resolve_events_with_store( + ROOM_ID, RoomVersions.V2.identifier, [state_at_event[n] for n in prev_events], event_map=event_map, @@ -565,6 +567,7 @@ class SimpleParamStateTestCase(unittest.TestCase): # Test that we correctly handle passing `None` as the event_map state_d = resolve_events_with_store( + ROOM_ID, RoomVersions.V2.identifier, [self.state_at_bob, self.state_at_charlie], event_map=None, -- cgit 1.4.1 From 83895316d4e18d4a52c43524942d98f864bac6f9 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Fri, 13 Dec 2019 12:55:32 +0000 Subject: sanity-checking for events used in state res (#6531) When we perform state resolution, check that all of the events involved are in the right room. --- changelog.d/6531.misc | 1 + synapse/handlers/federation.py | 1 + synapse/state/__init__.py | 32 ++++++++----- synapse/state/v1.py | 34 +++++++++++--- synapse/state/v2.py | 100 ++++++++++++++++++++++++++++++----------- tests/state/test_v2.py | 3 ++ 6 files changed, 128 insertions(+), 43 deletions(-) create mode 100644 changelog.d/6531.misc (limited to 'tests') diff --git a/changelog.d/6531.misc b/changelog.d/6531.misc new file mode 100644 index 0000000000..598efb79fc --- /dev/null +++ b/changelog.d/6531.misc @@ -0,0 +1 @@ +Improve sanity-checking when receiving events over federation. diff --git a/synapse/handlers/federation.py b/synapse/handlers/federation.py index ebeffbb768..fd3f5ced55 100644 --- a/synapse/handlers/federation.py +++ b/synapse/handlers/federation.py @@ -397,6 +397,7 @@ class FederationHandler(BaseHandler): event_map[x.event_id] = x state_map = yield resolve_events_with_store( + room_id, room_version, state_maps, event_map, diff --git a/synapse/state/__init__.py b/synapse/state/__init__.py index 139beef8ed..0e75e94c6f 100644 --- a/synapse/state/__init__.py +++ b/synapse/state/__init__.py @@ -16,7 +16,7 @@ import logging from collections import namedtuple -from typing import Iterable, Optional +from typing import Dict, Iterable, List, Optional, Tuple from six import iteritems, itervalues @@ -416,6 +416,7 @@ class StateHandler(object): with Measure(self.clock, "state._resolve_events"): new_state = yield resolve_events_with_store( + event.room_id, room_version, state_set_ids, event_map=state_map, @@ -461,7 +462,7 @@ class StateResolutionHandler(object): not be called for a single state group Args: - room_id (str): room we are resolving for (used for logging) + room_id (str): room we are resolving for (used for logging and sanity checks) room_version (str): version of the room state_groups_ids (dict[int, dict[(str, str), str]]): map from state group id to the state in that state group @@ -517,6 +518,7 @@ class StateResolutionHandler(object): logger.info("Resolving conflicted state for %r", room_id) with Measure(self.clock, "state._resolve_events"): new_state = yield resolve_events_with_store( + room_id, room_version, list(itervalues(state_groups_ids)), event_map=event_map, @@ -588,36 +590,44 @@ def _make_state_cache_entry(new_state, state_groups_ids): ) -def resolve_events_with_store(room_version, state_sets, event_map, state_res_store): +def resolve_events_with_store( + room_id: str, + room_version: str, + state_sets: List[Dict[Tuple[str, str], str]], + event_map: Optional[Dict[str, EventBase]], + state_res_store: "StateResolutionStore", +): """ Args: - room_version(str): Version of the room + room_id: the room we are working in + + room_version: Version of the room - state_sets(list): List of dicts of (type, state_key) -> event_id, + state_sets: List of dicts of (type, state_key) -> event_id, which are the different state groups to resolve. - event_map(dict[str,FrozenEvent]|None): + event_map: a dict from event_id to event, for any events that we happen to have in flight (eg, those currently being persisted). This will be used as a starting point fof finding the state we need; any missing events will be requested via state_map_factory. - If None, all events will be fetched via state_map_factory. + If None, all events will be fetched via state_res_store. - state_res_store (StateResolutionStore) + state_res_store: a place to fetch events from - Returns + Returns: Deferred[dict[(str, str), str]]: a map from (type, state_key) to event_id. """ v = KNOWN_ROOM_VERSIONS[room_version] if v.state_res == StateResolutionVersions.V1: return v1.resolve_events_with_store( - state_sets, event_map, state_res_store.get_events + room_id, state_sets, event_map, state_res_store.get_events ) else: return v2.resolve_events_with_store( - room_version, state_sets, event_map, state_res_store + room_id, room_version, state_sets, event_map, state_res_store ) diff --git a/synapse/state/v1.py b/synapse/state/v1.py index a2f92d9ff9..b2f9865f39 100644 --- a/synapse/state/v1.py +++ b/synapse/state/v1.py @@ -15,6 +15,7 @@ import hashlib import logging +from typing import Callable, Dict, List, Optional, Tuple from six import iteritems, iterkeys, itervalues @@ -24,6 +25,7 @@ from synapse import event_auth from synapse.api.constants import EventTypes from synapse.api.errors import AuthError from synapse.api.room_versions import RoomVersions +from synapse.events import EventBase logger = logging.getLogger(__name__) @@ -32,13 +34,20 @@ POWER_KEY = (EventTypes.PowerLevels, "") @defer.inlineCallbacks -def resolve_events_with_store(state_sets, event_map, state_map_factory): +def resolve_events_with_store( + room_id: str, + state_sets: List[Dict[Tuple[str, str], str]], + event_map: Optional[Dict[str, EventBase]], + state_map_factory: Callable, +): """ Args: - state_sets(list): List of dicts of (type, state_key) -> event_id, + room_id: the room we are working in + + state_sets: List of dicts of (type, state_key) -> event_id, which are the different state groups to resolve. - event_map(dict[str,FrozenEvent]|None): + event_map: a dict from event_id to event, for any events that we happen to have in flight (eg, those currently being persisted). This will be used as a starting point fof finding the state we need; any missing @@ -46,11 +55,11 @@ def resolve_events_with_store(state_sets, event_map, state_map_factory): If None, all events will be fetched via state_map_factory. - state_map_factory(func): will be called + state_map_factory: will be called with a list of event_ids that are needed, and should return with a Deferred of dict of event_id to event. - Returns + Returns: Deferred[dict[(str, str), str]]: a map from (type, state_key) to event_id. """ @@ -76,6 +85,14 @@ def resolve_events_with_store(state_sets, event_map, state_map_factory): if event_map is not None: state_map.update(event_map) + # everything in the state map should be in the right room + for event in state_map.values(): + if event.room_id != room_id: + raise Exception( + "Attempting to state-resolve for room %s with event %s which is in %s" + % (room_id, event.event_id, event.room_id,) + ) + # get the ids of the auth events which allow us to authenticate the # conflicted state, picking only from the unconflicting state. # @@ -95,6 +112,13 @@ def resolve_events_with_store(state_sets, event_map, state_map_factory): ) state_map_new = yield state_map_factory(new_needed_events) + for event in state_map_new.values(): + if event.room_id != room_id: + raise Exception( + "Attempting to state-resolve for room %s with event %s which is in %s" + % (room_id, event.event_id, event.room_id,) + ) + state_map.update(state_map_new) return _resolve_with_state( diff --git a/synapse/state/v2.py b/synapse/state/v2.py index b327c86f40..cb77ed5b78 100644 --- a/synapse/state/v2.py +++ b/synapse/state/v2.py @@ -16,29 +16,40 @@ import heapq import itertools import logging +from typing import Dict, List, Optional, Tuple from six import iteritems, itervalues from twisted.internet import defer +import synapse.state from synapse import event_auth from synapse.api.constants import EventTypes from synapse.api.errors import AuthError +from synapse.events import EventBase logger = logging.getLogger(__name__) @defer.inlineCallbacks -def resolve_events_with_store(room_version, state_sets, event_map, state_res_store): +def resolve_events_with_store( + room_id: str, + room_version: str, + state_sets: List[Dict[Tuple[str, str], str]], + event_map: Optional[Dict[str, EventBase]], + state_res_store: "synapse.state.StateResolutionStore", +): """Resolves the state using the v2 state resolution algorithm Args: - room_version (str): The room version + room_id: the room we are working in + + room_version: The room version - state_sets(list): List of dicts of (type, state_key) -> event_id, + state_sets: List of dicts of (type, state_key) -> event_id, which are the different state groups to resolve. - event_map(dict[str,FrozenEvent]|None): + event_map: a dict from event_id to event, for any events that we happen to have in flight (eg, those currently being persisted). This will be used as a starting point fof finding the state we need; any missing @@ -46,9 +57,9 @@ def resolve_events_with_store(room_version, state_sets, event_map, state_res_sto If None, all events will be fetched via state_res_store. - state_res_store (StateResolutionStore) + state_res_store: - Returns + Returns: Deferred[dict[(str, str), str]]: a map from (type, state_key) to event_id. """ @@ -84,6 +95,14 @@ def resolve_events_with_store(room_version, state_sets, event_map, state_res_sto ) event_map.update(events) + # everything in the event map should be in the right room + for event in event_map.values(): + if event.room_id != room_id: + raise Exception( + "Attempting to state-resolve for room %s with event %s which is in %s" + % (room_id, event.event_id, event.room_id,) + ) + full_conflicted_set = set(eid for eid in full_conflicted_set if eid in event_map) logger.debug("%d full_conflicted_set entries", len(full_conflicted_set)) @@ -94,13 +113,14 @@ def resolve_events_with_store(room_version, state_sets, event_map, state_res_sto ) sorted_power_events = yield _reverse_topological_power_sort( - power_events, event_map, state_res_store, full_conflicted_set + room_id, power_events, event_map, state_res_store, full_conflicted_set ) logger.debug("sorted %d power events", len(sorted_power_events)) # Now sequentially auth each one resolved_state = yield _iterative_auth_checks( + room_id, room_version, sorted_power_events, unconflicted_state, @@ -121,13 +141,18 @@ def resolve_events_with_store(room_version, state_sets, event_map, state_res_sto pl = resolved_state.get((EventTypes.PowerLevels, ""), None) leftover_events = yield _mainline_sort( - leftover_events, pl, event_map, state_res_store + room_id, leftover_events, pl, event_map, state_res_store ) logger.debug("resolving remaining events") resolved_state = yield _iterative_auth_checks( - room_version, leftover_events, resolved_state, event_map, state_res_store + room_id, + room_version, + leftover_events, + resolved_state, + event_map, + state_res_store, ) logger.debug("resolved") @@ -141,11 +166,12 @@ def resolve_events_with_store(room_version, state_sets, event_map, state_res_sto @defer.inlineCallbacks -def _get_power_level_for_sender(event_id, event_map, state_res_store): +def _get_power_level_for_sender(room_id, event_id, event_map, state_res_store): """Return the power level of the sender of the given event according to their auth events. Args: + room_id (str) event_id (str) event_map (dict[str,FrozenEvent]) state_res_store (StateResolutionStore) @@ -153,11 +179,11 @@ def _get_power_level_for_sender(event_id, event_map, state_res_store): Returns: Deferred[int] """ - event = yield _get_event(event_id, event_map, state_res_store) + event = yield _get_event(room_id, event_id, event_map, state_res_store) pl = None for aid in event.auth_event_ids(): - aev = yield _get_event(aid, event_map, state_res_store) + aev = yield _get_event(room_id, aid, event_map, state_res_store) if (aev.type, aev.state_key) == (EventTypes.PowerLevels, ""): pl = aev break @@ -165,7 +191,7 @@ def _get_power_level_for_sender(event_id, event_map, state_res_store): if pl is None: # Couldn't find power level. Check if they're the creator of the room for aid in event.auth_event_ids(): - aev = yield _get_event(aid, event_map, state_res_store) + aev = yield _get_event(room_id, aid, event_map, state_res_store) if (aev.type, aev.state_key) == (EventTypes.Create, ""): if aev.content.get("creator") == event.sender: return 100 @@ -279,7 +305,7 @@ def _is_power_event(event): @defer.inlineCallbacks def _add_event_and_auth_chain_to_graph( - graph, event_id, event_map, state_res_store, auth_diff + graph, room_id, event_id, event_map, state_res_store, auth_diff ): """Helper function for _reverse_topological_power_sort that add the event and its auth chain (that is in the auth diff) to the graph @@ -287,6 +313,7 @@ def _add_event_and_auth_chain_to_graph( Args: graph (dict[str, set[str]]): A map from event ID to the events auth event IDs + room_id (str): the room we are working in event_id (str): Event to add to the graph event_map (dict[str,FrozenEvent]) state_res_store (StateResolutionStore) @@ -298,7 +325,7 @@ def _add_event_and_auth_chain_to_graph( eid = state.pop() graph.setdefault(eid, set()) - event = yield _get_event(eid, event_map, state_res_store) + event = yield _get_event(room_id, eid, event_map, state_res_store) for aid in event.auth_event_ids(): if aid in auth_diff: if aid not in graph: @@ -308,11 +335,14 @@ def _add_event_and_auth_chain_to_graph( @defer.inlineCallbacks -def _reverse_topological_power_sort(event_ids, event_map, state_res_store, auth_diff): +def _reverse_topological_power_sort( + room_id, event_ids, event_map, state_res_store, auth_diff +): """Returns a list of the event_ids sorted by reverse topological ordering, and then by power level and origin_server_ts Args: + room_id (str): the room we are working in event_ids (list[str]): The events to sort event_map (dict[str,FrozenEvent]) state_res_store (StateResolutionStore) @@ -325,12 +355,14 @@ def _reverse_topological_power_sort(event_ids, event_map, state_res_store, auth_ graph = {} for event_id in event_ids: yield _add_event_and_auth_chain_to_graph( - graph, event_id, event_map, state_res_store, auth_diff + graph, room_id, event_id, event_map, state_res_store, auth_diff ) event_to_pl = {} for event_id in graph: - pl = yield _get_power_level_for_sender(event_id, event_map, state_res_store) + pl = yield _get_power_level_for_sender( + room_id, event_id, event_map, state_res_store + ) event_to_pl[event_id] = pl def _get_power_order(event_id): @@ -348,12 +380,13 @@ def _reverse_topological_power_sort(event_ids, event_map, state_res_store, auth_ @defer.inlineCallbacks def _iterative_auth_checks( - room_version, event_ids, base_state, event_map, state_res_store + room_id, room_version, event_ids, base_state, event_map, state_res_store ): """Sequentially apply auth checks to each event in given list, updating the state as it goes along. Args: + room_id (str) room_version (str) event_ids (list[str]): Ordered list of events to apply auth checks to base_state (dict[tuple[str, str], str]): The set of state to start with @@ -370,7 +403,7 @@ def _iterative_auth_checks( auth_events = {} for aid in event.auth_event_ids(): - ev = yield _get_event(aid, event_map, state_res_store) + ev = yield _get_event(room_id, aid, event_map, state_res_store) if ev.rejected_reason is None: auth_events[(ev.type, ev.state_key)] = ev @@ -378,7 +411,7 @@ def _iterative_auth_checks( for key in event_auth.auth_types_for_event(event): if key in resolved_state: ev_id = resolved_state[key] - ev = yield _get_event(ev_id, event_map, state_res_store) + ev = yield _get_event(room_id, ev_id, event_map, state_res_store) if ev.rejected_reason is None: auth_events[key] = event_map[ev_id] @@ -400,11 +433,14 @@ def _iterative_auth_checks( @defer.inlineCallbacks -def _mainline_sort(event_ids, resolved_power_event_id, event_map, state_res_store): +def _mainline_sort( + room_id, event_ids, resolved_power_event_id, event_map, state_res_store +): """Returns a sorted list of event_ids sorted by mainline ordering based on the given event resolved_power_event_id Args: + room_id (str): room we're working in event_ids (list[str]): Events to sort resolved_power_event_id (str): The final resolved power level event ID event_map (dict[str,FrozenEvent]) @@ -417,11 +453,11 @@ def _mainline_sort(event_ids, resolved_power_event_id, event_map, state_res_stor pl = resolved_power_event_id while pl: mainline.append(pl) - pl_ev = yield _get_event(pl, event_map, state_res_store) + pl_ev = yield _get_event(room_id, pl, event_map, state_res_store) auth_events = pl_ev.auth_event_ids() pl = None for aid in auth_events: - ev = yield _get_event(aid, event_map, state_res_store) + ev = yield _get_event(room_id, aid, event_map, state_res_store) if (ev.type, ev.state_key) == (EventTypes.PowerLevels, ""): pl = aid break @@ -457,6 +493,8 @@ def _get_mainline_depth_for_event(event, mainline_map, event_map, state_res_stor Deferred[int] """ + room_id = event.room_id + # We do an iterative search, replacing `event with the power level in its # auth events (if any) while event: @@ -468,7 +506,7 @@ def _get_mainline_depth_for_event(event, mainline_map, event_map, state_res_stor event = None for aid in auth_events: - aev = yield _get_event(aid, event_map, state_res_store) + aev = yield _get_event(room_id, aid, event_map, state_res_store) if (aev.type, aev.state_key) == (EventTypes.PowerLevels, ""): event = aev break @@ -478,11 +516,12 @@ def _get_mainline_depth_for_event(event, mainline_map, event_map, state_res_stor @defer.inlineCallbacks -def _get_event(event_id, event_map, state_res_store): +def _get_event(room_id, event_id, event_map, state_res_store): """Helper function to look up event in event_map, falling back to looking it up in the store Args: + room_id (str) event_id (str) event_map (dict[str,FrozenEvent]) state_res_store (StateResolutionStore) @@ -493,7 +532,14 @@ def _get_event(event_id, event_map, state_res_store): if event_id not in event_map: events = yield state_res_store.get_events([event_id], allow_rejected=True) event_map.update(events) - return event_map[event_id] + event = event_map[event_id] + assert event is not None + if event.room_id != room_id: + raise Exception( + "In state res for room %s, event %s is in %s" + % (room_id, event_id, event.room_id) + ) + return event def lexicographical_topological_sort(graph, key): diff --git a/tests/state/test_v2.py b/tests/state/test_v2.py index 8d3845c870..0f341d3ac3 100644 --- a/tests/state/test_v2.py +++ b/tests/state/test_v2.py @@ -58,6 +58,7 @@ class FakeEvent(object): self.type = type self.state_key = state_key self.content = content + self.room_id = ROOM_ID def to_event(self, auth_events, prev_events): """Given the auth_events and prev_events, convert to a Frozen Event @@ -418,6 +419,7 @@ class StateTestCase(unittest.TestCase): state_before = dict(state_at_event[prev_events[0]]) else: state_d = resolve_events_with_store( + ROOM_ID, RoomVersions.V2.identifier, [state_at_event[n] for n in prev_events], event_map=event_map, @@ -565,6 +567,7 @@ class SimpleParamStateTestCase(unittest.TestCase): # Test that we correctly handle passing `None` as the event_map state_d = resolve_events_with_store( + ROOM_ID, RoomVersions.V2.identifier, [self.state_at_bob, self.state_at_charlie], event_map=None, -- cgit 1.4.1 From 596dd9914dad1933ded1426bdec1e2b1e6874e39 Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Mon, 16 Dec 2019 14:53:21 +0000 Subject: Add test case --- tests/rest/client/v1/test_rooms.py | 133 +++++++++++++++++++++++++++++++++++++ 1 file changed, 133 insertions(+) (limited to 'tests') diff --git a/tests/rest/client/v1/test_rooms.py b/tests/rest/client/v1/test_rooms.py index 1ca7fa742f..9cb505f316 100644 --- a/tests/rest/client/v1/test_rooms.py +++ b/tests/rest/client/v1/test_rooms.py @@ -29,6 +29,7 @@ import synapse.rest.admin from synapse.api.constants import EventContentFields, EventTypes, Membership from synapse.handlers.pagination import PurgeStatus from synapse.rest.client.v1 import login, profile, room +from synapse.rest.client.v2_alpha import account from synapse.util.stringutils import random_string from tests import unittest @@ -1597,3 +1598,135 @@ class LabelsTestCase(unittest.HomeserverTestCase): ) return event_id + + +class ContextTestCase(unittest.HomeserverTestCase): + + servlets = [ + synapse.rest.admin.register_servlets_for_client_rest_resource, + room.register_servlets, + login.register_servlets, + account.register_servlets, + ] + + def make_homeserver(self, reactor, clock): + self.hs = self.setup_test_homeserver() + + return self.hs + + def prepare(self, reactor, clock, homeserver): + self.user_id = self.register_user("user", "password") + self.tok = self.login("user", "password") + self.room_id = self.helper.create_room_as(self.user_id, tok=self.tok) + + self.other_user_id = self.register_user("user2", "password") + self.other_tok = self.login("user2", "password") + + self.helper.invite(self.room_id, self.user_id, self.other_user_id, tok=self.tok) + self.helper.join(self.room_id, self.other_user_id, tok=self.other_tok) + + def test_erased_sender(self): + """Test that an erasure request results in the requester's events being hidden + from any new member of the room. + """ + + # Send a bunch of events in the room. + + self.helper.send(self.room_id, "message 1", tok=self.tok) + self.helper.send(self.room_id, "message 2", tok=self.tok) + event_id = self.helper.send(self.room_id, "message 3", tok=self.tok)["event_id"] + self.helper.send(self.room_id, "message 4", tok=self.tok) + self.helper.send(self.room_id, "message 5", tok=self.tok) + + # Check that we can still see the messages before the erasure request. + + request, channel = self.make_request( + "GET", + '/rooms/%s/context/%s?filter={"types":["m.room.message"]}' + % (self.room_id, event_id), + access_token=self.tok, + ) + self.render(request) + self.assertEqual(channel.code, 200, channel.result) + + events_before = channel.json_body["events_before"] + + self.assertEqual(len(events_before), 2, events_before) + self.assertEqual( + events_before[0].get("content", {}).get("body"), + "message 2", + events_before[0], + ) + self.assertEqual( + events_before[1].get("content", {}).get("body"), + "message 1", + events_before[1], + ) + + self.assertEqual( + channel.json_body["event"].get("content", {}).get("body"), + "message 3", + channel.json_body["event"], + ) + + events_after = channel.json_body["events_after"] + + self.assertEqual(len(events_after), 2, events_after) + self.assertEqual( + events_after[0].get("content", {}).get("body"), + "message 4", + events_after[0], + ) + self.assertEqual( + events_after[1].get("content", {}).get("body"), + "message 5", + events_after[1], + ) + + # Deactivate the first account and erase the user's data. + + deactivate_account_handler = self.hs.get_deactivate_account_handler() + self.get_success( + deactivate_account_handler.deactivate_account(self.user_id, erase_data=True) + ) + + # Invite another user in the room. This is needed because messages will be + # pruned only if the user wasn't a member of the room when the messages were + # sent. + + invited_user_id = self.register_user("user3", "password") + invited_tok = self.login("user3", "password") + + self.helper.invite( + self.room_id, self.other_user_id, invited_user_id, tok=self.other_tok + ) + self.helper.join(self.room_id, invited_user_id, tok=invited_tok) + + # Check that a user that joined the room after the erasure request can't see + # the messages anymore. + + request, channel = self.make_request( + 'GET', + '/rooms/%s/context/%s?filter={"types":["m.room.message"]}' + % (self.room_id, event_id), + access_token=invited_tok, + ) + self.render(request) + self.assertEqual(channel.code, 200, channel.result) + + events_before = channel.json_body["events_before"] + + self.assertEqual(len(events_before), 2, events_before) + self.assertDictEqual(events_before[0].get("content"), {}, events_before[0]) + self.assertDictEqual(events_before[1].get("content"), {}, events_before[1]) + + self.assertDictEqual( + channel.json_body["event"].get("content"), {}, channel.json_body["event"] + ) + + events_after = channel.json_body["events_after"] + + self.assertEqual(len(events_after), 2, events_after) + self.assertDictEqual(events_after[0].get("content"), {}, events_after[0]) + self.assertEqual(events_after[1].get("content"), {}, events_after[1]) + -- cgit 1.4.1 From a29420f9f449da72d1d38bcab4cedc182e9f2ba0 Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Mon, 16 Dec 2019 14:55:50 +0000 Subject: Lint --- tests/rest/client/v1/test_rooms.py | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'tests') diff --git a/tests/rest/client/v1/test_rooms.py b/tests/rest/client/v1/test_rooms.py index 9cb505f316..dca0fef97a 100644 --- a/tests/rest/client/v1/test_rooms.py +++ b/tests/rest/client/v1/test_rooms.py @@ -1706,7 +1706,7 @@ class ContextTestCase(unittest.HomeserverTestCase): # the messages anymore. request, channel = self.make_request( - 'GET', + "GET", '/rooms/%s/context/%s?filter={"types":["m.room.message"]}' % (self.room_id, event_id), access_token=invited_tok, @@ -1729,4 +1729,3 @@ class ContextTestCase(unittest.HomeserverTestCase): self.assertEqual(len(events_after), 2, events_after) self.assertDictEqual(events_after[0].get("content"), {}, events_after[0]) self.assertEqual(events_after[1].get("content"), {}, events_after[1]) - -- cgit 1.4.1 From a82006954912ed96b0d47db43db44e76e5b052d6 Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Mon, 16 Dec 2019 16:00:18 +0000 Subject: Incorporate review --- synapse/handlers/room.py | 2 +- tests/rest/client/v1/test_rooms.py | 5 ----- 2 files changed, 1 insertion(+), 6 deletions(-) (limited to 'tests') diff --git a/synapse/handlers/room.py b/synapse/handlers/room.py index 7f979e5812..60b8bbc7a5 100644 --- a/synapse/handlers/room.py +++ b/synapse/handlers/room.py @@ -941,7 +941,7 @@ class RoomContextHandler(object): if event_filter: state_events = event_filter.filter(state_events) - results["state"] = state_events + results["state"] = yield filter_evts(state_events) # We use a dummy token here as we only care about the room portion of # the token, which we replace. diff --git a/tests/rest/client/v1/test_rooms.py b/tests/rest/client/v1/test_rooms.py index dca0fef97a..e3af280ba6 100644 --- a/tests/rest/client/v1/test_rooms.py +++ b/tests/rest/client/v1/test_rooms.py @@ -1609,11 +1609,6 @@ class ContextTestCase(unittest.HomeserverTestCase): account.register_servlets, ] - def make_homeserver(self, reactor, clock): - self.hs = self.setup_test_homeserver() - - return self.hs - def prepare(self, reactor, clock, homeserver): self.user_id = self.register_user("user", "password") self.tok = self.login("user", "password") -- cgit 1.4.1 From bfb95654c97a8d3aa164eff96ecc13755c1c326d Mon Sep 17 00:00:00 2001 From: Will Hunt Date: Mon, 16 Dec 2019 16:11:55 +0000 Subject: Add option to allow profile queries without sharing a room (#6523) --- changelog.d/6523.feature | 1 + docs/sample_config.yaml | 7 +++++++ synapse/config/server.py | 13 +++++++++++++ synapse/handlers/profile.py | 6 +++++- tests/rest/client/v1/test_profile.py | 2 ++ 5 files changed, 28 insertions(+), 1 deletion(-) create mode 100644 changelog.d/6523.feature (limited to 'tests') diff --git a/changelog.d/6523.feature b/changelog.d/6523.feature new file mode 100644 index 0000000000..798fa143df --- /dev/null +++ b/changelog.d/6523.feature @@ -0,0 +1 @@ +Add option `limit_profile_requests_to_users_who_share_rooms` to prevent requirement of a local user sharing a room with another user to query their profile information. diff --git a/docs/sample_config.yaml b/docs/sample_config.yaml index 4d44e631d1..1787248f53 100644 --- a/docs/sample_config.yaml +++ b/docs/sample_config.yaml @@ -54,6 +54,13 @@ pid_file: DATADIR/homeserver.pid # #require_auth_for_profile_requests: true +# Uncomment to require a user to share a room with another user in order +# to retrieve their profile information. Only checked on Client-Server +# requests. Profile requests from other servers should be checked by the +# requesting server. Defaults to 'false'. +# +#limit_profile_requests_to_users_who_share_rooms: true + # If set to 'true', removes the need for authentication to access the server's # public rooms directory through the client API, meaning that anyone can # query the room directory. Defaults to 'false'. diff --git a/synapse/config/server.py b/synapse/config/server.py index 50af858c76..38f6ff9edc 100644 --- a/synapse/config/server.py +++ b/synapse/config/server.py @@ -102,6 +102,12 @@ class ServerConfig(Config): "require_auth_for_profile_requests", False ) + # Whether to require sharing a room with a user to retrieve their + # profile data + self.limit_profile_requests_to_users_who_share_rooms = config.get( + "limit_profile_requests_to_users_who_share_rooms", False, + ) + if "restrict_public_rooms_to_local_users" in config and ( "allow_public_rooms_without_auth" in config or "allow_public_rooms_over_federation" in config @@ -621,6 +627,13 @@ class ServerConfig(Config): # #require_auth_for_profile_requests: true + # Uncomment to require a user to share a room with another user in order + # to retrieve their profile information. Only checked on Client-Server + # requests. Profile requests from other servers should be checked by the + # requesting server. Defaults to 'false'. + # + #limit_profile_requests_to_users_who_share_rooms: true + # If set to 'true', removes the need for authentication to access the server's # public rooms directory through the client API, meaning that anyone can # query the room directory. Defaults to 'false'. diff --git a/synapse/handlers/profile.py b/synapse/handlers/profile.py index 1e5a4613c9..f9579d69ee 100644 --- a/synapse/handlers/profile.py +++ b/synapse/handlers/profile.py @@ -295,12 +295,16 @@ class BaseProfileHandler(BaseHandler): be found to be in any room the server is in, and therefore the query is denied. """ + # Implementation of MSC1301: don't allow looking up profiles if the # requester isn't in the same room as the target. We expect requester to # be None when this function is called outside of a profile query, e.g. # when building a membership event. In this case, we must allow the # lookup. - if not self.hs.config.require_auth_for_profile_requests or not requester: + if ( + not self.hs.config.limit_profile_requests_to_users_who_share_rooms + or not requester + ): return # Always allow the user to query their own profile. diff --git a/tests/rest/client/v1/test_profile.py b/tests/rest/client/v1/test_profile.py index 12c5e95cb5..8df58b4a63 100644 --- a/tests/rest/client/v1/test_profile.py +++ b/tests/rest/client/v1/test_profile.py @@ -237,6 +237,7 @@ class ProfilesRestrictedTestCase(unittest.HomeserverTestCase): config = self.default_config() config["require_auth_for_profile_requests"] = True + config["limit_profile_requests_to_users_who_share_rooms"] = True self.hs = self.setup_test_homeserver(config=config) return self.hs @@ -309,6 +310,7 @@ class OwnProfileUnrestrictedTestCase(unittest.HomeserverTestCase): def make_homeserver(self, reactor, clock): config = self.default_config() config["require_auth_for_profile_requests"] = True + config["limit_profile_requests_to_users_who_share_rooms"] = True self.hs = self.setup_test_homeserver(config=config) return self.hs -- cgit 1.4.1 From 2284eb3a533a2df04784df08da28e67d6588a5ea Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 18 Dec 2019 10:45:12 +0000 Subject: Add database config class (#6513) This encapsulates config for a given database and is the way to get new connections. --- changelog.d/6513.misc | 1 + scripts-dev/update_database | 9 +-- scripts/synapse_port_db | 58 ++++++++----------- synapse/config/database.py | 78 ++++++++++++++++++++------ synapse/handlers/presence.py | 2 +- synapse/server.py | 41 ++------------ synapse/storage/_base.py | 2 +- synapse/storage/data_stores/__init__.py | 40 ++++++++++--- synapse/storage/data_stores/main/client_ips.py | 2 +- synapse/storage/database.py | 45 ++++++++++++++- synapse/storage/engines/sqlite.py | 16 +++++- synapse/storage/prepare_database.py | 7 +-- tests/handlers/test_typing.py | 39 ++++++------- tests/replication/slave/storage/_base.py | 6 +- tests/server.py | 55 +++++++++--------- tests/storage/test_appservice.py | 37 ++++++++---- tests/storage/test_base.py | 14 +++-- tests/storage/test_registration.py | 1 - tests/utils.py | 43 +++++--------- 19 files changed, 287 insertions(+), 209 deletions(-) create mode 100644 changelog.d/6513.misc (limited to 'tests') diff --git a/changelog.d/6513.misc b/changelog.d/6513.misc new file mode 100644 index 0000000000..36700f5657 --- /dev/null +++ b/changelog.d/6513.misc @@ -0,0 +1 @@ +Remove all assumptions of there being a single phyiscal DB apart from the `synapse.config`. diff --git a/scripts-dev/update_database b/scripts-dev/update_database index 23017c21f8..1d62f0403a 100755 --- a/scripts-dev/update_database +++ b/scripts-dev/update_database @@ -26,7 +26,6 @@ from synapse.config.homeserver import HomeServerConfig from synapse.metrics.background_process_metrics import run_as_background_process from synapse.server import HomeServer from synapse.storage import DataStore -from synapse.storage.prepare_database import prepare_database logger = logging.getLogger("update_database") @@ -77,12 +76,8 @@ if __name__ == "__main__": # Instantiate and initialise the homeserver object. hs = MockHomeserver(config) - db_conn = hs.get_db_conn() - # Update the database to the latest schema. - prepare_database(db_conn, hs.database_engine, config=config) - db_conn.commit() - - # setup instantiates the store within the homeserver object. + # Setup instantiates the store within the homeserver object and updates the + # DB. hs.setup() store = hs.get_datastore() diff --git a/scripts/synapse_port_db b/scripts/synapse_port_db index e393a9b2f7..5b5368988c 100755 --- a/scripts/synapse_port_db +++ b/scripts/synapse_port_db @@ -30,6 +30,7 @@ import yaml from twisted.enterprise import adbapi from twisted.internet import defer, reactor +from synapse.config.database import DatabaseConnectionConfig from synapse.config.homeserver import HomeServerConfig from synapse.logging.context import PreserveLoggingContext from synapse.storage._base import LoggingTransaction @@ -55,7 +56,7 @@ from synapse.storage.data_stores.main.stats import StatsStore from synapse.storage.data_stores.main.user_directory import ( UserDirectoryBackgroundUpdateStore, ) -from synapse.storage.database import Database +from synapse.storage.database import Database, make_conn from synapse.storage.engines import create_engine from synapse.storage.prepare_database import prepare_database from synapse.util import Clock @@ -165,23 +166,17 @@ class Store( class MockHomeserver: - def __init__(self, config, database_engine, db_conn, db_pool): - self.database_engine = database_engine - self.db_conn = db_conn - self.db_pool = db_pool + def __init__(self, config): self.clock = Clock(reactor) self.config = config self.hostname = config.server_name - def get_db_conn(self): - return self.db_conn - - def get_db_pool(self): - return self.db_pool - def get_clock(self): return self.clock + def get_reactor(self): + return reactor + class Porter(object): def __init__(self, **kwargs): @@ -445,45 +440,36 @@ class Porter(object): else: return - def setup_db(self, db_config, database_engine): - db_conn = database_engine.module.connect( - **{ - k: v - for k, v in db_config.get("args", {}).items() - if not k.startswith("cp_") - } - ) - - prepare_database(db_conn, database_engine, config=None) + def setup_db(self, db_config: DatabaseConnectionConfig, engine): + db_conn = make_conn(db_config, engine) + prepare_database(db_conn, engine, config=None) db_conn.commit() return db_conn @defer.inlineCallbacks - def build_db_store(self, config): + def build_db_store(self, db_config: DatabaseConnectionConfig): """Builds and returns a database store using the provided configuration. Args: - config: The database configuration, i.e. a dict following the structure of - the "database" section of Synapse's configuration file. + config: The database configuration Returns: The built Store object. """ - engine = create_engine(config) - - self.progress.set_state("Preparing %s" % config["name"]) - conn = self.setup_db(config, engine) + self.progress.set_state("Preparing %s" % db_config.config["name"]) - db_pool = adbapi.ConnectionPool(config["name"], **config["args"]) + engine = create_engine(db_config.config) + conn = self.setup_db(db_config, engine) - hs = MockHomeserver(self.hs_config, engine, conn, db_pool) + hs = MockHomeserver(self.hs_config) - store = Store(Database(hs), conn, hs) + store = Store(Database(hs, db_config, engine), conn, hs) yield store.db.runInteraction( - "%s_engine.check_database" % config["name"], engine.check_database, + "%s_engine.check_database" % db_config.config["name"], + engine.check_database, ) return store @@ -509,7 +495,11 @@ class Porter(object): @defer.inlineCallbacks def run(self): try: - self.sqlite_store = yield self.build_db_store(self.sqlite_config) + self.sqlite_store = yield self.build_db_store( + DatabaseConnectionConfig( + "master", self.sqlite_config, data_stores=["main"] + ) + ) # Check if all background updates are done, abort if not. updates_complete = ( @@ -524,7 +514,7 @@ class Porter(object): defer.returnValue(None) self.postgres_store = yield self.build_db_store( - self.hs_config.database_config + self.hs_config.get_single_database() ) yield self.run_background_updates_on_postgres() diff --git a/synapse/config/database.py b/synapse/config/database.py index 0e2509f0b1..5f2f3c7cfd 100644 --- a/synapse/config/database.py +++ b/synapse/config/database.py @@ -12,12 +12,43 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. +import logging import os from textwrap import indent +from typing import List import yaml -from ._base import Config +from synapse.config._base import Config, ConfigError + +logger = logging.getLogger(__name__) + + +class DatabaseConnectionConfig: + """Contains the connection config for a particular database. + + Args: + name: A label for the database, used for logging. + db_config: The config for a particular database, as per `database` + section of main config. Has two fields: `name` for database + module name, and `args` for the args to give to the database + connector. + data_stores: The list of data stores that should be provisioned on the + database. + """ + + def __init__(self, name: str, db_config: dict, data_stores: List[str]): + if db_config["name"] not in ("sqlite3", "psycopg2"): + raise ConfigError("Unsupported database type %r" % (db_config["name"],)) + + if db_config["name"] == "sqlite3": + db_config.setdefault("args", {}).update( + {"cp_min": 1, "cp_max": 1, "check_same_thread": False} + ) + + self.name = name + self.config = db_config + self.data_stores = data_stores class DatabaseConfig(Config): @@ -26,20 +57,14 @@ class DatabaseConfig(Config): def read_config(self, config, **kwargs): self.event_cache_size = self.parse_size(config.get("event_cache_size", "10K")) - self.database_config = config.get("database") + database_config = config.get("database") - if self.database_config is None: - self.database_config = {"name": "sqlite3", "args": {}} + if database_config is None: + database_config = {"name": "sqlite3", "args": {}} - name = self.database_config.get("name", None) - if name == "psycopg2": - pass - elif name == "sqlite3": - self.database_config.setdefault("args", {}).update( - {"cp_min": 1, "cp_max": 1, "check_same_thread": False} - ) - else: - raise RuntimeError("Unsupported database type '%s'" % (name,)) + self.databases = [ + DatabaseConnectionConfig("master", database_config, data_stores=["main"]) + ] self.set_databasepath(config.get("database_path")) @@ -76,11 +101,24 @@ class DatabaseConfig(Config): self.set_databasepath(args.database_path) def set_databasepath(self, database_path): + if database_path is None: + return + if database_path != ":memory:": database_path = self.abspath(database_path) - if self.database_config.get("name", None) == "sqlite3": - if database_path is not None: - self.database_config["args"]["database"] = database_path + + # We only support setting a database path if we have a single sqlite3 + # database. + if len(self.databases) != 1: + raise ConfigError("Cannot specify 'database_path' with multiple databases") + + database = self.get_single_database() + if database.config["name"] != "sqlite3": + # We don't raise here as we haven't done so before for this case. + logger.warn("Ignoring 'database_path' for non-sqlite3 database") + return + + database.config["args"]["database"] = database_path @staticmethod def add_arguments(parser): @@ -91,3 +129,11 @@ class DatabaseConfig(Config): metavar="SQLITE_DATABASE_PATH", help="The path to a sqlite database to use.", ) + + def get_single_database(self) -> DatabaseConnectionConfig: + """Returns the database if there is only one, useful for e.g. tests + """ + if len(self.databases) != 1: + raise Exception("More than one database exists") + + return self.databases[0] diff --git a/synapse/handlers/presence.py b/synapse/handlers/presence.py index eda15bc623..240c4add12 100644 --- a/synapse/handlers/presence.py +++ b/synapse/handlers/presence.py @@ -230,7 +230,7 @@ class PresenceHandler(object): is some spurious presence changes that will self-correct. """ # If the DB pool has already terminated, don't try updating - if not self.hs.get_db_pool().running: + if not self.store.database.is_running(): return logger.info( diff --git a/synapse/server.py b/synapse/server.py index 5021068ce0..7926867b77 100644 --- a/synapse/server.py +++ b/synapse/server.py @@ -25,7 +25,6 @@ import abc import logging import os -from twisted.enterprise import adbapi from twisted.mail.smtp import sendmail from twisted.web.client import BrowserLikePolicyForHTTPS @@ -98,7 +97,6 @@ from synapse.server_notices.worker_server_notices_sender import ( ) from synapse.state import StateHandler, StateResolutionHandler from synapse.storage import DataStores, Storage -from synapse.storage.engines import create_engine from synapse.streams.events import EventSources from synapse.util import Clock from synapse.util.distributor import Distributor @@ -134,7 +132,6 @@ class HomeServer(object): DEPENDENCIES = [ "http_client", - "db_pool", "federation_client", "federation_server", "handlers", @@ -233,12 +230,6 @@ class HomeServer(object): self.admin_redaction_ratelimiter = Ratelimiter() self.registration_ratelimiter = Ratelimiter() - self.database_engine = create_engine(config.database_config) - config.database_config.setdefault("args", {})[ - "cp_openfun" - ] = self.database_engine.on_new_connection - self.db_config = config.database_config - self.datastores = None # Other kwargs are explicit dependencies @@ -247,10 +238,8 @@ class HomeServer(object): def setup(self): logger.info("Setting up.") - with self.get_db_conn() as conn: - self.datastores = DataStores(self.DATASTORE_CLASS, conn, self) - conn.commit() self.start_time = int(self.get_clock().time()) + self.datastores = DataStores(self.DATASTORE_CLASS, self) logger.info("Finished setting up.") def setup_master(self): @@ -284,6 +273,9 @@ class HomeServer(object): def get_datastore(self): return self.datastores.main + def get_datastores(self): + return self.datastores + def get_config(self): return self.config @@ -433,31 +425,6 @@ class HomeServer(object): ) return MatrixFederationHttpClient(self, tls_client_options_factory) - def build_db_pool(self): - name = self.db_config["name"] - - return adbapi.ConnectionPool( - name, cp_reactor=self.get_reactor(), **self.db_config.get("args", {}) - ) - - def get_db_conn(self, run_new_connection=True): - """Makes a new connection to the database, skipping the db pool - - Returns: - Connection: a connection object implementing the PEP-249 spec - """ - # Any param beginning with cp_ is a parameter for adbapi, and should - # not be passed to the database engine. - db_params = { - k: v - for k, v in self.db_config.get("args", {}).items() - if not k.startswith("cp_") - } - db_conn = self.database_engine.module.connect(**db_params) - if run_new_connection: - self.database_engine.on_new_connection(db_conn) - return db_conn - def build_media_repository_resource(self): # build the media repo resource. This indirects through the HomeServer # to ensure that we only have a single instance of diff --git a/synapse/storage/_base.py b/synapse/storage/_base.py index b7637b5dc0..88546ad614 100644 --- a/synapse/storage/_base.py +++ b/synapse/storage/_base.py @@ -40,7 +40,7 @@ class SQLBaseStore(object): def __init__(self, database: Database, db_conn, hs): self.hs = hs self._clock = hs.get_clock() - self.database_engine = hs.database_engine + self.database_engine = database.engine self.db = database self.rand = random.SystemRandom() diff --git a/synapse/storage/data_stores/__init__.py b/synapse/storage/data_stores/__init__.py index cafedd5c0d..0983e059c0 100644 --- a/synapse/storage/data_stores/__init__.py +++ b/synapse/storage/data_stores/__init__.py @@ -13,24 +13,50 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.storage.database import Database +import logging + +from synapse.storage.database import Database, make_conn +from synapse.storage.engines import create_engine from synapse.storage.prepare_database import prepare_database +logger = logging.getLogger(__name__) + class DataStores(object): """The various data stores. These are low level interfaces to physical databases. + + Attributes: + main (DataStore) """ - def __init__(self, main_store_class, db_conn, hs): + def __init__(self, main_store_class, hs): # Note we pass in the main store class here as workers use a different main # store. - database = Database(hs) - # Check that db is correctly configured. - database.engine.check_database(db_conn.cursor()) + self.databases = [] + + for database_config in hs.config.database.databases: + db_name = database_config.name + engine = create_engine(database_config.config) + + with make_conn(database_config, engine) as db_conn: + logger.info("Preparing database %r...", db_name) + + engine.check_database(db_conn.cursor()) + prepare_database( + db_conn, engine, hs.config, data_stores=database_config.data_stores, + ) + + database = Database(hs, database_config, engine) + + if "main" in database_config.data_stores: + logger.info("Starting 'main' data store") + self.main = main_store_class(database, db_conn, hs) + + db_conn.commit() - prepare_database(db_conn, database.engine, config=hs.config) + self.databases.append(database) - self.main = main_store_class(database, db_conn, hs) + logger.info("Database %r prepared", db_name) diff --git a/synapse/storage/data_stores/main/client_ips.py b/synapse/storage/data_stores/main/client_ips.py index add3037b69..13f4c9c72e 100644 --- a/synapse/storage/data_stores/main/client_ips.py +++ b/synapse/storage/data_stores/main/client_ips.py @@ -412,7 +412,7 @@ class ClientIpStore(ClientIpBackgroundUpdateStore): def _update_client_ips_batch(self): # If the DB pool has already terminated, don't try updating - if not self.hs.get_db_pool().running: + if not self.db.is_running(): return to_update = self._batch_row_update diff --git a/synapse/storage/database.py b/synapse/storage/database.py index ec19ae1d9d..1003dd84a5 100644 --- a/synapse/storage/database.py +++ b/synapse/storage/database.py @@ -24,9 +24,11 @@ from six.moves import intern, range from prometheus_client import Histogram +from twisted.enterprise import adbapi from twisted.internet import defer from synapse.api.errors import StoreError +from synapse.config.database import DatabaseConnectionConfig from synapse.logging.context import LoggingContext, make_deferred_yieldable from synapse.metrics.background_process_metrics import run_as_background_process from synapse.storage.background_updates import BackgroundUpdater @@ -74,6 +76,37 @@ UNIQUE_INDEX_BACKGROUND_UPDATES = { } +def make_pool( + reactor, db_config: DatabaseConnectionConfig, engine +) -> adbapi.ConnectionPool: + """Get the connection pool for the database. + """ + + return adbapi.ConnectionPool( + db_config.config["name"], + cp_reactor=reactor, + cp_openfun=engine.on_new_connection, + **db_config.config.get("args", {}) + ) + + +def make_conn(db_config: DatabaseConnectionConfig, engine): + """Make a new connection to the database and return it. + + Returns: + Connection + """ + + db_params = { + k: v + for k, v in db_config.config.get("args", {}).items() + if not k.startswith("cp_") + } + db_conn = engine.module.connect(**db_params) + engine.on_new_connection(db_conn) + return db_conn + + class LoggingTransaction(object): """An object that almost-transparently proxies for the 'txn' object passed to the constructor. Adds logging and metrics to the .execute() @@ -218,10 +251,11 @@ class Database(object): _TXN_ID = 0 - def __init__(self, hs): + def __init__(self, hs, database_config: DatabaseConnectionConfig, engine): self.hs = hs self._clock = hs.get_clock() - self._db_pool = hs.get_db_pool() + self._database_config = database_config + self._db_pool = make_pool(hs.get_reactor(), database_config, engine) self.updates = BackgroundUpdater(hs, self) @@ -234,7 +268,7 @@ class Database(object): # to watch it self._txn_perf_counters = PerformanceCounters() - self.engine = hs.database_engine + self.engine = engine # A set of tables that are not safe to use native upserts in. self._unsafe_to_upsert_tables = set(UNIQUE_INDEX_BACKGROUND_UPDATES.keys()) @@ -255,6 +289,11 @@ class Database(object): self._check_safe_to_upsert, ) + def is_running(self): + """Is the database pool currently running + """ + return self._db_pool.running + @defer.inlineCallbacks def _check_safe_to_upsert(self): """ diff --git a/synapse/storage/engines/sqlite.py b/synapse/storage/engines/sqlite.py index ddad17dc5a..df039a072d 100644 --- a/synapse/storage/engines/sqlite.py +++ b/synapse/storage/engines/sqlite.py @@ -16,8 +16,6 @@ import struct import threading -from synapse.storage.prepare_database import prepare_database - class Sqlite3Engine(object): single_threaded = True @@ -25,6 +23,9 @@ class Sqlite3Engine(object): def __init__(self, database_module, database_config): self.module = database_module + database = database_config.get("args", {}).get("database") + self._is_in_memory = database in (None, ":memory:",) + # The current max state_group, or None if we haven't looked # in the DB yet. self._current_state_group_id = None @@ -59,7 +60,16 @@ class Sqlite3Engine(object): return sql def on_new_connection(self, db_conn): - prepare_database(db_conn, self, config=None) + + # We need to import here to avoid an import loop. + from synapse.storage.prepare_database import prepare_database + + if self._is_in_memory: + # In memory databases need to be rebuilt each time. Ideally we'd + # reuse the same connection as we do when starting up, but that + # would involve using adbapi before we have started the reactor. + prepare_database(db_conn, self, config=None) + db_conn.create_function("rank", 1, _rank) def is_deadlock(self, error): diff --git a/synapse/storage/prepare_database.py b/synapse/storage/prepare_database.py index 731e1c9d9c..b4194b44ee 100644 --- a/synapse/storage/prepare_database.py +++ b/synapse/storage/prepare_database.py @@ -41,7 +41,7 @@ class UpgradeDatabaseException(PrepareDatabaseException): pass -def prepare_database(db_conn, database_engine, config): +def prepare_database(db_conn, database_engine, config, data_stores=["main"]): """Prepares a database for usage. Will either create all necessary tables or upgrade from an older schema version. @@ -54,11 +54,10 @@ def prepare_database(db_conn, database_engine, config): config (synapse.config.homeserver.HomeServerConfig|None): application config, or None if we are connecting to an existing database which we expect to be configured already + data_stores (list[str]): The name of the data stores that will be used + with this database. Defaults to all data stores. """ - # For now we only have the one datastore. - data_stores = ["main"] - try: cur = db_conn.cursor() version_info = _get_or_create_schema_state(cur, database_engine) diff --git a/tests/handlers/test_typing.py b/tests/handlers/test_typing.py index 92b8726093..596ddc6970 100644 --- a/tests/handlers/test_typing.py +++ b/tests/handlers/test_typing.py @@ -64,28 +64,29 @@ class TypingNotificationsTestCase(unittest.HomeserverTestCase): mock_federation_client = Mock(spec=["put_json"]) mock_federation_client.put_json.return_value = defer.succeed((200, "OK")) + datastores = Mock() + datastores.main = Mock( + spec=[ + # Bits that Federation needs + "prep_send_transaction", + "delivered_txn", + "get_received_txn_response", + "set_received_txn_response", + "get_destination_retry_timings", + "get_devices_by_remote", + # Bits that user_directory needs + "get_user_directory_stream_pos", + "get_current_state_deltas", + "get_device_updates_by_remote", + ] + ) + hs = self.setup_test_homeserver( - datastore=( - Mock( - spec=[ - # Bits that Federation needs - "prep_send_transaction", - "delivered_txn", - "get_received_txn_response", - "set_received_txn_response", - "get_destination_retry_timings", - "get_device_updates_by_remote", - # Bits that user_directory needs - "get_user_directory_stream_pos", - "get_current_state_deltas", - ] - ) - ), - notifier=Mock(), - http_client=mock_federation_client, - keyring=mock_keyring, + notifier=Mock(), http_client=mock_federation_client, keyring=mock_keyring ) + hs.datastores = datastores + return hs def prepare(self, reactor, clock, hs): diff --git a/tests/replication/slave/storage/_base.py b/tests/replication/slave/storage/_base.py index 3dae83c543..2a1e7c7166 100644 --- a/tests/replication/slave/storage/_base.py +++ b/tests/replication/slave/storage/_base.py @@ -20,7 +20,7 @@ from synapse.replication.tcp.client import ( ReplicationClientHandler, ) from synapse.replication.tcp.resource import ReplicationStreamProtocolFactory -from synapse.storage.database import Database +from synapse.storage.database import make_conn from tests import unittest from tests.server import FakeTransport @@ -41,10 +41,12 @@ class BaseSlavedStoreTestCase(unittest.HomeserverTestCase): def prepare(self, reactor, clock, hs): + db_config = hs.config.database.get_single_database() self.master_store = self.hs.get_datastore() self.storage = hs.get_storage() + database = hs.get_datastores().databases[0] self.slaved_store = self.STORE_TYPE( - Database(hs), self.hs.get_db_conn(), self.hs + database, make_conn(db_config, database.engine), self.hs ) self.event_id = 0 diff --git a/tests/server.py b/tests/server.py index 2b7cf4242e..a554dfdd57 100644 --- a/tests/server.py +++ b/tests/server.py @@ -302,41 +302,42 @@ def setup_test_homeserver(cleanup_func, *args, **kwargs): Set up a synchronous test server, driven by the reactor used by the homeserver. """ - d = _sth(cleanup_func, *args, **kwargs).result + server = _sth(cleanup_func, *args, **kwargs) - if isinstance(d, Failure): - d.raiseException() + database = server.config.database.get_single_database() # Make the thread pool synchronous. - clock = d.get_clock() - pool = d.get_db_pool() - - def runWithConnection(func, *args, **kwargs): - return threads.deferToThreadPool( - pool._reactor, - pool.threadpool, - pool._runWithConnection, - func, - *args, - **kwargs - ) - - def runInteraction(interaction, *args, **kwargs): - return threads.deferToThreadPool( - pool._reactor, - pool.threadpool, - pool._runInteraction, - interaction, - *args, - **kwargs - ) + clock = server.get_clock() + + for database in server.get_datastores().databases: + pool = database._db_pool + + def runWithConnection(func, *args, **kwargs): + return threads.deferToThreadPool( + pool._reactor, + pool.threadpool, + pool._runWithConnection, + func, + *args, + **kwargs + ) + + def runInteraction(interaction, *args, **kwargs): + return threads.deferToThreadPool( + pool._reactor, + pool.threadpool, + pool._runInteraction, + interaction, + *args, + **kwargs + ) - if pool: pool.runWithConnection = runWithConnection pool.runInteraction = runInteraction pool.threadpool = ThreadPool(clock._reactor) pool.running = True - return d + + return server def get_clock(): diff --git a/tests/storage/test_appservice.py b/tests/storage/test_appservice.py index 2e521e9ab7..fd52512696 100644 --- a/tests/storage/test_appservice.py +++ b/tests/storage/test_appservice.py @@ -28,7 +28,7 @@ from synapse.storage.data_stores.main.appservice import ( ApplicationServiceStore, ApplicationServiceTransactionStore, ) -from synapse.storage.database import Database +from synapse.storage.database import Database, make_conn from tests import unittest from tests.utils import setup_test_homeserver @@ -55,8 +55,10 @@ class ApplicationServiceStoreTestCase(unittest.TestCase): self._add_appservice("token2", "as2", "some_url", "some_hs_token", "bob") self._add_appservice("token3", "as3", "some_url", "some_hs_token", "bob") # must be done after inserts - database = Database(hs) - self.store = ApplicationServiceStore(database, hs.get_db_conn(), hs) + database = hs.get_datastores().databases[0] + self.store = ApplicationServiceStore( + database, make_conn(database._database_config, database.engine), hs + ) def tearDown(self): # TODO: suboptimal that we need to create files for tests! @@ -111,9 +113,6 @@ class ApplicationServiceTransactionStoreTestCase(unittest.TestCase): hs.config.event_cache_size = 1 hs.config.password_providers = [] - self.db_pool = hs.get_db_pool() - self.engine = hs.database_engine - self.as_list = [ {"token": "token1", "url": "https://matrix-as.org", "id": "id_1"}, {"token": "alpha_tok", "url": "https://alpha.com", "id": "id_alpha"}, @@ -125,8 +124,15 @@ class ApplicationServiceTransactionStoreTestCase(unittest.TestCase): self.as_yaml_files = [] - database = Database(hs) - self.store = TestTransactionStore(database, hs.get_db_conn(), hs) + # We assume there is only one database in these tests + database = hs.get_datastores().databases[0] + self.db_pool = database._db_pool + self.engine = database.engine + + db_config = hs.config.get_single_database() + self.store = TestTransactionStore( + database, make_conn(db_config, self.engine), hs + ) def _add_service(self, url, as_token, id): as_yaml = dict( @@ -419,7 +425,10 @@ class ApplicationServiceStoreConfigTestCase(unittest.TestCase): hs.config.event_cache_size = 1 hs.config.password_providers = [] - ApplicationServiceStore(Database(hs), hs.get_db_conn(), hs) + database = hs.get_datastores().databases[0] + ApplicationServiceStore( + database, make_conn(database._database_config, database.engine), hs + ) @defer.inlineCallbacks def test_duplicate_ids(self): @@ -435,7 +444,10 @@ class ApplicationServiceStoreConfigTestCase(unittest.TestCase): hs.config.password_providers = [] with self.assertRaises(ConfigError) as cm: - ApplicationServiceStore(Database(hs), hs.get_db_conn(), hs) + database = hs.get_datastores().databases[0] + ApplicationServiceStore( + database, make_conn(database._database_config, database.engine), hs + ) e = cm.exception self.assertIn(f1, str(e)) @@ -456,7 +468,10 @@ class ApplicationServiceStoreConfigTestCase(unittest.TestCase): hs.config.password_providers = [] with self.assertRaises(ConfigError) as cm: - ApplicationServiceStore(Database(hs), hs.get_db_conn(), hs) + database = hs.get_datastores().databases[0] + ApplicationServiceStore( + database, make_conn(database._database_config, database.engine), hs + ) e = cm.exception self.assertIn(f1, str(e)) diff --git a/tests/storage/test_base.py b/tests/storage/test_base.py index 537cfe9f64..cdee0a9e60 100644 --- a/tests/storage/test_base.py +++ b/tests/storage/test_base.py @@ -52,15 +52,17 @@ class SQLBaseStoreTestCase(unittest.TestCase): config = Mock() config._disable_native_upserts = True config.event_cache_size = 1 - config.database_config = {"name": "sqlite3"} - engine = create_engine(config.database_config) + hs = TestHomeServer("test", config=config) + + sqlite_config = {"name": "sqlite3"} + engine = create_engine(sqlite_config) fake_engine = Mock(wraps=engine) fake_engine.can_native_upsert = False - hs = TestHomeServer( - "test", db_pool=self.db_pool, config=config, database_engine=fake_engine - ) - self.datastore = SQLBaseStore(Database(hs), None, hs) + db = Database(Mock(), Mock(config=sqlite_config), fake_engine) + db._db_pool = self.db_pool + + self.datastore = SQLBaseStore(db, None, hs) @defer.inlineCallbacks def test_insert_1col(self): diff --git a/tests/storage/test_registration.py b/tests/storage/test_registration.py index 4578cc3b60..ed5786865a 100644 --- a/tests/storage/test_registration.py +++ b/tests/storage/test_registration.py @@ -26,7 +26,6 @@ class RegistrationStoreTestCase(unittest.TestCase): @defer.inlineCallbacks def setUp(self): hs = yield setup_test_homeserver(self.addCleanup) - self.db_pool = hs.get_db_pool() self.store = hs.get_datastore() diff --git a/tests/utils.py b/tests/utils.py index 585f305b9a..9f5bf40b4b 100644 --- a/tests/utils.py +++ b/tests/utils.py @@ -30,6 +30,7 @@ from twisted.internet import defer, reactor from synapse.api.constants import EventTypes from synapse.api.errors import CodeMessageException, cs_error from synapse.api.room_versions import RoomVersions +from synapse.config.database import DatabaseConnectionConfig from synapse.config.homeserver import HomeServerConfig from synapse.config.server import DEFAULT_ROOM_VERSION from synapse.federation.transport import server as federation_server @@ -177,7 +178,6 @@ class TestHomeServer(HomeServer): DATASTORE_CLASS = DataStore -@defer.inlineCallbacks def setup_test_homeserver( cleanup_func, name="test", @@ -214,7 +214,7 @@ def setup_test_homeserver( if USE_POSTGRES_FOR_TESTS: test_db = "synapse_test_%s" % uuid.uuid4().hex - config.database_config = { + database_config = { "name": "psycopg2", "args": { "database": test_db, @@ -226,12 +226,15 @@ def setup_test_homeserver( }, } else: - config.database_config = { + database_config = { "name": "sqlite3", "args": {"database": ":memory:", "cp_min": 1, "cp_max": 1}, } - db_engine = create_engine(config.database_config) + database = DatabaseConnectionConfig("master", database_config, ["main"]) + config.database.databases = [database] + + db_engine = create_engine(database.config) # Create the database before we actually try and connect to it, based off # the template database we generate in setupdb() @@ -251,11 +254,6 @@ def setup_test_homeserver( cur.close() db_conn.close() - # we need to configure the connection pool to run the on_new_connection - # function, so that we can test code that uses custom sqlite functions - # (like rank). - config.database_config["args"]["cp_openfun"] = db_engine.on_new_connection - if datastore is None: hs = homeserverToUse( name, @@ -267,21 +265,19 @@ def setup_test_homeserver( **kargs ) - # Prepare the DB on SQLite -- PostgreSQL is a copy of an already up to - # date db - if not isinstance(db_engine, PostgresEngine): - db_conn = hs.get_db_conn() - yield prepare_database(db_conn, db_engine, config) - db_conn.commit() - db_conn.close() + hs.setup() + if homeserverToUse.__name__ == "TestHomeServer": + hs.setup_master() + + if isinstance(db_engine, PostgresEngine): + database = hs.get_datastores().databases[0] - else: # We need to do cleanup on PostgreSQL def cleanup(): import psycopg2 # Close all the db pools - hs.get_db_pool().close() + database._db_pool.close() dropped = False @@ -320,23 +316,12 @@ def setup_test_homeserver( # Register the cleanup hook cleanup_func(cleanup) - hs.setup() - if homeserverToUse.__name__ == "TestHomeServer": - hs.setup_master() else: - # If we have been given an explicit datastore we probably want to mock - # out the DataStores somehow too. This all feels a bit wrong, but then - # mocking the stores feels wrong too. - datastores = Mock(datastore=datastore) - hs = homeserverToUse( name, - db_pool=None, datastore=datastore, - datastores=datastores, config=config, version_string="Synapse/tests", - database_engine=db_engine, tls_server_context_factory=Mock(), tls_client_options_factory=Mock(), reactor=reactor, -- cgit 1.4.1 From d6752ce5da38d35857fe324800d76a86ee1e64f1 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Wed, 18 Dec 2019 14:26:58 +0000 Subject: Clean up startup for the pusher (#6558) * Remove redundant python2 support code `str.decode()` doesn't exist on python3, so presumably this code was doing nothing * Filter out pushers with corrupt data When we get a row with unparsable json, drop the row, rather than returning a row with null `data`, which will then cause an explosion later on. * Improve logging when we can't start a pusher Log the ID to help us understand the problem * Make email pusher setup more robust We know we'll have a `data` member, since that comes from the database. What we *don't* know is if that is a dict, and if that has a `brand` member, and if that member is a string. --- changelog.d/6558.misc | 1 + synapse/push/pusher.py | 12 ++++++----- synapse/push/pusherpool.py | 10 +++++---- synapse/rest/client/v1/pusher.py | 33 +++++++++++++++--------------- synapse/storage/data_stores/main/pusher.py | 25 ++++++++-------------- tests/push/test_email.py | 3 +++ tests/push/test_http.py | 4 ++++ 7 files changed, 45 insertions(+), 43 deletions(-) create mode 100644 changelog.d/6558.misc (limited to 'tests') diff --git a/changelog.d/6558.misc b/changelog.d/6558.misc new file mode 100644 index 0000000000..a7572f1a85 --- /dev/null +++ b/changelog.d/6558.misc @@ -0,0 +1 @@ +Clean up logs from the push notifier at startup. \ No newline at end of file diff --git a/synapse/push/pusher.py b/synapse/push/pusher.py index f277aeb131..8ad0bf5936 100644 --- a/synapse/push/pusher.py +++ b/synapse/push/pusher.py @@ -80,9 +80,11 @@ class PusherFactory(object): return EmailPusher(self.hs, pusherdict, mailer) def _app_name_from_pusherdict(self, pusherdict): - if "data" in pusherdict and "brand" in pusherdict["data"]: - app_name = pusherdict["data"]["brand"] - else: - app_name = self.config.email_app_name + data = pusherdict["data"] - return app_name + if isinstance(data, dict): + brand = data.get("brand") + if isinstance(brand, str): + return brand + + return self.config.email_app_name diff --git a/synapse/push/pusherpool.py b/synapse/push/pusherpool.py index 0f6992202d..b9dca5bc63 100644 --- a/synapse/push/pusherpool.py +++ b/synapse/push/pusherpool.py @@ -232,7 +232,6 @@ class PusherPool: Deferred """ pushers = yield self.store.get_all_pushers() - logger.info("Starting %d pushers", len(pushers)) # Stagger starting up the pushers so we don't completely drown the # process on start up. @@ -245,7 +244,7 @@ class PusherPool: """Start the given pusher Args: - pusherdict (dict): + pusherdict (dict): dict with the values pulled from the db table Returns: Deferred[EmailPusher|HttpPusher] @@ -254,7 +253,8 @@ class PusherPool: p = self.pusher_factory.create_pusher(pusherdict) except PusherConfigException as e: logger.warning( - "Pusher incorrectly configured user=%s, appid=%s, pushkey=%s: %s", + "Pusher incorrectly configured id=%i, user=%s, appid=%s, pushkey=%s: %s", + pusherdict["id"], pusherdict.get("user_name"), pusherdict.get("app_id"), pusherdict.get("pushkey"), @@ -262,7 +262,9 @@ class PusherPool: ) return except Exception: - logger.exception("Couldn't start a pusher: caught Exception") + logger.exception( + "Couldn't start pusher id %i: caught Exception", pusherdict["id"], + ) return if not p: diff --git a/synapse/rest/client/v1/pusher.py b/synapse/rest/client/v1/pusher.py index 0791866f55..6f6b7aed6e 100644 --- a/synapse/rest/client/v1/pusher.py +++ b/synapse/rest/client/v1/pusher.py @@ -28,6 +28,17 @@ from synapse.rest.client.v2_alpha._base import client_patterns logger = logging.getLogger(__name__) +ALLOWED_KEYS = { + "app_display_name", + "app_id", + "data", + "device_display_name", + "kind", + "lang", + "profile_tag", + "pushkey", +} + class PushersRestServlet(RestServlet): PATTERNS = client_patterns("/pushers$", v1=True) @@ -43,23 +54,11 @@ class PushersRestServlet(RestServlet): pushers = await self.hs.get_datastore().get_pushers_by_user_id(user.to_string()) - allowed_keys = [ - "app_display_name", - "app_id", - "data", - "device_display_name", - "kind", - "lang", - "profile_tag", - "pushkey", - ] - - for p in pushers: - for k, v in list(p.items()): - if k not in allowed_keys: - del p[k] - - return 200, {"pushers": pushers} + filtered_pushers = list( + {k: v for k, v in p.items() if k in ALLOWED_KEYS} for p in pushers + ) + + return 200, {"pushers": filtered_pushers} def on_OPTIONS(self, _): return 200, {} diff --git a/synapse/storage/data_stores/main/pusher.py b/synapse/storage/data_stores/main/pusher.py index f07309ef09..6b03233262 100644 --- a/synapse/storage/data_stores/main/pusher.py +++ b/synapse/storage/data_stores/main/pusher.py @@ -15,8 +15,7 @@ # limitations under the License. import logging - -import six +from typing import Iterable, Iterator from canonicaljson import encode_canonical_json, json @@ -27,21 +26,16 @@ from synapse.util.caches.descriptors import cachedInlineCallbacks, cachedList logger = logging.getLogger(__name__) -if six.PY2: - db_binary_type = six.moves.builtins.buffer -else: - db_binary_type = memoryview - class PusherWorkerStore(SQLBaseStore): - def _decode_pushers_rows(self, rows): + def _decode_pushers_rows(self, rows: Iterable[dict]) -> Iterator[dict]: + """JSON-decode the data in the rows returned from the `pushers` table + + Drops any rows whose data cannot be decoded + """ for r in rows: dataJson = r["data"] - r["data"] = None try: - if isinstance(dataJson, db_binary_type): - dataJson = str(dataJson).decode("UTF8") - r["data"] = json.loads(dataJson) except Exception as e: logger.warning( @@ -50,12 +44,9 @@ class PusherWorkerStore(SQLBaseStore): dataJson, e.args[0], ) - pass - - if isinstance(r["pushkey"], db_binary_type): - r["pushkey"] = str(r["pushkey"]).decode("UTF8") + continue - return rows + yield r @defer.inlineCallbacks def user_has_pusher(self, user_id): diff --git a/tests/push/test_email.py b/tests/push/test_email.py index 358b593cd4..80187406bc 100644 --- a/tests/push/test_email.py +++ b/tests/push/test_email.py @@ -165,6 +165,7 @@ class EmailPusherTests(HomeserverTestCase): pushers = self.get_success( self.hs.get_datastore().get_pushers_by(dict(user_name=self.user_id)) ) + pushers = list(pushers) self.assertEqual(len(pushers), 1) last_stream_ordering = pushers[0]["last_stream_ordering"] @@ -175,6 +176,7 @@ class EmailPusherTests(HomeserverTestCase): pushers = self.get_success( self.hs.get_datastore().get_pushers_by(dict(user_name=self.user_id)) ) + pushers = list(pushers) self.assertEqual(len(pushers), 1) self.assertEqual(last_stream_ordering, pushers[0]["last_stream_ordering"]) @@ -192,5 +194,6 @@ class EmailPusherTests(HomeserverTestCase): pushers = self.get_success( self.hs.get_datastore().get_pushers_by(dict(user_name=self.user_id)) ) + pushers = list(pushers) self.assertEqual(len(pushers), 1) self.assertTrue(pushers[0]["last_stream_ordering"] > last_stream_ordering) diff --git a/tests/push/test_http.py b/tests/push/test_http.py index af2327fb66..fe3441f081 100644 --- a/tests/push/test_http.py +++ b/tests/push/test_http.py @@ -104,6 +104,7 @@ class HTTPPusherTests(HomeserverTestCase): pushers = self.get_success( self.hs.get_datastore().get_pushers_by(dict(user_name=user_id)) ) + pushers = list(pushers) self.assertEqual(len(pushers), 1) last_stream_ordering = pushers[0]["last_stream_ordering"] @@ -114,6 +115,7 @@ class HTTPPusherTests(HomeserverTestCase): pushers = self.get_success( self.hs.get_datastore().get_pushers_by(dict(user_name=user_id)) ) + pushers = list(pushers) self.assertEqual(len(pushers), 1) self.assertEqual(last_stream_ordering, pushers[0]["last_stream_ordering"]) @@ -132,6 +134,7 @@ class HTTPPusherTests(HomeserverTestCase): pushers = self.get_success( self.hs.get_datastore().get_pushers_by(dict(user_name=user_id)) ) + pushers = list(pushers) self.assertEqual(len(pushers), 1) self.assertTrue(pushers[0]["last_stream_ordering"] > last_stream_ordering) last_stream_ordering = pushers[0]["last_stream_ordering"] @@ -151,5 +154,6 @@ class HTTPPusherTests(HomeserverTestCase): pushers = self.get_success( self.hs.get_datastore().get_pushers_by(dict(user_name=user_id)) ) + pushers = list(pushers) self.assertEqual(len(pushers), 1) self.assertTrue(pushers[0]["last_stream_ordering"] > last_stream_ordering) -- cgit 1.4.1 From fa780e9721c940479a72eed9877ccad4fef78160 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Fri, 20 Dec 2019 10:32:02 +0000 Subject: Change EventContext to use the Storage class (#6564) --- changelog.d/6564.misc | 1 + synapse/api/auth.py | 2 +- synapse/events/snapshot.py | 36 +++++++++++++++----------- synapse/events/third_party_rules.py | 2 +- synapse/handlers/_base.py | 2 +- synapse/handlers/federation.py | 14 +++++----- synapse/handlers/message.py | 10 +++---- synapse/handlers/room.py | 2 +- synapse/handlers/room_member.py | 4 +-- synapse/push/bulk_push_rule_evaluator.py | 4 +-- synapse/replication/http/federation.py | 5 +++- synapse/replication/http/send_event.py | 3 ++- synapse/storage/data_stores/main/push_rule.py | 2 +- synapse/storage/data_stores/main/roommember.py | 2 +- tests/test_state.py | 28 ++++++++++---------- 15 files changed, 64 insertions(+), 53 deletions(-) create mode 100644 changelog.d/6564.misc (limited to 'tests') diff --git a/changelog.d/6564.misc b/changelog.d/6564.misc new file mode 100644 index 0000000000..f644f5868b --- /dev/null +++ b/changelog.d/6564.misc @@ -0,0 +1 @@ +Change `EventContext` to use the `Storage` class, in preparation for moving state database queries to a separate data store. diff --git a/synapse/api/auth.py b/synapse/api/auth.py index 9fd52a8c77..abbc7079a3 100644 --- a/synapse/api/auth.py +++ b/synapse/api/auth.py @@ -79,7 +79,7 @@ class Auth(object): @defer.inlineCallbacks def check_from_context(self, room_version, event, context, do_sig_check=True): - prev_state_ids = yield context.get_prev_state_ids(self.store) + prev_state_ids = yield context.get_prev_state_ids() auth_events_ids = yield self.compute_auth_events( event, prev_state_ids, for_verification=True ) diff --git a/synapse/events/snapshot.py b/synapse/events/snapshot.py index 64e898f40c..a44baea365 100644 --- a/synapse/events/snapshot.py +++ b/synapse/events/snapshot.py @@ -149,7 +149,7 @@ class EventContext: # the prev_state_ids, so if we're a state event we include the event # id that we replaced in the state. if event.is_state(): - prev_state_ids = yield self.get_prev_state_ids(store) + prev_state_ids = yield self.get_prev_state_ids() prev_state_id = prev_state_ids.get((event.type, event.state_key)) else: prev_state_id = None @@ -167,12 +167,13 @@ class EventContext: } @staticmethod - def deserialize(store, input): + def deserialize(storage, input): """Converts a dict that was produced by `serialize` back into a EventContext. Args: - store (DataStore): Used to convert AS ID to AS object + storage (Storage): Used to convert AS ID to AS object and fetch + state. input (dict): A dict produced by `serialize` Returns: @@ -181,6 +182,7 @@ class EventContext: context = _AsyncEventContextImpl( # We use the state_group and prev_state_id stuff to pull the # current_state_ids out of the DB and construct prev_state_ids. + storage=storage, prev_state_id=input["prev_state_id"], event_type=input["event_type"], event_state_key=input["event_state_key"], @@ -193,7 +195,7 @@ class EventContext: app_service_id = input["app_service_id"] if app_service_id: - context.app_service = store.get_app_service_by_id(app_service_id) + context.app_service = storage.main.get_app_service_by_id(app_service_id) return context @@ -216,7 +218,7 @@ class EventContext: return self._state_group @defer.inlineCallbacks - def get_current_state_ids(self, store): + def get_current_state_ids(self): """ Gets the room state map, including this event - ie, the state in ``state_group`` @@ -234,11 +236,11 @@ class EventContext: if self.rejected: raise RuntimeError("Attempt to access state_ids of rejected event") - yield self._ensure_fetched(store) + yield self._ensure_fetched() return self._current_state_ids @defer.inlineCallbacks - def get_prev_state_ids(self, store): + def get_prev_state_ids(self): """ Gets the room state map, excluding this event. @@ -250,7 +252,7 @@ class EventContext: Maps a (type, state_key) to the event ID of the state event matching this tuple. """ - yield self._ensure_fetched(store) + yield self._ensure_fetched() return self._prev_state_ids def get_cached_current_state_ids(self): @@ -270,7 +272,7 @@ class EventContext: return self._current_state_ids - def _ensure_fetched(self, store): + def _ensure_fetched(self): return defer.succeed(None) @@ -282,6 +284,8 @@ class _AsyncEventContextImpl(EventContext): Attributes: + _storage (Storage) + _fetching_state_deferred (Deferred|None): Resolves when *_state_ids have been calculated. None if we haven't started calculating yet @@ -295,28 +299,30 @@ class _AsyncEventContextImpl(EventContext): that was replaced. """ + # This needs to have a default as we're inheriting + _storage = attr.ib(default=None) _prev_state_id = attr.ib(default=None) _event_type = attr.ib(default=None) _event_state_key = attr.ib(default=None) _fetching_state_deferred = attr.ib(default=None) - def _ensure_fetched(self, store): + def _ensure_fetched(self): if not self._fetching_state_deferred: - self._fetching_state_deferred = run_in_background( - self._fill_out_state, store - ) + self._fetching_state_deferred = run_in_background(self._fill_out_state) return make_deferred_yieldable(self._fetching_state_deferred) @defer.inlineCallbacks - def _fill_out_state(self, store): + def _fill_out_state(self): """Called to populate the _current_state_ids and _prev_state_ids attributes by loading from the database. """ if self.state_group is None: return - self._current_state_ids = yield store.get_state_ids_for_group(self.state_group) + self._current_state_ids = yield self._storage.state.get_state_ids_for_group( + self.state_group + ) if self._prev_state_id and self._event_state_key is not None: self._prev_state_ids = dict(self._current_state_ids) diff --git a/synapse/events/third_party_rules.py b/synapse/events/third_party_rules.py index 714a9b1579..86f7e5f8aa 100644 --- a/synapse/events/third_party_rules.py +++ b/synapse/events/third_party_rules.py @@ -53,7 +53,7 @@ class ThirdPartyEventRules(object): if self.third_party_rules is None: return True - prev_state_ids = yield context.get_prev_state_ids(self.store) + prev_state_ids = yield context.get_prev_state_ids() # Retrieve the state events from the database. state_events = {} diff --git a/synapse/handlers/_base.py b/synapse/handlers/_base.py index d15c6282fb..51413d910e 100644 --- a/synapse/handlers/_base.py +++ b/synapse/handlers/_base.py @@ -134,7 +134,7 @@ class BaseHandler(object): guest_access = event.content.get("guest_access", "forbidden") if guest_access != "can_join": if context: - current_state_ids = yield context.get_current_state_ids(self.store) + current_state_ids = yield context.get_current_state_ids() current_state = yield self.store.get_events( list(current_state_ids.values()) ) diff --git a/synapse/handlers/federation.py b/synapse/handlers/federation.py index 60bb00fc6a..05ae40dde7 100644 --- a/synapse/handlers/federation.py +++ b/synapse/handlers/federation.py @@ -718,7 +718,7 @@ class FederationHandler(BaseHandler): # changing their profile info. newly_joined = True - prev_state_ids = await context.get_prev_state_ids(self.store) + prev_state_ids = await context.get_prev_state_ids() prev_state_id = prev_state_ids.get((event.type, event.state_key)) if prev_state_id: @@ -1418,7 +1418,7 @@ class FederationHandler(BaseHandler): user = UserID.from_string(event.state_key) yield self.user_joined_room(user, event.room_id) - prev_state_ids = yield context.get_prev_state_ids(self.store) + prev_state_ids = yield context.get_prev_state_ids() state_ids = list(prev_state_ids.values()) auth_chain = yield self.store.get_auth_chain(state_ids) @@ -1927,7 +1927,7 @@ class FederationHandler(BaseHandler): context = yield self.state_handler.compute_event_context(event, old_state=state) if not auth_events: - prev_state_ids = yield context.get_prev_state_ids(self.store) + prev_state_ids = yield context.get_prev_state_ids() auth_events_ids = yield self.auth.compute_auth_events( event, prev_state_ids, for_verification=True ) @@ -2336,12 +2336,12 @@ class FederationHandler(BaseHandler): k: a.event_id for k, a in iteritems(auth_events) if k != event_key } - current_state_ids = yield context.get_current_state_ids(self.store) + current_state_ids = yield context.get_current_state_ids() current_state_ids = dict(current_state_ids) current_state_ids.update(state_updates) - prev_state_ids = yield context.get_prev_state_ids(self.store) + prev_state_ids = yield context.get_prev_state_ids() prev_state_ids = dict(prev_state_ids) prev_state_ids.update({k: a.event_id for k, a in iteritems(auth_events)}) @@ -2625,7 +2625,7 @@ class FederationHandler(BaseHandler): event.content["third_party_invite"]["signed"]["token"], ) original_invite = None - prev_state_ids = yield context.get_prev_state_ids(self.store) + prev_state_ids = yield context.get_prev_state_ids() original_invite_id = prev_state_ids.get(key) if original_invite_id: original_invite = yield self.store.get_event( @@ -2673,7 +2673,7 @@ class FederationHandler(BaseHandler): signed = event.content["third_party_invite"]["signed"] token = signed["token"] - prev_state_ids = yield context.get_prev_state_ids(self.store) + prev_state_ids = yield context.get_prev_state_ids() invite_event_id = prev_state_ids.get((EventTypes.ThirdPartyInvite, token)) invite_event = None diff --git a/synapse/handlers/message.py b/synapse/handlers/message.py index bf9add7fe2..4ad752205f 100644 --- a/synapse/handlers/message.py +++ b/synapse/handlers/message.py @@ -515,7 +515,7 @@ class EventCreationHandler(object): # federation as well as those created locally. As of room v3, aliases events # can be created by users that are not in the room, therefore we have to # tolerate them in event_auth.check(). - prev_state_ids = yield context.get_prev_state_ids(self.store) + prev_state_ids = yield context.get_prev_state_ids() prev_event_id = prev_state_ids.get((EventTypes.Member, event.sender)) prev_event = ( yield self.store.get_event(prev_event_id, allow_none=True) @@ -665,7 +665,7 @@ class EventCreationHandler(object): If so, returns the version of the event in context. Otherwise, returns None. """ - prev_state_ids = yield context.get_prev_state_ids(self.store) + prev_state_ids = yield context.get_prev_state_ids() prev_event_id = prev_state_ids.get((event.type, event.state_key)) if not prev_event_id: return @@ -914,7 +914,7 @@ class EventCreationHandler(object): def is_inviter_member_event(e): return e.type == EventTypes.Member and e.sender == event.sender - current_state_ids = yield context.get_current_state_ids(self.store) + current_state_ids = yield context.get_current_state_ids() state_to_include_ids = [ e_id @@ -967,7 +967,7 @@ class EventCreationHandler(object): if original_event.room_id != event.room_id: raise SynapseError(400, "Cannot redact event from a different room") - prev_state_ids = yield context.get_prev_state_ids(self.store) + prev_state_ids = yield context.get_prev_state_ids() auth_events_ids = yield self.auth.compute_auth_events( event, prev_state_ids, for_verification=True ) @@ -989,7 +989,7 @@ class EventCreationHandler(object): event.internal_metadata.recheck_redaction = False if event.type == EventTypes.Create: - prev_state_ids = yield context.get_prev_state_ids(self.store) + prev_state_ids = yield context.get_prev_state_ids() if prev_state_ids: raise AuthError(403, "Changing the room create event is forbidden") diff --git a/synapse/handlers/room.py b/synapse/handlers/room.py index d3a1a7b4a6..89c9118b26 100644 --- a/synapse/handlers/room.py +++ b/synapse/handlers/room.py @@ -184,7 +184,7 @@ class RoomCreationHandler(BaseHandler): requester, tombstone_event, tombstone_context ) - old_room_state = yield tombstone_context.get_current_state_ids(self.store) + old_room_state = yield tombstone_context.get_current_state_ids() # update any aliases yield self._move_aliases_to_new_room( diff --git a/synapse/handlers/room_member.py b/synapse/handlers/room_member.py index 7b7270fc61..44c5e3239c 100644 --- a/synapse/handlers/room_member.py +++ b/synapse/handlers/room_member.py @@ -193,7 +193,7 @@ class RoomMemberHandler(object): requester, event, context, extra_users=[target], ratelimit=ratelimit ) - prev_state_ids = yield context.get_prev_state_ids(self.store) + prev_state_ids = yield context.get_prev_state_ids() prev_member_event_id = prev_state_ids.get((EventTypes.Member, user_id), None) @@ -601,7 +601,7 @@ class RoomMemberHandler(object): if prev_event is not None: return - prev_state_ids = yield context.get_prev_state_ids(self.store) + prev_state_ids = yield context.get_prev_state_ids() if event.membership == Membership.JOIN: if requester.is_guest: guest_can_join = yield self._can_guest_join(prev_state_ids) diff --git a/synapse/push/bulk_push_rule_evaluator.py b/synapse/push/bulk_push_rule_evaluator.py index 7881780760..7d9f5a38d9 100644 --- a/synapse/push/bulk_push_rule_evaluator.py +++ b/synapse/push/bulk_push_rule_evaluator.py @@ -116,7 +116,7 @@ class BulkPushRuleEvaluator(object): @defer.inlineCallbacks def _get_power_levels_and_sender_level(self, event, context): - prev_state_ids = yield context.get_prev_state_ids(self.store) + prev_state_ids = yield context.get_prev_state_ids() pl_event_id = prev_state_ids.get(POWER_KEY) if pl_event_id: # fastpath: if there's a power level event, that's all we need, and @@ -304,7 +304,7 @@ class RulesForRoom(object): push_rules_delta_state_cache_metric.inc_hits() else: - current_state_ids = yield context.get_current_state_ids(self.store) + current_state_ids = yield context.get_current_state_ids() push_rules_delta_state_cache_metric.inc_misses() push_rules_state_size_counter.inc(len(current_state_ids)) diff --git a/synapse/replication/http/federation.py b/synapse/replication/http/federation.py index 9af4e7e173..49a3251372 100644 --- a/synapse/replication/http/federation.py +++ b/synapse/replication/http/federation.py @@ -51,6 +51,7 @@ class ReplicationFederationSendEventsRestServlet(ReplicationEndpoint): super(ReplicationFederationSendEventsRestServlet, self).__init__(hs) self.store = hs.get_datastore() + self.storage = hs.get_storage() self.clock = hs.get_clock() self.federation_handler = hs.get_handlers().federation_handler @@ -100,7 +101,9 @@ class ReplicationFederationSendEventsRestServlet(ReplicationEndpoint): EventType = event_type_from_format_version(format_ver) event = EventType(event_dict, internal_metadata, rejected_reason) - context = EventContext.deserialize(self.store, event_payload["context"]) + context = EventContext.deserialize( + self.storage, event_payload["context"] + ) event_and_contexts.append((event, context)) diff --git a/synapse/replication/http/send_event.py b/synapse/replication/http/send_event.py index 9bafd60b14..84b92f16ad 100644 --- a/synapse/replication/http/send_event.py +++ b/synapse/replication/http/send_event.py @@ -54,6 +54,7 @@ class ReplicationSendEventRestServlet(ReplicationEndpoint): self.event_creation_handler = hs.get_event_creation_handler() self.store = hs.get_datastore() + self.storage = hs.get_storage() self.clock = hs.get_clock() @staticmethod @@ -100,7 +101,7 @@ class ReplicationSendEventRestServlet(ReplicationEndpoint): event = EventType(event_dict, internal_metadata, rejected_reason) requester = Requester.deserialize(self.store, content["requester"]) - context = EventContext.deserialize(self.store, content["context"]) + context = EventContext.deserialize(self.storage, content["context"]) ratelimit = content["ratelimit"] extra_users = [UserID.from_string(u) for u in content["extra_users"]] diff --git a/synapse/storage/data_stores/main/push_rule.py b/synapse/storage/data_stores/main/push_rule.py index 5ba13aa973..e2673ae073 100644 --- a/synapse/storage/data_stores/main/push_rule.py +++ b/synapse/storage/data_stores/main/push_rule.py @@ -244,7 +244,7 @@ class PushRulesWorkerStore( # To do this we set the state_group to a new object as object() != object() state_group = object() - current_state_ids = yield context.get_current_state_ids(self) + current_state_ids = yield context.get_current_state_ids() result = yield self._bulk_get_push_rules_for_room( event.room_id, state_group, current_state_ids, event=event ) diff --git a/synapse/storage/data_stores/main/roommember.py b/synapse/storage/data_stores/main/roommember.py index 92e3b9c512..70ff5751b6 100644 --- a/synapse/storage/data_stores/main/roommember.py +++ b/synapse/storage/data_stores/main/roommember.py @@ -477,7 +477,7 @@ class RoomMemberWorkerStore(EventsWorkerStore): # To do this we set the state_group to a new object as object() != object() state_group = object() - current_state_ids = yield context.get_current_state_ids(self) + current_state_ids = yield context.get_current_state_ids() result = yield self._get_joined_users_from_context( event.room_id, state_group, current_state_ids, event=event, context=context ) diff --git a/tests/test_state.py b/tests/test_state.py index 176535947a..e0aae06be4 100644 --- a/tests/test_state.py +++ b/tests/test_state.py @@ -209,7 +209,7 @@ class StateTestCase(unittest.TestCase): ctx_c = context_store["C"] ctx_d = context_store["D"] - prev_state_ids = yield ctx_d.get_prev_state_ids(self.store) + prev_state_ids = yield ctx_d.get_prev_state_ids() self.assertEqual(2, len(prev_state_ids)) self.assertEqual(ctx_c.state_group, ctx_d.state_group_before_event) @@ -253,7 +253,7 @@ class StateTestCase(unittest.TestCase): ctx_c = context_store["C"] ctx_d = context_store["D"] - prev_state_ids = yield ctx_d.get_prev_state_ids(self.store) + prev_state_ids = yield ctx_d.get_prev_state_ids() self.assertSetEqual( {"START", "A", "C"}, {e_id for e_id in prev_state_ids.values()} ) @@ -312,7 +312,7 @@ class StateTestCase(unittest.TestCase): ctx_c = context_store["C"] ctx_e = context_store["E"] - prev_state_ids = yield ctx_e.get_prev_state_ids(self.store) + prev_state_ids = yield ctx_e.get_prev_state_ids() self.assertSetEqual( {"START", "A", "B", "C"}, {e for e in prev_state_ids.values()} ) @@ -387,7 +387,7 @@ class StateTestCase(unittest.TestCase): ctx_b = context_store["B"] ctx_d = context_store["D"] - prev_state_ids = yield ctx_d.get_prev_state_ids(self.store) + prev_state_ids = yield ctx_d.get_prev_state_ids() self.assertSetEqual( {"A1", "A2", "A3", "A5", "B"}, {e for e in prev_state_ids.values()} ) @@ -419,10 +419,10 @@ class StateTestCase(unittest.TestCase): context = yield self.state.compute_event_context(event, old_state=old_state) - prev_state_ids = yield context.get_prev_state_ids(self.store) + prev_state_ids = yield context.get_prev_state_ids() self.assertCountEqual((e.event_id for e in old_state), prev_state_ids.values()) - current_state_ids = yield context.get_current_state_ids(self.store) + current_state_ids = yield context.get_current_state_ids() self.assertCountEqual( (e.event_id for e in old_state), current_state_ids.values() ) @@ -442,10 +442,10 @@ class StateTestCase(unittest.TestCase): context = yield self.state.compute_event_context(event, old_state=old_state) - prev_state_ids = yield context.get_prev_state_ids(self.store) + prev_state_ids = yield context.get_prev_state_ids() self.assertCountEqual((e.event_id for e in old_state), prev_state_ids.values()) - current_state_ids = yield context.get_current_state_ids(self.store) + current_state_ids = yield context.get_current_state_ids() self.assertCountEqual( (e.event_id for e in old_state + [event]), current_state_ids.values() ) @@ -479,7 +479,7 @@ class StateTestCase(unittest.TestCase): context = yield self.state.compute_event_context(event) - current_state_ids = yield context.get_current_state_ids(self.store) + current_state_ids = yield context.get_current_state_ids() self.assertEqual( set([e.event_id for e in old_state]), set(current_state_ids.values()) @@ -511,7 +511,7 @@ class StateTestCase(unittest.TestCase): context = yield self.state.compute_event_context(event) - prev_state_ids = yield context.get_prev_state_ids(self.store) + prev_state_ids = yield context.get_prev_state_ids() self.assertEqual( set([e.event_id for e in old_state]), set(prev_state_ids.values()) @@ -552,7 +552,7 @@ class StateTestCase(unittest.TestCase): event, prev_event_id1, old_state_1, prev_event_id2, old_state_2 ) - current_state_ids = yield context.get_current_state_ids(self.store) + current_state_ids = yield context.get_current_state_ids() self.assertEqual(len(current_state_ids), 6) @@ -594,7 +594,7 @@ class StateTestCase(unittest.TestCase): event, prev_event_id1, old_state_1, prev_event_id2, old_state_2 ) - current_state_ids = yield context.get_current_state_ids(self.store) + current_state_ids = yield context.get_current_state_ids() self.assertEqual(len(current_state_ids), 6) @@ -649,7 +649,7 @@ class StateTestCase(unittest.TestCase): event, prev_event_id1, old_state_1, prev_event_id2, old_state_2 ) - current_state_ids = yield context.get_current_state_ids(self.store) + current_state_ids = yield context.get_current_state_ids() self.assertEqual(old_state_2[3].event_id, current_state_ids[("test1", "1")]) @@ -677,7 +677,7 @@ class StateTestCase(unittest.TestCase): event, prev_event_id1, old_state_1, prev_event_id2, old_state_2 ) - current_state_ids = yield context.get_current_state_ids(self.store) + current_state_ids = yield context.get_current_state_ids() self.assertEqual(old_state_1[3].event_id, current_state_ids[("test1", "1")]) -- cgit 1.4.1 From 75d8f26ac85efd3816d454927f40b6e4c3032df1 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Fri, 20 Dec 2019 10:48:24 +0000 Subject: Split state groups into a separate data store (#6296) --- changelog.d/6245.misc | 1 + scripts/synapse_port_db | 8 +- synapse/config/database.py | 10 +- synapse/storage/data_stores/__init__.py | 5 + synapse/storage/data_stores/main/events.py | 157 ---- .../main/schema/delta/23/drop_state_index.sql | 16 - .../main/schema/delta/30/state_stream.sql | 33 - .../main/schema/delta/32/remove_indices.sql | 1 - .../main/schema/delta/35/add_state_index.sql | 17 - .../data_stores/main/schema/delta/35/state.sql | 22 - .../main/schema/delta/35/state_dedupe.sql | 17 - .../main/schema/delta/47/state_group_seq.py | 34 - .../main/schema/full_schemas/54/full.sql.postgres | 52 -- .../main/schema/full_schemas/54/full.sql.sqlite | 6 - synapse/storage/data_stores/main/state.py | 937 +-------------------- synapse/storage/data_stores/state/__init__.py | 16 + synapse/storage/data_stores/state/bg_updates.py | 374 ++++++++ .../state/schema/delta/23/drop_state_index.sql | 16 + .../state/schema/delta/30/state_stream.sql | 33 + .../state/schema/delta/32/remove_state_indices.sql | 19 + .../state/schema/delta/35/add_state_index.sql | 17 + .../data_stores/state/schema/delta/35/state.sql | 22 + .../state/schema/delta/35/state_dedupe.sql | 17 + .../state/schema/delta/47/state_group_seq.py | 34 + .../state/schema/delta/56/state_group_room_idx.sql | 17 + .../state/schema/full_schemas/54/full.sql | 37 + .../schema/full_schemas/54/sequence.sql.postgres | 21 + synapse/storage/data_stores/state/store.py | 640 ++++++++++++++ synapse/storage/persist_events.py | 2 +- synapse/storage/prepare_database.py | 2 +- synapse/storage/purge_events.py | 4 +- synapse/storage/state.py | 14 +- tests/storage/test_state.py | 2 +- tests/utils.py | 2 +- 34 files changed, 1298 insertions(+), 1307 deletions(-) create mode 100644 changelog.d/6245.misc delete mode 100644 synapse/storage/data_stores/main/schema/delta/23/drop_state_index.sql delete mode 100644 synapse/storage/data_stores/main/schema/delta/30/state_stream.sql delete mode 100644 synapse/storage/data_stores/main/schema/delta/35/add_state_index.sql delete mode 100644 synapse/storage/data_stores/main/schema/delta/35/state.sql delete mode 100644 synapse/storage/data_stores/main/schema/delta/35/state_dedupe.sql delete mode 100644 synapse/storage/data_stores/main/schema/delta/47/state_group_seq.py create mode 100644 synapse/storage/data_stores/state/__init__.py create mode 100644 synapse/storage/data_stores/state/bg_updates.py create mode 100644 synapse/storage/data_stores/state/schema/delta/23/drop_state_index.sql create mode 100644 synapse/storage/data_stores/state/schema/delta/30/state_stream.sql create mode 100644 synapse/storage/data_stores/state/schema/delta/32/remove_state_indices.sql create mode 100644 synapse/storage/data_stores/state/schema/delta/35/add_state_index.sql create mode 100644 synapse/storage/data_stores/state/schema/delta/35/state.sql create mode 100644 synapse/storage/data_stores/state/schema/delta/35/state_dedupe.sql create mode 100644 synapse/storage/data_stores/state/schema/delta/47/state_group_seq.py create mode 100644 synapse/storage/data_stores/state/schema/delta/56/state_group_room_idx.sql create mode 100644 synapse/storage/data_stores/state/schema/full_schemas/54/full.sql create mode 100644 synapse/storage/data_stores/state/schema/full_schemas/54/sequence.sql.postgres create mode 100644 synapse/storage/data_stores/state/store.py (limited to 'tests') diff --git a/changelog.d/6245.misc b/changelog.d/6245.misc new file mode 100644 index 0000000000..a3e6b8296e --- /dev/null +++ b/changelog.d/6245.misc @@ -0,0 +1 @@ +Split out state storage into separate data store. diff --git a/scripts/synapse_port_db b/scripts/synapse_port_db index 5b5368988c..eb927f2094 100755 --- a/scripts/synapse_port_db +++ b/scripts/synapse_port_db @@ -51,11 +51,12 @@ from synapse.storage.data_stores.main.registration import ( from synapse.storage.data_stores.main.room import RoomBackgroundUpdateStore from synapse.storage.data_stores.main.roommember import RoomMemberBackgroundUpdateStore from synapse.storage.data_stores.main.search import SearchBackgroundUpdateStore -from synapse.storage.data_stores.main.state import StateBackgroundUpdateStore +from synapse.storage.data_stores.main.state import MainStateBackgroundUpdateStore from synapse.storage.data_stores.main.stats import StatsStore from synapse.storage.data_stores.main.user_directory import ( UserDirectoryBackgroundUpdateStore, ) +from synapse.storage.data_stores.state.bg_updates import StateBackgroundUpdateStore from synapse.storage.database import Database, make_conn from synapse.storage.engines import create_engine from synapse.storage.prepare_database import prepare_database @@ -138,6 +139,7 @@ class Store( RoomMemberBackgroundUpdateStore, SearchBackgroundUpdateStore, StateBackgroundUpdateStore, + MainStateBackgroundUpdateStore, UserDirectoryBackgroundUpdateStore, StatsStore, ): @@ -496,9 +498,7 @@ class Porter(object): def run(self): try: self.sqlite_store = yield self.build_db_store( - DatabaseConnectionConfig( - "master", self.sqlite_config, data_stores=["main"] - ) + DatabaseConnectionConfig("master-sqlite", self.sqlite_config) ) # Check if all background updates are done, abort if not. diff --git a/synapse/config/database.py b/synapse/config/database.py index 5f2f3c7cfd..134824789c 100644 --- a/synapse/config/database.py +++ b/synapse/config/database.py @@ -34,10 +34,12 @@ class DatabaseConnectionConfig: module name, and `args` for the args to give to the database connector. data_stores: The list of data stores that should be provisioned on the - database. + database. Defaults to all data stores. """ - def __init__(self, name: str, db_config: dict, data_stores: List[str]): + def __init__( + self, name: str, db_config: dict, data_stores: List[str] = ["main", "state"] + ): if db_config["name"] not in ("sqlite3", "psycopg2"): raise ConfigError("Unsupported database type %r" % (db_config["name"],)) @@ -62,9 +64,7 @@ class DatabaseConfig(Config): if database_config is None: database_config = {"name": "sqlite3", "args": {}} - self.databases = [ - DatabaseConnectionConfig("master", database_config, data_stores=["main"]) - ] + self.databases = [DatabaseConnectionConfig("master", database_config)] self.set_databasepath(config.get("database_path")) diff --git a/synapse/storage/data_stores/__init__.py b/synapse/storage/data_stores/__init__.py index 0983e059c0..d20df5f076 100644 --- a/synapse/storage/data_stores/__init__.py +++ b/synapse/storage/data_stores/__init__.py @@ -15,6 +15,7 @@ import logging +from synapse.storage.data_stores.state import StateGroupDataStore from synapse.storage.database import Database, make_conn from synapse.storage.engines import create_engine from synapse.storage.prepare_database import prepare_database @@ -55,6 +56,10 @@ class DataStores(object): logger.info("Starting 'main' data store") self.main = main_store_class(database, db_conn, hs) + if "state" in database_config.data_stores: + logger.info("Starting 'state' data store") + self.state = StateGroupDataStore(database, db_conn, hs) + db_conn.commit() self.databases.append(database) diff --git a/synapse/storage/data_stores/main/events.py b/synapse/storage/data_stores/main/events.py index 998bba1aad..58f35d7f56 100644 --- a/synapse/storage/data_stores/main/events.py +++ b/synapse/storage/data_stores/main/events.py @@ -1757,163 +1757,6 @@ class EventsStore( return state_groups - def purge_unreferenced_state_groups( - self, room_id: str, state_groups_to_delete - ) -> defer.Deferred: - """Deletes no longer referenced state groups and de-deltas any state - groups that reference them. - - Args: - room_id: The room the state groups belong to (must all be in the - same room). - state_groups_to_delete (Collection[int]): Set of all state groups - to delete. - """ - - return self.db.runInteraction( - "purge_unreferenced_state_groups", - self._purge_unreferenced_state_groups, - room_id, - state_groups_to_delete, - ) - - def _purge_unreferenced_state_groups(self, txn, room_id, state_groups_to_delete): - logger.info( - "[purge] found %i state groups to delete", len(state_groups_to_delete) - ) - - rows = self.db.simple_select_many_txn( - txn, - table="state_group_edges", - column="prev_state_group", - iterable=state_groups_to_delete, - keyvalues={}, - retcols=("state_group",), - ) - - remaining_state_groups = set( - row["state_group"] - for row in rows - if row["state_group"] not in state_groups_to_delete - ) - - logger.info( - "[purge] de-delta-ing %i remaining state groups", - len(remaining_state_groups), - ) - - # Now we turn the state groups that reference to-be-deleted state - # groups to non delta versions. - for sg in remaining_state_groups: - logger.info("[purge] de-delta-ing remaining state group %s", sg) - curr_state = self._get_state_groups_from_groups_txn(txn, [sg]) - curr_state = curr_state[sg] - - self.db.simple_delete_txn( - txn, table="state_groups_state", keyvalues={"state_group": sg} - ) - - self.db.simple_delete_txn( - txn, table="state_group_edges", keyvalues={"state_group": sg} - ) - - self.db.simple_insert_many_txn( - txn, - table="state_groups_state", - values=[ - { - "state_group": sg, - "room_id": room_id, - "type": key[0], - "state_key": key[1], - "event_id": state_id, - } - for key, state_id in iteritems(curr_state) - ], - ) - - logger.info("[purge] removing redundant state groups") - txn.executemany( - "DELETE FROM state_groups_state WHERE state_group = ?", - ((sg,) for sg in state_groups_to_delete), - ) - txn.executemany( - "DELETE FROM state_groups WHERE id = ?", - ((sg,) for sg in state_groups_to_delete), - ) - - @defer.inlineCallbacks - def get_previous_state_groups(self, state_groups): - """Fetch the previous groups of the given state groups. - - Args: - state_groups (Iterable[int]) - - Returns: - Deferred[dict[int, int]]: mapping from state group to previous - state group. - """ - - rows = yield self.db.simple_select_many_batch( - table="state_group_edges", - column="prev_state_group", - iterable=state_groups, - keyvalues={}, - retcols=("prev_state_group", "state_group"), - desc="get_previous_state_groups", - ) - - return {row["state_group"]: row["prev_state_group"] for row in rows} - - def purge_room_state(self, room_id, state_groups_to_delete): - """Deletes all record of a room from state tables - - Args: - room_id (str): - state_groups_to_delete (list[int]): State groups to delete - """ - - return self.db.runInteraction( - "purge_room_state", - self._purge_room_state_txn, - room_id, - state_groups_to_delete, - ) - - def _purge_room_state_txn(self, txn, room_id, state_groups_to_delete): - # first we have to delete the state groups states - logger.info("[purge] removing %s from state_groups_state", room_id) - - self.db.simple_delete_many_txn( - txn, - table="state_groups_state", - column="state_group", - iterable=state_groups_to_delete, - keyvalues={}, - ) - - # ... and the state group edges - logger.info("[purge] removing %s from state_group_edges", room_id) - - self.db.simple_delete_many_txn( - txn, - table="state_group_edges", - column="state_group", - iterable=state_groups_to_delete, - keyvalues={}, - ) - - # ... and the state groups - logger.info("[purge] removing %s from state_groups", room_id) - - self.db.simple_delete_many_txn( - txn, - table="state_groups", - column="id", - iterable=state_groups_to_delete, - keyvalues={}, - ) - async def is_event_after(self, event_id1, event_id2): """Returns True if event_id1 is after event_id2 in the stream """ diff --git a/synapse/storage/data_stores/main/schema/delta/23/drop_state_index.sql b/synapse/storage/data_stores/main/schema/delta/23/drop_state_index.sql deleted file mode 100644 index ae09fa0065..0000000000 --- a/synapse/storage/data_stores/main/schema/delta/23/drop_state_index.sql +++ /dev/null @@ -1,16 +0,0 @@ -/* Copyright 2015, 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -DROP INDEX IF EXISTS state_groups_state_tuple; diff --git a/synapse/storage/data_stores/main/schema/delta/30/state_stream.sql b/synapse/storage/data_stores/main/schema/delta/30/state_stream.sql deleted file mode 100644 index e85699e82e..0000000000 --- a/synapse/storage/data_stores/main/schema/delta/30/state_stream.sql +++ /dev/null @@ -1,33 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - - -/* We used to create a table called current_state_resets, but this is no - * longer used and is removed in delta 54. - */ - -/* The outlier events that have aquired a state group typically through - * backfill. This is tracked separately to the events table, as assigning a - * state group change the position of the existing event in the stream - * ordering. - * However since a stream_ordering is assigned in persist_event for the - * (event, state) pair, we can use that stream_ordering to identify when - * the new state was assigned for the event. - */ -CREATE TABLE IF NOT EXISTS ex_outlier_stream( - event_stream_ordering BIGINT PRIMARY KEY NOT NULL, - event_id TEXT NOT NULL, - state_group BIGINT NOT NULL -); diff --git a/synapse/storage/data_stores/main/schema/delta/32/remove_indices.sql b/synapse/storage/data_stores/main/schema/delta/32/remove_indices.sql index 4219cdd06a..2de50d408c 100644 --- a/synapse/storage/data_stores/main/schema/delta/32/remove_indices.sql +++ b/synapse/storage/data_stores/main/schema/delta/32/remove_indices.sql @@ -20,7 +20,6 @@ DROP INDEX IF EXISTS events_room_id; -- Prefix of events_room_stream DROP INDEX IF EXISTS events_order; -- Prefix of events_order_topo_stream_room DROP INDEX IF EXISTS events_topological_ordering; -- Prefix of events_order_topo_stream_room DROP INDEX IF EXISTS events_stream_ordering; -- Duplicate of PRIMARY KEY -DROP INDEX IF EXISTS state_groups_id; -- Duplicate of PRIMARY KEY DROP INDEX IF EXISTS event_to_state_groups_id; -- Duplicate of PRIMARY KEY DROP INDEX IF EXISTS event_push_actions_room_id_event_id_user_id_profile_tag; -- Duplicate of UNIQUE CONSTRAINT diff --git a/synapse/storage/data_stores/main/schema/delta/35/add_state_index.sql b/synapse/storage/data_stores/main/schema/delta/35/add_state_index.sql deleted file mode 100644 index 33980d02f0..0000000000 --- a/synapse/storage/data_stores/main/schema/delta/35/add_state_index.sql +++ /dev/null @@ -1,17 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -INSERT into background_updates (update_name, progress_json, depends_on) - VALUES ('state_group_state_type_index', '{}', 'state_group_state_deduplication'); diff --git a/synapse/storage/data_stores/main/schema/delta/35/state.sql b/synapse/storage/data_stores/main/schema/delta/35/state.sql deleted file mode 100644 index 0f1fa68a89..0000000000 --- a/synapse/storage/data_stores/main/schema/delta/35/state.sql +++ /dev/null @@ -1,22 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -CREATE TABLE state_group_edges( - state_group BIGINT NOT NULL, - prev_state_group BIGINT NOT NULL -); - -CREATE INDEX state_group_edges_idx ON state_group_edges(state_group); -CREATE INDEX state_group_edges_prev_idx ON state_group_edges(prev_state_group); diff --git a/synapse/storage/data_stores/main/schema/delta/35/state_dedupe.sql b/synapse/storage/data_stores/main/schema/delta/35/state_dedupe.sql deleted file mode 100644 index 97e5067ef4..0000000000 --- a/synapse/storage/data_stores/main/schema/delta/35/state_dedupe.sql +++ /dev/null @@ -1,17 +0,0 @@ -/* Copyright 2016 OpenMarket Ltd - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -INSERT into background_updates (update_name, progress_json) - VALUES ('state_group_state_deduplication', '{}'); diff --git a/synapse/storage/data_stores/main/schema/delta/47/state_group_seq.py b/synapse/storage/data_stores/main/schema/delta/47/state_group_seq.py deleted file mode 100644 index 9fd1ccf6f7..0000000000 --- a/synapse/storage/data_stores/main/schema/delta/47/state_group_seq.py +++ /dev/null @@ -1,34 +0,0 @@ -# Copyright 2018 New Vector Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -from synapse.storage.engines import PostgresEngine - - -def run_create(cur, database_engine, *args, **kwargs): - if isinstance(database_engine, PostgresEngine): - # if we already have some state groups, we want to start making new - # ones with a higher id. - cur.execute("SELECT max(id) FROM state_groups") - row = cur.fetchone() - - if row[0] is None: - start_val = 1 - else: - start_val = row[0] + 1 - - cur.execute("CREATE SEQUENCE state_group_id_seq START WITH %s", (start_val,)) - - -def run_upgrade(*args, **kwargs): - pass diff --git a/synapse/storage/data_stores/main/schema/full_schemas/54/full.sql.postgres b/synapse/storage/data_stores/main/schema/full_schemas/54/full.sql.postgres index 4ad2929f32..889a9a0ce4 100644 --- a/synapse/storage/data_stores/main/schema/full_schemas/54/full.sql.postgres +++ b/synapse/storage/data_stores/main/schema/full_schemas/54/full.sql.postgres @@ -975,40 +975,6 @@ CREATE TABLE state_events ( -CREATE TABLE state_group_edges ( - state_group bigint NOT NULL, - prev_state_group bigint NOT NULL -); - - - -CREATE SEQUENCE state_group_id_seq - START WITH 1 - INCREMENT BY 1 - NO MINVALUE - NO MAXVALUE - CACHE 1; - - - -CREATE TABLE state_groups ( - id bigint NOT NULL, - room_id text NOT NULL, - event_id text NOT NULL -); - - - -CREATE TABLE state_groups_state ( - state_group bigint NOT NULL, - room_id text NOT NULL, - type text NOT NULL, - state_key text NOT NULL, - event_id text NOT NULL -); - - - CREATE TABLE stats_stream_pos ( lock character(1) DEFAULT 'X'::bpchar NOT NULL, stream_id bigint, @@ -1482,12 +1448,6 @@ ALTER TABLE ONLY state_events ADD CONSTRAINT state_events_event_id_key UNIQUE (event_id); - -ALTER TABLE ONLY state_groups - ADD CONSTRAINT state_groups_pkey PRIMARY KEY (id); - - - ALTER TABLE ONLY stats_stream_pos ADD CONSTRAINT stats_stream_pos_lock_key UNIQUE (lock); @@ -1928,18 +1888,6 @@ CREATE UNIQUE INDEX room_stats_room_ts ON room_stats USING btree (room_id, ts); -CREATE INDEX state_group_edges_idx ON state_group_edges USING btree (state_group); - - - -CREATE INDEX state_group_edges_prev_idx ON state_group_edges USING btree (prev_state_group); - - - -CREATE INDEX state_groups_state_type_idx ON state_groups_state USING btree (state_group, type, state_key); - - - CREATE INDEX stream_ordering_to_exterm_idx ON stream_ordering_to_exterm USING btree (stream_ordering); diff --git a/synapse/storage/data_stores/main/schema/full_schemas/54/full.sql.sqlite b/synapse/storage/data_stores/main/schema/full_schemas/54/full.sql.sqlite index bad33291e7..a0411ede7e 100644 --- a/synapse/storage/data_stores/main/schema/full_schemas/54/full.sql.sqlite +++ b/synapse/storage/data_stores/main/schema/full_schemas/54/full.sql.sqlite @@ -42,8 +42,6 @@ CREATE INDEX ev_edges_id ON event_edges(event_id); CREATE INDEX ev_edges_prev_id ON event_edges(prev_event_id); CREATE TABLE room_depth( room_id TEXT NOT NULL, min_depth INTEGER NOT NULL, UNIQUE (room_id) ); CREATE INDEX room_depth_room ON room_depth(room_id); -CREATE TABLE state_groups( id BIGINT PRIMARY KEY, room_id TEXT NOT NULL, event_id TEXT NOT NULL ); -CREATE TABLE state_groups_state( state_group BIGINT NOT NULL, room_id TEXT NOT NULL, type TEXT NOT NULL, state_key TEXT NOT NULL, event_id TEXT NOT NULL ); CREATE TABLE event_to_state_groups( event_id TEXT NOT NULL, state_group BIGINT NOT NULL, UNIQUE (event_id) ); CREATE TABLE local_media_repository ( media_id TEXT, media_type TEXT, media_length INTEGER, created_ts BIGINT, upload_name TEXT, user_id TEXT, quarantined_by TEXT, url_cache TEXT, last_access_ts BIGINT, UNIQUE (media_id) ); CREATE TABLE local_media_repository_thumbnails ( media_id TEXT, thumbnail_width INTEGER, thumbnail_height INTEGER, thumbnail_type TEXT, thumbnail_method TEXT, thumbnail_length INTEGER, UNIQUE ( media_id, thumbnail_width, thumbnail_height, thumbnail_type ) ); @@ -120,9 +118,6 @@ CREATE TABLE device_max_stream_id ( stream_id BIGINT NOT NULL ); CREATE TABLE public_room_list_stream ( stream_id BIGINT NOT NULL, room_id TEXT NOT NULL, visibility BOOLEAN NOT NULL , appservice_id TEXT, network_id TEXT); CREATE INDEX public_room_list_stream_idx on public_room_list_stream( stream_id ); CREATE INDEX public_room_list_stream_rm_idx on public_room_list_stream( room_id, stream_id ); -CREATE TABLE state_group_edges( state_group BIGINT NOT NULL, prev_state_group BIGINT NOT NULL ); -CREATE INDEX state_group_edges_idx ON state_group_edges(state_group); -CREATE INDEX state_group_edges_prev_idx ON state_group_edges(prev_state_group); CREATE TABLE stream_ordering_to_exterm ( stream_ordering BIGINT NOT NULL, room_id TEXT NOT NULL, event_id TEXT NOT NULL ); CREATE INDEX stream_ordering_to_exterm_idx on stream_ordering_to_exterm( stream_ordering ); CREATE INDEX stream_ordering_to_exterm_rm_idx on stream_ordering_to_exterm( room_id, stream_ordering ); @@ -254,6 +249,5 @@ CREATE INDEX user_ips_last_seen_only ON user_ips (last_seen); CREATE INDEX users_creation_ts ON users (creation_ts); CREATE INDEX event_to_state_groups_sg_index ON event_to_state_groups (state_group); CREATE UNIQUE INDEX device_lists_remote_cache_unique_id ON device_lists_remote_cache (user_id, device_id); -CREATE INDEX state_groups_state_type_idx ON state_groups_state(state_group, type, state_key); CREATE UNIQUE INDEX device_lists_remote_extremeties_unique_idx ON device_lists_remote_extremeties (user_id); CREATE UNIQUE INDEX user_ips_user_token_ip_unique_index ON user_ips (user_id, access_token, ip); diff --git a/synapse/storage/data_stores/main/state.py b/synapse/storage/data_stores/main/state.py index dcc6b43cdf..0dc39f139c 100644 --- a/synapse/storage/data_stores/main/state.py +++ b/synapse/storage/data_stores/main/state.py @@ -17,8 +17,7 @@ import logging from collections import namedtuple from typing import Iterable, Tuple -from six import iteritems, itervalues -from six.moves import range +from six import iteritems from twisted.internet import defer @@ -29,11 +28,9 @@ from synapse.events.snapshot import EventContext from synapse.storage._base import SQLBaseStore from synapse.storage.data_stores.main.events_worker import EventsWorkerStore from synapse.storage.database import Database -from synapse.storage.engines import PostgresEngine from synapse.storage.state import StateFilter -from synapse.util.caches import get_cache_factor_for, intern_string +from synapse.util.caches import intern_string from synapse.util.caches.descriptors import cached, cachedList -from synapse.util.caches.dictionary_cache import DictionaryCache from synapse.util.stringutils import to_ascii logger = logging.getLogger(__name__) @@ -55,207 +52,14 @@ class _GetStateGroupDelta( return len(self.delta_ids) if self.delta_ids else 0 -class StateGroupBackgroundUpdateStore(SQLBaseStore): - """Defines functions related to state groups needed to run the state backgroud - updates. - """ - - def _count_state_group_hops_txn(self, txn, state_group): - """Given a state group, count how many hops there are in the tree. - - This is used to ensure the delta chains don't get too long. - """ - if isinstance(self.database_engine, PostgresEngine): - sql = """ - WITH RECURSIVE state(state_group) AS ( - VALUES(?::bigint) - UNION ALL - SELECT prev_state_group FROM state_group_edges e, state s - WHERE s.state_group = e.state_group - ) - SELECT count(*) FROM state; - """ - - txn.execute(sql, (state_group,)) - row = txn.fetchone() - if row and row[0]: - return row[0] - else: - return 0 - else: - # We don't use WITH RECURSIVE on sqlite3 as there are distributions - # that ship with an sqlite3 version that doesn't support it (e.g. wheezy) - next_group = state_group - count = 0 - - while next_group: - next_group = self.db.simple_select_one_onecol_txn( - txn, - table="state_group_edges", - keyvalues={"state_group": next_group}, - retcol="prev_state_group", - allow_none=True, - ) - if next_group: - count += 1 - - return count - - def _get_state_groups_from_groups_txn( - self, txn, groups, state_filter=StateFilter.all() - ): - results = {group: {} for group in groups} - - where_clause, where_args = state_filter.make_sql_filter_clause() - - # Unless the filter clause is empty, we're going to append it after an - # existing where clause - if where_clause: - where_clause = " AND (%s)" % (where_clause,) - - if isinstance(self.database_engine, PostgresEngine): - # Temporarily disable sequential scans in this transaction. This is - # a temporary hack until we can add the right indices in - txn.execute("SET LOCAL enable_seqscan=off") - - # The below query walks the state_group tree so that the "state" - # table includes all state_groups in the tree. It then joins - # against `state_groups_state` to fetch the latest state. - # It assumes that previous state groups are always numerically - # lesser. - # The PARTITION is used to get the event_id in the greatest state - # group for the given type, state_key. - # This may return multiple rows per (type, state_key), but last_value - # should be the same. - sql = """ - WITH RECURSIVE state(state_group) AS ( - VALUES(?::bigint) - UNION ALL - SELECT prev_state_group FROM state_group_edges e, state s - WHERE s.state_group = e.state_group - ) - SELECT DISTINCT type, state_key, last_value(event_id) OVER ( - PARTITION BY type, state_key ORDER BY state_group ASC - ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING - ) AS event_id FROM state_groups_state - WHERE state_group IN ( - SELECT state_group FROM state - ) - """ - - for group in groups: - args = [group] - args.extend(where_args) - - txn.execute(sql + where_clause, args) - for row in txn: - typ, state_key, event_id = row - key = (typ, state_key) - results[group][key] = event_id - else: - max_entries_returned = state_filter.max_entries_returned() - - # We don't use WITH RECURSIVE on sqlite3 as there are distributions - # that ship with an sqlite3 version that doesn't support it (e.g. wheezy) - for group in groups: - next_group = group - - while next_group: - # We did this before by getting the list of group ids, and - # then passing that list to sqlite to get latest event for - # each (type, state_key). However, that was terribly slow - # without the right indices (which we can't add until - # after we finish deduping state, which requires this func) - args = [next_group] - args.extend(where_args) - - txn.execute( - "SELECT type, state_key, event_id FROM state_groups_state" - " WHERE state_group = ? " + where_clause, - args, - ) - results[group].update( - ((typ, state_key), event_id) - for typ, state_key, event_id in txn - if (typ, state_key) not in results[group] - ) - - # If the number of entries in the (type,state_key)->event_id dict - # matches the number of (type,state_keys) types we were searching - # for, then we must have found them all, so no need to go walk - # further down the tree... UNLESS our types filter contained - # wildcards (i.e. Nones) in which case we have to do an exhaustive - # search - if ( - max_entries_returned is not None - and len(results[group]) == max_entries_returned - ): - break - - next_group = self.db.simple_select_one_onecol_txn( - txn, - table="state_group_edges", - keyvalues={"state_group": next_group}, - retcol="prev_state_group", - allow_none=True, - ) - - return results - - # this inherits from EventsWorkerStore because it calls self.get_events -class StateGroupWorkerStore( - EventsWorkerStore, StateGroupBackgroundUpdateStore, SQLBaseStore -): +class StateGroupWorkerStore(EventsWorkerStore, SQLBaseStore): """The parts of StateGroupStore that can be called from workers. """ - STATE_GROUP_DEDUPLICATION_UPDATE_NAME = "state_group_state_deduplication" - STATE_GROUP_INDEX_UPDATE_NAME = "state_group_state_type_index" - CURRENT_STATE_INDEX_UPDATE_NAME = "current_state_members_idx" - def __init__(self, database: Database, db_conn, hs): super(StateGroupWorkerStore, self).__init__(database, db_conn, hs) - # Originally the state store used a single DictionaryCache to cache the - # event IDs for the state types in a given state group to avoid hammering - # on the state_group* tables. - # - # The point of using a DictionaryCache is that it can cache a subset - # of the state events for a given state group (i.e. a subset of the keys for a - # given dict which is an entry in the cache for a given state group ID). - # - # However, this poses problems when performing complicated queries - # on the store - for instance: "give me all the state for this group, but - # limit members to this subset of users", as DictionaryCache's API isn't - # rich enough to say "please cache any of these fields, apart from this subset". - # This is problematic when lazy loading members, which requires this behaviour, - # as without it the cache has no choice but to speculatively load all - # state events for the group, which negates the efficiency being sought. - # - # Rather than overcomplicating DictionaryCache's API, we instead split the - # state_group_cache into two halves - one for tracking non-member events, - # and the other for tracking member_events. This means that lazy loading - # queries can be made in a cache-friendly manner by querying both caches - # separately and then merging the result. So for the example above, you - # would query the members cache for a specific subset of state keys - # (which DictionaryCache will handle efficiently and fine) and the non-members - # cache for all state (which DictionaryCache will similarly handle fine) - # and then just merge the results together. - # - # We size the non-members cache to be smaller than the members cache as the - # vast majority of state in Matrix (today) is member events. - - self._state_group_cache = DictionaryCache( - "*stateGroupCache*", - # TODO: this hasn't been tuned yet - 50000 * get_cache_factor_for("stateGroupCache"), - ) - self._state_group_members_cache = DictionaryCache( - "*stateGroupMembersCache*", - 500000 * get_cache_factor_for("stateGroupMembersCache"), - ) - @defer.inlineCallbacks def get_room_version(self, room_id): """Get the room_version of a given room @@ -431,229 +235,6 @@ class StateGroupWorkerStore( return event.content.get("canonical_alias") - @cached(max_entries=10000, iterable=True) - def get_state_group_delta(self, state_group): - """Given a state group try to return a previous group and a delta between - the old and the new. - - Returns: - (prev_group, delta_ids), where both may be None. - """ - - def _get_state_group_delta_txn(txn): - prev_group = self.db.simple_select_one_onecol_txn( - txn, - table="state_group_edges", - keyvalues={"state_group": state_group}, - retcol="prev_state_group", - allow_none=True, - ) - - if not prev_group: - return _GetStateGroupDelta(None, None) - - delta_ids = self.db.simple_select_list_txn( - txn, - table="state_groups_state", - keyvalues={"state_group": state_group}, - retcols=("type", "state_key", "event_id"), - ) - - return _GetStateGroupDelta( - prev_group, - {(row["type"], row["state_key"]): row["event_id"] for row in delta_ids}, - ) - - return self.db.runInteraction( - "get_state_group_delta", _get_state_group_delta_txn - ) - - @defer.inlineCallbacks - def get_state_groups_ids(self, _room_id, event_ids): - """Get the event IDs of all the state for the state groups for the given events - - Args: - _room_id (str): id of the room for these events - event_ids (iterable[str]): ids of the events - - Returns: - Deferred[dict[int, dict[tuple[str, str], str]]]: - dict of state_group_id -> (dict of (type, state_key) -> event id) - """ - if not event_ids: - return {} - - event_to_groups = yield self._get_state_group_for_events(event_ids) - - groups = set(itervalues(event_to_groups)) - group_to_state = yield self._get_state_for_groups(groups) - - return group_to_state - - @defer.inlineCallbacks - def get_state_ids_for_group(self, state_group): - """Get the event IDs of all the state in the given state group - - Args: - state_group (int) - - Returns: - Deferred[dict]: Resolves to a map of (type, state_key) -> event_id - """ - group_to_state = yield self._get_state_for_groups((state_group,)) - - return group_to_state[state_group] - - @defer.inlineCallbacks - def get_state_groups(self, room_id, event_ids): - """ Get the state groups for the given list of event_ids - - Returns: - Deferred[dict[int, list[EventBase]]]: - dict of state_group_id -> list of state events. - """ - if not event_ids: - return {} - - group_to_ids = yield self.get_state_groups_ids(room_id, event_ids) - - state_event_map = yield self.get_events( - [ - ev_id - for group_ids in itervalues(group_to_ids) - for ev_id in itervalues(group_ids) - ], - get_prev_content=False, - ) - - return { - group: [ - state_event_map[v] - for v in itervalues(event_id_map) - if v in state_event_map - ] - for group, event_id_map in iteritems(group_to_ids) - } - - @defer.inlineCallbacks - def _get_state_groups_from_groups(self, groups, state_filter): - """Returns the state groups for a given set of groups, filtering on - types of state events. - - Args: - groups(list[int]): list of state group IDs to query - state_filter (StateFilter): The state filter used to fetch state - from the database. - Returns: - Deferred[dict[int, dict[tuple[str, str], str]]]: - dict of state_group_id -> (dict of (type, state_key) -> event id) - """ - results = {} - - chunks = [groups[i : i + 100] for i in range(0, len(groups), 100)] - for chunk in chunks: - res = yield self.db.runInteraction( - "_get_state_groups_from_groups", - self._get_state_groups_from_groups_txn, - chunk, - state_filter, - ) - results.update(res) - - return results - - @defer.inlineCallbacks - def get_state_for_events(self, event_ids, state_filter=StateFilter.all()): - """Given a list of event_ids and type tuples, return a list of state - dicts for each event. - - Args: - event_ids (list[string]) - state_filter (StateFilter): The state filter used to fetch state - from the database. - - Returns: - deferred: A dict of (event_id) -> (type, state_key) -> [state_events] - """ - event_to_groups = yield self._get_state_group_for_events(event_ids) - - groups = set(itervalues(event_to_groups)) - group_to_state = yield self._get_state_for_groups(groups, state_filter) - - state_event_map = yield self.get_events( - [ev_id for sd in itervalues(group_to_state) for ev_id in itervalues(sd)], - get_prev_content=False, - ) - - event_to_state = { - event_id: { - k: state_event_map[v] - for k, v in iteritems(group_to_state[group]) - if v in state_event_map - } - for event_id, group in iteritems(event_to_groups) - } - - return {event: event_to_state[event] for event in event_ids} - - @defer.inlineCallbacks - def get_state_ids_for_events(self, event_ids, state_filter=StateFilter.all()): - """ - Get the state dicts corresponding to a list of events, containing the event_ids - of the state events (as opposed to the events themselves) - - Args: - event_ids(list(str)): events whose state should be returned - state_filter (StateFilter): The state filter used to fetch state - from the database. - - Returns: - A deferred dict from event_id -> (type, state_key) -> event_id - """ - event_to_groups = yield self._get_state_group_for_events(event_ids) - - groups = set(itervalues(event_to_groups)) - group_to_state = yield self._get_state_for_groups(groups, state_filter) - - event_to_state = { - event_id: group_to_state[group] - for event_id, group in iteritems(event_to_groups) - } - - return {event: event_to_state[event] for event in event_ids} - - @defer.inlineCallbacks - def get_state_for_event(self, event_id, state_filter=StateFilter.all()): - """ - Get the state dict corresponding to a particular event - - Args: - event_id(str): event whose state should be returned - state_filter (StateFilter): The state filter used to fetch state - from the database. - - Returns: - A deferred dict from (type, state_key) -> state_event - """ - state_map = yield self.get_state_for_events([event_id], state_filter) - return state_map[event_id] - - @defer.inlineCallbacks - def get_state_ids_for_event(self, event_id, state_filter=StateFilter.all()): - """ - Get the state dict corresponding to a particular event - - Args: - event_id(str): event whose state should be returned - state_filter (StateFilter): The state filter used to fetch state - from the database. - - Returns: - A deferred dict from (type, state_key) -> state_event - """ - state_map = yield self.get_state_ids_for_events([event_id], state_filter) - return state_map[event_id] - @cached(max_entries=50000) def _get_state_group_for_event(self, event_id): return self.db.simple_select_one_onecol( @@ -684,329 +265,6 @@ class StateGroupWorkerStore( return {row["event_id"]: row["state_group"] for row in rows} - def _get_state_for_group_using_cache(self, cache, group, state_filter): - """Checks if group is in cache. See `_get_state_for_groups` - - Args: - cache(DictionaryCache): the state group cache to use - group(int): The state group to lookup - state_filter (StateFilter): The state filter used to fetch state - from the database. - - Returns 2-tuple (`state_dict`, `got_all`). - `got_all` is a bool indicating if we successfully retrieved all - requests state from the cache, if False we need to query the DB for the - missing state. - """ - is_all, known_absent, state_dict_ids = cache.get(group) - - if is_all or state_filter.is_full(): - # Either we have everything or want everything, either way - # `is_all` tells us whether we've gotten everything. - return state_filter.filter_state(state_dict_ids), is_all - - # tracks whether any of our requested types are missing from the cache - missing_types = False - - if state_filter.has_wildcards(): - # We don't know if we fetched all the state keys for the types in - # the filter that are wildcards, so we have to assume that we may - # have missed some. - missing_types = True - else: - # There aren't any wild cards, so `concrete_types()` returns the - # complete list of event types we're wanting. - for key in state_filter.concrete_types(): - if key not in state_dict_ids and key not in known_absent: - missing_types = True - break - - return state_filter.filter_state(state_dict_ids), not missing_types - - @defer.inlineCallbacks - def _get_state_for_groups(self, groups, state_filter=StateFilter.all()): - """Gets the state at each of a list of state groups, optionally - filtering by type/state_key - - Args: - groups (iterable[int]): list of state groups for which we want - to get the state. - state_filter (StateFilter): The state filter used to fetch state - from the database. - Returns: - Deferred[dict[int, dict[tuple[str, str], str]]]: - dict of state_group_id -> (dict of (type, state_key) -> event id) - """ - - member_filter, non_member_filter = state_filter.get_member_split() - - # Now we look them up in the member and non-member caches - ( - non_member_state, - incomplete_groups_nm, - ) = yield self._get_state_for_groups_using_cache( - groups, self._state_group_cache, state_filter=non_member_filter - ) - - ( - member_state, - incomplete_groups_m, - ) = yield self._get_state_for_groups_using_cache( - groups, self._state_group_members_cache, state_filter=member_filter - ) - - state = dict(non_member_state) - for group in groups: - state[group].update(member_state[group]) - - # Now fetch any missing groups from the database - - incomplete_groups = incomplete_groups_m | incomplete_groups_nm - - if not incomplete_groups: - return state - - cache_sequence_nm = self._state_group_cache.sequence - cache_sequence_m = self._state_group_members_cache.sequence - - # Help the cache hit ratio by expanding the filter a bit - db_state_filter = state_filter.return_expanded() - - group_to_state_dict = yield self._get_state_groups_from_groups( - list(incomplete_groups), state_filter=db_state_filter - ) - - # Now lets update the caches - self._insert_into_cache( - group_to_state_dict, - db_state_filter, - cache_seq_num_members=cache_sequence_m, - cache_seq_num_non_members=cache_sequence_nm, - ) - - # And finally update the result dict, by filtering out any extra - # stuff we pulled out of the database. - for group, group_state_dict in iteritems(group_to_state_dict): - # We just replace any existing entries, as we will have loaded - # everything we need from the database anyway. - state[group] = state_filter.filter_state(group_state_dict) - - return state - - def _get_state_for_groups_using_cache(self, groups, cache, state_filter): - """Gets the state at each of a list of state groups, optionally - filtering by type/state_key, querying from a specific cache. - - Args: - groups (iterable[int]): list of state groups for which we want - to get the state. - cache (DictionaryCache): the cache of group ids to state dicts which - we will pass through - either the normal state cache or the specific - members state cache. - state_filter (StateFilter): The state filter used to fetch state - from the database. - - Returns: - tuple[dict[int, dict[tuple[str, str], str]], set[int]]: Tuple of - dict of state_group_id -> (dict of (type, state_key) -> event id) - of entries in the cache, and the state group ids either missing - from the cache or incomplete. - """ - results = {} - incomplete_groups = set() - for group in set(groups): - state_dict_ids, got_all = self._get_state_for_group_using_cache( - cache, group, state_filter - ) - results[group] = state_dict_ids - - if not got_all: - incomplete_groups.add(group) - - return results, incomplete_groups - - def _insert_into_cache( - self, - group_to_state_dict, - state_filter, - cache_seq_num_members, - cache_seq_num_non_members, - ): - """Inserts results from querying the database into the relevant cache. - - Args: - group_to_state_dict (dict): The new entries pulled from database. - Map from state group to state dict - state_filter (StateFilter): The state filter used to fetch state - from the database. - cache_seq_num_members (int): Sequence number of member cache since - last lookup in cache - cache_seq_num_non_members (int): Sequence number of member cache since - last lookup in cache - """ - - # We need to work out which types we've fetched from the DB for the - # member vs non-member caches. This should be as accurate as possible, - # but can be an underestimate (e.g. when we have wild cards) - - member_filter, non_member_filter = state_filter.get_member_split() - if member_filter.is_full(): - # We fetched all member events - member_types = None - else: - # `concrete_types()` will only return a subset when there are wild - # cards in the filter, but that's fine. - member_types = member_filter.concrete_types() - - if non_member_filter.is_full(): - # We fetched all non member events - non_member_types = None - else: - non_member_types = non_member_filter.concrete_types() - - for group, group_state_dict in iteritems(group_to_state_dict): - state_dict_members = {} - state_dict_non_members = {} - - for k, v in iteritems(group_state_dict): - if k[0] == EventTypes.Member: - state_dict_members[k] = v - else: - state_dict_non_members[k] = v - - self._state_group_members_cache.update( - cache_seq_num_members, - key=group, - value=state_dict_members, - fetched_keys=member_types, - ) - - self._state_group_cache.update( - cache_seq_num_non_members, - key=group, - value=state_dict_non_members, - fetched_keys=non_member_types, - ) - - def store_state_group( - self, event_id, room_id, prev_group, delta_ids, current_state_ids - ): - """Store a new set of state, returning a newly assigned state group. - - Args: - event_id (str): The event ID for which the state was calculated - room_id (str) - prev_group (int|None): A previous state group for the room, optional. - delta_ids (dict|None): The delta between state at `prev_group` and - `current_state_ids`, if `prev_group` was given. Same format as - `current_state_ids`. - current_state_ids (dict): The state to store. Map of (type, state_key) - to event_id. - - Returns: - Deferred[int]: The state group ID - """ - - def _store_state_group_txn(txn): - if current_state_ids is None: - # AFAIK, this can never happen - raise Exception("current_state_ids cannot be None") - - state_group = self.database_engine.get_next_state_group_id(txn) - - self.db.simple_insert_txn( - txn, - table="state_groups", - values={"id": state_group, "room_id": room_id, "event_id": event_id}, - ) - - # We persist as a delta if we can, while also ensuring the chain - # of deltas isn't tooo long, as otherwise read performance degrades. - if prev_group: - is_in_db = self.db.simple_select_one_onecol_txn( - txn, - table="state_groups", - keyvalues={"id": prev_group}, - retcol="id", - allow_none=True, - ) - if not is_in_db: - raise Exception( - "Trying to persist state with unpersisted prev_group: %r" - % (prev_group,) - ) - - potential_hops = self._count_state_group_hops_txn(txn, prev_group) - if prev_group and potential_hops < MAX_STATE_DELTA_HOPS: - self.db.simple_insert_txn( - txn, - table="state_group_edges", - values={"state_group": state_group, "prev_state_group": prev_group}, - ) - - self.db.simple_insert_many_txn( - txn, - table="state_groups_state", - values=[ - { - "state_group": state_group, - "room_id": room_id, - "type": key[0], - "state_key": key[1], - "event_id": state_id, - } - for key, state_id in iteritems(delta_ids) - ], - ) - else: - self.db.simple_insert_many_txn( - txn, - table="state_groups_state", - values=[ - { - "state_group": state_group, - "room_id": room_id, - "type": key[0], - "state_key": key[1], - "event_id": state_id, - } - for key, state_id in iteritems(current_state_ids) - ], - ) - - # Prefill the state group caches with this group. - # It's fine to use the sequence like this as the state group map - # is immutable. (If the map wasn't immutable then this prefill could - # race with another update) - - current_member_state_ids = { - s: ev - for (s, ev) in iteritems(current_state_ids) - if s[0] == EventTypes.Member - } - txn.call_after( - self._state_group_members_cache.update, - self._state_group_members_cache.sequence, - key=state_group, - value=dict(current_member_state_ids), - ) - - current_non_member_state_ids = { - s: ev - for (s, ev) in iteritems(current_state_ids) - if s[0] != EventTypes.Member - } - txn.call_after( - self._state_group_cache.update, - self._state_group_cache.sequence, - key=state_group, - value=dict(current_non_member_state_ids), - ) - - return state_group - - return self.db.runInteraction("store_state_group", _store_state_group_txn) - @defer.inlineCallbacks def get_referenced_state_groups(self, state_groups): """Check if the state groups are referenced by events. @@ -1031,22 +289,14 @@ class StateGroupWorkerStore( return set(row["state_group"] for row in rows) -class StateBackgroundUpdateStore(StateGroupBackgroundUpdateStore): +class MainStateBackgroundUpdateStore(SQLBaseStore): - STATE_GROUP_DEDUPLICATION_UPDATE_NAME = "state_group_state_deduplication" - STATE_GROUP_INDEX_UPDATE_NAME = "state_group_state_type_index" CURRENT_STATE_INDEX_UPDATE_NAME = "current_state_members_idx" EVENT_STATE_GROUP_INDEX_UPDATE_NAME = "event_to_state_groups_sg_index" def __init__(self, database: Database, db_conn, hs): - super(StateBackgroundUpdateStore, self).__init__(database, db_conn, hs) - self.db.updates.register_background_update_handler( - self.STATE_GROUP_DEDUPLICATION_UPDATE_NAME, - self._background_deduplicate_state, - ) - self.db.updates.register_background_update_handler( - self.STATE_GROUP_INDEX_UPDATE_NAME, self._background_index_state - ) + super(MainStateBackgroundUpdateStore, self).__init__(database, db_conn, hs) + self.db.updates.register_background_index_update( self.CURRENT_STATE_INDEX_UPDATE_NAME, index_name="current_state_events_member_index", @@ -1061,181 +311,8 @@ class StateBackgroundUpdateStore(StateGroupBackgroundUpdateStore): columns=["state_group"], ) - @defer.inlineCallbacks - def _background_deduplicate_state(self, progress, batch_size): - """This background update will slowly deduplicate state by reencoding - them as deltas. - """ - last_state_group = progress.get("last_state_group", 0) - rows_inserted = progress.get("rows_inserted", 0) - max_group = progress.get("max_group", None) - - BATCH_SIZE_SCALE_FACTOR = 100 - batch_size = max(1, int(batch_size / BATCH_SIZE_SCALE_FACTOR)) - - if max_group is None: - rows = yield self.db.execute( - "_background_deduplicate_state", - None, - "SELECT coalesce(max(id), 0) FROM state_groups", - ) - max_group = rows[0][0] - - def reindex_txn(txn): - new_last_state_group = last_state_group - for count in range(batch_size): - txn.execute( - "SELECT id, room_id FROM state_groups" - " WHERE ? < id AND id <= ?" - " ORDER BY id ASC" - " LIMIT 1", - (new_last_state_group, max_group), - ) - row = txn.fetchone() - if row: - state_group, room_id = row - - if not row or not state_group: - return True, count - - txn.execute( - "SELECT state_group FROM state_group_edges" - " WHERE state_group = ?", - (state_group,), - ) - - # If we reach a point where we've already started inserting - # edges we should stop. - if txn.fetchall(): - return True, count - - txn.execute( - "SELECT coalesce(max(id), 0) FROM state_groups" - " WHERE id < ? AND room_id = ?", - (state_group, room_id), - ) - (prev_group,) = txn.fetchone() - new_last_state_group = state_group - - if prev_group: - potential_hops = self._count_state_group_hops_txn(txn, prev_group) - if potential_hops >= MAX_STATE_DELTA_HOPS: - # We want to ensure chains are at most this long,# - # otherwise read performance degrades. - continue - - prev_state = self._get_state_groups_from_groups_txn( - txn, [prev_group] - ) - prev_state = prev_state[prev_group] - - curr_state = self._get_state_groups_from_groups_txn( - txn, [state_group] - ) - curr_state = curr_state[state_group] - - if not set(prev_state.keys()) - set(curr_state.keys()): - # We can only do a delta if the current has a strict super set - # of keys - - delta_state = { - key: value - for key, value in iteritems(curr_state) - if prev_state.get(key, None) != value - } - - self.db.simple_delete_txn( - txn, - table="state_group_edges", - keyvalues={"state_group": state_group}, - ) - - self.db.simple_insert_txn( - txn, - table="state_group_edges", - values={ - "state_group": state_group, - "prev_state_group": prev_group, - }, - ) - - self.db.simple_delete_txn( - txn, - table="state_groups_state", - keyvalues={"state_group": state_group}, - ) - - self.db.simple_insert_many_txn( - txn, - table="state_groups_state", - values=[ - { - "state_group": state_group, - "room_id": room_id, - "type": key[0], - "state_key": key[1], - "event_id": state_id, - } - for key, state_id in iteritems(delta_state) - ], - ) - - progress = { - "last_state_group": state_group, - "rows_inserted": rows_inserted + batch_size, - "max_group": max_group, - } - - self.db.updates._background_update_progress_txn( - txn, self.STATE_GROUP_DEDUPLICATION_UPDATE_NAME, progress - ) - - return False, batch_size - - finished, result = yield self.db.runInteraction( - self.STATE_GROUP_DEDUPLICATION_UPDATE_NAME, reindex_txn - ) - - if finished: - yield self.db.updates._end_background_update( - self.STATE_GROUP_DEDUPLICATION_UPDATE_NAME - ) - - return result * BATCH_SIZE_SCALE_FACTOR - - @defer.inlineCallbacks - def _background_index_state(self, progress, batch_size): - def reindex_txn(conn): - conn.rollback() - if isinstance(self.database_engine, PostgresEngine): - # postgres insists on autocommit for the index - conn.set_session(autocommit=True) - try: - txn = conn.cursor() - txn.execute( - "CREATE INDEX CONCURRENTLY state_groups_state_type_idx" - " ON state_groups_state(state_group, type, state_key)" - ) - txn.execute("DROP INDEX IF EXISTS state_groups_state_id") - finally: - conn.set_session(autocommit=False) - else: - txn = conn.cursor() - txn.execute( - "CREATE INDEX state_groups_state_type_idx" - " ON state_groups_state(state_group, type, state_key)" - ) - txn.execute("DROP INDEX IF EXISTS state_groups_state_id") - - yield self.db.runWithConnection(reindex_txn) - - yield self.db.updates._end_background_update(self.STATE_GROUP_INDEX_UPDATE_NAME) - - return 1 - - -class StateStore(StateGroupWorkerStore, StateBackgroundUpdateStore): +class StateStore(StateGroupWorkerStore, MainStateBackgroundUpdateStore): """ Keeps track of the state at a given event. This is done by the concept of `state groups`. Every event is a assigned diff --git a/synapse/storage/data_stores/state/__init__.py b/synapse/storage/data_stores/state/__init__.py new file mode 100644 index 0000000000..86e09f6229 --- /dev/null +++ b/synapse/storage/data_stores/state/__init__.py @@ -0,0 +1,16 @@ +# -*- coding: utf-8 -*- +# Copyright 2019 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from synapse.storage.data_stores.state.store import StateGroupDataStore # noqa: F401 diff --git a/synapse/storage/data_stores/state/bg_updates.py b/synapse/storage/data_stores/state/bg_updates.py new file mode 100644 index 0000000000..e8edaf9f7b --- /dev/null +++ b/synapse/storage/data_stores/state/bg_updates.py @@ -0,0 +1,374 @@ +# -*- coding: utf-8 -*- +# Copyright 2014-2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging + +from six import iteritems + +from twisted.internet import defer + +from synapse.storage._base import SQLBaseStore +from synapse.storage.database import Database +from synapse.storage.engines import PostgresEngine +from synapse.storage.state import StateFilter + +logger = logging.getLogger(__name__) + + +MAX_STATE_DELTA_HOPS = 100 + + +class StateGroupBackgroundUpdateStore(SQLBaseStore): + """Defines functions related to state groups needed to run the state backgroud + updates. + """ + + def _count_state_group_hops_txn(self, txn, state_group): + """Given a state group, count how many hops there are in the tree. + + This is used to ensure the delta chains don't get too long. + """ + if isinstance(self.database_engine, PostgresEngine): + sql = """ + WITH RECURSIVE state(state_group) AS ( + VALUES(?::bigint) + UNION ALL + SELECT prev_state_group FROM state_group_edges e, state s + WHERE s.state_group = e.state_group + ) + SELECT count(*) FROM state; + """ + + txn.execute(sql, (state_group,)) + row = txn.fetchone() + if row and row[0]: + return row[0] + else: + return 0 + else: + # We don't use WITH RECURSIVE on sqlite3 as there are distributions + # that ship with an sqlite3 version that doesn't support it (e.g. wheezy) + next_group = state_group + count = 0 + + while next_group: + next_group = self.db.simple_select_one_onecol_txn( + txn, + table="state_group_edges", + keyvalues={"state_group": next_group}, + retcol="prev_state_group", + allow_none=True, + ) + if next_group: + count += 1 + + return count + + def _get_state_groups_from_groups_txn( + self, txn, groups, state_filter=StateFilter.all() + ): + results = {group: {} for group in groups} + + where_clause, where_args = state_filter.make_sql_filter_clause() + + # Unless the filter clause is empty, we're going to append it after an + # existing where clause + if where_clause: + where_clause = " AND (%s)" % (where_clause,) + + if isinstance(self.database_engine, PostgresEngine): + # Temporarily disable sequential scans in this transaction. This is + # a temporary hack until we can add the right indices in + txn.execute("SET LOCAL enable_seqscan=off") + + # The below query walks the state_group tree so that the "state" + # table includes all state_groups in the tree. It then joins + # against `state_groups_state` to fetch the latest state. + # It assumes that previous state groups are always numerically + # lesser. + # The PARTITION is used to get the event_id in the greatest state + # group for the given type, state_key. + # This may return multiple rows per (type, state_key), but last_value + # should be the same. + sql = """ + WITH RECURSIVE state(state_group) AS ( + VALUES(?::bigint) + UNION ALL + SELECT prev_state_group FROM state_group_edges e, state s + WHERE s.state_group = e.state_group + ) + SELECT DISTINCT type, state_key, last_value(event_id) OVER ( + PARTITION BY type, state_key ORDER BY state_group ASC + ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING + ) AS event_id FROM state_groups_state + WHERE state_group IN ( + SELECT state_group FROM state + ) + """ + + for group in groups: + args = [group] + args.extend(where_args) + + txn.execute(sql + where_clause, args) + for row in txn: + typ, state_key, event_id = row + key = (typ, state_key) + results[group][key] = event_id + else: + max_entries_returned = state_filter.max_entries_returned() + + # We don't use WITH RECURSIVE on sqlite3 as there are distributions + # that ship with an sqlite3 version that doesn't support it (e.g. wheezy) + for group in groups: + next_group = group + + while next_group: + # We did this before by getting the list of group ids, and + # then passing that list to sqlite to get latest event for + # each (type, state_key). However, that was terribly slow + # without the right indices (which we can't add until + # after we finish deduping state, which requires this func) + args = [next_group] + args.extend(where_args) + + txn.execute( + "SELECT type, state_key, event_id FROM state_groups_state" + " WHERE state_group = ? " + where_clause, + args, + ) + results[group].update( + ((typ, state_key), event_id) + for typ, state_key, event_id in txn + if (typ, state_key) not in results[group] + ) + + # If the number of entries in the (type,state_key)->event_id dict + # matches the number of (type,state_keys) types we were searching + # for, then we must have found them all, so no need to go walk + # further down the tree... UNLESS our types filter contained + # wildcards (i.e. Nones) in which case we have to do an exhaustive + # search + if ( + max_entries_returned is not None + and len(results[group]) == max_entries_returned + ): + break + + next_group = self.db.simple_select_one_onecol_txn( + txn, + table="state_group_edges", + keyvalues={"state_group": next_group}, + retcol="prev_state_group", + allow_none=True, + ) + + return results + + +class StateBackgroundUpdateStore(StateGroupBackgroundUpdateStore): + + STATE_GROUP_DEDUPLICATION_UPDATE_NAME = "state_group_state_deduplication" + STATE_GROUP_INDEX_UPDATE_NAME = "state_group_state_type_index" + STATE_GROUPS_ROOM_INDEX_UPDATE_NAME = "state_groups_room_id_idx" + + def __init__(self, database: Database, db_conn, hs): + super(StateBackgroundUpdateStore, self).__init__(database, db_conn, hs) + self.db.updates.register_background_update_handler( + self.STATE_GROUP_DEDUPLICATION_UPDATE_NAME, + self._background_deduplicate_state, + ) + self.db.updates.register_background_update_handler( + self.STATE_GROUP_INDEX_UPDATE_NAME, self._background_index_state + ) + self.db.updates.register_background_index_update( + self.STATE_GROUPS_ROOM_INDEX_UPDATE_NAME, + index_name="state_groups_room_id_idx", + table="state_groups", + columns=["room_id"], + ) + + @defer.inlineCallbacks + def _background_deduplicate_state(self, progress, batch_size): + """This background update will slowly deduplicate state by reencoding + them as deltas. + """ + last_state_group = progress.get("last_state_group", 0) + rows_inserted = progress.get("rows_inserted", 0) + max_group = progress.get("max_group", None) + + BATCH_SIZE_SCALE_FACTOR = 100 + + batch_size = max(1, int(batch_size / BATCH_SIZE_SCALE_FACTOR)) + + if max_group is None: + rows = yield self.db.execute( + "_background_deduplicate_state", + None, + "SELECT coalesce(max(id), 0) FROM state_groups", + ) + max_group = rows[0][0] + + def reindex_txn(txn): + new_last_state_group = last_state_group + for count in range(batch_size): + txn.execute( + "SELECT id, room_id FROM state_groups" + " WHERE ? < id AND id <= ?" + " ORDER BY id ASC" + " LIMIT 1", + (new_last_state_group, max_group), + ) + row = txn.fetchone() + if row: + state_group, room_id = row + + if not row or not state_group: + return True, count + + txn.execute( + "SELECT state_group FROM state_group_edges" + " WHERE state_group = ?", + (state_group,), + ) + + # If we reach a point where we've already started inserting + # edges we should stop. + if txn.fetchall(): + return True, count + + txn.execute( + "SELECT coalesce(max(id), 0) FROM state_groups" + " WHERE id < ? AND room_id = ?", + (state_group, room_id), + ) + (prev_group,) = txn.fetchone() + new_last_state_group = state_group + + if prev_group: + potential_hops = self._count_state_group_hops_txn(txn, prev_group) + if potential_hops >= MAX_STATE_DELTA_HOPS: + # We want to ensure chains are at most this long,# + # otherwise read performance degrades. + continue + + prev_state = self._get_state_groups_from_groups_txn( + txn, [prev_group] + ) + prev_state = prev_state[prev_group] + + curr_state = self._get_state_groups_from_groups_txn( + txn, [state_group] + ) + curr_state = curr_state[state_group] + + if not set(prev_state.keys()) - set(curr_state.keys()): + # We can only do a delta if the current has a strict super set + # of keys + + delta_state = { + key: value + for key, value in iteritems(curr_state) + if prev_state.get(key, None) != value + } + + self.db.simple_delete_txn( + txn, + table="state_group_edges", + keyvalues={"state_group": state_group}, + ) + + self.db.simple_insert_txn( + txn, + table="state_group_edges", + values={ + "state_group": state_group, + "prev_state_group": prev_group, + }, + ) + + self.db.simple_delete_txn( + txn, + table="state_groups_state", + keyvalues={"state_group": state_group}, + ) + + self.db.simple_insert_many_txn( + txn, + table="state_groups_state", + values=[ + { + "state_group": state_group, + "room_id": room_id, + "type": key[0], + "state_key": key[1], + "event_id": state_id, + } + for key, state_id in iteritems(delta_state) + ], + ) + + progress = { + "last_state_group": state_group, + "rows_inserted": rows_inserted + batch_size, + "max_group": max_group, + } + + self.db.updates._background_update_progress_txn( + txn, self.STATE_GROUP_DEDUPLICATION_UPDATE_NAME, progress + ) + + return False, batch_size + + finished, result = yield self.db.runInteraction( + self.STATE_GROUP_DEDUPLICATION_UPDATE_NAME, reindex_txn + ) + + if finished: + yield self.db.updates._end_background_update( + self.STATE_GROUP_DEDUPLICATION_UPDATE_NAME + ) + + return result * BATCH_SIZE_SCALE_FACTOR + + @defer.inlineCallbacks + def _background_index_state(self, progress, batch_size): + def reindex_txn(conn): + conn.rollback() + if isinstance(self.database_engine, PostgresEngine): + # postgres insists on autocommit for the index + conn.set_session(autocommit=True) + try: + txn = conn.cursor() + txn.execute( + "CREATE INDEX CONCURRENTLY state_groups_state_type_idx" + " ON state_groups_state(state_group, type, state_key)" + ) + txn.execute("DROP INDEX IF EXISTS state_groups_state_id") + finally: + conn.set_session(autocommit=False) + else: + txn = conn.cursor() + txn.execute( + "CREATE INDEX state_groups_state_type_idx" + " ON state_groups_state(state_group, type, state_key)" + ) + txn.execute("DROP INDEX IF EXISTS state_groups_state_id") + + yield self.db.runWithConnection(reindex_txn) + + yield self.db.updates._end_background_update(self.STATE_GROUP_INDEX_UPDATE_NAME) + + return 1 diff --git a/synapse/storage/data_stores/state/schema/delta/23/drop_state_index.sql b/synapse/storage/data_stores/state/schema/delta/23/drop_state_index.sql new file mode 100644 index 0000000000..ae09fa0065 --- /dev/null +++ b/synapse/storage/data_stores/state/schema/delta/23/drop_state_index.sql @@ -0,0 +1,16 @@ +/* Copyright 2015, 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +DROP INDEX IF EXISTS state_groups_state_tuple; diff --git a/synapse/storage/data_stores/state/schema/delta/30/state_stream.sql b/synapse/storage/data_stores/state/schema/delta/30/state_stream.sql new file mode 100644 index 0000000000..e85699e82e --- /dev/null +++ b/synapse/storage/data_stores/state/schema/delta/30/state_stream.sql @@ -0,0 +1,33 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + +/* We used to create a table called current_state_resets, but this is no + * longer used and is removed in delta 54. + */ + +/* The outlier events that have aquired a state group typically through + * backfill. This is tracked separately to the events table, as assigning a + * state group change the position of the existing event in the stream + * ordering. + * However since a stream_ordering is assigned in persist_event for the + * (event, state) pair, we can use that stream_ordering to identify when + * the new state was assigned for the event. + */ +CREATE TABLE IF NOT EXISTS ex_outlier_stream( + event_stream_ordering BIGINT PRIMARY KEY NOT NULL, + event_id TEXT NOT NULL, + state_group BIGINT NOT NULL +); diff --git a/synapse/storage/data_stores/state/schema/delta/32/remove_state_indices.sql b/synapse/storage/data_stores/state/schema/delta/32/remove_state_indices.sql new file mode 100644 index 0000000000..1450313bfa --- /dev/null +++ b/synapse/storage/data_stores/state/schema/delta/32/remove_state_indices.sql @@ -0,0 +1,19 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + +-- The following indices are redundant, other indices are equivalent or +-- supersets +DROP INDEX IF EXISTS state_groups_id; -- Duplicate of PRIMARY KEY diff --git a/synapse/storage/data_stores/state/schema/delta/35/add_state_index.sql b/synapse/storage/data_stores/state/schema/delta/35/add_state_index.sql new file mode 100644 index 0000000000..33980d02f0 --- /dev/null +++ b/synapse/storage/data_stores/state/schema/delta/35/add_state_index.sql @@ -0,0 +1,17 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +INSERT into background_updates (update_name, progress_json, depends_on) + VALUES ('state_group_state_type_index', '{}', 'state_group_state_deduplication'); diff --git a/synapse/storage/data_stores/state/schema/delta/35/state.sql b/synapse/storage/data_stores/state/schema/delta/35/state.sql new file mode 100644 index 0000000000..0f1fa68a89 --- /dev/null +++ b/synapse/storage/data_stores/state/schema/delta/35/state.sql @@ -0,0 +1,22 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE TABLE state_group_edges( + state_group BIGINT NOT NULL, + prev_state_group BIGINT NOT NULL +); + +CREATE INDEX state_group_edges_idx ON state_group_edges(state_group); +CREATE INDEX state_group_edges_prev_idx ON state_group_edges(prev_state_group); diff --git a/synapse/storage/data_stores/state/schema/delta/35/state_dedupe.sql b/synapse/storage/data_stores/state/schema/delta/35/state_dedupe.sql new file mode 100644 index 0000000000..97e5067ef4 --- /dev/null +++ b/synapse/storage/data_stores/state/schema/delta/35/state_dedupe.sql @@ -0,0 +1,17 @@ +/* Copyright 2016 OpenMarket Ltd + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +INSERT into background_updates (update_name, progress_json) + VALUES ('state_group_state_deduplication', '{}'); diff --git a/synapse/storage/data_stores/state/schema/delta/47/state_group_seq.py b/synapse/storage/data_stores/state/schema/delta/47/state_group_seq.py new file mode 100644 index 0000000000..9fd1ccf6f7 --- /dev/null +++ b/synapse/storage/data_stores/state/schema/delta/47/state_group_seq.py @@ -0,0 +1,34 @@ +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from synapse.storage.engines import PostgresEngine + + +def run_create(cur, database_engine, *args, **kwargs): + if isinstance(database_engine, PostgresEngine): + # if we already have some state groups, we want to start making new + # ones with a higher id. + cur.execute("SELECT max(id) FROM state_groups") + row = cur.fetchone() + + if row[0] is None: + start_val = 1 + else: + start_val = row[0] + 1 + + cur.execute("CREATE SEQUENCE state_group_id_seq START WITH %s", (start_val,)) + + +def run_upgrade(*args, **kwargs): + pass diff --git a/synapse/storage/data_stores/state/schema/delta/56/state_group_room_idx.sql b/synapse/storage/data_stores/state/schema/delta/56/state_group_room_idx.sql new file mode 100644 index 0000000000..7916ef18b2 --- /dev/null +++ b/synapse/storage/data_stores/state/schema/delta/56/state_group_room_idx.sql @@ -0,0 +1,17 @@ +/* Copyright 2019 The Matrix.org Foundation C.I.C. + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +INSERT INTO background_updates (update_name, progress_json) VALUES + ('state_groups_room_id_idx', '{}'); diff --git a/synapse/storage/data_stores/state/schema/full_schemas/54/full.sql b/synapse/storage/data_stores/state/schema/full_schemas/54/full.sql new file mode 100644 index 0000000000..35f97d6b3d --- /dev/null +++ b/synapse/storage/data_stores/state/schema/full_schemas/54/full.sql @@ -0,0 +1,37 @@ +/* Copyright 2019 The Matrix.org Foundation C.I.C + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE TABLE state_groups ( + id BIGINT PRIMARY KEY, + room_id TEXT NOT NULL, + event_id TEXT NOT NULL +); + +CREATE TABLE state_groups_state ( + state_group BIGINT NOT NULL, + room_id TEXT NOT NULL, + type TEXT NOT NULL, + state_key TEXT NOT NULL, + event_id TEXT NOT NULL +); + +CREATE TABLE state_group_edges ( + state_group BIGINT NOT NULL, + prev_state_group BIGINT NOT NULL +); + +CREATE INDEX state_group_edges_idx ON state_group_edges (state_group); +CREATE INDEX state_group_edges_prev_idx ON state_group_edges (prev_state_group); +CREATE INDEX state_groups_state_type_idx ON state_groups_state (state_group, type, state_key); diff --git a/synapse/storage/data_stores/state/schema/full_schemas/54/sequence.sql.postgres b/synapse/storage/data_stores/state/schema/full_schemas/54/sequence.sql.postgres new file mode 100644 index 0000000000..fcd926c9fb --- /dev/null +++ b/synapse/storage/data_stores/state/schema/full_schemas/54/sequence.sql.postgres @@ -0,0 +1,21 @@ +/* Copyright 2019 The Matrix.org Foundation C.I.C + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE SEQUENCE state_group_id_seq + START WITH 1 + INCREMENT BY 1 + NO MINVALUE + NO MAXVALUE + CACHE 1; diff --git a/synapse/storage/data_stores/state/store.py b/synapse/storage/data_stores/state/store.py new file mode 100644 index 0000000000..d53695f238 --- /dev/null +++ b/synapse/storage/data_stores/state/store.py @@ -0,0 +1,640 @@ +# -*- coding: utf-8 -*- +# Copyright 2014-2016 OpenMarket Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging +from collections import namedtuple + +from six import iteritems +from six.moves import range + +from twisted.internet import defer + +from synapse.api.constants import EventTypes +from synapse.storage._base import SQLBaseStore +from synapse.storage.data_stores.state.bg_updates import StateBackgroundUpdateStore +from synapse.storage.database import Database +from synapse.storage.state import StateFilter +from synapse.util.caches import get_cache_factor_for +from synapse.util.caches.descriptors import cached +from synapse.util.caches.dictionary_cache import DictionaryCache + +logger = logging.getLogger(__name__) + + +MAX_STATE_DELTA_HOPS = 100 + + +class _GetStateGroupDelta( + namedtuple("_GetStateGroupDelta", ("prev_group", "delta_ids")) +): + """Return type of get_state_group_delta that implements __len__, which lets + us use the itrable flag when caching + """ + + __slots__ = [] + + def __len__(self): + return len(self.delta_ids) if self.delta_ids else 0 + + +class StateGroupDataStore(StateBackgroundUpdateStore, SQLBaseStore): + """A data store for fetching/storing state groups. + """ + + def __init__(self, database: Database, db_conn, hs): + super(StateGroupDataStore, self).__init__(database, db_conn, hs) + + # Originally the state store used a single DictionaryCache to cache the + # event IDs for the state types in a given state group to avoid hammering + # on the state_group* tables. + # + # The point of using a DictionaryCache is that it can cache a subset + # of the state events for a given state group (i.e. a subset of the keys for a + # given dict which is an entry in the cache for a given state group ID). + # + # However, this poses problems when performing complicated queries + # on the store - for instance: "give me all the state for this group, but + # limit members to this subset of users", as DictionaryCache's API isn't + # rich enough to say "please cache any of these fields, apart from this subset". + # This is problematic when lazy loading members, which requires this behaviour, + # as without it the cache has no choice but to speculatively load all + # state events for the group, which negates the efficiency being sought. + # + # Rather than overcomplicating DictionaryCache's API, we instead split the + # state_group_cache into two halves - one for tracking non-member events, + # and the other for tracking member_events. This means that lazy loading + # queries can be made in a cache-friendly manner by querying both caches + # separately and then merging the result. So for the example above, you + # would query the members cache for a specific subset of state keys + # (which DictionaryCache will handle efficiently and fine) and the non-members + # cache for all state (which DictionaryCache will similarly handle fine) + # and then just merge the results together. + # + # We size the non-members cache to be smaller than the members cache as the + # vast majority of state in Matrix (today) is member events. + + self._state_group_cache = DictionaryCache( + "*stateGroupCache*", + # TODO: this hasn't been tuned yet + 50000 * get_cache_factor_for("stateGroupCache"), + ) + self._state_group_members_cache = DictionaryCache( + "*stateGroupMembersCache*", + 500000 * get_cache_factor_for("stateGroupMembersCache"), + ) + + @cached(max_entries=10000, iterable=True) + def get_state_group_delta(self, state_group): + """Given a state group try to return a previous group and a delta between + the old and the new. + + Returns: + (prev_group, delta_ids), where both may be None. + """ + + def _get_state_group_delta_txn(txn): + prev_group = self.db.simple_select_one_onecol_txn( + txn, + table="state_group_edges", + keyvalues={"state_group": state_group}, + retcol="prev_state_group", + allow_none=True, + ) + + if not prev_group: + return _GetStateGroupDelta(None, None) + + delta_ids = self.db.simple_select_list_txn( + txn, + table="state_groups_state", + keyvalues={"state_group": state_group}, + retcols=("type", "state_key", "event_id"), + ) + + return _GetStateGroupDelta( + prev_group, + {(row["type"], row["state_key"]): row["event_id"] for row in delta_ids}, + ) + + return self.db.runInteraction( + "get_state_group_delta", _get_state_group_delta_txn + ) + + @defer.inlineCallbacks + def _get_state_groups_from_groups(self, groups, state_filter): + """Returns the state groups for a given set of groups, filtering on + types of state events. + + Args: + groups(list[int]): list of state group IDs to query + state_filter (StateFilter): The state filter used to fetch state + from the database. + Returns: + Deferred[dict[int, dict[tuple[str, str], str]]]: + dict of state_group_id -> (dict of (type, state_key) -> event id) + """ + results = {} + + chunks = [groups[i : i + 100] for i in range(0, len(groups), 100)] + for chunk in chunks: + res = yield self.db.runInteraction( + "_get_state_groups_from_groups", + self._get_state_groups_from_groups_txn, + chunk, + state_filter, + ) + results.update(res) + + return results + + def _get_state_for_group_using_cache(self, cache, group, state_filter): + """Checks if group is in cache. See `_get_state_for_groups` + + Args: + cache(DictionaryCache): the state group cache to use + group(int): The state group to lookup + state_filter (StateFilter): The state filter used to fetch state + from the database. + + Returns 2-tuple (`state_dict`, `got_all`). + `got_all` is a bool indicating if we successfully retrieved all + requests state from the cache, if False we need to query the DB for the + missing state. + """ + is_all, known_absent, state_dict_ids = cache.get(group) + + if is_all or state_filter.is_full(): + # Either we have everything or want everything, either way + # `is_all` tells us whether we've gotten everything. + return state_filter.filter_state(state_dict_ids), is_all + + # tracks whether any of our requested types are missing from the cache + missing_types = False + + if state_filter.has_wildcards(): + # We don't know if we fetched all the state keys for the types in + # the filter that are wildcards, so we have to assume that we may + # have missed some. + missing_types = True + else: + # There aren't any wild cards, so `concrete_types()` returns the + # complete list of event types we're wanting. + for key in state_filter.concrete_types(): + if key not in state_dict_ids and key not in known_absent: + missing_types = True + break + + return state_filter.filter_state(state_dict_ids), not missing_types + + @defer.inlineCallbacks + def _get_state_for_groups(self, groups, state_filter=StateFilter.all()): + """Gets the state at each of a list of state groups, optionally + filtering by type/state_key + + Args: + groups (iterable[int]): list of state groups for which we want + to get the state. + state_filter (StateFilter): The state filter used to fetch state + from the database. + Returns: + Deferred[dict[int, dict[tuple[str, str], str]]]: + dict of state_group_id -> (dict of (type, state_key) -> event id) + """ + + member_filter, non_member_filter = state_filter.get_member_split() + + # Now we look them up in the member and non-member caches + ( + non_member_state, + incomplete_groups_nm, + ) = yield self._get_state_for_groups_using_cache( + groups, self._state_group_cache, state_filter=non_member_filter + ) + + ( + member_state, + incomplete_groups_m, + ) = yield self._get_state_for_groups_using_cache( + groups, self._state_group_members_cache, state_filter=member_filter + ) + + state = dict(non_member_state) + for group in groups: + state[group].update(member_state[group]) + + # Now fetch any missing groups from the database + + incomplete_groups = incomplete_groups_m | incomplete_groups_nm + + if not incomplete_groups: + return state + + cache_sequence_nm = self._state_group_cache.sequence + cache_sequence_m = self._state_group_members_cache.sequence + + # Help the cache hit ratio by expanding the filter a bit + db_state_filter = state_filter.return_expanded() + + group_to_state_dict = yield self._get_state_groups_from_groups( + list(incomplete_groups), state_filter=db_state_filter + ) + + # Now lets update the caches + self._insert_into_cache( + group_to_state_dict, + db_state_filter, + cache_seq_num_members=cache_sequence_m, + cache_seq_num_non_members=cache_sequence_nm, + ) + + # And finally update the result dict, by filtering out any extra + # stuff we pulled out of the database. + for group, group_state_dict in iteritems(group_to_state_dict): + # We just replace any existing entries, as we will have loaded + # everything we need from the database anyway. + state[group] = state_filter.filter_state(group_state_dict) + + return state + + def _get_state_for_groups_using_cache(self, groups, cache, state_filter): + """Gets the state at each of a list of state groups, optionally + filtering by type/state_key, querying from a specific cache. + + Args: + groups (iterable[int]): list of state groups for which we want + to get the state. + cache (DictionaryCache): the cache of group ids to state dicts which + we will pass through - either the normal state cache or the specific + members state cache. + state_filter (StateFilter): The state filter used to fetch state + from the database. + + Returns: + tuple[dict[int, dict[tuple[str, str], str]], set[int]]: Tuple of + dict of state_group_id -> (dict of (type, state_key) -> event id) + of entries in the cache, and the state group ids either missing + from the cache or incomplete. + """ + results = {} + incomplete_groups = set() + for group in set(groups): + state_dict_ids, got_all = self._get_state_for_group_using_cache( + cache, group, state_filter + ) + results[group] = state_dict_ids + + if not got_all: + incomplete_groups.add(group) + + return results, incomplete_groups + + def _insert_into_cache( + self, + group_to_state_dict, + state_filter, + cache_seq_num_members, + cache_seq_num_non_members, + ): + """Inserts results from querying the database into the relevant cache. + + Args: + group_to_state_dict (dict): The new entries pulled from database. + Map from state group to state dict + state_filter (StateFilter): The state filter used to fetch state + from the database. + cache_seq_num_members (int): Sequence number of member cache since + last lookup in cache + cache_seq_num_non_members (int): Sequence number of member cache since + last lookup in cache + """ + + # We need to work out which types we've fetched from the DB for the + # member vs non-member caches. This should be as accurate as possible, + # but can be an underestimate (e.g. when we have wild cards) + + member_filter, non_member_filter = state_filter.get_member_split() + if member_filter.is_full(): + # We fetched all member events + member_types = None + else: + # `concrete_types()` will only return a subset when there are wild + # cards in the filter, but that's fine. + member_types = member_filter.concrete_types() + + if non_member_filter.is_full(): + # We fetched all non member events + non_member_types = None + else: + non_member_types = non_member_filter.concrete_types() + + for group, group_state_dict in iteritems(group_to_state_dict): + state_dict_members = {} + state_dict_non_members = {} + + for k, v in iteritems(group_state_dict): + if k[0] == EventTypes.Member: + state_dict_members[k] = v + else: + state_dict_non_members[k] = v + + self._state_group_members_cache.update( + cache_seq_num_members, + key=group, + value=state_dict_members, + fetched_keys=member_types, + ) + + self._state_group_cache.update( + cache_seq_num_non_members, + key=group, + value=state_dict_non_members, + fetched_keys=non_member_types, + ) + + def store_state_group( + self, event_id, room_id, prev_group, delta_ids, current_state_ids + ): + """Store a new set of state, returning a newly assigned state group. + + Args: + event_id (str): The event ID for which the state was calculated + room_id (str) + prev_group (int|None): A previous state group for the room, optional. + delta_ids (dict|None): The delta between state at `prev_group` and + `current_state_ids`, if `prev_group` was given. Same format as + `current_state_ids`. + current_state_ids (dict): The state to store. Map of (type, state_key) + to event_id. + + Returns: + Deferred[int]: The state group ID + """ + + def _store_state_group_txn(txn): + if current_state_ids is None: + # AFAIK, this can never happen + raise Exception("current_state_ids cannot be None") + + state_group = self.database_engine.get_next_state_group_id(txn) + + self.db.simple_insert_txn( + txn, + table="state_groups", + values={"id": state_group, "room_id": room_id, "event_id": event_id}, + ) + + # We persist as a delta if we can, while also ensuring the chain + # of deltas isn't tooo long, as otherwise read performance degrades. + if prev_group: + is_in_db = self.db.simple_select_one_onecol_txn( + txn, + table="state_groups", + keyvalues={"id": prev_group}, + retcol="id", + allow_none=True, + ) + if not is_in_db: + raise Exception( + "Trying to persist state with unpersisted prev_group: %r" + % (prev_group,) + ) + + potential_hops = self._count_state_group_hops_txn(txn, prev_group) + if prev_group and potential_hops < MAX_STATE_DELTA_HOPS: + self.db.simple_insert_txn( + txn, + table="state_group_edges", + values={"state_group": state_group, "prev_state_group": prev_group}, + ) + + self.db.simple_insert_many_txn( + txn, + table="state_groups_state", + values=[ + { + "state_group": state_group, + "room_id": room_id, + "type": key[0], + "state_key": key[1], + "event_id": state_id, + } + for key, state_id in iteritems(delta_ids) + ], + ) + else: + self.db.simple_insert_many_txn( + txn, + table="state_groups_state", + values=[ + { + "state_group": state_group, + "room_id": room_id, + "type": key[0], + "state_key": key[1], + "event_id": state_id, + } + for key, state_id in iteritems(current_state_ids) + ], + ) + + # Prefill the state group caches with this group. + # It's fine to use the sequence like this as the state group map + # is immutable. (If the map wasn't immutable then this prefill could + # race with another update) + + current_member_state_ids = { + s: ev + for (s, ev) in iteritems(current_state_ids) + if s[0] == EventTypes.Member + } + txn.call_after( + self._state_group_members_cache.update, + self._state_group_members_cache.sequence, + key=state_group, + value=dict(current_member_state_ids), + ) + + current_non_member_state_ids = { + s: ev + for (s, ev) in iteritems(current_state_ids) + if s[0] != EventTypes.Member + } + txn.call_after( + self._state_group_cache.update, + self._state_group_cache.sequence, + key=state_group, + value=dict(current_non_member_state_ids), + ) + + return state_group + + return self.db.runInteraction("store_state_group", _store_state_group_txn) + + def purge_unreferenced_state_groups( + self, room_id: str, state_groups_to_delete + ) -> defer.Deferred: + """Deletes no longer referenced state groups and de-deltas any state + groups that reference them. + + Args: + room_id: The room the state groups belong to (must all be in the + same room). + state_groups_to_delete (Collection[int]): Set of all state groups + to delete. + """ + + return self.db.runInteraction( + "purge_unreferenced_state_groups", + self._purge_unreferenced_state_groups, + room_id, + state_groups_to_delete, + ) + + def _purge_unreferenced_state_groups(self, txn, room_id, state_groups_to_delete): + logger.info( + "[purge] found %i state groups to delete", len(state_groups_to_delete) + ) + + rows = self.db.simple_select_many_txn( + txn, + table="state_group_edges", + column="prev_state_group", + iterable=state_groups_to_delete, + keyvalues={}, + retcols=("state_group",), + ) + + remaining_state_groups = set( + row["state_group"] + for row in rows + if row["state_group"] not in state_groups_to_delete + ) + + logger.info( + "[purge] de-delta-ing %i remaining state groups", + len(remaining_state_groups), + ) + + # Now we turn the state groups that reference to-be-deleted state + # groups to non delta versions. + for sg in remaining_state_groups: + logger.info("[purge] de-delta-ing remaining state group %s", sg) + curr_state = self._get_state_groups_from_groups_txn(txn, [sg]) + curr_state = curr_state[sg] + + self.db.simple_delete_txn( + txn, table="state_groups_state", keyvalues={"state_group": sg} + ) + + self.db.simple_delete_txn( + txn, table="state_group_edges", keyvalues={"state_group": sg} + ) + + self.db.simple_insert_many_txn( + txn, + table="state_groups_state", + values=[ + { + "state_group": sg, + "room_id": room_id, + "type": key[0], + "state_key": key[1], + "event_id": state_id, + } + for key, state_id in iteritems(curr_state) + ], + ) + + logger.info("[purge] removing redundant state groups") + txn.executemany( + "DELETE FROM state_groups_state WHERE state_group = ?", + ((sg,) for sg in state_groups_to_delete), + ) + txn.executemany( + "DELETE FROM state_groups WHERE id = ?", + ((sg,) for sg in state_groups_to_delete), + ) + + @defer.inlineCallbacks + def get_previous_state_groups(self, state_groups): + """Fetch the previous groups of the given state groups. + + Args: + state_groups (Iterable[int]) + + Returns: + Deferred[dict[int, int]]: mapping from state group to previous + state group. + """ + + rows = yield self.db.simple_select_many_batch( + table="state_group_edges", + column="prev_state_group", + iterable=state_groups, + keyvalues={}, + retcols=("prev_state_group", "state_group"), + desc="get_previous_state_groups", + ) + + return {row["state_group"]: row["prev_state_group"] for row in rows} + + def purge_room_state(self, room_id, state_groups_to_delete): + """Deletes all record of a room from state tables + + Args: + room_id (str): + state_groups_to_delete (list[int]): State groups to delete + """ + + return self.db.runInteraction( + "purge_room_state", + self._purge_room_state_txn, + room_id, + state_groups_to_delete, + ) + + def _purge_room_state_txn(self, txn, room_id, state_groups_to_delete): + # first we have to delete the state groups states + logger.info("[purge] removing %s from state_groups_state", room_id) + + self.db.simple_delete_many_txn( + txn, + table="state_groups_state", + column="state_group", + iterable=state_groups_to_delete, + keyvalues={}, + ) + + # ... and the state group edges + logger.info("[purge] removing %s from state_group_edges", room_id) + + self.db.simple_delete_many_txn( + txn, + table="state_group_edges", + column="state_group", + iterable=state_groups_to_delete, + keyvalues={}, + ) + + # ... and the state groups + logger.info("[purge] removing %s from state_groups", room_id) + + self.db.simple_delete_many_txn( + txn, + table="state_groups", + column="id", + iterable=state_groups_to_delete, + keyvalues={}, + ) diff --git a/synapse/storage/persist_events.py b/synapse/storage/persist_events.py index fa03ca9ff7..1ed44925fc 100644 --- a/synapse/storage/persist_events.py +++ b/synapse/storage/persist_events.py @@ -183,7 +183,7 @@ class EventsPersistenceStorage(object): # so we use separate variables here even though they point to the same # store for now. self.main_store = stores.main - self.state_store = stores.main + self.state_store = stores.state self._clock = hs.get_clock() self.is_mine_id = hs.is_mine_id diff --git a/synapse/storage/prepare_database.py b/synapse/storage/prepare_database.py index 403848ad03..e70026b80a 100644 --- a/synapse/storage/prepare_database.py +++ b/synapse/storage/prepare_database.py @@ -42,7 +42,7 @@ class UpgradeDatabaseException(PrepareDatabaseException): pass -def prepare_database(db_conn, database_engine, config, data_stores=["main"]): +def prepare_database(db_conn, database_engine, config, data_stores=["main", "state"]): """Prepares a database for usage. Will either create all necessary tables or upgrade from an older schema version. diff --git a/synapse/storage/purge_events.py b/synapse/storage/purge_events.py index a368182034..d6a7bd7834 100644 --- a/synapse/storage/purge_events.py +++ b/synapse/storage/purge_events.py @@ -58,7 +58,7 @@ class PurgeEventsStorage(object): sg_to_delete = yield self._find_unreferenced_groups(state_groups) - yield self.stores.main.purge_unreferenced_state_groups(room_id, sg_to_delete) + yield self.stores.state.purge_unreferenced_state_groups(room_id, sg_to_delete) @defer.inlineCallbacks def _find_unreferenced_groups(self, state_groups): @@ -102,7 +102,7 @@ class PurgeEventsStorage(object): # groups that are referenced. current_search -= referenced - edges = yield self.stores.main.get_previous_state_groups(current_search) + edges = yield self.stores.state.get_previous_state_groups(current_search) prevs = set(edges.values()) # We don't bother re-handling groups we've already seen diff --git a/synapse/storage/state.py b/synapse/storage/state.py index 3735846899..cbeb586014 100644 --- a/synapse/storage/state.py +++ b/synapse/storage/state.py @@ -342,7 +342,7 @@ class StateGroupStorage(object): (prev_group, delta_ids) """ - return self.stores.main.get_state_group_delta(state_group) + return self.stores.state.get_state_group_delta(state_group) @defer.inlineCallbacks def get_state_groups_ids(self, _room_id, event_ids): @@ -362,7 +362,7 @@ class StateGroupStorage(object): event_to_groups = yield self.stores.main._get_state_group_for_events(event_ids) groups = set(itervalues(event_to_groups)) - group_to_state = yield self.stores.main._get_state_for_groups(groups) + group_to_state = yield self.stores.state._get_state_for_groups(groups) return group_to_state @@ -423,7 +423,7 @@ class StateGroupStorage(object): dict of state_group_id -> (dict of (type, state_key) -> event id) """ - return self.stores.main._get_state_groups_from_groups(groups, state_filter) + return self.stores.state._get_state_groups_from_groups(groups, state_filter) @defer.inlineCallbacks def get_state_for_events(self, event_ids, state_filter=StateFilter.all()): @@ -439,7 +439,7 @@ class StateGroupStorage(object): event_to_groups = yield self.stores.main._get_state_group_for_events(event_ids) groups = set(itervalues(event_to_groups)) - group_to_state = yield self.stores.main._get_state_for_groups( + group_to_state = yield self.stores.state._get_state_for_groups( groups, state_filter ) @@ -476,7 +476,7 @@ class StateGroupStorage(object): event_to_groups = yield self.stores.main._get_state_group_for_events(event_ids) groups = set(itervalues(event_to_groups)) - group_to_state = yield self.stores.main._get_state_for_groups( + group_to_state = yield self.stores.state._get_state_for_groups( groups, state_filter ) @@ -532,7 +532,7 @@ class StateGroupStorage(object): Deferred[dict[int, dict[tuple[str, str], str]]]: dict of state_group_id -> (dict of (type, state_key) -> event id) """ - return self.stores.main._get_state_for_groups(groups, state_filter) + return self.stores.state._get_state_for_groups(groups, state_filter) def store_state_group( self, event_id, room_id, prev_group, delta_ids, current_state_ids @@ -552,6 +552,6 @@ class StateGroupStorage(object): Returns: Deferred[int]: The state group ID """ - return self.stores.main.store_state_group( + return self.stores.state.store_state_group( event_id, room_id, prev_group, delta_ids, current_state_ids ) diff --git a/tests/storage/test_state.py b/tests/storage/test_state.py index 43200654f1..d6ecf102f8 100644 --- a/tests/storage/test_state.py +++ b/tests/storage/test_state.py @@ -35,7 +35,7 @@ class StateStoreTestCase(tests.unittest.TestCase): self.store = hs.get_datastore() self.storage = hs.get_storage() - self.state_datastore = self.store + self.state_datastore = self.storage.state.stores.state self.event_builder_factory = hs.get_event_builder_factory() self.event_creation_handler = hs.get_event_creation_handler() diff --git a/tests/utils.py b/tests/utils.py index 9f5bf40b4b..e2e9cafd79 100644 --- a/tests/utils.py +++ b/tests/utils.py @@ -231,7 +231,7 @@ def setup_test_homeserver( "args": {"database": ":memory:", "cp_min": 1, "cp_max": 1}, } - database = DatabaseConnectionConfig("master", database_config, ["main"]) + database = DatabaseConnectionConfig("master", database_config) config.database.databases = [database] db_engine = create_engine(database.config) -- cgit 1.4.1 From b6b57ecb4e845490fc26a537ff57df8cae1587b9 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Fri, 3 Jan 2020 14:19:48 +0000 Subject: Kill off redundant SynapseRequestFactory (#6619) We already get the Site via the Channel, so there's no need for a dedicated RequestFactory: we can just use the right constructor. --- changelog.d/6619.misc | 1 + synapse/http/site.py | 18 +++--------------- tests/server.py | 6 ++++-- 3 files changed, 8 insertions(+), 17 deletions(-) create mode 100644 changelog.d/6619.misc (limited to 'tests') diff --git a/changelog.d/6619.misc b/changelog.d/6619.misc new file mode 100644 index 0000000000..b608133219 --- /dev/null +++ b/changelog.d/6619.misc @@ -0,0 +1 @@ +Simplify http handling by removing redundant SynapseRequestFactory. diff --git a/synapse/http/site.py b/synapse/http/site.py index ff8184a3d0..9f2d035fa0 100644 --- a/synapse/http/site.py +++ b/synapse/http/site.py @@ -47,9 +47,9 @@ class SynapseRequest(Request): logcontext(LoggingContext) : the log context for this request """ - def __init__(self, site, channel, *args, **kw): + def __init__(self, channel, *args, **kw): Request.__init__(self, channel, *args, **kw) - self.site = site + self.site = channel.site self._channel = channel # this is used by the tests self.authenticated_entity = None self.start_time = 0 @@ -331,18 +331,6 @@ class XForwardedForRequest(SynapseRequest): ) -class SynapseRequestFactory(object): - def __init__(self, site, x_forwarded_for): - self.site = site - self.x_forwarded_for = x_forwarded_for - - def __call__(self, *args, **kwargs): - if self.x_forwarded_for: - return XForwardedForRequest(self.site, *args, **kwargs) - else: - return SynapseRequest(self.site, *args, **kwargs) - - class SynapseSite(Site): """ Subclass of a twisted http Site that does access logging with python's @@ -364,7 +352,7 @@ class SynapseSite(Site): self.site_tag = site_tag proxied = config.get("x_forwarded", False) - self.requestFactory = SynapseRequestFactory(self, proxied) + self.requestFactory = XForwardedForRequest if proxied else SynapseRequest self.access_logger = logging.getLogger(logger_name) self.server_version_string = server_version_string.encode("ascii") diff --git a/tests/server.py b/tests/server.py index a554dfdd57..1644710aa0 100644 --- a/tests/server.py +++ b/tests/server.py @@ -20,6 +20,7 @@ from twisted.python.failure import Failure from twisted.test.proto_helpers import AccumulatingProtocol, MemoryReactorClock from twisted.web.http import unquote from twisted.web.http_headers import Headers +from twisted.web.server import Site from synapse.http.site import SynapseRequest from synapse.util import Clock @@ -42,6 +43,7 @@ class FakeChannel(object): wire). """ + site = attr.ib(type=Site) _reactor = attr.ib() result = attr.ib(default=attr.Factory(dict)) _producer = None @@ -176,9 +178,9 @@ def make_request( content = content.encode("utf8") site = FakeSite() - channel = FakeChannel(reactor) + channel = FakeChannel(site, reactor) - req = request(site, channel) + req = request(channel) req.process = lambda: b"" req.content = BytesIO(content) req.postpath = list(map(unquote, path[1:].split(b"/"))) -- cgit 1.4.1 From 18674eebb1fa5d7445952d7e201afe33bd040523 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Mon, 6 Jan 2020 12:28:58 +0000 Subject: Workaround for error when fetching notary's own key (#6620) * Kill off redundant SynapseRequestFactory We already get the Site via the Channel, so there's no need for a dedicated RequestFactory: we can just use the right constructor. * Workaround for error when fetching notary's own key As a notary server, when we return our own keys, include all of our signing keys in verify_keys. This is a workaround for #6596. --- changelog.d/6620.misc | 1 + synapse/rest/key/v2/remote_key_resource.py | 30 ++++-- tests/rest/key/v2/test_remote_key_resource.py | 130 ++++++++++++++++++++++++++ tests/unittest.py | 11 ++- 4 files changed, 163 insertions(+), 9 deletions(-) create mode 100644 changelog.d/6620.misc create mode 100644 tests/rest/key/v2/test_remote_key_resource.py (limited to 'tests') diff --git a/changelog.d/6620.misc b/changelog.d/6620.misc new file mode 100644 index 0000000000..8bfb78fb20 --- /dev/null +++ b/changelog.d/6620.misc @@ -0,0 +1 @@ +Add a workaround for synapse raising exceptions when fetching the notary's own key from the notary. diff --git a/synapse/rest/key/v2/remote_key_resource.py b/synapse/rest/key/v2/remote_key_resource.py index e7fc3f0431..bf5e0eb844 100644 --- a/synapse/rest/key/v2/remote_key_resource.py +++ b/synapse/rest/key/v2/remote_key_resource.py @@ -15,6 +15,7 @@ import logging from canonicaljson import encode_canonical_json, json +from signedjson.key import encode_verify_key_base64 from signedjson.sign import sign_json from twisted.internet import defer @@ -216,15 +217,28 @@ class RemoteKey(DirectServeResource): if cache_misses and query_remote_on_cache_miss: yield self.fetcher.get_keys(cache_misses) yield self.query_keys(request, query, query_remote_on_cache_miss=False) - else: - signed_keys = [] - for key_json in json_results: - key_json = json.loads(key_json) + return + + signed_keys = [] + for key_json in json_results: + key_json = json.loads(key_json) + + # backwards-compatibility hack for #6596: if the requested key belongs + # to us, make sure that all of the signing keys appear in the + # "verify_keys" section. + if key_json["server_name"] == self.config.server_name: + verify_keys = key_json["verify_keys"] for signing_key in self.config.key_server_signing_keys: - key_json = sign_json(key_json, self.config.server_name, signing_key) + key_id = "%s:%s" % (signing_key.alg, signing_key.version) + verify_keys[key_id] = { + "key": encode_verify_key_base64(signing_key.verify_key) + } + + for signing_key in self.config.key_server_signing_keys: + key_json = sign_json(key_json, self.config.server_name, signing_key) - signed_keys.append(key_json) + signed_keys.append(key_json) - results = {"server_keys": signed_keys} + results = {"server_keys": signed_keys} - respond_with_json_bytes(request, 200, encode_canonical_json(results)) + respond_with_json_bytes(request, 200, encode_canonical_json(results)) diff --git a/tests/rest/key/v2/test_remote_key_resource.py b/tests/rest/key/v2/test_remote_key_resource.py new file mode 100644 index 0000000000..d8246b4e78 --- /dev/null +++ b/tests/rest/key/v2/test_remote_key_resource.py @@ -0,0 +1,130 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import urllib.parse +from io import BytesIO + +from mock import Mock + +import signedjson.key +from nacl.signing import SigningKey +from signedjson.sign import sign_json + +from twisted.web.resource import NoResource + +from synapse.http.site import SynapseRequest +from synapse.rest.key.v2 import KeyApiV2Resource +from synapse.util.httpresourcetree import create_resource_tree + +from tests import unittest +from tests.server import FakeChannel, wait_until_result + + +class RemoteKeyResourceTestCase(unittest.HomeserverTestCase): + def make_homeserver(self, reactor, clock): + self.http_client = Mock() + return self.setup_test_homeserver(http_client=self.http_client) + + def create_test_json_resource(self): + return create_resource_tree( + {"/_matrix/key/v2": KeyApiV2Resource(self.hs)}, root_resource=NoResource() + ) + + def expect_outgoing_key_request( + self, server_name: str, signing_key: SigningKey + ) -> None: + """ + Tell the mock http client to expect an outgoing GET request for the given key + """ + + def get_json(destination, path, ignore_backoff=False, **kwargs): + self.assertTrue(ignore_backoff) + self.assertEqual(destination, server_name) + key_id = "%s:%s" % (signing_key.alg, signing_key.version) + self.assertEqual( + path, "/_matrix/key/v2/server/%s" % (urllib.parse.quote(key_id),) + ) + + response = { + "server_name": server_name, + "old_verify_keys": {}, + "valid_until_ts": 200 * 1000, + "verify_keys": { + key_id: { + "key": signedjson.key.encode_verify_key_base64( + signing_key.verify_key + ) + } + }, + } + sign_json(response, server_name, signing_key) + return response + + self.http_client.get_json.side_effect = get_json + + def make_notary_request(self, server_name: str, key_id: str) -> dict: + """Send a GET request to the test server requesting the given key. + + Checks that the response is a 200 and returns the decoded json body. + """ + channel = FakeChannel(self.site, self.reactor) + req = SynapseRequest(channel) + req.content = BytesIO(b"") + req.requestReceived( + b"GET", + b"/_matrix/key/v2/query/%s/%s" + % (server_name.encode("utf-8"), key_id.encode("utf-8")), + b"1.1", + ) + wait_until_result(self.reactor, req) + self.assertEqual(channel.code, 200) + resp = channel.json_body + return resp + + def test_get_key(self): + """Fetch a remote key""" + SERVER_NAME = "remote.server" + testkey = signedjson.key.generate_signing_key("ver1") + self.expect_outgoing_key_request(SERVER_NAME, testkey) + + resp = self.make_notary_request(SERVER_NAME, "ed25519:ver1") + keys = resp["server_keys"] + self.assertEqual(len(keys), 1) + + self.assertIn("ed25519:ver1", keys[0]["verify_keys"]) + self.assertEqual(len(keys[0]["verify_keys"]), 1) + + # it should be signed by both the origin server and the notary + self.assertIn(SERVER_NAME, keys[0]["signatures"]) + self.assertIn(self.hs.hostname, keys[0]["signatures"]) + + def test_get_own_key(self): + """Fetch our own key""" + testkey = signedjson.key.generate_signing_key("ver1") + self.expect_outgoing_key_request(self.hs.hostname, testkey) + + resp = self.make_notary_request(self.hs.hostname, "ed25519:ver1") + keys = resp["server_keys"] + self.assertEqual(len(keys), 1) + + # it should be signed by both itself, and the notary signing key + sigs = keys[0]["signatures"] + self.assertEqual(len(sigs), 1) + self.assertIn(self.hs.hostname, sigs) + oursigs = sigs[self.hs.hostname] + self.assertEqual(len(oursigs), 2) + + # and both keys should be present in the verify_keys section + self.assertIn("ed25519:ver1", keys[0]["verify_keys"]) + self.assertIn("ed25519:a_lPym", keys[0]["verify_keys"]) diff --git a/tests/unittest.py b/tests/unittest.py index b30b7d1718..cbda237278 100644 --- a/tests/unittest.py +++ b/tests/unittest.py @@ -36,7 +36,7 @@ from synapse.config.homeserver import HomeServerConfig from synapse.config.ratelimiting import FederationRateLimitConfig from synapse.federation.transport import server as federation_server from synapse.http.server import JsonResource -from synapse.http.site import SynapseRequest +from synapse.http.site import SynapseRequest, SynapseSite from synapse.logging.context import LoggingContext from synapse.server import HomeServer from synapse.types import Requester, UserID, create_requester @@ -210,6 +210,15 @@ class HomeserverTestCase(TestCase): # Register the resources self.resource = self.create_test_json_resource() + # create a site to wrap the resource. + self.site = SynapseSite( + logger_name="synapse.access.http.fake", + site_tag="test", + config={}, + resource=self.resource, + server_version_string="1", + ) + from tests.rest.client.v1.utils import RestHelper self.helper = RestHelper(self.hs, self.resource, getattr(self, "user_id", None)) -- cgit 1.4.1 From 4b36b482e0cc1a63db27534c4ea5d9608cdb6a79 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Mon, 6 Jan 2020 12:33:56 +0000 Subject: Fix exception when fetching notary server's old keys (#6625) Lift the restriction that *all* the keys used for signing v2 key responses be present in verify_keys. Fixes #6596. --- changelog.d/6625.bugfix | 1 + synapse/crypto/keyring.py | 13 ++-- tests/crypto/test_keyring.py | 139 +++++++++++++++++++++++++++++-------------- 3 files changed, 103 insertions(+), 50 deletions(-) create mode 100644 changelog.d/6625.bugfix (limited to 'tests') diff --git a/changelog.d/6625.bugfix b/changelog.d/6625.bugfix new file mode 100644 index 0000000000..a8dc5587dc --- /dev/null +++ b/changelog.d/6625.bugfix @@ -0,0 +1 @@ +Fix exception when fetching the `matrix.org:ed25519:auto` key. diff --git a/synapse/crypto/keyring.py b/synapse/crypto/keyring.py index 7cfad192e8..6fe5a6a26a 100644 --- a/synapse/crypto/keyring.py +++ b/synapse/crypto/keyring.py @@ -511,17 +511,18 @@ class BaseV2KeyFetcher(object): server_name = response_json["server_name"] verified = False for key_id in response_json["signatures"].get(server_name, {}): - # each of the keys used for the signature must be present in the response - # json. key = verify_keys.get(key_id) if not key: - raise KeyLookupError( - "Key response is signed by key id %s:%s but that key is not " - "present in the response" % (server_name, key_id) - ) + # the key may not be present in verify_keys if: + # * we got the key from the notary server, and: + # * the key belongs to the notary server, and: + # * the notary server is using a different key to sign notary + # responses. + continue verify_signed_json(response_json, server_name, key.verify_key) verified = True + break if not verified: raise KeyLookupError( diff --git a/tests/crypto/test_keyring.py b/tests/crypto/test_keyring.py index 8efd39c7f7..34d5895f18 100644 --- a/tests/crypto/test_keyring.py +++ b/tests/crypto/test_keyring.py @@ -19,6 +19,7 @@ from mock import Mock import canonicaljson import signedjson.key import signedjson.sign +from nacl.signing import SigningKey from signedjson.key import encode_verify_key_base64, get_verify_key from twisted.internet import defer @@ -412,34 +413,37 @@ class PerspectivesKeyFetcherTestCase(unittest.HomeserverTestCase): handlers=None, http_client=self.http_client, config=config ) - def test_get_keys_from_perspectives(self): - # arbitrarily advance the clock a bit - self.reactor.advance(100) - - fetcher = PerspectivesKeyFetcher(self.hs) - - SERVER_NAME = "server2" - testkey = signedjson.key.generate_signing_key("ver1") - testverifykey = signedjson.key.get_verify_key(testkey) - testverifykey_id = "ed25519:ver1" - VALID_UNTIL_TS = 200 * 1000 + def build_perspectives_response( + self, server_name: str, signing_key: SigningKey, valid_until_ts: int, + ) -> dict: + """ + Build a valid perspectives server response to a request for the given key + """ + verify_key = signedjson.key.get_verify_key(signing_key) + verifykey_id = "%s:%s" % (verify_key.alg, verify_key.version) - # valid response response = { - "server_name": SERVER_NAME, + "server_name": server_name, "old_verify_keys": {}, - "valid_until_ts": VALID_UNTIL_TS, + "valid_until_ts": valid_until_ts, "verify_keys": { - testverifykey_id: { - "key": signedjson.key.encode_verify_key_base64(testverifykey) + verifykey_id: { + "key": signedjson.key.encode_verify_key_base64(verify_key) } }, } - # the response must be signed by both the origin server and the perspectives # server. - signedjson.sign.sign_json(response, SERVER_NAME, testkey) + signedjson.sign.sign_json(response, server_name, signing_key) self.mock_perspective_server.sign_response(response) + return response + + def expect_outgoing_key_query( + self, expected_server_name: str, expected_key_id: str, response: dict + ) -> None: + """ + Tell the mock http client to expect a perspectives-server key query + """ def post_json(destination, path, data, **kwargs): self.assertEqual(destination, self.mock_perspective_server.server_name) @@ -447,11 +451,79 @@ class PerspectivesKeyFetcherTestCase(unittest.HomeserverTestCase): # check that the request is for the expected key q = data["server_keys"] - self.assertEqual(list(q[SERVER_NAME].keys()), ["key1"]) + self.assertEqual(list(q[expected_server_name].keys()), [expected_key_id]) return {"server_keys": [response]} self.http_client.post_json.side_effect = post_json + def test_get_keys_from_perspectives(self): + # arbitrarily advance the clock a bit + self.reactor.advance(100) + + fetcher = PerspectivesKeyFetcher(self.hs) + + SERVER_NAME = "server2" + testkey = signedjson.key.generate_signing_key("ver1") + testverifykey = signedjson.key.get_verify_key(testkey) + testverifykey_id = "ed25519:ver1" + VALID_UNTIL_TS = 200 * 1000 + + response = self.build_perspectives_response( + SERVER_NAME, testkey, VALID_UNTIL_TS, + ) + + self.expect_outgoing_key_query(SERVER_NAME, "key1", response) + + keys_to_fetch = {SERVER_NAME: {"key1": 0}} + keys = self.get_success(fetcher.get_keys(keys_to_fetch)) + self.assertIn(SERVER_NAME, keys) + k = keys[SERVER_NAME][testverifykey_id] + self.assertEqual(k.valid_until_ts, VALID_UNTIL_TS) + self.assertEqual(k.verify_key, testverifykey) + self.assertEqual(k.verify_key.alg, "ed25519") + self.assertEqual(k.verify_key.version, "ver1") + + # check that the perspectives store is correctly updated + lookup_triplet = (SERVER_NAME, testverifykey_id, None) + key_json = self.get_success( + self.hs.get_datastore().get_server_keys_json([lookup_triplet]) + ) + res = key_json[lookup_triplet] + self.assertEqual(len(res), 1) + res = res[0] + self.assertEqual(res["key_id"], testverifykey_id) + self.assertEqual(res["from_server"], self.mock_perspective_server.server_name) + self.assertEqual(res["ts_added_ms"], self.reactor.seconds() * 1000) + self.assertEqual(res["ts_valid_until_ms"], VALID_UNTIL_TS) + + self.assertEqual( + bytes(res["key_json"]), canonicaljson.encode_canonical_json(response) + ) + + def test_get_perspectives_own_key(self): + """Check that we can get the perspectives server's own keys + + This is slightly complicated by the fact that the perspectives server may + use different keys for signing notary responses. + """ + + # arbitrarily advance the clock a bit + self.reactor.advance(100) + + fetcher = PerspectivesKeyFetcher(self.hs) + + SERVER_NAME = self.mock_perspective_server.server_name + testkey = signedjson.key.generate_signing_key("ver1") + testverifykey = signedjson.key.get_verify_key(testkey) + testverifykey_id = "ed25519:ver1" + VALID_UNTIL_TS = 200 * 1000 + + response = self.build_perspectives_response( + SERVER_NAME, testkey, VALID_UNTIL_TS + ) + + self.expect_outgoing_key_query(SERVER_NAME, "key1", response) + keys_to_fetch = {SERVER_NAME: {"key1": 0}} keys = self.get_success(fetcher.get_keys(keys_to_fetch)) self.assertIn(SERVER_NAME, keys) @@ -490,35 +562,14 @@ class PerspectivesKeyFetcherTestCase(unittest.HomeserverTestCase): VALID_UNTIL_TS = 200 * 1000 def build_response(): - # valid response - response = { - "server_name": SERVER_NAME, - "old_verify_keys": {}, - "valid_until_ts": VALID_UNTIL_TS, - "verify_keys": { - testverifykey_id: { - "key": signedjson.key.encode_verify_key_base64(testverifykey) - } - }, - } - - # the response must be signed by both the origin server and the perspectives - # server. - signedjson.sign.sign_json(response, SERVER_NAME, testkey) - self.mock_perspective_server.sign_response(response) - return response + return self.build_perspectives_response( + SERVER_NAME, testkey, VALID_UNTIL_TS + ) def get_key_from_perspectives(response): fetcher = PerspectivesKeyFetcher(self.hs) keys_to_fetch = {SERVER_NAME: {"key1": 0}} - - def post_json(destination, path, data, **kwargs): - self.assertEqual(destination, self.mock_perspective_server.server_name) - self.assertEqual(path, "/_matrix/key/v2/query") - return {"server_keys": [response]} - - self.http_client.post_json.side_effect = post_json - + self.expect_outgoing_key_query(SERVER_NAME, "key1", response) return self.get_success(fetcher.get_keys(keys_to_fetch)) # start with a valid response so we can check we are testing the right thing -- cgit 1.4.1 From 5a047816434e2ce2df8b80eb63a49c17dc3085fb Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Fri, 3 Jan 2020 15:31:09 +0000 Subject: rename get_prev_events_for_room to get_prev_events_and_hashes_for_room ... to make way for a new method which just returns the event ids --- synapse/handlers/message.py | 6 ++++-- synapse/handlers/room_member.py | 4 +++- synapse/storage/data_stores/main/event_federation.py | 5 +++-- tests/storage/test_event_federation.py | 4 ++-- 4 files changed, 12 insertions(+), 7 deletions(-) (limited to 'tests') diff --git a/synapse/handlers/message.py b/synapse/handlers/message.py index 4ad752205f..2695975a16 100644 --- a/synapse/handlers/message.py +++ b/synapse/handlers/message.py @@ -740,7 +740,7 @@ class EventCreationHandler(object): % (len(prev_events_and_hashes),) ) else: - prev_events_and_hashes = yield self.store.get_prev_events_for_room( + prev_events_and_hashes = yield self.store.get_prev_events_and_hashes_for_room( builder.room_id ) @@ -1042,7 +1042,9 @@ class EventCreationHandler(object): # For each room we need to find a joined member we can use to send # the dummy event with. - prev_events_and_hashes = yield self.store.get_prev_events_for_room(room_id) + prev_events_and_hashes = yield self.store.get_prev_events_and_hashes_for_room( + room_id + ) latest_event_ids = (event_id for (event_id, _, _) in prev_events_and_hashes) diff --git a/synapse/handlers/room_member.py b/synapse/handlers/room_member.py index 44c5e3239c..91bb34cd55 100644 --- a/synapse/handlers/room_member.py +++ b/synapse/handlers/room_member.py @@ -370,7 +370,9 @@ class RoomMemberHandler(object): if block_invite: raise SynapseError(403, "Invites have been disabled on this server") - prev_events_and_hashes = yield self.store.get_prev_events_for_room(room_id) + prev_events_and_hashes = yield self.store.get_prev_events_and_hashes_for_room( + room_id + ) latest_event_ids = (event_id for (event_id, _, _) in prev_events_and_hashes) current_state_ids = yield self.state_handler.get_current_state_ids( diff --git a/synapse/storage/data_stores/main/event_federation.py b/synapse/storage/data_stores/main/event_federation.py index 1f517e8fad..266fc9715f 100644 --- a/synapse/storage/data_stores/main/event_federation.py +++ b/synapse/storage/data_stores/main/event_federation.py @@ -149,9 +149,10 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBas ) @defer.inlineCallbacks - def get_prev_events_for_room(self, room_id): + def get_prev_events_and_hashes_for_room(self, room_id): """ - Gets a subset of the current forward extremities in the given room. + Gets a subset of the current forward extremities in the given room, + along with their depths and hashes. Limits the result to 10 extremities, so that we can avoid creating events which refer to hundreds of prev_events. diff --git a/tests/storage/test_event_federation.py b/tests/storage/test_event_federation.py index eadfb90a22..3a68bf3274 100644 --- a/tests/storage/test_event_federation.py +++ b/tests/storage/test_event_federation.py @@ -26,7 +26,7 @@ class EventFederationWorkerStoreTestCase(tests.unittest.TestCase): self.store = hs.get_datastore() @defer.inlineCallbacks - def test_get_prev_events_for_room(self): + def test_get_prev_events_and_hashes_for_room(self): room_id = "@ROOM:local" # add a bunch of events and hashes to act as forward extremities @@ -64,7 +64,7 @@ class EventFederationWorkerStoreTestCase(tests.unittest.TestCase): yield self.store.db.runInteraction("insert", insert_event, i) # this should get the last five and five others - r = yield self.store.get_prev_events_for_room(room_id) + r = yield self.store.get_prev_events_and_hashes_for_room(room_id) self.assertEqual(10, len(r)) for i in range(0, 5): el = r[i] -- cgit 1.4.1 From 3bef62488e5cff4dfb33454f2f2e18cc928f319b Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Fri, 3 Jan 2020 16:19:55 +0000 Subject: Remove unused hashes and depths from create_event params --- synapse/handlers/message.py | 21 +++++---------------- synapse/handlers/room_member.py | 8 +++++++- tests/unittest.py | 6 +----- 3 files changed, 13 insertions(+), 22 deletions(-) (limited to 'tests') diff --git a/synapse/handlers/message.py b/synapse/handlers/message.py index 5415b0c9ee..8ea3aca2f4 100644 --- a/synapse/handlers/message.py +++ b/synapse/handlers/message.py @@ -422,7 +422,7 @@ class EventCreationHandler(object): event_dict, token_id=None, txn_id=None, - prev_events_and_hashes=None, + prev_event_ids: Optional[Collection[str]] = None, require_consent=True, ): """ @@ -439,10 +439,9 @@ class EventCreationHandler(object): token_id (str) txn_id (str) - prev_events_and_hashes (list[(str, dict[str, str], int)]|None): + prev_event_ids: the forward extremities to use as the prev_events for the - new event. For each event, a tuple of (event_id, hashes, depth) - where *hashes* is a map from algorithm to hash. + new event. If None, they will be requested from the database. @@ -497,12 +496,6 @@ class EventCreationHandler(object): if txn_id is not None: builder.internal_metadata.txn_id = txn_id - prev_event_ids = ( - None - if prev_events_and_hashes is None - else [event_id for event_id, _, _ in prev_events_and_hashes] - ) - event, context = yield self.create_new_client_event( builder=builder, requester=requester, prev_event_ids=prev_event_ids, ) @@ -1038,11 +1031,7 @@ class EventCreationHandler(object): # For each room we need to find a joined member we can use to send # the dummy event with. - prev_events_and_hashes = yield self.store.get_prev_events_and_hashes_for_room( - room_id - ) - - latest_event_ids = (event_id for (event_id, _, _) in prev_events_and_hashes) + latest_event_ids = yield self.store.get_prev_events_for_room(room_id) members = yield self.state.get_current_users_in_room( room_id, latest_event_ids=latest_event_ids @@ -1061,7 +1050,7 @@ class EventCreationHandler(object): "room_id": room_id, "sender": user_id, }, - prev_events_and_hashes=prev_events_and_hashes, + prev_event_ids=latest_event_ids, ) event.internal_metadata.proactively_send = False diff --git a/synapse/handlers/room_member.py b/synapse/handlers/room_member.py index 91bb34cd55..d550ba8ab4 100644 --- a/synapse/handlers/room_member.py +++ b/synapse/handlers/room_member.py @@ -164,6 +164,12 @@ class RoomMemberHandler(object): if requester.is_guest: content["kind"] = "guest" + prev_event_ids = ( + None + if prev_events_and_hashes is None + else [event_id for event_id, _, _ in prev_events_and_hashes] + ) + event, context = yield self.event_creation_handler.create_event( requester, { @@ -177,7 +183,7 @@ class RoomMemberHandler(object): }, token_id=requester.access_token_id, txn_id=txn_id, - prev_events_and_hashes=prev_events_and_hashes, + prev_event_ids=prev_event_ids, require_consent=require_consent, ) diff --git a/tests/unittest.py b/tests/unittest.py index b30b7d1718..07b50c0ccd 100644 --- a/tests/unittest.py +++ b/tests/unittest.py @@ -522,10 +522,6 @@ class HomeserverTestCase(TestCase): secrets = self.hs.get_secrets() requester = Requester(user, None, False, None, None) - prev_events_and_hashes = None - if prev_event_ids: - prev_events_and_hashes = [[p, {}, 0] for p in prev_event_ids] - event, context = self.get_success( event_creator.create_event( requester, @@ -535,7 +531,7 @@ class HomeserverTestCase(TestCase): "sender": user.to_string(), "content": {"body": secrets.token_hex(), "msgtype": "m.text"}, }, - prev_events_and_hashes=prev_events_and_hashes, + prev_event_ids=prev_event_ids, ) ) -- cgit 1.4.1 From dc41fbf0dda981df117d8cf1938e023a38836cda Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Fri, 3 Jan 2020 16:30:51 +0000 Subject: Remove unused get_prev_events_and_hashes_for_room --- .../storage/data_stores/main/event_federation.py | 30 ---------------------- tests/storage/test_event_federation.py | 19 +++++--------- 2 files changed, 6 insertions(+), 43 deletions(-) (limited to 'tests') diff --git a/synapse/storage/data_stores/main/event_federation.py b/synapse/storage/data_stores/main/event_federation.py index 88e6489576..32e76621a7 100644 --- a/synapse/storage/data_stores/main/event_federation.py +++ b/synapse/storage/data_stores/main/event_federation.py @@ -14,7 +14,6 @@ # limitations under the License. import itertools import logging -import random from six.moves import range from six.moves.queue import Empty, PriorityQueue @@ -148,35 +147,6 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBas retcol="event_id", ) - @defer.inlineCallbacks - def get_prev_events_and_hashes_for_room(self, room_id): - """ - Gets a subset of the current forward extremities in the given room, - along with their depths and hashes. - - Limits the result to 10 extremities, so that we can avoid creating - events which refer to hundreds of prev_events. - - Args: - room_id (str): room_id - - Returns: - Deferred[list[(str, dict[str, str], int)]] - for each event, a tuple of (event_id, hashes, depth) - where *hashes* is a map from algorithm to hash. - """ - res = yield self.get_latest_event_ids_and_hashes_in_room(room_id) - if len(res) > 10: - # Sort by reverse depth, so we point to the most recent. - res.sort(key=lambda a: -a[2]) - - # we use half of the limit for the actual most recent events, and - # the other half to randomly point to some of the older events, to - # make sure that we don't completely ignore the older events. - res = res[0:5] + random.sample(res[5:], 5) - - return res - def get_prev_events_for_room(self, room_id: str): """ Gets a subset of the current forward extremities in the given room. diff --git a/tests/storage/test_event_federation.py b/tests/storage/test_event_federation.py index 3a68bf3274..a331517f4d 100644 --- a/tests/storage/test_event_federation.py +++ b/tests/storage/test_event_federation.py @@ -26,7 +26,7 @@ class EventFederationWorkerStoreTestCase(tests.unittest.TestCase): self.store = hs.get_datastore() @defer.inlineCallbacks - def test_get_prev_events_and_hashes_for_room(self): + def test_get_prev_events_for_room(self): room_id = "@ROOM:local" # add a bunch of events and hashes to act as forward extremities @@ -60,21 +60,14 @@ class EventFederationWorkerStoreTestCase(tests.unittest.TestCase): (event_id, bytearray(b"ffff")), ) - for i in range(0, 11): + for i in range(0, 20): yield self.store.db.runInteraction("insert", insert_event, i) - # this should get the last five and five others - r = yield self.store.get_prev_events_and_hashes_for_room(room_id) + # this should get the last ten + r = yield self.store.get_prev_events_for_room(room_id) self.assertEqual(10, len(r)) - for i in range(0, 5): - el = r[i] - depth = el[2] - self.assertEqual(10 - i, depth) - - for i in range(5, 5): - el = r[i] - depth = el[2] - self.assertLessEqual(5, depth) + for i in range(0, 10): + self.assertEqual("$event_%i:local" % (19 - i), r[i]) @defer.inlineCallbacks def test_get_rooms_with_many_extremities(self): -- cgit 1.4.1 From d20c3465441cd64ba3a1e84ee399bbadc0997bdf Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Tue, 7 Jan 2020 14:09:07 +0000 Subject: port BackgroundUpdateTestCase to HomeserverTestCase (#6653) --- changelog.d/6653.misc | 1 + tests/storage/test_background_update.py | 72 +++++++++++++++++---------------- 2 files changed, 38 insertions(+), 35 deletions(-) create mode 100644 changelog.d/6653.misc (limited to 'tests') diff --git a/changelog.d/6653.misc b/changelog.d/6653.misc new file mode 100644 index 0000000000..fbe7c0e7db --- /dev/null +++ b/changelog.d/6653.misc @@ -0,0 +1 @@ +Port core background update routines to async/await. diff --git a/tests/storage/test_background_update.py b/tests/storage/test_background_update.py index aec76f4ab1..ae14fb407d 100644 --- a/tests/storage/test_background_update.py +++ b/tests/storage/test_background_update.py @@ -2,44 +2,37 @@ from mock import Mock from twisted.internet import defer +from synapse.storage.background_updates import BackgroundUpdater + from tests import unittest -from tests.utils import setup_test_homeserver -class BackgroundUpdateTestCase(unittest.TestCase): - @defer.inlineCallbacks - def setUp(self): - hs = yield setup_test_homeserver(self.addCleanup) - self.store = hs.get_datastore() - self.clock = hs.get_clock() +class BackgroundUpdateTestCase(unittest.HomeserverTestCase): + def prepare(self, reactor, clock, homeserver): + self.updates = self.hs.get_datastore().db.updates # type: BackgroundUpdater + # the base test class should have run the real bg updates for us + self.assertTrue(self.updates.has_completed_background_updates()) self.update_handler = Mock() - - yield self.store.db.updates.register_background_update_handler( + self.updates.register_background_update_handler( "test_update", self.update_handler ) - # run the real background updates, to get them out the way - # (perhaps we should run them as part of the test HS setup, since we - # run all of the other schema setup stuff there?) - while True: - res = yield self.store.db.updates.do_next_background_update(1000) - if res is None: - break - - @defer.inlineCallbacks def test_do_background_update(self): - desired_count = 1000 + # the time we claim each update takes duration_ms = 42 + # the target runtime for each bg update + target_background_update_duration_ms = 50000 + # first step: make a bit of progress @defer.inlineCallbacks def update(progress, count): - self.clock.advance_time_msec(count * duration_ms) + yield self.clock.sleep((count * duration_ms) / 1000) progress = {"my_key": progress["my_key"] + 1} - yield self.store.db.runInteraction( + yield self.hs.get_datastore().db.runInteraction( "update_progress", - self.store.db.updates._background_update_progress_txn, + self.updates._background_update_progress_txn, "test_update", progress, ) @@ -47,37 +40,46 @@ class BackgroundUpdateTestCase(unittest.TestCase): self.update_handler.side_effect = update - yield self.store.db.updates.start_background_update( - "test_update", {"my_key": 1} + self.get_success( + self.updates.start_background_update("test_update", {"my_key": 1}) ) - self.update_handler.reset_mock() - result = yield self.store.db.updates.do_next_background_update( - duration_ms * desired_count + res = self.get_success( + self.updates.do_next_background_update( + target_background_update_duration_ms + ), + by=0.1, ) - self.assertIsNotNone(result) + self.assertIsNotNone(res) + + # on the first call, we should get run with the default background update size self.update_handler.assert_called_once_with( - {"my_key": 1}, self.store.db.updates.DEFAULT_BACKGROUND_BATCH_SIZE + {"my_key": 1}, self.updates.DEFAULT_BACKGROUND_BATCH_SIZE ) # second step: complete the update + # we should now get run with a much bigger number of items to update @defer.inlineCallbacks def update(progress, count): - yield self.store.db.updates._end_background_update("test_update") + self.assertEqual(progress, {"my_key": 2}) + self.assertAlmostEqual( + count, target_background_update_duration_ms / duration_ms, places=0, + ) + yield self.updates._end_background_update("test_update") return count self.update_handler.side_effect = update self.update_handler.reset_mock() - result = yield self.store.db.updates.do_next_background_update( - duration_ms * desired_count + result = self.get_success( + self.updates.do_next_background_update(target_background_update_duration_ms) ) self.assertIsNotNone(result) - self.update_handler.assert_called_once_with({"my_key": 2}, desired_count) + self.update_handler.assert_called_once() # third step: we don't expect to be called any more self.update_handler.reset_mock() - result = yield self.store.db.updates.do_next_background_update( - duration_ms * desired_count + result = self.get_success( + self.updates.do_next_background_update(target_background_update_duration_ms) ) self.assertIsNone(result) self.assertFalse(self.update_handler.called) -- cgit 1.4.1 From 573fee759cbd76fca93bf90783cd013a11b9b4e5 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Wed, 8 Jan 2020 13:24:10 +0000 Subject: Back out ill-advised notary server hackery (#6657) This was ill-advised. We can't modify verify_keys here, because the response object has already been signed by the requested key. Furthermore, it's somewhat unnecessary because existing versions of Synapse (which get upset that the notary key isn't present in verify_keys) will fall back to a direct fetch via `/key/v2/server`. Also: more tests for fetching keys via perspectives: it would be nice if we actually tested when our fetcher can't talk to our notary impl. --- changelog.d/6657.bugfix | 1 + synapse/rest/key/v2/remote_key_resource.py | 30 ++---- tests/rest/key/__init__.py | 0 tests/rest/key/v2/__init__.py | 0 tests/rest/key/v2/test_remote_key_resource.py | 135 +++++++++++++++++++++++++- 5 files changed, 140 insertions(+), 26 deletions(-) create mode 100644 changelog.d/6657.bugfix create mode 100644 tests/rest/key/__init__.py create mode 100644 tests/rest/key/v2/__init__.py (limited to 'tests') diff --git a/changelog.d/6657.bugfix b/changelog.d/6657.bugfix new file mode 100644 index 0000000000..94e51a9896 --- /dev/null +++ b/changelog.d/6657.bugfix @@ -0,0 +1 @@ +Fix incorrect signing of responses from the key server implementation. \ No newline at end of file diff --git a/synapse/rest/key/v2/remote_key_resource.py b/synapse/rest/key/v2/remote_key_resource.py index bf5e0eb844..e7fc3f0431 100644 --- a/synapse/rest/key/v2/remote_key_resource.py +++ b/synapse/rest/key/v2/remote_key_resource.py @@ -15,7 +15,6 @@ import logging from canonicaljson import encode_canonical_json, json -from signedjson.key import encode_verify_key_base64 from signedjson.sign import sign_json from twisted.internet import defer @@ -217,28 +216,15 @@ class RemoteKey(DirectServeResource): if cache_misses and query_remote_on_cache_miss: yield self.fetcher.get_keys(cache_misses) yield self.query_keys(request, query, query_remote_on_cache_miss=False) - return - - signed_keys = [] - for key_json in json_results: - key_json = json.loads(key_json) - - # backwards-compatibility hack for #6596: if the requested key belongs - # to us, make sure that all of the signing keys appear in the - # "verify_keys" section. - if key_json["server_name"] == self.config.server_name: - verify_keys = key_json["verify_keys"] + else: + signed_keys = [] + for key_json in json_results: + key_json = json.loads(key_json) for signing_key in self.config.key_server_signing_keys: - key_id = "%s:%s" % (signing_key.alg, signing_key.version) - verify_keys[key_id] = { - "key": encode_verify_key_base64(signing_key.verify_key) - } - - for signing_key in self.config.key_server_signing_keys: - key_json = sign_json(key_json, self.config.server_name, signing_key) + key_json = sign_json(key_json, self.config.server_name, signing_key) - signed_keys.append(key_json) + signed_keys.append(key_json) - results = {"server_keys": signed_keys} + results = {"server_keys": signed_keys} - respond_with_json_bytes(request, 200, encode_canonical_json(results)) + respond_with_json_bytes(request, 200, encode_canonical_json(results)) diff --git a/tests/rest/key/__init__.py b/tests/rest/key/__init__.py new file mode 100644 index 0000000000..e69de29bb2 diff --git a/tests/rest/key/v2/__init__.py b/tests/rest/key/v2/__init__.py new file mode 100644 index 0000000000..e69de29bb2 diff --git a/tests/rest/key/v2/test_remote_key_resource.py b/tests/rest/key/v2/test_remote_key_resource.py index d8246b4e78..6776a56cad 100644 --- a/tests/rest/key/v2/test_remote_key_resource.py +++ b/tests/rest/key/v2/test_remote_key_resource.py @@ -13,25 +13,30 @@ # See the License for the specific language governing permissions and # limitations under the License. import urllib.parse -from io import BytesIO +from io import BytesIO, StringIO from mock import Mock import signedjson.key +from canonicaljson import encode_canonical_json from nacl.signing import SigningKey from signedjson.sign import sign_json from twisted.web.resource import NoResource +from synapse.crypto.keyring import PerspectivesKeyFetcher from synapse.http.site import SynapseRequest from synapse.rest.key.v2 import KeyApiV2Resource +from synapse.storage.keys import FetchKeyResult from synapse.util.httpresourcetree import create_resource_tree +from synapse.util.stringutils import random_string from tests import unittest from tests.server import FakeChannel, wait_until_result +from tests.utils import default_config -class RemoteKeyResourceTestCase(unittest.HomeserverTestCase): +class BaseRemoteKeyResourceTestCase(unittest.HomeserverTestCase): def make_homeserver(self, reactor, clock): self.http_client = Mock() return self.setup_test_homeserver(http_client=self.http_client) @@ -73,6 +78,8 @@ class RemoteKeyResourceTestCase(unittest.HomeserverTestCase): self.http_client.get_json.side_effect = get_json + +class RemoteKeyResourceTestCase(BaseRemoteKeyResourceTestCase): def make_notary_request(self, server_name: str, key_id: str) -> dict: """Send a GET request to the test server requesting the given key. @@ -125,6 +132,126 @@ class RemoteKeyResourceTestCase(unittest.HomeserverTestCase): oursigs = sigs[self.hs.hostname] self.assertEqual(len(oursigs), 2) - # and both keys should be present in the verify_keys section + # the requested key should be present in the verify_keys section self.assertIn("ed25519:ver1", keys[0]["verify_keys"]) - self.assertIn("ed25519:a_lPym", keys[0]["verify_keys"]) + + +class EndToEndPerspectivesTests(BaseRemoteKeyResourceTestCase): + """End-to-end tests of the perspectives fetch case + + The idea here is to actually wire up a PerspectivesKeyFetcher to the notary + endpoint, to check that the two implementations are compatible. + """ + + def default_config(self, *args, **kwargs): + config = super().default_config(*args, **kwargs) + + # replace the signing key with our own + self.hs_signing_key = signedjson.key.generate_signing_key("kssk") + strm = StringIO() + signedjson.key.write_signing_keys(strm, [self.hs_signing_key]) + config["signing_key"] = strm.getvalue() + + return config + + def prepare(self, reactor, clock, homeserver): + # make a second homeserver, configured to use the first one as a key notary + self.http_client2 = Mock() + config = default_config(name="keyclient") + config["trusted_key_servers"] = [ + { + "server_name": self.hs.hostname, + "verify_keys": { + "ed25519:%s" + % ( + self.hs_signing_key.version, + ): signedjson.key.encode_verify_key_base64( + self.hs_signing_key.verify_key + ) + }, + } + ] + self.hs2 = self.setup_test_homeserver( + http_client=self.http_client2, config=config + ) + + # wire up outbound POST /key/v2/query requests from hs2 so that they + # will be forwarded to hs1 + def post_json(destination, path, data): + self.assertEqual(destination, self.hs.hostname) + self.assertEqual( + path, "/_matrix/key/v2/query", + ) + + channel = FakeChannel(self.site, self.reactor) + req = SynapseRequest(channel) + req.content = BytesIO(encode_canonical_json(data)) + + req.requestReceived( + b"POST", path.encode("utf-8"), b"1.1", + ) + wait_until_result(self.reactor, req) + self.assertEqual(channel.code, 200) + resp = channel.json_body + return resp + + self.http_client2.post_json.side_effect = post_json + + def test_get_key(self): + """Fetch a key belonging to a random server""" + # make up a key to be fetched. + testkey = signedjson.key.generate_signing_key("abc") + + # we expect hs1 to make a regular key request to the target server + self.expect_outgoing_key_request("targetserver", testkey) + keyid = "ed25519:%s" % (testkey.version,) + + fetcher = PerspectivesKeyFetcher(self.hs2) + d = fetcher.get_keys({"targetserver": {keyid: 1000}}) + res = self.get_success(d) + self.assertIn("targetserver", res) + keyres = res["targetserver"][keyid] + assert isinstance(keyres, FetchKeyResult) + self.assertEqual( + signedjson.key.encode_verify_key_base64(keyres.verify_key), + signedjson.key.encode_verify_key_base64(testkey.verify_key), + ) + + def test_get_notary_key(self): + """Fetch a key belonging to the notary server""" + # make up a key to be fetched. We randomise the keyid to try to get it to + # appear before the key server signing key sometimes (otherwise we bail out + # before fetching its signature) + testkey = signedjson.key.generate_signing_key(random_string(5)) + + # we expect hs1 to make a regular key request to itself + self.expect_outgoing_key_request(self.hs.hostname, testkey) + keyid = "ed25519:%s" % (testkey.version,) + + fetcher = PerspectivesKeyFetcher(self.hs2) + d = fetcher.get_keys({self.hs.hostname: {keyid: 1000}}) + res = self.get_success(d) + self.assertIn(self.hs.hostname, res) + keyres = res[self.hs.hostname][keyid] + assert isinstance(keyres, FetchKeyResult) + self.assertEqual( + signedjson.key.encode_verify_key_base64(keyres.verify_key), + signedjson.key.encode_verify_key_base64(testkey.verify_key), + ) + + def test_get_notary_keyserver_key(self): + """Fetch the notary's keyserver key""" + # we expect hs1 to make a regular key request to itself + self.expect_outgoing_key_request(self.hs.hostname, self.hs_signing_key) + keyid = "ed25519:%s" % (self.hs_signing_key.version,) + + fetcher = PerspectivesKeyFetcher(self.hs2) + d = fetcher.get_keys({self.hs.hostname: {keyid: 1000}}) + res = self.get_success(d) + self.assertIn(self.hs.hostname, res) + keyres = res[self.hs.hostname][keyid] + assert isinstance(keyres, FetchKeyResult) + self.assertEqual( + signedjson.key.encode_verify_key_base64(keyres.verify_key), + signedjson.key.encode_verify_key_base64(self.hs_signing_key.verify_key), + ) -- cgit 1.4.1 From 7caaa29daab7bde331dcec0bb760a8fe5870f18e Mon Sep 17 00:00:00 2001 From: Manuel Stahl <37705355+awesome-manuel@users.noreply.github.com> Date: Wed, 8 Jan 2020 14:26:40 +0100 Subject: Fix GET request on /_synapse/admin/v2/users endpoint (#6563) Fixes #6552 --- changelog.d/6563.bugfix | 1 + synapse/storage/data_stores/main/__init__.py | 4 +-- tests/rest/admin/test_admin.py | 41 ++++++++++++++++++++++++++++ 3 files changed, 44 insertions(+), 2 deletions(-) create mode 100644 changelog.d/6563.bugfix (limited to 'tests') diff --git a/changelog.d/6563.bugfix b/changelog.d/6563.bugfix new file mode 100644 index 0000000000..3325fb1dcf --- /dev/null +++ b/changelog.d/6563.bugfix @@ -0,0 +1 @@ +Fix GET request on /_synapse/admin/v2/users endpoint. Contributed by Awesome Technologies Innovationslabor GmbH. \ No newline at end of file diff --git a/synapse/storage/data_stores/main/__init__.py b/synapse/storage/data_stores/main/__init__.py index c577c0df5f..2700cca822 100644 --- a/synapse/storage/data_stores/main/__init__.py +++ b/synapse/storage/data_stores/main/__init__.py @@ -526,9 +526,9 @@ class DataStore( attr_filter = {} if not guests: - attr_filter["is_guest"] = False + attr_filter["is_guest"] = 0 if not deactivated: - attr_filter["deactivated"] = False + attr_filter["deactivated"] = 0 return self.db.simple_select_list_paginate( desc="get_users_paginate", diff --git a/tests/rest/admin/test_admin.py b/tests/rest/admin/test_admin.py index 0ed2594381..325bd6a608 100644 --- a/tests/rest/admin/test_admin.py +++ b/tests/rest/admin/test_admin.py @@ -341,6 +341,47 @@ class UserRegisterTestCase(unittest.HomeserverTestCase): self.assertEqual("Invalid user type", channel.json_body["error"]) +class UsersListTestCase(unittest.HomeserverTestCase): + + servlets = [ + synapse.rest.admin.register_servlets, + login.register_servlets, + ] + url = "/_synapse/admin/v2/users" + + def prepare(self, reactor, clock, hs): + self.admin_user = self.register_user("admin", "pass", admin=True) + self.admin_user_tok = self.login("admin", "pass") + + self.register_user("user1", "pass1", admin=False) + self.register_user("user2", "pass2", admin=False) + + def test_no_auth(self): + """ + Try to list users without authentication. + """ + request, channel = self.make_request("GET", self.url, b"{}") + self.render(request) + + self.assertEqual(401, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("M_MISSING_TOKEN", channel.json_body["errcode"]) + + def test_all_users(self): + """ + List all users, including deactivated users. + """ + request, channel = self.make_request( + "GET", + self.url + "?deactivated=true", + b"{}", + access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual(3, len(channel.json_body["users"])) + + class ShutdownRoomTestCase(unittest.HomeserverTestCase): servlets = [ synapse.rest.admin.register_servlets_for_client_rest_resource, -- cgit 1.4.1 From d2906fe6667d3384f37ef03ca87172d643d49587 Mon Sep 17 00:00:00 2001 From: Manuel Stahl <37705355+awesome-manuel@users.noreply.github.com> Date: Thu, 9 Jan 2020 14:31:00 +0100 Subject: Allow admin users to create or modify users without a shared secret (#6495) Signed-off-by: Manuel Stahl --- changelog.d/5742.feature | 1 + docs/admin_api/user_admin_api.rst | 33 +- synapse/handlers/admin.py | 9 + synapse/rest/admin/__init__.py | 2 + synapse/rest/admin/users.py | 142 +++++++ synapse/storage/data_stores/main/registration.py | 2 + tests/rest/admin/test_admin.py | 338 ---------------- tests/rest/admin/test_user.py | 465 +++++++++++++++++++++++ tests/storage/test_registration.py | 2 + 9 files changed, 655 insertions(+), 339 deletions(-) create mode 100644 changelog.d/5742.feature create mode 100644 tests/rest/admin/test_user.py (limited to 'tests') diff --git a/changelog.d/5742.feature b/changelog.d/5742.feature new file mode 100644 index 0000000000..de10302275 --- /dev/null +++ b/changelog.d/5742.feature @@ -0,0 +1 @@ +Allow admin to create or modify a user. Contributed by Awesome Technologies Innovationslabor GmbH. diff --git a/docs/admin_api/user_admin_api.rst b/docs/admin_api/user_admin_api.rst index b451dc5014..0b3d09d694 100644 --- a/docs/admin_api/user_admin_api.rst +++ b/docs/admin_api/user_admin_api.rst @@ -1,3 +1,33 @@ +Create or modify Account +======================== + +This API allows an administrator to create or modify a user account with a +specific ``user_id``. + +This api is:: + + PUT /_synapse/admin/v2/users/ + +with a body of: + +.. code:: json + + { + "password": "user_password", + "displayname": "User", + "avatar_url": "", + "admin": false, + "deactivated": false + } + +including an ``access_token`` of a server admin. + +The parameter ``displayname`` is optional and defaults to ``user_id``. +The parameter ``avatar_url`` is optional. +The parameter ``admin`` is optional and defaults to 'false'. +The parameter ``deactivated`` is optional and defaults to 'false'. +If the user already exists then optional parameters default to the current value. + List Accounts ============= @@ -50,7 +80,8 @@ This API returns information about a specific user account. The api is:: - GET /_synapse/admin/v1/whois/ + GET /_synapse/admin/v1/whois/ (deprecated) + GET /_synapse/admin/v2/users/ including an ``access_token`` of a server admin. diff --git a/synapse/handlers/admin.py b/synapse/handlers/admin.py index 1a4ba12385..76d18a8ba8 100644 --- a/synapse/handlers/admin.py +++ b/synapse/handlers/admin.py @@ -51,6 +51,15 @@ class AdminHandler(BaseHandler): return ret + async def get_user(self, user): + """Function to get user details""" + ret = await self.store.get_user_by_id(user.to_string()) + if ret: + profile = await self.store.get_profileinfo(user.localpart) + ret["displayname"] = profile.display_name + ret["avatar_url"] = profile.avatar_url + return ret + async def get_users(self): """Function to retrieve a list of users in users table. diff --git a/synapse/rest/admin/__init__.py b/synapse/rest/admin/__init__.py index c122c449f4..a10b4a9b72 100644 --- a/synapse/rest/admin/__init__.py +++ b/synapse/rest/admin/__init__.py @@ -38,6 +38,7 @@ from synapse.rest.admin.users import ( SearchUsersRestServlet, UserAdminServlet, UserRegisterServlet, + UserRestServletV2, UsersRestServlet, UsersRestServletV2, WhoisRestServlet, @@ -191,6 +192,7 @@ def register_servlets(hs, http_server): SendServerNoticeServlet(hs).register(http_server) VersionServlet(hs).register(http_server) UserAdminServlet(hs).register(http_server) + UserRestServletV2(hs).register(http_server) UsersRestServletV2(hs).register(http_server) diff --git a/synapse/rest/admin/users.py b/synapse/rest/admin/users.py index 1937879dbe..574cb90c74 100644 --- a/synapse/rest/admin/users.py +++ b/synapse/rest/admin/users.py @@ -102,6 +102,148 @@ class UsersRestServletV2(RestServlet): return 200, ret +class UserRestServletV2(RestServlet): + PATTERNS = (re.compile("^/_synapse/admin/v2/users/(?P@[^/]+)$"),) + + """Get request to list user details. + This needs user to have administrator access in Synapse. + + GET /_synapse/admin/v2/users/ + + returns: + 200 OK with user details if success otherwise an error. + + Put request to allow an administrator to add or modify a user. + This needs user to have administrator access in Synapse. + We use PUT instead of POST since we already know the id of the user + object to create. POST could be used to create guests. + + PUT /_synapse/admin/v2/users/ + { + "password": "secret", + "displayname": "User" + } + + returns: + 201 OK with new user object if user was created or + 200 OK with modified user object if user was modified + otherwise an error. + """ + + def __init__(self, hs): + self.hs = hs + self.auth = hs.get_auth() + self.admin_handler = hs.get_handlers().admin_handler + self.profile_handler = hs.get_profile_handler() + self.set_password_handler = hs.get_set_password_handler() + self.deactivate_account_handler = hs.get_deactivate_account_handler() + self.registration_handler = hs.get_registration_handler() + + async def on_GET(self, request, user_id): + await assert_requester_is_admin(self.auth, request) + + target_user = UserID.from_string(user_id) + if not self.hs.is_mine(target_user): + raise SynapseError(400, "Can only lookup local users") + + ret = await self.admin_handler.get_user(target_user) + + return 200, ret + + async def on_PUT(self, request, user_id): + await assert_requester_is_admin(self.auth, request) + + target_user = UserID.from_string(user_id) + body = parse_json_object_from_request(request) + + if not self.hs.is_mine(target_user): + raise SynapseError(400, "This endpoint can only be used with local users") + + user = await self.admin_handler.get_user(target_user) + + if user: # modify user + requester = await self.auth.get_user_by_req(request) + + if "displayname" in body: + await self.profile_handler.set_displayname( + target_user, requester, body["displayname"], True + ) + + if "avatar_url" in body: + await self.profile_handler.set_avatar_url( + target_user, requester, body["avatar_url"], True + ) + + if "admin" in body: + set_admin_to = bool(body["admin"]) + if set_admin_to != user["admin"]: + auth_user = requester.user + if target_user == auth_user and not set_admin_to: + raise SynapseError(400, "You may not demote yourself.") + + await self.admin_handler.set_user_server_admin( + target_user, set_admin_to + ) + + if "password" in body: + if ( + not isinstance(body["password"], text_type) + or len(body["password"]) > 512 + ): + raise SynapseError(400, "Invalid password") + else: + new_password = body["password"] + await self._set_password_handler.set_password( + target_user, new_password, requester + ) + + if "deactivated" in body: + deactivate = bool(body["deactivated"]) + if deactivate and not user["deactivated"]: + result = await self.deactivate_account_handler.deactivate_account( + target_user.to_string(), False + ) + if not result: + raise SynapseError(500, "Could not deactivate user") + + user = await self.admin_handler.get_user(target_user) + return 200, user + + else: # create user + if "password" not in body: + raise SynapseError( + 400, "password must be specified", errcode=Codes.BAD_JSON + ) + elif ( + not isinstance(body["password"], text_type) + or len(body["password"]) > 512 + ): + raise SynapseError(400, "Invalid password") + + admin = body.get("admin", None) + user_type = body.get("user_type", None) + displayname = body.get("displayname", None) + + if user_type is not None and user_type not in UserTypes.ALL_USER_TYPES: + raise SynapseError(400, "Invalid user type") + + user_id = await self.registration_handler.register_user( + localpart=target_user.localpart, + password=body["password"], + admin=bool(admin), + default_display_name=displayname, + user_type=user_type, + ) + if "avatar_url" in body: + await self.profile_handler.set_avatar_url( + user_id, requester, body["avatar_url"], True + ) + + ret = await self.admin_handler.get_user(target_user) + + return 201, ret + + class UserRegisterServlet(RestServlet): """ Attributes: diff --git a/synapse/storage/data_stores/main/registration.py b/synapse/storage/data_stores/main/registration.py index 5e8ecac0ea..cb4b2b39a0 100644 --- a/synapse/storage/data_stores/main/registration.py +++ b/synapse/storage/data_stores/main/registration.py @@ -52,11 +52,13 @@ class RegistrationWorkerStore(SQLBaseStore): "name", "password_hash", "is_guest", + "admin", "consent_version", "consent_server_notice_sent", "appservice_id", "creation_ts", "user_type", + "deactivated", ], allow_none=True, desc="get_user_by_id", diff --git a/tests/rest/admin/test_admin.py b/tests/rest/admin/test_admin.py index 325bd6a608..6ceb483aa8 100644 --- a/tests/rest/admin/test_admin.py +++ b/tests/rest/admin/test_admin.py @@ -13,14 +13,11 @@ # See the License for the specific language governing permissions and # limitations under the License. -import hashlib -import hmac import json from mock import Mock import synapse.rest.admin -from synapse.api.constants import UserTypes from synapse.http.server import JsonResource from synapse.rest.admin import VersionServlet from synapse.rest.client.v1 import events, login, room @@ -47,341 +44,6 @@ class VersionTestCase(unittest.HomeserverTestCase): ) -class UserRegisterTestCase(unittest.HomeserverTestCase): - - servlets = [synapse.rest.admin.register_servlets_for_client_rest_resource] - - def make_homeserver(self, reactor, clock): - - self.url = "/_matrix/client/r0/admin/register" - - self.registration_handler = Mock() - self.identity_handler = Mock() - self.login_handler = Mock() - self.device_handler = Mock() - self.device_handler.check_device_registered = Mock(return_value="FAKE") - - self.datastore = Mock(return_value=Mock()) - self.datastore.get_current_state_deltas = Mock(return_value=(0, [])) - - self.secrets = Mock() - - self.hs = self.setup_test_homeserver() - - self.hs.config.registration_shared_secret = "shared" - - self.hs.get_media_repository = Mock() - self.hs.get_deactivate_account_handler = Mock() - - return self.hs - - def test_disabled(self): - """ - If there is no shared secret, registration through this method will be - prevented. - """ - self.hs.config.registration_shared_secret = None - - request, channel = self.make_request("POST", self.url, b"{}") - self.render(request) - - self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual( - "Shared secret registration is not enabled", channel.json_body["error"] - ) - - def test_get_nonce(self): - """ - Calling GET on the endpoint will return a randomised nonce, using the - homeserver's secrets provider. - """ - secrets = Mock() - secrets.token_hex = Mock(return_value="abcd") - - self.hs.get_secrets = Mock(return_value=secrets) - - request, channel = self.make_request("GET", self.url) - self.render(request) - - self.assertEqual(channel.json_body, {"nonce": "abcd"}) - - def test_expired_nonce(self): - """ - Calling GET on the endpoint will return a randomised nonce, which will - only last for SALT_TIMEOUT (60s). - """ - request, channel = self.make_request("GET", self.url) - self.render(request) - nonce = channel.json_body["nonce"] - - # 59 seconds - self.reactor.advance(59) - - body = json.dumps({"nonce": nonce}) - request, channel = self.make_request("POST", self.url, body.encode("utf8")) - self.render(request) - - self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual("username must be specified", channel.json_body["error"]) - - # 61 seconds - self.reactor.advance(2) - - request, channel = self.make_request("POST", self.url, body.encode("utf8")) - self.render(request) - - self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual("unrecognised nonce", channel.json_body["error"]) - - def test_register_incorrect_nonce(self): - """ - Only the provided nonce can be used, as it's checked in the MAC. - """ - request, channel = self.make_request("GET", self.url) - self.render(request) - nonce = channel.json_body["nonce"] - - want_mac = hmac.new(key=b"shared", digestmod=hashlib.sha1) - want_mac.update(b"notthenonce\x00bob\x00abc123\x00admin") - want_mac = want_mac.hexdigest() - - body = json.dumps( - { - "nonce": nonce, - "username": "bob", - "password": "abc123", - "admin": True, - "mac": want_mac, - } - ) - request, channel = self.make_request("POST", self.url, body.encode("utf8")) - self.render(request) - - self.assertEqual(403, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual("HMAC incorrect", channel.json_body["error"]) - - def test_register_correct_nonce(self): - """ - When the correct nonce is provided, and the right key is provided, the - user is registered. - """ - request, channel = self.make_request("GET", self.url) - self.render(request) - nonce = channel.json_body["nonce"] - - want_mac = hmac.new(key=b"shared", digestmod=hashlib.sha1) - want_mac.update( - nonce.encode("ascii") + b"\x00bob\x00abc123\x00admin\x00support" - ) - want_mac = want_mac.hexdigest() - - body = json.dumps( - { - "nonce": nonce, - "username": "bob", - "password": "abc123", - "admin": True, - "user_type": UserTypes.SUPPORT, - "mac": want_mac, - } - ) - request, channel = self.make_request("POST", self.url, body.encode("utf8")) - self.render(request) - - self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual("@bob:test", channel.json_body["user_id"]) - - def test_nonce_reuse(self): - """ - A valid unrecognised nonce. - """ - request, channel = self.make_request("GET", self.url) - self.render(request) - nonce = channel.json_body["nonce"] - - want_mac = hmac.new(key=b"shared", digestmod=hashlib.sha1) - want_mac.update(nonce.encode("ascii") + b"\x00bob\x00abc123\x00admin") - want_mac = want_mac.hexdigest() - - body = json.dumps( - { - "nonce": nonce, - "username": "bob", - "password": "abc123", - "admin": True, - "mac": want_mac, - } - ) - request, channel = self.make_request("POST", self.url, body.encode("utf8")) - self.render(request) - - self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual("@bob:test", channel.json_body["user_id"]) - - # Now, try and reuse it - request, channel = self.make_request("POST", self.url, body.encode("utf8")) - self.render(request) - - self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual("unrecognised nonce", channel.json_body["error"]) - - def test_missing_parts(self): - """ - Synapse will complain if you don't give nonce, username, password, and - mac. Admin and user_types are optional. Additional checks are done for length - and type. - """ - - def nonce(): - request, channel = self.make_request("GET", self.url) - self.render(request) - return channel.json_body["nonce"] - - # - # Nonce check - # - - # Must be present - body = json.dumps({}) - request, channel = self.make_request("POST", self.url, body.encode("utf8")) - self.render(request) - - self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual("nonce must be specified", channel.json_body["error"]) - - # - # Username checks - # - - # Must be present - body = json.dumps({"nonce": nonce()}) - request, channel = self.make_request("POST", self.url, body.encode("utf8")) - self.render(request) - - self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual("username must be specified", channel.json_body["error"]) - - # Must be a string - body = json.dumps({"nonce": nonce(), "username": 1234}) - request, channel = self.make_request("POST", self.url, body.encode("utf8")) - self.render(request) - - self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual("Invalid username", channel.json_body["error"]) - - # Must not have null bytes - body = json.dumps({"nonce": nonce(), "username": "abcd\u0000"}) - request, channel = self.make_request("POST", self.url, body.encode("utf8")) - self.render(request) - - self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual("Invalid username", channel.json_body["error"]) - - # Must not have null bytes - body = json.dumps({"nonce": nonce(), "username": "a" * 1000}) - request, channel = self.make_request("POST", self.url, body.encode("utf8")) - self.render(request) - - self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual("Invalid username", channel.json_body["error"]) - - # - # Password checks - # - - # Must be present - body = json.dumps({"nonce": nonce(), "username": "a"}) - request, channel = self.make_request("POST", self.url, body.encode("utf8")) - self.render(request) - - self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual("password must be specified", channel.json_body["error"]) - - # Must be a string - body = json.dumps({"nonce": nonce(), "username": "a", "password": 1234}) - request, channel = self.make_request("POST", self.url, body.encode("utf8")) - self.render(request) - - self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual("Invalid password", channel.json_body["error"]) - - # Must not have null bytes - body = json.dumps({"nonce": nonce(), "username": "a", "password": "abcd\u0000"}) - request, channel = self.make_request("POST", self.url, body.encode("utf8")) - self.render(request) - - self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual("Invalid password", channel.json_body["error"]) - - # Super long - body = json.dumps({"nonce": nonce(), "username": "a", "password": "A" * 1000}) - request, channel = self.make_request("POST", self.url, body.encode("utf8")) - self.render(request) - - self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual("Invalid password", channel.json_body["error"]) - - # - # user_type check - # - - # Invalid user_type - body = json.dumps( - { - "nonce": nonce(), - "username": "a", - "password": "1234", - "user_type": "invalid", - } - ) - request, channel = self.make_request("POST", self.url, body.encode("utf8")) - self.render(request) - - self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual("Invalid user type", channel.json_body["error"]) - - -class UsersListTestCase(unittest.HomeserverTestCase): - - servlets = [ - synapse.rest.admin.register_servlets, - login.register_servlets, - ] - url = "/_synapse/admin/v2/users" - - def prepare(self, reactor, clock, hs): - self.admin_user = self.register_user("admin", "pass", admin=True) - self.admin_user_tok = self.login("admin", "pass") - - self.register_user("user1", "pass1", admin=False) - self.register_user("user2", "pass2", admin=False) - - def test_no_auth(self): - """ - Try to list users without authentication. - """ - request, channel = self.make_request("GET", self.url, b"{}") - self.render(request) - - self.assertEqual(401, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual("M_MISSING_TOKEN", channel.json_body["errcode"]) - - def test_all_users(self): - """ - List all users, including deactivated users. - """ - request, channel = self.make_request( - "GET", - self.url + "?deactivated=true", - b"{}", - access_token=self.admin_user_tok, - ) - self.render(request) - - self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual(3, len(channel.json_body["users"])) - - class ShutdownRoomTestCase(unittest.HomeserverTestCase): servlets = [ synapse.rest.admin.register_servlets_for_client_rest_resource, diff --git a/tests/rest/admin/test_user.py b/tests/rest/admin/test_user.py new file mode 100644 index 0000000000..7352d609e6 --- /dev/null +++ b/tests/rest/admin/test_user.py @@ -0,0 +1,465 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import hashlib +import hmac +import json + +from mock import Mock + +import synapse.rest.admin +from synapse.api.constants import UserTypes +from synapse.rest.client.v1 import login + +from tests import unittest + + +class UserRegisterTestCase(unittest.HomeserverTestCase): + + servlets = [synapse.rest.admin.register_servlets_for_client_rest_resource] + + def make_homeserver(self, reactor, clock): + + self.url = "/_matrix/client/r0/admin/register" + + self.registration_handler = Mock() + self.identity_handler = Mock() + self.login_handler = Mock() + self.device_handler = Mock() + self.device_handler.check_device_registered = Mock(return_value="FAKE") + + self.datastore = Mock(return_value=Mock()) + self.datastore.get_current_state_deltas = Mock(return_value=(0, [])) + + self.secrets = Mock() + + self.hs = self.setup_test_homeserver() + + self.hs.config.registration_shared_secret = "shared" + + self.hs.get_media_repository = Mock() + self.hs.get_deactivate_account_handler = Mock() + + return self.hs + + def test_disabled(self): + """ + If there is no shared secret, registration through this method will be + prevented. + """ + self.hs.config.registration_shared_secret = None + + request, channel = self.make_request("POST", self.url, b"{}") + self.render(request) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual( + "Shared secret registration is not enabled", channel.json_body["error"] + ) + + def test_get_nonce(self): + """ + Calling GET on the endpoint will return a randomised nonce, using the + homeserver's secrets provider. + """ + secrets = Mock() + secrets.token_hex = Mock(return_value="abcd") + + self.hs.get_secrets = Mock(return_value=secrets) + + request, channel = self.make_request("GET", self.url) + self.render(request) + + self.assertEqual(channel.json_body, {"nonce": "abcd"}) + + def test_expired_nonce(self): + """ + Calling GET on the endpoint will return a randomised nonce, which will + only last for SALT_TIMEOUT (60s). + """ + request, channel = self.make_request("GET", self.url) + self.render(request) + nonce = channel.json_body["nonce"] + + # 59 seconds + self.reactor.advance(59) + + body = json.dumps({"nonce": nonce}) + request, channel = self.make_request("POST", self.url, body.encode("utf8")) + self.render(request) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("username must be specified", channel.json_body["error"]) + + # 61 seconds + self.reactor.advance(2) + + request, channel = self.make_request("POST", self.url, body.encode("utf8")) + self.render(request) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("unrecognised nonce", channel.json_body["error"]) + + def test_register_incorrect_nonce(self): + """ + Only the provided nonce can be used, as it's checked in the MAC. + """ + request, channel = self.make_request("GET", self.url) + self.render(request) + nonce = channel.json_body["nonce"] + + want_mac = hmac.new(key=b"shared", digestmod=hashlib.sha1) + want_mac.update(b"notthenonce\x00bob\x00abc123\x00admin") + want_mac = want_mac.hexdigest() + + body = json.dumps( + { + "nonce": nonce, + "username": "bob", + "password": "abc123", + "admin": True, + "mac": want_mac, + } + ) + request, channel = self.make_request("POST", self.url, body.encode("utf8")) + self.render(request) + + self.assertEqual(403, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("HMAC incorrect", channel.json_body["error"]) + + def test_register_correct_nonce(self): + """ + When the correct nonce is provided, and the right key is provided, the + user is registered. + """ + request, channel = self.make_request("GET", self.url) + self.render(request) + nonce = channel.json_body["nonce"] + + want_mac = hmac.new(key=b"shared", digestmod=hashlib.sha1) + want_mac.update( + nonce.encode("ascii") + b"\x00bob\x00abc123\x00admin\x00support" + ) + want_mac = want_mac.hexdigest() + + body = json.dumps( + { + "nonce": nonce, + "username": "bob", + "password": "abc123", + "admin": True, + "user_type": UserTypes.SUPPORT, + "mac": want_mac, + } + ) + request, channel = self.make_request("POST", self.url, body.encode("utf8")) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@bob:test", channel.json_body["user_id"]) + + def test_nonce_reuse(self): + """ + A valid unrecognised nonce. + """ + request, channel = self.make_request("GET", self.url) + self.render(request) + nonce = channel.json_body["nonce"] + + want_mac = hmac.new(key=b"shared", digestmod=hashlib.sha1) + want_mac.update(nonce.encode("ascii") + b"\x00bob\x00abc123\x00admin") + want_mac = want_mac.hexdigest() + + body = json.dumps( + { + "nonce": nonce, + "username": "bob", + "password": "abc123", + "admin": True, + "mac": want_mac, + } + ) + request, channel = self.make_request("POST", self.url, body.encode("utf8")) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@bob:test", channel.json_body["user_id"]) + + # Now, try and reuse it + request, channel = self.make_request("POST", self.url, body.encode("utf8")) + self.render(request) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("unrecognised nonce", channel.json_body["error"]) + + def test_missing_parts(self): + """ + Synapse will complain if you don't give nonce, username, password, and + mac. Admin and user_types are optional. Additional checks are done for length + and type. + """ + + def nonce(): + request, channel = self.make_request("GET", self.url) + self.render(request) + return channel.json_body["nonce"] + + # + # Nonce check + # + + # Must be present + body = json.dumps({}) + request, channel = self.make_request("POST", self.url, body.encode("utf8")) + self.render(request) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("nonce must be specified", channel.json_body["error"]) + + # + # Username checks + # + + # Must be present + body = json.dumps({"nonce": nonce()}) + request, channel = self.make_request("POST", self.url, body.encode("utf8")) + self.render(request) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("username must be specified", channel.json_body["error"]) + + # Must be a string + body = json.dumps({"nonce": nonce(), "username": 1234}) + request, channel = self.make_request("POST", self.url, body.encode("utf8")) + self.render(request) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("Invalid username", channel.json_body["error"]) + + # Must not have null bytes + body = json.dumps({"nonce": nonce(), "username": "abcd\u0000"}) + request, channel = self.make_request("POST", self.url, body.encode("utf8")) + self.render(request) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("Invalid username", channel.json_body["error"]) + + # Must not have null bytes + body = json.dumps({"nonce": nonce(), "username": "a" * 1000}) + request, channel = self.make_request("POST", self.url, body.encode("utf8")) + self.render(request) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("Invalid username", channel.json_body["error"]) + + # + # Password checks + # + + # Must be present + body = json.dumps({"nonce": nonce(), "username": "a"}) + request, channel = self.make_request("POST", self.url, body.encode("utf8")) + self.render(request) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("password must be specified", channel.json_body["error"]) + + # Must be a string + body = json.dumps({"nonce": nonce(), "username": "a", "password": 1234}) + request, channel = self.make_request("POST", self.url, body.encode("utf8")) + self.render(request) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("Invalid password", channel.json_body["error"]) + + # Must not have null bytes + body = json.dumps({"nonce": nonce(), "username": "a", "password": "abcd\u0000"}) + request, channel = self.make_request("POST", self.url, body.encode("utf8")) + self.render(request) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("Invalid password", channel.json_body["error"]) + + # Super long + body = json.dumps({"nonce": nonce(), "username": "a", "password": "A" * 1000}) + request, channel = self.make_request("POST", self.url, body.encode("utf8")) + self.render(request) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("Invalid password", channel.json_body["error"]) + + # + # user_type check + # + + # Invalid user_type + body = json.dumps( + { + "nonce": nonce(), + "username": "a", + "password": "1234", + "user_type": "invalid", + } + ) + request, channel = self.make_request("POST", self.url, body.encode("utf8")) + self.render(request) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("Invalid user type", channel.json_body["error"]) + + +class UsersListTestCase(unittest.HomeserverTestCase): + + servlets = [ + synapse.rest.admin.register_servlets, + login.register_servlets, + ] + url = "/_synapse/admin/v2/users" + + def prepare(self, reactor, clock, hs): + self.admin_user = self.register_user("admin", "pass", admin=True) + self.admin_user_tok = self.login("admin", "pass") + + self.register_user("user1", "pass1", admin=False) + self.register_user("user2", "pass2", admin=False) + + def test_no_auth(self): + """ + Try to list users without authentication. + """ + request, channel = self.make_request("GET", self.url, b"{}") + self.render(request) + + self.assertEqual(401, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("M_MISSING_TOKEN", channel.json_body["errcode"]) + + def test_all_users(self): + """ + List all users, including deactivated users. + """ + request, channel = self.make_request( + "GET", + self.url + "?deactivated=true", + b"{}", + access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual(3, len(channel.json_body["users"])) + + +class UserRestTestCase(unittest.HomeserverTestCase): + + servlets = [ + synapse.rest.admin.register_servlets, + login.register_servlets, + ] + + def prepare(self, reactor, clock, hs): + self.store = hs.get_datastore() + + self.url = "/_synapse/admin/v2/users/@bob:test" + + self.admin_user = self.register_user("admin", "pass", admin=True) + self.admin_user_tok = self.login("admin", "pass") + + self.other_user = self.register_user("user", "pass") + self.other_user_token = self.login("user", "pass") + + def test_requester_is_no_admin(self): + """ + If the user is not a server admin, an error is returned. + """ + self.hs.config.registration_shared_secret = None + + request, channel = self.make_request( + "GET", self.url, access_token=self.other_user_token, + ) + self.render(request) + + self.assertEqual(403, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("You are not a server admin", channel.json_body["error"]) + + request, channel = self.make_request( + "PUT", self.url, access_token=self.other_user_token, content=b"{}", + ) + self.render(request) + + self.assertEqual(403, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("You are not a server admin", channel.json_body["error"]) + + def test_requester_is_admin(self): + """ + If the user is a server admin, a new user is created. + """ + self.hs.config.registration_shared_secret = None + + body = json.dumps({"password": "abc123", "admin": True}) + + # Create user + request, channel = self.make_request( + "PUT", + self.url, + access_token=self.admin_user_tok, + content=body.encode(encoding="utf_8"), + ) + self.render(request) + + self.assertEqual(201, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@bob:test", channel.json_body["name"]) + self.assertEqual("bob", channel.json_body["displayname"]) + + # Get user + request, channel = self.make_request( + "GET", self.url, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@bob:test", channel.json_body["name"]) + self.assertEqual("bob", channel.json_body["displayname"]) + self.assertEqual(1, channel.json_body["admin"]) + self.assertEqual(0, channel.json_body["is_guest"]) + self.assertEqual(0, channel.json_body["deactivated"]) + + # Modify user + body = json.dumps({"displayname": "foobar", "deactivated": True}) + + request, channel = self.make_request( + "PUT", + self.url, + access_token=self.admin_user_tok, + content=body.encode(encoding="utf_8"), + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@bob:test", channel.json_body["name"]) + self.assertEqual("foobar", channel.json_body["displayname"]) + self.assertEqual(True, channel.json_body["deactivated"]) + + # Get user + request, channel = self.make_request( + "GET", self.url, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@bob:test", channel.json_body["name"]) + self.assertEqual("foobar", channel.json_body["displayname"]) + self.assertEqual(1, channel.json_body["admin"]) + self.assertEqual(0, channel.json_body["is_guest"]) + self.assertEqual(1, channel.json_body["deactivated"]) diff --git a/tests/storage/test_registration.py b/tests/storage/test_registration.py index ed5786865a..71a40a0a49 100644 --- a/tests/storage/test_registration.py +++ b/tests/storage/test_registration.py @@ -43,12 +43,14 @@ class RegistrationStoreTestCase(unittest.TestCase): # TODO(paul): Surely this field should be 'user_id', not 'name' "name": self.user_id, "password_hash": self.pwhash, + "admin": 0, "is_guest": 0, "consent_version": None, "consent_server_notice_sent": None, "appservice_id": None, "creation_ts": 1000, "user_type": None, + "deactivated": 0, }, (yield self.store.get_user_by_id(self.user_id)), ) -- cgit 1.4.1 From 326c893d24e0f03b62bd9d1136a335f329bb8528 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Mon, 13 Jan 2020 12:48:22 +0000 Subject: Kill off RegistrationError (#6691) This is pretty pointless. Let's just use SynapseError. --- changelog.d/6691.misc | 1 + synapse/api/errors.py | 6 ------ synapse/handlers/register.py | 12 +++--------- tests/handlers/test_register.py | 2 -- 4 files changed, 4 insertions(+), 17 deletions(-) create mode 100644 changelog.d/6691.misc (limited to 'tests') diff --git a/changelog.d/6691.misc b/changelog.d/6691.misc new file mode 100644 index 0000000000..104e9ce648 --- /dev/null +++ b/changelog.d/6691.misc @@ -0,0 +1 @@ +Remove redundant RegistrationError class. diff --git a/synapse/api/errors.py b/synapse/api/errors.py index 5853a54c95..9e9844b47c 100644 --- a/synapse/api/errors.py +++ b/synapse/api/errors.py @@ -158,12 +158,6 @@ class UserDeactivatedError(SynapseError): ) -class RegistrationError(SynapseError): - """An error raised when a registration event fails.""" - - pass - - class FederationDeniedError(SynapseError): """An error raised when the server tries to federate with a server which is not on its federation whitelist. diff --git a/synapse/handlers/register.py b/synapse/handlers/register.py index 885da82985..7ffc194f0c 100644 --- a/synapse/handlers/register.py +++ b/synapse/handlers/register.py @@ -20,13 +20,7 @@ from twisted.internet import defer from synapse import types from synapse.api.constants import MAX_USERID_LENGTH, LoginType -from synapse.api.errors import ( - AuthError, - Codes, - ConsentNotGivenError, - RegistrationError, - SynapseError, -) +from synapse.api.errors import AuthError, Codes, ConsentNotGivenError, SynapseError from synapse.config.server import is_threepid_reserved from synapse.http.servlet import assert_params_in_dict from synapse.replication.http.login import RegisterDeviceReplicationServlet @@ -165,7 +159,7 @@ class RegistrationHandler(BaseHandler): Returns: Deferred[str]: user_id Raises: - RegistrationError if there was a problem registering. + SynapseError if there was a problem registering. """ yield self.check_registration_ratelimit(address) @@ -182,7 +176,7 @@ class RegistrationHandler(BaseHandler): if not was_guest: try: int(localpart) - raise RegistrationError( + raise SynapseError( 400, "Numeric user IDs are reserved for guest users." ) except ValueError: diff --git a/tests/handlers/test_register.py b/tests/handlers/test_register.py index 1e9ba3a201..e2915eb7b1 100644 --- a/tests/handlers/test_register.py +++ b/tests/handlers/test_register.py @@ -269,8 +269,6 @@ class RegistrationTestCase(unittest.HomeserverTestCase): one will be randomly generated. Returns: A tuple of (user_id, access_token). - Raises: - RegistrationError if there was a problem registering. """ if localpart is None: raise SynapseError(400, "Request must include user id") -- cgit 1.4.1 From 1177d3f3a33bd3ae1eef46fba360d319598359ad Mon Sep 17 00:00:00 2001 From: Andrew Morgan <1342360+anoadragon453@users.noreply.github.com> Date: Mon, 13 Jan 2020 18:10:43 +0000 Subject: Quarantine media by ID or user ID (#6681) --- changelog.d/6681.feature | 1 + docs/admin_api/media_admin_api.md | 76 ++++++- docs/workers.md | 4 +- synapse/rest/admin/media.py | 68 +++++- synapse/storage/data_stores/main/room.py | 116 ++++++++++- tests/rest/admin/test_admin.py | 341 +++++++++++++++++++++++++++++++ tests/rest/client/v1/utils.py | 37 ++++ 7 files changed, 632 insertions(+), 11 deletions(-) create mode 100644 changelog.d/6681.feature (limited to 'tests') diff --git a/changelog.d/6681.feature b/changelog.d/6681.feature new file mode 100644 index 0000000000..5cf19a4e0e --- /dev/null +++ b/changelog.d/6681.feature @@ -0,0 +1 @@ +Add new quarantine media admin APIs to quarantine by media ID or by user who uploaded the media. diff --git a/docs/admin_api/media_admin_api.md b/docs/admin_api/media_admin_api.md index 8b3666d5f5..46ba7a1a71 100644 --- a/docs/admin_api/media_admin_api.md +++ b/docs/admin_api/media_admin_api.md @@ -22,19 +22,81 @@ It returns a JSON body like the following: } ``` -# Quarantine media in a room +# Quarantine media -This API 'quarantines' all the media in a room. +Quarantining media means that it is marked as inaccessible by users. It applies +to any local media, and any locally-cached copies of remote media. -The API is: +The media file itself (and any thumbnails) is not deleted from the server. + +## Quarantining media by ID + +This API quarantines a single piece of local or remote media. + +Request: ``` -POST /_synapse/admin/v1/quarantine_media/ +POST /_synapse/admin/v1/media/quarantine// {} ``` -Quarantining media means that it is marked as inaccessible by users. It applies -to any local media, and any locally-cached copies of remote media. +Where `server_name` is in the form of `example.org`, and `media_id` is in the +form of `abcdefg12345...`. + +Response: + +``` +{} +``` + +## Quarantining media in a room + +This API quarantines all local and remote media in a room. + +Request: + +``` +POST /_synapse/admin/v1/room//media/quarantine + +{} +``` + +Where `room_id` is in the form of `!roomid12345:example.org`. + +Response: + +``` +{ + "num_quarantined": 10 # The number of media items successfully quarantined +} +``` + +Note that there is a legacy endpoint, `POST +/_synapse/admin/v1/quarantine_media/`, that operates the same. +However, it is deprecated and may be removed in a future release. + +## Quarantining all media of a user + +This API quarantines all *local* media that a *local* user has uploaded. That is to say, if +you would like to quarantine media uploaded by a user on a remote homeserver, you should +instead use one of the other APIs. + +Request: + +``` +POST /_synapse/admin/v1/user//media/quarantine + +{} +``` + +Where `user_id` is in the form of `@bob:example.org`. + +Response: + +``` +{ + "num_quarantined": 10 # The number of media items successfully quarantined +} +``` -The media file itself (and any thumbnails) is not deleted from the server. diff --git a/docs/workers.md b/docs/workers.md index f4283aeb05..0ab269fd96 100644 --- a/docs/workers.md +++ b/docs/workers.md @@ -202,7 +202,9 @@ Handles the media repository. It can handle all endpoints starting with: ... and the following regular expressions matching media-specific administration APIs: ^/_synapse/admin/v1/purge_media_cache$ - ^/_synapse/admin/v1/room/.*/media$ + ^/_synapse/admin/v1/room/.*/media.*$ + ^/_synapse/admin/v1/user/.*/media.*$ + ^/_synapse/admin/v1/media/.*$ ^/_synapse/admin/v1/quarantine_media/.*$ You should also set `enable_media_repo: False` in the shared configuration diff --git a/synapse/rest/admin/media.py b/synapse/rest/admin/media.py index fa833e54cf..3a445d6eed 100644 --- a/synapse/rest/admin/media.py +++ b/synapse/rest/admin/media.py @@ -32,16 +32,24 @@ class QuarantineMediaInRoom(RestServlet): this server. """ - PATTERNS = historical_admin_path_patterns("/quarantine_media/(?P[^/]+)") + PATTERNS = ( + historical_admin_path_patterns("/room/(?P[^/]+)/media/quarantine") + + + # This path kept around for legacy reasons + historical_admin_path_patterns("/quarantine_media/(?P![^/]+)") + ) def __init__(self, hs): self.store = hs.get_datastore() self.auth = hs.get_auth() - async def on_POST(self, request, room_id): + async def on_POST(self, request, room_id: str): requester = await self.auth.get_user_by_req(request) await assert_user_is_admin(self.auth, requester.user) + logging.info("Quarantining room: %s", room_id) + + # Quarantine all media in this room num_quarantined = await self.store.quarantine_media_ids_in_room( room_id, requester.user.to_string() ) @@ -49,6 +57,60 @@ class QuarantineMediaInRoom(RestServlet): return 200, {"num_quarantined": num_quarantined} +class QuarantineMediaByUser(RestServlet): + """Quarantines all local media by a given user so that no one can download it via + this server. + """ + + PATTERNS = historical_admin_path_patterns( + "/user/(?P[^/]+)/media/quarantine" + ) + + def __init__(self, hs): + self.store = hs.get_datastore() + self.auth = hs.get_auth() + + async def on_POST(self, request, user_id: str): + requester = await self.auth.get_user_by_req(request) + await assert_user_is_admin(self.auth, requester.user) + + logging.info("Quarantining local media by user: %s", user_id) + + # Quarantine all media this user has uploaded + num_quarantined = await self.store.quarantine_media_ids_by_user( + user_id, requester.user.to_string() + ) + + return 200, {"num_quarantined": num_quarantined} + + +class QuarantineMediaByID(RestServlet): + """Quarantines local or remote media by a given ID so that no one can download + it via this server. + """ + + PATTERNS = historical_admin_path_patterns( + "/media/quarantine/(?P[^/]+)/(?P[^/]+)" + ) + + def __init__(self, hs): + self.store = hs.get_datastore() + self.auth = hs.get_auth() + + async def on_POST(self, request, server_name: str, media_id: str): + requester = await self.auth.get_user_by_req(request) + await assert_user_is_admin(self.auth, requester.user) + + logging.info("Quarantining local media by ID: %s/%s", server_name, media_id) + + # Quarantine this media id + await self.store.quarantine_media_by_id( + server_name, media_id, requester.user.to_string() + ) + + return 200, {} + + class ListMediaInRoom(RestServlet): """Lists all of the media in a given room. """ @@ -94,4 +156,6 @@ def register_servlets_for_media_repo(hs, http_server): """ PurgeMediaCacheRestServlet(hs).register(http_server) QuarantineMediaInRoom(hs).register(http_server) + QuarantineMediaByID(hs).register(http_server) + QuarantineMediaByUser(hs).register(http_server) ListMediaInRoom(hs).register(http_server) diff --git a/synapse/storage/data_stores/main/room.py b/synapse/storage/data_stores/main/room.py index 8636d75030..49bab62be3 100644 --- a/synapse/storage/data_stores/main/room.py +++ b/synapse/storage/data_stores/main/room.py @@ -18,7 +18,7 @@ import collections import logging import re from abc import abstractmethod -from typing import Optional, Tuple +from typing import List, Optional, Tuple from six import integer_types @@ -399,6 +399,8 @@ class RoomWorkerStore(SQLBaseStore): the associated media """ + logger.info("Quarantining media in room: %s", room_id) + def _quarantine_media_in_room_txn(txn): local_mxcs, remote_mxcs = self._get_media_mxcs_in_room_txn(txn, room_id) total_media_quarantined = 0 @@ -494,6 +496,118 @@ class RoomWorkerStore(SQLBaseStore): return local_media_mxcs, remote_media_mxcs + def quarantine_media_by_id( + self, server_name: str, media_id: str, quarantined_by: str, + ): + """quarantines a single local or remote media id + + Args: + server_name: The name of the server that holds this media + media_id: The ID of the media to be quarantined + quarantined_by: The user ID that initiated the quarantine request + """ + logger.info("Quarantining media: %s/%s", server_name, media_id) + is_local = server_name == self.config.server_name + + def _quarantine_media_by_id_txn(txn): + local_mxcs = [media_id] if is_local else [] + remote_mxcs = [(server_name, media_id)] if not is_local else [] + + return self._quarantine_media_txn( + txn, local_mxcs, remote_mxcs, quarantined_by + ) + + return self.db.runInteraction( + "quarantine_media_by_user", _quarantine_media_by_id_txn + ) + + def quarantine_media_ids_by_user(self, user_id: str, quarantined_by: str): + """quarantines all local media associated with a single user + + Args: + user_id: The ID of the user to quarantine media of + quarantined_by: The ID of the user who made the quarantine request + """ + + def _quarantine_media_by_user_txn(txn): + local_media_ids = self._get_media_ids_by_user_txn(txn, user_id) + return self._quarantine_media_txn(txn, local_media_ids, [], quarantined_by) + + return self.db.runInteraction( + "quarantine_media_by_user", _quarantine_media_by_user_txn + ) + + def _get_media_ids_by_user_txn(self, txn, user_id: str, filter_quarantined=True): + """Retrieves local media IDs by a given user + + Args: + txn (cursor) + user_id: The ID of the user to retrieve media IDs of + + Returns: + The local and remote media as a lists of tuples where the key is + the hostname and the value is the media ID. + """ + # Local media + sql = """ + SELECT media_id + FROM local_media_repository + WHERE user_id = ? + """ + if filter_quarantined: + sql += "AND quarantined_by IS NULL" + txn.execute(sql, (user_id,)) + + local_media_ids = [row[0] for row in txn] + + # TODO: Figure out all remote media a user has referenced in a message + + return local_media_ids + + def _quarantine_media_txn( + self, + txn, + local_mxcs: List[str], + remote_mxcs: List[Tuple[str, str]], + quarantined_by: str, + ) -> int: + """Quarantine local and remote media items + + Args: + txn (cursor) + local_mxcs: A list of local mxc URLs + remote_mxcs: A list of (remote server, media id) tuples representing + remote mxc URLs + quarantined_by: The ID of the user who initiated the quarantine request + Returns: + The total number of media items quarantined + """ + total_media_quarantined = 0 + + # Update all the tables to set the quarantined_by flag + txn.executemany( + """ + UPDATE local_media_repository + SET quarantined_by = ? + WHERE media_id = ? + """, + ((quarantined_by, media_id) for media_id in local_mxcs), + ) + + txn.executemany( + """ + UPDATE remote_media_cache + SET quarantined_by = ? + WHERE media_origin = ? AND media_id = ? + """, + ((quarantined_by, origin, media_id) for origin, media_id in remote_mxcs), + ) + + total_media_quarantined += len(local_mxcs) + total_media_quarantined += len(remote_mxcs) + + return total_media_quarantined + class RoomBackgroundUpdateStore(SQLBaseStore): REMOVE_TOMESTONED_ROOMS_BG_UPDATE = "remove_tombstoned_rooms_from_directory" diff --git a/tests/rest/admin/test_admin.py b/tests/rest/admin/test_admin.py index 6ceb483aa8..7a7e898843 100644 --- a/tests/rest/admin/test_admin.py +++ b/tests/rest/admin/test_admin.py @@ -14,11 +14,17 @@ # limitations under the License. import json +import os +import urllib.parse +from binascii import unhexlify from mock import Mock +from twisted.internet.defer import Deferred + import synapse.rest.admin from synapse.http.server import JsonResource +from synapse.logging.context import make_deferred_yieldable from synapse.rest.admin import VersionServlet from synapse.rest.client.v1 import events, login, room from synapse.rest.client.v2_alpha import groups @@ -346,3 +352,338 @@ class PurgeRoomTestCase(unittest.HomeserverTestCase): self.assertEqual(count, 0, msg="Rows not purged in {}".format(table)) test_purge_room.skip = "Disabled because it's currently broken" + + +class QuarantineMediaTestCase(unittest.HomeserverTestCase): + """Test /quarantine_media admin API. + """ + + servlets = [ + synapse.rest.admin.register_servlets, + synapse.rest.admin.register_servlets_for_media_repo, + login.register_servlets, + room.register_servlets, + ] + + def prepare(self, reactor, clock, hs): + self.store = hs.get_datastore() + self.hs = hs + + # Allow for uploading and downloading to/from the media repo + self.media_repo = hs.get_media_repository_resource() + self.download_resource = self.media_repo.children[b"download"] + self.upload_resource = self.media_repo.children[b"upload"] + self.image_data = unhexlify( + b"89504e470d0a1a0a0000000d4948445200000001000000010806" + b"0000001f15c4890000000a49444154789c63000100000500010d" + b"0a2db40000000049454e44ae426082" + ) + + def make_homeserver(self, reactor, clock): + + self.fetches = [] + + def get_file(destination, path, output_stream, args=None, max_size=None): + """ + Returns tuple[int,dict,str,int] of file length, response headers, + absolute URI, and response code. + """ + + def write_to(r): + data, response = r + output_stream.write(data) + return response + + d = Deferred() + d.addCallback(write_to) + self.fetches.append((d, destination, path, args)) + return make_deferred_yieldable(d) + + client = Mock() + client.get_file = get_file + + self.storage_path = self.mktemp() + self.media_store_path = self.mktemp() + os.mkdir(self.storage_path) + os.mkdir(self.media_store_path) + + config = self.default_config() + config["media_store_path"] = self.media_store_path + config["thumbnail_requirements"] = {} + config["max_image_pixels"] = 2000000 + + provider_config = { + "module": "synapse.rest.media.v1.storage_provider.FileStorageProviderBackend", + "store_local": True, + "store_synchronous": False, + "store_remote": True, + "config": {"directory": self.storage_path}, + } + config["media_storage_providers"] = [provider_config] + + hs = self.setup_test_homeserver(config=config, http_client=client) + + return hs + + def test_quarantine_media_requires_admin(self): + self.register_user("nonadmin", "pass", admin=False) + non_admin_user_tok = self.login("nonadmin", "pass") + + # Attempt quarantine media APIs as non-admin + url = "/_synapse/admin/v1/media/quarantine/example.org/abcde12345" + request, channel = self.make_request( + "POST", url.encode("ascii"), access_token=non_admin_user_tok, + ) + self.render(request) + + # Expect a forbidden error + self.assertEqual( + 403, + int(channel.result["code"]), + msg="Expected forbidden on quarantining media as a non-admin", + ) + + # And the roomID/userID endpoint + url = "/_synapse/admin/v1/room/!room%3Aexample.com/media/quarantine" + request, channel = self.make_request( + "POST", url.encode("ascii"), access_token=non_admin_user_tok, + ) + self.render(request) + + # Expect a forbidden error + self.assertEqual( + 403, + int(channel.result["code"]), + msg="Expected forbidden on quarantining media as a non-admin", + ) + + def test_quarantine_media_by_id(self): + self.register_user("id_admin", "pass", admin=True) + admin_user_tok = self.login("id_admin", "pass") + + self.register_user("id_nonadmin", "pass", admin=False) + non_admin_user_tok = self.login("id_nonadmin", "pass") + + # Upload some media into the room + response = self.helper.upload_media( + self.upload_resource, self.image_data, tok=admin_user_tok + ) + + # Extract media ID from the response + server_name_and_media_id = response["content_uri"][ + 6: + ] # Cut off the 'mxc://' bit + server_name, media_id = server_name_and_media_id.split("/") + + # Attempt to access the media + request, channel = self.make_request( + "GET", + server_name_and_media_id, + shorthand=False, + access_token=non_admin_user_tok, + ) + request.render(self.download_resource) + self.pump(1.0) + + # Should be successful + self.assertEqual(200, int(channel.code), msg=channel.result["body"]) + + # Quarantine the media + url = "/_synapse/admin/v1/media/quarantine/%s/%s" % ( + urllib.parse.quote(server_name), + urllib.parse.quote(media_id), + ) + request, channel = self.make_request("POST", url, access_token=admin_user_tok,) + self.render(request) + self.pump(1.0) + self.assertEqual(200, int(channel.code), msg=channel.result["body"]) + + # Attempt to access the media + request, channel = self.make_request( + "GET", + server_name_and_media_id, + shorthand=False, + access_token=admin_user_tok, + ) + request.render(self.download_resource) + self.pump(1.0) + + # Should be quarantined + self.assertEqual( + 404, + int(channel.code), + msg=( + "Expected to receive a 404 on accessing quarantined media: %s" + % server_name_and_media_id + ), + ) + + def test_quarantine_all_media_in_room(self): + self.register_user("room_admin", "pass", admin=True) + admin_user_tok = self.login("room_admin", "pass") + + non_admin_user = self.register_user("room_nonadmin", "pass", admin=False) + non_admin_user_tok = self.login("room_nonadmin", "pass") + + room_id = self.helper.create_room_as(non_admin_user, tok=admin_user_tok) + self.helper.join(room_id, non_admin_user, tok=non_admin_user_tok) + + # Upload some media + response_1 = self.helper.upload_media( + self.upload_resource, self.image_data, tok=non_admin_user_tok + ) + response_2 = self.helper.upload_media( + self.upload_resource, self.image_data, tok=non_admin_user_tok + ) + + # Extract mxcs + mxc_1 = response_1["content_uri"] + mxc_2 = response_2["content_uri"] + + # Send it into the room + self.helper.send_event( + room_id, + "m.room.message", + content={"body": "image-1", "msgtype": "m.image", "url": mxc_1}, + txn_id="111", + tok=non_admin_user_tok, + ) + self.helper.send_event( + room_id, + "m.room.message", + content={"body": "image-2", "msgtype": "m.image", "url": mxc_2}, + txn_id="222", + tok=non_admin_user_tok, + ) + + # Quarantine all media in the room + url = "/_synapse/admin/v1/room/%s/media/quarantine" % urllib.parse.quote( + room_id + ) + request, channel = self.make_request("POST", url, access_token=admin_user_tok,) + self.render(request) + self.pump(1.0) + self.assertEqual(200, int(channel.code), msg=channel.result["body"]) + self.assertEqual( + json.loads(channel.result["body"].decode("utf-8")), + {"num_quarantined": 2}, + "Expected 2 quarantined items", + ) + + # Convert mxc URLs to server/media_id strings + server_and_media_id_1 = mxc_1[6:] + server_and_media_id_2 = mxc_2[6:] + + # Test that we cannot download any of the media anymore + request, channel = self.make_request( + "GET", + server_and_media_id_1, + shorthand=False, + access_token=non_admin_user_tok, + ) + request.render(self.download_resource) + self.pump(1.0) + + # Should be quarantined + self.assertEqual( + 404, + int(channel.code), + msg=( + "Expected to receive a 404 on accessing quarantined media: %s" + % server_and_media_id_1 + ), + ) + + request, channel = self.make_request( + "GET", + server_and_media_id_2, + shorthand=False, + access_token=non_admin_user_tok, + ) + request.render(self.download_resource) + self.pump(1.0) + + # Should be quarantined + self.assertEqual( + 404, + int(channel.code), + msg=( + "Expected to receive a 404 on accessing quarantined media: %s" + % server_and_media_id_2 + ), + ) + + def test_quarantine_all_media_by_user(self): + self.register_user("user_admin", "pass", admin=True) + admin_user_tok = self.login("user_admin", "pass") + + non_admin_user = self.register_user("user_nonadmin", "pass", admin=False) + non_admin_user_tok = self.login("user_nonadmin", "pass") + + # Upload some media + response_1 = self.helper.upload_media( + self.upload_resource, self.image_data, tok=non_admin_user_tok + ) + response_2 = self.helper.upload_media( + self.upload_resource, self.image_data, tok=non_admin_user_tok + ) + + # Extract media IDs + server_and_media_id_1 = response_1["content_uri"][6:] + server_and_media_id_2 = response_2["content_uri"][6:] + + # Quarantine all media by this user + url = "/_synapse/admin/v1/user/%s/media/quarantine" % urllib.parse.quote( + non_admin_user + ) + request, channel = self.make_request( + "POST", url.encode("ascii"), access_token=admin_user_tok, + ) + self.render(request) + self.pump(1.0) + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual( + json.loads(channel.result["body"].decode("utf-8")), + {"num_quarantined": 2}, + "Expected 2 quarantined items", + ) + + # Attempt to access each piece of media + request, channel = self.make_request( + "GET", + server_and_media_id_1, + shorthand=False, + access_token=non_admin_user_tok, + ) + request.render(self.download_resource) + self.pump(1.0) + + # Should be quarantined + self.assertEqual( + 404, + int(channel.code), + msg=( + "Expected to receive a 404 on accessing quarantined media: %s" + % server_and_media_id_1, + ), + ) + + # Attempt to access each piece of media + request, channel = self.make_request( + "GET", + server_and_media_id_2, + shorthand=False, + access_token=non_admin_user_tok, + ) + request.render(self.download_resource) + self.pump(1.0) + + # Should be quarantined + self.assertEqual( + 404, + int(channel.code), + msg=( + "Expected to receive a 404 on accessing quarantined media: %s" + % server_and_media_id_2 + ), + ) diff --git a/tests/rest/client/v1/utils.py b/tests/rest/client/v1/utils.py index e7417b3d14..873d5ef99c 100644 --- a/tests/rest/client/v1/utils.py +++ b/tests/rest/client/v1/utils.py @@ -21,6 +21,8 @@ import time import attr +from twisted.web.resource import Resource + from synapse.api.constants import Membership from tests.server import make_request, render @@ -160,3 +162,38 @@ class RestHelper(object): ) return channel.json_body + + def upload_media( + self, + resource: Resource, + image_data: bytes, + tok: str, + filename: str = "test.png", + expect_code: int = 200, + ) -> dict: + """Upload a piece of test media to the media repo + Args: + resource: The resource that will handle the upload request + image_data: The image data to upload + tok: The user token to use during the upload + filename: The filename of the media to be uploaded + expect_code: The return code to expect from attempting to upload the media + """ + image_length = len(image_data) + path = "/_matrix/media/r0/upload?filename=%s" % (filename,) + request, channel = make_request( + self.hs.get_reactor(), "POST", path, content=image_data, access_token=tok + ) + request.requestHeaders.addRawHeader( + b"Content-Length", str(image_length).encode("UTF-8") + ) + request.render(resource) + self.hs.get_reactor().pump([100]) + + assert channel.code == expect_code, "Expected: %d, got: %d, resp: %r" % ( + expect_code, + int(channel.result["code"]), + channel.result["body"], + ) + + return channel.json_body -- cgit 1.4.1 From 28c98e51ffa166bd717646b0b34228e59f253485 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 15 Jan 2020 14:59:33 +0000 Subject: Add `local_current_membership` table (#6655) Currently we rely on `current_state_events` to figure out what rooms a user was in and their last membership event in there. However, if the server leaves the room then the table may be cleaned up and that information is lost. So lets add a table that separately holds that information. --- changelog.d/6655.misc | 1 + scripts/synapse_port_db | 2 +- synapse/handlers/admin.py | 2 +- synapse/handlers/deactivate_account.py | 2 +- synapse/handlers/initial_sync.py | 2 +- synapse/handlers/room_member.py | 2 +- synapse/handlers/search.py | 2 +- synapse/handlers/sync.py | 2 +- synapse/push/push_tools.py | 2 +- synapse/replication/slave/storage/events.py | 2 +- synapse/server_notices/server_notices_manager.py | 2 +- synapse/storage/data_stores/main/events.py | 30 ++++ synapse/storage/data_stores/main/roommember.py | 189 ++++++++++++--------- .../schema/delta/57/local_current_membership.py | 97 +++++++++++ synapse/storage/prepare_database.py | 2 +- tests/handlers/test_sync.py | 4 +- tests/replication/slave/storage/test_events.py | 4 +- tests/rest/client/v2_alpha/test_account.py | 12 +- tests/rest/client/v2_alpha/test_sync.py | 9 - tests/storage/test_roommember.py | 2 +- 20 files changed, 263 insertions(+), 107 deletions(-) create mode 100644 changelog.d/6655.misc create mode 100644 synapse/storage/data_stores/main/schema/delta/57/local_current_membership.py (limited to 'tests') diff --git a/changelog.d/6655.misc b/changelog.d/6655.misc new file mode 100644 index 0000000000..01e78bc84e --- /dev/null +++ b/changelog.d/6655.misc @@ -0,0 +1 @@ +Add `local_current_membership` table for tracking local user membership state in rooms. diff --git a/scripts/synapse_port_db b/scripts/synapse_port_db index f135c8bc54..5e69104b97 100755 --- a/scripts/synapse_port_db +++ b/scripts/synapse_port_db @@ -470,7 +470,7 @@ class Porter(object): engine.check_database( db_conn, allow_outdated_version=allow_outdated_version ) - prepare_database(db_conn, engine, config=None) + prepare_database(db_conn, engine, config=self.hs_config) store = Store(Database(hs, db_config, engine), db_conn, hs) db_conn.commit() diff --git a/synapse/handlers/admin.py b/synapse/handlers/admin.py index 76d18a8ba8..a9407553b4 100644 --- a/synapse/handlers/admin.py +++ b/synapse/handlers/admin.py @@ -134,7 +134,7 @@ class AdminHandler(BaseHandler): The returned value is that returned by `writer.finished()`. """ # Get all rooms the user is in or has been in - rooms = await self.store.get_rooms_for_user_where_membership_is( + rooms = await self.store.get_rooms_for_local_user_where_membership_is( user_id, membership_list=( Membership.JOIN, diff --git a/synapse/handlers/deactivate_account.py b/synapse/handlers/deactivate_account.py index 4426967f88..2afb390a92 100644 --- a/synapse/handlers/deactivate_account.py +++ b/synapse/handlers/deactivate_account.py @@ -140,7 +140,7 @@ class DeactivateAccountHandler(BaseHandler): user_id (str): The user ID to reject pending invites for. """ user = UserID.from_string(user_id) - pending_invites = await self.store.get_invited_rooms_for_user(user_id) + pending_invites = await self.store.get_invited_rooms_for_local_user(user_id) for room in pending_invites: try: diff --git a/synapse/handlers/initial_sync.py b/synapse/handlers/initial_sync.py index 44ec3e66ae..2e6755f19c 100644 --- a/synapse/handlers/initial_sync.py +++ b/synapse/handlers/initial_sync.py @@ -101,7 +101,7 @@ class InitialSyncHandler(BaseHandler): if include_archived: memberships.append(Membership.LEAVE) - room_list = await self.store.get_rooms_for_user_where_membership_is( + room_list = await self.store.get_rooms_for_local_user_where_membership_is( user_id=user_id, membership_list=memberships ) diff --git a/synapse/handlers/room_member.py b/synapse/handlers/room_member.py index 03bb52ccfb..15e8aa5249 100644 --- a/synapse/handlers/room_member.py +++ b/synapse/handlers/room_member.py @@ -690,7 +690,7 @@ class RoomMemberHandler(object): @defer.inlineCallbacks def _get_inviter(self, user_id, room_id): - invite = yield self.store.get_invite_for_user_in_room( + invite = yield self.store.get_invite_for_local_user_in_room( user_id=user_id, room_id=room_id ) if invite: diff --git a/synapse/handlers/search.py b/synapse/handlers/search.py index ef750d1497..110097eab9 100644 --- a/synapse/handlers/search.py +++ b/synapse/handlers/search.py @@ -179,7 +179,7 @@ class SearchHandler(BaseHandler): search_filter = Filter(filter_dict) # TODO: Search through left rooms too - rooms = yield self.store.get_rooms_for_user_where_membership_is( + rooms = yield self.store.get_rooms_for_local_user_where_membership_is( user.to_string(), membership_list=[Membership.JOIN], # membership_list=[Membership.JOIN, Membership.LEAVE, Membership.Ban], diff --git a/synapse/handlers/sync.py b/synapse/handlers/sync.py index 2d3b8ba73c..cd95f85e3f 100644 --- a/synapse/handlers/sync.py +++ b/synapse/handlers/sync.py @@ -1662,7 +1662,7 @@ class SyncHandler(object): Membership.BAN, ) - room_list = await self.store.get_rooms_for_user_where_membership_is( + room_list = await self.store.get_rooms_for_local_user_where_membership_is( user_id=user_id, membership_list=membership_list ) diff --git a/synapse/push/push_tools.py b/synapse/push/push_tools.py index de5c101a58..5dae4648c0 100644 --- a/synapse/push/push_tools.py +++ b/synapse/push/push_tools.py @@ -21,7 +21,7 @@ from synapse.storage import Storage @defer.inlineCallbacks def get_badge_count(store, user_id): - invites = yield store.get_invited_rooms_for_user(user_id) + invites = yield store.get_invited_rooms_for_local_user(user_id) joins = yield store.get_rooms_for_user(user_id) my_receipts_by_room = yield store.get_receipts_for_user(user_id, "m.read") diff --git a/synapse/replication/slave/storage/events.py b/synapse/replication/slave/storage/events.py index 29f35b9915..3aa6cb8b96 100644 --- a/synapse/replication/slave/storage/events.py +++ b/synapse/replication/slave/storage/events.py @@ -152,7 +152,7 @@ class SlavedEventStore( if etype == EventTypes.Member: self._membership_stream_cache.entity_has_changed(state_key, stream_ordering) - self.get_invited_rooms_for_user.invalidate((state_key,)) + self.get_invited_rooms_for_local_user.invalidate((state_key,)) if relates_to: self.get_relations_for_event.invalidate_many((relates_to,)) diff --git a/synapse/server_notices/server_notices_manager.py b/synapse/server_notices/server_notices_manager.py index 2dac90578c..f7432c8d2f 100644 --- a/synapse/server_notices/server_notices_manager.py +++ b/synapse/server_notices/server_notices_manager.py @@ -105,7 +105,7 @@ class ServerNoticesManager(object): assert self._is_mine_id(user_id), "Cannot send server notices to remote users" - rooms = yield self._store.get_rooms_for_user_where_membership_is( + rooms = yield self._store.get_rooms_for_local_user_where_membership_is( user_id, [Membership.INVITE, Membership.JOIN] ) system_mxid = self._config.server_notices_mxid diff --git a/synapse/storage/data_stores/main/events.py b/synapse/storage/data_stores/main/events.py index 58f35d7f56..e9fe63037b 100644 --- a/synapse/storage/data_stores/main/events.py +++ b/synapse/storage/data_stores/main/events.py @@ -128,6 +128,7 @@ class EventsStore( hs.get_clock().looping_call(_censor_redactions, 5 * 60 * 1000) self._ephemeral_messages_enabled = hs.config.enable_ephemeral_messages + self.is_mine_id = hs.is_mine_id @defer.inlineCallbacks def _read_forward_extremities(self): @@ -547,6 +548,34 @@ class EventsStore( ], ) + # Note: Do we really want to delete rows here (that we do not + # subsequently reinsert below)? While technically correct it means + # we have no record of the fact the user *was* a member of the + # room but got, say, state reset out of it. + if to_delete or to_insert: + txn.executemany( + "DELETE FROM local_current_membership" + " WHERE room_id = ? AND user_id = ?", + ( + (room_id, state_key) + for etype, state_key in itertools.chain(to_delete, to_insert) + if etype == EventTypes.Member and self.is_mine_id(state_key) + ), + ) + + if to_insert: + txn.executemany( + """INSERT INTO local_current_membership + (room_id, user_id, event_id, membership) + VALUES (?, ?, ?, (SELECT membership FROM room_memberships WHERE event_id = ?)) + """, + [ + (room_id, key[1], ev_id, ev_id) + for key, ev_id in to_insert.items() + if key[0] == EventTypes.Member and self.is_mine_id(key[1]) + ], + ) + txn.call_after( self._curr_state_delta_stream_cache.entity_has_changed, room_id, @@ -1724,6 +1753,7 @@ class EventsStore( "local_invites", "room_account_data", "room_tags", + "local_current_membership", ): logger.info("[purge] removing %s from %s", room_id, table) txn.execute("DELETE FROM %s WHERE room_id=?" % (table,), (room_id,)) diff --git a/synapse/storage/data_stores/main/roommember.py b/synapse/storage/data_stores/main/roommember.py index 70ff5751b6..9acef7c950 100644 --- a/synapse/storage/data_stores/main/roommember.py +++ b/synapse/storage/data_stores/main/roommember.py @@ -297,19 +297,22 @@ class RoomMemberWorkerStore(EventsWorkerStore): return {row[0]: row[1] for row in txn} @cached() - def get_invited_rooms_for_user(self, user_id): - """ Get all the rooms the user is invited to + def get_invited_rooms_for_local_user(self, user_id): + """ Get all the rooms the *local* user is invited to + Args: user_id (str): The user ID. Returns: A deferred list of RoomsForUser. """ - return self.get_rooms_for_user_where_membership_is(user_id, [Membership.INVITE]) + return self.get_rooms_for_local_user_where_membership_is( + user_id, [Membership.INVITE] + ) @defer.inlineCallbacks - def get_invite_for_user_in_room(self, user_id, room_id): - """Gets the invite for the given user and room + def get_invite_for_local_user_in_room(self, user_id, room_id): + """Gets the invite for the given *local* user and room Args: user_id (str) @@ -319,15 +322,15 @@ class RoomMemberWorkerStore(EventsWorkerStore): Deferred: Resolves to either a RoomsForUser or None if no invite was found. """ - invites = yield self.get_invited_rooms_for_user(user_id) + invites = yield self.get_invited_rooms_for_local_user(user_id) for invite in invites: if invite.room_id == room_id: return invite return None @defer.inlineCallbacks - def get_rooms_for_user_where_membership_is(self, user_id, membership_list): - """ Get all the rooms for this user where the membership for this user + def get_rooms_for_local_user_where_membership_is(self, user_id, membership_list): + """ Get all the rooms for this *local* user where the membership for this user matches one in the membership list. Filters out forgotten rooms. @@ -344,8 +347,8 @@ class RoomMemberWorkerStore(EventsWorkerStore): return defer.succeed(None) rooms = yield self.db.runInteraction( - "get_rooms_for_user_where_membership_is", - self._get_rooms_for_user_where_membership_is_txn, + "get_rooms_for_local_user_where_membership_is", + self._get_rooms_for_local_user_where_membership_is_txn, user_id, membership_list, ) @@ -354,76 +357,42 @@ class RoomMemberWorkerStore(EventsWorkerStore): forgotten_rooms = yield self.get_forgotten_rooms_for_user(user_id) return [room for room in rooms if room.room_id not in forgotten_rooms] - def _get_rooms_for_user_where_membership_is_txn( + def _get_rooms_for_local_user_where_membership_is_txn( self, txn, user_id, membership_list ): + # Paranoia check. + if not self.hs.is_mine_id(user_id): + raise Exception( + "Cannot call 'get_rooms_for_local_user_where_membership_is' on non-local user %r" + % (user_id,), + ) - do_invite = Membership.INVITE in membership_list - membership_list = [m for m in membership_list if m != Membership.INVITE] - - results = [] - if membership_list: - if self._current_state_events_membership_up_to_date: - clause, args = make_in_list_sql_clause( - self.database_engine, "c.membership", membership_list - ) - sql = """ - SELECT room_id, e.sender, c.membership, event_id, e.stream_ordering - FROM current_state_events AS c - INNER JOIN events AS e USING (room_id, event_id) - WHERE - c.type = 'm.room.member' - AND state_key = ? - AND %s - """ % ( - clause, - ) - else: - clause, args = make_in_list_sql_clause( - self.database_engine, "m.membership", membership_list - ) - sql = """ - SELECT room_id, e.sender, m.membership, event_id, e.stream_ordering - FROM current_state_events AS c - INNER JOIN room_memberships AS m USING (room_id, event_id) - INNER JOIN events AS e USING (room_id, event_id) - WHERE - c.type = 'm.room.member' - AND state_key = ? - AND %s - """ % ( - clause, - ) - - txn.execute(sql, (user_id, *args)) - results = [RoomsForUser(**r) for r in self.db.cursor_to_dict(txn)] + clause, args = make_in_list_sql_clause( + self.database_engine, "c.membership", membership_list + ) - if do_invite: - sql = ( - "SELECT i.room_id, inviter, i.event_id, e.stream_ordering" - " FROM local_invites as i" - " INNER JOIN events as e USING (event_id)" - " WHERE invitee = ? AND locally_rejected is NULL" - " AND replaced_by is NULL" - ) + sql = """ + SELECT room_id, e.sender, c.membership, event_id, e.stream_ordering + FROM local_current_membership AS c + INNER JOIN events AS e USING (room_id, event_id) + WHERE + user_id = ? + AND %s + """ % ( + clause, + ) - txn.execute(sql, (user_id,)) - results.extend( - RoomsForUser( - room_id=r["room_id"], - sender=r["inviter"], - event_id=r["event_id"], - stream_ordering=r["stream_ordering"], - membership=Membership.INVITE, - ) - for r in self.db.cursor_to_dict(txn) - ) + txn.execute(sql, (user_id, *args)) + results = [RoomsForUser(**r) for r in self.db.cursor_to_dict(txn)] return results - @cachedInlineCallbacks(max_entries=500000, iterable=True) + @cached(max_entries=500000, iterable=True) def get_rooms_for_user_with_stream_ordering(self, user_id): - """Returns a set of room_ids the user is currently joined to + """Returns a set of room_ids the user is currently joined to. + + If a remote user only returns rooms this server is currently + participating in. Args: user_id (str) @@ -433,17 +402,49 @@ class RoomMemberWorkerStore(EventsWorkerStore): the rooms the user is in currently, along with the stream ordering of the most recent join for that user and room. """ - rooms = yield self.get_rooms_for_user_where_membership_is( - user_id, membership_list=[Membership.JOIN] - ) - return frozenset( - GetRoomsForUserWithStreamOrdering(r.room_id, r.stream_ordering) - for r in rooms + return self.db.runInteraction( + "get_rooms_for_user_with_stream_ordering", + self._get_rooms_for_user_with_stream_ordering_txn, + user_id, ) + def _get_rooms_for_user_with_stream_ordering_txn(self, txn, user_id): + # We use `current_state_events` here and not `local_current_membership` + # as a) this gets called with remote users and b) this only gets called + # for rooms the server is participating in. + if self._current_state_events_membership_up_to_date: + sql = """ + SELECT room_id, e.stream_ordering + FROM current_state_events AS c + INNER JOIN events AS e USING (room_id, event_id) + WHERE + c.type = 'm.room.member' + AND state_key = ? + AND c.membership = ? + """ + else: + sql = """ + SELECT room_id, e.stream_ordering + FROM current_state_events AS c + INNER JOIN room_memberships AS m USING (room_id, event_id) + INNER JOIN events AS e USING (room_id, event_id) + WHERE + c.type = 'm.room.member' + AND state_key = ? + AND m.membership = ? + """ + + txn.execute(sql, (user_id, Membership.JOIN)) + results = frozenset(GetRoomsForUserWithStreamOrdering(*row) for row in txn) + + return results + @defer.inlineCallbacks def get_rooms_for_user(self, user_id, on_invalidate=None): - """Returns a set of room_ids the user is currently joined to + """Returns a set of room_ids the user is currently joined to. + + If a remote user only returns rooms this server is currently + participating in. """ rooms = yield self.get_rooms_for_user_with_stream_ordering( user_id, on_invalidate=on_invalidate @@ -1022,7 +1023,7 @@ class RoomMemberStore(RoomMemberWorkerStore, RoomMemberBackgroundUpdateStore): event.internal_metadata.stream_ordering, ) txn.call_after( - self.get_invited_rooms_for_user.invalidate, (event.state_key,) + self.get_invited_rooms_for_local_user.invalidate, (event.state_key,) ) # We update the local_invites table only if the event is "current", @@ -1064,6 +1065,27 @@ class RoomMemberStore(RoomMemberWorkerStore, RoomMemberBackgroundUpdateStore): ), ) + # We also update the `local_current_membership` table with + # latest invite info. This will usually get updated by the + # `current_state_events` handling, unless its an outlier. + if event.internal_metadata.is_outlier(): + # This should only happen for out of band memberships, so + # we add a paranoia check. + assert event.internal_metadata.is_out_of_band_membership() + + self.db.simple_upsert_txn( + txn, + table="local_current_membership", + keyvalues={ + "room_id": event.room_id, + "user_id": event.state_key, + }, + values={ + "event_id": event.event_id, + "membership": event.membership, + }, + ) + @defer.inlineCallbacks def locally_reject_invite(self, user_id, room_id): sql = ( @@ -1075,6 +1097,15 @@ class RoomMemberStore(RoomMemberWorkerStore, RoomMemberBackgroundUpdateStore): def f(txn, stream_ordering): txn.execute(sql, (stream_ordering, True, room_id, user_id)) + # We also clear this entry from `local_current_membership`. + # Ideally we'd point to a leave event, but we don't have one, so + # nevermind. + self.db.simple_delete_txn( + txn, + table="local_current_membership", + keyvalues={"room_id": room_id, "user_id": user_id}, + ) + with self._stream_id_gen.get_next() as stream_ordering: yield self.db.runInteraction("locally_reject_invite", f, stream_ordering) diff --git a/synapse/storage/data_stores/main/schema/delta/57/local_current_membership.py b/synapse/storage/data_stores/main/schema/delta/57/local_current_membership.py new file mode 100644 index 0000000000..601c236c4a --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/57/local_current_membership.py @@ -0,0 +1,97 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + + +# We create a new table called `local_current_membership` that stores the latest +# membership state of local users in rooms, which helps track leaves/bans/etc +# even if the server has left the room (and so has deleted the room from +# `current_state_events`). This will also include outstanding invites for local +# users for rooms the server isn't in. +# +# If the server isn't and hasn't been in the room then it will only include +# outsstanding invites, and not e.g. pre-emptive bans of local users. +# +# If the server later rejoins a room `local_current_membership` can simply be +# replaced with the new current state of the room (which results in the +# equivalent behaviour as if the server had remained in the room). + + +def run_upgrade(cur, database_engine, config, *args, **kwargs): + # We need to do the insert in `run_upgrade` section as we don't have access + # to `config` in `run_create`. + + # This upgrade may take a bit of time for large servers (e.g. one minute for + # matrix.org) but means we avoid a lots of book keeping required to do it as + # a background update. + + # We check if the `current_state_events.membership` is up to date by + # checking if the relevant background update has finished. If it has + # finished we can avoid doing a join against `room_memberships`, which + # speesd things up. + cur.execute( + """SELECT 1 FROM background_updates + WHERE update_name = 'current_state_events_membership' + """ + ) + current_state_membership_up_to_date = not bool(cur.fetchone()) + + # Cheekily drop and recreate indices, as that is faster. + cur.execute("DROP INDEX local_current_membership_idx") + cur.execute("DROP INDEX local_current_membership_room_idx") + + if current_state_membership_up_to_date: + sql = """ + INSERT INTO local_current_membership (room_id, user_id, event_id, membership) + SELECT c.room_id, state_key AS user_id, event_id, c.membership + FROM current_state_events AS c + WHERE type = 'm.room.member' AND c.membership IS NOT NULL AND state_key like '%' || ? + """ + else: + # We can't rely on the membership column, so we need to join against + # `room_memberships`. + sql = """ + INSERT INTO local_current_membership (room_id, user_id, event_id, membership) + SELECT c.room_id, state_key AS user_id, event_id, r.membership + FROM current_state_events AS c + INNER JOIN room_memberships AS r USING (event_id) + WHERE type = 'm.room.member' and state_key like '%' || ? + """ + cur.execute(sql, (config.server_name,)) + + cur.execute( + "CREATE UNIQUE INDEX local_current_membership_idx ON local_current_membership(user_id, room_id)" + ) + cur.execute( + "CREATE INDEX local_current_membership_room_idx ON local_current_membership(room_id)" + ) + + +def run_create(cur, database_engine, *args, **kwargs): + cur.execute( + """ + CREATE TABLE local_current_membership ( + room_id TEXT NOT NULL, + user_id TEXT NOT NULL, + event_id TEXT NOT NULL, + membership TEXT NOT NULL + )""" + ) + + cur.execute( + "CREATE UNIQUE INDEX local_current_membership_idx ON local_current_membership(user_id, room_id)" + ) + cur.execute( + "CREATE INDEX local_current_membership_room_idx ON local_current_membership(room_id)" + ) diff --git a/synapse/storage/prepare_database.py b/synapse/storage/prepare_database.py index e70026b80a..e86984cd50 100644 --- a/synapse/storage/prepare_database.py +++ b/synapse/storage/prepare_database.py @@ -29,7 +29,7 @@ logger = logging.getLogger(__name__) # Remember to update this number every time a change is made to database # schema files, so the users will be informed on server restarts. -SCHEMA_VERSION = 56 +SCHEMA_VERSION = 57 dir_path = os.path.abspath(os.path.dirname(__file__)) diff --git a/tests/handlers/test_sync.py b/tests/handlers/test_sync.py index 758ee071a5..4cbe9784ed 100644 --- a/tests/handlers/test_sync.py +++ b/tests/handlers/test_sync.py @@ -32,8 +32,8 @@ class SyncTestCase(tests.unittest.HomeserverTestCase): def test_wait_for_sync_for_user_auth_blocking(self): - user_id1 = "@user1:server" - user_id2 = "@user2:server" + user_id1 = "@user1:test" + user_id2 = "@user2:test" sync_config = self._generate_sync_config(user_id1) self.reactor.advance(100) # So we get not 0 time diff --git a/tests/replication/slave/storage/test_events.py b/tests/replication/slave/storage/test_events.py index b68e9fe082..b1b037006d 100644 --- a/tests/replication/slave/storage/test_events.py +++ b/tests/replication/slave/storage/test_events.py @@ -115,13 +115,13 @@ class SlavedEventStoreTestCase(BaseSlavedStoreTestCase): def test_invites(self): self.persist(type="m.room.create", key="", creator=USER_ID) - self.check("get_invited_rooms_for_user", [USER_ID_2], []) + self.check("get_invited_rooms_for_local_user", [USER_ID_2], []) event = self.persist(type="m.room.member", key=USER_ID_2, membership="invite") self.replicate() self.check( - "get_invited_rooms_for_user", + "get_invited_rooms_for_local_user", [USER_ID_2], [ RoomsForUser( diff --git a/tests/rest/client/v2_alpha/test_account.py b/tests/rest/client/v2_alpha/test_account.py index 0f51895b81..c3facc00eb 100644 --- a/tests/rest/client/v2_alpha/test_account.py +++ b/tests/rest/client/v2_alpha/test_account.py @@ -285,7 +285,9 @@ class DeactivateTestCase(unittest.HomeserverTestCase): ) # Make sure the invite is here. - pending_invites = self.get_success(store.get_invited_rooms_for_user(invitee_id)) + pending_invites = self.get_success( + store.get_invited_rooms_for_local_user(invitee_id) + ) self.assertEqual(len(pending_invites), 1, pending_invites) self.assertEqual(pending_invites[0].room_id, room_id, pending_invites) @@ -293,12 +295,16 @@ class DeactivateTestCase(unittest.HomeserverTestCase): self.deactivate(invitee_id, invitee_tok) # Check that the invite isn't there anymore. - pending_invites = self.get_success(store.get_invited_rooms_for_user(invitee_id)) + pending_invites = self.get_success( + store.get_invited_rooms_for_local_user(invitee_id) + ) self.assertEqual(len(pending_invites), 0, pending_invites) # Check that the membership of @invitee:test in the room is now "leave". memberships = self.get_success( - store.get_rooms_for_user_where_membership_is(invitee_id, [Membership.LEAVE]) + store.get_rooms_for_local_user_where_membership_is( + invitee_id, [Membership.LEAVE] + ) ) self.assertEqual(len(memberships), 1, memberships) self.assertEqual(memberships[0].room_id, room_id, memberships) diff --git a/tests/rest/client/v2_alpha/test_sync.py b/tests/rest/client/v2_alpha/test_sync.py index 661c1f88b9..9c13a13786 100644 --- a/tests/rest/client/v2_alpha/test_sync.py +++ b/tests/rest/client/v2_alpha/test_sync.py @@ -15,8 +15,6 @@ # limitations under the License. import json -from mock import Mock - import synapse.rest.admin from synapse.api.constants import EventContentFields, EventTypes from synapse.rest.client.v1 import login, room @@ -36,13 +34,6 @@ class FilterTestCase(unittest.HomeserverTestCase): sync.register_servlets, ] - def make_homeserver(self, reactor, clock): - - hs = self.setup_test_homeserver( - "red", http_client=None, federation_client=Mock() - ) - return hs - def test_sync_argless(self): request, channel = self.make_request("GET", "/sync") self.render(request) diff --git a/tests/storage/test_roommember.py b/tests/storage/test_roommember.py index 7840f63fe3..00df0ea68e 100644 --- a/tests/storage/test_roommember.py +++ b/tests/storage/test_roommember.py @@ -57,7 +57,7 @@ class RoomMemberStoreTestCase(unittest.HomeserverTestCase): self.room = self.helper.create_room_as(self.u_alice, tok=self.t_alice) rooms_for_user = self.get_success( - self.store.get_rooms_for_user_where_membership_is( + self.store.get_rooms_for_local_user_where_membership_is( self.u_alice, [Membership.JOIN] ) ) -- cgit 1.4.1 From 8f5d7302acb7f6d15ba7051df7fd7fda7375a29e Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Wed, 15 Jan 2020 15:58:55 +0000 Subject: Implement RedirectException (#6687) Allow REST endpoint implemnentations to raise a RedirectException, which will redirect the user's browser to a given location. --- changelog.d/6687.misc | 1 + synapse/api/errors.py | 27 ++++++++++++++++- synapse/http/server.py | 13 ++++++--- tests/test_server.py | 79 ++++++++++++++++++++++++++++++++++++++++++++++++-- 4 files changed, 113 insertions(+), 7 deletions(-) create mode 100644 changelog.d/6687.misc (limited to 'tests') diff --git a/changelog.d/6687.misc b/changelog.d/6687.misc new file mode 100644 index 0000000000..deb0454602 --- /dev/null +++ b/changelog.d/6687.misc @@ -0,0 +1 @@ +Allow REST endpoint implementations to raise a RedirectException, which will redirect the user's browser to a given location. diff --git a/synapse/api/errors.py b/synapse/api/errors.py index 9e9844b47c..1c9456e583 100644 --- a/synapse/api/errors.py +++ b/synapse/api/errors.py @@ -17,13 +17,15 @@ """Contains exceptions and error codes.""" import logging -from typing import Dict +from typing import Dict, List from six import iteritems from six.moves import http_client from canonicaljson import json +from twisted.web import http + logger = logging.getLogger(__name__) @@ -80,6 +82,29 @@ class CodeMessageException(RuntimeError): self.msg = msg +class RedirectException(CodeMessageException): + """A pseudo-error indicating that we want to redirect the client to a different + location + + Attributes: + cookies: a list of set-cookies values to add to the response. For example: + b"sessionId=a3fWa; Expires=Wed, 21 Oct 2015 07:28:00 GMT" + """ + + def __init__(self, location: bytes, http_code: int = http.FOUND): + """ + + Args: + location: the URI to redirect to + http_code: the HTTP response code + """ + msg = "Redirect to %s" % (location.decode("utf-8"),) + super().__init__(code=http_code, msg=msg) + self.location = location + + self.cookies = [] # type: List[bytes] + + class SynapseError(CodeMessageException): """A base exception type for matrix errors which have an errcode and error message (as well as an HTTP status code). diff --git a/synapse/http/server.py b/synapse/http/server.py index 943d12c907..04bc2385a2 100644 --- a/synapse/http/server.py +++ b/synapse/http/server.py @@ -14,8 +14,8 @@ # See the License for the specific language governing permissions and # limitations under the License. -import cgi import collections +import html import http.client import logging import types @@ -36,6 +36,7 @@ import synapse.metrics from synapse.api.errors import ( CodeMessageException, Codes, + RedirectException, SynapseError, UnrecognizedRequestError, ) @@ -153,14 +154,18 @@ def _return_html_error(f, request): Args: f (twisted.python.failure.Failure): - request (twisted.web.iweb.IRequest): + request (twisted.web.server.Request): """ if f.check(CodeMessageException): cme = f.value code = cme.code msg = cme.msg - if isinstance(cme, SynapseError): + if isinstance(cme, RedirectException): + logger.info("%s redirect to %s", request, cme.location) + request.setHeader(b"location", cme.location) + request.cookies.extend(cme.cookies) + elif isinstance(cme, SynapseError): logger.info("%s SynapseError: %s - %s", request, code, msg) else: logger.error( @@ -178,7 +183,7 @@ def _return_html_error(f, request): exc_info=(f.type, f.value, f.getTracebackObject()), ) - body = HTML_ERROR_TEMPLATE.format(code=code, msg=cgi.escape(msg)).encode("utf-8") + body = HTML_ERROR_TEMPLATE.format(code=code, msg=html.escape(msg)).encode("utf-8") request.setResponseCode(code) request.setHeader(b"Content-Type", b"text/html; charset=utf-8") request.setHeader(b"Content-Length", b"%i" % (len(body),)) diff --git a/tests/test_server.py b/tests/test_server.py index 98fef21d55..0d57eed268 100644 --- a/tests/test_server.py +++ b/tests/test_server.py @@ -23,8 +23,12 @@ from twisted.test.proto_helpers import AccumulatingProtocol from twisted.web.resource import Resource from twisted.web.server import NOT_DONE_YET -from synapse.api.errors import Codes, SynapseError -from synapse.http.server import JsonResource +from synapse.api.errors import Codes, RedirectException, SynapseError +from synapse.http.server import ( + DirectServeResource, + JsonResource, + wrap_html_request_handler, +) from synapse.http.site import SynapseSite, logger from synapse.logging.context import make_deferred_yieldable from synapse.util import Clock @@ -164,6 +168,77 @@ class JsonResourceTests(unittest.TestCase): self.assertEqual(channel.json_body["errcode"], "M_UNRECOGNIZED") +class WrapHtmlRequestHandlerTests(unittest.TestCase): + class TestResource(DirectServeResource): + callback = None + + @wrap_html_request_handler + async def _async_render_GET(self, request): + return await self.callback(request) + + def setUp(self): + self.reactor = ThreadedMemoryReactorClock() + + def test_good_response(self): + def callback(request): + request.write(b"response") + request.finish() + + res = WrapHtmlRequestHandlerTests.TestResource() + res.callback = callback + + request, channel = make_request(self.reactor, b"GET", b"/path") + render(request, res, self.reactor) + + self.assertEqual(channel.result["code"], b"200") + body = channel.result["body"] + self.assertEqual(body, b"response") + + def test_redirect_exception(self): + """ + If the callback raises a RedirectException, it is turned into a 30x + with the right location. + """ + + def callback(request, **kwargs): + raise RedirectException(b"/look/an/eagle", 301) + + res = WrapHtmlRequestHandlerTests.TestResource() + res.callback = callback + + request, channel = make_request(self.reactor, b"GET", b"/path") + render(request, res, self.reactor) + + self.assertEqual(channel.result["code"], b"301") + headers = channel.result["headers"] + location_headers = [v for k, v in headers if k == b"Location"] + self.assertEqual(location_headers, [b"/look/an/eagle"]) + + def test_redirect_exception_with_cookie(self): + """ + If the callback raises a RedirectException which sets a cookie, that is + returned too + """ + + def callback(request, **kwargs): + e = RedirectException(b"/no/over/there", 304) + e.cookies.append(b"session=yespls") + raise e + + res = WrapHtmlRequestHandlerTests.TestResource() + res.callback = callback + + request, channel = make_request(self.reactor, b"GET", b"/path") + render(request, res, self.reactor) + + self.assertEqual(channel.result["code"], b"304") + headers = channel.result["headers"] + location_headers = [v for k, v in headers if k == b"Location"] + self.assertEqual(location_headers, [b"/no/over/there"]) + cookies_headers = [v for k, v in headers if k == b"Set-Cookie"] + self.assertEqual(cookies_headers, [b"session=yespls"]) + + class SiteTestCase(unittest.HomeserverTestCase): def test_lose_connection(self): """ -- cgit 1.4.1 From 19a1aac48cc83fe41287a97bb0a96280a0e8c565 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 15 Jan 2020 18:13:47 +0000 Subject: Fix purge_room admin API (#6711) --- changelog.d/6711.bugfix | 1 + synapse/storage/purge_events.py | 2 +- tests/rest/admin/test_admin.py | 4 +--- 3 files changed, 3 insertions(+), 4 deletions(-) create mode 100644 changelog.d/6711.bugfix (limited to 'tests') diff --git a/changelog.d/6711.bugfix b/changelog.d/6711.bugfix new file mode 100644 index 0000000000..c70506bd88 --- /dev/null +++ b/changelog.d/6711.bugfix @@ -0,0 +1 @@ +Fix `purge_room` admin API. diff --git a/synapse/storage/purge_events.py b/synapse/storage/purge_events.py index d6a7bd7834..fdc0abf5cf 100644 --- a/synapse/storage/purge_events.py +++ b/synapse/storage/purge_events.py @@ -34,7 +34,7 @@ class PurgeEventsStorage(object): """ state_groups_to_delete = yield self.stores.main.purge_room(room_id) - yield self.stores.main.purge_room_state(room_id, state_groups_to_delete) + yield self.stores.state.purge_room_state(room_id, state_groups_to_delete) @defer.inlineCallbacks def purge_history(self, room_id, token, delete_local_events): diff --git a/tests/rest/admin/test_admin.py b/tests/rest/admin/test_admin.py index 7a7e898843..f3b4a31e21 100644 --- a/tests/rest/admin/test_admin.py +++ b/tests/rest/admin/test_admin.py @@ -337,7 +337,7 @@ class PurgeRoomTestCase(unittest.HomeserverTestCase): "local_invites", "room_account_data", "room_tags", - "state_groups", + # "state_groups", # Current impl leaves orphaned state groups around. "state_groups_state", ): count = self.get_success( @@ -351,8 +351,6 @@ class PurgeRoomTestCase(unittest.HomeserverTestCase): self.assertEqual(count, 0, msg="Rows not purged in {}".format(table)) - test_purge_room.skip = "Disabled because it's currently broken" - class QuarantineMediaTestCase(unittest.HomeserverTestCase): """Test /quarantine_media admin API. -- cgit 1.4.1 From 48c3a96886de64f3141ad68b8163cd2fc0c197ff Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Thu, 16 Jan 2020 09:16:12 +0000 Subject: Port synapse.replication.tcp to async/await (#6666) * Port synapse.replication.tcp to async/await * Newsfile * Correctly document type of on_ functions as async * Don't be overenthusiastic with the asyncing.... --- changelog.d/6666.misc | 1 + synapse/app/admin_cmd.py | 3 +- synapse/app/appservice.py | 5 +-- synapse/app/federation_sender.py | 5 +-- synapse/app/pusher.py | 5 +-- synapse/app/synchrotron.py | 5 +-- synapse/app/user_dir.py | 5 +-- synapse/federation/send_queue.py | 4 +- synapse/handlers/typing.py | 2 +- synapse/replication/tcp/client.py | 11 ++--- synapse/replication/tcp/protocol.py | 72 ++++++++++++++----------------- synapse/replication/tcp/resource.py | 31 ++++++------- synapse/replication/tcp/streams/_base.py | 25 +++++------ synapse/replication/tcp/streams/events.py | 9 ++-- tests/replication/tcp/streams/_base.py | 2 +- 15 files changed, 80 insertions(+), 105 deletions(-) create mode 100644 changelog.d/6666.misc (limited to 'tests') diff --git a/changelog.d/6666.misc b/changelog.d/6666.misc new file mode 100644 index 0000000000..e79c23d2d2 --- /dev/null +++ b/changelog.d/6666.misc @@ -0,0 +1 @@ +Port `synapse.replication.tcp` to async/await. diff --git a/synapse/app/admin_cmd.py b/synapse/app/admin_cmd.py index 8e36bc57d3..1c7c6ec0c8 100644 --- a/synapse/app/admin_cmd.py +++ b/synapse/app/admin_cmd.py @@ -84,8 +84,7 @@ class AdminCmdServer(HomeServer): class AdminCmdReplicationHandler(ReplicationClientHandler): - @defer.inlineCallbacks - def on_rdata(self, stream_name, token, rows): + async def on_rdata(self, stream_name, token, rows): pass def get_streams_to_replicate(self): diff --git a/synapse/app/appservice.py b/synapse/app/appservice.py index e82e0f11e3..2217d4a4fb 100644 --- a/synapse/app/appservice.py +++ b/synapse/app/appservice.py @@ -115,9 +115,8 @@ class ASReplicationHandler(ReplicationClientHandler): super(ASReplicationHandler, self).__init__(hs.get_datastore()) self.appservice_handler = hs.get_application_service_handler() - @defer.inlineCallbacks - def on_rdata(self, stream_name, token, rows): - yield super(ASReplicationHandler, self).on_rdata(stream_name, token, rows) + async def on_rdata(self, stream_name, token, rows): + await super(ASReplicationHandler, self).on_rdata(stream_name, token, rows) if stream_name == "events": max_stream_id = self.store.get_room_max_stream_ordering() diff --git a/synapse/app/federation_sender.py b/synapse/app/federation_sender.py index 83c436229c..a57cf991ac 100644 --- a/synapse/app/federation_sender.py +++ b/synapse/app/federation_sender.py @@ -145,9 +145,8 @@ class FederationSenderReplicationHandler(ReplicationClientHandler): super(FederationSenderReplicationHandler, self).__init__(hs.get_datastore()) self.send_handler = FederationSenderHandler(hs, self) - @defer.inlineCallbacks - def on_rdata(self, stream_name, token, rows): - yield super(FederationSenderReplicationHandler, self).on_rdata( + async def on_rdata(self, stream_name, token, rows): + await super(FederationSenderReplicationHandler, self).on_rdata( stream_name, token, rows ) self.send_handler.process_replication_rows(stream_name, token, rows) diff --git a/synapse/app/pusher.py b/synapse/app/pusher.py index 09e639040a..e46b6ac598 100644 --- a/synapse/app/pusher.py +++ b/synapse/app/pusher.py @@ -141,9 +141,8 @@ class PusherReplicationHandler(ReplicationClientHandler): self.pusher_pool = hs.get_pusherpool() - @defer.inlineCallbacks - def on_rdata(self, stream_name, token, rows): - yield super(PusherReplicationHandler, self).on_rdata(stream_name, token, rows) + async def on_rdata(self, stream_name, token, rows): + await super(PusherReplicationHandler, self).on_rdata(stream_name, token, rows) run_in_background(self.poke_pushers, stream_name, token, rows) @defer.inlineCallbacks diff --git a/synapse/app/synchrotron.py b/synapse/app/synchrotron.py index 03031ee34d..3218da07bd 100644 --- a/synapse/app/synchrotron.py +++ b/synapse/app/synchrotron.py @@ -358,9 +358,8 @@ class SyncReplicationHandler(ReplicationClientHandler): self.presence_handler = hs.get_presence_handler() self.notifier = hs.get_notifier() - @defer.inlineCallbacks - def on_rdata(self, stream_name, token, rows): - yield super(SyncReplicationHandler, self).on_rdata(stream_name, token, rows) + async def on_rdata(self, stream_name, token, rows): + await super(SyncReplicationHandler, self).on_rdata(stream_name, token, rows) run_in_background(self.process_and_notify, stream_name, token, rows) def get_streams_to_replicate(self): diff --git a/synapse/app/user_dir.py b/synapse/app/user_dir.py index 1257098f92..ba536d6f04 100644 --- a/synapse/app/user_dir.py +++ b/synapse/app/user_dir.py @@ -172,9 +172,8 @@ class UserDirectoryReplicationHandler(ReplicationClientHandler): super(UserDirectoryReplicationHandler, self).__init__(hs.get_datastore()) self.user_directory = hs.get_user_directory_handler() - @defer.inlineCallbacks - def on_rdata(self, stream_name, token, rows): - yield super(UserDirectoryReplicationHandler, self).on_rdata( + async def on_rdata(self, stream_name, token, rows): + await super(UserDirectoryReplicationHandler, self).on_rdata( stream_name, token, rows ) if stream_name == EventsStream.NAME: diff --git a/synapse/federation/send_queue.py b/synapse/federation/send_queue.py index ced4925a98..174f6e42be 100644 --- a/synapse/federation/send_queue.py +++ b/synapse/federation/send_queue.py @@ -259,7 +259,9 @@ class FederationRemoteSendQueue(object): def federation_ack(self, token): self._clear_queue_before_pos(token) - def get_replication_rows(self, from_token, to_token, limit, federation_ack=None): + async def get_replication_rows( + self, from_token, to_token, limit, federation_ack=None + ): """Get rows to be sent over federation between the two tokens Args: diff --git a/synapse/handlers/typing.py b/synapse/handlers/typing.py index b635c339ed..d5ca9cb07b 100644 --- a/synapse/handlers/typing.py +++ b/synapse/handlers/typing.py @@ -257,7 +257,7 @@ class TypingHandler(object): "typing_key", self._latest_room_serial, rooms=[member.room_id] ) - def get_all_typing_updates(self, last_id, current_id): + async def get_all_typing_updates(self, last_id, current_id): if last_id == current_id: return [] diff --git a/synapse/replication/tcp/client.py b/synapse/replication/tcp/client.py index aa7fd90e26..52a0aefe68 100644 --- a/synapse/replication/tcp/client.py +++ b/synapse/replication/tcp/client.py @@ -110,7 +110,7 @@ class ReplicationClientHandler(AbstractReplicationClientHandler): port = hs.config.worker_replication_port hs.get_reactor().connectTCP(host, port, self.factory) - def on_rdata(self, stream_name, token, rows): + async def on_rdata(self, stream_name, token, rows): """Called to handle a batch of replication data with a given stream token. By default this just pokes the slave store. Can be overridden in subclasses to @@ -121,20 +121,17 @@ class ReplicationClientHandler(AbstractReplicationClientHandler): token (int): stream token for this batch of rows rows (list): a list of Stream.ROW_TYPE objects as returned by Stream.parse_row. - - Returns: - Deferred|None """ logger.debug("Received rdata %s -> %s", stream_name, token) - return self.store.process_replication_rows(stream_name, token, rows) + self.store.process_replication_rows(stream_name, token, rows) - def on_position(self, stream_name, token): + async def on_position(self, stream_name, token): """Called when we get new position data. By default this just pokes the slave store. Can be overriden in subclasses to handle more. """ - return self.store.process_replication_rows(stream_name, token, []) + self.store.process_replication_rows(stream_name, token, []) def on_sync(self, data): """When we received a SYNC we wake up any deferreds that were waiting diff --git a/synapse/replication/tcp/protocol.py b/synapse/replication/tcp/protocol.py index db0353c996..5f4bdf84d2 100644 --- a/synapse/replication/tcp/protocol.py +++ b/synapse/replication/tcp/protocol.py @@ -81,12 +81,11 @@ from synapse.replication.tcp.commands import ( SyncCommand, UserSyncCommand, ) +from synapse.replication.tcp.streams import STREAMS_MAP from synapse.types import Collection from synapse.util import Clock from synapse.util.stringutils import random_string -from .streams import STREAMS_MAP - connection_close_counter = Counter( "synapse_replication_tcp_protocol_close_reason", "", ["reason_type"] ) @@ -241,19 +240,16 @@ class BaseReplicationStreamProtocol(LineOnlyReceiver): "replication-" + cmd.get_logcontext_id(), self.handle_command, cmd ) - def handle_command(self, cmd): + async def handle_command(self, cmd: Command): """Handle a command we have received over the replication stream. - By default delegates to on_ + By default delegates to on_, which should return an awaitable. Args: - cmd (synapse.replication.tcp.commands.Command): received command - - Returns: - Deferred + cmd: received command """ handler = getattr(self, "on_%s" % (cmd.NAME,)) - return handler(cmd) + await handler(cmd) def close(self): logger.warning("[%s] Closing connection", self.id()) @@ -326,10 +322,10 @@ class BaseReplicationStreamProtocol(LineOnlyReceiver): for cmd in pending: self.send_command(cmd) - def on_PING(self, line): + async def on_PING(self, line): self.received_ping = True - def on_ERROR(self, cmd): + async def on_ERROR(self, cmd): logger.error("[%s] Remote reported error: %r", self.id(), cmd.data) def pauseProducing(self): @@ -429,16 +425,16 @@ class ServerReplicationStreamProtocol(BaseReplicationStreamProtocol): BaseReplicationStreamProtocol.connectionMade(self) self.streamer.new_connection(self) - def on_NAME(self, cmd): + async def on_NAME(self, cmd): logger.info("[%s] Renamed to %r", self.id(), cmd.data) self.name = cmd.data - def on_USER_SYNC(self, cmd): - return self.streamer.on_user_sync( + async def on_USER_SYNC(self, cmd): + await self.streamer.on_user_sync( self.conn_id, cmd.user_id, cmd.is_syncing, cmd.last_sync_ms ) - def on_REPLICATE(self, cmd): + async def on_REPLICATE(self, cmd): stream_name = cmd.stream_name token = cmd.token @@ -449,23 +445,23 @@ class ServerReplicationStreamProtocol(BaseReplicationStreamProtocol): for stream in iterkeys(self.streamer.streams_by_name) ] - return make_deferred_yieldable( + await make_deferred_yieldable( defer.gatherResults(deferreds, consumeErrors=True) ) else: - return self.subscribe_to_stream(stream_name, token) + await self.subscribe_to_stream(stream_name, token) - def on_FEDERATION_ACK(self, cmd): - return self.streamer.federation_ack(cmd.token) + async def on_FEDERATION_ACK(self, cmd): + self.streamer.federation_ack(cmd.token) - def on_REMOVE_PUSHER(self, cmd): - return self.streamer.on_remove_pusher(cmd.app_id, cmd.push_key, cmd.user_id) + async def on_REMOVE_PUSHER(self, cmd): + await self.streamer.on_remove_pusher(cmd.app_id, cmd.push_key, cmd.user_id) - def on_INVALIDATE_CACHE(self, cmd): - return self.streamer.on_invalidate_cache(cmd.cache_func, cmd.keys) + async def on_INVALIDATE_CACHE(self, cmd): + self.streamer.on_invalidate_cache(cmd.cache_func, cmd.keys) - def on_USER_IP(self, cmd): - return self.streamer.on_user_ip( + async def on_USER_IP(self, cmd): + self.streamer.on_user_ip( cmd.user_id, cmd.access_token, cmd.ip, @@ -474,8 +470,7 @@ class ServerReplicationStreamProtocol(BaseReplicationStreamProtocol): cmd.last_seen, ) - @defer.inlineCallbacks - def subscribe_to_stream(self, stream_name, token): + async def subscribe_to_stream(self, stream_name, token): """Subscribe the remote to a stream. This invloves checking if they've missed anything and sending those @@ -487,7 +482,7 @@ class ServerReplicationStreamProtocol(BaseReplicationStreamProtocol): try: # Get missing updates - updates, current_token = yield self.streamer.get_stream_updates( + updates, current_token = await self.streamer.get_stream_updates( stream_name, token ) @@ -572,7 +567,7 @@ class AbstractReplicationClientHandler(metaclass=abc.ABCMeta): """ @abc.abstractmethod - def on_rdata(self, stream_name, token, rows): + async def on_rdata(self, stream_name, token, rows): """Called to handle a batch of replication data with a given stream token. Args: @@ -580,14 +575,11 @@ class AbstractReplicationClientHandler(metaclass=abc.ABCMeta): token (int): stream token for this batch of rows rows (list): a list of Stream.ROW_TYPE objects as returned by Stream.parse_row. - - Returns: - Deferred|None """ raise NotImplementedError() @abc.abstractmethod - def on_position(self, stream_name, token): + async def on_position(self, stream_name, token): """Called when we get new position data.""" raise NotImplementedError() @@ -676,12 +668,12 @@ class ClientReplicationStreamProtocol(BaseReplicationStreamProtocol): if not self.streams_connecting: self.handler.finished_connecting() - def on_SERVER(self, cmd): + async def on_SERVER(self, cmd): if cmd.data != self.server_name: logger.error("[%s] Connected to wrong remote: %r", self.id(), cmd.data) self.send_error("Wrong remote") - def on_RDATA(self, cmd): + async def on_RDATA(self, cmd): stream_name = cmd.stream_name inbound_rdata_count.labels(stream_name).inc() @@ -701,19 +693,19 @@ class ClientReplicationStreamProtocol(BaseReplicationStreamProtocol): # Check if this is the last of a batch of updates rows = self.pending_batches.pop(stream_name, []) rows.append(row) - return self.handler.on_rdata(stream_name, cmd.token, rows) + await self.handler.on_rdata(stream_name, cmd.token, rows) - def on_POSITION(self, cmd): + async def on_POSITION(self, cmd): # When we get a `POSITION` command it means we've finished getting # missing updates for the given stream, and are now up to date. self.streams_connecting.discard(cmd.stream_name) if not self.streams_connecting: self.handler.finished_connecting() - return self.handler.on_position(cmd.stream_name, cmd.token) + await self.handler.on_position(cmd.stream_name, cmd.token) - def on_SYNC(self, cmd): - return self.handler.on_sync(cmd.data) + async def on_SYNC(self, cmd): + self.handler.on_sync(cmd.data) def replicate(self, stream_name, token): """Send the subscription request to the server diff --git a/synapse/replication/tcp/resource.py b/synapse/replication/tcp/resource.py index cbfdaf5773..b1752e88cd 100644 --- a/synapse/replication/tcp/resource.py +++ b/synapse/replication/tcp/resource.py @@ -23,7 +23,6 @@ from six import itervalues from prometheus_client import Counter -from twisted.internet import defer from twisted.internet.protocol import Factory from synapse.metrics import LaterGauge @@ -155,8 +154,7 @@ class ReplicationStreamer(object): run_as_background_process("replication_notifier", self._run_notifier_loop) - @defer.inlineCallbacks - def _run_notifier_loop(self): + async def _run_notifier_loop(self): self.is_looping = True try: @@ -185,7 +183,7 @@ class ReplicationStreamer(object): continue if self._replication_torture_level: - yield self.clock.sleep( + await self.clock.sleep( self._replication_torture_level / 1000.0 ) @@ -196,7 +194,7 @@ class ReplicationStreamer(object): stream.upto_token, ) try: - updates, current_token = yield stream.get_updates() + updates, current_token = await stream.get_updates() except Exception: logger.info("Failed to handle stream %s", stream.NAME) raise @@ -233,7 +231,7 @@ class ReplicationStreamer(object): self.is_looping = False @measure_func("repl.get_stream_updates") - def get_stream_updates(self, stream_name, token): + async def get_stream_updates(self, stream_name, token): """For a given stream get all updates since token. This is called when a client first subscribes to a stream. """ @@ -241,7 +239,7 @@ class ReplicationStreamer(object): if not stream: raise Exception("unknown stream %s", stream_name) - return stream.get_updates_since(token) + return await stream.get_updates_since(token) @measure_func("repl.federation_ack") def federation_ack(self, token): @@ -252,22 +250,20 @@ class ReplicationStreamer(object): self.federation_sender.federation_ack(token) @measure_func("repl.on_user_sync") - @defer.inlineCallbacks - def on_user_sync(self, conn_id, user_id, is_syncing, last_sync_ms): + async def on_user_sync(self, conn_id, user_id, is_syncing, last_sync_ms): """A client has started/stopped syncing on a worker. """ user_sync_counter.inc() - yield self.presence_handler.update_external_syncs_row( + await self.presence_handler.update_external_syncs_row( conn_id, user_id, is_syncing, last_sync_ms ) @measure_func("repl.on_remove_pusher") - @defer.inlineCallbacks - def on_remove_pusher(self, app_id, push_key, user_id): + async def on_remove_pusher(self, app_id, push_key, user_id): """A client has asked us to remove a pusher """ remove_pusher_counter.inc() - yield self.store.delete_pusher_by_app_id_pushkey_user_id( + await self.store.delete_pusher_by_app_id_pushkey_user_id( app_id=app_id, pushkey=push_key, user_id=user_id ) @@ -281,15 +277,16 @@ class ReplicationStreamer(object): getattr(self.store, cache_func).invalidate(tuple(keys)) @measure_func("repl.on_user_ip") - @defer.inlineCallbacks - def on_user_ip(self, user_id, access_token, ip, user_agent, device_id, last_seen): + async def on_user_ip( + self, user_id, access_token, ip, user_agent, device_id, last_seen + ): """The client saw a user request """ user_ip_cache_counter.inc() - yield self.store.insert_client_ip( + await self.store.insert_client_ip( user_id, access_token, ip, user_agent, device_id, last_seen ) - yield self._server_notices_sender.on_user_ip(user_id) + await self._server_notices_sender.on_user_ip(user_id) def send_sync_to_all_connections(self, data): """Sends a SYNC command to all clients. diff --git a/synapse/replication/tcp/streams/_base.py b/synapse/replication/tcp/streams/_base.py index 4ab0334fc1..e03e77199b 100644 --- a/synapse/replication/tcp/streams/_base.py +++ b/synapse/replication/tcp/streams/_base.py @@ -19,8 +19,6 @@ import logging from collections import namedtuple from typing import Any -from twisted.internet import defer - logger = logging.getLogger(__name__) @@ -144,8 +142,7 @@ class Stream(object): self.upto_token = self.current_token() self.last_token = self.upto_token - @defer.inlineCallbacks - def get_updates(self): + async def get_updates(self): """Gets all updates since the last time this function was called (or since the stream was constructed if it hadn't been called before), until the `upto_token` @@ -156,13 +153,12 @@ class Stream(object): list of ``(token, row)`` entries. ``row`` will be json-serialised and sent over the replication steam. """ - updates, current_token = yield self.get_updates_since(self.last_token) + updates, current_token = await self.get_updates_since(self.last_token) self.last_token = current_token return updates, current_token - @defer.inlineCallbacks - def get_updates_since(self, from_token): + async def get_updates_since(self, from_token): """Like get_updates except allows specifying from when we should stream updates @@ -182,15 +178,16 @@ class Stream(object): if from_token == current_token: return [], current_token + logger.info("get_updates_since: %s", self.__class__) if self._LIMITED: - rows = yield self.update_function( + rows = await self.update_function( from_token, current_token, limit=MAX_EVENTS_BEHIND + 1 ) # never turn more than MAX_EVENTS_BEHIND + 1 into updates. rows = itertools.islice(rows, MAX_EVENTS_BEHIND + 1) else: - rows = yield self.update_function(from_token, current_token) + rows = await self.update_function(from_token, current_token) updates = [(row[0], row[1:]) for row in rows] @@ -295,9 +292,8 @@ class PushRulesStream(Stream): push_rules_token, _ = self.store.get_push_rules_stream_token() return push_rules_token - @defer.inlineCallbacks - def update_function(self, from_token, to_token, limit): - rows = yield self.store.get_all_push_rule_updates(from_token, to_token, limit) + async def update_function(self, from_token, to_token, limit): + rows = await self.store.get_all_push_rule_updates(from_token, to_token, limit) return [(row[0], row[2]) for row in rows] @@ -413,9 +409,8 @@ class AccountDataStream(Stream): super(AccountDataStream, self).__init__(hs) - @defer.inlineCallbacks - def update_function(self, from_token, to_token, limit): - global_results, room_results = yield self.store.get_all_updated_account_data( + async def update_function(self, from_token, to_token, limit): + global_results, room_results = await self.store.get_all_updated_account_data( from_token, from_token, to_token, limit ) diff --git a/synapse/replication/tcp/streams/events.py b/synapse/replication/tcp/streams/events.py index 0843e5aa90..b3afabb8cd 100644 --- a/synapse/replication/tcp/streams/events.py +++ b/synapse/replication/tcp/streams/events.py @@ -19,8 +19,6 @@ from typing import Tuple, Type import attr -from twisted.internet import defer - from ._base import Stream @@ -122,16 +120,15 @@ class EventsStream(Stream): super(EventsStream, self).__init__(hs) - @defer.inlineCallbacks - def update_function(self, from_token, current_token, limit=None): - event_rows = yield self._store.get_all_new_forward_event_rows( + async def update_function(self, from_token, current_token, limit=None): + event_rows = await self._store.get_all_new_forward_event_rows( from_token, current_token, limit ) event_updates = ( (row[0], EventsStreamEventRow.TypeId, row[1:]) for row in event_rows ) - state_rows = yield self._store.get_all_updated_current_state_deltas( + state_rows = await self._store.get_all_updated_current_state_deltas( from_token, current_token, limit ) state_updates = ( diff --git a/tests/replication/tcp/streams/_base.py b/tests/replication/tcp/streams/_base.py index 1d14e77255..e96ad4ca4e 100644 --- a/tests/replication/tcp/streams/_base.py +++ b/tests/replication/tcp/streams/_base.py @@ -73,6 +73,6 @@ class TestReplicationClientHandler(object): def finished_connecting(self): pass - def on_rdata(self, stream_name, token, rows): + async def on_rdata(self, stream_name, token, rows): for r in rows: self.received_rdata_rows.append((stream_name, token, r)) -- cgit 1.4.1 From acc7820574426cf27673d941b1b0362272113351 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Thu, 16 Jan 2020 22:26:34 +0000 Subject: Log saml assertions rather than the whole response ... since the whole response is huge. We even need to break up the assertions, since kibana otherwise truncates them. --- synapse/handlers/saml_handler.py | 13 ++++++++++- synapse/util/iterutils.py | 13 +++++++++++ tests/util/test_itertools.py | 47 ++++++++++++++++++++++++++++++++++++++++ 3 files changed, 72 insertions(+), 1 deletion(-) create mode 100644 tests/util/test_itertools.py (limited to 'tests') diff --git a/synapse/handlers/saml_handler.py b/synapse/handlers/saml_handler.py index 107f97032b..32638671c9 100644 --- a/synapse/handlers/saml_handler.py +++ b/synapse/handlers/saml_handler.py @@ -32,6 +32,7 @@ from synapse.types import ( mxid_localpart_allowed_characters, ) from synapse.util.async_helpers import Linearizer +from synapse.util.iterutils import chunk_seq logger = logging.getLogger(__name__) @@ -132,7 +133,17 @@ class SamlHandler: logger.warning("SAML2 response was not signed") raise SynapseError(400, "SAML2 response was not signed") - logger.info("SAML2 response: %s", saml2_auth.origxml) + logger.debug("SAML2 response: %s", saml2_auth.origxml) + for assertion in saml2_auth.assertions: + # kibana limits the length of a log field, whereas this is all rather + # useful, so split it up. + count = 0 + for part in chunk_seq(str(assertion), 10000): + logger.info( + "SAML2 assertion: %s%s", "(%i)..." % (count,) if count else "", part + ) + count += 1 + logger.info("SAML2 mapped attributes: %s", saml2_auth.ava) try: diff --git a/synapse/util/iterutils.py b/synapse/util/iterutils.py index c10016fbc5..06faeebe7f 100644 --- a/synapse/util/iterutils.py +++ b/synapse/util/iterutils.py @@ -33,3 +33,16 @@ def batch_iter(iterable: Iterable[T], size: int) -> Iterator[Tuple[T]]: sourceiter = iter(iterable) # call islice until it returns an empty tuple return iter(lambda: tuple(islice(sourceiter, size)), ()) + + +ISeq = TypeVar("ISeq", bound=Sequence, covariant=True) + + +def chunk_seq(iseq: ISeq, maxlen: int) -> Iterable[ISeq]: + """Split the given sequence into chunks of the given size + + The last chunk may be shorter than the given size. + + If the input is empty, no chunks are returned. + """ + return (iseq[i : i + maxlen] for i in range(0, len(iseq), maxlen)) diff --git a/tests/util/test_itertools.py b/tests/util/test_itertools.py new file mode 100644 index 0000000000..0ab0a91483 --- /dev/null +++ b/tests/util/test_itertools.py @@ -0,0 +1,47 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +from synapse.util.iterutils import chunk_seq + +from tests.unittest import TestCase + + +class ChunkSeqTests(TestCase): + def test_short_seq(self): + parts = chunk_seq("123", 8) + + self.assertEqual( + list(parts), ["123"], + ) + + def test_long_seq(self): + parts = chunk_seq("abcdefghijklmnop", 8) + + self.assertEqual( + list(parts), ["abcdefgh", "ijklmnop"], + ) + + def test_uneven_parts(self): + parts = chunk_seq("abcdefghijklmnop", 5) + + self.assertEqual( + list(parts), ["abcde", "fghij", "klmno", "p"], + ) + + def test_empty_input(self): + parts = chunk_seq([], 5) + + self.assertEqual( + list(parts), [], + ) -- cgit 1.4.1 From ceecedc68ba1af25b0ee60c5cf927fd1fd245b9f Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Mon, 20 Jan 2020 17:23:59 +0000 Subject: Fix changing password via user admin API. (#6730) --- changelog.d/6730.bugfix | 1 + synapse/rest/admin/users.py | 4 ++-- tests/rest/admin/test_user.py | 13 +++++++++++++ 3 files changed, 16 insertions(+), 2 deletions(-) create mode 100644 changelog.d/6730.bugfix (limited to 'tests') diff --git a/changelog.d/6730.bugfix b/changelog.d/6730.bugfix new file mode 100644 index 0000000000..beb444ca66 --- /dev/null +++ b/changelog.d/6730.bugfix @@ -0,0 +1 @@ +Fix changing password via user admin API. diff --git a/synapse/rest/admin/users.py b/synapse/rest/admin/users.py index 574cb90c74..c178c960c5 100644 --- a/synapse/rest/admin/users.py +++ b/synapse/rest/admin/users.py @@ -193,8 +193,8 @@ class UserRestServletV2(RestServlet): raise SynapseError(400, "Invalid password") else: new_password = body["password"] - await self._set_password_handler.set_password( - target_user, new_password, requester + await self.set_password_handler.set_password( + target_user.to_string(), new_password, requester ) if "deactivated" in body: diff --git a/tests/rest/admin/test_user.py b/tests/rest/admin/test_user.py index 7352d609e6..8f09f51c61 100644 --- a/tests/rest/admin/test_user.py +++ b/tests/rest/admin/test_user.py @@ -435,6 +435,19 @@ class UserRestTestCase(unittest.HomeserverTestCase): self.assertEqual(0, channel.json_body["is_guest"]) self.assertEqual(0, channel.json_body["deactivated"]) + # Change password + body = json.dumps({"password": "hahaha"}) + + request, channel = self.make_request( + "PUT", + self.url, + access_token=self.admin_user_tok, + content=body.encode(encoding="utf_8"), + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + # Modify user body = json.dumps({"displayname": "foobar", "deactivated": True}) -- cgit 1.4.1 From 74b74462f1c8b2db9b0995cbf64d879cbfce0dc4 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Mon, 20 Jan 2020 17:38:09 +0000 Subject: Fix `/events/:event_id` deprecated API. (#6731) --- changelog.d/6731.bugfix | 1 + synapse/rest/client/v1/events.py | 2 +- tests/rest/client/v1/test_events.py | 27 +++++++++++++++++++++++++++ tests/unittest.py | 2 +- 4 files changed, 30 insertions(+), 2 deletions(-) create mode 100644 changelog.d/6731.bugfix (limited to 'tests') diff --git a/changelog.d/6731.bugfix b/changelog.d/6731.bugfix new file mode 100644 index 0000000000..21f6e15cbd --- /dev/null +++ b/changelog.d/6731.bugfix @@ -0,0 +1 @@ +Fix `/events/:event_id` deprecated API. diff --git a/synapse/rest/client/v1/events.py b/synapse/rest/client/v1/events.py index 4beb617733..25effd0261 100644 --- a/synapse/rest/client/v1/events.py +++ b/synapse/rest/client/v1/events.py @@ -70,7 +70,6 @@ class EventStreamRestServlet(RestServlet): return 200, {} -# TODO: Unit test gets, with and without auth, with different kinds of events. class EventRestServlet(RestServlet): PATTERNS = client_patterns("/events/(?P[^/]*)$", v1=True) @@ -78,6 +77,7 @@ class EventRestServlet(RestServlet): super(EventRestServlet, self).__init__() self.clock = hs.get_clock() self.event_handler = hs.get_event_handler() + self.auth = hs.get_auth() self._event_serializer = hs.get_event_client_serializer() async def on_GET(self, request, event_id): diff --git a/tests/rest/client/v1/test_events.py b/tests/rest/client/v1/test_events.py index f340b7e851..ffb2de1505 100644 --- a/tests/rest/client/v1/test_events.py +++ b/tests/rest/client/v1/test_events.py @@ -134,3 +134,30 @@ class EventStreamPermissionsTestCase(unittest.HomeserverTestCase): # someone else set topic, expect 6 (join,send,topic,join,send,topic) pass + + +class GetEventsTestCase(unittest.HomeserverTestCase): + servlets = [ + events.register_servlets, + room.register_servlets, + synapse.rest.admin.register_servlets_for_client_rest_resource, + login.register_servlets, + ] + + def prepare(self, hs, reactor, clock): + + # register an account + self.user_id = self.register_user("sid1", "pass") + self.token = self.login(self.user_id, "pass") + + self.room_id = self.helper.create_room_as(self.user_id, tok=self.token) + + def test_get_event_via_events(self): + resp = self.helper.send(self.room_id, tok=self.token) + event_id = resp["event_id"] + + request, channel = self.make_request( + "GET", "/events/" + event_id, access_token=self.token, + ) + self.render(request) + self.assertEquals(channel.code, 200, msg=channel.result) diff --git a/tests/unittest.py b/tests/unittest.py index ddcd4becfe..b56e249386 100644 --- a/tests/unittest.py +++ b/tests/unittest.py @@ -463,7 +463,7 @@ class HomeserverTestCase(TestCase): # Create the user request, channel = self.make_request("GET", "/_matrix/client/r0/admin/register") self.render(request) - self.assertEqual(channel.code, 200) + self.assertEqual(channel.code, 200, msg=channel.result) nonce = channel.json_body["nonce"] want_mac = hmac.new(key=b"shared", digestmod=hashlib.sha1) -- cgit 1.4.1 From aa9b00fb2f9a7718d67fb11621a83035492ed9fb Mon Sep 17 00:00:00 2001 From: Andrew Morgan <1342360+anoadragon453@users.noreply.github.com> Date: Wed, 22 Jan 2020 11:05:50 +0000 Subject: Fix and add test to deprecated quarantine media admin api (#6756) --- changelog.d/6756.feature | 1 + synapse/rest/admin/media.py | 2 +- tests/rest/admin/test_admin.py | 15 +++++++++++---- 3 files changed, 13 insertions(+), 5 deletions(-) create mode 100644 changelog.d/6756.feature (limited to 'tests') diff --git a/changelog.d/6756.feature b/changelog.d/6756.feature new file mode 100644 index 0000000000..6328c868f2 --- /dev/null +++ b/changelog.d/6756.feature @@ -0,0 +1 @@ +Add new quarantine media admin APIs to quarantine by media ID or by user who uploaded the media. \ No newline at end of file diff --git a/synapse/rest/admin/media.py b/synapse/rest/admin/media.py index 3a445d6eed..ee75095c0e 100644 --- a/synapse/rest/admin/media.py +++ b/synapse/rest/admin/media.py @@ -36,7 +36,7 @@ class QuarantineMediaInRoom(RestServlet): historical_admin_path_patterns("/room/(?P[^/]+)/media/quarantine") + # This path kept around for legacy reasons - historical_admin_path_patterns("/quarantine_media/(?P![^/]+)") + historical_admin_path_patterns("/quarantine_media/(?P[^/]+)") ) def __init__(self, hs): diff --git a/tests/rest/admin/test_admin.py b/tests/rest/admin/test_admin.py index f3b4a31e21..af4d604e50 100644 --- a/tests/rest/admin/test_admin.py +++ b/tests/rest/admin/test_admin.py @@ -516,7 +516,7 @@ class QuarantineMediaTestCase(unittest.HomeserverTestCase): ), ) - def test_quarantine_all_media_in_room(self): + def test_quarantine_all_media_in_room(self, override_url_template=None): self.register_user("room_admin", "pass", admin=True) admin_user_tok = self.login("room_admin", "pass") @@ -555,9 +555,12 @@ class QuarantineMediaTestCase(unittest.HomeserverTestCase): ) # Quarantine all media in the room - url = "/_synapse/admin/v1/room/%s/media/quarantine" % urllib.parse.quote( - room_id - ) + if override_url_template: + url = override_url_template % urllib.parse.quote(room_id) + else: + url = "/_synapse/admin/v1/room/%s/media/quarantine" % urllib.parse.quote( + room_id + ) request, channel = self.make_request("POST", url, access_token=admin_user_tok,) self.render(request) self.pump(1.0) @@ -611,6 +614,10 @@ class QuarantineMediaTestCase(unittest.HomeserverTestCase): ), ) + def test_quaraantine_all_media_in_room_deprecated_api_path(self): + # Perform the above test with the deprecated API path + self.test_quarantine_all_media_in_room("/_synapse/admin/v1/quarantine_media/%s") + def test_quarantine_all_media_by_user(self): self.register_user("user_admin", "pass", admin=True) admin_user_tok = self.login("user_admin", "pass") -- cgit 1.4.1 From 67aa18e8dc81d6d5c0645fc330e049ed7d2e786e Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Wed, 22 Jan 2020 12:28:07 +0000 Subject: Add tests for thumbnailing --- tests/rest/media/v1/test_media_storage.py | 48 +++++++++++++++++++++++++++++-- 1 file changed, 45 insertions(+), 3 deletions(-) (limited to 'tests') diff --git a/tests/rest/media/v1/test_media_storage.py b/tests/rest/media/v1/test_media_storage.py index bc662b61db..d80a7daed3 100644 --- a/tests/rest/media/v1/test_media_storage.py +++ b/tests/rest/media/v1/test_media_storage.py @@ -17,7 +17,7 @@ import os import shutil import tempfile -from binascii import unhexlify +from binascii import unhexlify, hexlify from mock import Mock from six.moves.urllib import parse @@ -149,6 +149,7 @@ class MediaRepoTests(unittest.HomeserverTestCase): self.media_repo = hs.get_media_repository_resource() self.download_resource = self.media_repo.children[b"download"] + self.thumbnail_resource = self.media_repo.children[b"thumbnail"] # smol png self.end_content = unhexlify( @@ -157,10 +158,12 @@ class MediaRepoTests(unittest.HomeserverTestCase): b"0a2db40000000049454e44ae426082" ) + self.media_id = "example.com/12345" + def _req(self, content_disposition): request, channel = self.make_request( - "GET", "example.com/12345", shorthand=False + "GET", self.media_id, shorthand=False ) request.render(self.download_resource) self.pump() @@ -170,7 +173,7 @@ class MediaRepoTests(unittest.HomeserverTestCase): self.assertEqual(len(self.fetches), 1) self.assertEqual(self.fetches[0][1], "example.com") self.assertEqual( - self.fetches[0][2], "/_matrix/media/v1/download/example.com/12345" + self.fetches[0][2], "/_matrix/media/v1/download/" + self.media_id ) self.assertEqual(self.fetches[0][3], {"allow_remote": "false"}) @@ -229,3 +232,42 @@ class MediaRepoTests(unittest.HomeserverTestCase): headers = channel.headers self.assertEqual(headers.getRawHeaders(b"Content-Type"), [b"image/png"]) self.assertEqual(headers.getRawHeaders(b"Content-Disposition"), None) + + def test_thumbnail_crop(self): + expected_body = unhexlify( + b"89504e470d0a1a0a0000000d4948445200000020000000200806" + b"000000737a7af40000001a49444154789cedc101010000008220" + b"ffaf6e484001000000ef0610200001194334ee0000000049454e" + b"44ae426082" + ) + + self._test_thumbnail("crop", expected_body) + + def test_thumbnail_scale(self): + expected_body = unhexlify( + b"89504e470d0a1a0a0000000d4948445200000001000000010806" + b"0000001f15c4890000000d49444154789c636060606000000005" + b"0001a5f645400000000049454e44ae426082" + ) + + self._test_thumbnail("scale", expected_body) + + def _test_thumbnail(self, method, expected_body): + params = "?width=32&height=32&method=" + method + request, channel = self.make_request( + "GET", self.media_id + params, shorthand=False + ) + request.render(self.thumbnail_resource) + self.pump() + + headers = { + b"Content-Length": [b"%d" % (len(self.end_content))], + b"Content-Type": [b"image/png"], + } + self.fetches[0][0].callback( + (self.end_content, (len(self.end_content), headers)) + ) + self.pump() + + self.assertEqual(channel.code, 200) + self.assertEqual(channel.result["body"], expected_body, channel.result["body"]) -- cgit 1.4.1 From d9a8728b1183c416a89a79856f3b5185386600b2 Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Wed, 22 Jan 2020 12:30:49 +0000 Subject: Remove unused import --- tests/rest/media/v1/test_media_storage.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tests') diff --git a/tests/rest/media/v1/test_media_storage.py b/tests/rest/media/v1/test_media_storage.py index d80a7daed3..6345cc7637 100644 --- a/tests/rest/media/v1/test_media_storage.py +++ b/tests/rest/media/v1/test_media_storage.py @@ -17,7 +17,7 @@ import os import shutil import tempfile -from binascii import unhexlify, hexlify +from binascii import unhexlify from mock import Mock from six.moves.urllib import parse -- cgit 1.4.1 From 6ae0c8db3335faa9b5f0e4407f7a4a3713c84062 Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Wed, 22 Jan 2020 12:38:18 +0000 Subject: Lint + changelog --- changelog.d/6764.misc | 1 + tests/rest/media/v1/test_media_storage.py | 4 +--- 2 files changed, 2 insertions(+), 3 deletions(-) create mode 100644 changelog.d/6764.misc (limited to 'tests') diff --git a/changelog.d/6764.misc b/changelog.d/6764.misc new file mode 100644 index 0000000000..8edd767405 --- /dev/null +++ b/changelog.d/6764.misc @@ -0,0 +1 @@ +Fixup `synapse.rest` to pass mypy. diff --git a/tests/rest/media/v1/test_media_storage.py b/tests/rest/media/v1/test_media_storage.py index 6345cc7637..1809ceb839 100644 --- a/tests/rest/media/v1/test_media_storage.py +++ b/tests/rest/media/v1/test_media_storage.py @@ -162,9 +162,7 @@ class MediaRepoTests(unittest.HomeserverTestCase): def _req(self, content_disposition): - request, channel = self.make_request( - "GET", self.media_id, shorthand=False - ) + request, channel = self.make_request("GET", self.media_id, shorthand=False) request.render(self.download_resource) self.pump() -- cgit 1.4.1 From 90a28fb475a29daa9e7a9ee7204f6f76cc8af441 Mon Sep 17 00:00:00 2001 From: Andrew Morgan <1342360+anoadragon453@users.noreply.github.com> Date: Wed, 22 Jan 2020 13:36:43 +0000 Subject: Admin API to list, filter and sort rooms (#6720) --- changelog.d/6720.feature | 1 + docs/admin_api/rooms.md | 173 ++++++++++++++ synapse/rest/admin/__init__.py | 3 +- synapse/rest/admin/_base.py | 15 ++ synapse/rest/admin/rooms.py | 82 +++++++ synapse/rest/client/v2_alpha/_base.py | 2 +- synapse/storage/data_stores/main/room.py | 125 +++++++++- tests/rest/admin/test_admin.py | 393 ++++++++++++++++++++++++++++++- 8 files changed, 787 insertions(+), 7 deletions(-) create mode 100644 changelog.d/6720.feature create mode 100644 docs/admin_api/rooms.md (limited to 'tests') diff --git a/changelog.d/6720.feature b/changelog.d/6720.feature new file mode 100644 index 0000000000..dfc1b74d62 --- /dev/null +++ b/changelog.d/6720.feature @@ -0,0 +1 @@ +Add a new admin API to list and filter rooms on the server. \ No newline at end of file diff --git a/docs/admin_api/rooms.md b/docs/admin_api/rooms.md new file mode 100644 index 0000000000..082721ea95 --- /dev/null +++ b/docs/admin_api/rooms.md @@ -0,0 +1,173 @@ +# List Room API + +The List Room admin API allows server admins to get a list of rooms on their +server. There are various parameters available that allow for filtering and +sorting the returned list. This API supports pagination. + +## Parameters + +The following query parameters are available: + +* `from` - Offset in the returned list. Defaults to `0`. +* `limit` - Maximum amount of rooms to return. Defaults to `100`. +* `order_by` - The method in which to sort the returned list of rooms. Valid values are: + - `alphabetical` - Rooms are ordered alphabetically by room name. This is the default. + - `size` - Rooms are ordered by the number of members. Largest to smallest. +* `dir` - Direction of room order. Either `f` for forwards or `b` for backwards. Setting + this value to `b` will reverse the above sort order. Defaults to `f`. +* `search_term` - Filter rooms by their room name. Search term can be contained in any + part of the room name. Defaults to no filtering. + +The following fields are possible in the JSON response body: + +* `rooms` - An array of objects, each containing information about a room. + - Room objects contain the following fields: + - `room_id` - The ID of the room. + - `name` - The name of the room. + - `canonical_alias` - The canonical (main) alias address of the room. + - `joined_members` - How many users are currently in the room. +* `offset` - The current pagination offset in rooms. This parameter should be + used instead of `next_token` for room offset as `next_token` is + not intended to be parsed. +* `total_rooms` - The total number of rooms this query can return. Using this + and `offset`, you have enough information to know the current + progression through the list. +* `next_batch` - If this field is present, we know that there are potentially + more rooms on the server that did not all fit into this response. + We can use `next_batch` to get the "next page" of results. To do + so, simply repeat your request, setting the `from` parameter to + the value of `next_batch`. +* `prev_batch` - If this field is present, it is possible to paginate backwards. + Use `prev_batch` for the `from` value in the next request to + get the "previous page" of results. + +## Usage + +A standard request with no filtering: + +``` +GET /_synapse/admin/rooms + +{} +``` + +Response: + +``` +{ + "rooms": [ + { + "room_id": "!OGEhHVWSdvArJzumhm:matrix.org", + "name": "Matrix HQ", + "canonical_alias": "#matrix:matrix.org", + "joined_members": 8326 + }, + ... (8 hidden items) ... + { + "room_id": "!xYvNcQPhnkrdUmYczI:matrix.org", + "name": "This Week In Matrix (TWIM)", + "canonical_alias": "#twim:matrix.org", + "joined_members": 314 + } + ], + "offset": 0, + "total_rooms": 10 +} +``` + +Filtering by room name: + +``` +GET /_synapse/admin/rooms?search_term=TWIM + +{} +``` + +Response: + +``` +{ + "rooms": [ + { + "room_id": "!xYvNcQPhnkrdUmYczI:matrix.org", + "name": "This Week In Matrix (TWIM)", + "canonical_alias": "#twim:matrix.org", + "joined_members": 314 + } + ], + "offset": 0, + "total_rooms": 1 +} +``` + +Paginating through a list of rooms: + +``` +GET /_synapse/admin/rooms?order_by=size + +{} +``` + +Response: + +``` +{ + "rooms": [ + { + "room_id": "!OGEhHVWSdvArJzumhm:matrix.org", + "name": "Matrix HQ", + "canonical_alias": "#matrix:matrix.org", + "joined_members": 8326 + }, + ... (98 hidden items) ... + { + "room_id": "!xYvNcQPhnkrdUmYczI:matrix.org", + "name": "This Week In Matrix (TWIM)", + "canonical_alias": "#twim:matrix.org", + "joined_members": 314 + } + ], + "offset": 0, + "total_rooms": 150 + "next_token": 100 +} +``` + +The presence of the `next_token` parameter tells us that there are more rooms +than returned in this request, and we need to make another request to get them. +To get the next batch of room results, we repeat our request, setting the `from` +parameter to the value of `next_token`. + +``` +GET /_synapse/admin/rooms?order_by=size&from=100 + +{} +``` + +Response: + +``` +{ + "rooms": [ + { + "room_id": "!mscvqgqpHYjBGDxNym:matrix.org", + "name": "Music Theory", + "canonical_alias": "#musictheory:matrix.org", + "joined_members": 127 + }, + ... (48 hidden items) ... + { + "room_id": "!twcBhHVdZlQWuuxBhN:termina.org.uk", + "name": "weechat-matrix", + "canonical_alias": "#weechat-matrix:termina.org.uk", + "joined_members": 137 + } + ], + "offset": 100, + "prev_batch": 0, + "total_rooms": 150 +} +``` + +Once the `next_token` parameter is no longer present, we know we've reached the +end of the list. diff --git a/synapse/rest/admin/__init__.py b/synapse/rest/admin/__init__.py index 2932fe2123..42cc2b062a 100644 --- a/synapse/rest/admin/__init__.py +++ b/synapse/rest/admin/__init__.py @@ -29,7 +29,7 @@ from synapse.rest.admin._base import ( from synapse.rest.admin.groups import DeleteGroupAdminRestServlet from synapse.rest.admin.media import ListMediaInRoom, register_servlets_for_media_repo from synapse.rest.admin.purge_room_servlet import PurgeRoomServlet -from synapse.rest.admin.rooms import ShutdownRoomRestServlet +from synapse.rest.admin.rooms import ListRoomRestServlet, ShutdownRoomRestServlet from synapse.rest.admin.server_notice_servlet import SendServerNoticeServlet from synapse.rest.admin.users import ( AccountValidityRenewServlet, @@ -188,6 +188,7 @@ def register_servlets(hs, http_server): Register all the admin servlets. """ register_servlets_for_client_rest_resource(hs, http_server) + ListRoomRestServlet(hs).register(http_server) PurgeRoomServlet(hs).register(http_server) SendServerNoticeServlet(hs).register(http_server) VersionServlet(hs).register(http_server) diff --git a/synapse/rest/admin/_base.py b/synapse/rest/admin/_base.py index afd0647205..459482eb6d 100644 --- a/synapse/rest/admin/_base.py +++ b/synapse/rest/admin/_base.py @@ -40,6 +40,21 @@ def historical_admin_path_patterns(path_regex): ) +def admin_patterns(path_regex: str): + """Returns the list of patterns for an admin endpoint + + Args: + path_regex: The regex string to match. This should NOT have a ^ + as this will be prefixed. + + Returns: + A list of regex patterns. + """ + admin_prefix = "^/_synapse/admin/v1" + patterns = [re.compile(admin_prefix + path_regex)] + return patterns + + async def assert_requester_is_admin(auth, request): """Verify that the requester is an admin user diff --git a/synapse/rest/admin/rooms.py b/synapse/rest/admin/rooms.py index f7cc5e9be9..f9b8c0a4f0 100644 --- a/synapse/rest/admin/rooms.py +++ b/synapse/rest/admin/rooms.py @@ -15,15 +15,20 @@ import logging from synapse.api.constants import Membership +from synapse.api.errors import Codes, SynapseError from synapse.http.servlet import ( RestServlet, assert_params_in_dict, + parse_integer, parse_json_object_from_request, + parse_string, ) from synapse.rest.admin._base import ( + admin_patterns, assert_user_is_admin, historical_admin_path_patterns, ) +from synapse.storage.data_stores.main.room import RoomSortOrder from synapse.types import create_requester from synapse.util.async_helpers import maybe_awaitable @@ -155,3 +160,80 @@ class ShutdownRoomRestServlet(RestServlet): "new_room_id": new_room_id, }, ) + + +class ListRoomRestServlet(RestServlet): + """ + List all rooms that are known to the homeserver. Results are returned + in a dictionary containing room information. Supports pagination. + """ + + PATTERNS = admin_patterns("/rooms") + + def __init__(self, hs): + self.store = hs.get_datastore() + self.auth = hs.get_auth() + self.admin_handler = hs.get_handlers().admin_handler + + async def on_GET(self, request): + requester = await self.auth.get_user_by_req(request) + await assert_user_is_admin(self.auth, requester.user) + + # Extract query parameters + start = parse_integer(request, "from", default=0) + limit = parse_integer(request, "limit", default=100) + order_by = parse_string(request, "order_by", default="alphabetical") + if order_by not in ( + RoomSortOrder.ALPHABETICAL.value, + RoomSortOrder.SIZE.value, + ): + raise SynapseError( + 400, + "Unknown value for order_by: %s" % (order_by,), + errcode=Codes.INVALID_PARAM, + ) + + search_term = parse_string(request, "search_term") + if search_term == "": + raise SynapseError( + 400, + "search_term cannot be an empty string", + errcode=Codes.INVALID_PARAM, + ) + + direction = parse_string(request, "dir", default="f") + if direction not in ("f", "b"): + raise SynapseError( + 400, "Unknown direction: %s" % (direction,), errcode=Codes.INVALID_PARAM + ) + + reverse_order = True if direction == "b" else False + + # Return list of rooms according to parameters + rooms, total_rooms = await self.store.get_rooms_paginate( + start, limit, order_by, reverse_order, search_term + ) + response = { + # next_token should be opaque, so return a value the client can parse + "offset": start, + "rooms": rooms, + "total_rooms": total_rooms, + } + + # Are there more rooms to paginate through after this? + if (start + limit) < total_rooms: + # There are. Calculate where the query should start from next time + # to get the next part of the list + response["next_batch"] = start + limit + + # Is it possible to paginate backwards? Check if we currently have an + # offset + if start > 0: + if start > limit: + # Going back one iteration won't take us to the start. + # Calculate new offset + response["prev_batch"] = start - limit + else: + response["prev_batch"] = 0 + + return 200, response diff --git a/synapse/rest/client/v2_alpha/_base.py b/synapse/rest/client/v2_alpha/_base.py index 2a3f4dd58f..bc11b4dda4 100644 --- a/synapse/rest/client/v2_alpha/_base.py +++ b/synapse/rest/client/v2_alpha/_base.py @@ -32,7 +32,7 @@ def client_patterns(path_regex, releases=(0,), unstable=True, v1=False): Args: path_regex (str): The regex string to match. This should NOT have a ^ - as this will be prefixed. + as this will be prefixed. Returns: SRE_Pattern """ diff --git a/synapse/storage/data_stores/main/room.py b/synapse/storage/data_stores/main/room.py index 49bab62be3..d968803ad2 100644 --- a/synapse/storage/data_stores/main/room.py +++ b/synapse/storage/data_stores/main/room.py @@ -18,7 +18,8 @@ import collections import logging import re from abc import abstractmethod -from typing import List, Optional, Tuple +from enum import Enum +from typing import Any, Dict, List, Optional, Tuple from six import integer_types @@ -46,6 +47,18 @@ RatelimitOverride = collections.namedtuple( ) +class RoomSortOrder(Enum): + """ + Enum to define the sorting method used when returning rooms with get_rooms_paginate + + ALPHABETICAL = sort rooms alphabetically by name + SIZE = sort rooms by membership size, highest to lowest + """ + + ALPHABETICAL = "alphabetical" + SIZE = "size" + + class RoomWorkerStore(SQLBaseStore): def __init__(self, database: Database, db_conn, hs): super(RoomWorkerStore, self).__init__(database, db_conn, hs) @@ -281,6 +294,116 @@ class RoomWorkerStore(SQLBaseStore): desc="is_room_blocked", ) + async def get_rooms_paginate( + self, + start: int, + limit: int, + order_by: RoomSortOrder, + reverse_order: bool, + search_term: Optional[str], + ) -> Tuple[List[Dict[str, Any]], int]: + """Function to retrieve a paginated list of rooms as json. + + Args: + start: offset in the list + limit: maximum amount of rooms to retrieve + order_by: the sort order of the returned list + reverse_order: whether to reverse the room list + search_term: a string to filter room names by + Returns: + A list of room dicts and an integer representing the total number of + rooms that exist given this query + """ + # Filter room names by a string + where_statement = "" + if search_term: + where_statement = "WHERE state.name LIKE ?" + + # Our postgres db driver converts ? -> %s in SQL strings as that's the + # placeholder for postgres. + # HOWEVER, if you put a % into your SQL then everything goes wibbly. + # To get around this, we're going to surround search_term with %'s + # before giving it to the database in python instead + search_term = "%" + search_term + "%" + + # Set ordering + if RoomSortOrder(order_by) == RoomSortOrder.SIZE: + order_by_column = "curr.joined_members" + order_by_asc = False + elif RoomSortOrder(order_by) == RoomSortOrder.ALPHABETICAL: + # Sort alphabetically + order_by_column = "state.name" + order_by_asc = True + else: + raise StoreError( + 500, "Incorrect value for order_by provided: %s" % order_by + ) + + # Whether to return the list in reverse order + if reverse_order: + # Flip the boolean + order_by_asc = not order_by_asc + + # Create one query for getting the limited number of events that the user asked + # for, and another query for getting the total number of events that could be + # returned. Thus allowing us to see if there are more events to paginate through + info_sql = """ + SELECT state.room_id, state.name, state.canonical_alias, curr.joined_members + FROM room_stats_state state + INNER JOIN room_stats_current curr USING (room_id) + %s + ORDER BY %s %s + LIMIT ? + OFFSET ? + """ % ( + where_statement, + order_by_column, + "ASC" if order_by_asc else "DESC", + ) + + # Use a nested SELECT statement as SQL can't count(*) with an OFFSET + count_sql = """ + SELECT count(*) FROM ( + SELECT room_id FROM room_stats_state state + %s + ) AS get_room_ids + """ % ( + where_statement, + ) + + def _get_rooms_paginate_txn(txn): + # Execute the data query + sql_values = (limit, start) + if search_term: + # Add the search term into the WHERE clause + sql_values = (search_term,) + sql_values + txn.execute(info_sql, sql_values) + + # Refactor room query data into a structured dictionary + rooms = [] + for room in txn: + rooms.append( + { + "room_id": room[0], + "name": room[1], + "canonical_alias": room[2], + "joined_members": room[3], + } + ) + + # Execute the count query + + # Add the search term into the WHERE clause if present + sql_values = (search_term,) if search_term else () + txn.execute(count_sql, sql_values) + + room_count = txn.fetchone() + return rooms, room_count[0] + + return await self.db.runInteraction( + "get_rooms_paginate", _get_rooms_paginate_txn, + ) + @cachedInlineCallbacks(max_entries=10000) def get_ratelimit_for_user(self, user_id): """Check if there are any overrides for ratelimiting for the given diff --git a/tests/rest/admin/test_admin.py b/tests/rest/admin/test_admin.py index af4d604e50..0342aed416 100644 --- a/tests/rest/admin/test_admin.py +++ b/tests/rest/admin/test_admin.py @@ -17,6 +17,7 @@ import json import os import urllib.parse from binascii import unhexlify +from typing import List, Optional from mock import Mock @@ -26,7 +27,7 @@ import synapse.rest.admin from synapse.http.server import JsonResource from synapse.logging.context import make_deferred_yieldable from synapse.rest.admin import VersionServlet -from synapse.rest.client.v1 import events, login, room +from synapse.rest.client.v1 import directory, events, login, room from synapse.rest.client.v2_alpha import groups from tests import unittest @@ -468,9 +469,7 @@ class QuarantineMediaTestCase(unittest.HomeserverTestCase): ) # Extract media ID from the response - server_name_and_media_id = response["content_uri"][ - 6: - ] # Cut off the 'mxc://' bit + server_name_and_media_id = response["content_uri"][6:] # Cut off 'mxc://' server_name, media_id = server_name_and_media_id.split("/") # Attempt to access the media @@ -692,3 +691,389 @@ class QuarantineMediaTestCase(unittest.HomeserverTestCase): % server_and_media_id_2 ), ) + + +class RoomTestCase(unittest.HomeserverTestCase): + """Test /room admin API. + """ + + servlets = [ + synapse.rest.admin.register_servlets, + login.register_servlets, + room.register_servlets, + directory.register_servlets, + ] + + def prepare(self, reactor, clock, hs): + self.store = hs.get_datastore() + + # Create user + self.admin_user = self.register_user("admin", "pass", admin=True) + self.admin_user_tok = self.login("admin", "pass") + + def test_list_rooms(self): + """Test that we can list rooms""" + # Create 3 test rooms + total_rooms = 3 + room_ids = [] + for x in range(total_rooms): + room_id = self.helper.create_room_as( + self.admin_user, tok=self.admin_user_tok + ) + room_ids.append(room_id) + + # Request the list of rooms + url = "/_synapse/admin/v1/rooms" + request, channel = self.make_request( + "GET", url.encode("ascii"), access_token=self.admin_user_tok, + ) + self.render(request) + + # Check request completed successfully + self.assertEqual(200, int(channel.code), msg=channel.json_body) + + # Check that response json body contains a "rooms" key + self.assertTrue( + "rooms" in channel.json_body, + msg="Response body does not " "contain a 'rooms' key", + ) + + # Check that 3 rooms were returned + self.assertEqual(3, len(channel.json_body["rooms"]), msg=channel.json_body) + + # Check their room_ids match + returned_room_ids = [room["room_id"] for room in channel.json_body["rooms"]] + self.assertEqual(room_ids, returned_room_ids) + + # Check that all fields are available + for r in channel.json_body["rooms"]: + self.assertIn("name", r) + self.assertIn("canonical_alias", r) + self.assertIn("joined_members", r) + + # Check that the correct number of total rooms was returned + self.assertEqual(channel.json_body["total_rooms"], total_rooms) + + # Check that the offset is correct + # Should be 0 as we aren't paginating + self.assertEqual(channel.json_body["offset"], 0) + + # Check that the prev_batch parameter is not present + self.assertNotIn("prev_batch", channel.json_body) + + # We shouldn't receive a next token here as there's no further rooms to show + self.assertNotIn("next_batch", channel.json_body) + + def test_list_rooms_pagination(self): + """Test that we can get a full list of rooms through pagination""" + # Create 5 test rooms + total_rooms = 5 + room_ids = [] + for x in range(total_rooms): + room_id = self.helper.create_room_as( + self.admin_user, tok=self.admin_user_tok + ) + room_ids.append(room_id) + + # Set the name of the rooms so we get a consistent returned ordering + for idx, room_id in enumerate(room_ids): + self.helper.send_state( + room_id, "m.room.name", {"name": str(idx)}, tok=self.admin_user_tok, + ) + + # Request the list of rooms + returned_room_ids = [] + start = 0 + limit = 2 + + run_count = 0 + should_repeat = True + while should_repeat: + run_count += 1 + + url = "/_synapse/admin/v1/rooms?from=%d&limit=%d&order_by=%s" % ( + start, + limit, + "alphabetical", + ) + request, channel = self.make_request( + "GET", url.encode("ascii"), access_token=self.admin_user_tok, + ) + self.render(request) + self.assertEqual( + 200, int(channel.result["code"]), msg=channel.result["body"] + ) + + self.assertTrue("rooms" in channel.json_body) + for r in channel.json_body["rooms"]: + returned_room_ids.append(r["room_id"]) + + # Check that the correct number of total rooms was returned + self.assertEqual(channel.json_body["total_rooms"], total_rooms) + + # Check that the offset is correct + # We're only getting 2 rooms each page, so should be 2 * last run_count + self.assertEqual(channel.json_body["offset"], 2 * (run_count - 1)) + + if run_count > 1: + # Check the value of prev_batch is correct + self.assertEqual(channel.json_body["prev_batch"], 2 * (run_count - 2)) + + if "next_batch" not in channel.json_body: + # We have reached the end of the list + should_repeat = False + else: + # Make another query with an updated start value + start = channel.json_body["next_batch"] + + # We should've queried the endpoint 3 times + self.assertEqual( + run_count, + 3, + msg="Should've queried 3 times for 5 rooms with limit 2 per query", + ) + + # Check that we received all of the room ids + self.assertEqual(room_ids, returned_room_ids) + + url = "/_synapse/admin/v1/rooms?from=%d&limit=%d" % (start, limit) + request, channel = self.make_request( + "GET", url.encode("ascii"), access_token=self.admin_user_tok, + ) + self.render(request) + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + + def test_correct_room_attributes(self): + """Test the correct attributes for a room are returned""" + # Create a test room + room_id = self.helper.create_room_as(self.admin_user, tok=self.admin_user_tok) + + test_alias = "#test:test" + test_room_name = "something" + + # Have another user join the room + user_2 = self.register_user("user4", "pass") + user_tok_2 = self.login("user4", "pass") + self.helper.join(room_id, user_2, tok=user_tok_2) + + # Create a new alias to this room + url = "/_matrix/client/r0/directory/room/%s" % (urllib.parse.quote(test_alias),) + request, channel = self.make_request( + "PUT", + url.encode("ascii"), + {"room_id": room_id}, + access_token=self.admin_user_tok, + ) + self.render(request) + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + + # Set this new alias as the canonical alias for this room + self.helper.send_state( + room_id, + "m.room.aliases", + {"aliases": [test_alias]}, + tok=self.admin_user_tok, + state_key="test", + ) + self.helper.send_state( + room_id, + "m.room.canonical_alias", + {"alias": test_alias}, + tok=self.admin_user_tok, + ) + + # Set a name for the room + self.helper.send_state( + room_id, "m.room.name", {"name": test_room_name}, tok=self.admin_user_tok, + ) + + # Request the list of rooms + url = "/_synapse/admin/v1/rooms" + request, channel = self.make_request( + "GET", url.encode("ascii"), access_token=self.admin_user_tok, + ) + self.render(request) + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + + # Check that rooms were returned + self.assertTrue("rooms" in channel.json_body) + rooms = channel.json_body["rooms"] + + # Check that only one room was returned + self.assertEqual(len(rooms), 1) + + # And that the value of the total_rooms key was correct + self.assertEqual(channel.json_body["total_rooms"], 1) + + # Check that the offset is correct + # We're not paginating, so should be 0 + self.assertEqual(channel.json_body["offset"], 0) + + # Check that there is no `prev_batch` + self.assertNotIn("prev_batch", channel.json_body) + + # Check that there is no `next_batch` + self.assertNotIn("next_batch", channel.json_body) + + # Check that all provided attributes are set + r = rooms[0] + self.assertEqual(room_id, r["room_id"]) + self.assertEqual(test_room_name, r["name"]) + self.assertEqual(test_alias, r["canonical_alias"]) + + def test_room_list_sort_order(self): + """Test room list sort ordering. alphabetical versus number of members, + reversing the order, etc. + """ + # Create 3 test rooms + room_id_1 = self.helper.create_room_as(self.admin_user, tok=self.admin_user_tok) + room_id_2 = self.helper.create_room_as(self.admin_user, tok=self.admin_user_tok) + room_id_3 = self.helper.create_room_as(self.admin_user, tok=self.admin_user_tok) + + # Set room names in alphabetical order. room 1 -> A, 2 -> B, 3 -> C + self.helper.send_state( + room_id_1, "m.room.name", {"name": "A"}, tok=self.admin_user_tok, + ) + self.helper.send_state( + room_id_2, "m.room.name", {"name": "B"}, tok=self.admin_user_tok, + ) + self.helper.send_state( + room_id_3, "m.room.name", {"name": "C"}, tok=self.admin_user_tok, + ) + + # Set room member size in the reverse order. room 1 -> 1 member, 2 -> 2, 3 -> 3 + user_1 = self.register_user("bob1", "pass") + user_1_tok = self.login("bob1", "pass") + self.helper.join(room_id_2, user_1, tok=user_1_tok) + + user_2 = self.register_user("bob2", "pass") + user_2_tok = self.login("bob2", "pass") + self.helper.join(room_id_3, user_2, tok=user_2_tok) + + user_3 = self.register_user("bob3", "pass") + user_3_tok = self.login("bob3", "pass") + self.helper.join(room_id_3, user_3, tok=user_3_tok) + + def _order_test( + order_type: str, expected_room_list: List[str], reverse: bool = False, + ): + """Request the list of rooms in a certain order. Assert that order is what + we expect + + Args: + order_type: The type of ordering to give the server + expected_room_list: The list of room_ids in the order we expect to get + back from the server + """ + # Request the list of rooms in the given order + url = "/_synapse/admin/v1/rooms?order_by=%s" % (order_type,) + if reverse: + url += "&dir=b" + request, channel = self.make_request( + "GET", url.encode("ascii"), access_token=self.admin_user_tok, + ) + self.render(request) + self.assertEqual(200, channel.code, msg=channel.json_body) + + # Check that rooms were returned + self.assertTrue("rooms" in channel.json_body) + rooms = channel.json_body["rooms"] + + # Check for the correct total_rooms value + self.assertEqual(channel.json_body["total_rooms"], 3) + + # Check that the offset is correct + # We're not paginating, so should be 0 + self.assertEqual(channel.json_body["offset"], 0) + + # Check that there is no `prev_batch` + self.assertNotIn("prev_batch", channel.json_body) + + # Check that there is no `next_batch` + self.assertNotIn("next_batch", channel.json_body) + + # Check that rooms were returned in alphabetical order + returned_order = [r["room_id"] for r in rooms] + self.assertListEqual(expected_room_list, returned_order) # order is checked + + # Test different sort orders, with forward and reverse directions + _order_test("alphabetical", [room_id_1, room_id_2, room_id_3]) + _order_test("alphabetical", [room_id_3, room_id_2, room_id_1], reverse=True) + + _order_test("size", [room_id_3, room_id_2, room_id_1]) + _order_test("size", [room_id_1, room_id_2, room_id_3], reverse=True) + + def test_search_term(self): + """Test that searching for a room works correctly""" + # Create two test rooms + room_id_1 = self.helper.create_room_as(self.admin_user, tok=self.admin_user_tok) + room_id_2 = self.helper.create_room_as(self.admin_user, tok=self.admin_user_tok) + + room_name_1 = "something" + room_name_2 = "else" + + # Set the name for each room + self.helper.send_state( + room_id_1, "m.room.name", {"name": room_name_1}, tok=self.admin_user_tok, + ) + self.helper.send_state( + room_id_2, "m.room.name", {"name": room_name_2}, tok=self.admin_user_tok, + ) + + def _search_test( + expected_room_id: Optional[str], + search_term: str, + expected_http_code: int = 200, + ): + """Search for a room and check that the returned room's id is a match + + Args: + expected_room_id: The room_id expected to be returned by the API. Set + to None to expect zero results for the search + search_term: The term to search for room names with + expected_http_code: The expected http code for the request + """ + url = "/_synapse/admin/v1/rooms?search_term=%s" % (search_term,) + request, channel = self.make_request( + "GET", url.encode("ascii"), access_token=self.admin_user_tok, + ) + self.render(request) + self.assertEqual(expected_http_code, channel.code, msg=channel.json_body) + + if expected_http_code != 200: + return + + # Check that rooms were returned + self.assertTrue("rooms" in channel.json_body) + rooms = channel.json_body["rooms"] + + # Check that the expected number of rooms were returned + expected_room_count = 1 if expected_room_id else 0 + self.assertEqual(len(rooms), expected_room_count) + self.assertEqual(channel.json_body["total_rooms"], expected_room_count) + + # Check that the offset is correct + # We're not paginating, so should be 0 + self.assertEqual(channel.json_body["offset"], 0) + + # Check that there is no `prev_batch` + self.assertNotIn("prev_batch", channel.json_body) + + # Check that there is no `next_batch` + self.assertNotIn("next_batch", channel.json_body) + + if expected_room_id: + # Check that the first returned room id is correct + r = rooms[0] + self.assertEqual(expected_room_id, r["room_id"]) + + # Perform search tests + _search_test(room_id_1, "something") + _search_test(room_id_1, "thing") + + _search_test(room_id_2, "else") + _search_test(room_id_2, "se") + + _search_test(None, "foo") + _search_test(None, "bar") + _search_test(None, "", expected_http_code=400) -- cgit 1.4.1 From fa4d609e20318821e2ffbeb35bfddbc86be81be0 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Thu, 23 Jan 2020 15:19:03 +0000 Subject: Make 'event.redacts' never raise. (#6771) There are quite a few places that we assume that a redaction event has a corresponding `redacts` key, which is not always the case. So lets cheekily make it so that event.redacts just returns None instead. --- changelog.d/6771.bugfix | 1 + synapse/events/__init__.py | 28 +++++++++++++++--- synapse/storage/data_stores/main/events.py | 2 +- synapse/storage/data_stores/main/events_worker.py | 2 +- tests/storage/test_redaction.py | 35 +++++++++++++++++++++++ 5 files changed, 62 insertions(+), 6 deletions(-) create mode 100644 changelog.d/6771.bugfix (limited to 'tests') diff --git a/changelog.d/6771.bugfix b/changelog.d/6771.bugfix new file mode 100644 index 0000000000..623ba24acb --- /dev/null +++ b/changelog.d/6771.bugfix @@ -0,0 +1 @@ +Fix persisting redaction events that have been redacted (or otherwise don't have a redacts key). diff --git a/synapse/events/__init__.py b/synapse/events/__init__.py index 88ed6d764f..72c09327f4 100644 --- a/synapse/events/__init__.py +++ b/synapse/events/__init__.py @@ -116,16 +116,32 @@ class _EventInternalMetadata(object): return getattr(self, "redacted", False) -def _event_dict_property(key): +_SENTINEL = object() + + +def _event_dict_property(key, default=_SENTINEL): + """Creates a new property for the given key that delegates access to + `self._event_dict`. + + The default is used if the key is missing from the `_event_dict`, if given, + otherwise an AttributeError will be raised. + + Note: If a default is given then `hasattr` will always return true. + """ + # We want to be able to use hasattr with the event dict properties. # However, (on python3) hasattr expects AttributeError to be raised. Hence, # we need to transform the KeyError into an AttributeError - def getter(self): + + def getter_raises(self): try: return self._event_dict[key] except KeyError: raise AttributeError(key) + def getter_default(self): + return self._event_dict.get(key, default) + def setter(self, v): try: self._event_dict[key] = v @@ -138,7 +154,11 @@ def _event_dict_property(key): except KeyError: raise AttributeError(key) - return property(getter, setter, delete) + if default is _SENTINEL: + # No default given, so use the getter that raises + return property(getter_raises, setter, delete) + else: + return property(getter_default, setter, delete) class EventBase(object): @@ -165,7 +185,7 @@ class EventBase(object): origin = _event_dict_property("origin") origin_server_ts = _event_dict_property("origin_server_ts") prev_events = _event_dict_property("prev_events") - redacts = _event_dict_property("redacts") + redacts = _event_dict_property("redacts", None) room_id = _event_dict_property("room_id") sender = _event_dict_property("sender") user_id = _event_dict_property("sender") diff --git a/synapse/storage/data_stores/main/events.py b/synapse/storage/data_stores/main/events.py index 596daf8909..ce553566a5 100644 --- a/synapse/storage/data_stores/main/events.py +++ b/synapse/storage/data_stores/main/events.py @@ -951,7 +951,7 @@ class EventsStore( elif event.type == EventTypes.Message: # Insert into the event_search table. self._store_room_message_txn(txn, event) - elif event.type == EventTypes.Redaction: + elif event.type == EventTypes.Redaction and event.redacts is not None: # Insert into the redactions table. self._store_redaction(txn, event) elif event.type == EventTypes.Retention: diff --git a/synapse/storage/data_stores/main/events_worker.py b/synapse/storage/data_stores/main/events_worker.py index 3b93e0597a..7251e819f5 100644 --- a/synapse/storage/data_stores/main/events_worker.py +++ b/synapse/storage/data_stores/main/events_worker.py @@ -287,7 +287,7 @@ class EventsWorkerStore(SQLBaseStore): # we have to recheck auth now. if not allow_rejected and entry.event.type == EventTypes.Redaction: - if not hasattr(entry.event, "redacts"): + if entry.event.redacts is None: # A redacted redaction doesn't have a `redacts` key, in # which case lets just withhold the event. # diff --git a/tests/storage/test_redaction.py b/tests/storage/test_redaction.py index dc45173355..feb1c07cb2 100644 --- a/tests/storage/test_redaction.py +++ b/tests/storage/test_redaction.py @@ -398,3 +398,38 @@ class RedactionTestCase(unittest.HomeserverTestCase): self.get_success( self.store.get_event(first_redact_event.event_id, allow_none=True) ) + + def test_store_redacted_redaction(self): + """Tests that we can store a redacted redaction. + """ + + self.get_success( + self.inject_room_member(self.room1, self.u_alice, Membership.JOIN) + ) + + builder = self.event_builder_factory.for_room_version( + RoomVersions.V1, + { + "type": EventTypes.Redaction, + "sender": self.u_alice.to_string(), + "room_id": self.room1.to_string(), + "content": {"reason": "foo"}, + }, + ) + + redaction_event, context = self.get_success( + self.event_creation_handler.create_new_client_event(builder) + ) + + self.get_success( + self.storage.persistence.persist_event(redaction_event, context) + ) + + # Now lets jump to the future where we have censored the redaction event + # in the DB. + self.reactor.advance(60 * 60 * 24 * 31) + + # We just want to check that fetching the event doesn't raise an exception. + self.get_success( + self.store.get_event(redaction_event.event_id, allow_none=True) + ) -- cgit 1.4.1 From 9f7aaf90b5ef76416852f35201a851d45eccc0a1 Mon Sep 17 00:00:00 2001 From: Andrew Morgan <1342360+anoadragon453@users.noreply.github.com> Date: Fri, 24 Jan 2020 14:28:40 +0000 Subject: Validate client_secret parameter (#6767) --- changelog.d/6767.bugfix | 1 + synapse/handlers/identity.py | 4 ++- synapse/rest/client/v2_alpha/account.py | 23 ++++++++++---- synapse/rest/client/v2_alpha/register.py | 3 ++ synapse/util/stringutils.py | 17 +++++++++++ tests/util/test_stringutils.py | 51 ++++++++++++++++++++++++++++++++ 6 files changed, 93 insertions(+), 6 deletions(-) create mode 100644 changelog.d/6767.bugfix create mode 100644 tests/util/test_stringutils.py (limited to 'tests') diff --git a/changelog.d/6767.bugfix b/changelog.d/6767.bugfix new file mode 100644 index 0000000000..63c7c63315 --- /dev/null +++ b/changelog.d/6767.bugfix @@ -0,0 +1 @@ +Validate `client_secret` parameter using the regex provided by the Client-Server API, temporarily allowing `:` characters for older clients. The `:` character will be removed in a future release. diff --git a/synapse/handlers/identity.py b/synapse/handlers/identity.py index 000fbf090f..23f07832e7 100644 --- a/synapse/handlers/identity.py +++ b/synapse/handlers/identity.py @@ -38,7 +38,7 @@ from synapse.api.errors import ( from synapse.config.emailconfig import ThreepidBehaviour from synapse.http.client import SimpleHttpClient from synapse.util.hash import sha256_and_url_safe_base64 -from synapse.util.stringutils import random_string +from synapse.util.stringutils import assert_valid_client_secret, random_string from ._base import BaseHandler @@ -84,6 +84,8 @@ class IdentityHandler(BaseHandler): raise SynapseError( 400, "Missing param client_secret in creds", errcode=Codes.MISSING_PARAM ) + assert_valid_client_secret(client_secret) + session_id = creds.get("sid") if not session_id: raise SynapseError( diff --git a/synapse/rest/client/v2_alpha/account.py b/synapse/rest/client/v2_alpha/account.py index fc240f5cf8..dc837d6c75 100644 --- a/synapse/rest/client/v2_alpha/account.py +++ b/synapse/rest/client/v2_alpha/account.py @@ -30,6 +30,7 @@ from synapse.http.servlet import ( ) from synapse.push.mailer import Mailer, load_jinja2_templates from synapse.util.msisdn import phone_number_to_msisdn +from synapse.util.stringutils import assert_valid_client_secret from synapse.util.threepids import check_3pid_allowed from ._base import client_patterns, interactive_auth_handler @@ -81,6 +82,8 @@ class EmailPasswordRequestTokenRestServlet(RestServlet): # Extract params from body client_secret = body["client_secret"] + assert_valid_client_secret(client_secret) + email = body["email"] send_attempt = body["send_attempt"] next_link = body.get("next_link") # Optional param @@ -166,8 +169,9 @@ class PasswordResetSubmitTokenServlet(RestServlet): ) sid = parse_string(request, "sid", required=True) - client_secret = parse_string(request, "client_secret", required=True) token = parse_string(request, "token", required=True) + client_secret = parse_string(request, "client_secret", required=True) + assert_valid_client_secret(client_secret) # Attempt to validate a 3PID session try: @@ -353,6 +357,8 @@ class EmailThreepidRequestTokenRestServlet(RestServlet): body = parse_json_object_from_request(request) assert_params_in_dict(body, ["client_secret", "email", "send_attempt"]) client_secret = body["client_secret"] + assert_valid_client_secret(client_secret) + email = body["email"] send_attempt = body["send_attempt"] next_link = body.get("next_link") # Optional param @@ -413,6 +419,8 @@ class MsisdnThreepidRequestTokenRestServlet(RestServlet): body, ["client_secret", "country", "phone_number", "send_attempt"] ) client_secret = body["client_secret"] + assert_valid_client_secret(client_secret) + country = body["country"] phone_number = body["phone_number"] send_attempt = body["send_attempt"] @@ -493,8 +501,9 @@ class AddThreepidEmailSubmitTokenServlet(RestServlet): ) sid = parse_string(request, "sid", required=True) - client_secret = parse_string(request, "client_secret", required=True) token = parse_string(request, "token", required=True) + client_secret = parse_string(request, "client_secret", required=True) + assert_valid_client_secret(client_secret) # Attempt to validate a 3PID session try: @@ -559,6 +568,7 @@ class AddThreepidMsisdnSubmitTokenServlet(RestServlet): body = parse_json_object_from_request(request) assert_params_in_dict(body, ["client_secret", "sid", "token"]) + assert_valid_client_secret(body["client_secret"]) # Proxy submit_token request to msisdn threepid delegate response = await self.identity_handler.proxy_msisdn_submit_token( @@ -600,8 +610,9 @@ class ThreepidRestServlet(RestServlet): ) assert_params_in_dict(threepid_creds, ["client_secret", "sid"]) - client_secret = threepid_creds["client_secret"] sid = threepid_creds["sid"] + client_secret = threepid_creds["client_secret"] + assert_valid_client_secret(client_secret) validation_session = await self.identity_handler.validate_threepid_session( client_secret, sid @@ -637,8 +648,9 @@ class ThreepidAddRestServlet(RestServlet): body = parse_json_object_from_request(request) assert_params_in_dict(body, ["client_secret", "sid"]) - client_secret = body["client_secret"] sid = body["sid"] + client_secret = body["client_secret"] + assert_valid_client_secret(client_secret) await self.auth_handler.validate_user_via_ui_auth( requester, body, self.hs.get_ip_from_request(request) @@ -676,8 +688,9 @@ class ThreepidBindRestServlet(RestServlet): assert_params_in_dict(body, ["id_server", "sid", "client_secret"]) id_server = body["id_server"] sid = body["sid"] - client_secret = body["client_secret"] id_access_token = body.get("id_access_token") # optional + client_secret = body["client_secret"] + assert_valid_client_secret(client_secret) requester = await self.auth.get_user_by_req(request) user_id = requester.user.to_string() diff --git a/synapse/rest/client/v2_alpha/register.py b/synapse/rest/client/v2_alpha/register.py index 1bda9aec7e..a09189b1b4 100644 --- a/synapse/rest/client/v2_alpha/register.py +++ b/synapse/rest/client/v2_alpha/register.py @@ -49,6 +49,7 @@ from synapse.http.servlet import ( from synapse.push.mailer import load_jinja2_templates from synapse.util.msisdn import phone_number_to_msisdn from synapse.util.ratelimitutils import FederationRateLimiter +from synapse.util.stringutils import assert_valid_client_secret from synapse.util.threepids import check_3pid_allowed from ._base import client_patterns, interactive_auth_handler @@ -116,6 +117,8 @@ class EmailRegisterRequestTokenRestServlet(RestServlet): # Extract params from body client_secret = body["client_secret"] + assert_valid_client_secret(client_secret) + email = body["email"] send_attempt = body["send_attempt"] next_link = body.get("next_link") # Optional param diff --git a/synapse/util/stringutils.py b/synapse/util/stringutils.py index 982c6d81ca..2c0dcb5208 100644 --- a/synapse/util/stringutils.py +++ b/synapse/util/stringutils.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2020 The Matrix.org Foundation C.I.C. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -14,14 +15,22 @@ # limitations under the License. import random +import re import string import six from six import PY2, PY3 from six.moves import range +from synapse.api.errors import Codes, SynapseError + _string_with_symbols = string.digits + string.ascii_letters + ".,;:^&*-_+=#~@" +# https://matrix.org/docs/spec/client_server/r0.6.0#post-matrix-client-r0-register-email-requesttoken +# Note: The : character is allowed here for older clients, but will be removed in a +# future release. Context: https://github.com/matrix-org/synapse/issues/6766 +client_secret_regex = re.compile(r"^[0-9a-zA-Z\.\=\_\-\:]+$") + # random_string and random_string_with_symbols are used for a range of things, # some cryptographically important, some less so. We use SystemRandom to make sure # we get cryptographically-secure randoms. @@ -109,3 +118,11 @@ def exception_to_unicode(e): return msg.decode("utf-8", errors="replace") else: return msg + + +def assert_valid_client_secret(client_secret): + """Validate that a given string matches the client_secret regex defined by the spec""" + if client_secret_regex.match(client_secret) is None: + raise SynapseError( + 400, "Invalid client_secret parameter", errcode=Codes.INVALID_PARAM + ) diff --git a/tests/util/test_stringutils.py b/tests/util/test_stringutils.py new file mode 100644 index 0000000000..4f4da29a98 --- /dev/null +++ b/tests/util/test_stringutils.py @@ -0,0 +1,51 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from synapse.api.errors import SynapseError +from synapse.util.stringutils import assert_valid_client_secret + +from .. import unittest + + +class StringUtilsTestCase(unittest.TestCase): + def test_client_secret_regex(self): + """Ensure that client_secret does not contain illegal characters""" + good = [ + "abcde12345", + "ABCabc123", + "_--something==_", + "...--==-18913", + "8Dj2odd-e9asd.cd==_--ddas-secret-", + # We temporarily allow : characters: https://github.com/matrix-org/synapse/issues/6766 + # To be removed in a future release + "SECRET:1234567890", + ] + + bad = [ + "--+-/secret", + "\\dx--dsa288", + "", + "AAS//", + "asdj**", + ">X> Date: Mon, 27 Jan 2020 14:30:57 +0000 Subject: Add `rooms.room_version` column (#6729) This is so that we don't have to rely on pulling it out from `current_state_events` table. --- changelog.d/6729.misc | 1 + synapse/federation/federation_client.py | 50 ++++++++---- synapse/handlers/federation.py | 65 +++++++++++---- synapse/handlers/room.py | 52 +++++++----- .../client/v2_alpha/room_upgrade_rest_servlet.py | 3 +- synapse/storage/data_stores/main/room.py | 94 ++++++++++++++++++++-- .../main/schema/delta/57/rooms_version_column.sql | 24 ++++++ synapse/storage/data_stores/main/state.py | 34 +++++--- tests/storage/test_room.py | 7 +- tests/storage/test_state.py | 5 +- tests/utils.py | 8 ++ 11 files changed, 270 insertions(+), 73 deletions(-) create mode 100644 changelog.d/6729.misc create mode 100644 synapse/storage/data_stores/main/schema/delta/57/rooms_version_column.sql (limited to 'tests') diff --git a/changelog.d/6729.misc b/changelog.d/6729.misc new file mode 100644 index 0000000000..5537355bea --- /dev/null +++ b/changelog.d/6729.misc @@ -0,0 +1 @@ +Record room versions in the `rooms` table. diff --git a/synapse/federation/federation_client.py b/synapse/federation/federation_client.py index af652a7659..d57e8ca7a2 100644 --- a/synapse/federation/federation_client.py +++ b/synapse/federation/federation_client.py @@ -17,6 +17,7 @@ import copy import itertools import logging +from typing import Dict, Iterable from prometheus_client import Counter @@ -29,6 +30,7 @@ from synapse.api.errors import ( FederationDeniedError, HttpResponseException, SynapseError, + UnsupportedRoomVersionError, ) from synapse.api.room_versions import ( KNOWN_ROOM_VERSIONS, @@ -385,6 +387,8 @@ class FederationClient(FederationBase): return res except InvalidResponseError as e: logger.warning("Failed to %s via %s: %s", description, destination, e) + except UnsupportedRoomVersionError: + raise except HttpResponseException as e: if not 500 <= e.code < 600: raise e.to_synapse_error() @@ -404,7 +408,13 @@ class FederationClient(FederationBase): raise SynapseError(502, "Failed to %s via any server" % (description,)) def make_membership_event( - self, destinations, room_id, user_id, membership, content, params + self, + destinations: Iterable[str], + room_id: str, + user_id: str, + membership: str, + content: dict, + params: Dict[str, str], ): """ Creates an m.room.member event, with context, without participating in the room. @@ -417,21 +427,23 @@ class FederationClient(FederationBase): Note that this does not append any events to any graphs. Args: - destinations (Iterable[str]): Candidate homeservers which are probably + destinations: Candidate homeservers which are probably participating in the room. - room_id (str): The room in which the event will happen. - user_id (str): The user whose membership is being evented. - membership (str): The "membership" property of the event. Must be - one of "join" or "leave". - content (dict): Any additional data to put into the content field - of the event. - params (dict[str, str|Iterable[str]]): Query parameters to include in the - request. + room_id: The room in which the event will happen. + user_id: The user whose membership is being evented. + membership: The "membership" property of the event. Must be one of + "join" or "leave". + content: Any additional data to put into the content field of the + event. + params: Query parameters to include in the request. Return: - Deferred[tuple[str, FrozenEvent, int]]: resolves to a tuple of - `(origin, event, event_format)` where origin is the remote - homeserver which generated the event, and event_format is one of - `synapse.api.room_versions.EventFormatVersions`. + Deferred[Tuple[str, FrozenEvent, RoomVersion]]: resolves to a tuple of + `(origin, event, room_version)` where origin is the remote + homeserver which generated the event, and room_version is the + version of the room. + + Fails with a `UnsupportedRoomVersionError` if remote responds with + a room version we don't understand. Fails with a ``SynapseError`` if the chosen remote server returns a 300/400 code. @@ -453,8 +465,12 @@ class FederationClient(FederationBase): # Note: If not supplied, the room version may be either v1 or v2, # however either way the event format version will be v1. - room_version = ret.get("room_version", RoomVersions.V1.identifier) - event_format = room_version_to_event_format(room_version) + room_version_id = ret.get("room_version", RoomVersions.V1.identifier) + room_version = KNOWN_ROOM_VERSIONS.get(room_version_id) + if not room_version: + raise UnsupportedRoomVersionError() + + event_format = room_version_to_event_format(room_version_id) pdu_dict = ret.get("event", None) if not isinstance(pdu_dict, dict): @@ -478,7 +494,7 @@ class FederationClient(FederationBase): event_dict=pdu_dict, ) - return (destination, ev, event_format) + return (destination, ev, room_version) return self._try_destination_list( "make_" + membership, destinations, send_request diff --git a/synapse/handlers/federation.py b/synapse/handlers/federation.py index d4f9a792fc..f824ee79a0 100644 --- a/synapse/handlers/federation.py +++ b/synapse/handlers/federation.py @@ -44,10 +44,10 @@ from synapse.api.errors import ( StoreError, SynapseError, ) -from synapse.api.room_versions import KNOWN_ROOM_VERSIONS, RoomVersions +from synapse.api.room_versions import KNOWN_ROOM_VERSIONS, RoomVersion, RoomVersions from synapse.crypto.event_signing import compute_event_signature from synapse.event_auth import auth_types_for_event -from synapse.events import EventBase +from synapse.events import EventBase, room_version_to_event_format from synapse.events.snapshot import EventContext from synapse.events.validator import EventValidator from synapse.logging.context import ( @@ -703,8 +703,20 @@ class FederationHandler(BaseHandler): if not room: try: + prev_state_ids = await context.get_prev_state_ids() + create_event = await self.store.get_event( + prev_state_ids[(EventTypes.Create, "")] + ) + + room_version_id = create_event.content.get( + "room_version", RoomVersions.V1.identifier + ) + await self.store.store_room( - room_id=room_id, room_creator_user_id="", is_public=False + room_id=room_id, + room_creator_user_id="", + is_public=False, + room_version=KNOWN_ROOM_VERSIONS[room_version_id], ) except StoreError: logger.exception("Failed to store room.") @@ -1186,7 +1198,7 @@ class FederationHandler(BaseHandler): """ logger.debug("Joining %s to %s", joinee, room_id) - origin, event, event_format_version = yield self._make_and_verify_event( + origin, event, room_version = yield self._make_and_verify_event( target_hosts, room_id, joinee, @@ -1214,6 +1226,8 @@ class FederationHandler(BaseHandler): target_hosts.insert(0, origin) except ValueError: pass + + event_format_version = room_version_to_event_format(room_version.identifier) ret = yield self.federation_client.send_join( target_hosts, event, event_format_version ) @@ -1234,13 +1248,18 @@ class FederationHandler(BaseHandler): try: yield self.store.store_room( - room_id=room_id, room_creator_user_id="", is_public=False + room_id=room_id, + room_creator_user_id="", + is_public=False, + room_version=room_version, ) except Exception: # FIXME pass - yield self._persist_auth_tree(origin, auth_chain, state, event) + yield self._persist_auth_tree( + origin, auth_chain, state, event, room_version + ) # Check whether this room is the result of an upgrade of a room we already know # about. If so, migrate over user information @@ -1486,7 +1505,7 @@ class FederationHandler(BaseHandler): @defer.inlineCallbacks def do_remotely_reject_invite(self, target_hosts, room_id, user_id, content): - origin, event, event_format_version = yield self._make_and_verify_event( + origin, event, room_version = yield self._make_and_verify_event( target_hosts, room_id, user_id, "leave", content=content ) # Mark as outlier as we don't have any state for this event; we're not @@ -1513,7 +1532,11 @@ class FederationHandler(BaseHandler): def _make_and_verify_event( self, target_hosts, room_id, user_id, membership, content={}, params=None ): - origin, event, format_ver = yield self.federation_client.make_membership_event( + ( + origin, + event, + room_version, + ) = yield self.federation_client.make_membership_event( target_hosts, room_id, user_id, membership, content, params=params ) @@ -1525,7 +1548,7 @@ class FederationHandler(BaseHandler): assert event.user_id == user_id assert event.state_key == user_id assert event.room_id == room_id - return origin, event, format_ver + return origin, event, room_version @defer.inlineCallbacks @log_function @@ -1810,7 +1833,14 @@ class FederationHandler(BaseHandler): ) @defer.inlineCallbacks - def _persist_auth_tree(self, origin, auth_events, state, event): + def _persist_auth_tree( + self, + origin: str, + auth_events: List[EventBase], + state: List[EventBase], + event: EventBase, + room_version: RoomVersion, + ): """Checks the auth chain is valid (and passes auth checks) for the state and event. Then persists the auth chain and state atomically. Persists the event separately. Notifies about the persisted events @@ -1819,10 +1849,12 @@ class FederationHandler(BaseHandler): Will attempt to fetch missing auth events. Args: - origin (str): Where the events came from - auth_events (list) - state (list) - event (Event) + origin: Where the events came from + auth_events + state + event + room_version: The room version we expect this room to have, and + will raise if it doesn't match the version in the create event. Returns: Deferred @@ -1848,10 +1880,13 @@ class FederationHandler(BaseHandler): # invalid, and it would fail auth checks anyway. raise SynapseError(400, "No create event in state") - room_version = create_event.content.get( + room_version_id = create_event.content.get( "room_version", RoomVersions.V1.identifier ) + if room_version.identifier != room_version_id: + raise SynapseError(400, "Room version mismatch") + missing_auth_events = set() for e in itertools.chain(auth_events, state, [event]): for e_id in e.auth_event_ids(): diff --git a/synapse/handlers/room.py b/synapse/handlers/room.py index 9f50196ea7..a9490782b7 100644 --- a/synapse/handlers/room.py +++ b/synapse/handlers/room.py @@ -29,7 +29,7 @@ from twisted.internet import defer from synapse.api.constants import EventTypes, JoinRules, RoomCreationPreset from synapse.api.errors import AuthError, Codes, NotFoundError, StoreError, SynapseError -from synapse.api.room_versions import KNOWN_ROOM_VERSIONS +from synapse.api.room_versions import KNOWN_ROOM_VERSIONS, RoomVersion from synapse.http.endpoint import parse_and_validate_server_name from synapse.storage.state import StateFilter from synapse.types import ( @@ -100,13 +100,15 @@ class RoomCreationHandler(BaseHandler): self.third_party_event_rules = hs.get_third_party_event_rules() @defer.inlineCallbacks - def upgrade_room(self, requester, old_room_id, new_version): + def upgrade_room( + self, requester: Requester, old_room_id: str, new_version: RoomVersion + ): """Replace a room with a new room with a different version Args: - requester (synapse.types.Requester): the user requesting the upgrade - old_room_id (unicode): the id of the room to be replaced - new_version (unicode): the new room version to use + requester: the user requesting the upgrade + old_room_id: the id of the room to be replaced + new_version: the new room version to use Returns: Deferred[unicode]: the new room id @@ -151,7 +153,7 @@ class RoomCreationHandler(BaseHandler): if r is None: raise NotFoundError("Unknown room id %s" % (old_room_id,)) new_room_id = yield self._generate_room_id( - creator_id=user_id, is_public=r["is_public"] + creator_id=user_id, is_public=r["is_public"], room_version=new_version, ) logger.info("Creating new room %s to replace %s", new_room_id, old_room_id) @@ -299,18 +301,22 @@ class RoomCreationHandler(BaseHandler): @defer.inlineCallbacks def clone_existing_room( - self, requester, old_room_id, new_room_id, new_room_version, tombstone_event_id + self, + requester: Requester, + old_room_id: str, + new_room_id: str, + new_room_version: RoomVersion, + tombstone_event_id: str, ): """Populate a new room based on an old room Args: - requester (synapse.types.Requester): the user requesting the upgrade - old_room_id (unicode): the id of the room to be replaced - new_room_id (unicode): the id to give the new room (should already have been + requester: the user requesting the upgrade + old_room_id : the id of the room to be replaced + new_room_id: the id to give the new room (should already have been created with _gemerate_room_id()) - new_room_version (unicode): the new room version to use - tombstone_event_id (unicode|str): the ID of the tombstone event in the old - room. + new_room_version: the new room version to use + tombstone_event_id: the ID of the tombstone event in the old room. Returns: Deferred """ @@ -320,7 +326,7 @@ class RoomCreationHandler(BaseHandler): raise SynapseError(403, "You are not permitted to create rooms") creation_content = { - "room_version": new_room_version, + "room_version": new_room_version.identifier, "predecessor": {"room_id": old_room_id, "event_id": tombstone_event_id}, } @@ -577,14 +583,15 @@ class RoomCreationHandler(BaseHandler): if ratelimit: yield self.ratelimit(requester) - room_version = config.get( + room_version_id = config.get( "room_version", self.config.default_room_version.identifier ) - if not isinstance(room_version, string_types): + if not isinstance(room_version_id, string_types): raise SynapseError(400, "room_version must be a string", Codes.BAD_JSON) - if room_version not in KNOWN_ROOM_VERSIONS: + room_version = KNOWN_ROOM_VERSIONS.get(room_version_id) + if room_version is None: raise SynapseError( 400, "Your homeserver does not support this room version", @@ -631,7 +638,9 @@ class RoomCreationHandler(BaseHandler): visibility = config.get("visibility", None) is_public = visibility == "public" - room_id = yield self._generate_room_id(creator_id=user_id, is_public=is_public) + room_id = yield self._generate_room_id( + creator_id=user_id, is_public=is_public, room_version=room_version, + ) directory_handler = self.hs.get_handlers().directory_handler if room_alias: @@ -660,7 +669,7 @@ class RoomCreationHandler(BaseHandler): creation_content = config.get("creation_content", {}) # override any attempt to set room versions via the creation_content - creation_content["room_version"] = room_version + creation_content["room_version"] = room_version.identifier yield self._send_events_for_new_room( requester, @@ -849,7 +858,9 @@ class RoomCreationHandler(BaseHandler): yield send(etype=etype, state_key=state_key, content=content) @defer.inlineCallbacks - def _generate_room_id(self, creator_id, is_public): + def _generate_room_id( + self, creator_id: str, is_public: str, room_version: RoomVersion, + ): # autogen room IDs and try to create it. We may clash, so just # try a few times till one goes through, giving up eventually. attempts = 0 @@ -863,6 +874,7 @@ class RoomCreationHandler(BaseHandler): room_id=gen_room_id, room_creator_user_id=creator_id, is_public=is_public, + room_version=room_version, ) return gen_room_id except StoreError: diff --git a/synapse/rest/client/v2_alpha/room_upgrade_rest_servlet.py b/synapse/rest/client/v2_alpha/room_upgrade_rest_servlet.py index ca97330797..f357015a70 100644 --- a/synapse/rest/client/v2_alpha/room_upgrade_rest_servlet.py +++ b/synapse/rest/client/v2_alpha/room_upgrade_rest_servlet.py @@ -64,7 +64,8 @@ class RoomUpgradeRestServlet(RestServlet): assert_params_in_dict(content, ("new_version",)) new_version = content["new_version"] - if new_version not in KNOWN_ROOM_VERSIONS: + new_version = KNOWN_ROOM_VERSIONS.get(content["new_version"]) + if new_version is None: raise SynapseError( 400, "Your homeserver does not support this room version", diff --git a/synapse/storage/data_stores/main/room.py b/synapse/storage/data_stores/main/room.py index d968803ad2..9a17e336ba 100644 --- a/synapse/storage/data_stores/main/room.py +++ b/synapse/storage/data_stores/main/room.py @@ -29,9 +29,10 @@ from twisted.internet import defer from synapse.api.constants import EventTypes from synapse.api.errors import StoreError +from synapse.api.room_versions import RoomVersion, RoomVersions from synapse.storage._base import SQLBaseStore from synapse.storage.data_stores.main.search import SearchStore -from synapse.storage.database import Database +from synapse.storage.database import Database, LoggingTransaction from synapse.types import ThirdPartyInstanceID from synapse.util.caches.descriptors import cached, cachedInlineCallbacks @@ -734,6 +735,7 @@ class RoomWorkerStore(SQLBaseStore): class RoomBackgroundUpdateStore(SQLBaseStore): REMOVE_TOMESTONED_ROOMS_BG_UPDATE = "remove_tombstoned_rooms_from_directory" + ADD_ROOMS_ROOM_VERSION_COLUMN = "add_rooms_room_version_column" def __init__(self, database: Database, db_conn, hs): super(RoomBackgroundUpdateStore, self).__init__(database, db_conn, hs) @@ -749,6 +751,11 @@ class RoomBackgroundUpdateStore(SQLBaseStore): self._remove_tombstoned_rooms_from_directory, ) + self.db.updates.register_background_update_handler( + self.ADD_ROOMS_ROOM_VERSION_COLUMN, + self._background_add_rooms_room_version_column, + ) + @defer.inlineCallbacks def _background_insert_retention(self, progress, batch_size): """Retrieves a list of all rooms within a range and inserts an entry for each of @@ -817,6 +824,73 @@ class RoomBackgroundUpdateStore(SQLBaseStore): defer.returnValue(batch_size) + async def _background_add_rooms_room_version_column( + self, progress: dict, batch_size: int + ): + """Background update to go and add room version inforamtion to `rooms` + table from `current_state_events` table. + """ + + last_room_id = progress.get("room_id", "") + + def _background_add_rooms_room_version_column_txn(txn: LoggingTransaction): + sql = """ + SELECT room_id, json FROM current_state_events + INNER JOIN event_json USING (room_id, event_id) + WHERE room_id > ? AND type = 'm.room.create' AND state_key = '' + ORDER BY room_id + LIMIT ? + """ + + txn.execute(sql, (last_room_id, batch_size)) + + updates = [] + for room_id, event_json in txn: + event_dict = json.loads(event_json) + room_version_id = event_dict.get("content", {}).get( + "room_version", RoomVersions.V1.identifier + ) + + creator = event_dict.get("content").get("creator") + + updates.append((room_id, creator, room_version_id)) + + if not updates: + return True + + new_last_room_id = "" + for room_id, creator, room_version_id in updates: + # We upsert here just in case we don't already have a row, + # mainly for paranoia as much badness would happen if we don't + # insert the row and then try and get the room version for the + # room. + self.db.simple_upsert_txn( + txn, + table="rooms", + keyvalues={"room_id": room_id}, + values={"room_version": room_version_id}, + insertion_values={"is_public": False, "creator": creator}, + ) + new_last_room_id = room_id + + self.db.updates._background_update_progress_txn( + txn, self.ADD_ROOMS_ROOM_VERSION_COLUMN, {"room_id": new_last_room_id} + ) + + return False + + end = await self.db.runInteraction( + "_background_add_rooms_room_version_column", + _background_add_rooms_room_version_column_txn, + ) + + if end: + await self.db.updates._end_background_update( + self.ADD_ROOMS_ROOM_VERSION_COLUMN + ) + + return batch_size + async def _remove_tombstoned_rooms_from_directory( self, progress, batch_size ) -> int: @@ -881,14 +955,21 @@ class RoomStore(RoomBackgroundUpdateStore, RoomWorkerStore, SearchStore): self.config = hs.config @defer.inlineCallbacks - def store_room(self, room_id, room_creator_user_id, is_public): + def store_room( + self, + room_id: str, + room_creator_user_id: str, + is_public: bool, + room_version: RoomVersion, + ): """Stores a room. Args: - room_id (str): The desired room ID, can be None. - room_creator_user_id (str): The user ID of the room creator. - is_public (bool): True to indicate that this room should appear in - public room lists. + room_id: The desired room ID, can be None. + room_creator_user_id: The user ID of the room creator. + is_public: True to indicate that this room should appear in + public room lists. + room_version: The version of the room Raises: StoreError if the room could not be stored. """ @@ -902,6 +983,7 @@ class RoomStore(RoomBackgroundUpdateStore, RoomWorkerStore, SearchStore): "room_id": room_id, "creator": room_creator_user_id, "is_public": is_public, + "room_version": room_version.identifier, }, ) if is_public: diff --git a/synapse/storage/data_stores/main/schema/delta/57/rooms_version_column.sql b/synapse/storage/data_stores/main/schema/delta/57/rooms_version_column.sql new file mode 100644 index 0000000000..352a66f5b0 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/57/rooms_version_column.sql @@ -0,0 +1,24 @@ +/* Copyright 2020 The Matrix.org Foundation C.I.C + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + +-- We want to start storing the room version independently of +-- `current_state_events` so that we can delete stale entries from it without +-- losing the information. +ALTER TABLE rooms ADD COLUMN room_version TEXT; + + +INSERT into background_updates (update_name, progress_json) + VALUES ('add_rooms_room_version_column', '{}'); diff --git a/synapse/storage/data_stores/main/state.py b/synapse/storage/data_stores/main/state.py index 33bebd1c48..bd7b0276f1 100644 --- a/synapse/storage/data_stores/main/state.py +++ b/synapse/storage/data_stores/main/state.py @@ -60,24 +60,34 @@ class StateGroupWorkerStore(EventsWorkerStore, SQLBaseStore): def __init__(self, database: Database, db_conn, hs): super(StateGroupWorkerStore, self).__init__(database, db_conn, hs) - @defer.inlineCallbacks - def get_room_version(self, room_id): + @cached(max_entries=10000) + async def get_room_version(self, room_id: str) -> str: """Get the room_version of a given room - Args: - room_id (str) - - Returns: - Deferred[str] - Raises: - NotFoundError if the room is unknown + NotFoundError: if the room is unknown """ - # for now we do this by looking at the create event. We may want to cache this - # more intelligently in future. + + # First we try looking up room version from the database, but for old + # rooms we might not have added the room version to it yet so we fall + # back to previous behaviour and look in current state events. + + # We really should have an entry in the rooms table for every room we + # care about, but let's be a bit paranoid (at least while the background + # update is happening) to avoid breaking existing rooms. + version = await self.db.simple_select_one_onecol( + table="rooms", + keyvalues={"room_id": room_id}, + retcol="room_version", + desc="get_room_version", + allow_none=True, + ) + + if version is not None: + return version # Retrieve the room's create event - create_event = yield self.get_create_event_for_room(room_id) + create_event = await self.get_create_event_for_room(room_id) return create_event.content.get("room_version", "1") @defer.inlineCallbacks diff --git a/tests/storage/test_room.py b/tests/storage/test_room.py index 3ddaa151fe..086adeb8fd 100644 --- a/tests/storage/test_room.py +++ b/tests/storage/test_room.py @@ -17,6 +17,7 @@ from twisted.internet import defer from synapse.api.constants import EventTypes +from synapse.api.room_versions import RoomVersions from synapse.types import RoomAlias, RoomID, UserID from tests import unittest @@ -40,6 +41,7 @@ class RoomStoreTestCase(unittest.TestCase): self.room.to_string(), room_creator_user_id=self.u_creator.to_string(), is_public=True, + room_version=RoomVersions.V1, ) @defer.inlineCallbacks @@ -68,7 +70,10 @@ class RoomEventsStoreTestCase(unittest.TestCase): self.room = RoomID.from_string("!abcde:test") yield self.store.store_room( - self.room.to_string(), room_creator_user_id="@creator:text", is_public=True + self.room.to_string(), + room_creator_user_id="@creator:text", + is_public=True, + room_version=RoomVersions.V1, ) @defer.inlineCallbacks diff --git a/tests/storage/test_state.py b/tests/storage/test_state.py index d6ecf102f8..04d58fbf24 100644 --- a/tests/storage/test_state.py +++ b/tests/storage/test_state.py @@ -45,7 +45,10 @@ class StateStoreTestCase(tests.unittest.TestCase): self.room = RoomID.from_string("!abc123:test") yield self.store.store_room( - self.room.to_string(), room_creator_user_id="@creator:text", is_public=True + self.room.to_string(), + room_creator_user_id="@creator:text", + is_public=True, + room_version=RoomVersions.V1, ) @defer.inlineCallbacks diff --git a/tests/utils.py b/tests/utils.py index e2e9cafd79..513f358f4f 100644 --- a/tests/utils.py +++ b/tests/utils.py @@ -639,9 +639,17 @@ def create_room(hs, room_id, creator_id): """ persistence_store = hs.get_storage().persistence + store = hs.get_datastore() event_builder_factory = hs.get_event_builder_factory() event_creation_handler = hs.get_event_creation_handler() + yield store.store_room( + room_id=room_id, + room_creator_user_id=creator_id, + is_public=False, + room_version=RoomVersions.V1, + ) + builder = event_builder_factory.for_room_version( RoomVersions.V1, { -- cgit 1.4.1 From a8ce7aeb433e08f46306797a1252668c178a7825 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Tue, 28 Jan 2020 14:18:29 +0000 Subject: Pass room version object into event_auth.check and check_redaction (#6788) These are easier to work with than the strings and we normally have one around. This fixes `FederationHander._persist_auth_tree` which was passing a RoomVersion object into event_auth.check instead of a string. --- changelog.d/6788.misc | 1 + synapse/api/auth.py | 7 +++++-- synapse/event_auth.py | 34 +++++++++++++++++++++------------- synapse/handlers/federation.py | 18 +++++++++++------- synapse/handlers/message.py | 8 ++++++-- synapse/state/v1.py | 4 ++-- synapse/state/v2.py | 4 +++- tests/test_event_auth.py | 11 ++++------- 8 files changed, 53 insertions(+), 34 deletions(-) create mode 100644 changelog.d/6788.misc (limited to 'tests') diff --git a/changelog.d/6788.misc b/changelog.d/6788.misc new file mode 100644 index 0000000000..5537355bea --- /dev/null +++ b/changelog.d/6788.misc @@ -0,0 +1 @@ +Record room versions in the `rooms` table. diff --git a/synapse/api/auth.py b/synapse/api/auth.py index 2cbfab2569..8b1277ad02 100644 --- a/synapse/api/auth.py +++ b/synapse/api/auth.py @@ -33,6 +33,7 @@ from synapse.api.errors import ( MissingClientTokenError, ResourceLimitError, ) +from synapse.api.room_versions import KNOWN_ROOM_VERSIONS from synapse.config.server import is_threepid_reserved from synapse.types import StateMap, UserID from synapse.util.caches import CACHE_SIZE_FACTOR, register_cache @@ -77,15 +78,17 @@ class Auth(object): self._account_validity = hs.config.account_validity @defer.inlineCallbacks - def check_from_context(self, room_version, event, context, do_sig_check=True): + def check_from_context(self, room_version: str, event, context, do_sig_check=True): prev_state_ids = yield context.get_prev_state_ids() auth_events_ids = yield self.compute_auth_events( event, prev_state_ids, for_verification=True ) auth_events = yield self.store.get_events(auth_events_ids) auth_events = {(e.type, e.state_key): e for e in itervalues(auth_events)} + + room_version_obj = KNOWN_ROOM_VERSIONS[room_version] event_auth.check( - room_version, event, auth_events=auth_events, do_sig_check=do_sig_check + room_version_obj, event, auth_events=auth_events, do_sig_check=do_sig_check ) @defer.inlineCallbacks diff --git a/synapse/event_auth.py b/synapse/event_auth.py index e3a1ba47a0..016d5678e5 100644 --- a/synapse/event_auth.py +++ b/synapse/event_auth.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014 - 2016 OpenMarket Ltd +# Copyright 2020 The Matrix.org Foundation C.I.C. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -23,17 +24,27 @@ from unpaddedbase64 import decode_base64 from synapse.api.constants import EventTypes, JoinRules, Membership from synapse.api.errors import AuthError, EventSizeError, SynapseError -from synapse.api.room_versions import KNOWN_ROOM_VERSIONS, EventFormatVersions +from synapse.api.room_versions import ( + KNOWN_ROOM_VERSIONS, + EventFormatVersions, + RoomVersion, +) from synapse.types import UserID, get_domain_from_id logger = logging.getLogger(__name__) -def check(room_version, event, auth_events, do_sig_check=True, do_size_check=True): +def check( + room_version_obj: RoomVersion, + event, + auth_events, + do_sig_check=True, + do_size_check=True, +): """ Checks if this event is correctly authed. Args: - room_version (str): the version of the room + room_version_obj: the version of the room event: the event being checked. auth_events (dict: event-key -> event): the existing room state. @@ -97,10 +108,11 @@ def check(room_version, event, auth_events, do_sig_check=True, do_size_check=Tru 403, "Creation event's room_id domain does not match sender's" ) - room_version = event.content.get("room_version", "1") - if room_version not in KNOWN_ROOM_VERSIONS: + room_version_prop = event.content.get("room_version", "1") + if room_version_prop not in KNOWN_ROOM_VERSIONS: raise AuthError( - 403, "room appears to have unsupported version %s" % (room_version,) + 403, + "room appears to have unsupported version %s" % (room_version_prop,), ) # FIXME logger.debug("Allowing! %s", event) @@ -160,7 +172,7 @@ def check(room_version, event, auth_events, do_sig_check=True, do_size_check=Tru _check_power_levels(event, auth_events) if event.type == EventTypes.Redaction: - check_redaction(room_version, event, auth_events) + check_redaction(room_version_obj, event, auth_events) logger.debug("Allowing! %s", event) @@ -386,7 +398,7 @@ def _can_send_event(event, auth_events): return True -def check_redaction(room_version, event, auth_events): +def check_redaction(room_version_obj: RoomVersion, event, auth_events): """Check whether the event sender is allowed to redact the target event. Returns: @@ -406,11 +418,7 @@ def check_redaction(room_version, event, auth_events): if user_level >= redact_level: return False - v = KNOWN_ROOM_VERSIONS.get(room_version) - if not v: - raise RuntimeError("Unrecognized room version %r" % (room_version,)) - - if v.event_format == EventFormatVersions.V1: + if room_version_obj.event_format == EventFormatVersions.V1: redacter_domain = get_domain_from_id(event.event_id) redactee_domain = get_domain_from_id(event.redacts) if redacter_domain == redactee_domain: diff --git a/synapse/handlers/federation.py b/synapse/handlers/federation.py index f824ee79a0..180f165a7a 100644 --- a/synapse/handlers/federation.py +++ b/synapse/handlers/federation.py @@ -47,7 +47,7 @@ from synapse.api.errors import ( from synapse.api.room_versions import KNOWN_ROOM_VERSIONS, RoomVersion, RoomVersions from synapse.crypto.event_signing import compute_event_signature from synapse.event_auth import auth_types_for_event -from synapse.events import EventBase, room_version_to_event_format +from synapse.events import EventBase from synapse.events.snapshot import EventContext from synapse.events.validator import EventValidator from synapse.logging.context import ( @@ -1198,7 +1198,7 @@ class FederationHandler(BaseHandler): """ logger.debug("Joining %s to %s", joinee, room_id) - origin, event, room_version = yield self._make_and_verify_event( + origin, event, room_version_obj = yield self._make_and_verify_event( target_hosts, room_id, joinee, @@ -1227,7 +1227,7 @@ class FederationHandler(BaseHandler): except ValueError: pass - event_format_version = room_version_to_event_format(room_version.identifier) + event_format_version = room_version_obj.event_format ret = yield self.federation_client.send_join( target_hosts, event, event_format_version ) @@ -1251,14 +1251,14 @@ class FederationHandler(BaseHandler): room_id=room_id, room_creator_user_id="", is_public=False, - room_version=room_version, + room_version=room_version_obj, ) except Exception: # FIXME pass yield self._persist_auth_tree( - origin, auth_chain, state, event, room_version + origin, auth_chain, state, event, room_version_obj ) # Check whether this room is the result of an upgrade of a room we already know @@ -2022,6 +2022,7 @@ class FederationHandler(BaseHandler): if do_soft_fail_check: room_version = yield self.store.get_room_version(event.room_id) + room_version_obj = KNOWN_ROOM_VERSIONS[room_version] # Calculate the "current state". if state is not None: @@ -2071,7 +2072,9 @@ class FederationHandler(BaseHandler): } try: - event_auth.check(room_version, event, auth_events=current_auth_events) + event_auth.check( + room_version_obj, event, auth_events=current_auth_events + ) except AuthError as e: logger.warning("Soft-failing %r because %s", event, e) event.internal_metadata.soft_failed = True @@ -2155,6 +2158,7 @@ class FederationHandler(BaseHandler): defer.Deferred[EventContext]: updated context object """ room_version = yield self.store.get_room_version(event.room_id) + room_version_obj = KNOWN_ROOM_VERSIONS[room_version] try: context = yield self._update_auth_events_and_context_for_auth( @@ -2172,7 +2176,7 @@ class FederationHandler(BaseHandler): ) try: - event_auth.check(room_version, event, auth_events=auth_events) + event_auth.check(room_version_obj, event, auth_events=auth_events) except AuthError as e: logger.warning("Failed auth resolution for %r because %s", event, e) context.rejected = RejectedReason.AUTH_ERROR diff --git a/synapse/handlers/message.py b/synapse/handlers/message.py index 8ea3aca2f4..9a0f661b9b 100644 --- a/synapse/handlers/message.py +++ b/synapse/handlers/message.py @@ -40,7 +40,7 @@ from synapse.api.errors import ( NotFoundError, SynapseError, ) -from synapse.api.room_versions import RoomVersions +from synapse.api.room_versions import KNOWN_ROOM_VERSIONS, RoomVersions from synapse.api.urls import ConsentURIBuilder from synapse.events.validator import EventValidator from synapse.logging.context import run_in_background @@ -962,9 +962,13 @@ class EventCreationHandler(object): ) auth_events = yield self.store.get_events(auth_events_ids) auth_events = {(e.type, e.state_key): e for e in auth_events.values()} + room_version = yield self.store.get_room_version(event.room_id) + room_version_obj = KNOWN_ROOM_VERSIONS[room_version] - if event_auth.check_redaction(room_version, event, auth_events=auth_events): + if event_auth.check_redaction( + room_version_obj, event, auth_events=auth_events + ): # this user doesn't have 'redact' rights, so we need to do some more # checks on the original event. Let's start by checking the original # event exists. diff --git a/synapse/state/v1.py b/synapse/state/v1.py index d6c34ce3b7..24b7c0faef 100644 --- a/synapse/state/v1.py +++ b/synapse/state/v1.py @@ -281,7 +281,7 @@ def _resolve_auth_events(events, auth_events): try: # The signatures have already been checked at this point event_auth.check( - RoomVersions.V1.identifier, + RoomVersions.V1, event, auth_events, do_sig_check=False, @@ -299,7 +299,7 @@ def _resolve_normal_events(events, auth_events): try: # The signatures have already been checked at this point event_auth.check( - RoomVersions.V1.identifier, + RoomVersions.V1, event, auth_events, do_sig_check=False, diff --git a/synapse/state/v2.py b/synapse/state/v2.py index 6216fdd204..531018c6a5 100644 --- a/synapse/state/v2.py +++ b/synapse/state/v2.py @@ -26,6 +26,7 @@ import synapse.state from synapse import event_auth from synapse.api.constants import EventTypes from synapse.api.errors import AuthError +from synapse.api.room_versions import KNOWN_ROOM_VERSIONS from synapse.events import EventBase from synapse.types import StateMap @@ -402,6 +403,7 @@ def _iterative_auth_checks( Deferred[StateMap[str]]: Returns the final updated state """ resolved_state = base_state.copy() + room_version_obj = KNOWN_ROOM_VERSIONS[room_version] for event_id in event_ids: event = event_map[event_id] @@ -430,7 +432,7 @@ def _iterative_auth_checks( try: event_auth.check( - room_version, + room_version_obj, event, auth_events, do_sig_check=False, diff --git a/tests/test_event_auth.py b/tests/test_event_auth.py index 8b2741d277..ca20b085a2 100644 --- a/tests/test_event_auth.py +++ b/tests/test_event_auth.py @@ -37,7 +37,7 @@ class EventAuthTestCase(unittest.TestCase): # creator should be able to send state event_auth.check( - RoomVersions.V1.identifier, + RoomVersions.V1, _random_state_event(creator), auth_events, do_sig_check=False, @@ -47,7 +47,7 @@ class EventAuthTestCase(unittest.TestCase): self.assertRaises( AuthError, event_auth.check, - RoomVersions.V1.identifier, + RoomVersions.V1, _random_state_event(joiner), auth_events, do_sig_check=False, @@ -76,7 +76,7 @@ class EventAuthTestCase(unittest.TestCase): self.assertRaises( AuthError, event_auth.check, - RoomVersions.V1.identifier, + RoomVersions.V1, _random_state_event(pleb), auth_events, do_sig_check=False, @@ -84,10 +84,7 @@ class EventAuthTestCase(unittest.TestCase): # king should be able to send state event_auth.check( - RoomVersions.V1.identifier, - _random_state_event(king), - auth_events, - do_sig_check=False, + RoomVersions.V1, _random_state_event(king), auth_events, do_sig_check=False, ) -- cgit 1.4.1 From ee42a5513e020424daa736962fee7fb69bd2373a Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Tue, 28 Jan 2020 11:02:55 +0000 Subject: Factor out a `copy_power_levels_contents` method I'm going to need another copy (hah!) of this. --- synapse/events/utils.py | 37 ++++++++++++++++++++++++++++++++++++- synapse/handlers/room.py | 23 +++++++++++------------ tests/events/test_utils.py | 45 +++++++++++++++++++++++++++++++++++++++++++-- 3 files changed, 90 insertions(+), 15 deletions(-) (limited to 'tests') diff --git a/synapse/events/utils.py b/synapse/events/utils.py index 07d1c5bcf0..be57c6d9be 100644 --- a/synapse/events/utils.py +++ b/synapse/events/utils.py @@ -12,8 +12,9 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. - +import collections import re +from typing import Mapping, Union from six import string_types @@ -422,3 +423,37 @@ class EventClientSerializer(object): return yieldable_gather_results( self.serialize_event, events, time_now=time_now, **kwargs ) + + +def copy_power_levels_contents( + old_power_levels: Mapping[str, Union[int, Mapping[str, int]]] +): + """Copy the content of a power_levels event, unfreezing frozendicts along the way + + Raises: + TypeError if the input does not look like a valid power levels event content + """ + if not isinstance(old_power_levels, collections.Mapping): + raise TypeError("Not a valid power-levels content: %r" % (old_power_levels,)) + + power_levels = {} + for k, v in old_power_levels.items(): + + if isinstance(v, int): + power_levels[k] = v + continue + + if isinstance(v, collections.Mapping): + power_levels[k] = h = {} + for k1, v1 in v.items(): + # we should only have one level of nesting + if not isinstance(v1, int): + raise TypeError( + "Invalid power_levels value for %s.%s: %r" % (k, k1, v) + ) + h[k1] = v1 + continue + + raise TypeError("Invalid power_levels value for %s: %r" % (k, v)) + + return power_levels diff --git a/synapse/handlers/room.py b/synapse/handlers/room.py index a9490782b7..532ee22fa4 100644 --- a/synapse/handlers/room.py +++ b/synapse/handlers/room.py @@ -30,6 +30,7 @@ from twisted.internet import defer from synapse.api.constants import EventTypes, JoinRules, RoomCreationPreset from synapse.api.errors import AuthError, Codes, NotFoundError, StoreError, SynapseError from synapse.api.room_versions import KNOWN_ROOM_VERSIONS, RoomVersion +from synapse.events.utils import copy_power_levels_contents from synapse.http.endpoint import parse_and_validate_server_name from synapse.storage.state import StateFilter from synapse.types import ( @@ -367,6 +368,15 @@ class RoomCreationHandler(BaseHandler): if old_event: initial_state[k] = old_event.content + # deep-copy the power-levels event before we start modifying it + # note that if frozen_dicts are enabled, `power_levels` will be a frozen + # dict so we can't just copy.deepcopy it. + initial_state[ + (EventTypes.PowerLevels, "") + ] = power_levels = copy_power_levels_contents( + initial_state[(EventTypes.PowerLevels, "")] + ) + # Resolve the minimum power level required to send any state event # We will give the upgrading user this power level temporarily (if necessary) such that # they are able to copy all of the state events over, then revert them back to their @@ -375,8 +385,6 @@ class RoomCreationHandler(BaseHandler): # Copy over user power levels now as this will not be possible with >100PL users once # the room has been created - power_levels = initial_state[(EventTypes.PowerLevels, "")] - # Calculate the minimum power level needed to clone the room event_power_levels = power_levels.get("events", {}) state_default = power_levels.get("state_default", 0) @@ -386,16 +394,7 @@ class RoomCreationHandler(BaseHandler): # Raise the requester's power level in the new room if necessary current_power_level = power_levels["users"][user_id] if current_power_level < needed_power_level: - # make sure we copy the event content rather than overwriting it. - # note that if frozen_dicts are enabled, `power_levels` will be a frozen - # dict so we can't just copy.deepcopy it. - - new_power_levels = {k: v for k, v in power_levels.items() if k != "users"} - new_power_levels["users"] = { - k: v for k, v in power_levels.get("users", {}).items() if k != user_id - } - new_power_levels["users"][user_id] = needed_power_level - initial_state[(EventTypes.PowerLevels, "")] = new_power_levels + power_levels["users"][user_id] = needed_power_level yield self._send_events_for_new_room( requester, diff --git a/tests/events/test_utils.py b/tests/events/test_utils.py index 9e3d4d0f47..2b13980dfd 100644 --- a/tests/events/test_utils.py +++ b/tests/events/test_utils.py @@ -15,9 +15,14 @@ from synapse.events import FrozenEvent -from synapse.events.utils import prune_event, serialize_event +from synapse.events.utils import ( + copy_power_levels_contents, + prune_event, + serialize_event, +) +from synapse.util.frozenutils import freeze -from .. import unittest +from tests import unittest def MockEvent(**kwargs): @@ -241,3 +246,39 @@ class SerializeEventTestCase(unittest.TestCase): self.serialize( MockEvent(room_id="!foo:bar", content={"foo": "bar"}), ["room_id", 4] ) + + +class CopyPowerLevelsContentTestCase(unittest.TestCase): + def setUp(self) -> None: + self.test_content = { + "ban": 50, + "events": {"m.room.name": 100, "m.room.power_levels": 100}, + "events_default": 0, + "invite": 50, + "kick": 50, + "notifications": {"room": 20}, + "redact": 50, + "state_default": 50, + "users": {"@example:localhost": 100}, + "users_default": 0, + } + + def _test(self, input): + a = copy_power_levels_contents(input) + + self.assertEqual(a["ban"], 50) + self.assertEqual(a["events"]["m.room.name"], 100) + + # make sure that changing the copy changes the copy and not the orig + a["ban"] = 10 + a["events"]["m.room.power_levels"] = 20 + + self.assertEqual(input["ban"], 50) + self.assertEqual(input["events"]["m.room.power_levels"], 100) + + def test_unfrozen(self): + self._test(self.test_content) + + def test_frozen(self): + input = freeze(self.test_content) + self._test(input) -- cgit 1.4.1 From 5a246611e3cbf27cf1dd7e4453adc8040cddd6a2 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Thu, 30 Jan 2020 11:25:59 +0000 Subject: Type defintions for use in refactoring for redaction changes (#6803) * Bump signedjson to 1.1 ... so that we can use the type definitions * Fix breakage caused by upgrade to signedjson 1.1 Thanks, @illicitonion... --- changelog.d/6803.misc | 1 + synapse/events/__init__.py | 5 +++-- synapse/python_dependencies.py | 4 +++- synapse/types.py | 7 ++++++- tests/storage/test_keys.py | 15 +++++++++++---- 5 files changed, 24 insertions(+), 8 deletions(-) create mode 100644 changelog.d/6803.misc (limited to 'tests') diff --git a/changelog.d/6803.misc b/changelog.d/6803.misc new file mode 100644 index 0000000000..08aa80bcd9 --- /dev/null +++ b/changelog.d/6803.misc @@ -0,0 +1 @@ +Refactoring work in preparation for changing the event redaction algorithm. diff --git a/synapse/events/__init__.py b/synapse/events/__init__.py index 72c09327f4..f813fa2fe7 100644 --- a/synapse/events/__init__.py +++ b/synapse/events/__init__.py @@ -23,6 +23,7 @@ from unpaddedbase64 import encode_base64 from synapse.api.errors import UnsupportedRoomVersionError from synapse.api.room_versions import KNOWN_ROOM_VERSIONS, EventFormatVersions +from synapse.types import JsonDict from synapse.util.caches import intern_dict from synapse.util.frozenutils import freeze @@ -197,7 +198,7 @@ class EventBase(object): def is_state(self): return hasattr(self, "state_key") and self.state_key is not None - def get_dict(self): + def get_dict(self) -> JsonDict: d = dict(self._event_dict) d.update({"signatures": self.signatures, "unsigned": dict(self.unsigned)}) @@ -209,7 +210,7 @@ class EventBase(object): def get_internal_metadata_dict(self): return self.internal_metadata.get_dict() - def get_pdu_json(self, time_now=None): + def get_pdu_json(self, time_now=None) -> JsonDict: pdu_json = self.get_dict() if time_now is not None and "age_ts" in pdu_json["unsigned"]: diff --git a/synapse/python_dependencies.py b/synapse/python_dependencies.py index 5871feaafd..8de8cb2c12 100644 --- a/synapse/python_dependencies.py +++ b/synapse/python_dependencies.py @@ -1,6 +1,7 @@ # Copyright 2015, 2016 OpenMarket Ltd # Copyright 2017 Vector Creations Ltd # Copyright 2018 New Vector Ltd +# Copyright 2020 The Matrix.org Foundation C.I.C. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -43,7 +44,8 @@ REQUIREMENTS = [ "frozendict>=1", "unpaddedbase64>=1.1.0", "canonicaljson>=1.1.3", - "signedjson>=1.0.0", + # we use the type definitions added in signedjson 1.1. + "signedjson>=1.1.0", "pynacl>=1.2.1", "idna>=2.5", # validating SSL certs for IP addresses requires service_identity 18.1. diff --git a/synapse/types.py b/synapse/types.py index 65e4d8c181..f3cd465735 100644 --- a/synapse/types.py +++ b/synapse/types.py @@ -17,7 +17,7 @@ import re import string import sys from collections import namedtuple -from typing import Dict, Tuple, TypeVar +from typing import Any, Dict, Tuple, TypeVar import attr from signedjson.key import decode_verify_key_bytes @@ -43,6 +43,11 @@ T = TypeVar("T") StateMap = Dict[Tuple[str, str], T] +# the type of a JSON-serialisable dict. This could be made stronger, but it will +# do for now. +JsonDict = Dict[str, Any] + + class Requester( namedtuple( "Requester", ["user", "access_token_id", "is_guest", "device_id", "app_service"] diff --git a/tests/storage/test_keys.py b/tests/storage/test_keys.py index e07ff01201..95f309fbbc 100644 --- a/tests/storage/test_keys.py +++ b/tests/storage/test_keys.py @@ -14,6 +14,7 @@ # limitations under the License. import signedjson.key +import unpaddedbase64 from twisted.internet.defer import Deferred @@ -21,11 +22,17 @@ from synapse.storage.keys import FetchKeyResult import tests.unittest -KEY_1 = signedjson.key.decode_verify_key_base64( - "ed25519", "key1", "fP5l4JzpZPq/zdbBg5xx6lQGAAOM9/3w94cqiJ5jPrw" + +def decode_verify_key_base64(key_id: str, key_base64: str): + key_bytes = unpaddedbase64.decode_base64(key_base64) + return signedjson.key.decode_verify_key_bytes(key_id, key_bytes) + + +KEY_1 = decode_verify_key_base64( + "ed25519:key1", "fP5l4JzpZPq/zdbBg5xx6lQGAAOM9/3w94cqiJ5jPrw" ) -KEY_2 = signedjson.key.decode_verify_key_base64( - "ed25519", "key2", "Noi6WqcDj0QmPxCNQqgezwTlBKrfqehY1u2FyWP9uYw" +KEY_2 = decode_verify_key_base64( + "ed25519:key2", "Noi6WqcDj0QmPxCNQqgezwTlBKrfqehY1u2FyWP9uYw" ) -- cgit 1.4.1 From b660327056cdced860d532ab2404a26946da7ef5 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Thu, 30 Jan 2020 17:06:38 +0000 Subject: Resync remote device list when detected as stale. (#6786) --- changelog.d/6786.misc | 1 + synapse/handlers/devicemessage.py | 10 ++++++++-- synapse/handlers/federation.py | 18 ++++++++++++++++-- tests/handlers/test_typing.py | 6 +++--- 4 files changed, 28 insertions(+), 7 deletions(-) create mode 100644 changelog.d/6786.misc (limited to 'tests') diff --git a/changelog.d/6786.misc b/changelog.d/6786.misc new file mode 100644 index 0000000000..94c692e53a --- /dev/null +++ b/changelog.d/6786.misc @@ -0,0 +1 @@ +Attempt to resync remote users' devices when detected as stale. diff --git a/synapse/handlers/devicemessage.py b/synapse/handlers/devicemessage.py index 5c5fe77be2..05c4b3eec0 100644 --- a/synapse/handlers/devicemessage.py +++ b/synapse/handlers/devicemessage.py @@ -21,6 +21,7 @@ from canonicaljson import json from twisted.internet import defer from synapse.api.errors import SynapseError +from synapse.logging.context import run_in_background from synapse.logging.opentracing import ( get_active_span_text_map, log_kv, @@ -48,6 +49,8 @@ class DeviceMessageHandler(object): "m.direct_to_device", self.on_direct_to_device_edu ) + self._device_list_updater = hs.get_device_handler().device_list_updater + @defer.inlineCallbacks def on_direct_to_device_edu(self, origin, content): local_messages = {} @@ -134,8 +137,11 @@ class DeviceMessageHandler(object): unknown_devices, ) yield self.store.mark_remote_user_device_cache_as_stale(sender_user_id) - # TODO: Poke something to start trying to refetch user's - # keys. + + # Immediately attempt a resync in the background + run_in_background( + self._device_list_updater.user_device_resync, sender_user_id + ) @defer.inlineCallbacks def send_device_message(self, sender_user_id, message_type, messages): diff --git a/synapse/handlers/federation.py b/synapse/handlers/federation.py index a67020a259..ca484e5458 100644 --- a/synapse/handlers/federation.py +++ b/synapse/handlers/federation.py @@ -57,6 +57,7 @@ from synapse.logging.context import ( run_in_background, ) from synapse.logging.utils import log_function +from synapse.replication.http.devices import ReplicationUserDevicesResyncRestServlet from synapse.replication.http.federation import ( ReplicationCleanRoomRestServlet, ReplicationFederationSendEventsRestServlet, @@ -156,6 +157,13 @@ class FederationHandler(BaseHandler): hs ) + if hs.config.worker_app: + self._user_device_resync = ReplicationUserDevicesResyncRestServlet.make_client( + hs + ) + else: + self._device_list_updater = hs.get_device_handler().device_list_updater + # When joining a room we need to queue any events for that room up self.room_queues = {} self._room_pdu_linearizer = Linearizer("fed_room_pdu") @@ -759,8 +767,14 @@ class FederationHandler(BaseHandler): await self.store.mark_remote_user_device_cache_as_stale( event.sender ) - # TODO: Poke something to start trying to refetch user's - # keys. + + # Immediately attempt a resync in the background + if self.config.worker_app: + return run_in_background(self._user_device_resync, event.sender) + else: + return run_in_background( + self._device_list_updater.user_device_resync, event.sender + ) @log_function async def backfill(self, dest, room_id, limit, extremities): diff --git a/tests/handlers/test_typing.py b/tests/handlers/test_typing.py index 596ddc6970..68b9847bd2 100644 --- a/tests/handlers/test_typing.py +++ b/tests/handlers/test_typing.py @@ -81,6 +81,9 @@ class TypingNotificationsTestCase(unittest.HomeserverTestCase): ] ) + # the tests assume that we are starting at unix time 1000 + reactor.pump((1000,)) + hs = self.setup_test_homeserver( notifier=Mock(), http_client=mock_federation_client, keyring=mock_keyring ) @@ -90,9 +93,6 @@ class TypingNotificationsTestCase(unittest.HomeserverTestCase): return hs def prepare(self, reactor, clock, hs): - # the tests assume that we are starting at unix time 1000 - reactor.pump((1000,)) - mock_notifier = hs.get_notifier() self.on_new_event = mock_notifier.on_new_event -- cgit 1.4.1 From 184303b8650a90256f84bc9801b749a5b81b6d4b Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Thu, 30 Jan 2020 17:20:55 +0000 Subject: MSC2260: Block direct sends of m.room.aliases events (#6794) as per MSC2260 --- changelog.d/6794.feature | 1 + synapse/rest/client/v1/room.py | 12 ++++++++++ tests/rest/admin/test_admin.py | 7 ------ tests/rest/client/v1/test_directory.py | 41 +++++++++++++--------------------- 4 files changed, 28 insertions(+), 33 deletions(-) create mode 100644 changelog.d/6794.feature (limited to 'tests') diff --git a/changelog.d/6794.feature b/changelog.d/6794.feature new file mode 100644 index 0000000000..df9e4b77ab --- /dev/null +++ b/changelog.d/6794.feature @@ -0,0 +1 @@ +Implement updated authorization rules for aliases events, from [MSC2260](https://github.com/matrix-org/matrix-doc/pull/2260). diff --git a/synapse/rest/client/v1/room.py b/synapse/rest/client/v1/room.py index 5aef8238b8..6f31584c51 100644 --- a/synapse/rest/client/v1/room.py +++ b/synapse/rest/client/v1/room.py @@ -184,6 +184,12 @@ class RoomStateEventRestServlet(TransactionRestServlet): content = parse_json_object_from_request(request) + if event_type == EventTypes.Aliases: + # MSC2260 + raise SynapseError( + 400, "Cannot send m.room.aliases events via /rooms/{room_id}/state" + ) + event_dict = { "type": event_type, "content": content, @@ -231,6 +237,12 @@ class RoomSendEventRestServlet(TransactionRestServlet): requester = await self.auth.get_user_by_req(request, allow_guest=True) content = parse_json_object_from_request(request) + if event_type == EventTypes.Aliases: + # MSC2260 + raise SynapseError( + 400, "Cannot send m.room.aliases events via /rooms/{room_id}/send" + ) + event_dict = { "type": event_type, "content": content, diff --git a/tests/rest/admin/test_admin.py b/tests/rest/admin/test_admin.py index 0342aed416..e5984aaad8 100644 --- a/tests/rest/admin/test_admin.py +++ b/tests/rest/admin/test_admin.py @@ -868,13 +868,6 @@ class RoomTestCase(unittest.HomeserverTestCase): self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) # Set this new alias as the canonical alias for this room - self.helper.send_state( - room_id, - "m.room.aliases", - {"aliases": [test_alias]}, - tok=self.admin_user_tok, - state_key="test", - ) self.helper.send_state( room_id, "m.room.canonical_alias", diff --git a/tests/rest/client/v1/test_directory.py b/tests/rest/client/v1/test_directory.py index 633b7dbda0..914cf54927 100644 --- a/tests/rest/client/v1/test_directory.py +++ b/tests/rest/client/v1/test_directory.py @@ -51,26 +51,30 @@ class DirectoryTestCase(unittest.HomeserverTestCase): self.user = self.register_user("user", "test") self.user_tok = self.login("user", "test") - def test_state_event_not_in_room(self): - self.ensure_user_left_room() - self.set_alias_via_state_event(403) + def test_cannot_set_alias_via_state_event(self): + self.ensure_user_joined_room() + url = "/_matrix/client/r0/rooms/%s/state/m.room.aliases/%s" % ( + self.room_id, + self.hs.hostname, + ) + + data = {"aliases": [self.random_alias(5)]} + request_data = json.dumps(data) + + request, channel = self.make_request( + "PUT", url, request_data, access_token=self.user_tok + ) + self.render(request) + self.assertEqual(channel.code, 400, channel.result) def test_directory_endpoint_not_in_room(self): self.ensure_user_left_room() self.set_alias_via_directory(403) - def test_state_event_in_room_too_long(self): - self.ensure_user_joined_room() - self.set_alias_via_state_event(400, alias_length=256) - def test_directory_in_room_too_long(self): self.ensure_user_joined_room() self.set_alias_via_directory(400, alias_length=256) - def test_state_event_in_room(self): - self.ensure_user_joined_room() - self.set_alias_via_state_event(200) - def test_directory_in_room(self): self.ensure_user_joined_room() self.set_alias_via_directory(200) @@ -102,21 +106,6 @@ class DirectoryTestCase(unittest.HomeserverTestCase): self.render(request) self.assertEqual(channel.code, 200, channel.result) - def set_alias_via_state_event(self, expected_code, alias_length=5): - url = "/_matrix/client/r0/rooms/%s/state/m.room.aliases/%s" % ( - self.room_id, - self.hs.hostname, - ) - - data = {"aliases": [self.random_alias(alias_length)]} - request_data = json.dumps(data) - - request, channel = self.make_request( - "PUT", url, request_data, access_token=self.user_tok - ) - self.render(request) - self.assertEqual(channel.code, expected_code, channel.result) - def set_alias_via_directory(self, expected_code, alias_length=5): url = "/_matrix/client/r0/directory/room/%s" % self.random_alias(alias_length) data = {"room_id": self.room_id} -- cgit 1.4.1 From ef6bdafb29abe14cb40f6b83a46e70d82cd3e041 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Wed, 29 Jan 2020 17:55:48 +0000 Subject: Store the room version in EventBuilder --- synapse/events/builder.py | 12 +++++++----- tests/handlers/test_presence.py | 4 ++-- 2 files changed, 9 insertions(+), 7 deletions(-) (limited to 'tests') diff --git a/synapse/events/builder.py b/synapse/events/builder.py index 3997751337..291fb38a26 100644 --- a/synapse/events/builder.py +++ b/synapse/events/builder.py @@ -23,6 +23,7 @@ from synapse.api.room_versions import ( KNOWN_EVENT_FORMAT_VERSIONS, KNOWN_ROOM_VERSIONS, EventFormatVersions, + RoomVersion, ) from synapse.crypto.event_signing import add_hashes_and_signatures from synapse.types import EventID @@ -40,7 +41,7 @@ class EventBuilder(object): content/unsigned/internal_metadata fields are still mutable) Attributes: - format_version (int): Event format version + room_version: Version of the target room room_id (str) type (str) sender (str) @@ -63,7 +64,7 @@ class EventBuilder(object): _hostname = attr.ib() _signing_key = attr.ib() - format_version = attr.ib() + room_version = attr.ib(type=RoomVersion) room_id = attr.ib() type = attr.ib() @@ -108,7 +109,8 @@ class EventBuilder(object): ) auth_ids = yield self._auth.compute_auth_events(self, state_ids) - if self.format_version == EventFormatVersions.V1: + format_version = self.room_version.event_format + if format_version == EventFormatVersions.V1: auth_events = yield self._store.add_event_hashes(auth_ids) prev_events = yield self._store.add_event_hashes(prev_event_ids) else: @@ -148,7 +150,7 @@ class EventBuilder(object): clock=self._clock, hostname=self._hostname, signing_key=self._signing_key, - format_version=self.format_version, + format_version=format_version, event_dict=event_dict, internal_metadata_dict=self.internal_metadata.get_dict(), ) @@ -201,7 +203,7 @@ class EventBuilderFactory(object): clock=self.clock, hostname=self.hostname, signing_key=self.signing_key, - format_version=room_version.event_format, + room_version=room_version, type=key_values["type"], state_key=key_values.get("state_key"), room_id=key_values["room_id"], diff --git a/tests/handlers/test_presence.py b/tests/handlers/test_presence.py index d4293b4312..69914428e2 100644 --- a/tests/handlers/test_presence.py +++ b/tests/handlers/test_presence.py @@ -19,7 +19,7 @@ from mock import Mock, call from signedjson.key import generate_signing_key from synapse.api.constants import EventTypes, Membership, PresenceState -from synapse.events import room_version_to_event_format +from synapse.api.room_versions import KNOWN_ROOM_VERSIONS from synapse.events.builder import EventBuilder from synapse.handlers.presence import ( EXTERNAL_PROCESS_EXPIRY, @@ -597,7 +597,7 @@ class PresenceJoinTestCase(unittest.HomeserverTestCase): clock=self.clock, hostname=hostname, signing_key=self.random_signing_key, - format_version=room_version_to_event_format(room_version), + room_version=KNOWN_ROOM_VERSIONS[room_version], room_id=room_id, type=EventTypes.Member, sender=user_id, -- cgit 1.4.1 From 2a81393a4b905c8bd4c31da04a8b4407462948b9 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Wed, 29 Jan 2020 17:40:33 +0000 Subject: Pass room_version into add_hashes_and_signatures --- synapse/crypto/event_signing.py | 20 +++++++++++++------- synapse/events/builder.py | 2 +- tests/crypto/test_event_signing.py | 9 +++++++-- 3 files changed, 21 insertions(+), 10 deletions(-) (limited to 'tests') diff --git a/synapse/crypto/event_signing.py b/synapse/crypto/event_signing.py index e65bd61d97..1f2bccf700 100644 --- a/synapse/crypto/event_signing.py +++ b/synapse/crypto/event_signing.py @@ -20,10 +20,13 @@ import logging from canonicaljson import encode_canonical_json from signedjson.sign import sign_json +from signedjson.types import SigningKey from unpaddedbase64 import decode_base64, encode_base64 from synapse.api.errors import Codes, SynapseError +from synapse.api.room_versions import RoomVersion from synapse.events.utils import prune_event, prune_event_dict +from synapse.types import JsonDict logger = logging.getLogger(__name__) @@ -137,20 +140,23 @@ def compute_event_signature(event_dict, signature_name, signing_key): def add_hashes_and_signatures( - event_dict, signature_name, signing_key, hash_algorithm=hashlib.sha256 + room_version: RoomVersion, + event_dict: JsonDict, + signature_name: str, + signing_key: SigningKey, ): """Add content hash and sign the event Args: - event_dict (dict): The event to add hashes to and sign - signature_name (str): The name of the entity signing the event + room_version: the version of the room this event is in + + event_dict: The event to add hashes to and sign + signature_name: The name of the entity signing the event (typically the server's hostname). - signing_key (syutil.crypto.SigningKey): The key to sign with - hash_algorithm: A hasher from `hashlib`, e.g. hashlib.sha256, to use - to hash the event + signing_key: The key to sign with """ - name, digest = compute_content_hash(event_dict, hash_algorithm=hash_algorithm) + name, digest = compute_content_hash(event_dict, hash_algorithm=hashlib.sha256) event_dict.setdefault("hashes", {})[name] = encode_base64(digest) diff --git a/synapse/events/builder.py b/synapse/events/builder.py index a26f4c9044..8d63ad6dc3 100644 --- a/synapse/events/builder.py +++ b/synapse/events/builder.py @@ -255,7 +255,7 @@ def create_local_event_from_event_dict( event_dict.setdefault("signatures", {}) - add_hashes_and_signatures(event_dict, hostname, signing_key) + add_hashes_and_signatures(room_version, event_dict, hostname, signing_key) return event_type_from_format_version(format_version)( event_dict, internal_metadata_dict=internal_metadata_dict ) diff --git a/tests/crypto/test_event_signing.py b/tests/crypto/test_event_signing.py index 126e176004..6143a50ab2 100644 --- a/tests/crypto/test_event_signing.py +++ b/tests/crypto/test_event_signing.py @@ -17,6 +17,7 @@ import nacl.signing from unpaddedbase64 import decode_base64 +from synapse.api.room_versions import RoomVersions from synapse.crypto.event_signing import add_hashes_and_signatures from synapse.events import FrozenEvent @@ -49,7 +50,9 @@ class EventSigningTestCase(unittest.TestCase): "unsigned": {"age_ts": 1000000}, } - add_hashes_and_signatures(event_dict, HOSTNAME, self.signing_key) + add_hashes_and_signatures( + RoomVersions.V1, event_dict, HOSTNAME, self.signing_key + ) event = FrozenEvent(event_dict) @@ -81,7 +84,9 @@ class EventSigningTestCase(unittest.TestCase): "unsigned": {"age_ts": 1000000}, } - add_hashes_and_signatures(event_dict, HOSTNAME, self.signing_key) + add_hashes_and_signatures( + RoomVersions.V1, event_dict, HOSTNAME, self.signing_key + ) event = FrozenEvent(event_dict) -- cgit 1.4.1 From d7bf793cc1e3f5268285286341835ac54753eff6 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Fri, 31 Jan 2020 10:06:21 +0000 Subject: s/get_room_version/get_room_version_id/ ... to make way for a forthcoming get_room_version which returns a RoomVersion object. --- synapse/federation/federation_client.py | 10 +++++----- synapse/federation/federation_server.py | 16 ++++++++-------- synapse/handlers/federation.py | 18 +++++++++--------- synapse/handlers/message.py | 8 +++++--- synapse/handlers/pagination.py | 2 +- synapse/handlers/room.py | 2 +- synapse/state/__init__.py | 2 +- synapse/storage/data_stores/main/state.py | 2 +- synapse/storage/persist_events.py | 2 +- tests/handlers/test_presence.py | 2 +- tests/test_state.py | 2 +- tests/unittest.py | 4 +++- 12 files changed, 37 insertions(+), 33 deletions(-) (limited to 'tests') diff --git a/synapse/federation/federation_client.py b/synapse/federation/federation_client.py index d57e8ca7a2..4ac3d81cba 100644 --- a/synapse/federation/federation_client.py +++ b/synapse/federation/federation_client.py @@ -198,7 +198,7 @@ class FederationClient(FederationBase): logger.debug("backfill transaction_data=%r", transaction_data) - room_version = yield self.store.get_room_version(room_id) + room_version = yield self.store.get_room_version_id(room_id) format_ver = room_version_to_event_format(room_version) pdus = [ @@ -336,7 +336,7 @@ class FederationClient(FederationBase): def get_event_auth(self, destination, room_id, event_id): res = yield self.transport_layer.get_event_auth(destination, room_id, event_id) - room_version = yield self.store.get_room_version(room_id) + room_version = yield self.store.get_room_version_id(room_id) format_ver = room_version_to_event_format(room_version) auth_chain = [ @@ -649,7 +649,7 @@ class FederationClient(FederationBase): @defer.inlineCallbacks def send_invite(self, destination, room_id, event_id, pdu): - room_version = yield self.store.get_room_version(room_id) + room_version = yield self.store.get_room_version_id(room_id) content = yield self._do_send_invite(destination, pdu, room_version) @@ -657,7 +657,7 @@ class FederationClient(FederationBase): logger.debug("Got response to send_invite: %s", pdu_dict) - room_version = yield self.store.get_room_version(room_id) + room_version = yield self.store.get_room_version_id(room_id) format_ver = room_version_to_event_format(room_version) pdu = event_from_pdu_json(pdu_dict, format_ver) @@ -859,7 +859,7 @@ class FederationClient(FederationBase): timeout=timeout, ) - room_version = yield self.store.get_room_version(room_id) + room_version = yield self.store.get_room_version_id(room_id) format_ver = room_version_to_event_format(room_version) events = [ diff --git a/synapse/federation/federation_server.py b/synapse/federation/federation_server.py index 9562faa3ee..a4c97ed458 100644 --- a/synapse/federation/federation_server.py +++ b/synapse/federation/federation_server.py @@ -234,7 +234,7 @@ class FederationServer(FederationBase): continue try: - room_version = await self.store.get_room_version(room_id) + room_version = await self.store.get_room_version_id(room_id) except NotFoundError: logger.info("Ignoring PDU for unknown room_id: %s", room_id) continue @@ -334,7 +334,7 @@ class FederationServer(FederationBase): ) ) - room_version = await self.store.get_room_version(room_id) + room_version = await self.store.get_room_version_id(room_id) resp["room_version"] = room_version return 200, resp @@ -385,7 +385,7 @@ class FederationServer(FederationBase): origin_host, _ = parse_server_name(origin) await self.check_server_matches_acl(origin_host, room_id) - room_version = await self.store.get_room_version(room_id) + room_version = await self.store.get_room_version_id(room_id) if room_version not in supported_versions: logger.warning( "Room version %s not in %s", room_version, supported_versions @@ -417,7 +417,7 @@ class FederationServer(FederationBase): async def on_send_join_request(self, origin, content, room_id): logger.debug("on_send_join_request: content: %s", content) - room_version = await self.store.get_room_version(room_id) + room_version = await self.store.get_room_version_id(room_id) format_ver = room_version_to_event_format(room_version) pdu = event_from_pdu_json(content, format_ver) @@ -440,7 +440,7 @@ class FederationServer(FederationBase): await self.check_server_matches_acl(origin_host, room_id) pdu = await self.handler.on_make_leave_request(origin, room_id, user_id) - room_version = await self.store.get_room_version(room_id) + room_version = await self.store.get_room_version_id(room_id) time_now = self._clock.time_msec() return {"event": pdu.get_pdu_json(time_now), "room_version": room_version} @@ -448,7 +448,7 @@ class FederationServer(FederationBase): async def on_send_leave_request(self, origin, content, room_id): logger.debug("on_send_leave_request: content: %s", content) - room_version = await self.store.get_room_version(room_id) + room_version = await self.store.get_room_version_id(room_id) format_ver = room_version_to_event_format(room_version) pdu = event_from_pdu_json(content, format_ver) @@ -495,7 +495,7 @@ class FederationServer(FederationBase): origin_host, _ = parse_server_name(origin) await self.check_server_matches_acl(origin_host, room_id) - room_version = await self.store.get_room_version(room_id) + room_version = await self.store.get_room_version_id(room_id) format_ver = room_version_to_event_format(room_version) auth_chain = [ @@ -664,7 +664,7 @@ class FederationServer(FederationBase): logger.info("Accepting join PDU %s from %s", pdu.event_id, origin) # We've already checked that we know the room version by this point - room_version = await self.store.get_room_version(pdu.room_id) + room_version = await self.store.get_room_version_id(pdu.room_id) # Check signature. try: diff --git a/synapse/handlers/federation.py b/synapse/handlers/federation.py index 01372f6d47..30c720f093 100644 --- a/synapse/handlers/federation.py +++ b/synapse/handlers/federation.py @@ -388,7 +388,7 @@ class FederationHandler(BaseHandler): for x in remote_state: event_map[x.event_id] = x - room_version = await self.store.get_room_version(room_id) + room_version = await self.store.get_room_version_id(room_id) state_map = await resolve_events_with_store( room_id, room_version, @@ -1110,7 +1110,7 @@ class FederationHandler(BaseHandler): Logs a warning if we can't find the given event. """ - room_version = await self.store.get_room_version(room_id) + room_version = await self.store.get_room_version_id(room_id) event_infos = [] @@ -1373,7 +1373,7 @@ class FederationHandler(BaseHandler): event_content = {"membership": Membership.JOIN} - room_version = yield self.store.get_room_version(room_id) + room_version = yield self.store.get_room_version_id(room_id) builder = self.event_builder_factory.new( room_version, @@ -1607,7 +1607,7 @@ class FederationHandler(BaseHandler): ) raise SynapseError(403, "User not from origin", Codes.FORBIDDEN) - room_version = yield self.store.get_room_version(room_id) + room_version = yield self.store.get_room_version_id(room_id) builder = self.event_builder_factory.new( room_version, { @@ -2055,7 +2055,7 @@ class FederationHandler(BaseHandler): do_soft_fail_check = False if do_soft_fail_check: - room_version = yield self.store.get_room_version(event.room_id) + room_version = yield self.store.get_room_version_id(event.room_id) room_version_obj = KNOWN_ROOM_VERSIONS[room_version] # Calculate the "current state". @@ -2191,7 +2191,7 @@ class FederationHandler(BaseHandler): Returns: defer.Deferred[EventContext]: updated context object """ - room_version = yield self.store.get_room_version(event.room_id) + room_version = yield self.store.get_room_version_id(event.room_id) room_version_obj = KNOWN_ROOM_VERSIONS[room_version] try: @@ -2363,7 +2363,7 @@ class FederationHandler(BaseHandler): remote_auth_events.update({(d.type, d.state_key): d for d in different_events}) remote_state = remote_auth_events.values() - room_version = yield self.store.get_room_version(event.room_id) + room_version = yield self.store.get_room_version_id(event.room_id) new_state = yield self.state_handler.resolve_events( room_version, (local_state, remote_state), event ) @@ -2587,7 +2587,7 @@ class FederationHandler(BaseHandler): } if (yield self.auth.check_host_in_room(room_id, self.hs.hostname)): - room_version = yield self.store.get_room_version(room_id) + room_version = yield self.store.get_room_version_id(room_id) builder = self.event_builder_factory.new(room_version, event_dict) EventValidator().validate_builder(builder) @@ -2650,7 +2650,7 @@ class FederationHandler(BaseHandler): Returns: Deferred: resolves (to None) """ - room_version = yield self.store.get_room_version(room_id) + room_version = yield self.store.get_room_version_id(room_id) # NB: event_dict has a particular specced format we might need to fudge # if we change event formats too much. diff --git a/synapse/handlers/message.py b/synapse/handlers/message.py index 9a0f661b9b..bdf16c84d3 100644 --- a/synapse/handlers/message.py +++ b/synapse/handlers/message.py @@ -459,7 +459,9 @@ class EventCreationHandler(object): room_version = event_dict["content"]["room_version"] else: try: - room_version = yield self.store.get_room_version(event_dict["room_id"]) + room_version = yield self.store.get_room_version_id( + event_dict["room_id"] + ) except NotFoundError: raise AuthError(403, "Unknown room") @@ -788,7 +790,7 @@ class EventCreationHandler(object): ): room_version = event.content.get("room_version", RoomVersions.V1.identifier) else: - room_version = yield self.store.get_room_version(event.room_id) + room_version = yield self.store.get_room_version_id(event.room_id) event_allowed = yield self.third_party_event_rules.check_event_allowed( event, context @@ -963,7 +965,7 @@ class EventCreationHandler(object): auth_events = yield self.store.get_events(auth_events_ids) auth_events = {(e.type, e.state_key): e for e in auth_events.values()} - room_version = yield self.store.get_room_version(event.room_id) + room_version = yield self.store.get_room_version_id(event.room_id) room_version_obj = KNOWN_ROOM_VERSIONS[room_version] if event_auth.check_redaction( diff --git a/synapse/handlers/pagination.py b/synapse/handlers/pagination.py index 71d76202c9..caf841a643 100644 --- a/synapse/handlers/pagination.py +++ b/synapse/handlers/pagination.py @@ -281,7 +281,7 @@ class PaginationHandler(object): """Purge the given room from the database""" with (await self.pagination_lock.write(room_id)): # check we know about the room - await self.store.get_room_version(room_id) + await self.store.get_room_version_id(room_id) # first check that we have no users in this room joined = await defer.maybeDeferred( diff --git a/synapse/handlers/room.py b/synapse/handlers/room.py index a95b45d791..1382399557 100644 --- a/synapse/handlers/room.py +++ b/synapse/handlers/room.py @@ -178,7 +178,7 @@ class RoomCreationHandler(BaseHandler): }, token_id=requester.access_token_id, ) - old_room_version = yield self.store.get_room_version(old_room_id) + old_room_version = yield self.store.get_room_version_id(old_room_id) yield self.auth.check_from_context( old_room_version, tombstone_event, tombstone_context ) diff --git a/synapse/state/__init__.py b/synapse/state/__init__.py index cacd0c0c2b..fdd6bef6b4 100644 --- a/synapse/state/__init__.py +++ b/synapse/state/__init__.py @@ -394,7 +394,7 @@ class StateHandler(object): delta_ids=delta_ids, ) - room_version = yield self.store.get_room_version(room_id) + room_version = yield self.store.get_room_version_id(room_id) result = yield self._state_resolution_handler.resolve_state_groups( room_id, diff --git a/synapse/storage/data_stores/main/state.py b/synapse/storage/data_stores/main/state.py index 4167f83c9b..6700942523 100644 --- a/synapse/storage/data_stores/main/state.py +++ b/synapse/storage/data_stores/main/state.py @@ -62,7 +62,7 @@ class StateGroupWorkerStore(EventsWorkerStore, SQLBaseStore): super(StateGroupWorkerStore, self).__init__(database, db_conn, hs) @cached(max_entries=10000) - async def get_room_version(self, room_id: str) -> str: + async def get_room_version_id(self, room_id: str) -> str: """Get the room_version of a given room Raises: diff --git a/synapse/storage/persist_events.py b/synapse/storage/persist_events.py index 86166fd4c1..af3fd67ab9 100644 --- a/synapse/storage/persist_events.py +++ b/synapse/storage/persist_events.py @@ -661,7 +661,7 @@ class EventsPersistenceStorage(object): break if not room_version: - room_version = await self.main_store.get_room_version(room_id) + room_version = await self.main_store.get_room_version_id(room_id) logger.debug("calling resolve_state_groups from preserve_events") res = await self._state_resolution_handler.resolve_state_groups( diff --git a/tests/handlers/test_presence.py b/tests/handlers/test_presence.py index d4293b4312..e92e090c3c 100644 --- a/tests/handlers/test_presence.py +++ b/tests/handlers/test_presence.py @@ -588,7 +588,7 @@ class PresenceJoinTestCase(unittest.HomeserverTestCase): hostname = get_domain_from_id(user_id) - room_version = self.get_success(self.store.get_room_version(room_id)) + room_version = self.get_success(self.store.get_room_version_id(room_id)) builder = EventBuilder( state=self.state, diff --git a/tests/test_state.py b/tests/test_state.py index e0aae06be4..1e4449fa1c 100644 --- a/tests/test_state.py +++ b/tests/test_state.py @@ -119,7 +119,7 @@ class StateGroupStore(object): def register_event_id_state_group(self, event_id, state_group): self._event_to_state_group[event_id] = state_group - def get_room_version(self, room_id): + def get_room_version_id(self, room_id): return RoomVersions.V1.identifier diff --git a/tests/unittest.py b/tests/unittest.py index b56e249386..98bf27d39c 100644 --- a/tests/unittest.py +++ b/tests/unittest.py @@ -589,7 +589,9 @@ class HomeserverTestCase(TestCase): event_builder_factory = self.hs.get_event_builder_factory() event_creation_handler = self.hs.get_event_creation_handler() - room_version = self.get_success(self.hs.get_datastore().get_room_version(room)) + room_version = self.get_success( + self.hs.get_datastore().get_room_version_id(room) + ) builder = event_builder_factory.for_room_version( KNOWN_ROOM_VERSIONS[room_version], -- cgit 1.4.1 From b9391c957572224c3a7c22870102fcbd24dea4e0 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Mon, 3 Feb 2020 18:05:44 +0000 Subject: Add typing to SyncHandler (#6821) Co-authored-by: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> --- changelog.d/6821.misc | 1 + synapse/events/__init__.py | 18 +- synapse/handlers/sync.py | 705 +++++++++++++++++++++------------------- tests/storage/test_redaction.py | 5 +- tox.ini | 1 + 5 files changed, 381 insertions(+), 349 deletions(-) create mode 100644 changelog.d/6821.misc (limited to 'tests') diff --git a/changelog.d/6821.misc b/changelog.d/6821.misc new file mode 100644 index 0000000000..1d5265d5e2 --- /dev/null +++ b/changelog.d/6821.misc @@ -0,0 +1 @@ +Add type hints to `SyncHandler`. diff --git a/synapse/events/__init__.py b/synapse/events/__init__.py index f813fa2fe7..92f76703b3 100644 --- a/synapse/events/__init__.py +++ b/synapse/events/__init__.py @@ -189,8 +189,14 @@ class EventBase(object): redacts = _event_dict_property("redacts", None) room_id = _event_dict_property("room_id") sender = _event_dict_property("sender") + state_key = _event_dict_property("state_key") + type = _event_dict_property("type") user_id = _event_dict_property("sender") + @property + def event_id(self) -> str: + raise NotImplementedError() + @property def membership(self): return self.content["membership"] @@ -281,10 +287,7 @@ class FrozenEvent(EventBase): else: frozen_dict = event_dict - self.event_id = event_dict["event_id"] - self.type = event_dict["type"] - if "state_key" in event_dict: - self.state_key = event_dict["state_key"] + self._event_id = event_dict["event_id"] super(FrozenEvent, self).__init__( frozen_dict, @@ -294,6 +297,10 @@ class FrozenEvent(EventBase): rejected_reason=rejected_reason, ) + @property + def event_id(self) -> str: + return self._event_id + def __str__(self): return self.__repr__() @@ -332,9 +339,6 @@ class FrozenEventV2(EventBase): frozen_dict = event_dict self._event_id = None - self.type = event_dict["type"] - if "state_key" in event_dict: - self.state_key = event_dict["state_key"] super(FrozenEventV2, self).__init__( frozen_dict, diff --git a/synapse/handlers/sync.py b/synapse/handlers/sync.py index cd95f85e3f..5f060241b4 100644 --- a/synapse/handlers/sync.py +++ b/synapse/handlers/sync.py @@ -14,20 +14,30 @@ # See the License for the specific language governing permissions and # limitations under the License. -import collections import itertools import logging +from typing import Any, Dict, FrozenSet, List, Optional, Set, Tuple from six import iteritems, itervalues +import attr from prometheus_client import Counter from synapse.api.constants import EventTypes, Membership +from synapse.api.filtering import FilterCollection +from synapse.events import EventBase from synapse.logging.context import LoggingContext from synapse.push.clientformat import format_push_rules_for_user from synapse.storage.roommember import MemberSummary from synapse.storage.state import StateFilter -from synapse.types import RoomStreamToken +from synapse.types import ( + Collection, + JsonDict, + RoomStreamToken, + StateMap, + StreamToken, + UserID, +) from synapse.util.async_helpers import concurrently_execute from synapse.util.caches.expiringcache import ExpiringCache from synapse.util.caches.lrucache import LruCache @@ -62,17 +72,22 @@ LAZY_LOADED_MEMBERS_CACHE_MAX_AGE = 30 * 60 * 1000 LAZY_LOADED_MEMBERS_CACHE_MAX_SIZE = 100 -SyncConfig = collections.namedtuple( - "SyncConfig", ["user", "filter_collection", "is_guest", "request_key", "device_id"] -) +@attr.s(slots=True, frozen=True) +class SyncConfig: + user = attr.ib(type=UserID) + filter_collection = attr.ib(type=FilterCollection) + is_guest = attr.ib(type=bool) + request_key = attr.ib(type=Tuple[Any, ...]) + device_id = attr.ib(type=str) -class TimelineBatch( - collections.namedtuple("TimelineBatch", ["prev_batch", "events", "limited"]) -): - __slots__ = [] +@attr.s(slots=True, frozen=True) +class TimelineBatch: + prev_batch = attr.ib(type=StreamToken) + events = attr.ib(type=List[EventBase]) + limited = attr.ib(bool) - def __nonzero__(self): + def __nonzero__(self) -> bool: """Make the result appear empty if there are no updates. This is used to tell if room needs to be part of the sync result. """ @@ -81,23 +96,17 @@ class TimelineBatch( __bool__ = __nonzero__ # python3 -class JoinedSyncResult( - collections.namedtuple( - "JoinedSyncResult", - [ - "room_id", # str - "timeline", # TimelineBatch - "state", # dict[(str, str), FrozenEvent] - "ephemeral", - "account_data", - "unread_notifications", - "summary", - ], - ) -): - __slots__ = [] - - def __nonzero__(self): +@attr.s(slots=True, frozen=True) +class JoinedSyncResult: + room_id = attr.ib(type=str) + timeline = attr.ib(type=TimelineBatch) + state = attr.ib(type=StateMap[EventBase]) + ephemeral = attr.ib(type=List[JsonDict]) + account_data = attr.ib(type=List[JsonDict]) + unread_notifications = attr.ib(type=JsonDict) + summary = attr.ib(type=Optional[JsonDict]) + + def __nonzero__(self) -> bool: """Make the result appear empty if there are no updates. This is used to tell if room needs to be part of the sync result. """ @@ -113,20 +122,14 @@ class JoinedSyncResult( __bool__ = __nonzero__ # python3 -class ArchivedSyncResult( - collections.namedtuple( - "ArchivedSyncResult", - [ - "room_id", # str - "timeline", # TimelineBatch - "state", # dict[(str, str), FrozenEvent] - "account_data", - ], - ) -): - __slots__ = [] - - def __nonzero__(self): +@attr.s(slots=True, frozen=True) +class ArchivedSyncResult: + room_id = attr.ib(type=str) + timeline = attr.ib(type=TimelineBatch) + state = attr.ib(type=StateMap[EventBase]) + account_data = attr.ib(type=List[JsonDict]) + + def __nonzero__(self) -> bool: """Make the result appear empty if there are no updates. This is used to tell if room needs to be part of the sync result. """ @@ -135,70 +138,88 @@ class ArchivedSyncResult( __bool__ = __nonzero__ # python3 -class InvitedSyncResult( - collections.namedtuple( - "InvitedSyncResult", - ["room_id", "invite"], # str # FrozenEvent: the invite event - ) -): - __slots__ = [] +@attr.s(slots=True, frozen=True) +class InvitedSyncResult: + room_id = attr.ib(type=str) + invite = attr.ib(type=EventBase) - def __nonzero__(self): + def __nonzero__(self) -> bool: """Invited rooms should always be reported to the client""" return True __bool__ = __nonzero__ # python3 -class GroupsSyncResult( - collections.namedtuple("GroupsSyncResult", ["join", "invite", "leave"]) -): - __slots__ = [] +@attr.s(slots=True, frozen=True) +class GroupsSyncResult: + join = attr.ib(type=JsonDict) + invite = attr.ib(type=JsonDict) + leave = attr.ib(type=JsonDict) - def __nonzero__(self): + def __nonzero__(self) -> bool: return bool(self.join or self.invite or self.leave) __bool__ = __nonzero__ # python3 -class DeviceLists( - collections.namedtuple( - "DeviceLists", - [ - "changed", # list of user_ids whose devices may have changed - "left", # list of user_ids whose devices we no longer track - ], - ) -): - __slots__ = [] +@attr.s(slots=True, frozen=True) +class DeviceLists: + """ + Attributes: + changed: List of user_ids whose devices may have changed + left: List of user_ids whose devices we no longer track + """ + + changed = attr.ib(type=Collection[str]) + left = attr.ib(type=Collection[str]) - def __nonzero__(self): + def __nonzero__(self) -> bool: return bool(self.changed or self.left) __bool__ = __nonzero__ # python3 -class SyncResult( - collections.namedtuple( - "SyncResult", - [ - "next_batch", # Token for the next sync - "presence", # List of presence events for the user. - "account_data", # List of account_data events for the user. - "joined", # JoinedSyncResult for each joined room. - "invited", # InvitedSyncResult for each invited room. - "archived", # ArchivedSyncResult for each archived room. - "to_device", # List of direct messages for the device. - "device_lists", # List of user_ids whose devices have changed - "device_one_time_keys_count", # Dict of algorithm to count for one time keys - # for this device - "groups", - ], - ) -): - __slots__ = [] - - def __nonzero__(self): +@attr.s +class _RoomChanges: + """The set of room entries to include in the sync, plus the set of joined + and left room IDs since last sync. + """ + + room_entries = attr.ib(type=List["RoomSyncResultBuilder"]) + invited = attr.ib(type=List[InvitedSyncResult]) + newly_joined_rooms = attr.ib(type=List[str]) + newly_left_rooms = attr.ib(type=List[str]) + + +@attr.s(slots=True, frozen=True) +class SyncResult: + """ + Attributes: + next_batch: Token for the next sync + presence: List of presence events for the user. + account_data: List of account_data events for the user. + joined: JoinedSyncResult for each joined room. + invited: InvitedSyncResult for each invited room. + archived: ArchivedSyncResult for each archived room. + to_device: List of direct messages for the device. + device_lists: List of user_ids whose devices have changed + device_one_time_keys_count: Dict of algorithm to count for one time keys + for this device + groups: Group updates, if any + """ + + next_batch = attr.ib(type=StreamToken) + presence = attr.ib(type=List[JsonDict]) + account_data = attr.ib(type=List[JsonDict]) + joined = attr.ib(type=List[JoinedSyncResult]) + invited = attr.ib(type=List[InvitedSyncResult]) + archived = attr.ib(type=List[ArchivedSyncResult]) + to_device = attr.ib(type=List[JsonDict]) + device_lists = attr.ib(type=DeviceLists) + device_one_time_keys_count = attr.ib(type=JsonDict) + groups = attr.ib(type=Optional[GroupsSyncResult]) + + def __nonzero__(self) -> bool: """Make the result appear empty if there are no updates. This is used to tell if the notifier needs to wait for more events when polling for events. @@ -240,13 +261,15 @@ class SyncHandler(object): ) async def wait_for_sync_for_user( - self, sync_config, since_token=None, timeout=0, full_state=False - ): + self, + sync_config: SyncConfig, + since_token: Optional[StreamToken] = None, + timeout: int = 0, + full_state: bool = False, + ) -> SyncResult: """Get the sync for a client if we have new data for it now. Otherwise wait for new data to arrive on the server. If the timeout expires, then return an empty sync result. - Returns: - Deferred[SyncResult] """ # If the user is not part of the mau group, then check that limits have # not been exceeded (if not part of the group by this point, almost certain @@ -265,8 +288,12 @@ class SyncHandler(object): return res async def _wait_for_sync_for_user( - self, sync_config, since_token, timeout, full_state - ): + self, + sync_config: SyncConfig, + since_token: Optional[StreamToken] = None, + timeout: int = 0, + full_state: bool = False, + ) -> SyncResult: if since_token is None: sync_type = "initial_sync" elif full_state: @@ -305,25 +332,33 @@ class SyncHandler(object): return result - def current_sync_for_user(self, sync_config, since_token=None, full_state=False): + async def current_sync_for_user( + self, + sync_config: SyncConfig, + since_token: Optional[StreamToken] = None, + full_state: bool = False, + ) -> SyncResult: """Get the sync for client needed to match what the server has now. - Returns: - A Deferred SyncResult. """ - return self.generate_sync_result(sync_config, since_token, full_state) + return await self.generate_sync_result(sync_config, since_token, full_state) - async def push_rules_for_user(self, user): + async def push_rules_for_user(self, user: UserID) -> JsonDict: user_id = user.to_string() rules = await self.store.get_push_rules_for_user(user_id) rules = format_push_rules_for_user(user, rules) return rules - async def ephemeral_by_room(self, sync_result_builder, now_token, since_token=None): + async def ephemeral_by_room( + self, + sync_result_builder: "SyncResultBuilder", + now_token: StreamToken, + since_token: Optional[StreamToken] = None, + ) -> Tuple[StreamToken, Dict[str, List[JsonDict]]]: """Get the ephemeral events for each room the user is in Args: - sync_result_builder(SyncResultBuilder) - now_token (StreamToken): Where the server is currently up to. - since_token (StreamToken): Where the server was when the client + sync_result_builder + now_token: Where the server is currently up to. + since_token: Where the server was when the client last synced. Returns: A tuple of the now StreamToken, updated to reflect the which typing @@ -348,7 +383,7 @@ class SyncHandler(object): ) now_token = now_token.copy_and_replace("typing_key", typing_key) - ephemeral_by_room = {} + ephemeral_by_room = {} # type: JsonDict for event in typing: # we want to exclude the room_id from the event, but modifying the @@ -380,13 +415,13 @@ class SyncHandler(object): async def _load_filtered_recents( self, - room_id, - sync_config, - now_token, - since_token=None, - recents=None, - newly_joined_room=False, - ): + room_id: str, + sync_config: SyncConfig, + now_token: StreamToken, + since_token: Optional[StreamToken] = None, + potential_recents: Optional[List[EventBase]] = None, + newly_joined_room: bool = False, + ) -> TimelineBatch: """ Returns: a Deferred TimelineBatch @@ -397,21 +432,29 @@ class SyncHandler(object): sync_config.filter_collection.blocks_all_room_timeline() ) - if recents is None or newly_joined_room or timeline_limit < len(recents): + if ( + potential_recents is None + or newly_joined_room + or timeline_limit < len(potential_recents) + ): limited = True else: limited = False - if recents: - recents = sync_config.filter_collection.filter_room_timeline(recents) + if potential_recents: + recents = sync_config.filter_collection.filter_room_timeline( + potential_recents + ) # We check if there are any state events, if there are then we pass # all current state events to the filter_events function. This is to # ensure that we always include current state in the timeline - current_state_ids = frozenset() + current_state_ids = frozenset() # type: FrozenSet[str] if any(e.is_state() for e in recents): - current_state_ids = await self.state.get_current_state_ids(room_id) - current_state_ids = frozenset(itervalues(current_state_ids)) + current_state_ids_map = await self.state.get_current_state_ids( + room_id + ) + current_state_ids = frozenset(itervalues(current_state_ids_map)) recents = await filter_events_for_client( self.storage, @@ -463,8 +506,10 @@ class SyncHandler(object): # ensure that we always include current state in the timeline current_state_ids = frozenset() if any(e.is_state() for e in loaded_recents): - current_state_ids = await self.state.get_current_state_ids(room_id) - current_state_ids = frozenset(itervalues(current_state_ids)) + current_state_ids_map = await self.state.get_current_state_ids( + room_id + ) + current_state_ids = frozenset(itervalues(current_state_ids_map)) loaded_recents = await filter_events_for_client( self.storage, @@ -493,17 +538,15 @@ class SyncHandler(object): limited=limited or newly_joined_room, ) - async def get_state_after_event(self, event, state_filter=StateFilter.all()): + async def get_state_after_event( + self, event: EventBase, state_filter: StateFilter = StateFilter.all() + ) -> StateMap[str]: """ Get the room state after the given event Args: - event(synapse.events.EventBase): event of interest - state_filter (StateFilter): The state filter used to fetch state - from the database. - - Returns: - A Deferred map from ((type, state_key)->Event) + event: event of interest + state_filter: The state filter used to fetch state from the database. """ state_ids = await self.state_store.get_state_ids_for_event( event.event_id, state_filter=state_filter @@ -514,18 +557,17 @@ class SyncHandler(object): return state_ids async def get_state_at( - self, room_id, stream_position, state_filter=StateFilter.all() - ): + self, + room_id: str, + stream_position: StreamToken, + state_filter: StateFilter = StateFilter.all(), + ) -> StateMap[str]: """ Get the room state at a particular stream position Args: - room_id(str): room for which to get state - stream_position(StreamToken): point at which to get state - state_filter (StateFilter): The state filter used to fetch state - from the database. - - Returns: - A Deferred map from ((type, state_key)->Event) + room_id: room for which to get state + stream_position: point at which to get state + state_filter: The state filter used to fetch state from the database. """ # FIXME this claims to get the state at a stream position, but # get_recent_events_for_room operates by topo ordering. This therefore @@ -546,23 +588,25 @@ class SyncHandler(object): state = {} return state - async def compute_summary(self, room_id, sync_config, batch, state, now_token): + async def compute_summary( + self, + room_id: str, + sync_config: SyncConfig, + batch: TimelineBatch, + state: StateMap[EventBase], + now_token: StreamToken, + ) -> Optional[JsonDict]: """ Works out a room summary block for this room, summarising the number of joined members in the room, and providing the 'hero' members if the room has no name so clients can consistently name rooms. Also adds state events to 'state' if needed to describe the heroes. - Args: - room_id(str): - sync_config(synapse.handlers.sync.SyncConfig): - batch(synapse.handlers.sync.TimelineBatch): The timeline batch for - the room that will be sent to the user. - state(dict): dict of (type, state_key) -> Event as returned by - compute_state_delta - now_token(str): Token of the end of the current batch. - - Returns: - A deferred dict describing the room summary + Args + room_id + sync_config + batch: The timeline batch for the room that will be sent to the user. + state: State as returned by compute_state_delta + now_token: Token of the end of the current batch. """ # FIXME: we could/should get this from room_stats when matthew/stats lands @@ -681,7 +725,7 @@ class SyncHandler(object): return summary - def get_lazy_loaded_members_cache(self, cache_key): + def get_lazy_loaded_members_cache(self, cache_key: Tuple[str, str]) -> LruCache: cache = self.lazy_loaded_members_cache.get(cache_key) if cache is None: logger.debug("creating LruCache for %r", cache_key) @@ -692,23 +736,24 @@ class SyncHandler(object): return cache async def compute_state_delta( - self, room_id, batch, sync_config, since_token, now_token, full_state - ): + self, + room_id: str, + batch: TimelineBatch, + sync_config: SyncConfig, + since_token: Optional[StreamToken], + now_token: StreamToken, + full_state: bool, + ) -> StateMap[EventBase]: """ Works out the difference in state between the start of the timeline and the previous sync. Args: - room_id(str): - batch(synapse.handlers.sync.TimelineBatch): The timeline batch for - the room that will be sent to the user. - sync_config(synapse.handlers.sync.SyncConfig): - since_token(str|None): Token of the end of the previous batch. May - be None. - now_token(str): Token of the end of the current batch. - full_state(bool): Whether to force returning the full state. - - Returns: - A deferred dict of (type, state_key) -> Event + room_id: + batch: The timeline batch for the room that will be sent to the user. + sync_config: + since_token: Token of the end of the previous batch. May be None. + now_token: Token of the end of the current batch. + full_state: Whether to force returning the full state. """ # TODO(mjark) Check if the state events were received by the server # after the previous sync, since we need to include those state @@ -800,6 +845,10 @@ class SyncHandler(object): # about them). state_filter = StateFilter.all() + # If this is an initial sync then full_state should be set, and + # that case is handled above. We assert here to ensure that this + # is indeed the case. + assert since_token is not None state_at_previous_sync = await self.get_state_at( room_id, stream_position=since_token, state_filter=state_filter ) @@ -874,7 +923,7 @@ class SyncHandler(object): if t[0] == EventTypes.Member: cache.set(t[1], event_id) - state = {} + state = {} # type: Dict[str, EventBase] if state_ids: state = await self.store.get_events(list(state_ids.values())) @@ -885,7 +934,9 @@ class SyncHandler(object): ) } - async def unread_notifs_for_room_id(self, room_id, sync_config): + async def unread_notifs_for_room_id( + self, room_id: str, sync_config: SyncConfig + ) -> Optional[Dict[str, str]]: with Measure(self.clock, "unread_notifs_for_room_id"): last_unread_event_id = await self.store.get_last_receipt_event_id_for_user( user_id=sync_config.user.to_string(), @@ -893,7 +944,6 @@ class SyncHandler(object): receipt_type="m.read", ) - notifs = [] if last_unread_event_id: notifs = await self.store.get_unread_event_push_actions_by_room_for_user( room_id, sync_config.user.to_string(), last_unread_event_id @@ -905,17 +955,12 @@ class SyncHandler(object): return None async def generate_sync_result( - self, sync_config, since_token=None, full_state=False - ): + self, + sync_config: SyncConfig, + since_token: Optional[StreamToken] = None, + full_state: bool = False, + ) -> SyncResult: """Generates a sync result. - - Args: - sync_config (SyncConfig) - since_token (StreamToken) - full_state (bool) - - Returns: - Deferred(SyncResult) """ # NB: The now_token gets changed by some of the generate_sync_* methods, # this is due to some of the underlying streams not supporting the ability @@ -977,7 +1022,7 @@ class SyncHandler(object): ) device_id = sync_config.device_id - one_time_key_counts = {} + one_time_key_counts = {} # type: JsonDict if device_id: one_time_key_counts = await self.store.count_e2e_one_time_keys( user_id, device_id @@ -1007,7 +1052,9 @@ class SyncHandler(object): ) @measure_func("_generate_sync_entry_for_groups") - async def _generate_sync_entry_for_groups(self, sync_result_builder): + async def _generate_sync_entry_for_groups( + self, sync_result_builder: "SyncResultBuilder" + ) -> None: user_id = sync_result_builder.sync_config.user.to_string() since_token = sync_result_builder.since_token now_token = sync_result_builder.now_token @@ -1052,27 +1099,22 @@ class SyncHandler(object): @measure_func("_generate_sync_entry_for_device_list") async def _generate_sync_entry_for_device_list( self, - sync_result_builder, - newly_joined_rooms, - newly_joined_or_invited_users, - newly_left_rooms, - newly_left_users, - ): + sync_result_builder: "SyncResultBuilder", + newly_joined_rooms: Set[str], + newly_joined_or_invited_users: Set[str], + newly_left_rooms: Set[str], + newly_left_users: Set[str], + ) -> DeviceLists: """Generate the DeviceLists section of sync Args: - sync_result_builder (SyncResultBuilder) - newly_joined_rooms (set[str]): Set of rooms user has joined since - previous sync - newly_joined_or_invited_users (set[str]): Set of users that have - joined or been invited to a room since previous sync. - newly_left_rooms (set[str]): Set of rooms user has left since + sync_result_builder + newly_joined_rooms: Set of rooms user has joined since previous sync + newly_joined_or_invited_users: Set of users that have joined or + been invited to a room since previous sync. + newly_left_rooms: Set of rooms user has left since previous sync + newly_left_users: Set of users that have left a room we're in since previous sync - newly_left_users (set[str]): Set of users that have left a room - we're in since previous sync - - Returns: - Deferred[DeviceLists] """ user_id = sync_result_builder.sync_config.user.to_string() @@ -1133,15 +1175,11 @@ class SyncHandler(object): else: return DeviceLists(changed=[], left=[]) - async def _generate_sync_entry_for_to_device(self, sync_result_builder): + async def _generate_sync_entry_for_to_device( + self, sync_result_builder: "SyncResultBuilder" + ) -> None: """Generates the portion of the sync response. Populates `sync_result_builder` with the result. - - Args: - sync_result_builder(SyncResultBuilder) - - Returns: - Deferred(dict): A dictionary containing the per room account data. """ user_id = sync_result_builder.sync_config.user.to_string() device_id = sync_result_builder.sync_config.device_id @@ -1179,15 +1217,17 @@ class SyncHandler(object): else: sync_result_builder.to_device = [] - async def _generate_sync_entry_for_account_data(self, sync_result_builder): + async def _generate_sync_entry_for_account_data( + self, sync_result_builder: "SyncResultBuilder" + ) -> Dict[str, Dict[str, JsonDict]]: """Generates the account data portion of the sync response. Populates `sync_result_builder` with the result. Args: - sync_result_builder(SyncResultBuilder) + sync_result_builder Returns: - Deferred(dict): A dictionary containing the per room account data. + A dictionary containing the per room account data. """ sync_config = sync_result_builder.sync_config user_id = sync_result_builder.sync_config.user.to_string() @@ -1231,18 +1271,21 @@ class SyncHandler(object): return account_data_by_room async def _generate_sync_entry_for_presence( - self, sync_result_builder, newly_joined_rooms, newly_joined_or_invited_users - ): + self, + sync_result_builder: "SyncResultBuilder", + newly_joined_rooms: Set[str], + newly_joined_or_invited_users: Set[str], + ) -> None: """Generates the presence portion of the sync response. Populates the `sync_result_builder` with the result. Args: - sync_result_builder(SyncResultBuilder) - newly_joined_rooms(list): List of rooms that the user has joined - since the last sync (or empty if an initial sync) - newly_joined_or_invited_users(list): List of users that have joined - or been invited to rooms since the last sync (or empty if an initial - sync) + sync_result_builder + newly_joined_rooms: Set of rooms that the user has joined since + the last sync (or empty if an initial sync) + newly_joined_or_invited_users: Set of users that have joined or + been invited to rooms since the last sync (or empty if an + initial sync) """ now_token = sync_result_builder.now_token sync_config = sync_result_builder.sync_config @@ -1286,17 +1329,19 @@ class SyncHandler(object): sync_result_builder.presence = presence async def _generate_sync_entry_for_rooms( - self, sync_result_builder, account_data_by_room - ): + self, + sync_result_builder: "SyncResultBuilder", + account_data_by_room: Dict[str, Dict[str, JsonDict]], + ) -> Tuple[Set[str], Set[str], Set[str], Set[str]]: """Generates the rooms portion of the sync response. Populates the `sync_result_builder` with the result. Args: - sync_result_builder(SyncResultBuilder) - account_data_by_room(dict): Dictionary of per room account data + sync_result_builder + account_data_by_room: Dictionary of per room account data Returns: - Deferred(tuple): Returns a 4-tuple of + Returns a 4-tuple of `(newly_joined_rooms, newly_joined_or_invited_users, newly_left_rooms, newly_left_users)` """ @@ -1307,7 +1352,7 @@ class SyncHandler(object): ) if block_all_room_ephemeral: - ephemeral_by_room = {} + ephemeral_by_room = {} # type: Dict[str, List[JsonDict]] else: now_token, ephemeral_by_room = await self.ephemeral_by_room( sync_result_builder, @@ -1328,7 +1373,7 @@ class SyncHandler(object): ) if not tags_by_room: logger.debug("no-oping sync") - return [], [], [], [] + return set(), set(), set(), set() ignored_account_data = await self.store.get_global_account_data_by_type_for_user( "m.ignored_user_list", user_id=user_id @@ -1340,19 +1385,22 @@ class SyncHandler(object): ignored_users = frozenset() if since_token: - res = await self._get_rooms_changed(sync_result_builder, ignored_users) - room_entries, invited, newly_joined_rooms, newly_left_rooms = res - + room_changes = await self._get_rooms_changed( + sync_result_builder, ignored_users + ) tags_by_room = await self.store.get_updated_tags( user_id, since_token.account_data_key ) else: - res = await self._get_all_rooms(sync_result_builder, ignored_users) - room_entries, invited, newly_joined_rooms = res - newly_left_rooms = [] + room_changes = await self._get_all_rooms(sync_result_builder, ignored_users) tags_by_room = await self.store.get_tags_for_user(user_id) + room_entries = room_changes.room_entries + invited = room_changes.invited + newly_joined_rooms = room_changes.newly_joined_rooms + newly_left_rooms = room_changes.newly_left_rooms + def handle_room_entries(room_entry): return self._generate_room_entry( sync_result_builder, @@ -1392,13 +1440,15 @@ class SyncHandler(object): newly_left_users -= newly_joined_or_invited_users return ( - newly_joined_rooms, + set(newly_joined_rooms), newly_joined_or_invited_users, - newly_left_rooms, + set(newly_left_rooms), newly_left_users, ) - async def _have_rooms_changed(self, sync_result_builder): + async def _have_rooms_changed( + self, sync_result_builder: "SyncResultBuilder" + ) -> bool: """Returns whether there may be any new events that should be sent down the sync. Returns True if there are. """ @@ -1422,22 +1472,10 @@ class SyncHandler(object): return True return False - async def _get_rooms_changed(self, sync_result_builder, ignored_users): + async def _get_rooms_changed( + self, sync_result_builder: "SyncResultBuilder", ignored_users: Set[str] + ) -> _RoomChanges: """Gets the the changes that have happened since the last sync. - - Args: - sync_result_builder(SyncResultBuilder) - ignored_users(set(str)): Set of users ignored by user. - - Returns: - Deferred(tuple): Returns a tuple of the form: - `(room_entries, invited_rooms, newly_joined_rooms, newly_left_rooms)` - - where: - room_entries is a list [RoomSyncResultBuilder] - invited_rooms is a list [InvitedSyncResult] - newly_joined_rooms is a list[str] of room ids - newly_left_rooms is a list[str] of room ids """ user_id = sync_result_builder.sync_config.user.to_string() since_token = sync_result_builder.since_token @@ -1451,7 +1489,7 @@ class SyncHandler(object): user_id, since_token.room_key, now_token.room_key ) - mem_change_events_by_room_id = {} + mem_change_events_by_room_id = {} # type: Dict[str, List[EventBase]] for event in rooms_changed: mem_change_events_by_room_id.setdefault(event.room_id, []).append(event) @@ -1570,7 +1608,7 @@ class SyncHandler(object): # This is all screaming out for a refactor, as the logic here is # subtle and the moving parts numerous. if leave_event.internal_metadata.is_out_of_band_membership(): - batch_events = [leave_event] + batch_events = [leave_event] # type: Optional[List[EventBase]] else: batch_events = None @@ -1636,18 +1674,17 @@ class SyncHandler(object): ) room_entries.append(entry) - return room_entries, invited, newly_joined_rooms, newly_left_rooms + return _RoomChanges(room_entries, invited, newly_joined_rooms, newly_left_rooms) - async def _get_all_rooms(self, sync_result_builder, ignored_users): + async def _get_all_rooms( + self, sync_result_builder: "SyncResultBuilder", ignored_users: Set[str] + ) -> _RoomChanges: """Returns entries for all rooms for the user. Args: - sync_result_builder(SyncResultBuilder) - ignored_users(set(str)): Set of users ignored by user. + sync_result_builder + ignored_users: Set of users ignored by user. - Returns: - Deferred(tuple): Returns a tuple of the form: - `([RoomSyncResultBuilder], [InvitedSyncResult], [])` """ user_id = sync_result_builder.sync_config.user.to_string() @@ -1709,30 +1746,30 @@ class SyncHandler(object): ) ) - return room_entries, invited, [] + return _RoomChanges(room_entries, invited, [], []) async def _generate_room_entry( self, - sync_result_builder, - ignored_users, - room_builder, - ephemeral, - tags, - account_data, - always_include=False, + sync_result_builder: "SyncResultBuilder", + ignored_users: Set[str], + room_builder: "RoomSyncResultBuilder", + ephemeral: List[JsonDict], + tags: Optional[List[JsonDict]], + account_data: Dict[str, JsonDict], + always_include: bool = False, ): """Populates the `joined` and `archived` section of `sync_result_builder` based on the `room_builder`. Args: - sync_result_builder(SyncResultBuilder) - ignored_users(set(str)): Set of users ignored by user. - room_builder(RoomSyncResultBuilder) - ephemeral(list): List of new ephemeral events for room - tags(list): List of *all* tags for room, or None if there has been + sync_result_builder + ignored_users: Set of users ignored by user. + room_builder + ephemeral: List of new ephemeral events for room + tags: List of *all* tags for room, or None if there has been no change. - account_data(list): List of new account data for room - always_include(bool): Always include this room in the sync response, + account_data: List of new account data for room + always_include: Always include this room in the sync response, even if empty. """ newly_joined = room_builder.newly_joined @@ -1758,7 +1795,7 @@ class SyncHandler(object): sync_config, now_token=upto_token, since_token=since_token, - recents=events, + potential_recents=events, newly_joined_room=newly_joined, ) @@ -1809,7 +1846,7 @@ class SyncHandler(object): room_id, batch, sync_config, since_token, now_token, full_state=full_state ) - summary = {} + summary = {} # type: Optional[JsonDict] # we include a summary in room responses when we're lazy loading # members (as the client otherwise doesn't have enough info to form @@ -1833,7 +1870,7 @@ class SyncHandler(object): ) if room_builder.rtype == "joined": - unread_notifications = {} + unread_notifications = {} # type: Dict[str, str] room_sync = JoinedSyncResult( room_id=room_id, timeline=batch, @@ -1860,18 +1897,20 @@ class SyncHandler(object): % (room_id, user_id, len(state)) ) elif room_builder.rtype == "archived": - room_sync = ArchivedSyncResult( + archived_room_sync = ArchivedSyncResult( room_id=room_id, timeline=batch, state=state, account_data=account_data_events, ) - if room_sync or always_include: - sync_result_builder.archived.append(room_sync) + if archived_room_sync or always_include: + sync_result_builder.archived.append(archived_room_sync) else: raise Exception("Unrecognized rtype: %r", room_builder.rtype) - async def get_rooms_for_user_at(self, user_id, stream_ordering): + async def get_rooms_for_user_at( + self, user_id: str, stream_ordering: int + ) -> FrozenSet[str]: """Get set of joined rooms for a user at the given stream ordering. The stream ordering *must* be recent, otherwise this may throw an @@ -1879,12 +1918,11 @@ class SyncHandler(object): current token, which should be perfectly fine). Args: - user_id (str) - stream_ordering (int) + user_id + stream_ordering ReturnValue: - Deferred[frozenset[str]]: Set of room_ids the user is in at given - stream_ordering. + Set of room_ids the user is in at given stream_ordering. """ joined_rooms = await self.store.get_rooms_for_user_with_stream_ordering(user_id) @@ -1911,11 +1949,10 @@ class SyncHandler(object): if user_id in users_in_room: joined_room_ids.add(room_id) - joined_room_ids = frozenset(joined_room_ids) - return joined_room_ids + return frozenset(joined_room_ids) -def _action_has_highlight(actions): +def _action_has_highlight(actions: List[JsonDict]) -> bool: for action in actions: try: if action.get("set_tweak", None) == "highlight": @@ -1927,22 +1964,23 @@ def _action_has_highlight(actions): def _calculate_state( - timeline_contains, timeline_start, previous, current, lazy_load_members -): + timeline_contains: StateMap[str], + timeline_start: StateMap[str], + previous: StateMap[str], + current: StateMap[str], + lazy_load_members: bool, +) -> StateMap[str]: """Works out what state to include in a sync response. Args: - timeline_contains (dict): state in the timeline - timeline_start (dict): state at the start of the timeline - previous (dict): state at the end of the previous sync (or empty dict + timeline_contains: state in the timeline + timeline_start: state at the start of the timeline + previous: state at the end of the previous sync (or empty dict if this is an initial sync) - current (dict): state at the end of the timeline - lazy_load_members (bool): whether to return members from timeline_start + current: state at the end of the timeline + lazy_load_members: whether to return members from timeline_start or not. assumes that timeline_start has already been filtered to include only the members the client needs to know about. - - Returns: - dict """ event_id_to_key = { e: key @@ -1979,15 +2017,16 @@ def _calculate_state( return {event_id_to_key[e]: e for e in state_ids} -class SyncResultBuilder(object): +@attr.s +class SyncResultBuilder: """Used to help build up a new SyncResult for a user Attributes: - sync_config (SyncConfig) - full_state (bool) - since_token (StreamToken) - now_token (StreamToken) - joined_room_ids (list[str]) + sync_config + full_state: The full_state flag as specified by user + since_token: The token supplied by user, or None. + now_token: The token to sync up to. + joined_room_ids: List of rooms the user is joined to # The following mirror the fields in a sync response presence (list) @@ -1995,61 +2034,45 @@ class SyncResultBuilder(object): joined (list[JoinedSyncResult]) invited (list[InvitedSyncResult]) archived (list[ArchivedSyncResult]) - device (list) groups (GroupsSyncResult|None) to_device (list) """ - def __init__( - self, sync_config, full_state, since_token, now_token, joined_room_ids - ): - """ - Args: - sync_config (SyncConfig) - full_state (bool): The full_state flag as specified by user - since_token (StreamToken): The token supplied by user, or None. - now_token (StreamToken): The token to sync up to. - joined_room_ids (list[str]): List of rooms the user is joined to - """ - self.sync_config = sync_config - self.full_state = full_state - self.since_token = since_token - self.now_token = now_token - self.joined_room_ids = joined_room_ids - - self.presence = [] - self.account_data = [] - self.joined = [] - self.invited = [] - self.archived = [] - self.device = [] - self.groups = None - self.to_device = [] + sync_config = attr.ib(type=SyncConfig) + full_state = attr.ib(type=bool) + since_token = attr.ib(type=Optional[StreamToken]) + now_token = attr.ib(type=StreamToken) + joined_room_ids = attr.ib(type=FrozenSet[str]) + + presence = attr.ib(type=List[JsonDict], default=attr.Factory(list)) + account_data = attr.ib(type=List[JsonDict], default=attr.Factory(list)) + joined = attr.ib(type=List[JoinedSyncResult], default=attr.Factory(list)) + invited = attr.ib(type=List[InvitedSyncResult], default=attr.Factory(list)) + archived = attr.ib(type=List[ArchivedSyncResult], default=attr.Factory(list)) + groups = attr.ib(type=Optional[GroupsSyncResult], default=None) + to_device = attr.ib(type=List[JsonDict], default=attr.Factory(list)) +@attr.s class RoomSyncResultBuilder(object): """Stores information needed to create either a `JoinedSyncResult` or `ArchivedSyncResult`. + + Attributes: + room_id + rtype: One of `"joined"` or `"archived"` + events: List of events to include in the room (more events may be added + when generating result). + newly_joined: If the user has newly joined the room + full_state: Whether the full state should be sent in result + since_token: Earliest point to return events from, or None + upto_token: Latest point to return events from. """ - def __init__( - self, room_id, rtype, events, newly_joined, full_state, since_token, upto_token - ): - """ - Args: - room_id(str) - rtype(str): One of `"joined"` or `"archived"` - events(list[FrozenEvent]): List of events to include in the room - (more events may be added when generating result). - newly_joined(bool): If the user has newly joined the room - full_state(bool): Whether the full state should be sent in result - since_token(StreamToken): Earliest point to return events from, or None - upto_token(StreamToken): Latest point to return events from. - """ - self.room_id = room_id - self.rtype = rtype - self.events = events - self.newly_joined = newly_joined - self.full_state = full_state - self.since_token = since_token - self.upto_token = upto_token + room_id = attr.ib(type=str) + rtype = attr.ib(type=str) + events = attr.ib(type=Optional[List[EventBase]]) + newly_joined = attr.ib(type=bool) + full_state = attr.ib(type=bool) + since_token = attr.ib(type=Optional[StreamToken]) + upto_token = attr.ib(type=StreamToken) diff --git a/tests/storage/test_redaction.py b/tests/storage/test_redaction.py index feb1c07cb2..b9ee6ec1ec 100644 --- a/tests/storage/test_redaction.py +++ b/tests/storage/test_redaction.py @@ -238,8 +238,11 @@ class RedactionTestCase(unittest.HomeserverTestCase): @defer.inlineCallbacks def build(self, prev_event_ids): built_event = yield self._base_builder.build(prev_event_ids) - built_event.event_id = self._event_id + + built_event._event_id = self._event_id built_event._event_dict["event_id"] = self._event_id + assert built_event.event_id == self._event_id + return built_event @property diff --git a/tox.ini b/tox.ini index 88ef12bebd..ef22368cf1 100644 --- a/tox.ini +++ b/tox.ini @@ -180,6 +180,7 @@ commands = mypy \ synapse/api \ synapse/config/ \ synapse/federation/transport \ + synapse/handlers/sync.py \ synapse/handlers/ui_auth \ synapse/logging/ \ synapse/module_api \ -- cgit 1.4.1 From 928edef9793bf10fa6156a42c4babbfaaaa17f88 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Fri, 31 Jan 2020 16:50:13 +0000 Subject: Pass room_version into `event_from_pdu_json` It's called from all over the shop, so this one's a bit messy. --- changelog.d/6856.misc | 1 + synapse/federation/federation_base.py | 28 ++++++++++++---------- synapse/federation/federation_client.py | 35 +++++++++++++--------------- synapse/federation/federation_server.py | 41 +++++++++++---------------------- tests/handlers/test_federation.py | 6 +++-- 5 files changed, 51 insertions(+), 60 deletions(-) create mode 100644 changelog.d/6856.misc (limited to 'tests') diff --git a/changelog.d/6856.misc b/changelog.d/6856.misc new file mode 100644 index 0000000000..08aa80bcd9 --- /dev/null +++ b/changelog.d/6856.misc @@ -0,0 +1 @@ +Refactoring work in preparation for changing the event redaction algorithm. diff --git a/synapse/federation/federation_base.py b/synapse/federation/federation_base.py index 0e22183280..ebe8b8e9fe 100644 --- a/synapse/federation/federation_base.py +++ b/synapse/federation/federation_base.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2015, 2016 OpenMarket Ltd +# Copyright 2020 The Matrix.org Foundation C.I.C. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -22,9 +23,13 @@ from twisted.internet.defer import DeferredList from synapse.api.constants import MAX_DEPTH, EventTypes, Membership from synapse.api.errors import Codes, SynapseError -from synapse.api.room_versions import KNOWN_ROOM_VERSIONS, EventFormatVersions +from synapse.api.room_versions import ( + KNOWN_ROOM_VERSIONS, + EventFormatVersions, + RoomVersion, +) from synapse.crypto.event_signing import check_event_content_hash -from synapse.events import event_type_from_format_version +from synapse.events import EventBase, event_type_from_format_version from synapse.events.utils import prune_event from synapse.http.servlet import assert_params_in_dict from synapse.logging.context import ( @@ -33,7 +38,7 @@ from synapse.logging.context import ( make_deferred_yieldable, preserve_fn, ) -from synapse.types import get_domain_from_id +from synapse.types import JsonDict, get_domain_from_id from synapse.util import unwrapFirstError logger = logging.getLogger(__name__) @@ -342,16 +347,15 @@ def _is_invite_via_3pid(event): ) -def event_from_pdu_json(pdu_json, event_format_version, outlier=False): - """Construct a FrozenEvent from an event json received over federation +def event_from_pdu_json( + pdu_json: JsonDict, room_version: RoomVersion, outlier: bool = False +) -> EventBase: + """Construct an EventBase from an event json received over federation Args: - pdu_json (object): pdu as received over federation - event_format_version (int): The event format version - outlier (bool): True to mark this event as an outlier - - Returns: - FrozenEvent + pdu_json: pdu as received over federation + room_version: The version of the room this event belongs to + outlier: True to mark this event as an outlier Raises: SynapseError: if the pdu is missing required fields or is otherwise @@ -370,7 +374,7 @@ def event_from_pdu_json(pdu_json, event_format_version, outlier=False): elif depth > MAX_DEPTH: raise SynapseError(400, "Depth too large", Codes.BAD_JSON) - event = event_type_from_format_version(event_format_version)(pdu_json) + event = event_type_from_format_version(room_version.event_format)(pdu_json) event.internal_metadata.outlier = outlier diff --git a/synapse/federation/federation_client.py b/synapse/federation/federation_client.py index 5fb4bd414c..4870e39652 100644 --- a/synapse/federation/federation_client.py +++ b/synapse/federation/federation_client.py @@ -49,7 +49,7 @@ from synapse.api.room_versions import ( RoomVersion, RoomVersions, ) -from synapse.events import EventBase, builder, room_version_to_event_format +from synapse.events import EventBase, builder from synapse.federation.federation_base import FederationBase, event_from_pdu_json from synapse.logging.context import make_deferred_yieldable from synapse.logging.utils import log_function @@ -209,18 +209,18 @@ class FederationClient(FederationBase): logger.debug("backfill transaction_data=%r", transaction_data) - room_version = await self.store.get_room_version_id(room_id) - format_ver = room_version_to_event_format(room_version) + room_version = await self.store.get_room_version(room_id) pdus = [ - event_from_pdu_json(p, format_ver, outlier=False) + event_from_pdu_json(p, room_version, outlier=False) for p in transaction_data["pdus"] ] # FIXME: We should handle signature failures more gracefully. pdus[:] = await make_deferred_yieldable( defer.gatherResults( - self._check_sigs_and_hashes(room_version, pdus), consumeErrors=True + self._check_sigs_and_hashes(room_version.identifier, pdus), + consumeErrors=True, ).addErrback(unwrapFirstError) ) @@ -262,8 +262,6 @@ class FederationClient(FederationBase): pdu_attempts = self.pdu_destination_tried.setdefault(event_id, {}) - format_ver = room_version.event_format - signed_pdu = None for destination in destinations: now = self._clock.time_msec() @@ -284,7 +282,7 @@ class FederationClient(FederationBase): ) pdu_list = [ - event_from_pdu_json(p, format_ver, outlier=outlier) + event_from_pdu_json(p, room_version, outlier=outlier) for p in transaction_data["pdus"] ] @@ -350,15 +348,15 @@ class FederationClient(FederationBase): async def get_event_auth(self, destination, room_id, event_id): res = await self.transport_layer.get_event_auth(destination, room_id, event_id) - room_version = await self.store.get_room_version_id(room_id) - format_ver = room_version_to_event_format(room_version) + room_version = await self.store.get_room_version(room_id) auth_chain = [ - event_from_pdu_json(p, format_ver, outlier=True) for p in res["auth_chain"] + event_from_pdu_json(p, room_version, outlier=True) + for p in res["auth_chain"] ] signed_auth = await self._check_sigs_and_hash_and_fetch( - destination, auth_chain, outlier=True, room_version=room_version + destination, auth_chain, outlier=True, room_version=room_version.identifier ) signed_auth.sort(key=lambda e: e.depth) @@ -547,12 +545,12 @@ class FederationClient(FederationBase): logger.debug("Got content: %s", content) state = [ - event_from_pdu_json(p, room_version.event_format, outlier=True) + event_from_pdu_json(p, room_version, outlier=True) for p in content.get("state", []) ] auth_chain = [ - event_from_pdu_json(p, room_version.event_format, outlier=True) + event_from_pdu_json(p, room_version, outlier=True) for p in content.get("auth_chain", []) ] @@ -677,7 +675,7 @@ class FederationClient(FederationBase): logger.debug("Got response to send_invite: %s", pdu_dict) - pdu = event_from_pdu_json(pdu_dict, room_version.event_format) + pdu = event_from_pdu_json(pdu_dict, room_version) # Check signatures are correct. pdu = await self._check_sigs_and_hash(room_version.identifier, pdu) @@ -865,15 +863,14 @@ class FederationClient(FederationBase): timeout=timeout, ) - room_version = await self.store.get_room_version_id(room_id) - format_ver = room_version_to_event_format(room_version) + room_version = await self.store.get_room_version(room_id) events = [ - event_from_pdu_json(e, format_ver) for e in content.get("events", []) + event_from_pdu_json(e, room_version) for e in content.get("events", []) ] signed_events = await self._check_sigs_and_hash_and_fetch( - destination, events, outlier=False, room_version=room_version + destination, events, outlier=False, room_version=room_version.identifier ) except HttpResponseException as e: if not e.code == 400: diff --git a/synapse/federation/federation_server.py b/synapse/federation/federation_server.py index 8e3933b6c5..2489832a11 100644 --- a/synapse/federation/federation_server.py +++ b/synapse/federation/federation_server.py @@ -38,7 +38,6 @@ from synapse.api.errors import ( UnsupportedRoomVersionError, ) from synapse.api.room_versions import KNOWN_ROOM_VERSIONS -from synapse.events import room_version_to_event_format from synapse.federation.federation_base import FederationBase, event_from_pdu_json from synapse.federation.persistence import TransactionActions from synapse.federation.units import Edu, Transaction @@ -234,24 +233,17 @@ class FederationServer(FederationBase): continue try: - room_version = await self.store.get_room_version_id(room_id) + room_version = await self.store.get_room_version(room_id) except NotFoundError: logger.info("Ignoring PDU for unknown room_id: %s", room_id) continue - - try: - format_ver = room_version_to_event_format(room_version) - except UnsupportedRoomVersionError: + except UnsupportedRoomVersionError as e: # this can happen if support for a given room version is withdrawn, # so that we still get events for said room. - logger.info( - "Ignoring PDU for room %s with unknown version %s", - room_id, - room_version, - ) + logger.info("Ignoring PDU: %s", e) continue - event = event_from_pdu_json(p, format_ver) + event = event_from_pdu_json(p, room_version) pdus_by_room.setdefault(room_id, []).append(event) pdu_results = {} @@ -407,9 +399,7 @@ class FederationServer(FederationBase): Codes.UNSUPPORTED_ROOM_VERSION, ) - format_ver = room_version.event_format - - pdu = event_from_pdu_json(content, format_ver) + pdu = event_from_pdu_json(content, room_version) origin_host, _ = parse_server_name(origin) await self.check_server_matches_acl(origin_host, pdu.room_id) pdu = await self._check_sigs_and_hash(room_version.identifier, pdu) @@ -420,16 +410,15 @@ class FederationServer(FederationBase): async def on_send_join_request(self, origin, content, room_id): logger.debug("on_send_join_request: content: %s", content) - room_version = await self.store.get_room_version_id(room_id) - format_ver = room_version_to_event_format(room_version) - pdu = event_from_pdu_json(content, format_ver) + room_version = await self.store.get_room_version(room_id) + pdu = event_from_pdu_json(content, room_version) origin_host, _ = parse_server_name(origin) await self.check_server_matches_acl(origin_host, pdu.room_id) logger.debug("on_send_join_request: pdu sigs: %s", pdu.signatures) - pdu = await self._check_sigs_and_hash(room_version, pdu) + pdu = await self._check_sigs_and_hash(room_version.identifier, pdu) res_pdus = await self.handler.on_send_join_request(origin, pdu) time_now = self._clock.time_msec() @@ -451,16 +440,15 @@ class FederationServer(FederationBase): async def on_send_leave_request(self, origin, content, room_id): logger.debug("on_send_leave_request: content: %s", content) - room_version = await self.store.get_room_version_id(room_id) - format_ver = room_version_to_event_format(room_version) - pdu = event_from_pdu_json(content, format_ver) + room_version = await self.store.get_room_version(room_id) + pdu = event_from_pdu_json(content, room_version) origin_host, _ = parse_server_name(origin) await self.check_server_matches_acl(origin_host, pdu.room_id) logger.debug("on_send_leave_request: pdu sigs: %s", pdu.signatures) - pdu = await self._check_sigs_and_hash(room_version, pdu) + pdu = await self._check_sigs_and_hash(room_version.identifier, pdu) await self.handler.on_send_leave_request(origin, pdu) return {} @@ -498,15 +486,14 @@ class FederationServer(FederationBase): origin_host, _ = parse_server_name(origin) await self.check_server_matches_acl(origin_host, room_id) - room_version = await self.store.get_room_version_id(room_id) - format_ver = room_version_to_event_format(room_version) + room_version = await self.store.get_room_version(room_id) auth_chain = [ - event_from_pdu_json(e, format_ver) for e in content["auth_chain"] + event_from_pdu_json(e, room_version) for e in content["auth_chain"] ] signed_auth = await self._check_sigs_and_hash_and_fetch( - origin, auth_chain, outlier=True, room_version=room_version + origin, auth_chain, outlier=True, room_version=room_version.identifier ) ret = await self.handler.on_query_auth( diff --git a/tests/handlers/test_federation.py b/tests/handlers/test_federation.py index b4d92cf732..132e35651d 100644 --- a/tests/handlers/test_federation.py +++ b/tests/handlers/test_federation.py @@ -99,6 +99,7 @@ class FederationTestCase(unittest.HomeserverTestCase): user_id = self.register_user("kermit", "test") tok = self.login("kermit", "test") room_id = self.helper.create_room_as(room_creator=user_id, tok=tok) + room_version = self.get_success(self.store.get_room_version(room_id)) # pretend that another server has joined join_event = self._build_and_send_join_event(OTHER_SERVER, OTHER_USER, room_id) @@ -120,7 +121,7 @@ class FederationTestCase(unittest.HomeserverTestCase): "auth_events": [], "origin_server_ts": self.clock.time_msec(), }, - join_event.format_version, + room_version, ) with LoggingContext(request="send_rejected"): @@ -149,6 +150,7 @@ class FederationTestCase(unittest.HomeserverTestCase): user_id = self.register_user("kermit", "test") tok = self.login("kermit", "test") room_id = self.helper.create_room_as(room_creator=user_id, tok=tok) + room_version = self.get_success(self.store.get_room_version(room_id)) # pretend that another server has joined join_event = self._build_and_send_join_event(OTHER_SERVER, OTHER_USER, room_id) @@ -171,7 +173,7 @@ class FederationTestCase(unittest.HomeserverTestCase): "auth_events": [], "origin_server_ts": self.clock.time_msec(), }, - join_event.format_version, + room_version, ) with LoggingContext(request="send_rejected"): -- cgit 1.4.1 From 56ca93ef5941b5dfcda368f373a6bcd80d177acd Mon Sep 17 00:00:00 2001 From: Dirk Klimpel <5740567+dklimpel@users.noreply.github.com> Date: Fri, 7 Feb 2020 11:29:36 +0100 Subject: Admin api to add an email address (#6789) --- changelog.d/6769.feature | 1 + docs/admin_api/user_admin_api.rst | 11 +++++++++++ synapse/handlers/admin.py | 2 ++ synapse/handlers/auth.py | 8 ++++++++ synapse/rest/admin/users.py | 39 +++++++++++++++++++++++++++++++++++++++ tests/rest/admin/test_user.py | 19 +++++++++++++++++-- 6 files changed, 78 insertions(+), 2 deletions(-) create mode 100644 changelog.d/6769.feature (limited to 'tests') diff --git a/changelog.d/6769.feature b/changelog.d/6769.feature new file mode 100644 index 0000000000..8a60e12907 --- /dev/null +++ b/changelog.d/6769.feature @@ -0,0 +1 @@ +Admin API to add or modify threepids of user accounts. \ No newline at end of file diff --git a/docs/admin_api/user_admin_api.rst b/docs/admin_api/user_admin_api.rst index 0b3d09d694..eb146095de 100644 --- a/docs/admin_api/user_admin_api.rst +++ b/docs/admin_api/user_admin_api.rst @@ -15,6 +15,16 @@ with a body of: { "password": "user_password", "displayname": "User", + "threepids": [ + { + "medium": "email", + "address": "" + }, + { + "medium": "email", + "address": "" + } + ], "avatar_url": "", "admin": false, "deactivated": false @@ -23,6 +33,7 @@ with a body of: including an ``access_token`` of a server admin. The parameter ``displayname`` is optional and defaults to ``user_id``. +The parameter ``threepids`` is optional. The parameter ``avatar_url`` is optional. The parameter ``admin`` is optional and defaults to 'false'. The parameter ``deactivated`` is optional and defaults to 'false'. diff --git a/synapse/handlers/admin.py b/synapse/handlers/admin.py index 9205865231..f3c0aeceb6 100644 --- a/synapse/handlers/admin.py +++ b/synapse/handlers/admin.py @@ -58,8 +58,10 @@ class AdminHandler(BaseHandler): ret = await self.store.get_user_by_id(user.to_string()) if ret: profile = await self.store.get_profileinfo(user.localpart) + threepids = await self.store.user_get_threepids(user.to_string()) ret["displayname"] = profile.display_name ret["avatar_url"] = profile.avatar_url + ret["threepids"] = threepids return ret async def export_user_data(self, user_id, writer): diff --git a/synapse/handlers/auth.py b/synapse/handlers/auth.py index 54a71c49d2..48a88d3c2a 100644 --- a/synapse/handlers/auth.py +++ b/synapse/handlers/auth.py @@ -816,6 +816,14 @@ class AuthHandler(BaseHandler): @defer.inlineCallbacks def add_threepid(self, user_id, medium, address, validated_at): + # check if medium has a valid value + if medium not in ["email", "msisdn"]: + raise SynapseError( + code=400, + msg=("'%s' is not a valid value for 'medium'" % (medium,)), + errcode=Codes.INVALID_PARAM, + ) + # 'Canonicalise' email addresses down to lower case. # We've now moving towards the homeserver being the entity that # is responsible for validating threepids used for resetting passwords diff --git a/synapse/rest/admin/users.py b/synapse/rest/admin/users.py index f1c4434f5c..e75c5f1370 100644 --- a/synapse/rest/admin/users.py +++ b/synapse/rest/admin/users.py @@ -136,6 +136,8 @@ class UserRestServletV2(RestServlet): self.hs = hs self.auth = hs.get_auth() self.admin_handler = hs.get_handlers().admin_handler + self.store = hs.get_datastore() + self.auth_handler = hs.get_auth_handler() self.profile_handler = hs.get_profile_handler() self.set_password_handler = hs.get_set_password_handler() self.deactivate_account_handler = hs.get_deactivate_account_handler() @@ -163,6 +165,7 @@ class UserRestServletV2(RestServlet): raise SynapseError(400, "This endpoint can only be used with local users") user = await self.admin_handler.get_user(target_user) + user_id = target_user.to_string() if user: # modify user if "displayname" in body: @@ -170,6 +173,29 @@ class UserRestServletV2(RestServlet): target_user, requester, body["displayname"], True ) + if "threepids" in body: + # check for required parameters for each threepid + for threepid in body["threepids"]: + assert_params_in_dict(threepid, ["medium", "address"]) + + # remove old threepids from user + threepids = await self.store.user_get_threepids(user_id) + for threepid in threepids: + try: + await self.auth_handler.delete_threepid( + user_id, threepid["medium"], threepid["address"], None + ) + except Exception: + logger.exception("Failed to remove threepids") + raise SynapseError(500, "Failed to remove threepids") + + # add new threepids to user + current_time = self.hs.get_clock().time_msec() + for threepid in body["threepids"]: + await self.auth_handler.add_threepid( + user_id, threepid["medium"], threepid["address"], current_time + ) + if "avatar_url" in body: await self.profile_handler.set_avatar_url( target_user, requester, body["avatar_url"], True @@ -221,6 +247,7 @@ class UserRestServletV2(RestServlet): admin = body.get("admin", None) user_type = body.get("user_type", None) displayname = body.get("displayname", None) + threepids = body.get("threepids", None) if user_type is not None and user_type not in UserTypes.ALL_USER_TYPES: raise SynapseError(400, "Invalid user type") @@ -232,6 +259,18 @@ class UserRestServletV2(RestServlet): default_display_name=displayname, user_type=user_type, ) + + if "threepids" in body: + # check for required parameters for each threepid + for threepid in body["threepids"]: + assert_params_in_dict(threepid, ["medium", "address"]) + + current_time = self.hs.get_clock().time_msec() + for threepid in body["threepids"]: + await self.auth_handler.add_threepid( + user_id, threepid["medium"], threepid["address"], current_time + ) + if "avatar_url" in body: await self.profile_handler.set_avatar_url( user_id, requester, body["avatar_url"], True diff --git a/tests/rest/admin/test_user.py b/tests/rest/admin/test_user.py index 8f09f51c61..3b5169b38d 100644 --- a/tests/rest/admin/test_user.py +++ b/tests/rest/admin/test_user.py @@ -407,7 +407,13 @@ class UserRestTestCase(unittest.HomeserverTestCase): """ self.hs.config.registration_shared_secret = None - body = json.dumps({"password": "abc123", "admin": True}) + body = json.dumps( + { + "password": "abc123", + "admin": True, + "threepids": [{"medium": "email", "address": "bob@bob.bob"}], + } + ) # Create user request, channel = self.make_request( @@ -421,6 +427,8 @@ class UserRestTestCase(unittest.HomeserverTestCase): self.assertEqual(201, int(channel.result["code"]), msg=channel.result["body"]) self.assertEqual("@bob:test", channel.json_body["name"]) self.assertEqual("bob", channel.json_body["displayname"]) + self.assertEqual("email", channel.json_body["threepids"][0]["medium"]) + self.assertEqual("bob@bob.bob", channel.json_body["threepids"][0]["address"]) # Get user request, channel = self.make_request( @@ -449,7 +457,13 @@ class UserRestTestCase(unittest.HomeserverTestCase): self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) # Modify user - body = json.dumps({"displayname": "foobar", "deactivated": True}) + body = json.dumps( + { + "displayname": "foobar", + "deactivated": True, + "threepids": [{"medium": "email", "address": "bob2@bob.bob"}], + } + ) request, channel = self.make_request( "PUT", @@ -463,6 +477,7 @@ class UserRestTestCase(unittest.HomeserverTestCase): self.assertEqual("@bob:test", channel.json_body["name"]) self.assertEqual("foobar", channel.json_body["displayname"]) self.assertEqual(True, channel.json_body["deactivated"]) + # the user is deactivated, the threepid will be deleted # Get user request, channel = self.make_request( -- cgit 1.4.1 From b08b0a22d505b1555f511e3f38935a62930ea25d Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Fri, 7 Feb 2020 13:56:38 +0000 Subject: Add typing to synapse.federation.sender (#6871) --- changelog.d/6871.misc | 1 + synapse/federation/federation_server.py | 7 +- synapse/federation/sender/__init__.py | 99 +++++++++++----------- synapse/federation/sender/per_destination_queue.py | 88 +++++++++---------- synapse/federation/sender/transaction_manager.py | 16 ++-- synapse/federation/units.py | 23 ++++- synapse/server.pyi | 2 + tests/handlers/test_typing.py | 8 +- tox.ini | 1 + 9 files changed, 138 insertions(+), 107 deletions(-) create mode 100644 changelog.d/6871.misc (limited to 'tests') diff --git a/changelog.d/6871.misc b/changelog.d/6871.misc new file mode 100644 index 0000000000..5161af9983 --- /dev/null +++ b/changelog.d/6871.misc @@ -0,0 +1 @@ +Add typing to `synapse.federation.sender` and port to async/await. diff --git a/synapse/federation/federation_server.py b/synapse/federation/federation_server.py index 2489832a11..a6c966a393 100644 --- a/synapse/federation/federation_server.py +++ b/synapse/federation/federation_server.py @@ -294,7 +294,12 @@ class FederationServer(FederationBase): async def _process_edu(edu_dict): received_edus_counter.inc() - edu = Edu(**edu_dict) + edu = Edu( + origin=origin, + destination=self.server_name, + edu_type=edu_dict["edu_type"], + content=edu_dict["content"], + ) await self.registry.on_edu(edu.edu_type, origin, edu.content) await concurrently_execute( diff --git a/synapse/federation/sender/__init__.py b/synapse/federation/sender/__init__.py index 36c83c3027..233cb33daf 100644 --- a/synapse/federation/sender/__init__.py +++ b/synapse/federation/sender/__init__.py @@ -14,6 +14,7 @@ # limitations under the License. import logging +from typing import Dict, Hashable, Iterable, List, Optional, Set from six import itervalues @@ -23,6 +24,7 @@ from twisted.internet import defer import synapse import synapse.metrics +from synapse.events import EventBase from synapse.federation.sender.per_destination_queue import PerDestinationQueue from synapse.federation.sender.transaction_manager import TransactionManager from synapse.federation.units import Edu @@ -39,6 +41,8 @@ from synapse.metrics import ( events_processed_counter, ) from synapse.metrics.background_process_metrics import run_as_background_process +from synapse.storage.presence import UserPresenceState +from synapse.types import ReadReceipt from synapse.util.metrics import Measure, measure_func logger = logging.getLogger(__name__) @@ -68,7 +72,7 @@ class FederationSender(object): self._transaction_manager = TransactionManager(hs) # map from destination to PerDestinationQueue - self._per_destination_queues = {} # type: dict[str, PerDestinationQueue] + self._per_destination_queues = {} # type: Dict[str, PerDestinationQueue] LaterGauge( "synapse_federation_transaction_queue_pending_destinations", @@ -84,7 +88,7 @@ class FederationSender(object): # Map of user_id -> UserPresenceState for all the pending presence # to be sent out by user_id. Entries here get processed and put in # pending_presence_by_dest - self.pending_presence = {} + self.pending_presence = {} # type: Dict[str, UserPresenceState] LaterGauge( "synapse_federation_transaction_queue_pending_pdus", @@ -116,20 +120,17 @@ class FederationSender(object): # and that there is a pending call to _flush_rrs_for_room in the system. self._queues_awaiting_rr_flush_by_room = ( {} - ) # type: dict[str, set[PerDestinationQueue]] + ) # type: Dict[str, Set[PerDestinationQueue]] self._rr_txn_interval_per_room_ms = ( - 1000.0 / hs.get_config().federation_rr_transactions_per_room_per_second + 1000.0 / hs.config.federation_rr_transactions_per_room_per_second ) - def _get_per_destination_queue(self, destination): + def _get_per_destination_queue(self, destination: str) -> PerDestinationQueue: """Get or create a PerDestinationQueue for the given destination Args: - destination (str): server_name of remote server - - Returns: - PerDestinationQueue + destination: server_name of remote server """ queue = self._per_destination_queues.get(destination) if not queue: @@ -137,7 +138,7 @@ class FederationSender(object): self._per_destination_queues[destination] = queue return queue - def notify_new_events(self, current_id): + def notify_new_events(self, current_id: int) -> None: """This gets called when we have some new events we might want to send out to other servers. """ @@ -151,13 +152,12 @@ class FederationSender(object): "process_event_queue_for_federation", self._process_event_queue_loop ) - @defer.inlineCallbacks - def _process_event_queue_loop(self): + async def _process_event_queue_loop(self) -> None: try: self._is_processing = True while True: - last_token = yield self.store.get_federation_out_pos("events") - next_token, events = yield self.store.get_all_new_events_stream( + last_token = await self.store.get_federation_out_pos("events") + next_token, events = await self.store.get_all_new_events_stream( last_token, self._last_poked_id, limit=100 ) @@ -166,8 +166,7 @@ class FederationSender(object): if not events and next_token >= self._last_poked_id: break - @defer.inlineCallbacks - def handle_event(event): + async def handle_event(event: EventBase) -> None: # Only send events for this server. send_on_behalf_of = event.internal_metadata.get_send_on_behalf_of() is_mine = self.is_mine_id(event.sender) @@ -184,7 +183,7 @@ class FederationSender(object): # Otherwise if the last member on a server in a room is # banned then it won't receive the event because it won't # be in the room after the ban. - destinations = yield self.state.get_hosts_in_room_at_events( + destinations = await self.state.get_hosts_in_room_at_events( event.room_id, event_ids=event.prev_event_ids() ) except Exception: @@ -206,17 +205,16 @@ class FederationSender(object): self._send_pdu(event, destinations) - @defer.inlineCallbacks - def handle_room_events(events): + async def handle_room_events(events: Iterable[EventBase]) -> None: with Measure(self.clock, "handle_room_events"): for event in events: - yield handle_event(event) + await handle_event(event) - events_by_room = {} + events_by_room = {} # type: Dict[str, List[EventBase]] for event in events: events_by_room.setdefault(event.room_id, []).append(event) - yield make_deferred_yieldable( + await make_deferred_yieldable( defer.gatherResults( [ run_in_background(handle_room_events, evs) @@ -226,11 +224,11 @@ class FederationSender(object): ) ) - yield self.store.update_federation_out_pos("events", next_token) + await self.store.update_federation_out_pos("events", next_token) if events: now = self.clock.time_msec() - ts = yield self.store.get_received_ts(events[-1].event_id) + ts = await self.store.get_received_ts(events[-1].event_id) synapse.metrics.event_processing_lag.labels( "federation_sender" @@ -254,7 +252,7 @@ class FederationSender(object): finally: self._is_processing = False - def _send_pdu(self, pdu, destinations): + def _send_pdu(self, pdu: EventBase, destinations: Iterable[str]) -> None: # We loop through all destinations to see whether we already have # a transaction in progress. If we do, stick it in the pending_pdus # table and we'll get back to it later. @@ -276,11 +274,11 @@ class FederationSender(object): self._get_per_destination_queue(destination).send_pdu(pdu, order) @defer.inlineCallbacks - def send_read_receipt(self, receipt): + def send_read_receipt(self, receipt: ReadReceipt): """Send a RR to any other servers in the room Args: - receipt (synapse.types.ReadReceipt): receipt to be sent + receipt: receipt to be sent """ # Some background on the rate-limiting going on here. @@ -343,7 +341,7 @@ class FederationSender(object): else: queue.flush_read_receipts_for_room(room_id) - def _schedule_rr_flush_for_room(self, room_id, n_domains): + def _schedule_rr_flush_for_room(self, room_id: str, n_domains: int) -> None: # that is going to cause approximately len(domains) transactions, so now back # off for that multiplied by RR_TXN_INTERVAL_PER_ROOM backoff_ms = self._rr_txn_interval_per_room_ms * n_domains @@ -352,7 +350,7 @@ class FederationSender(object): self.clock.call_later(backoff_ms, self._flush_rrs_for_room, room_id) self._queues_awaiting_rr_flush_by_room[room_id] = set() - def _flush_rrs_for_room(self, room_id): + def _flush_rrs_for_room(self, room_id: str) -> None: queues = self._queues_awaiting_rr_flush_by_room.pop(room_id) logger.debug("Flushing RRs in %s to %s", room_id, queues) @@ -368,14 +366,11 @@ class FederationSender(object): @preserve_fn # the caller should not yield on this @defer.inlineCallbacks - def send_presence(self, states): + def send_presence(self, states: List[UserPresenceState]): """Send the new presence states to the appropriate destinations. This actually queues up the presence states ready for sending and triggers a background task to process them and send out the transactions. - - Args: - states (list(UserPresenceState)) """ if not self.hs.config.use_presence: # No-op if presence is disabled. @@ -412,11 +407,10 @@ class FederationSender(object): finally: self._processing_pending_presence = False - def send_presence_to_destinations(self, states, destinations): + def send_presence_to_destinations( + self, states: List[UserPresenceState], destinations: List[str] + ) -> None: """Send the given presence states to the given destinations. - - Args: - states (list[UserPresenceState]) destinations (list[str]) """ @@ -431,12 +425,9 @@ class FederationSender(object): @measure_func("txnqueue._process_presence") @defer.inlineCallbacks - def _process_presence_inner(self, states): + def _process_presence_inner(self, states: List[UserPresenceState]): """Given a list of states populate self.pending_presence_by_dest and poke to send a new transaction to each destination - - Args: - states (list(UserPresenceState)) """ hosts_and_states = yield get_interested_remotes(self.store, states, self.state) @@ -446,14 +437,20 @@ class FederationSender(object): continue self._get_per_destination_queue(destination).send_presence(states) - def build_and_send_edu(self, destination, edu_type, content, key=None): + def build_and_send_edu( + self, + destination: str, + edu_type: str, + content: dict, + key: Optional[Hashable] = None, + ): """Construct an Edu object, and queue it for sending Args: - destination (str): name of server to send to - edu_type (str): type of EDU to send - content (dict): content of EDU - key (Any|None): clobbering key for this edu + destination: name of server to send to + edu_type: type of EDU to send + content: content of EDU + key: clobbering key for this edu """ if destination == self.server_name: logger.info("Not sending EDU to ourselves") @@ -468,12 +465,12 @@ class FederationSender(object): self.send_edu(edu, key) - def send_edu(self, edu, key): + def send_edu(self, edu: Edu, key: Optional[Hashable]): """Queue an EDU for sending Args: - edu (Edu): edu to send - key (Any|None): clobbering key for this edu + edu: edu to send + key: clobbering key for this edu """ queue = self._get_per_destination_queue(edu.destination) if key: @@ -481,7 +478,7 @@ class FederationSender(object): else: queue.send_edu(edu) - def send_device_messages(self, destination): + def send_device_messages(self, destination: str): if destination == self.server_name: logger.warning("Not sending device update to ourselves") return @@ -501,5 +498,5 @@ class FederationSender(object): self._get_per_destination_queue(destination).attempt_new_transaction() - def get_current_token(self): + def get_current_token(self) -> int: return 0 diff --git a/synapse/federation/sender/per_destination_queue.py b/synapse/federation/sender/per_destination_queue.py index 5012aaea35..e13cd20ffa 100644 --- a/synapse/federation/sender/per_destination_queue.py +++ b/synapse/federation/sender/per_destination_queue.py @@ -15,11 +15,11 @@ # limitations under the License. import datetime import logging +from typing import Dict, Hashable, Iterable, List, Tuple from prometheus_client import Counter -from twisted.internet import defer - +import synapse.server from synapse.api.errors import ( FederationDeniedError, HttpResponseException, @@ -31,7 +31,7 @@ from synapse.handlers.presence import format_user_presence_state from synapse.metrics import sent_transactions_counter from synapse.metrics.background_process_metrics import run_as_background_process from synapse.storage.presence import UserPresenceState -from synapse.types import StateMap +from synapse.types import ReadReceipt from synapse.util.retryutils import NotRetryingDestination, get_retry_limiter # This is defined in the Matrix spec and enforced by the receiver. @@ -56,13 +56,18 @@ class PerDestinationQueue(object): Manages the per-destination transmission queues. Args: - hs (synapse.HomeServer): - transaction_sender (TransactionManager): - destination (str): the server_name of the destination that we are managing + hs + transaction_sender + destination: the server_name of the destination that we are managing transmission for. """ - def __init__(self, hs, transaction_manager, destination): + def __init__( + self, + hs: "synapse.server.HomeServer", + transaction_manager: "synapse.federation.sender.TransactionManager", + destination: str, + ): self._server_name = hs.hostname self._clock = hs.get_clock() self._store = hs.get_datastore() @@ -72,20 +77,20 @@ class PerDestinationQueue(object): self.transmission_loop_running = False # a list of tuples of (pending pdu, order) - self._pending_pdus = [] # type: list[tuple[EventBase, int]] - self._pending_edus = [] # type: list[Edu] + self._pending_pdus = [] # type: List[Tuple[EventBase, int]] + self._pending_edus = [] # type: List[Edu] # Pending EDUs by their "key". Keyed EDUs are EDUs that get clobbered # based on their key (e.g. typing events by room_id) # Map of (edu_type, key) -> Edu - self._pending_edus_keyed = {} # type: StateMap[Edu] + self._pending_edus_keyed = {} # type: Dict[Tuple[str, Hashable], Edu] # Map of user_id -> UserPresenceState of pending presence to be sent to this # destination - self._pending_presence = {} # type: dict[str, UserPresenceState] + self._pending_presence = {} # type: Dict[str, UserPresenceState] # room_id -> receipt_type -> user_id -> receipt_dict - self._pending_rrs = {} + self._pending_rrs = {} # type: Dict[str, Dict[str, Dict[str, dict]]] self._rrs_pending_flush = False # stream_id of last successfully sent to-device message. @@ -95,50 +100,50 @@ class PerDestinationQueue(object): # stream_id of last successfully sent device list update. self._last_device_list_stream_id = 0 - def __str__(self): + def __str__(self) -> str: return "PerDestinationQueue[%s]" % self._destination - def pending_pdu_count(self): + def pending_pdu_count(self) -> int: return len(self._pending_pdus) - def pending_edu_count(self): + def pending_edu_count(self) -> int: return ( len(self._pending_edus) + len(self._pending_presence) + len(self._pending_edus_keyed) ) - def send_pdu(self, pdu, order): + def send_pdu(self, pdu: EventBase, order: int) -> None: """Add a PDU to the queue, and start the transmission loop if neccessary Args: - pdu (EventBase): pdu to send - order (int): + pdu: pdu to send + order """ self._pending_pdus.append((pdu, order)) self.attempt_new_transaction() - def send_presence(self, states): + def send_presence(self, states: Iterable[UserPresenceState]) -> None: """Add presence updates to the queue. Start the transmission loop if neccessary. Args: - states (iterable[UserPresenceState]): presence to send + states: presence to send """ self._pending_presence.update({state.user_id: state for state in states}) self.attempt_new_transaction() - def queue_read_receipt(self, receipt): + def queue_read_receipt(self, receipt: ReadReceipt) -> None: """Add a RR to the list to be sent. Doesn't start the transmission loop yet (see flush_read_receipts_for_room) Args: - receipt (synapse.api.receipt_info.ReceiptInfo): receipt to be queued + receipt: receipt to be queued """ self._pending_rrs.setdefault(receipt.room_id, {}).setdefault( receipt.receipt_type, {} )[receipt.user_id] = {"event_ids": receipt.event_ids, "data": receipt.data} - def flush_read_receipts_for_room(self, room_id): + def flush_read_receipts_for_room(self, room_id: str) -> None: # if we don't have any read-receipts for this room, it may be that we've already # sent them out, so we don't need to flush. if room_id not in self._pending_rrs: @@ -146,15 +151,15 @@ class PerDestinationQueue(object): self._rrs_pending_flush = True self.attempt_new_transaction() - def send_keyed_edu(self, edu, key): + def send_keyed_edu(self, edu: Edu, key: Hashable) -> None: self._pending_edus_keyed[(edu.edu_type, key)] = edu self.attempt_new_transaction() - def send_edu(self, edu): + def send_edu(self, edu) -> None: self._pending_edus.append(edu) self.attempt_new_transaction() - def attempt_new_transaction(self): + def attempt_new_transaction(self) -> None: """Try to start a new transaction to this destination If there is already a transaction in progress to this destination, @@ -177,23 +182,22 @@ class PerDestinationQueue(object): self._transaction_transmission_loop, ) - @defer.inlineCallbacks - def _transaction_transmission_loop(self): - pending_pdus = [] + async def _transaction_transmission_loop(self) -> None: + pending_pdus = [] # type: List[Tuple[EventBase, int]] try: self.transmission_loop_running = True # This will throw if we wouldn't retry. We do this here so we fail # quickly, but we will later check this again in the http client, # hence why we throw the result away. - yield get_retry_limiter(self._destination, self._clock, self._store) + await get_retry_limiter(self._destination, self._clock, self._store) pending_pdus = [] while True: # We have to keep 2 free slots for presence and rr_edus limit = MAX_EDUS_PER_TRANSACTION - 2 - device_update_edus, dev_list_id = yield self._get_device_update_edus( + device_update_edus, dev_list_id = await self._get_device_update_edus( limit ) @@ -202,7 +206,7 @@ class PerDestinationQueue(object): ( to_device_edus, device_stream_id, - ) = yield self._get_to_device_message_edus(limit) + ) = await self._get_to_device_message_edus(limit) pending_edus = device_update_edus + to_device_edus @@ -269,7 +273,7 @@ class PerDestinationQueue(object): # END CRITICAL SECTION - success = yield self._transaction_manager.send_new_transaction( + success = await self._transaction_manager.send_new_transaction( self._destination, pending_pdus, pending_edus ) if success: @@ -280,7 +284,7 @@ class PerDestinationQueue(object): # Remove the acknowledged device messages from the database # Only bother if we actually sent some device messages if to_device_edus: - yield self._store.delete_device_msgs_for_remote( + await self._store.delete_device_msgs_for_remote( self._destination, device_stream_id ) @@ -289,7 +293,7 @@ class PerDestinationQueue(object): logger.info( "Marking as sent %r %r", self._destination, dev_list_id ) - yield self._store.mark_as_sent_devices_by_remote( + await self._store.mark_as_sent_devices_by_remote( self._destination, dev_list_id ) @@ -334,7 +338,7 @@ class PerDestinationQueue(object): # We want to be *very* sure we clear this after we stop processing self.transmission_loop_running = False - def _get_rr_edus(self, force_flush): + def _get_rr_edus(self, force_flush: bool) -> Iterable[Edu]: if not self._pending_rrs: return if not force_flush and not self._rrs_pending_flush: @@ -351,17 +355,16 @@ class PerDestinationQueue(object): self._rrs_pending_flush = False yield edu - def _pop_pending_edus(self, limit): + def _pop_pending_edus(self, limit: int) -> List[Edu]: pending_edus = self._pending_edus pending_edus, self._pending_edus = pending_edus[:limit], pending_edus[limit:] return pending_edus - @defer.inlineCallbacks - def _get_device_update_edus(self, limit): + async def _get_device_update_edus(self, limit: int) -> Tuple[List[Edu], int]: last_device_list = self._last_device_list_stream_id # Retrieve list of new device updates to send to the destination - now_stream_id, results = yield self._store.get_device_updates_by_remote( + now_stream_id, results = await self._store.get_device_updates_by_remote( self._destination, last_device_list, limit=limit ) edus = [ @@ -378,11 +381,10 @@ class PerDestinationQueue(object): return (edus, now_stream_id) - @defer.inlineCallbacks - def _get_to_device_message_edus(self, limit): + async def _get_to_device_message_edus(self, limit: int) -> Tuple[List[Edu], int]: last_device_stream_id = self._last_device_stream_id to_device_stream_id = self._store.get_to_device_stream_token() - contents, stream_id = yield self._store.get_new_device_msgs_for_remote( + contents, stream_id = await self._store.get_new_device_msgs_for_remote( self._destination, last_device_stream_id, to_device_stream_id, limit ) edus = [ diff --git a/synapse/federation/sender/transaction_manager.py b/synapse/federation/sender/transaction_manager.py index 5fed626d5b..3c2a02a3b3 100644 --- a/synapse/federation/sender/transaction_manager.py +++ b/synapse/federation/sender/transaction_manager.py @@ -13,14 +13,15 @@ # See the License for the specific language governing permissions and # limitations under the License. import logging +from typing import List from canonicaljson import json -from twisted.internet import defer - +import synapse.server from synapse.api.errors import HttpResponseException +from synapse.events import EventBase from synapse.federation.persistence import TransactionActions -from synapse.federation.units import Transaction +from synapse.federation.units import Edu, Transaction from synapse.logging.opentracing import ( extract_text_map, set_tag, @@ -39,7 +40,7 @@ class TransactionManager(object): shared between PerDestinationQueue objects """ - def __init__(self, hs): + def __init__(self, hs: "synapse.server.HomeServer"): self._server_name = hs.hostname self.clock = hs.get_clock() # nb must be called this for @measure_func self._store = hs.get_datastore() @@ -50,8 +51,9 @@ class TransactionManager(object): self._next_txn_id = int(self.clock.time_msec()) @measure_func("_send_new_transaction") - @defer.inlineCallbacks - def send_new_transaction(self, destination, pending_pdus, pending_edus): + async def send_new_transaction( + self, destination: str, pending_pdus: List[EventBase], pending_edus: List[Edu] + ): # Make a transaction-sending opentracing span. This span follows on from # all the edus in that transaction. This needs to be done since there is @@ -127,7 +129,7 @@ class TransactionManager(object): return data try: - response = yield self._transport_layer.send_transaction( + response = await self._transport_layer.send_transaction( transaction, json_data_cb ) code = 200 diff --git a/synapse/federation/units.py b/synapse/federation/units.py index b4d743cde7..6b32e0dcbf 100644 --- a/synapse/federation/units.py +++ b/synapse/federation/units.py @@ -19,11 +19,15 @@ server protocol. import logging +import attr + +from synapse.types import JsonDict from synapse.util.jsonobject import JsonEncodedObject logger = logging.getLogger(__name__) +@attr.s(slots=True) class Edu(JsonEncodedObject): """ An Edu represents a piece of data sent from one homeserver to another. @@ -32,11 +36,24 @@ class Edu(JsonEncodedObject): internal ID or previous references graph. """ - valid_keys = ["origin", "destination", "edu_type", "content"] + edu_type = attr.ib(type=str) + content = attr.ib(type=dict) + origin = attr.ib(type=str) + destination = attr.ib(type=str) - required_keys = ["edu_type"] + def get_dict(self) -> JsonDict: + return { + "edu_type": self.edu_type, + "content": self.content, + } - internal_keys = ["origin", "destination"] + def get_internal_dict(self) -> JsonDict: + return { + "edu_type": self.edu_type, + "content": self.content, + "origin": self.origin, + "destination": self.destination, + } def get_context(self): return getattr(self, "content", {}).get("org.matrix.opentracing_context", "{}") diff --git a/synapse/server.pyi b/synapse/server.pyi index 90347ac23e..40eabfe5d9 100644 --- a/synapse/server.pyi +++ b/synapse/server.pyi @@ -107,3 +107,5 @@ class HomeServer(object): self, ) -> synapse.replication.tcp.client.ReplicationClientHandler: pass + def is_mine_id(self, domain_id: str) -> bool: + pass diff --git a/tests/handlers/test_typing.py b/tests/handlers/test_typing.py index 68b9847bd2..2767b0497a 100644 --- a/tests/handlers/test_typing.py +++ b/tests/handlers/test_typing.py @@ -111,7 +111,9 @@ class TypingNotificationsTestCase(unittest.HomeserverTestCase): retry_timings_res ) - self.datastore.get_device_updates_by_remote.return_value = (0, []) + self.datastore.get_device_updates_by_remote.return_value = defer.succeed( + (0, []) + ) def get_received_txn_response(*args): return defer.succeed(None) @@ -144,7 +146,9 @@ class TypingNotificationsTestCase(unittest.HomeserverTestCase): self.datastore.get_current_state_deltas.return_value = (0, None) self.datastore.get_to_device_stream_token = lambda: 0 - self.datastore.get_new_device_msgs_for_remote = lambda *args, **kargs: ([], 0) + self.datastore.get_new_device_msgs_for_remote = lambda *args, **kargs: defer.succeed( + ([], 0) + ) self.datastore.delete_device_msgs_for_remote = lambda *args, **kargs: None self.datastore.set_received_txn_response = lambda *args, **kwargs: defer.succeed( None diff --git a/tox.ini b/tox.ini index ef22368cf1..f8229eba88 100644 --- a/tox.ini +++ b/tox.ini @@ -179,6 +179,7 @@ extras = all commands = mypy \ synapse/api \ synapse/config/ \ + synapse/federation/sender \ synapse/federation/transport \ synapse/handlers/sync.py \ synapse/handlers/ui_auth \ -- cgit 1.4.1 From 799001f2c0b31d72b95a252a3808da25987e1ed3 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Fri, 7 Feb 2020 15:30:04 +0000 Subject: Add a `make_event_from_dict` method (#6858) ... and use it in places where it's trivial to do so. This will make it easier to pass room versions into the FrozenEvent constructors. --- changelog.d/6858.misc | 1 + synapse/events/__init__.py | 16 ++++++++++++++-- synapse/events/builder.py | 10 +++------- synapse/federation/federation_base.py | 5 ++--- tests/api/test_filtering.py | 4 ++-- tests/crypto/test_event_signing.py | 6 +++--- tests/events/test_utils.py | 9 +++++---- tests/federation/test_federation_server.py | 4 ++-- tests/replication/slave/storage/test_events.py | 12 ++++++++---- tests/state/test_v2.py | 4 ++-- tests/test_event_auth.py | 10 +++++----- tests/test_federation.py | 6 +++--- tests/test_state.py | 4 ++-- 13 files changed, 52 insertions(+), 39 deletions(-) create mode 100644 changelog.d/6858.misc (limited to 'tests') diff --git a/changelog.d/6858.misc b/changelog.d/6858.misc new file mode 100644 index 0000000000..08aa80bcd9 --- /dev/null +++ b/changelog.d/6858.misc @@ -0,0 +1 @@ +Refactoring work in preparation for changing the event redaction algorithm. diff --git a/synapse/events/__init__.py b/synapse/events/__init__.py index 89d41d82b6..a842661a90 100644 --- a/synapse/events/__init__.py +++ b/synapse/events/__init__.py @@ -16,12 +16,13 @@ import os from distutils.util import strtobool +from typing import Optional, Type import six from unpaddedbase64 import encode_base64 -from synapse.api.room_versions import EventFormatVersions +from synapse.api.room_versions import EventFormatVersions, RoomVersion, RoomVersions from synapse.types import JsonDict from synapse.util.caches import intern_dict from synapse.util.frozenutils import freeze @@ -407,7 +408,7 @@ class FrozenEventV3(FrozenEventV2): return self._event_id -def event_type_from_format_version(format_version): +def event_type_from_format_version(format_version: int) -> Type[EventBase]: """Returns the python type to use to construct an Event object for the given event format version. @@ -427,3 +428,14 @@ def event_type_from_format_version(format_version): return FrozenEventV3 else: raise Exception("No event format %r" % (format_version,)) + + +def make_event_from_dict( + event_dict: JsonDict, + room_version: RoomVersion = RoomVersions.V1, + internal_metadata_dict: JsonDict = {}, + rejected_reason: Optional[str] = None, +) -> EventBase: + """Construct an EventBase from the given event dict""" + event_type = event_type_from_format_version(room_version.event_format) + return event_type(event_dict, internal_metadata_dict, rejected_reason) diff --git a/synapse/events/builder.py b/synapse/events/builder.py index 8d63ad6dc3..a0c4a40c27 100644 --- a/synapse/events/builder.py +++ b/synapse/events/builder.py @@ -28,11 +28,7 @@ from synapse.api.room_versions import ( RoomVersion, ) from synapse.crypto.event_signing import add_hashes_and_signatures -from synapse.events import ( - EventBase, - _EventInternalMetadata, - event_type_from_format_version, -) +from synapse.events import EventBase, _EventInternalMetadata, make_event_from_dict from synapse.types import EventID, JsonDict from synapse.util import Clock from synapse.util.stringutils import random_string @@ -256,8 +252,8 @@ def create_local_event_from_event_dict( event_dict.setdefault("signatures", {}) add_hashes_and_signatures(room_version, event_dict, hostname, signing_key) - return event_type_from_format_version(format_version)( - event_dict, internal_metadata_dict=internal_metadata_dict + return make_event_from_dict( + event_dict, room_version, internal_metadata_dict=internal_metadata_dict ) diff --git a/synapse/federation/federation_base.py b/synapse/federation/federation_base.py index ebe8b8e9fe..eea64c1c9f 100644 --- a/synapse/federation/federation_base.py +++ b/synapse/federation/federation_base.py @@ -29,7 +29,7 @@ from synapse.api.room_versions import ( RoomVersion, ) from synapse.crypto.event_signing import check_event_content_hash -from synapse.events import EventBase, event_type_from_format_version +from synapse.events import EventBase, make_event_from_dict from synapse.events.utils import prune_event from synapse.http.servlet import assert_params_in_dict from synapse.logging.context import ( @@ -374,8 +374,7 @@ def event_from_pdu_json( elif depth > MAX_DEPTH: raise SynapseError(400, "Depth too large", Codes.BAD_JSON) - event = event_type_from_format_version(room_version.event_format)(pdu_json) - + event = make_event_from_dict(pdu_json, room_version) event.internal_metadata.outlier = outlier return event diff --git a/tests/api/test_filtering.py b/tests/api/test_filtering.py index 63d8633582..4e67503cf0 100644 --- a/tests/api/test_filtering.py +++ b/tests/api/test_filtering.py @@ -25,7 +25,7 @@ from twisted.internet import defer from synapse.api.constants import EventContentFields from synapse.api.errors import SynapseError from synapse.api.filtering import Filter -from synapse.events import FrozenEvent +from synapse.events import make_event_from_dict from tests import unittest from tests.utils import DeferredMockCallable, MockHttpResource, setup_test_homeserver @@ -38,7 +38,7 @@ def MockEvent(**kwargs): kwargs["event_id"] = "fake_event_id" if "type" not in kwargs: kwargs["type"] = "fake_type" - return FrozenEvent(kwargs) + return make_event_from_dict(kwargs) class FilteringTestCase(unittest.TestCase): diff --git a/tests/crypto/test_event_signing.py b/tests/crypto/test_event_signing.py index 6143a50ab2..62f639a18d 100644 --- a/tests/crypto/test_event_signing.py +++ b/tests/crypto/test_event_signing.py @@ -19,7 +19,7 @@ from unpaddedbase64 import decode_base64 from synapse.api.room_versions import RoomVersions from synapse.crypto.event_signing import add_hashes_and_signatures -from synapse.events import FrozenEvent +from synapse.events import make_event_from_dict from tests import unittest @@ -54,7 +54,7 @@ class EventSigningTestCase(unittest.TestCase): RoomVersions.V1, event_dict, HOSTNAME, self.signing_key ) - event = FrozenEvent(event_dict) + event = make_event_from_dict(event_dict) self.assertTrue(hasattr(event, "hashes")) self.assertIn("sha256", event.hashes) @@ -88,7 +88,7 @@ class EventSigningTestCase(unittest.TestCase): RoomVersions.V1, event_dict, HOSTNAME, self.signing_key ) - event = FrozenEvent(event_dict) + event = make_event_from_dict(event_dict) self.assertTrue(hasattr(event, "hashes")) self.assertIn("sha256", event.hashes) diff --git a/tests/events/test_utils.py b/tests/events/test_utils.py index 2b13980dfd..45d55b9e94 100644 --- a/tests/events/test_utils.py +++ b/tests/events/test_utils.py @@ -13,8 +13,7 @@ # See the License for the specific language governing permissions and # limitations under the License. - -from synapse.events import FrozenEvent +from synapse.events import make_event_from_dict from synapse.events.utils import ( copy_power_levels_contents, prune_event, @@ -30,7 +29,7 @@ def MockEvent(**kwargs): kwargs["event_id"] = "fake_event_id" if "type" not in kwargs: kwargs["type"] = "fake_type" - return FrozenEvent(kwargs) + return make_event_from_dict(kwargs) class PruneEventTestCase(unittest.TestCase): @@ -38,7 +37,9 @@ class PruneEventTestCase(unittest.TestCase): `matchdict` when it is redacted. """ def run_test(self, evdict, matchdict): - self.assertEquals(prune_event(FrozenEvent(evdict)).get_dict(), matchdict) + self.assertEquals( + prune_event(make_event_from_dict(evdict)).get_dict(), matchdict + ) def test_minimal(self): self.run_test( diff --git a/tests/federation/test_federation_server.py b/tests/federation/test_federation_server.py index 1ec8c40901..e7d8699040 100644 --- a/tests/federation/test_federation_server.py +++ b/tests/federation/test_federation_server.py @@ -15,7 +15,7 @@ # limitations under the License. import logging -from synapse.events import FrozenEvent +from synapse.events import make_event_from_dict from synapse.federation.federation_server import server_matches_acl_event from synapse.rest import admin from synapse.rest.client.v1 import login, room @@ -105,7 +105,7 @@ class StateQueryTests(unittest.FederatingHomeserverTestCase): def _create_acl_event(content): - return FrozenEvent( + return make_event_from_dict( { "room_id": "!a:b", "event_id": "$a:b", diff --git a/tests/replication/slave/storage/test_events.py b/tests/replication/slave/storage/test_events.py index b1b037006d..d31210fbe4 100644 --- a/tests/replication/slave/storage/test_events.py +++ b/tests/replication/slave/storage/test_events.py @@ -15,7 +15,7 @@ import logging from canonicaljson import encode_canonical_json -from synapse.events import FrozenEvent, _EventInternalMetadata +from synapse.events import FrozenEvent, _EventInternalMetadata, make_event_from_dict from synapse.events.snapshot import EventContext from synapse.handlers.room import RoomEventSource from synapse.replication.slave.storage.events import SlavedEventStore @@ -90,7 +90,9 @@ class SlavedEventStoreTestCase(BaseSlavedStoreTestCase): msg_dict["content"] = {} msg_dict["unsigned"]["redacted_by"] = redaction.event_id msg_dict["unsigned"]["redacted_because"] = redaction - redacted = FrozenEvent(msg_dict, msg.internal_metadata.get_dict()) + redacted = make_event_from_dict( + msg_dict, internal_metadata_dict=msg.internal_metadata.get_dict() + ) self.check("get_event", [msg.event_id], redacted) def test_backfilled_redactions(self): @@ -110,7 +112,9 @@ class SlavedEventStoreTestCase(BaseSlavedStoreTestCase): msg_dict["content"] = {} msg_dict["unsigned"]["redacted_by"] = redaction.event_id msg_dict["unsigned"]["redacted_because"] = redaction - redacted = FrozenEvent(msg_dict, msg.internal_metadata.get_dict()) + redacted = make_event_from_dict( + msg_dict, internal_metadata_dict=msg.internal_metadata.get_dict() + ) self.check("get_event", [msg.event_id], redacted) def test_invites(self): @@ -345,7 +349,7 @@ class SlavedEventStoreTestCase(BaseSlavedStoreTestCase): if redacts is not None: event_dict["redacts"] = redacts - event = FrozenEvent(event_dict, internal_metadata_dict=internal) + event = make_event_from_dict(event_dict, internal_metadata_dict=internal) self.event_id += 1 diff --git a/tests/state/test_v2.py b/tests/state/test_v2.py index 0f341d3ac3..5bafad9f19 100644 --- a/tests/state/test_v2.py +++ b/tests/state/test_v2.py @@ -22,7 +22,7 @@ import attr from synapse.api.constants import EventTypes, JoinRules, Membership from synapse.api.room_versions import RoomVersions from synapse.event_auth import auth_types_for_event -from synapse.events import FrozenEvent +from synapse.events import make_event_from_dict from synapse.state.v2 import lexicographical_topological_sort, resolve_events_with_store from synapse.types import EventID @@ -89,7 +89,7 @@ class FakeEvent(object): if self.state_key is not None: event_dict["state_key"] = self.state_key - return FrozenEvent(event_dict) + return make_event_from_dict(event_dict) # All graphs start with this set of events diff --git a/tests/test_event_auth.py b/tests/test_event_auth.py index ca20b085a2..bfa5d6f510 100644 --- a/tests/test_event_auth.py +++ b/tests/test_event_auth.py @@ -18,7 +18,7 @@ import unittest from synapse import event_auth from synapse.api.errors import AuthError from synapse.api.room_versions import RoomVersions -from synapse.events import FrozenEvent +from synapse.events import make_event_from_dict class EventAuthTestCase(unittest.TestCase): @@ -94,7 +94,7 @@ TEST_ROOM_ID = "!test:room" def _create_event(user_id): - return FrozenEvent( + return make_event_from_dict( { "room_id": TEST_ROOM_ID, "event_id": _get_event_id(), @@ -106,7 +106,7 @@ def _create_event(user_id): def _join_event(user_id): - return FrozenEvent( + return make_event_from_dict( { "room_id": TEST_ROOM_ID, "event_id": _get_event_id(), @@ -119,7 +119,7 @@ def _join_event(user_id): def _power_levels_event(sender, content): - return FrozenEvent( + return make_event_from_dict( { "room_id": TEST_ROOM_ID, "event_id": _get_event_id(), @@ -132,7 +132,7 @@ def _power_levels_event(sender, content): def _random_state_event(sender): - return FrozenEvent( + return make_event_from_dict( { "room_id": TEST_ROOM_ID, "event_id": _get_event_id(), diff --git a/tests/test_federation.py b/tests/test_federation.py index 68684460c6..9b5cf562f3 100644 --- a/tests/test_federation.py +++ b/tests/test_federation.py @@ -2,7 +2,7 @@ from mock import Mock from twisted.internet.defer import ensureDeferred, maybeDeferred, succeed -from synapse.events import FrozenEvent +from synapse.events import make_event_from_dict from synapse.logging.context import LoggingContext from synapse.types import Requester, UserID from synapse.util import Clock @@ -43,7 +43,7 @@ class MessageAcceptTests(unittest.TestCase): ) )[0] - join_event = FrozenEvent( + join_event = make_event_from_dict( { "room_id": self.room_id, "sender": "@baduser:test.serv", @@ -105,7 +105,7 @@ class MessageAcceptTests(unittest.TestCase): )[0] # Now lie about an event - lying_event = FrozenEvent( + lying_event = make_event_from_dict( { "room_id": self.room_id, "sender": "@baduser:test.serv", diff --git a/tests/test_state.py b/tests/test_state.py index 1e4449fa1c..d1578fe581 100644 --- a/tests/test_state.py +++ b/tests/test_state.py @@ -20,7 +20,7 @@ from twisted.internet import defer from synapse.api.auth import Auth from synapse.api.constants import EventTypes, Membership from synapse.api.room_versions import RoomVersions -from synapse.events import FrozenEvent +from synapse.events import make_event_from_dict from synapse.events.snapshot import EventContext from synapse.state import StateHandler, StateResolutionHandler @@ -66,7 +66,7 @@ def create_event( d.update(kwargs) - event = FrozenEvent(d) + event = make_event_from_dict(d) return event -- cgit 1.4.1 From a92e703ab9d78aecc062e797f941bb7e206650a5 Mon Sep 17 00:00:00 2001 From: Patrick Cloke Date: Mon, 10 Feb 2020 16:35:26 -0500 Subject: Reject device display names that are too long (#6882) * Reject device display names that are too long. Too long is currently defined as 100 characters in length. * Add a regression test for rejecting a too long device display name. --- changelog.d/6882.misc | 1 + synapse/handlers/device.py | 14 +++++++++++++- tests/handlers/test_device.py | 18 ++++++++++++++++++ 3 files changed, 32 insertions(+), 1 deletion(-) create mode 100644 changelog.d/6882.misc (limited to 'tests') diff --git a/changelog.d/6882.misc b/changelog.d/6882.misc new file mode 100644 index 0000000000..e8382e36ae --- /dev/null +++ b/changelog.d/6882.misc @@ -0,0 +1 @@ +Reject device display names over 100 characters in length. diff --git a/synapse/handlers/device.py b/synapse/handlers/device.py index 6d8e48ed39..50cea3f378 100644 --- a/synapse/handlers/device.py +++ b/synapse/handlers/device.py @@ -26,6 +26,7 @@ from synapse.api.errors import ( FederationDeniedError, HttpResponseException, RequestSendFailed, + SynapseError, ) from synapse.logging.opentracing import log_kv, set_tag, trace from synapse.types import RoomStreamToken, get_domain_from_id @@ -39,6 +40,8 @@ from ._base import BaseHandler logger = logging.getLogger(__name__) +MAX_DEVICE_DISPLAY_NAME_LEN = 100 + class DeviceWorkerHandler(BaseHandler): def __init__(self, hs): @@ -404,9 +407,18 @@ class DeviceHandler(DeviceWorkerHandler): defer.Deferred: """ + # Reject a new displayname which is too long. + new_display_name = content.get("display_name") + if new_display_name and len(new_display_name) > MAX_DEVICE_DISPLAY_NAME_LEN: + raise SynapseError( + 400, + "Device display name is too long (max %i)" + % (MAX_DEVICE_DISPLAY_NAME_LEN,), + ) + try: yield self.store.update_device( - user_id, device_id, new_display_name=content.get("display_name") + user_id, device_id, new_display_name=new_display_name ) yield self.notify_device_update(user_id, [device_id]) except errors.StoreError as e: diff --git a/tests/handlers/test_device.py b/tests/handlers/test_device.py index a3aa0a1cf2..62b47f6574 100644 --- a/tests/handlers/test_device.py +++ b/tests/handlers/test_device.py @@ -160,6 +160,24 @@ class DeviceTestCase(unittest.HomeserverTestCase): res = self.get_success(self.handler.get_device(user1, "abc")) self.assertEqual(res["display_name"], "new display") + def test_update_device_too_long_display_name(self): + """Update a device with a display name that is invalid (too long).""" + self._record_users() + + # Request to update a device display name with a new value that is longer than allowed. + update = { + "display_name": "a" + * (synapse.handlers.device.MAX_DEVICE_DISPLAY_NAME_LEN + 1) + } + self.get_failure( + self.handler.update_device(user1, "abc", update), + synapse.api.errors.SynapseError, + ) + + # Ensure the display name was not updated. + res = self.get_success(self.handler.get_device(user1, "abc")) + self.assertEqual(res["display_name"], "display 2") + def test_update_unknown_device(self): update = {"display_name": "new_display"} res = self.handler.update_device("user_id", "unknown_device_id", update) -- cgit 1.4.1 From d8994942f28f5028e560f6aba52512fae3ca1a6a Mon Sep 17 00:00:00 2001 From: Andrew Morgan <1342360+anoadragon453@users.noreply.github.com> Date: Wed, 12 Feb 2020 18:14:10 +0000 Subject: Return a 404 for admin api user lookup if user not found (#6901) --- changelog.d/6901.misc | 1 + synapse/rest/admin/users.py | 5 ++++- tests/rest/admin/test_user.py | 16 ++++++++++++++++ 3 files changed, 21 insertions(+), 1 deletion(-) create mode 100644 changelog.d/6901.misc (limited to 'tests') diff --git a/changelog.d/6901.misc b/changelog.d/6901.misc new file mode 100644 index 0000000000..b2f12bbe86 --- /dev/null +++ b/changelog.d/6901.misc @@ -0,0 +1 @@ +Return a 404 instead of 200 for querying information of a non-existant user through the admin API. \ No newline at end of file diff --git a/synapse/rest/admin/users.py b/synapse/rest/admin/users.py index e75c5f1370..2107b5dc56 100644 --- a/synapse/rest/admin/users.py +++ b/synapse/rest/admin/users.py @@ -21,7 +21,7 @@ from six import text_type from six.moves import http_client from synapse.api.constants import UserTypes -from synapse.api.errors import Codes, SynapseError +from synapse.api.errors import Codes, NotFoundError, SynapseError from synapse.http.servlet import ( RestServlet, assert_params_in_dict, @@ -152,6 +152,9 @@ class UserRestServletV2(RestServlet): ret = await self.admin_handler.get_user(target_user) + if not ret: + raise NotFoundError("User not found") + return 200, ret async def on_PUT(self, request, user_id): diff --git a/tests/rest/admin/test_user.py b/tests/rest/admin/test_user.py index 3b5169b38d..490ce8f55d 100644 --- a/tests/rest/admin/test_user.py +++ b/tests/rest/admin/test_user.py @@ -401,6 +401,22 @@ class UserRestTestCase(unittest.HomeserverTestCase): self.assertEqual(403, int(channel.result["code"]), msg=channel.result["body"]) self.assertEqual("You are not a server admin", channel.json_body["error"]) + def test_user_does_not_exist(self): + """ + Tests that a lookup for a user that does not exist returns a 404 + """ + self.hs.config.registration_shared_secret = None + + request, channel = self.make_request( + "GET", + "/_synapse/admin/v2/users/@unknown_person:test", + access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(404, channel.code, msg=channel.json_body) + self.assertEqual("M_NOT_FOUND", channel.json_body["errcode"]) + def test_requester_is_admin(self): """ If the user is a server admin, a new user is created. -- cgit 1.4.1 From 49f877d32efc79cb40b2766cb052cf35bad31de5 Mon Sep 17 00:00:00 2001 From: Patrick Cloke Date: Fri, 14 Feb 2020 07:17:54 -0500 Subject: Filter the results of user directory searching via the spam checker (#6888) Add a method to the spam checker to filter the user directory results. --- changelog.d/6888.feature | 1 + docs/spam_checker.md | 3 ++ synapse/events/spamcheck.py | 27 ++++++++++ synapse/handlers/user_directory.py | 14 +++++- tests/handlers/test_user_directory.py | 92 +++++++++++++++++++++++++++++++++++ 5 files changed, 135 insertions(+), 2 deletions(-) create mode 100644 changelog.d/6888.feature (limited to 'tests') diff --git a/changelog.d/6888.feature b/changelog.d/6888.feature new file mode 100644 index 0000000000..1b7ac0c823 --- /dev/null +++ b/changelog.d/6888.feature @@ -0,0 +1 @@ +The result of a user directory search can now be filtered via the spam checker. diff --git a/docs/spam_checker.md b/docs/spam_checker.md index 97ff17f952..5b5f5000b7 100644 --- a/docs/spam_checker.md +++ b/docs/spam_checker.md @@ -54,6 +54,9 @@ class ExampleSpamChecker: def user_may_publish_room(self, userid, room_id): return True # allow publishing of all rooms + + def check_username_for_spam(self, user_profile): + return False # allow all usernames ``` ## Configuration diff --git a/synapse/events/spamcheck.py b/synapse/events/spamcheck.py index 5a907718d6..0a13fca9a4 100644 --- a/synapse/events/spamcheck.py +++ b/synapse/events/spamcheck.py @@ -15,6 +15,7 @@ # limitations under the License. import inspect +from typing import Dict from synapse.spam_checker_api import SpamCheckerApi @@ -125,3 +126,29 @@ class SpamChecker(object): return True return self.spam_checker.user_may_publish_room(userid, room_id) + + def check_username_for_spam(self, user_profile: Dict[str, str]) -> bool: + """Checks if a user ID or display name are considered "spammy" by this server. + + If the server considers a username spammy, then it will not be included in + user directory results. + + Args: + user_profile: The user information to check, it contains the keys: + * user_id + * display_name + * avatar_url + + Returns: + True if the user is spammy. + """ + if self.spam_checker is None: + return False + + # For backwards compatibility, if the method does not exist on the spam checker, fallback to not interfering. + checker = getattr(self.spam_checker, "check_username_for_spam", None) + if not checker: + return False + # Make a copy of the user profile object to ensure the spam checker + # cannot modify it. + return checker(user_profile.copy()) diff --git a/synapse/handlers/user_directory.py b/synapse/handlers/user_directory.py index 81aa58dc8c..722760c59d 100644 --- a/synapse/handlers/user_directory.py +++ b/synapse/handlers/user_directory.py @@ -52,6 +52,7 @@ class UserDirectoryHandler(StateDeltasHandler): self.is_mine_id = hs.is_mine_id self.update_user_directory = hs.config.update_user_directory self.search_all_users = hs.config.user_directory_search_all_users + self.spam_checker = hs.get_spam_checker() # The current position in the current_state_delta stream self.pos = None @@ -65,7 +66,7 @@ class UserDirectoryHandler(StateDeltasHandler): # we start populating the user directory self.clock.call_later(0, self.notify_new_event) - def search_users(self, user_id, search_term, limit): + async def search_users(self, user_id, search_term, limit): """Searches for users in directory Returns: @@ -82,7 +83,16 @@ class UserDirectoryHandler(StateDeltasHandler): ] } """ - return self.store.search_user_dir(user_id, search_term, limit) + results = await self.store.search_user_dir(user_id, search_term, limit) + + # Remove any spammy users from the results. + results["results"] = [ + user + for user in results["results"] + if not self.spam_checker.check_username_for_spam(user) + ] + + return results def notify_new_event(self): """Called when there may be more deltas to process diff --git a/tests/handlers/test_user_directory.py b/tests/handlers/test_user_directory.py index 26071059d2..0a4765fff4 100644 --- a/tests/handlers/test_user_directory.py +++ b/tests/handlers/test_user_directory.py @@ -147,6 +147,98 @@ class UserDirectoryTestCase(unittest.HomeserverTestCase): s = self.get_success(self.handler.search_users(u1, "user3", 10)) self.assertEqual(len(s["results"]), 0) + def test_spam_checker(self): + """ + A user which fails to the spam checks will not appear in search results. + """ + u1 = self.register_user("user1", "pass") + u1_token = self.login(u1, "pass") + u2 = self.register_user("user2", "pass") + u2_token = self.login(u2, "pass") + + # We do not add users to the directory until they join a room. + s = self.get_success(self.handler.search_users(u1, "user2", 10)) + self.assertEqual(len(s["results"]), 0) + + room = self.helper.create_room_as(u1, is_public=False, tok=u1_token) + self.helper.invite(room, src=u1, targ=u2, tok=u1_token) + self.helper.join(room, user=u2, tok=u2_token) + + # Check we have populated the database correctly. + shares_private = self.get_users_who_share_private_rooms() + public_users = self.get_users_in_public_rooms() + + self.assertEqual( + self._compress_shared(shares_private), set([(u1, u2, room), (u2, u1, room)]) + ) + self.assertEqual(public_users, []) + + # We get one search result when searching for user2 by user1. + s = self.get_success(self.handler.search_users(u1, "user2", 10)) + self.assertEqual(len(s["results"]), 1) + + # Configure a spam checker that does not filter any users. + spam_checker = self.hs.get_spam_checker() + + class AllowAll(object): + def check_username_for_spam(self, user_profile): + # Allow all users. + return False + + spam_checker.spam_checker = AllowAll() + + # The results do not change: + # We get one search result when searching for user2 by user1. + s = self.get_success(self.handler.search_users(u1, "user2", 10)) + self.assertEqual(len(s["results"]), 1) + + # Configure a spam checker that filters all users. + class BlockAll(object): + def check_username_for_spam(self, user_profile): + # All users are spammy. + return True + + spam_checker.spam_checker = BlockAll() + + # User1 now gets no search results for any of the other users. + s = self.get_success(self.handler.search_users(u1, "user2", 10)) + self.assertEqual(len(s["results"]), 0) + + def test_legacy_spam_checker(self): + """ + A spam checker without the expected method should be ignored. + """ + u1 = self.register_user("user1", "pass") + u1_token = self.login(u1, "pass") + u2 = self.register_user("user2", "pass") + u2_token = self.login(u2, "pass") + + # We do not add users to the directory until they join a room. + s = self.get_success(self.handler.search_users(u1, "user2", 10)) + self.assertEqual(len(s["results"]), 0) + + room = self.helper.create_room_as(u1, is_public=False, tok=u1_token) + self.helper.invite(room, src=u1, targ=u2, tok=u1_token) + self.helper.join(room, user=u2, tok=u2_token) + + # Check we have populated the database correctly. + shares_private = self.get_users_who_share_private_rooms() + public_users = self.get_users_in_public_rooms() + + self.assertEqual( + self._compress_shared(shares_private), set([(u1, u2, room), (u2, u1, room)]) + ) + self.assertEqual(public_users, []) + + # Configure a spam checker. + spam_checker = self.hs.get_spam_checker() + # The spam checker doesn't need any methods, so create a bare object. + spam_checker.spam_checker = object() + + # We get one search result when searching for user2 by user1. + s = self.get_success(self.handler.search_users(u1, "user2", 10)) + self.assertEqual(len(s["results"]), 1) + def _compress_shared(self, shared): """ Compress a list of users who share rooms dicts to a list of tuples. -- cgit 1.4.1 From 02e89021f58f931068ab0337de039181cc7f6569 Mon Sep 17 00:00:00 2001 From: Patrick Cloke Date: Fri, 14 Feb 2020 09:05:43 -0500 Subject: Convert the directory handler tests to use HomeserverTestCase (#6919) Convert directory handler tests to use HomeserverTestCase. --- changelog.d/6919.misc | 1 + tests/handlers/test_directory.py | 41 +++++++++++++++++----------------------- 2 files changed, 18 insertions(+), 24 deletions(-) create mode 100644 changelog.d/6919.misc (limited to 'tests') diff --git a/changelog.d/6919.misc b/changelog.d/6919.misc new file mode 100644 index 0000000000..aa2cd89998 --- /dev/null +++ b/changelog.d/6919.misc @@ -0,0 +1 @@ +Convert the directory handler tests to use HomeserverTestCase. diff --git a/tests/handlers/test_directory.py b/tests/handlers/test_directory.py index 91c7a17070..ee88cf5a4b 100644 --- a/tests/handlers/test_directory.py +++ b/tests/handlers/test_directory.py @@ -19,24 +19,16 @@ from mock import Mock from twisted.internet import defer from synapse.config.room_directory import RoomDirectoryConfig -from synapse.handlers.directory import DirectoryHandler from synapse.rest.client.v1 import directory, room from synapse.types import RoomAlias from tests import unittest -from tests.utils import setup_test_homeserver -class DirectoryHandlers(object): - def __init__(self, hs): - self.directory_handler = DirectoryHandler(hs) - - -class DirectoryTestCase(unittest.TestCase): +class DirectoryTestCase(unittest.HomeserverTestCase): """ Tests the directory service. """ - @defer.inlineCallbacks - def setUp(self): + def make_homeserver(self, reactor, clock): self.mock_federation = Mock() self.mock_registry = Mock() @@ -47,14 +39,12 @@ class DirectoryTestCase(unittest.TestCase): self.mock_registry.register_query_handler = register_query_handler - hs = yield setup_test_homeserver( - self.addCleanup, + hs = self.setup_test_homeserver( http_client=None, resource_for_federation=Mock(), federation_client=self.mock_federation, federation_registry=self.mock_registry, ) - hs.handlers = DirectoryHandlers(hs) self.handler = hs.get_handlers().directory_handler @@ -64,23 +54,25 @@ class DirectoryTestCase(unittest.TestCase): self.your_room = RoomAlias.from_string("#your-room:test") self.remote_room = RoomAlias.from_string("#another:remote") - @defer.inlineCallbacks + return hs + def test_get_local_association(self): - yield self.store.create_room_alias_association( - self.my_room, "!8765qwer:test", ["test"] + self.get_success( + self.store.create_room_alias_association( + self.my_room, "!8765qwer:test", ["test"] + ) ) - result = yield self.handler.get_association(self.my_room) + result = self.get_success(self.handler.get_association(self.my_room)) self.assertEquals({"room_id": "!8765qwer:test", "servers": ["test"]}, result) - @defer.inlineCallbacks def test_get_remote_association(self): self.mock_federation.make_query.return_value = defer.succeed( {"room_id": "!8765qwer:test", "servers": ["test", "remote"]} ) - result = yield self.handler.get_association(self.remote_room) + result = self.get_success(self.handler.get_association(self.remote_room)) self.assertEquals( {"room_id": "!8765qwer:test", "servers": ["test", "remote"]}, result @@ -93,14 +85,15 @@ class DirectoryTestCase(unittest.TestCase): ignore_backoff=True, ) - @defer.inlineCallbacks def test_incoming_fed_query(self): - yield self.store.create_room_alias_association( - self.your_room, "!8765asdf:test", ["test"] + self.get_success( + self.store.create_room_alias_association( + self.your_room, "!8765asdf:test", ["test"] + ) ) - response = yield self.query_handlers["directory"]( - {"room_alias": "#your-room:test"} + response = self.get_success( + self.handler.on_directory_query({"room_alias": "#your-room:test"}) ) self.assertEquals({"room_id": "!8765asdf:test", "servers": ["test"]}, response) -- cgit 1.4.1 From 43b2be9764bf244122f25a0f2bf8ec8c761f5cad Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Fri, 7 Feb 2020 13:08:34 +0000 Subject: Replace _event_dict_property with DictProperty this amounts to the same thing, but replaces `_event_dict` with `_dict`, and removes some of the function layers generated by `property`. --- synapse/events/__init__.py | 144 ++++++++++++++++++++++------------------ tests/storage/test_redaction.py | 2 +- 2 files changed, 80 insertions(+), 66 deletions(-) (limited to 'tests') diff --git a/synapse/events/__init__.py b/synapse/events/__init__.py index a842661a90..512254f65d 100644 --- a/synapse/events/__init__.py +++ b/synapse/events/__init__.py @@ -37,6 +37,65 @@ from synapse.util.frozenutils import freeze USE_FROZEN_DICTS = strtobool(os.environ.get("SYNAPSE_USE_FROZEN_DICTS", "0")) +class DictProperty: + """An object property which delegates to the `_dict` within its parent object.""" + + __slots__ = ["key"] + + def __init__(self, key: str): + self.key = key + + def __get__(self, instance, owner=None): + # if the property is accessed as a class property rather than an instance + # property, return the property itself rather than the value + if instance is None: + return self + try: + return instance._dict[self.key] + except KeyError as e1: + # We want this to look like a regular attribute error (mostly so that + # hasattr() works correctly), so we convert the KeyError into an + # AttributeError. + # + # To exclude the KeyError from the traceback, we explicitly + # 'raise from e1.__context__' (which is better than 'raise from None', + # becuase that would omit any *earlier* exceptions). + # + raise AttributeError( + "'%s' has no '%s' property" % (type(instance), self.key) + ) from e1.__context__ + + def __set__(self, instance, v): + instance._dict[self.key] = v + + def __delete__(self, instance): + try: + del instance._dict[self.key] + except KeyError as e1: + raise AttributeError( + "'%s' has no '%s' property" % (type(instance), self.key) + ) from e1.__context__ + + +class DefaultDictProperty(DictProperty): + """An extension of DictProperty which provides a default if the property is + not present in the parent's _dict. + + Note that this means that hasattr() on the property always returns True. + """ + + __slots__ = ["default"] + + def __init__(self, key, default): + super().__init__(key) + self.default = default + + def __get__(self, instance, owner=None): + if instance is None: + return self + return instance._dict.get(self.key, self.default) + + class _EventInternalMetadata(object): def __init__(self, internal_metadata_dict): self.__dict__ = dict(internal_metadata_dict) @@ -117,51 +176,6 @@ class _EventInternalMetadata(object): return getattr(self, "redacted", False) -_SENTINEL = object() - - -def _event_dict_property(key, default=_SENTINEL): - """Creates a new property for the given key that delegates access to - `self._event_dict`. - - The default is used if the key is missing from the `_event_dict`, if given, - otherwise an AttributeError will be raised. - - Note: If a default is given then `hasattr` will always return true. - """ - - # We want to be able to use hasattr with the event dict properties. - # However, (on python3) hasattr expects AttributeError to be raised. Hence, - # we need to transform the KeyError into an AttributeError - - def getter_raises(self): - try: - return self._event_dict[key] - except KeyError: - raise AttributeError(key) - - def getter_default(self): - return self._event_dict.get(key, default) - - def setter(self, v): - try: - self._event_dict[key] = v - except KeyError: - raise AttributeError(key) - - def delete(self): - try: - del self._event_dict[key] - except KeyError: - raise AttributeError(key) - - if default is _SENTINEL: - # No default given, so use the getter that raises - return property(getter_raises, setter, delete) - else: - return property(getter_default, setter, delete) - - class EventBase(object): def __init__( self, @@ -175,23 +189,23 @@ class EventBase(object): self.unsigned = unsigned self.rejected_reason = rejected_reason - self._event_dict = event_dict + self._dict = event_dict self.internal_metadata = _EventInternalMetadata(internal_metadata_dict) - auth_events = _event_dict_property("auth_events") - depth = _event_dict_property("depth") - content = _event_dict_property("content") - hashes = _event_dict_property("hashes") - origin = _event_dict_property("origin") - origin_server_ts = _event_dict_property("origin_server_ts") - prev_events = _event_dict_property("prev_events") - redacts = _event_dict_property("redacts", None) - room_id = _event_dict_property("room_id") - sender = _event_dict_property("sender") - state_key = _event_dict_property("state_key") - type = _event_dict_property("type") - user_id = _event_dict_property("sender") + auth_events = DictProperty("auth_events") + depth = DictProperty("depth") + content = DictProperty("content") + hashes = DictProperty("hashes") + origin = DictProperty("origin") + origin_server_ts = DictProperty("origin_server_ts") + prev_events = DictProperty("prev_events") + redacts = DefaultDictProperty("redacts", None) + room_id = DictProperty("room_id") + sender = DictProperty("sender") + state_key = DictProperty("state_key") + type = DictProperty("type") + user_id = DictProperty("sender") @property def event_id(self) -> str: @@ -205,13 +219,13 @@ class EventBase(object): return hasattr(self, "state_key") and self.state_key is not None def get_dict(self) -> JsonDict: - d = dict(self._event_dict) + d = dict(self._dict) d.update({"signatures": self.signatures, "unsigned": dict(self.unsigned)}) return d def get(self, key, default=None): - return self._event_dict.get(key, default) + return self._dict.get(key, default) def get_internal_metadata_dict(self): return self.internal_metadata.get_dict() @@ -233,16 +247,16 @@ class EventBase(object): raise AttributeError("Unrecognized attribute %s" % (instance,)) def __getitem__(self, field): - return self._event_dict[field] + return self._dict[field] def __contains__(self, field): - return field in self._event_dict + return field in self._dict def items(self): - return list(self._event_dict.items()) + return list(self._dict.items()) def keys(self): - return six.iterkeys(self._event_dict) + return six.iterkeys(self._dict) def prev_event_ids(self): """Returns the list of prev event IDs. The order matches the order diff --git a/tests/storage/test_redaction.py b/tests/storage/test_redaction.py index b9ee6ec1ec..db3667dc43 100644 --- a/tests/storage/test_redaction.py +++ b/tests/storage/test_redaction.py @@ -240,7 +240,7 @@ class RedactionTestCase(unittest.HomeserverTestCase): built_event = yield self._base_builder.build(prev_event_ids) built_event._event_id = self._event_id - built_event._event_dict["event_id"] = self._event_id + built_event._dict["event_id"] = self._event_id assert built_event.event_id == self._event_id return built_event -- cgit 1.4.1 From 3404ad289b1d2e5bc5c7f277f519b9698dbdaa15 Mon Sep 17 00:00:00 2001 From: Andrew Morgan <1342360+anoadragon453@users.noreply.github.com> Date: Mon, 17 Feb 2020 13:23:37 +0000 Subject: Raise the default power levels for invites, tombstones and server acls (#6834) --- changelog.d/6834.misc | 1 + synapse/handlers/room.py | 10 +++++++++- tests/rest/client/v1/test_rooms.py | 4 +++- 3 files changed, 13 insertions(+), 2 deletions(-) create mode 100644 changelog.d/6834.misc (limited to 'tests') diff --git a/changelog.d/6834.misc b/changelog.d/6834.misc new file mode 100644 index 0000000000..79acebe516 --- /dev/null +++ b/changelog.d/6834.misc @@ -0,0 +1 @@ +Change the default power levels of invites, tombstones and server ACLs for new rooms. \ No newline at end of file diff --git a/synapse/handlers/room.py b/synapse/handlers/room.py index ab07edd2fc..033083acac 100644 --- a/synapse/handlers/room.py +++ b/synapse/handlers/room.py @@ -64,18 +64,21 @@ class RoomCreationHandler(BaseHandler): "history_visibility": "shared", "original_invitees_have_ops": False, "guest_can_join": True, + "power_level_content_override": {"invite": 0}, }, RoomCreationPreset.TRUSTED_PRIVATE_CHAT: { "join_rules": JoinRules.INVITE, "history_visibility": "shared", "original_invitees_have_ops": True, "guest_can_join": True, + "power_level_content_override": {"invite": 0}, }, RoomCreationPreset.PUBLIC_CHAT: { "join_rules": JoinRules.PUBLIC, "history_visibility": "shared", "original_invitees_have_ops": False, "guest_can_join": False, + "power_level_content_override": {}, }, } @@ -829,19 +832,24 @@ class RoomCreationHandler(BaseHandler): # This will be reudundant on pre-MSC2260 rooms, since the # aliases event is special-cased. EventTypes.Aliases: 0, + EventTypes.Tombstone: 100, + EventTypes.ServerACL: 100, }, "events_default": 0, "state_default": 50, "ban": 50, "kick": 50, "redact": 50, - "invite": 0, + "invite": 50, } if config["original_invitees_have_ops"]: for invitee in invite_list: power_level_content["users"][invitee] = 100 + # Power levels overrides are defined per chat preset + power_level_content.update(config["power_level_content_override"]) + if power_level_content_override: power_level_content.update(power_level_content_override) diff --git a/tests/rest/client/v1/test_rooms.py b/tests/rest/client/v1/test_rooms.py index e3af280ba6..fb681a1db9 100644 --- a/tests/rest/client/v1/test_rooms.py +++ b/tests/rest/client/v1/test_rooms.py @@ -1612,7 +1612,9 @@ class ContextTestCase(unittest.HomeserverTestCase): def prepare(self, reactor, clock, homeserver): self.user_id = self.register_user("user", "password") self.tok = self.login("user", "password") - self.room_id = self.helper.create_room_as(self.user_id, tok=self.tok) + self.room_id = self.helper.create_room_as( + self.user_id, tok=self.tok, is_public=False + ) self.other_user_id = self.register_user("user2", "password") self.other_tok = self.login("user2", "password") -- cgit 1.4.1 From fe3941f6e33a17fa7cdf209a4370f4e805341db4 Mon Sep 17 00:00:00 2001 From: Patrick Cloke Date: Tue, 18 Feb 2020 07:29:44 -0500 Subject: Stop sending events when creating or deleting aliases (#6904) Stop sending events when creating or deleting associations (room aliases). Send an updated canonical alias event if one of the alt_aliases is deleted. --- changelog.d/6904.removal | 1 + synapse/handlers/directory.py | 75 ++++++++++--------- synapse/handlers/room.py | 6 +- tests/handlers/test_directory.py | 154 ++++++++++++++++++++++++++++++++++++++- 4 files changed, 194 insertions(+), 42 deletions(-) create mode 100644 changelog.d/6904.removal (limited to 'tests') diff --git a/changelog.d/6904.removal b/changelog.d/6904.removal new file mode 100644 index 0000000000..a5cc0c3605 --- /dev/null +++ b/changelog.d/6904.removal @@ -0,0 +1 @@ +Stop sending alias events during adding / removing aliases. Check alt_aliases in the latest canonical aliases event when deleting an alias. diff --git a/synapse/handlers/directory.py b/synapse/handlers/directory.py index 8c5980cb0c..f718388884 100644 --- a/synapse/handlers/directory.py +++ b/synapse/handlers/directory.py @@ -81,13 +81,7 @@ class DirectoryHandler(BaseHandler): @defer.inlineCallbacks def create_association( - self, - requester, - room_alias, - room_id, - servers=None, - send_event=True, - check_membership=True, + self, requester, room_alias, room_id, servers=None, check_membership=True, ): """Attempt to create a new alias @@ -97,7 +91,6 @@ class DirectoryHandler(BaseHandler): room_id (str) servers (list[str]|None): List of servers that others servers should try and join via - send_event (bool): Whether to send an updated m.room.aliases event check_membership (bool): Whether to check if the user is in the room before the alias can be set (if the server's config requires it). @@ -150,16 +143,9 @@ class DirectoryHandler(BaseHandler): ) yield self._create_association(room_alias, room_id, servers, creator=user_id) - if send_event: - try: - yield self.send_room_alias_update_event(requester, room_id) - except AuthError as e: - # sending the aliases event may fail due to the user not having - # permission in the room; this is permitted. - logger.info("Skipping updating aliases event due to auth error %s", e) @defer.inlineCallbacks - def delete_association(self, requester, room_alias, send_event=True): + def delete_association(self, requester, room_alias): """Remove an alias from the directory (this is only meant for human users; AS users should call @@ -168,9 +154,6 @@ class DirectoryHandler(BaseHandler): Args: requester (Requester): room_alias (RoomAlias): - send_event (bool): Whether to send an updated m.room.aliases event. - Note that, if we delete the canonical alias, we will always attempt - to send an m.room.canonical_alias event Returns: Deferred[unicode]: room id that the alias used to point to @@ -206,9 +189,6 @@ class DirectoryHandler(BaseHandler): room_id = yield self._delete_association(room_alias) try: - if send_event: - yield self.send_room_alias_update_event(requester, room_id) - yield self._update_canonical_alias( requester, requester.user.to_string(), room_id, room_alias ) @@ -319,25 +299,50 @@ class DirectoryHandler(BaseHandler): @defer.inlineCallbacks def _update_canonical_alias(self, requester, user_id, room_id, room_alias): + """ + Send an updated canonical alias event if the removed alias was set as + the canonical alias or listed in the alt_aliases field. + """ alias_event = yield self.state.get_current_state( room_id, EventTypes.CanonicalAlias, "" ) - alias_str = room_alias.to_string() - if not alias_event or alias_event.content.get("alias", "") != alias_str: + # There is no canonical alias, nothing to do. + if not alias_event: return - yield self.event_creation_handler.create_and_send_nonmember_event( - requester, - { - "type": EventTypes.CanonicalAlias, - "state_key": "", - "room_id": room_id, - "sender": user_id, - "content": {}, - }, - ratelimit=False, - ) + # Obtain a mutable version of the event content. + content = dict(alias_event.content) + send_update = False + + # Remove the alias property if it matches the removed alias. + alias_str = room_alias.to_string() + if alias_event.content.get("alias", "") == alias_str: + send_update = True + content.pop("alias", "") + + # Filter alt_aliases for the removed alias. + alt_aliases = content.pop("alt_aliases", None) + # If the aliases are not a list (or not found) do not attempt to modify + # the list. + if isinstance(alt_aliases, list): + send_update = True + alt_aliases = [alias for alias in alt_aliases if alias != alias_str] + if alt_aliases: + content["alt_aliases"] = alt_aliases + + if send_update: + yield self.event_creation_handler.create_and_send_nonmember_event( + requester, + { + "type": EventTypes.CanonicalAlias, + "state_key": "", + "room_id": room_id, + "sender": user_id, + "content": content, + }, + ratelimit=False, + ) @defer.inlineCallbacks def get_association_from_room_alias(self, room_alias): diff --git a/synapse/handlers/room.py b/synapse/handlers/room.py index 033083acac..49ec2f48bc 100644 --- a/synapse/handlers/room.py +++ b/synapse/handlers/room.py @@ -478,9 +478,7 @@ class RoomCreationHandler(BaseHandler): for alias_str in aliases: alias = RoomAlias.from_string(alias_str) try: - yield directory_handler.delete_association( - requester, alias, send_event=False - ) + yield directory_handler.delete_association(requester, alias) removed_aliases.append(alias_str) except SynapseError as e: logger.warning("Unable to remove alias %s from old room: %s", alias, e) @@ -511,7 +509,6 @@ class RoomCreationHandler(BaseHandler): RoomAlias.from_string(alias), new_room_id, servers=(self.hs.hostname,), - send_event=False, check_membership=False, ) logger.info("Moved alias %s to new room", alias) @@ -664,7 +661,6 @@ class RoomCreationHandler(BaseHandler): room_id=room_id, room_alias=room_alias, servers=[self.hs.hostname], - send_event=False, check_membership=False, ) diff --git a/tests/handlers/test_directory.py b/tests/handlers/test_directory.py index ee88cf5a4b..27b916aed4 100644 --- a/tests/handlers/test_directory.py +++ b/tests/handlers/test_directory.py @@ -18,9 +18,11 @@ from mock import Mock from twisted.internet import defer +import synapse.api.errors +from synapse.api.constants import EventTypes from synapse.config.room_directory import RoomDirectoryConfig -from synapse.rest.client.v1 import directory, room -from synapse.types import RoomAlias +from synapse.rest.client.v1 import directory, login, room +from synapse.types import RoomAlias, create_requester from tests import unittest @@ -85,6 +87,38 @@ class DirectoryTestCase(unittest.HomeserverTestCase): ignore_backoff=True, ) + def test_delete_alias_not_allowed(self): + room_id = "!8765qwer:test" + self.get_success( + self.store.create_room_alias_association(self.my_room, room_id, ["test"]) + ) + + self.get_failure( + self.handler.delete_association( + create_requester("@user:test"), self.my_room + ), + synapse.api.errors.AuthError, + ) + + def test_delete_alias(self): + room_id = "!8765qwer:test" + user_id = "@user:test" + self.get_success( + self.store.create_room_alias_association( + self.my_room, room_id, ["test"], user_id + ) + ) + + result = self.get_success( + self.handler.delete_association(create_requester(user_id), self.my_room) + ) + self.assertEquals(room_id, result) + + # The alias should not be found. + self.get_failure( + self.handler.get_association(self.my_room), synapse.api.errors.SynapseError + ) + def test_incoming_fed_query(self): self.get_success( self.store.create_room_alias_association( @@ -99,6 +133,122 @@ class DirectoryTestCase(unittest.HomeserverTestCase): self.assertEquals({"room_id": "!8765asdf:test", "servers": ["test"]}, response) +class CanonicalAliasTestCase(unittest.HomeserverTestCase): + """Test modifications of the canonical alias when delete aliases. + """ + + servlets = [ + synapse.rest.admin.register_servlets, + login.register_servlets, + room.register_servlets, + directory.register_servlets, + ] + + def prepare(self, reactor, clock, hs): + self.store = hs.get_datastore() + self.handler = hs.get_handlers().directory_handler + self.state_handler = hs.get_state_handler() + + # Create user + self.admin_user = self.register_user("admin", "pass", admin=True) + self.admin_user_tok = self.login("admin", "pass") + + # Create a test room + self.room_id = self.helper.create_room_as( + self.admin_user, tok=self.admin_user_tok + ) + + self.test_alias = "#test:test" + self.room_alias = RoomAlias.from_string(self.test_alias) + + # Create a new alias to this room. + self.get_success( + self.store.create_room_alias_association( + self.room_alias, self.room_id, ["test"], self.admin_user + ) + ) + + def test_remove_alias(self): + """Removing an alias that is the canonical alias should remove it there too.""" + # Set this new alias as the canonical alias for this room + self.helper.send_state( + self.room_id, + "m.room.canonical_alias", + {"alias": self.test_alias, "alt_aliases": [self.test_alias]}, + tok=self.admin_user_tok, + ) + + data = self.get_success( + self.state_handler.get_current_state( + self.room_id, EventTypes.CanonicalAlias, "" + ) + ) + self.assertEqual(data["content"]["alias"], self.test_alias) + self.assertEqual(data["content"]["alt_aliases"], [self.test_alias]) + + # Finally, delete the alias. + self.get_success( + self.handler.delete_association( + create_requester(self.admin_user), self.room_alias + ) + ) + + data = self.get_success( + self.state_handler.get_current_state( + self.room_id, EventTypes.CanonicalAlias, "" + ) + ) + self.assertNotIn("alias", data["content"]) + self.assertNotIn("alt_aliases", data["content"]) + + def test_remove_other_alias(self): + """Removing an alias listed as in alt_aliases should remove it there too.""" + # Create a second alias. + other_test_alias = "#test2:test" + other_room_alias = RoomAlias.from_string(other_test_alias) + self.get_success( + self.store.create_room_alias_association( + other_room_alias, self.room_id, ["test"], self.admin_user + ) + ) + + # Set the alias as the canonical alias for this room. + self.helper.send_state( + self.room_id, + "m.room.canonical_alias", + { + "alias": self.test_alias, + "alt_aliases": [self.test_alias, other_test_alias], + }, + tok=self.admin_user_tok, + ) + + data = self.get_success( + self.state_handler.get_current_state( + self.room_id, EventTypes.CanonicalAlias, "" + ) + ) + self.assertEqual(data["content"]["alias"], self.test_alias) + self.assertEqual( + data["content"]["alt_aliases"], [self.test_alias, other_test_alias] + ) + + # Delete the second alias. + self.get_success( + self.handler.delete_association( + create_requester(self.admin_user), other_room_alias + ) + ) + + data = self.get_success( + self.state_handler.get_current_state( + self.room_id, EventTypes.CanonicalAlias, "" + ) + ) + self.assertEqual(data["content"]["alias"], self.test_alias) + self.assertEqual(data["content"]["alt_aliases"], [self.test_alias]) + + class TestCreateAliasACL(unittest.HomeserverTestCase): user_id = "@test:test" -- cgit 1.4.1 From adfaea8c698a38ffe14ac682a946abc9f8152635 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Tue, 18 Feb 2020 16:23:25 +0000 Subject: Implement GET /_matrix/client/r0/rooms/{roomId}/aliases (#6939) per matrix-org/matrix-doc#2432 --- changelog.d/6939.feature | 1 + synapse/handlers/directory.py | 17 ++++++++- synapse/rest/client/v1/room.py | 23 +++++++++++++ tests/rest/client/v1/test_rooms.py | 70 +++++++++++++++++++++++++++++++++++++- tests/unittest.py | 28 ++++++++++----- 5 files changed, 128 insertions(+), 11 deletions(-) create mode 100644 changelog.d/6939.feature (limited to 'tests') diff --git a/changelog.d/6939.feature b/changelog.d/6939.feature new file mode 100644 index 0000000000..40fe7fc9a9 --- /dev/null +++ b/changelog.d/6939.feature @@ -0,0 +1 @@ +Implement `GET /_matrix/client/r0/rooms/{roomId}/aliases` endpoint as per [MSC2432](https://github.com/matrix-org/matrix-doc/pull/2432). diff --git a/synapse/handlers/directory.py b/synapse/handlers/directory.py index f718388884..3f8c792149 100644 --- a/synapse/handlers/directory.py +++ b/synapse/handlers/directory.py @@ -16,6 +16,7 @@ import logging import string +from typing import List from twisted.internet import defer @@ -28,7 +29,7 @@ from synapse.api.errors import ( StoreError, SynapseError, ) -from synapse.types import RoomAlias, UserID, get_domain_from_id +from synapse.types import Requester, RoomAlias, UserID, get_domain_from_id from ._base import BaseHandler @@ -452,3 +453,17 @@ class DirectoryHandler(BaseHandler): yield self.store.set_room_is_public_appservice( room_id, appservice_id, network_id, visibility == "public" ) + + async def get_aliases_for_room( + self, requester: Requester, room_id: str + ) -> List[str]: + """ + Get a list of the aliases that currently point to this room on this server + """ + # allow access to server admins and current members of the room + is_admin = await self.auth.is_server_admin(requester.user) + if not is_admin: + await self.auth.check_joined_room(room_id, requester.user.to_string()) + + aliases = await self.store.get_aliases_for_room(room_id) + return aliases diff --git a/synapse/rest/client/v1/room.py b/synapse/rest/client/v1/room.py index 6f31584c51..143dc738c6 100644 --- a/synapse/rest/client/v1/room.py +++ b/synapse/rest/client/v1/room.py @@ -45,6 +45,10 @@ from synapse.storage.state import StateFilter from synapse.streams.config import PaginationConfig from synapse.types import RoomAlias, RoomID, StreamToken, ThirdPartyInstanceID, UserID +MYPY = False +if MYPY: + import synapse.server + logger = logging.getLogger(__name__) @@ -843,6 +847,24 @@ class RoomTypingRestServlet(RestServlet): return 200, {} +class RoomAliasListServlet(RestServlet): + PATTERNS = client_patterns("/rooms/(?P[^/]*)/aliases", unstable=False) + + def __init__(self, hs: "synapse.server.HomeServer"): + super().__init__() + self.auth = hs.get_auth() + self.directory_handler = hs.get_handlers().directory_handler + + async def on_GET(self, request, room_id): + requester = await self.auth.get_user_by_req(request) + + alias_list = await self.directory_handler.get_aliases_for_room( + requester, room_id + ) + + return 200, {"aliases": alias_list} + + class SearchRestServlet(RestServlet): PATTERNS = client_patterns("/search$", v1=True) @@ -931,6 +953,7 @@ def register_servlets(hs, http_server): JoinedRoomsRestServlet(hs).register(http_server) RoomEventServlet(hs).register(http_server) RoomEventContextServlet(hs).register(http_server) + RoomAliasListServlet(hs).register(http_server) def register_deprecated_servlets(hs, http_server): diff --git a/tests/rest/client/v1/test_rooms.py b/tests/rest/client/v1/test_rooms.py index fb681a1db9..fb08a45d27 100644 --- a/tests/rest/client/v1/test_rooms.py +++ b/tests/rest/client/v1/test_rooms.py @@ -28,8 +28,9 @@ from twisted.internet import defer import synapse.rest.admin from synapse.api.constants import EventContentFields, EventTypes, Membership from synapse.handlers.pagination import PurgeStatus -from synapse.rest.client.v1 import login, profile, room +from synapse.rest.client.v1 import directory, login, profile, room from synapse.rest.client.v2_alpha import account +from synapse.types import JsonDict, RoomAlias from synapse.util.stringutils import random_string from tests import unittest @@ -1726,3 +1727,70 @@ class ContextTestCase(unittest.HomeserverTestCase): self.assertEqual(len(events_after), 2, events_after) self.assertDictEqual(events_after[0].get("content"), {}, events_after[0]) self.assertEqual(events_after[1].get("content"), {}, events_after[1]) + + +class DirectoryTestCase(unittest.HomeserverTestCase): + + servlets = [ + synapse.rest.admin.register_servlets_for_client_rest_resource, + directory.register_servlets, + login.register_servlets, + room.register_servlets, + ] + + def prepare(self, reactor, clock, homeserver): + self.room_owner = self.register_user("room_owner", "test") + self.room_owner_tok = self.login("room_owner", "test") + + self.room_id = self.helper.create_room_as( + self.room_owner, tok=self.room_owner_tok + ) + + def test_no_aliases(self): + res = self._get_aliases(self.room_owner_tok) + self.assertEqual(res["aliases"], []) + + def test_not_in_room(self): + self.register_user("user", "test") + user_tok = self.login("user", "test") + res = self._get_aliases(user_tok, expected_code=403) + self.assertEqual(res["errcode"], "M_FORBIDDEN") + + def test_with_aliases(self): + alias1 = self._random_alias() + alias2 = self._random_alias() + + self._set_alias_via_directory(alias1) + self._set_alias_via_directory(alias2) + + res = self._get_aliases(self.room_owner_tok) + self.assertEqual(set(res["aliases"]), {alias1, alias2}) + + def _get_aliases(self, access_token: str, expected_code: int = 200) -> JsonDict: + """Calls the endpoint under test. returns the json response object.""" + request, channel = self.make_request( + "GET", + "/_matrix/client/r0/rooms/%s/aliases" % (self.room_id,), + access_token=access_token, + ) + self.render(request) + self.assertEqual(channel.code, expected_code, channel.result) + res = channel.json_body + self.assertIsInstance(res, dict) + if expected_code == 200: + self.assertIsInstance(res["aliases"], list) + return res + + def _random_alias(self) -> str: + return RoomAlias(random_string(5), self.hs.hostname).to_string() + + def _set_alias_via_directory(self, alias: str, expected_code: int = 200): + url = "/_matrix/client/r0/directory/room/" + alias + data = {"room_id": self.room_id} + request_data = json.dumps(data) + + request, channel = self.make_request( + "PUT", url, request_data, access_token=self.room_owner_tok + ) + self.render(request) + self.assertEqual(channel.code, expected_code, channel.result) diff --git a/tests/unittest.py b/tests/unittest.py index 98bf27d39c..8816a4d152 100644 --- a/tests/unittest.py +++ b/tests/unittest.py @@ -21,6 +21,7 @@ import hmac import inspect import logging import time +from typing import Optional, Tuple, Type, TypeVar, Union from mock import Mock @@ -42,7 +43,13 @@ from synapse.server import HomeServer from synapse.types import Requester, UserID, create_requester from synapse.util.ratelimitutils import FederationRateLimiter -from tests.server import get_clock, make_request, render, setup_test_homeserver +from tests.server import ( + FakeChannel, + get_clock, + make_request, + render, + setup_test_homeserver, +) from tests.test_utils.logging_setup import setup_logging from tests.utils import default_config, setupdb @@ -71,6 +78,9 @@ def around(target): return _around +T = TypeVar("T") + + class TestCase(unittest.TestCase): """A subclass of twisted.trial's TestCase which looks for 'loglevel' attributes on both itself and its individual test methods, to override the @@ -334,14 +344,14 @@ class HomeserverTestCase(TestCase): def make_request( self, - method, - path, - content=b"", - access_token=None, - request=SynapseRequest, - shorthand=True, - federation_auth_origin=None, - ): + method: Union[bytes, str], + path: Union[bytes, str], + content: Union[bytes, dict] = b"", + access_token: Optional[str] = None, + request: Type[T] = SynapseRequest, + shorthand: bool = True, + federation_auth_origin: str = None, + ) -> Tuple[T, FakeChannel]: """ Create a SynapseRequest at the path using the method and containing the given content. -- cgit 1.4.1 From b58d17e44f9d9ff7e70578e0f4e328bb9113ec7e Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Tue, 18 Feb 2020 23:13:29 +0000 Subject: Refactor the membership check methods in Auth these were getting a bit unwieldy, so let's combine `check_joined_room` and `check_user_was_in_room` into a single `check_user_in_room`. --- synapse/api/auth.py | 80 +++++++++++++++++++--------------------- synapse/handlers/initial_sync.py | 31 +++------------- synapse/handlers/typing.py | 4 +- tests/handlers/test_typing.py | 4 +- 4 files changed, 46 insertions(+), 73 deletions(-) (limited to 'tests') diff --git a/synapse/api/auth.py b/synapse/api/auth.py index 8b1277ad02..de7b75ca36 100644 --- a/synapse/api/auth.py +++ b/synapse/api/auth.py @@ -14,6 +14,7 @@ # limitations under the License. import logging +from typing import Optional from six import itervalues @@ -35,6 +36,7 @@ from synapse.api.errors import ( ) from synapse.api.room_versions import KNOWN_ROOM_VERSIONS from synapse.config.server import is_threepid_reserved +from synapse.events import EventBase from synapse.types import StateMap, UserID from synapse.util.caches import CACHE_SIZE_FACTOR, register_cache from synapse.util.caches.lrucache import LruCache @@ -92,20 +94,34 @@ class Auth(object): ) @defer.inlineCallbacks - def check_joined_room(self, room_id, user_id, current_state=None): - """Check if the user is currently joined in the room + def check_user_in_room( + self, + room_id: str, + user_id: str, + current_state: Optional[StateMap[EventBase]] = None, + allow_departed_users: bool = False, + ): + """Check if the user is in the room, or was at some point. Args: - room_id(str): The room to check. - user_id(str): The user to check. - current_state(dict): Optional map of the current state of the room. + room_id: The room to check. + + user_id: The user to check. + + current_state: Optional map of the current state of the room. If provided then that map is used to check whether they are a member of the room. Otherwise the current membership is loaded from the database. + + allow_departed_users: if True, accept users that were previously + members but have now departed. + Raises: - AuthError if the user is not in the room. + AuthError if the user is/was not in the room. Returns: - A deferred membership event for the user if the user is in - the room. + Deferred[Optional[EventBase]]: + Membership event for the user if the user was in the + room. This will be the join event if they are currently joined to + the room. This will be the leave event if they have left the room. """ if current_state: member = current_state.get((EventTypes.Member, user_id), None) @@ -113,37 +129,19 @@ class Auth(object): member = yield self.state.get_current_state( room_id=room_id, event_type=EventTypes.Member, state_key=user_id ) - - self._check_joined_room(member, user_id, room_id) - return member - - @defer.inlineCallbacks - def check_user_was_in_room(self, room_id, user_id): - """Check if the user was in the room at some point. - Args: - room_id(str): The room to check. - user_id(str): The user to check. - Raises: - AuthError if the user was never in the room. - Returns: - A deferred membership event for the user if the user was in the - room. This will be the join event if they are currently joined to - the room. This will be the leave event if they have left the room. - """ - member = yield self.state.get_current_state( - room_id=room_id, event_type=EventTypes.Member, state_key=user_id - ) membership = member.membership if member else None - if membership not in (Membership.JOIN, Membership.LEAVE): - raise AuthError(403, "User %s not in room %s" % (user_id, room_id)) + if membership == Membership.JOIN: + return member - if membership == Membership.LEAVE: + # XXX this looks totally bogus. Why do we not allow users who have been banned, + # or those who were members previously and have been re-invited? + if allow_departed_users and membership == Membership.LEAVE: forgot = yield self.store.did_forget(user_id, room_id) - if forgot: - raise AuthError(403, "User %s not in room %s" % (user_id, room_id)) + if not forgot: + return member - return member + raise AuthError(403, "User %s not in room %s" % (user_id, room_id)) @defer.inlineCallbacks def check_host_in_room(self, room_id, host): @@ -151,12 +149,6 @@ class Auth(object): latest_event_ids = yield self.store.is_host_joined(room_id, host) return latest_event_ids - def _check_joined_room(self, member, user_id, room_id): - if not member or member.membership != Membership.JOIN: - raise AuthError( - 403, "User %s not in room %s (%s)" % (user_id, room_id, repr(member)) - ) - def can_federate(self, event, auth_events): creation_event = auth_events.get((EventTypes.Create, "")) @@ -560,7 +552,7 @@ class Auth(object): return True user_id = user.to_string() - yield self.check_joined_room(room_id, user_id) + yield self.check_user_in_room(room_id, user_id) # We currently require the user is a "moderator" in the room. We do this # by checking if they would (theoretically) be able to change the @@ -645,12 +637,14 @@ class Auth(object): """ try: - # check_user_was_in_room will return the most recent membership + # check_user_in_room will return the most recent membership # event for the user if: # * The user is a non-guest user, and was ever in the room # * The user is a guest user, and has joined the room # else it will throw. - member_event = yield self.check_user_was_in_room(room_id, user_id) + member_event = yield self.check_user_in_room( + room_id, user_id, allow_departed_users=True + ) return member_event.membership, member_event.event_id except AuthError: visibility = yield self.state.get_current_state( diff --git a/synapse/handlers/initial_sync.py b/synapse/handlers/initial_sync.py index 2e6755f19c..b7c6a921d9 100644 --- a/synapse/handlers/initial_sync.py +++ b/synapse/handlers/initial_sync.py @@ -18,7 +18,7 @@ import logging from twisted.internet import defer from synapse.api.constants import EventTypes, Membership -from synapse.api.errors import AuthError, Codes, SynapseError +from synapse.api.errors import SynapseError from synapse.events.validator import EventValidator from synapse.handlers.presence import format_user_presence_state from synapse.logging.context import make_deferred_yieldable, run_in_background @@ -274,9 +274,10 @@ class InitialSyncHandler(BaseHandler): user_id = requester.user.to_string() - membership, member_event_id = await self._check_in_room_or_world_readable( - room_id, user_id - ) + ( + membership, + member_event_id, + ) = await self.auth.check_user_in_room_or_world_readable(room_id, user_id) is_peeking = member_event_id is None if membership == Membership.JOIN: @@ -433,25 +434,3 @@ class InitialSyncHandler(BaseHandler): ret["membership"] = membership return ret - - async def _check_in_room_or_world_readable(self, room_id, user_id): - try: - # check_user_was_in_room will return the most recent membership - # event for the user if: - # * The user is a non-guest user, and was ever in the room - # * The user is a guest user, and has joined the room - # else it will throw. - member_event = await self.auth.check_user_was_in_room(room_id, user_id) - return member_event.membership, member_event.event_id - except AuthError: - visibility = await self.state_handler.get_current_state( - room_id, EventTypes.RoomHistoryVisibility, "" - ) - if ( - visibility - and visibility.content["history_visibility"] == "world_readable" - ): - return Membership.JOIN, None - raise AuthError( - 403, "Guest access not allowed", errcode=Codes.GUEST_ACCESS_FORBIDDEN - ) diff --git a/synapse/handlers/typing.py b/synapse/handlers/typing.py index d5ca9cb07b..5406618431 100644 --- a/synapse/handlers/typing.py +++ b/synapse/handlers/typing.py @@ -125,7 +125,7 @@ class TypingHandler(object): if target_user_id != auth_user_id: raise AuthError(400, "Cannot set another user's typing state") - yield self.auth.check_joined_room(room_id, target_user_id) + yield self.auth.check_user_in_room(room_id, target_user_id) logger.debug("%s has started typing in %s", target_user_id, room_id) @@ -155,7 +155,7 @@ class TypingHandler(object): if target_user_id != auth_user_id: raise AuthError(400, "Cannot set another user's typing state") - yield self.auth.check_joined_room(room_id, target_user_id) + yield self.auth.check_user_in_room(room_id, target_user_id) logger.debug("%s has stopped typing in %s", target_user_id, room_id) diff --git a/tests/handlers/test_typing.py b/tests/handlers/test_typing.py index 2767b0497a..140cc0a3c2 100644 --- a/tests/handlers/test_typing.py +++ b/tests/handlers/test_typing.py @@ -122,11 +122,11 @@ class TypingNotificationsTestCase(unittest.HomeserverTestCase): self.room_members = [] - def check_joined_room(room_id, user_id): + def check_user_in_room(room_id, user_id): if user_id not in [u.to_string() for u in self.room_members]: raise AuthError(401, "User is not in the room") - hs.get_auth().check_joined_room = check_joined_room + hs.get_auth().check_user_in_room = check_user_in_room def get_joined_hosts_for_room(room_id): return set(member.domain for member in self.room_members) -- cgit 1.4.1 From 709e81f5183d8ff67d86f4569234cb4a8be7a8d4 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Tue, 18 Feb 2020 23:15:54 +0000 Subject: Make room alias lists peekable As per https://github.com/matrix-org/matrix-doc/pull/2432#pullrequestreview-360566830, make room alias lists accessible to users outside world_readable rooms. --- synapse/handlers/directory.py | 4 +++- tests/rest/client/v1/test_rooms.py | 17 +++++++++++++++++ 2 files changed, 20 insertions(+), 1 deletion(-) (limited to 'tests') diff --git a/synapse/handlers/directory.py b/synapse/handlers/directory.py index 3f8c792149..db2104c5f6 100644 --- a/synapse/handlers/directory.py +++ b/synapse/handlers/directory.py @@ -463,7 +463,9 @@ class DirectoryHandler(BaseHandler): # allow access to server admins and current members of the room is_admin = await self.auth.is_server_admin(requester.user) if not is_admin: - await self.auth.check_joined_room(room_id, requester.user.to_string()) + await self.auth.check_user_in_room_or_world_readable( + room_id, requester.user.to_string() + ) aliases = await self.store.get_aliases_for_room(room_id) return aliases diff --git a/tests/rest/client/v1/test_rooms.py b/tests/rest/client/v1/test_rooms.py index fb08a45d27..8e389eb6c9 100644 --- a/tests/rest/client/v1/test_rooms.py +++ b/tests/rest/client/v1/test_rooms.py @@ -1766,6 +1766,23 @@ class DirectoryTestCase(unittest.HomeserverTestCase): res = self._get_aliases(self.room_owner_tok) self.assertEqual(set(res["aliases"]), {alias1, alias2}) + def test_peekable_room(self): + alias1 = self._random_alias() + self._set_alias_via_directory(alias1) + + self.helper.send_state( + self.room_id, + EventTypes.RoomHistoryVisibility, + body={"history_visibility": "world_readable"}, + tok=self.room_owner_tok, + ) + + self.register_user("user", "test") + user_tok = self.login("user", "test") + + res = self._get_aliases(user_tok) + self.assertEqual(res["aliases"], [alias1]) + def _get_aliases(self, access_token: str, expected_code: int = 200) -> JsonDict: """Calls the endpoint under test. returns the json response object.""" request, channel = self.make_request( -- cgit 1.4.1 From 880aaac1d82695b1a89f22f1f86c7f295ca205e0 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Wed, 19 Feb 2020 10:40:27 +0000 Subject: Move MSC2432 stuff onto unstable prefix (#6948) it's not in the spec yet, so needs to be unstable. Also add a feature flag for it. Also add a test for admin users. --- changelog.d/6948.feature | 1 + synapse/rest/client/v1/room.py | 8 +++++++- synapse/rest/client/versions.py | 2 ++ tests/rest/client/v1/test_rooms.py | 16 +++++++++++++--- 4 files changed, 23 insertions(+), 4 deletions(-) create mode 100644 changelog.d/6948.feature (limited to 'tests') diff --git a/changelog.d/6948.feature b/changelog.d/6948.feature new file mode 100644 index 0000000000..40fe7fc9a9 --- /dev/null +++ b/changelog.d/6948.feature @@ -0,0 +1 @@ +Implement `GET /_matrix/client/r0/rooms/{roomId}/aliases` endpoint as per [MSC2432](https://github.com/matrix-org/matrix-doc/pull/2432). diff --git a/synapse/rest/client/v1/room.py b/synapse/rest/client/v1/room.py index 143dc738c6..64f51406fb 100644 --- a/synapse/rest/client/v1/room.py +++ b/synapse/rest/client/v1/room.py @@ -16,6 +16,7 @@ """ This module contains REST servlets to do with rooms: /rooms/ """ import logging +import re from typing import List, Optional from six.moves.urllib import parse as urlparse @@ -848,7 +849,12 @@ class RoomTypingRestServlet(RestServlet): class RoomAliasListServlet(RestServlet): - PATTERNS = client_patterns("/rooms/(?P[^/]*)/aliases", unstable=False) + PATTERNS = [ + re.compile( + r"^/_matrix/client/unstable/org\.matrix\.msc2432" + r"/rooms/(?P[^/]*)/aliases" + ), + ] def __init__(self, hs: "synapse.server.HomeServer"): super().__init__() diff --git a/synapse/rest/client/versions.py b/synapse/rest/client/versions.py index 3eeb3607f4..d90a6a890b 100644 --- a/synapse/rest/client/versions.py +++ b/synapse/rest/client/versions.py @@ -72,6 +72,8 @@ class VersionsRestServlet(RestServlet): "org.matrix.label_based_filtering": True, # Implements support for cross signing as described in MSC1756 "org.matrix.e2e_cross_signing": True, + # Implements additional endpoints as described in MSC2432 + "org.matrix.msc2432": True, }, }, ) diff --git a/tests/rest/client/v1/test_rooms.py b/tests/rest/client/v1/test_rooms.py index fb08a45d27..f82655677c 100644 --- a/tests/rest/client/v1/test_rooms.py +++ b/tests/rest/client/v1/test_rooms.py @@ -1729,8 +1729,7 @@ class ContextTestCase(unittest.HomeserverTestCase): self.assertEqual(events_after[1].get("content"), {}, events_after[1]) -class DirectoryTestCase(unittest.HomeserverTestCase): - +class RoomAliasListTestCase(unittest.HomeserverTestCase): servlets = [ synapse.rest.admin.register_servlets_for_client_rest_resource, directory.register_servlets, @@ -1756,6 +1755,16 @@ class DirectoryTestCase(unittest.HomeserverTestCase): res = self._get_aliases(user_tok, expected_code=403) self.assertEqual(res["errcode"], "M_FORBIDDEN") + def test_admin_user(self): + alias1 = self._random_alias() + self._set_alias_via_directory(alias1) + + self.register_user("user", "test", admin=True) + user_tok = self.login("user", "test") + + res = self._get_aliases(user_tok) + self.assertEqual(res["aliases"], [alias1]) + def test_with_aliases(self): alias1 = self._random_alias() alias2 = self._random_alias() @@ -1770,7 +1779,8 @@ class DirectoryTestCase(unittest.HomeserverTestCase): """Calls the endpoint under test. returns the json response object.""" request, channel = self.make_request( "GET", - "/_matrix/client/r0/rooms/%s/aliases" % (self.room_id,), + "/_matrix/client/unstable/org.matrix.msc2432/rooms/%s/aliases" + % (self.room_id,), access_token=access_token, ) self.render(request) -- cgit 1.4.1 From 2b37eabca1e9355e2e2ab8f65bbdda12431ecc28 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 19 Feb 2020 15:04:47 +0000 Subject: Reduce auth chains fetched during v2 state res. (#6952) The state res v2 algorithm only cares about the difference between auth chains, so we can pass in the known common state to the `get_auth_chain` storage function so that it can ignore those events. --- changelog.d/6952.misc | 1 + synapse/state/__init__.py | 15 ++++++++---- synapse/state/v2.py | 2 +- .../storage/data_stores/main/event_federation.py | 28 ++++++++++++++++++---- tests/state/test_v2.py | 6 +++-- 5 files changed, 39 insertions(+), 13 deletions(-) create mode 100644 changelog.d/6952.misc (limited to 'tests') diff --git a/changelog.d/6952.misc b/changelog.d/6952.misc new file mode 100644 index 0000000000..e26dc5cab8 --- /dev/null +++ b/changelog.d/6952.misc @@ -0,0 +1 @@ +Improve perf of v2 state res for large rooms. diff --git a/synapse/state/__init__.py b/synapse/state/__init__.py index fdd6bef6b4..df7a4f6a89 100644 --- a/synapse/state/__init__.py +++ b/synapse/state/__init__.py @@ -16,7 +16,7 @@ import logging from collections import namedtuple -from typing import Dict, Iterable, List, Optional +from typing import Dict, Iterable, List, Optional, Set from six import iteritems, itervalues @@ -662,7 +662,7 @@ class StateResolutionStore(object): allow_rejected=allow_rejected, ) - def get_auth_chain(self, event_ids): + def get_auth_chain(self, event_ids: List[str], ignore_events: Set[str]): """Gets the full auth chain for a set of events (including rejected events). @@ -674,11 +674,16 @@ class StateResolutionStore(object): presence of rejected events Args: - event_ids (list): The event IDs of the events to fetch the auth - chain for. Must be state events. + event_ids: The event IDs of the events to fetch the auth chain for. + Must be state events. + ignore_events: Set of events to exclude from the returned auth + chain. + Returns: Deferred[list[str]]: List of event IDs of the auth chain. """ - return self.store.get_auth_chain_ids(event_ids, include_given=True) + return self.store.get_auth_chain_ids( + event_ids, include_given=True, ignore_events=ignore_events, + ) diff --git a/synapse/state/v2.py b/synapse/state/v2.py index 531018c6a5..75fe58305a 100644 --- a/synapse/state/v2.py +++ b/synapse/state/v2.py @@ -248,7 +248,7 @@ def _get_auth_chain_difference(state_sets, event_map, state_res_store): and eid not in common ) - auth_chain = yield state_res_store.get_auth_chain(auth_ids) + auth_chain = yield state_res_store.get_auth_chain(auth_ids, common) auth_ids.update(auth_chain) auth_sets.append(auth_ids) diff --git a/synapse/storage/data_stores/main/event_federation.py b/synapse/storage/data_stores/main/event_federation.py index 60c67457b4..e16da2577d 100644 --- a/synapse/storage/data_stores/main/event_federation.py +++ b/synapse/storage/data_stores/main/event_federation.py @@ -14,6 +14,7 @@ # limitations under the License. import itertools import logging +from typing import List, Optional, Set from six.moves import range from six.moves.queue import Empty, PriorityQueue @@ -46,21 +47,37 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBas event_ids, include_given=include_given ).addCallback(self.get_events_as_list) - def get_auth_chain_ids(self, event_ids, include_given=False): + def get_auth_chain_ids( + self, + event_ids: List[str], + include_given: bool = False, + ignore_events: Optional[Set[str]] = None, + ): """Get auth events for given event_ids. The events *must* be state events. Args: - event_ids (list): state events - include_given (bool): include the given events in result + event_ids: state events + include_given: include the given events in result + ignore_events: Set of events to exclude from the returned auth + chain. This is useful if the caller will just discard the + given events anyway, and saves us from figuring out their auth + chains if not required. Returns: list of event_ids """ return self.db.runInteraction( - "get_auth_chain_ids", self._get_auth_chain_ids_txn, event_ids, include_given + "get_auth_chain_ids", + self._get_auth_chain_ids_txn, + event_ids, + include_given, + ignore_events, ) - def _get_auth_chain_ids_txn(self, txn, event_ids, include_given): + def _get_auth_chain_ids_txn(self, txn, event_ids, include_given, ignore_events): + if ignore_events is None: + ignore_events = set() + if include_given: results = set(event_ids) else: @@ -80,6 +97,7 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBas txn.execute(base_sql + clause, list(args)) new_front.update([r[0] for r in txn]) + new_front -= ignore_events new_front -= results front = new_front diff --git a/tests/state/test_v2.py b/tests/state/test_v2.py index 5bafad9f19..5059ade850 100644 --- a/tests/state/test_v2.py +++ b/tests/state/test_v2.py @@ -603,7 +603,7 @@ class TestStateResolutionStore(object): return {eid: self.event_map[eid] for eid in event_ids if eid in self.event_map} - def get_auth_chain(self, event_ids): + def get_auth_chain(self, event_ids, ignore_events): """Gets the full auth chain for a set of events (including rejected events). @@ -617,6 +617,8 @@ class TestStateResolutionStore(object): Args: event_ids (list): The event IDs of the events to fetch the auth chain for. Must be state events. + ignore_events: Set of events to exclude from the returned auth + chain. Returns: Deferred[list[str]]: List of event IDs of the auth chain. @@ -627,7 +629,7 @@ class TestStateResolutionStore(object): stack = list(event_ids) while stack: event_id = stack.pop() - if event_id in result: + if event_id in result or event_id in ignore_events: continue result.add(event_id) -- cgit 1.4.1 From 509e381afa8c656e72f5fef3d651a9819794174a Mon Sep 17 00:00:00 2001 From: Patrick Cloke Date: Fri, 21 Feb 2020 07:15:07 -0500 Subject: Clarify list/set/dict/tuple comprehensions and enforce via flake8 (#6957) Ensure good comprehension hygiene using flake8-comprehensions. --- CONTRIBUTING.md | 2 +- changelog.d/6957.misc | 1 + docs/code_style.md | 2 +- scripts-dev/convert_server_keys.py | 2 +- synapse/app/_base.py | 2 +- synapse/app/federation_sender.py | 4 +-- synapse/app/pusher.py | 2 +- synapse/config/server.py | 4 +-- synapse/config/tls.py | 2 +- synapse/crypto/keyring.py | 6 ++-- synapse/federation/send_queue.py | 4 +-- synapse/groups/groups_server.py | 2 +- synapse/handlers/device.py | 2 +- synapse/handlers/directory.py | 4 +-- synapse/handlers/federation.py | 18 +++++----- synapse/handlers/presence.py | 6 ++-- synapse/handlers/receipts.py | 2 +- synapse/handlers/room.py | 2 +- synapse/handlers/search.py | 8 ++--- synapse/handlers/sync.py | 22 ++++++------ synapse/handlers/typing.py | 4 +-- synapse/logging/utils.py | 2 +- synapse/metrics/__init__.py | 2 +- synapse/metrics/background_process_metrics.py | 4 +-- synapse/push/bulk_push_rule_evaluator.py | 8 ++--- synapse/push/emailpusher.py | 2 +- synapse/push/mailer.py | 20 +++++------ synapse/push/pusherpool.py | 2 +- synapse/rest/admin/_base.py | 4 +-- synapse/rest/client/v1/push_rule.py | 6 ++-- synapse/rest/client/v1/pusher.py | 4 +-- synapse/rest/client/v2_alpha/sync.py | 2 +- synapse/rest/key/v2/remote_key_resource.py | 2 +- synapse/rest/media/v1/_base.py | 40 ++++++++++------------ synapse/state/v1.py | 10 +++--- synapse/state/v2.py | 8 ++--- synapse/storage/_base.py | 2 +- synapse/storage/background_updates.py | 2 +- synapse/storage/data_stores/main/appservice.py | 14 ++++---- synapse/storage/data_stores/main/client_ips.py | 4 +-- synapse/storage/data_stores/main/devices.py | 13 ++++--- .../storage/data_stores/main/event_federation.py | 2 +- synapse/storage/data_stores/main/events.py | 8 ++--- .../storage/data_stores/main/events_bg_updates.py | 2 +- synapse/storage/data_stores/main/events_worker.py | 6 ++-- synapse/storage/data_stores/main/push_rule.py | 8 ++--- synapse/storage/data_stores/main/receipts.py | 4 +-- synapse/storage/data_stores/main/roommember.py | 4 +-- synapse/storage/data_stores/main/state.py | 8 ++--- synapse/storage/data_stores/main/stream.py | 8 ++--- .../storage/data_stores/main/user_erasure_store.py | 4 +-- synapse/storage/data_stores/state/store.py | 4 +-- synapse/storage/database.py | 4 +-- synapse/storage/persist_events.py | 8 ++--- synapse/storage/prepare_database.py | 6 ++-- synapse/util/frozenutils.py | 2 +- synapse/visibility.py | 4 +-- tests/config/test_generate.py | 2 +- tests/federation/test_federation_server.py | 2 +- tests/handlers/test_presence.py | 4 +-- tests/handlers/test_typing.py | 6 ++-- tests/handlers/test_user_directory.py | 12 +++---- tests/push/test_email.py | 6 ++-- tests/push/test_http.py | 8 ++--- tests/rest/client/v2_alpha/test_sync.py | 28 ++++++++------- tests/storage/test__base.py | 4 +-- tests/storage/test_appservice.py | 36 +++++++++---------- tests/storage/test_cleanup_extrems.py | 10 +++--- tests/storage/test_event_metrics.py | 36 +++++++++---------- tests/storage/test_state.py | 2 +- tests/test_state.py | 18 +++------- tests/util/test_stream_change_cache.py | 18 +++------- tox.ini | 1 + 73 files changed, 251 insertions(+), 276 deletions(-) create mode 100644 changelog.d/6957.misc (limited to 'tests') diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md index 4b01b6ac8c..253a0ca648 100644 --- a/CONTRIBUTING.md +++ b/CONTRIBUTING.md @@ -60,7 +60,7 @@ python 3.6 and to install each tool: ``` # Install the dependencies -pip install -U black flake8 isort +pip install -U black flake8 flake8-comprehensions isort # Run the linter script ./scripts-dev/lint.sh diff --git a/changelog.d/6957.misc b/changelog.d/6957.misc new file mode 100644 index 0000000000..4f98030110 --- /dev/null +++ b/changelog.d/6957.misc @@ -0,0 +1 @@ +Use flake8-comprehensions to enforce good hygiene of list/set/dict comprehensions. diff --git a/docs/code_style.md b/docs/code_style.md index 71aecd41f7..6ef6f80290 100644 --- a/docs/code_style.md +++ b/docs/code_style.md @@ -30,7 +30,7 @@ The necessary tools are detailed below. Install `flake8` with: - pip install --upgrade flake8 + pip install --upgrade flake8 flake8-comprehensions Check all application and test code with: diff --git a/scripts-dev/convert_server_keys.py b/scripts-dev/convert_server_keys.py index 179be61c30..06b4c1e2ff 100644 --- a/scripts-dev/convert_server_keys.py +++ b/scripts-dev/convert_server_keys.py @@ -103,7 +103,7 @@ def main(): yaml.safe_dump(result, sys.stdout, default_flow_style=False) - rows = list(row for server, json in result.items() for row in rows_v2(server, json)) + rows = [row for server, json in result.items() for row in rows_v2(server, json)] cursor = connection.cursor() cursor.executemany( diff --git a/synapse/app/_base.py b/synapse/app/_base.py index 109b1e2fb5..9ffd23c6df 100644 --- a/synapse/app/_base.py +++ b/synapse/app/_base.py @@ -141,7 +141,7 @@ def start_reactor( def quit_with_error(error_string): message_lines = error_string.split("\n") - line_length = max([len(l) for l in message_lines if len(l) < 80]) + 2 + line_length = max(len(l) for l in message_lines if len(l) < 80) + 2 sys.stderr.write("*" * line_length + "\n") for line in message_lines: sys.stderr.write(" %s\n" % (line.rstrip(),)) diff --git a/synapse/app/federation_sender.py b/synapse/app/federation_sender.py index 63a91f1177..b7fcf80ddc 100644 --- a/synapse/app/federation_sender.py +++ b/synapse/app/federation_sender.py @@ -262,7 +262,7 @@ class FederationSenderHandler(object): # ... as well as device updates and messages elif stream_name == DeviceListsStream.NAME: - hosts = set(row.destination for row in rows) + hosts = {row.destination for row in rows} for host in hosts: self.federation_sender.send_device_messages(host) @@ -270,7 +270,7 @@ class FederationSenderHandler(object): # The to_device stream includes stuff to be pushed to both local # clients and remote servers, so we ignore entities that start with # '@' (since they'll be local users rather than destinations). - hosts = set(row.entity for row in rows if not row.entity.startswith("@")) + hosts = {row.entity for row in rows if not row.entity.startswith("@")} for host in hosts: self.federation_sender.send_device_messages(host) diff --git a/synapse/app/pusher.py b/synapse/app/pusher.py index e46b6ac598..84e9f8d5e2 100644 --- a/synapse/app/pusher.py +++ b/synapse/app/pusher.py @@ -158,7 +158,7 @@ class PusherReplicationHandler(ReplicationClientHandler): yield self.pusher_pool.on_new_notifications(token, token) elif stream_name == "receipts": yield self.pusher_pool.on_new_receipts( - token, token, set(row.room_id for row in rows) + token, token, {row.room_id for row in rows} ) except Exception: logger.exception("Error poking pushers") diff --git a/synapse/config/server.py b/synapse/config/server.py index 0ec1b0fadd..7525765fee 100644 --- a/synapse/config/server.py +++ b/synapse/config/server.py @@ -1066,12 +1066,12 @@ KNOWN_RESOURCES = ( def _check_resource_config(listeners): - resource_names = set( + resource_names = { res_name for listener in listeners for res in listener.get("resources", []) for res_name in res.get("names", []) - ) + } for resource in resource_names: if resource not in KNOWN_RESOURCES: diff --git a/synapse/config/tls.py b/synapse/config/tls.py index 97a12d51f6..a65538562b 100644 --- a/synapse/config/tls.py +++ b/synapse/config/tls.py @@ -260,7 +260,7 @@ class TlsConfig(Config): crypto.FILETYPE_ASN1, self.tls_certificate ) sha256_fingerprint = encode_base64(sha256(x509_certificate_bytes).digest()) - sha256_fingerprints = set(f["sha256"] for f in self.tls_fingerprints) + sha256_fingerprints = {f["sha256"] for f in self.tls_fingerprints} if sha256_fingerprint not in sha256_fingerprints: self.tls_fingerprints.append({"sha256": sha256_fingerprint}) diff --git a/synapse/crypto/keyring.py b/synapse/crypto/keyring.py index 6fe5a6a26a..983f0ead8c 100644 --- a/synapse/crypto/keyring.py +++ b/synapse/crypto/keyring.py @@ -326,9 +326,7 @@ class Keyring(object): verify_requests (list[VerifyJsonRequest]): list of verify requests """ - remaining_requests = set( - (rq for rq in verify_requests if not rq.key_ready.called) - ) + remaining_requests = {rq for rq in verify_requests if not rq.key_ready.called} @defer.inlineCallbacks def do_iterations(): @@ -396,7 +394,7 @@ class Keyring(object): results = yield fetcher.get_keys(missing_keys) - completed = list() + completed = [] for verify_request in remaining_requests: server_name = verify_request.server_name diff --git a/synapse/federation/send_queue.py b/synapse/federation/send_queue.py index 001bb304ae..876fb0e245 100644 --- a/synapse/federation/send_queue.py +++ b/synapse/federation/send_queue.py @@ -129,9 +129,9 @@ class FederationRemoteSendQueue(object): for key in keys[:i]: del self.presence_changed[key] - user_ids = set( + user_ids = { user_id for uids in self.presence_changed.values() for user_id in uids - ) + } keys = self.presence_destinations.keys() i = self.presence_destinations.bisect_left(position_to_delete) diff --git a/synapse/groups/groups_server.py b/synapse/groups/groups_server.py index c106abae21..4f0dc0a209 100644 --- a/synapse/groups/groups_server.py +++ b/synapse/groups/groups_server.py @@ -608,7 +608,7 @@ class GroupsServerHandler(GroupsServerWorkerHandler): user_results = yield self.store.get_users_in_group( group_id, include_private=True ) - if user_id in [user_result["user_id"] for user_result in user_results]: + if user_id in (user_result["user_id"] for user_result in user_results): raise SynapseError(400, "User already in group") content = { diff --git a/synapse/handlers/device.py b/synapse/handlers/device.py index 50cea3f378..a514c30714 100644 --- a/synapse/handlers/device.py +++ b/synapse/handlers/device.py @@ -742,6 +742,6 @@ class DeviceListUpdater(object): # We clobber the seen updates since we've re-synced from a given # point. - self._seen_updates[user_id] = set([stream_id]) + self._seen_updates[user_id] = {stream_id} defer.returnValue(result) diff --git a/synapse/handlers/directory.py b/synapse/handlers/directory.py index 921d887b24..0b23ca919a 100644 --- a/synapse/handlers/directory.py +++ b/synapse/handlers/directory.py @@ -72,7 +72,7 @@ class DirectoryHandler(BaseHandler): # TODO(erikj): Check if there is a current association. if not servers: users = yield self.state.get_current_users_in_room(room_id) - servers = set(get_domain_from_id(u) for u in users) + servers = {get_domain_from_id(u) for u in users} if not servers: raise SynapseError(400, "Failed to get server list") @@ -255,7 +255,7 @@ class DirectoryHandler(BaseHandler): ) users = yield self.state.get_current_users_in_room(room_id) - extra_servers = set(get_domain_from_id(u) for u in users) + extra_servers = {get_domain_from_id(u) for u in users} servers = set(extra_servers) | set(servers) # If this server is in the list of servers, return it first. diff --git a/synapse/handlers/federation.py b/synapse/handlers/federation.py index eb20ef4aec..a689065f89 100644 --- a/synapse/handlers/federation.py +++ b/synapse/handlers/federation.py @@ -659,11 +659,11 @@ class FederationHandler(BaseHandler): # this can happen if a remote server claims that the state or # auth_events at an event in room A are actually events in room B - bad_events = list( + bad_events = [ (event_id, event.room_id) for event_id, event in fetched_events.items() if event.room_id != room_id - ) + ] for bad_event_id, bad_room_id in bad_events: # This is a bogus situation, but since we may only discover it a long time @@ -856,7 +856,7 @@ class FederationHandler(BaseHandler): # Don't bother processing events we already have. seen_events = await self.store.have_events_in_timeline( - set(e.event_id for e in events) + {e.event_id for e in events} ) events = [e for e in events if e.event_id not in seen_events] @@ -866,7 +866,7 @@ class FederationHandler(BaseHandler): event_map = {e.event_id: e for e in events} - event_ids = set(e.event_id for e in events) + event_ids = {e.event_id for e in events} # build a list of events whose prev_events weren't in the batch. # (XXX: this will include events whose prev_events we already have; that doesn't @@ -892,13 +892,13 @@ class FederationHandler(BaseHandler): state_events.update({s.event_id: s for s in state}) events_to_state[e_id] = state - required_auth = set( + required_auth = { a_id for event in events + list(state_events.values()) + list(auth_events.values()) for a_id in event.auth_event_ids() - ) + } auth_events.update( {e_id: event_map[e_id] for e_id in required_auth if e_id in event_map} ) @@ -1247,7 +1247,7 @@ class FederationHandler(BaseHandler): async def on_event_auth(self, event_id: str) -> List[EventBase]: event = await self.store.get_event(event_id) auth = await self.store.get_auth_chain( - [auth_id for auth_id in event.auth_event_ids()], include_given=True + list(event.auth_event_ids()), include_given=True ) return list(auth) @@ -2152,7 +2152,7 @@ class FederationHandler(BaseHandler): # Now get the current auth_chain for the event. local_auth_chain = await self.store.get_auth_chain( - [auth_id for auth_id in event.auth_event_ids()], include_given=True + list(event.auth_event_ids()), include_given=True ) # TODO: Check if we would now reject event_id. If so we need to tell @@ -2654,7 +2654,7 @@ class FederationHandler(BaseHandler): member_handler = self.hs.get_room_member_handler() yield member_handler.send_membership_event(None, event, context) else: - destinations = set(x.split(":", 1)[-1] for x in (sender_user_id, room_id)) + destinations = {x.split(":", 1)[-1] for x in (sender_user_id, room_id)} yield self.federation_client.forward_third_party_invite( destinations, room_id, event_dict ) diff --git a/synapse/handlers/presence.py b/synapse/handlers/presence.py index 202aa9294f..0d6cf2b008 100644 --- a/synapse/handlers/presence.py +++ b/synapse/handlers/presence.py @@ -313,7 +313,7 @@ class PresenceHandler(object): notified_presence_counter.inc(len(to_notify)) yield self._persist_and_notify(list(to_notify.values())) - self.unpersisted_users_changes |= set(s.user_id for s in new_states) + self.unpersisted_users_changes |= {s.user_id for s in new_states} self.unpersisted_users_changes -= set(to_notify.keys()) to_federation_ping = { @@ -698,7 +698,7 @@ class PresenceHandler(object): updates = yield self.current_state_for_users(target_user_ids) updates = list(updates.values()) - for user_id in set(target_user_ids) - set(u.user_id for u in updates): + for user_id in set(target_user_ids) - {u.user_id for u in updates}: updates.append(UserPresenceState.default(user_id)) now = self.clock.time_msec() @@ -886,7 +886,7 @@ class PresenceHandler(object): hosts = yield self.state.get_current_hosts_in_room(room_id) # Filter out ourselves. - hosts = set(host for host in hosts if host != self.server_name) + hosts = {host for host in hosts if host != self.server_name} self.federation.send_presence_to_destinations( states=[state], destinations=hosts diff --git a/synapse/handlers/receipts.py b/synapse/handlers/receipts.py index 9283c039e3..8bc100db42 100644 --- a/synapse/handlers/receipts.py +++ b/synapse/handlers/receipts.py @@ -94,7 +94,7 @@ class ReceiptsHandler(BaseHandler): # no new receipts return False - affected_room_ids = list(set([r.room_id for r in receipts])) + affected_room_ids = list({r.room_id for r in receipts}) self.notifier.on_new_event("receipt_key", max_batch_id, rooms=affected_room_ids) # Note that the min here shouldn't be relied upon to be accurate. diff --git a/synapse/handlers/room.py b/synapse/handlers/room.py index 76e8f61b74..8ee870f0bb 100644 --- a/synapse/handlers/room.py +++ b/synapse/handlers/room.py @@ -355,7 +355,7 @@ class RoomCreationHandler(BaseHandler): # If so, mark the new room as non-federatable as well creation_content["m.federate"] = False - initial_state = dict() + initial_state = {} # Replicate relevant room events types_to_copy = ( diff --git a/synapse/handlers/search.py b/synapse/handlers/search.py index 110097eab9..ec1542d416 100644 --- a/synapse/handlers/search.py +++ b/synapse/handlers/search.py @@ -184,7 +184,7 @@ class SearchHandler(BaseHandler): membership_list=[Membership.JOIN], # membership_list=[Membership.JOIN, Membership.LEAVE, Membership.Ban], ) - room_ids = set(r.room_id for r in rooms) + room_ids = {r.room_id for r in rooms} # If doing a subset of all rooms seearch, check if any of the rooms # are from an upgraded room, and search their contents as well @@ -374,12 +374,12 @@ class SearchHandler(BaseHandler): ).to_string() if include_profile: - senders = set( + senders = { ev.sender for ev in itertools.chain( res["events_before"], [event], res["events_after"] ) - ) + } if res["events_after"]: last_event_id = res["events_after"][-1].event_id @@ -421,7 +421,7 @@ class SearchHandler(BaseHandler): state_results = {} if include_state: - rooms = set(e.room_id for e in allowed_events) + rooms = {e.room_id for e in allowed_events} for room_id in rooms: state = yield self.state_handler.get_current_state(room_id) state_results[room_id] = list(state.values()) diff --git a/synapse/handlers/sync.py b/synapse/handlers/sync.py index 4324bc702e..669dbc8a48 100644 --- a/synapse/handlers/sync.py +++ b/synapse/handlers/sync.py @@ -682,11 +682,9 @@ class SyncHandler(object): # FIXME: order by stream ordering rather than as returned by SQL if joined_user_ids or invited_user_ids: - summary["m.heroes"] = sorted( - [user_id for user_id in (joined_user_ids + invited_user_ids)] - )[0:5] + summary["m.heroes"] = sorted(joined_user_ids + invited_user_ids)[0:5] else: - summary["m.heroes"] = sorted([user_id for user_id in gone_user_ids])[0:5] + summary["m.heroes"] = sorted(gone_user_ids)[0:5] if not sync_config.filter_collection.lazy_load_members(): return summary @@ -697,9 +695,9 @@ class SyncHandler(object): # track which members the client should already know about via LL: # Ones which are already in state... - existing_members = set( + existing_members = { user_id for (typ, user_id) in state.keys() if typ == EventTypes.Member - ) + } # ...or ones which are in the timeline... for ev in batch.events: @@ -773,10 +771,10 @@ class SyncHandler(object): # We only request state for the members needed to display the # timeline: - members_to_fetch = set( + members_to_fetch = { event.sender # FIXME: we also care about invite targets etc. for event in batch.events - ) + } if full_state: # always make sure we LL ourselves so we know we're in the room @@ -1993,10 +1991,10 @@ def _calculate_state( ) } - c_ids = set(e for e in itervalues(current)) - ts_ids = set(e for e in itervalues(timeline_start)) - p_ids = set(e for e in itervalues(previous)) - tc_ids = set(e for e in itervalues(timeline_contains)) + c_ids = set(itervalues(current)) + ts_ids = set(itervalues(timeline_start)) + p_ids = set(itervalues(previous)) + tc_ids = set(itervalues(timeline_contains)) # If we are lazyloading room members, we explicitly add the membership events # for the senders in the timeline into the state block returned by /sync, diff --git a/synapse/handlers/typing.py b/synapse/handlers/typing.py index 5406618431..391bceb0c4 100644 --- a/synapse/handlers/typing.py +++ b/synapse/handlers/typing.py @@ -198,7 +198,7 @@ class TypingHandler(object): now=now, obj=member, then=now + FEDERATION_PING_INTERVAL ) - for domain in set(get_domain_from_id(u) for u in users): + for domain in {get_domain_from_id(u) for u in users}: if domain != self.server_name: logger.debug("sending typing update to %s", domain) self.federation.build_and_send_edu( @@ -231,7 +231,7 @@ class TypingHandler(object): return users = yield self.state.get_current_users_in_room(room_id) - domains = set(get_domain_from_id(u) for u in users) + domains = {get_domain_from_id(u) for u in users} if self.server_name in domains: logger.info("Got typing update from %s: %r", user_id, content) diff --git a/synapse/logging/utils.py b/synapse/logging/utils.py index 6073fc2725..0c2527bd86 100644 --- a/synapse/logging/utils.py +++ b/synapse/logging/utils.py @@ -148,7 +148,7 @@ def trace_function(f): pathname=pathname, lineno=lineno, msg=msg, - args=tuple(), + args=(), exc_info=None, ) diff --git a/synapse/metrics/__init__.py b/synapse/metrics/__init__.py index 0b45e1f52a..0dba997a23 100644 --- a/synapse/metrics/__init__.py +++ b/synapse/metrics/__init__.py @@ -240,7 +240,7 @@ class BucketCollector(object): res.append(["+Inf", sum(data.values())]) metric = HistogramMetricFamily( - self.name, "", buckets=res, sum_value=sum([x * y for x, y in data.items()]) + self.name, "", buckets=res, sum_value=sum(x * y for x, y in data.items()) ) yield metric diff --git a/synapse/metrics/background_process_metrics.py b/synapse/metrics/background_process_metrics.py index c53d2a0d40..b65bcd8806 100644 --- a/synapse/metrics/background_process_metrics.py +++ b/synapse/metrics/background_process_metrics.py @@ -80,13 +80,13 @@ _background_process_db_sched_duration = Counter( # map from description to a counter, so that we can name our logcontexts # incrementally. (It actually duplicates _background_process_start_count, but # it's much simpler to do so than to try to combine them.) -_background_process_counts = dict() # type: dict[str, int] +_background_process_counts = {} # type: dict[str, int] # map from description to the currently running background processes. # # it's kept as a dict of sets rather than a big set so that we can keep track # of process descriptions that no longer have any active processes. -_background_processes = dict() # type: dict[str, set[_BackgroundProcess]] +_background_processes = {} # type: dict[str, set[_BackgroundProcess]] # A lock that covers the above dicts _bg_metrics_lock = threading.Lock() diff --git a/synapse/push/bulk_push_rule_evaluator.py b/synapse/push/bulk_push_rule_evaluator.py index 7d9f5a38d9..433ca2f416 100644 --- a/synapse/push/bulk_push_rule_evaluator.py +++ b/synapse/push/bulk_push_rule_evaluator.py @@ -400,11 +400,11 @@ class RulesForRoom(object): if logger.isEnabledFor(logging.DEBUG): logger.debug("Found members %r: %r", self.room_id, members.values()) - interested_in_user_ids = set( + interested_in_user_ids = { user_id for user_id, membership in itervalues(members) if membership == Membership.JOIN - ) + } logger.debug("Joined: %r", interested_in_user_ids) @@ -412,9 +412,9 @@ class RulesForRoom(object): interested_in_user_ids, on_invalidate=self.invalidate_all_cb ) - user_ids = set( + user_ids = { uid for uid, have_pusher in iteritems(if_users_with_pushers) if have_pusher - ) + } logger.debug("With pushers: %r", user_ids) diff --git a/synapse/push/emailpusher.py b/synapse/push/emailpusher.py index 8c818a86bf..ba4551d619 100644 --- a/synapse/push/emailpusher.py +++ b/synapse/push/emailpusher.py @@ -204,7 +204,7 @@ class EmailPusher(object): yield self.send_notification(unprocessed, reason) yield self.save_last_stream_ordering_and_success( - max([ea["stream_ordering"] for ea in unprocessed]) + max(ea["stream_ordering"] for ea in unprocessed) ) # we update the throttle on all the possible unprocessed push actions diff --git a/synapse/push/mailer.py b/synapse/push/mailer.py index b13b646bfd..4ccaf178ce 100644 --- a/synapse/push/mailer.py +++ b/synapse/push/mailer.py @@ -526,12 +526,10 @@ class Mailer(object): # If the room doesn't have a name, say who the messages # are from explicitly to avoid, "messages in the Bob room" sender_ids = list( - set( - [ - notif_events[n["event_id"]].sender - for n in notifs_by_room[room_id] - ] - ) + { + notif_events[n["event_id"]].sender + for n in notifs_by_room[room_id] + } ) member_events = yield self.store.get_events( @@ -558,12 +556,10 @@ class Mailer(object): # If the reason room doesn't have a name, say who the messages # are from explicitly to avoid, "messages in the Bob room" sender_ids = list( - set( - [ - notif_events[n["event_id"]].sender - for n in notifs_by_room[reason["room_id"]] - ] - ) + { + notif_events[n["event_id"]].sender + for n in notifs_by_room[reason["room_id"]] + } ) member_events = yield self.store.get_events( diff --git a/synapse/push/pusherpool.py b/synapse/push/pusherpool.py index b9dca5bc63..01789a9fb4 100644 --- a/synapse/push/pusherpool.py +++ b/synapse/push/pusherpool.py @@ -191,7 +191,7 @@ class PusherPool: min_stream_id - 1, max_stream_id ) # This returns a tuple, user_id is at index 3 - users_affected = set([r[3] for r in updated_receipts]) + users_affected = {r[3] for r in updated_receipts} for u in users_affected: if u in self.pushers: diff --git a/synapse/rest/admin/_base.py b/synapse/rest/admin/_base.py index 459482eb6d..a96f75ce26 100644 --- a/synapse/rest/admin/_base.py +++ b/synapse/rest/admin/_base.py @@ -29,7 +29,7 @@ def historical_admin_path_patterns(path_regex): Note that this should only be used for existing endpoints: new ones should just register for the /_synapse/admin path. """ - return list( + return [ re.compile(prefix + path_regex) for prefix in ( "^/_synapse/admin/v1", @@ -37,7 +37,7 @@ def historical_admin_path_patterns(path_regex): "^/_matrix/client/unstable/admin", "^/_matrix/client/r0/admin", ) - ) + ] def admin_patterns(path_regex: str): diff --git a/synapse/rest/client/v1/push_rule.py b/synapse/rest/client/v1/push_rule.py index 4f74600239..9fd4908136 100644 --- a/synapse/rest/client/v1/push_rule.py +++ b/synapse/rest/client/v1/push_rule.py @@ -49,7 +49,7 @@ class PushRuleRestServlet(RestServlet): if self._is_worker: raise Exception("Cannot handle PUT /push_rules on worker") - spec = _rule_spec_from_path([x for x in path.split("/")]) + spec = _rule_spec_from_path(path.split("/")) try: priority_class = _priority_class_from_spec(spec) except InvalidRuleException as e: @@ -110,7 +110,7 @@ class PushRuleRestServlet(RestServlet): if self._is_worker: raise Exception("Cannot handle DELETE /push_rules on worker") - spec = _rule_spec_from_path([x for x in path.split("/")]) + spec = _rule_spec_from_path(path.split("/")) requester = await self.auth.get_user_by_req(request) user_id = requester.user.to_string() @@ -138,7 +138,7 @@ class PushRuleRestServlet(RestServlet): rules = format_push_rules_for_user(requester.user, rules) - path = [x for x in path.split("/")][1:] + path = path.split("/")[1:] if path == []: # we're a reference impl: pedantry is our job. diff --git a/synapse/rest/client/v1/pusher.py b/synapse/rest/client/v1/pusher.py index 6f6b7aed6e..550a2f1b44 100644 --- a/synapse/rest/client/v1/pusher.py +++ b/synapse/rest/client/v1/pusher.py @@ -54,9 +54,9 @@ class PushersRestServlet(RestServlet): pushers = await self.hs.get_datastore().get_pushers_by_user_id(user.to_string()) - filtered_pushers = list( + filtered_pushers = [ {k: v for k, v in p.items() if k in ALLOWED_KEYS} for p in pushers - ) + ] return 200, {"pushers": filtered_pushers} diff --git a/synapse/rest/client/v2_alpha/sync.py b/synapse/rest/client/v2_alpha/sync.py index d8292ce29f..8fa68dd37f 100644 --- a/synapse/rest/client/v2_alpha/sync.py +++ b/synapse/rest/client/v2_alpha/sync.py @@ -72,7 +72,7 @@ class SyncRestServlet(RestServlet): """ PATTERNS = client_patterns("/sync$") - ALLOWED_PRESENCE = set(["online", "offline", "unavailable"]) + ALLOWED_PRESENCE = {"online", "offline", "unavailable"} def __init__(self, hs): super(SyncRestServlet, self).__init__() diff --git a/synapse/rest/key/v2/remote_key_resource.py b/synapse/rest/key/v2/remote_key_resource.py index 9d6813a047..4b6d030a57 100644 --- a/synapse/rest/key/v2/remote_key_resource.py +++ b/synapse/rest/key/v2/remote_key_resource.py @@ -149,7 +149,7 @@ class RemoteKey(DirectServeResource): time_now_ms = self.clock.time_msec() - cache_misses = dict() # type: Dict[str, Set[str]] + cache_misses = {} # type: Dict[str, Set[str]] for (server_name, key_id, from_server), results in cached.items(): results = [(result["ts_added_ms"], result) for result in results] diff --git a/synapse/rest/media/v1/_base.py b/synapse/rest/media/v1/_base.py index 65bbf00073..ba28dd089d 100644 --- a/synapse/rest/media/v1/_base.py +++ b/synapse/rest/media/v1/_base.py @@ -135,27 +135,25 @@ def add_file_headers(request, media_type, file_size, upload_name): # separators as defined in RFC2616. SP and HT are handled separately. # see _can_encode_filename_as_token. -_FILENAME_SEPARATOR_CHARS = set( - ( - "(", - ")", - "<", - ">", - "@", - ",", - ";", - ":", - "\\", - '"', - "/", - "[", - "]", - "?", - "=", - "{", - "}", - ) -) +_FILENAME_SEPARATOR_CHARS = { + "(", + ")", + "<", + ">", + "@", + ",", + ";", + ":", + "\\", + '"', + "/", + "[", + "]", + "?", + "=", + "{", + "}", +} def _can_encode_filename_as_token(x): diff --git a/synapse/state/v1.py b/synapse/state/v1.py index 24b7c0faef..9bf98d06f2 100644 --- a/synapse/state/v1.py +++ b/synapse/state/v1.py @@ -69,9 +69,9 @@ def resolve_events_with_store( unconflicted_state, conflicted_state = _seperate(state_sets) - needed_events = set( + needed_events = { event_id for event_ids in itervalues(conflicted_state) for event_id in event_ids - ) + } needed_event_count = len(needed_events) if event_map is not None: needed_events -= set(iterkeys(event_map)) @@ -261,11 +261,11 @@ def _resolve_state_events(conflicted_state, auth_events): def _resolve_auth_events(events, auth_events): - reverse = [i for i in reversed(_ordered_events(events))] + reverse = list(reversed(_ordered_events(events))) - auth_keys = set( + auth_keys = { key for event in events for key in event_auth.auth_types_for_event(event) - ) + } new_auth_events = {} for key in auth_keys: diff --git a/synapse/state/v2.py b/synapse/state/v2.py index 75fe58305a..0ffe6d8c14 100644 --- a/synapse/state/v2.py +++ b/synapse/state/v2.py @@ -105,7 +105,7 @@ def resolve_events_with_store( % (room_id, event.event_id, event.room_id,) ) - full_conflicted_set = set(eid for eid in full_conflicted_set if eid in event_map) + full_conflicted_set = {eid for eid in full_conflicted_set if eid in event_map} logger.debug("%d full_conflicted_set entries", len(full_conflicted_set)) @@ -233,7 +233,7 @@ def _get_auth_chain_difference(state_sets, event_map, state_res_store): auth_sets = [] for state_set in state_sets: - auth_ids = set( + auth_ids = { eid for key, eid in iteritems(state_set) if ( @@ -246,7 +246,7 @@ def _get_auth_chain_difference(state_sets, event_map, state_res_store): ) ) and eid not in common - ) + } auth_chain = yield state_res_store.get_auth_chain(auth_ids, common) auth_ids.update(auth_chain) @@ -275,7 +275,7 @@ def _seperate(state_sets): conflicted_state = {} for key in set(itertools.chain.from_iterable(state_sets)): - event_ids = set(state_set.get(key) for state_set in state_sets) + event_ids = {state_set.get(key) for state_set in state_sets} if len(event_ids) == 1: unconflicted_state[key] = event_ids.pop() else: diff --git a/synapse/storage/_base.py b/synapse/storage/_base.py index da3b99f93d..13de5f1f62 100644 --- a/synapse/storage/_base.py +++ b/synapse/storage/_base.py @@ -56,7 +56,7 @@ class SQLBaseStore(metaclass=ABCMeta): members_changed (iterable[str]): The user_ids of members that have changed """ - for host in set(get_domain_from_id(u) for u in members_changed): + for host in {get_domain_from_id(u) for u in members_changed}: self._attempt_to_invalidate_cache("is_host_joined", (room_id, host)) self._attempt_to_invalidate_cache("was_host_joined", (room_id, host)) diff --git a/synapse/storage/background_updates.py b/synapse/storage/background_updates.py index bd547f35cf..eb1a7e5002 100644 --- a/synapse/storage/background_updates.py +++ b/synapse/storage/background_updates.py @@ -189,7 +189,7 @@ class BackgroundUpdater(object): keyvalues=None, retcols=("update_name", "depends_on"), ) - in_flight = set(update["update_name"] for update in updates) + in_flight = {update["update_name"] for update in updates} for update in updates: if update["depends_on"] not in in_flight: self._background_update_queue.append(update["update_name"]) diff --git a/synapse/storage/data_stores/main/appservice.py b/synapse/storage/data_stores/main/appservice.py index b2f39649fd..efbc06c796 100644 --- a/synapse/storage/data_stores/main/appservice.py +++ b/synapse/storage/data_stores/main/appservice.py @@ -135,7 +135,7 @@ class ApplicationServiceTransactionWorkerStore( may be empty. """ results = yield self.db.simple_select_list( - "application_services_state", dict(state=state), ["as_id"] + "application_services_state", {"state": state}, ["as_id"] ) # NB: This assumes this class is linked with ApplicationServiceStore as_list = self.get_app_services() @@ -158,7 +158,7 @@ class ApplicationServiceTransactionWorkerStore( """ result = yield self.db.simple_select_one( "application_services_state", - dict(as_id=service.id), + {"as_id": service.id}, ["state"], allow_none=True, desc="get_appservice_state", @@ -177,7 +177,7 @@ class ApplicationServiceTransactionWorkerStore( A Deferred which resolves when the state was set successfully. """ return self.db.simple_upsert( - "application_services_state", dict(as_id=service.id), dict(state=state) + "application_services_state", {"as_id": service.id}, {"state": state} ) def create_appservice_txn(self, service, events): @@ -253,13 +253,15 @@ class ApplicationServiceTransactionWorkerStore( self.db.simple_upsert_txn( txn, "application_services_state", - dict(as_id=service.id), - dict(last_txn=txn_id), + {"as_id": service.id}, + {"last_txn": txn_id}, ) # Delete txn self.db.simple_delete_txn( - txn, "application_services_txns", dict(txn_id=txn_id, as_id=service.id) + txn, + "application_services_txns", + {"txn_id": txn_id, "as_id": service.id}, ) return self.db.runInteraction( diff --git a/synapse/storage/data_stores/main/client_ips.py b/synapse/storage/data_stores/main/client_ips.py index 13f4c9c72e..e1ccb27142 100644 --- a/synapse/storage/data_stores/main/client_ips.py +++ b/synapse/storage/data_stores/main/client_ips.py @@ -530,7 +530,7 @@ class ClientIpStore(ClientIpBackgroundUpdateStore): ((row["access_token"], row["ip"]), (row["user_agent"], row["last_seen"])) for row in rows ) - return list( + return [ { "access_token": access_token, "ip": ip, @@ -538,7 +538,7 @@ class ClientIpStore(ClientIpBackgroundUpdateStore): "last_seen": last_seen, } for (access_token, ip), (user_agent, last_seen) in iteritems(results) - ) + ] @wrap_as_background_process("prune_old_user_ips") async def _prune_old_user_ips(self): diff --git a/synapse/storage/data_stores/main/devices.py b/synapse/storage/data_stores/main/devices.py index b7617efb80..d55733a4cd 100644 --- a/synapse/storage/data_stores/main/devices.py +++ b/synapse/storage/data_stores/main/devices.py @@ -137,7 +137,7 @@ class DeviceWorkerStore(SQLBaseStore): # get the cross-signing keys of the users in the list, so that we can # determine which of the device changes were cross-signing keys - users = set(r[0] for r in updates) + users = {r[0] for r in updates} master_key_by_user = {} self_signing_key_by_user = {} for user in users: @@ -446,7 +446,7 @@ class DeviceWorkerStore(SQLBaseStore): a set of user_ids and results_map is a mapping of user_id -> device_id -> device_info """ - user_ids = set(user_id for user_id, _ in query_list) + user_ids = {user_id for user_id, _ in query_list} user_map = yield self.get_device_list_last_stream_id_for_remotes(list(user_ids)) # We go and check if any of the users need to have their device lists @@ -454,10 +454,9 @@ class DeviceWorkerStore(SQLBaseStore): users_needing_resync = yield self.get_user_ids_requiring_device_list_resync( user_ids ) - user_ids_in_cache = ( - set(user_id for user_id, stream_id in user_map.items() if stream_id) - - users_needing_resync - ) + user_ids_in_cache = { + user_id for user_id, stream_id in user_map.items() if stream_id + } - users_needing_resync user_ids_not_in_cache = user_ids - user_ids_in_cache results = {} @@ -604,7 +603,7 @@ class DeviceWorkerStore(SQLBaseStore): rows = yield self.db.execute( "get_users_whose_signatures_changed", None, sql, user_id, from_key ) - return set(user for row in rows for user in json.loads(row[0])) + return {user for row in rows for user in json.loads(row[0])} else: return set() diff --git a/synapse/storage/data_stores/main/event_federation.py b/synapse/storage/data_stores/main/event_federation.py index 750ec1b70d..49a7b8b433 100644 --- a/synapse/storage/data_stores/main/event_federation.py +++ b/synapse/storage/data_stores/main/event_federation.py @@ -426,7 +426,7 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBas query, (room_id, event_id, False, limit - len(event_results)) ) - new_results = set(t[0] for t in txn) - seen_events + new_results = {t[0] for t in txn} - seen_events new_front |= new_results seen_events |= new_results diff --git a/synapse/storage/data_stores/main/events.py b/synapse/storage/data_stores/main/events.py index c9d0d68c3a..8ae23df00a 100644 --- a/synapse/storage/data_stores/main/events.py +++ b/synapse/storage/data_stores/main/events.py @@ -145,7 +145,7 @@ class EventsStore( return txn.fetchall() res = yield self.db.runInteraction("read_forward_extremities", fetch) - self._current_forward_extremities_amount = c_counter(list(x[0] for x in res)) + self._current_forward_extremities_amount = c_counter([x[0] for x in res]) @_retry_on_integrity_error @defer.inlineCallbacks @@ -598,11 +598,11 @@ class EventsStore( # We find out which membership events we may have deleted # and which we have added, then we invlidate the caches for all # those users. - members_changed = set( + members_changed = { state_key for ev_type, state_key in itertools.chain(to_delete, to_insert) if ev_type == EventTypes.Member - ) + } for member in members_changed: txn.call_after( @@ -1615,7 +1615,7 @@ class EventsStore( """ ) - referenced_state_groups = set(sg for sg, in txn) + referenced_state_groups = {sg for sg, in txn} logger.info( "[purge] found %i referenced state groups", len(referenced_state_groups) ) diff --git a/synapse/storage/data_stores/main/events_bg_updates.py b/synapse/storage/data_stores/main/events_bg_updates.py index 5177b71016..f54c8b1ee0 100644 --- a/synapse/storage/data_stores/main/events_bg_updates.py +++ b/synapse/storage/data_stores/main/events_bg_updates.py @@ -402,7 +402,7 @@ class EventsBackgroundUpdatesStore(SQLBaseStore): keyvalues={}, retcols=("room_id",), ) - room_ids = set(row["room_id"] for row in rows) + room_ids = {row["room_id"] for row in rows} for room_id in room_ids: txn.call_after( self.get_latest_event_ids_in_room.invalidate, (room_id,) diff --git a/synapse/storage/data_stores/main/events_worker.py b/synapse/storage/data_stores/main/events_worker.py index 7251e819f5..47a3a26072 100644 --- a/synapse/storage/data_stores/main/events_worker.py +++ b/synapse/storage/data_stores/main/events_worker.py @@ -494,9 +494,9 @@ class EventsWorkerStore(SQLBaseStore): """ with Measure(self._clock, "_fetch_event_list"): try: - events_to_fetch = set( + events_to_fetch = { event_id for events, _ in event_list for event_id in events - ) + } row_dict = self.db.new_transaction( conn, "do_fetch", [], [], self._fetch_event_rows, events_to_fetch @@ -804,7 +804,7 @@ class EventsWorkerStore(SQLBaseStore): desc="have_events_in_timeline", ) - return set(r["event_id"] for r in rows) + return {r["event_id"] for r in rows} @defer.inlineCallbacks def have_seen_events(self, event_ids): diff --git a/synapse/storage/data_stores/main/push_rule.py b/synapse/storage/data_stores/main/push_rule.py index e2673ae073..62ac88d9f2 100644 --- a/synapse/storage/data_stores/main/push_rule.py +++ b/synapse/storage/data_stores/main/push_rule.py @@ -276,21 +276,21 @@ class PushRulesWorkerStore( # We ignore app service users for now. This is so that we don't fill # up the `get_if_users_have_pushers` cache with AS entries that we # know don't have pushers, nor even read receipts. - local_users_in_room = set( + local_users_in_room = { u for u in users_in_room if self.hs.is_mine_id(u) and not self.get_if_app_services_interested_in_user(u) - ) + } # users in the room who have pushers need to get push rules run because # that's how their pushers work if_users_with_pushers = yield self.get_if_users_have_pushers( local_users_in_room, on_invalidate=cache_context.invalidate ) - user_ids = set( + user_ids = { uid for uid, have_pusher in if_users_with_pushers.items() if have_pusher - ) + } users_with_receipts = yield self.get_users_with_read_receipts_in_room( room_id, on_invalidate=cache_context.invalidate diff --git a/synapse/storage/data_stores/main/receipts.py b/synapse/storage/data_stores/main/receipts.py index 96e54d145e..0d932a0672 100644 --- a/synapse/storage/data_stores/main/receipts.py +++ b/synapse/storage/data_stores/main/receipts.py @@ -58,7 +58,7 @@ class ReceiptsWorkerStore(SQLBaseStore): @cachedInlineCallbacks() def get_users_with_read_receipts_in_room(self, room_id): receipts = yield self.get_receipts_for_room(room_id, "m.read") - return set(r["user_id"] for r in receipts) + return {r["user_id"] for r in receipts} @cached(num_args=2) def get_receipts_for_room(self, room_id, receipt_type): @@ -283,7 +283,7 @@ class ReceiptsWorkerStore(SQLBaseStore): args.append(limit) txn.execute(sql, args) - return list(r[0:5] + (json.loads(r[5]),) for r in txn) + return [r[0:5] + (json.loads(r[5]),) for r in txn] return self.db.runInteraction( "get_all_updated_receipts", get_all_updated_receipts_txn diff --git a/synapse/storage/data_stores/main/roommember.py b/synapse/storage/data_stores/main/roommember.py index d5ced05701..d5bd0cb5cf 100644 --- a/synapse/storage/data_stores/main/roommember.py +++ b/synapse/storage/data_stores/main/roommember.py @@ -465,7 +465,7 @@ class RoomMemberWorkerStore(EventsWorkerStore): txn.execute(sql % (clause,), args) - return set(row[0] for row in txn) + return {row[0] for row in txn} return await self.db.runInteraction( "get_users_server_still_shares_room_with", @@ -826,7 +826,7 @@ class RoomMemberWorkerStore(EventsWorkerStore): GROUP BY room_id, user_id; """ txn.execute(sql, (user_id,)) - return set(row[0] for row in txn if row[1] == 0) + return {row[0] for row in txn if row[1] == 0} return self.db.runInteraction( "get_forgotten_rooms_for_user", _get_forgotten_rooms_for_user_txn diff --git a/synapse/storage/data_stores/main/state.py b/synapse/storage/data_stores/main/state.py index 3d34103e67..3a3b9a8e72 100644 --- a/synapse/storage/data_stores/main/state.py +++ b/synapse/storage/data_stores/main/state.py @@ -321,7 +321,7 @@ class StateGroupWorkerStore(EventsWorkerStore, SQLBaseStore): desc="get_referenced_state_groups", ) - return set(row["state_group"] for row in rows) + return {row["state_group"] for row in rows} class MainStateBackgroundUpdateStore(RoomMemberWorkerStore): @@ -367,7 +367,7 @@ class MainStateBackgroundUpdateStore(RoomMemberWorkerStore): """ txn.execute(sql, (last_room_id, batch_size)) - room_ids = list(row[0] for row in txn) + room_ids = [row[0] for row in txn] if not room_ids: return True, set() @@ -384,7 +384,7 @@ class MainStateBackgroundUpdateStore(RoomMemberWorkerStore): txn.execute(sql, (last_room_id, room_ids[-1], "%:" + self.server_name)) - joined_room_ids = set(row[0] for row in txn) + joined_room_ids = {row[0] for row in txn} left_rooms = set(room_ids) - joined_room_ids @@ -404,7 +404,7 @@ class MainStateBackgroundUpdateStore(RoomMemberWorkerStore): retcols=("state_key",), ) - potentially_left_users = set(row["state_key"] for row in rows) + potentially_left_users = {row["state_key"] for row in rows} # Now lets actually delete the rooms from the DB. self.db.simple_delete_many_txn( diff --git a/synapse/storage/data_stores/main/stream.py b/synapse/storage/data_stores/main/stream.py index 056b25b13a..ada5cce6c2 100644 --- a/synapse/storage/data_stores/main/stream.py +++ b/synapse/storage/data_stores/main/stream.py @@ -346,11 +346,11 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): from_key (str): The room_key portion of a StreamToken """ from_key = RoomStreamToken.parse_stream_token(from_key).stream - return set( + return { room_id for room_id in room_ids if self._events_stream_cache.has_entity_changed(room_id, from_key) - ) + } @defer.inlineCallbacks def get_room_events_stream_for_room( @@ -679,11 +679,11 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): ) events_before = yield self.get_events_as_list( - [e for e in results["before"]["event_ids"]], get_prev_content=True + list(results["before"]["event_ids"]), get_prev_content=True ) events_after = yield self.get_events_as_list( - [e for e in results["after"]["event_ids"]], get_prev_content=True + list(results["after"]["event_ids"]), get_prev_content=True ) return { diff --git a/synapse/storage/data_stores/main/user_erasure_store.py b/synapse/storage/data_stores/main/user_erasure_store.py index af8025bc17..ec6b8a4ffd 100644 --- a/synapse/storage/data_stores/main/user_erasure_store.py +++ b/synapse/storage/data_stores/main/user_erasure_store.py @@ -63,9 +63,9 @@ class UserErasureWorkerStore(SQLBaseStore): retcols=("user_id",), desc="are_users_erased", ) - erased_users = set(row["user_id"] for row in rows) + erased_users = {row["user_id"] for row in rows} - res = dict((u, u in erased_users) for u in user_ids) + res = {u: u in erased_users for u in user_ids} return res diff --git a/synapse/storage/data_stores/state/store.py b/synapse/storage/data_stores/state/store.py index c4ee9b7ccb..57a5267663 100644 --- a/synapse/storage/data_stores/state/store.py +++ b/synapse/storage/data_stores/state/store.py @@ -520,11 +520,11 @@ class StateGroupDataStore(StateBackgroundUpdateStore, SQLBaseStore): retcols=("state_group",), ) - remaining_state_groups = set( + remaining_state_groups = { row["state_group"] for row in rows if row["state_group"] not in state_groups_to_delete - ) + } logger.info( "[purge] de-delta-ing %i remaining state groups", diff --git a/synapse/storage/database.py b/synapse/storage/database.py index 6dcb5c04da..1953614401 100644 --- a/synapse/storage/database.py +++ b/synapse/storage/database.py @@ -554,8 +554,8 @@ class Database(object): Returns: A list of dicts where the key is the column header. """ - col_headers = list(intern(str(column[0])) for column in cursor.description) - results = list(dict(zip(col_headers, row)) for row in cursor) + col_headers = [intern(str(column[0])) for column in cursor.description] + results = [dict(zip(col_headers, row)) for row in cursor] return results def execute(self, desc, decoder, query, *args): diff --git a/synapse/storage/persist_events.py b/synapse/storage/persist_events.py index b950550f23..0f9ac1cf09 100644 --- a/synapse/storage/persist_events.py +++ b/synapse/storage/persist_events.py @@ -602,14 +602,14 @@ class EventsPersistenceStorage(object): event_id_to_state_group.update(event_to_groups) # State groups of old_latest_event_ids - old_state_groups = set( + old_state_groups = { event_id_to_state_group[evid] for evid in old_latest_event_ids - ) + } # State groups of new_latest_event_ids - new_state_groups = set( + new_state_groups = { event_id_to_state_group[evid] for evid in new_latest_event_ids - ) + } # If they old and new groups are the same then we don't need to do # anything. diff --git a/synapse/storage/prepare_database.py b/synapse/storage/prepare_database.py index c285ef52a0..fc69c32a0a 100644 --- a/synapse/storage/prepare_database.py +++ b/synapse/storage/prepare_database.py @@ -345,9 +345,9 @@ def _upgrade_existing_database( "Could not open delta dir for version %d: %s" % (v, directory) ) - duplicates = set( + duplicates = { file_name for file_name, count in file_name_counter.items() if count > 1 - ) + } if duplicates: # We don't support using the same file name in the same delta version. raise PrepareDatabaseException( @@ -454,7 +454,7 @@ def _apply_module_schema_files(cur, database_engine, modname, names_and_streams) ), (modname,), ) - applied_deltas = set(d for d, in cur) + applied_deltas = {d for d, in cur} for (name, stream) in names_and_streams: if name in applied_deltas: continue diff --git a/synapse/util/frozenutils.py b/synapse/util/frozenutils.py index 635b897d6c..f2ccd5e7c6 100644 --- a/synapse/util/frozenutils.py +++ b/synapse/util/frozenutils.py @@ -30,7 +30,7 @@ def freeze(o): return o try: - return tuple([freeze(i) for i in o]) + return tuple(freeze(i) for i in o) except TypeError: pass diff --git a/synapse/visibility.py b/synapse/visibility.py index d0abd8f04f..e60d9756b7 100644 --- a/synapse/visibility.py +++ b/synapse/visibility.py @@ -75,7 +75,7 @@ def filter_events_for_client( """ # Filter out events that have been soft failed so that we don't relay them # to clients. - events = list(e for e in events if not e.internal_metadata.is_soft_failed()) + events = [e for e in events if not e.internal_metadata.is_soft_failed()] types = ((EventTypes.RoomHistoryVisibility, ""), (EventTypes.Member, user_id)) event_id_to_state = yield storage.state.get_state_for_events( @@ -97,7 +97,7 @@ def filter_events_for_client( erased_senders = yield storage.main.are_users_erased((e.sender for e in events)) if apply_retention_policies: - room_ids = set(e.room_id for e in events) + room_ids = {e.room_id for e in events} retention_policies = {} for room_id in room_ids: diff --git a/tests/config/test_generate.py b/tests/config/test_generate.py index 2684e662de..463855ecc8 100644 --- a/tests/config/test_generate.py +++ b/tests/config/test_generate.py @@ -48,7 +48,7 @@ class ConfigGenerationTestCase(unittest.TestCase): ) self.assertSetEqual( - set(["homeserver.yaml", "lemurs.win.log.config", "lemurs.win.signing.key"]), + {"homeserver.yaml", "lemurs.win.log.config", "lemurs.win.signing.key"}, set(os.listdir(self.dir)), ) diff --git a/tests/federation/test_federation_server.py b/tests/federation/test_federation_server.py index e7d8699040..296dc887be 100644 --- a/tests/federation/test_federation_server.py +++ b/tests/federation/test_federation_server.py @@ -83,7 +83,7 @@ class StateQueryTests(unittest.FederatingHomeserverTestCase): ) ) - self.assertEqual(members, set(["@user:other.example.com", u1])) + self.assertEqual(members, {"@user:other.example.com", u1}) self.assertEqual(len(channel.json_body["pdus"]), 6) def test_needs_to_be_in_room(self): diff --git a/tests/handlers/test_presence.py b/tests/handlers/test_presence.py index c171038df8..64915bafcd 100644 --- a/tests/handlers/test_presence.py +++ b/tests/handlers/test_presence.py @@ -338,7 +338,7 @@ class PresenceTimeoutTestCase(unittest.TestCase): ) new_state = handle_timeout( - state, is_mine=True, syncing_user_ids=set([user_id]), now=now + state, is_mine=True, syncing_user_ids={user_id}, now=now ) self.assertIsNotNone(new_state) @@ -579,7 +579,7 @@ class PresenceJoinTestCase(unittest.HomeserverTestCase): ) self.assertEqual(expected_state.state, PresenceState.ONLINE) self.federation_sender.send_presence_to_destinations.assert_called_once_with( - destinations=set(("server2", "server3")), states=[expected_state] + destinations={"server2", "server3"}, states=[expected_state] ) def _add_new_user(self, room_id, user_id): diff --git a/tests/handlers/test_typing.py b/tests/handlers/test_typing.py index 140cc0a3c2..07b204666e 100644 --- a/tests/handlers/test_typing.py +++ b/tests/handlers/test_typing.py @@ -129,12 +129,12 @@ class TypingNotificationsTestCase(unittest.HomeserverTestCase): hs.get_auth().check_user_in_room = check_user_in_room def get_joined_hosts_for_room(room_id): - return set(member.domain for member in self.room_members) + return {member.domain for member in self.room_members} self.datastore.get_joined_hosts_for_room = get_joined_hosts_for_room def get_current_users_in_room(room_id): - return set(str(u) for u in self.room_members) + return {str(u) for u in self.room_members} hs.get_state_handler().get_current_users_in_room = get_current_users_in_room @@ -257,7 +257,7 @@ class TypingNotificationsTestCase(unittest.HomeserverTestCase): member = RoomMember(ROOM_ID, U_APPLE.to_string()) self.handler._member_typing_until[member] = 1002000 - self.handler._room_typing[ROOM_ID] = set([U_APPLE.to_string()]) + self.handler._room_typing[ROOM_ID] = {U_APPLE.to_string()} self.assertEquals(self.event_source.get_current_key(), 0) diff --git a/tests/handlers/test_user_directory.py b/tests/handlers/test_user_directory.py index 0a4765fff4..7b92bdbc47 100644 --- a/tests/handlers/test_user_directory.py +++ b/tests/handlers/test_user_directory.py @@ -114,7 +114,7 @@ class UserDirectoryTestCase(unittest.HomeserverTestCase): public_users = self.get_users_in_public_rooms() self.assertEqual( - self._compress_shared(shares_private), set([(u1, u2, room), (u2, u1, room)]) + self._compress_shared(shares_private), {(u1, u2, room), (u2, u1, room)} ) self.assertEqual(public_users, []) @@ -169,7 +169,7 @@ class UserDirectoryTestCase(unittest.HomeserverTestCase): public_users = self.get_users_in_public_rooms() self.assertEqual( - self._compress_shared(shares_private), set([(u1, u2, room), (u2, u1, room)]) + self._compress_shared(shares_private), {(u1, u2, room), (u2, u1, room)} ) self.assertEqual(public_users, []) @@ -226,7 +226,7 @@ class UserDirectoryTestCase(unittest.HomeserverTestCase): public_users = self.get_users_in_public_rooms() self.assertEqual( - self._compress_shared(shares_private), set([(u1, u2, room), (u2, u1, room)]) + self._compress_shared(shares_private), {(u1, u2, room), (u2, u1, room)} ) self.assertEqual(public_users, []) @@ -358,12 +358,12 @@ class UserDirectoryTestCase(unittest.HomeserverTestCase): public_users = self.get_users_in_public_rooms() # User 1 and User 2 are in the same public room - self.assertEqual(set(public_users), set([(u1, room), (u2, room)])) + self.assertEqual(set(public_users), {(u1, room), (u2, room)}) # User 1 and User 3 share private rooms self.assertEqual( self._compress_shared(shares_private), - set([(u1, u3, private_room), (u3, u1, private_room)]), + {(u1, u3, private_room), (u3, u1, private_room)}, ) def test_initial_share_all_users(self): @@ -398,7 +398,7 @@ class UserDirectoryTestCase(unittest.HomeserverTestCase): # No users share rooms self.assertEqual(public_users, []) - self.assertEqual(self._compress_shared(shares_private), set([])) + self.assertEqual(self._compress_shared(shares_private), set()) # Despite not sharing a room, search_all_users means we get a search # result. diff --git a/tests/push/test_email.py b/tests/push/test_email.py index 80187406bc..83032cc9ea 100644 --- a/tests/push/test_email.py +++ b/tests/push/test_email.py @@ -163,7 +163,7 @@ class EmailPusherTests(HomeserverTestCase): # Get the stream ordering before it gets sent pushers = self.get_success( - self.hs.get_datastore().get_pushers_by(dict(user_name=self.user_id)) + self.hs.get_datastore().get_pushers_by({"user_name": self.user_id}) ) pushers = list(pushers) self.assertEqual(len(pushers), 1) @@ -174,7 +174,7 @@ class EmailPusherTests(HomeserverTestCase): # It hasn't succeeded yet, so the stream ordering shouldn't have moved pushers = self.get_success( - self.hs.get_datastore().get_pushers_by(dict(user_name=self.user_id)) + self.hs.get_datastore().get_pushers_by({"user_name": self.user_id}) ) pushers = list(pushers) self.assertEqual(len(pushers), 1) @@ -192,7 +192,7 @@ class EmailPusherTests(HomeserverTestCase): # The stream ordering has increased pushers = self.get_success( - self.hs.get_datastore().get_pushers_by(dict(user_name=self.user_id)) + self.hs.get_datastore().get_pushers_by({"user_name": self.user_id}) ) pushers = list(pushers) self.assertEqual(len(pushers), 1) diff --git a/tests/push/test_http.py b/tests/push/test_http.py index fe3441f081..baf9c785f4 100644 --- a/tests/push/test_http.py +++ b/tests/push/test_http.py @@ -102,7 +102,7 @@ class HTTPPusherTests(HomeserverTestCase): # Get the stream ordering before it gets sent pushers = self.get_success( - self.hs.get_datastore().get_pushers_by(dict(user_name=user_id)) + self.hs.get_datastore().get_pushers_by({"user_name": user_id}) ) pushers = list(pushers) self.assertEqual(len(pushers), 1) @@ -113,7 +113,7 @@ class HTTPPusherTests(HomeserverTestCase): # It hasn't succeeded yet, so the stream ordering shouldn't have moved pushers = self.get_success( - self.hs.get_datastore().get_pushers_by(dict(user_name=user_id)) + self.hs.get_datastore().get_pushers_by({"user_name": user_id}) ) pushers = list(pushers) self.assertEqual(len(pushers), 1) @@ -132,7 +132,7 @@ class HTTPPusherTests(HomeserverTestCase): # The stream ordering has increased pushers = self.get_success( - self.hs.get_datastore().get_pushers_by(dict(user_name=user_id)) + self.hs.get_datastore().get_pushers_by({"user_name": user_id}) ) pushers = list(pushers) self.assertEqual(len(pushers), 1) @@ -152,7 +152,7 @@ class HTTPPusherTests(HomeserverTestCase): # The stream ordering has increased, again pushers = self.get_success( - self.hs.get_datastore().get_pushers_by(dict(user_name=user_id)) + self.hs.get_datastore().get_pushers_by({"user_name": user_id}) ) pushers = list(pushers) self.assertEqual(len(pushers), 1) diff --git a/tests/rest/client/v2_alpha/test_sync.py b/tests/rest/client/v2_alpha/test_sync.py index 9c13a13786..fa3a3ec1bd 100644 --- a/tests/rest/client/v2_alpha/test_sync.py +++ b/tests/rest/client/v2_alpha/test_sync.py @@ -40,16 +40,14 @@ class FilterTestCase(unittest.HomeserverTestCase): self.assertEqual(channel.code, 200) self.assertTrue( - set( - [ - "next_batch", - "rooms", - "presence", - "account_data", - "to_device", - "device_lists", - ] - ).issubset(set(channel.json_body.keys())) + { + "next_batch", + "rooms", + "presence", + "account_data", + "to_device", + "device_lists", + }.issubset(set(channel.json_body.keys())) ) def test_sync_presence_disabled(self): @@ -63,9 +61,13 @@ class FilterTestCase(unittest.HomeserverTestCase): self.assertEqual(channel.code, 200) self.assertTrue( - set( - ["next_batch", "rooms", "account_data", "to_device", "device_lists"] - ).issubset(set(channel.json_body.keys())) + { + "next_batch", + "rooms", + "account_data", + "to_device", + "device_lists", + }.issubset(set(channel.json_body.keys())) ) diff --git a/tests/storage/test__base.py b/tests/storage/test__base.py index d491ea2924..e37260a820 100644 --- a/tests/storage/test__base.py +++ b/tests/storage/test__base.py @@ -373,7 +373,7 @@ class UpsertManyTests(unittest.HomeserverTestCase): ) self.assertEqual( set(self._dump_to_tuple(res)), - set([(1, "user1", "hello"), (2, "user2", "there")]), + {(1, "user1", "hello"), (2, "user2", "there")}, ) # Update only user2 @@ -400,5 +400,5 @@ class UpsertManyTests(unittest.HomeserverTestCase): ) self.assertEqual( set(self._dump_to_tuple(res)), - set([(1, "user1", "hello"), (2, "user2", "bleb")]), + {(1, "user1", "hello"), (2, "user2", "bleb")}, ) diff --git a/tests/storage/test_appservice.py b/tests/storage/test_appservice.py index fd52512696..31710949a8 100644 --- a/tests/storage/test_appservice.py +++ b/tests/storage/test_appservice.py @@ -69,14 +69,14 @@ class ApplicationServiceStoreTestCase(unittest.TestCase): pass def _add_appservice(self, as_token, id, url, hs_token, sender): - as_yaml = dict( - url=url, - as_token=as_token, - hs_token=hs_token, - id=id, - sender_localpart=sender, - namespaces={}, - ) + as_yaml = { + "url": url, + "as_token": as_token, + "hs_token": hs_token, + "id": id, + "sender_localpart": sender, + "namespaces": {}, + } # use the token as the filename with open(as_token, "w") as outfile: outfile.write(yaml.dump(as_yaml)) @@ -135,14 +135,14 @@ class ApplicationServiceTransactionStoreTestCase(unittest.TestCase): ) def _add_service(self, url, as_token, id): - as_yaml = dict( - url=url, - as_token=as_token, - hs_token="something", - id=id, - sender_localpart="a_sender", - namespaces={}, - ) + as_yaml = { + "url": url, + "as_token": as_token, + "hs_token": "something", + "id": id, + "sender_localpart": "a_sender", + "namespaces": {}, + } # use the token as the filename with open(as_token, "w") as outfile: outfile.write(yaml.dump(as_yaml)) @@ -384,8 +384,8 @@ class ApplicationServiceTransactionStoreTestCase(unittest.TestCase): ) self.assertEquals(2, len(services)) self.assertEquals( - set([self.as_list[2]["id"], self.as_list[0]["id"]]), - set([services[0].id, services[1].id]), + {self.as_list[2]["id"], self.as_list[0]["id"]}, + {services[0].id, services[1].id}, ) diff --git a/tests/storage/test_cleanup_extrems.py b/tests/storage/test_cleanup_extrems.py index 029ac26454..0e04b2cf92 100644 --- a/tests/storage/test_cleanup_extrems.py +++ b/tests/storage/test_cleanup_extrems.py @@ -134,7 +134,7 @@ class CleanupExtremBackgroundUpdateStoreTestCase(HomeserverTestCase): latest_event_ids = self.get_success( self.store.get_latest_event_ids_in_room(self.room_id) ) - self.assertEqual(set(latest_event_ids), set((event_id_a, event_id_b))) + self.assertEqual(set(latest_event_ids), {event_id_a, event_id_b}) # Run the background update and check it did the right thing self.run_background_update() @@ -172,7 +172,7 @@ class CleanupExtremBackgroundUpdateStoreTestCase(HomeserverTestCase): latest_event_ids = self.get_success( self.store.get_latest_event_ids_in_room(self.room_id) ) - self.assertEqual(set(latest_event_ids), set((event_id_a, event_id_b))) + self.assertEqual(set(latest_event_ids), {event_id_a, event_id_b}) # Run the background update and check it did the right thing self.run_background_update() @@ -227,9 +227,7 @@ class CleanupExtremBackgroundUpdateStoreTestCase(HomeserverTestCase): latest_event_ids = self.get_success( self.store.get_latest_event_ids_in_room(self.room_id) ) - self.assertEqual( - set(latest_event_ids), set((event_id_a, event_id_b, event_id_c)) - ) + self.assertEqual(set(latest_event_ids), {event_id_a, event_id_b, event_id_c}) # Run the background update and check it did the right thing self.run_background_update() @@ -237,7 +235,7 @@ class CleanupExtremBackgroundUpdateStoreTestCase(HomeserverTestCase): latest_event_ids = self.get_success( self.store.get_latest_event_ids_in_room(self.room_id) ) - self.assertEqual(set(latest_event_ids), set([event_id_b, event_id_c])) + self.assertEqual(set(latest_event_ids), {event_id_b, event_id_c}) class CleanupExtremDummyEventsTestCase(HomeserverTestCase): diff --git a/tests/storage/test_event_metrics.py b/tests/storage/test_event_metrics.py index f26ff57a18..a7b7fd36d3 100644 --- a/tests/storage/test_event_metrics.py +++ b/tests/storage/test_event_metrics.py @@ -59,24 +59,22 @@ class ExtremStatisticsTestCase(HomeserverTestCase): ) ) - expected = set( - [ - b'synapse_forward_extremities_bucket{le="1.0"} 0.0', - b'synapse_forward_extremities_bucket{le="2.0"} 2.0', - b'synapse_forward_extremities_bucket{le="3.0"} 2.0', - b'synapse_forward_extremities_bucket{le="5.0"} 2.0', - b'synapse_forward_extremities_bucket{le="7.0"} 3.0', - b'synapse_forward_extremities_bucket{le="10.0"} 3.0', - b'synapse_forward_extremities_bucket{le="15.0"} 3.0', - b'synapse_forward_extremities_bucket{le="20.0"} 3.0', - b'synapse_forward_extremities_bucket{le="50.0"} 3.0', - b'synapse_forward_extremities_bucket{le="100.0"} 3.0', - b'synapse_forward_extremities_bucket{le="200.0"} 3.0', - b'synapse_forward_extremities_bucket{le="500.0"} 3.0', - b'synapse_forward_extremities_bucket{le="+Inf"} 3.0', - b"synapse_forward_extremities_count 3.0", - b"synapse_forward_extremities_sum 10.0", - ] - ) + expected = { + b'synapse_forward_extremities_bucket{le="1.0"} 0.0', + b'synapse_forward_extremities_bucket{le="2.0"} 2.0', + b'synapse_forward_extremities_bucket{le="3.0"} 2.0', + b'synapse_forward_extremities_bucket{le="5.0"} 2.0', + b'synapse_forward_extremities_bucket{le="7.0"} 3.0', + b'synapse_forward_extremities_bucket{le="10.0"} 3.0', + b'synapse_forward_extremities_bucket{le="15.0"} 3.0', + b'synapse_forward_extremities_bucket{le="20.0"} 3.0', + b'synapse_forward_extremities_bucket{le="50.0"} 3.0', + b'synapse_forward_extremities_bucket{le="100.0"} 3.0', + b'synapse_forward_extremities_bucket{le="200.0"} 3.0', + b'synapse_forward_extremities_bucket{le="500.0"} 3.0', + b'synapse_forward_extremities_bucket{le="+Inf"} 3.0', + b"synapse_forward_extremities_count 3.0", + b"synapse_forward_extremities_sum 10.0", + } self.assertEqual(items, expected) diff --git a/tests/storage/test_state.py b/tests/storage/test_state.py index 04d58fbf24..0b88308ff4 100644 --- a/tests/storage/test_state.py +++ b/tests/storage/test_state.py @@ -394,7 +394,7 @@ class StateStoreTestCase(tests.unittest.TestCase): ) = self.state_datastore._state_group_cache.get(group) self.assertEqual(is_all, False) - self.assertEqual(known_absent, set([(e1.type, e1.state_key)])) + self.assertEqual(known_absent, {(e1.type, e1.state_key)}) self.assertDictEqual(state_dict_ids, {(e1.type, e1.state_key): e1.event_id}) ############################################ diff --git a/tests/test_state.py b/tests/test_state.py index d1578fe581..66f22f6813 100644 --- a/tests/test_state.py +++ b/tests/test_state.py @@ -254,9 +254,7 @@ class StateTestCase(unittest.TestCase): ctx_d = context_store["D"] prev_state_ids = yield ctx_d.get_prev_state_ids() - self.assertSetEqual( - {"START", "A", "C"}, {e_id for e_id in prev_state_ids.values()} - ) + self.assertSetEqual({"START", "A", "C"}, set(prev_state_ids.values())) self.assertEqual(ctx_c.state_group, ctx_d.state_group_before_event) self.assertEqual(ctx_d.state_group_before_event, ctx_d.state_group) @@ -313,9 +311,7 @@ class StateTestCase(unittest.TestCase): ctx_e = context_store["E"] prev_state_ids = yield ctx_e.get_prev_state_ids() - self.assertSetEqual( - {"START", "A", "B", "C"}, {e for e in prev_state_ids.values()} - ) + self.assertSetEqual({"START", "A", "B", "C"}, set(prev_state_ids.values())) self.assertEqual(ctx_c.state_group, ctx_e.state_group_before_event) self.assertEqual(ctx_e.state_group_before_event, ctx_e.state_group) @@ -388,9 +384,7 @@ class StateTestCase(unittest.TestCase): ctx_d = context_store["D"] prev_state_ids = yield ctx_d.get_prev_state_ids() - self.assertSetEqual( - {"A1", "A2", "A3", "A5", "B"}, {e for e in prev_state_ids.values()} - ) + self.assertSetEqual({"A1", "A2", "A3", "A5", "B"}, set(prev_state_ids.values())) self.assertEqual(ctx_b.state_group, ctx_d.state_group_before_event) self.assertEqual(ctx_d.state_group_before_event, ctx_d.state_group) @@ -482,7 +476,7 @@ class StateTestCase(unittest.TestCase): current_state_ids = yield context.get_current_state_ids() self.assertEqual( - set([e.event_id for e in old_state]), set(current_state_ids.values()) + {e.event_id for e in old_state}, set(current_state_ids.values()) ) self.assertEqual(group_name, context.state_group) @@ -513,9 +507,7 @@ class StateTestCase(unittest.TestCase): prev_state_ids = yield context.get_prev_state_ids() - self.assertEqual( - set([e.event_id for e in old_state]), set(prev_state_ids.values()) - ) + self.assertEqual({e.event_id for e in old_state}, set(prev_state_ids.values())) self.assertIsNotNone(context.state_group) diff --git a/tests/util/test_stream_change_cache.py b/tests/util/test_stream_change_cache.py index f2be63706b..72a9de5370 100644 --- a/tests/util/test_stream_change_cache.py +++ b/tests/util/test_stream_change_cache.py @@ -67,7 +67,7 @@ class StreamChangeCacheTests(unittest.TestCase): # If we update an existing entity, it keeps the two existing entities cache.entity_has_changed("bar@baz.net", 5) self.assertEqual( - set(["bar@baz.net", "user@elsewhere.org"]), set(cache._entity_to_key) + {"bar@baz.net", "user@elsewhere.org"}, set(cache._entity_to_key) ) def test_get_all_entities_changed(self): @@ -137,7 +137,7 @@ class StreamChangeCacheTests(unittest.TestCase): cache.get_entities_changed( ["user@foo.com", "bar@baz.net", "user@elsewhere.org"], stream_pos=2 ), - set(["bar@baz.net", "user@elsewhere.org"]), + {"bar@baz.net", "user@elsewhere.org"}, ) # Query all the entries mid-way through the stream, but include one @@ -153,7 +153,7 @@ class StreamChangeCacheTests(unittest.TestCase): ], stream_pos=2, ), - set(["bar@baz.net", "user@elsewhere.org"]), + {"bar@baz.net", "user@elsewhere.org"}, ) # Query all the entries, but before the first known point. We will get @@ -168,21 +168,13 @@ class StreamChangeCacheTests(unittest.TestCase): ], stream_pos=0, ), - set( - [ - "user@foo.com", - "bar@baz.net", - "user@elsewhere.org", - "not@here.website", - ] - ), + {"user@foo.com", "bar@baz.net", "user@elsewhere.org", "not@here.website"}, ) # Query a subset of the entries mid-way through the stream. We should # only get back the subset. self.assertEqual( - cache.get_entities_changed(["bar@baz.net"], stream_pos=2), - set(["bar@baz.net"]), + cache.get_entities_changed(["bar@baz.net"], stream_pos=2), {"bar@baz.net"}, ) def test_max_pos(self): diff --git a/tox.ini b/tox.ini index b9132a3177..b715ea0bff 100644 --- a/tox.ini +++ b/tox.ini @@ -123,6 +123,7 @@ skip_install = True basepython = python3.6 deps = flake8 + flake8-comprehensions black==19.10b0 # We pin so that our tests don't start failing on new releases of black. commands = python -m black --check --diff . -- cgit 1.4.1 From bbf8886a05be6a929556d6f09a1b6ce053a3c403 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Tue, 25 Feb 2020 16:56:55 +0000 Subject: Merge worker apps into one. (#6964) --- changelog.d/6964.misc | 1 + synapse/app/appservice.py | 156 +---- synapse/app/client_reader.py | 190 +----- synapse/app/event_creator.py | 186 +----- synapse/app/federation_reader.py | 172 +----- synapse/app/federation_sender.py | 303 +-------- synapse/app/frontend_proxy.py | 236 +------ synapse/app/generic_worker.py | 917 ++++++++++++++++++++++++++++ synapse/app/media_repository.py | 157 +---- synapse/app/pusher.py | 209 +------ synapse/app/synchrotron.py | 449 +------------- synapse/app/user_dir.py | 211 +------ synapse/replication/slave/storage/events.py | 20 + synapse/storage/data_stores/main/pusher.py | 156 ++--- tests/app/test_frontend_proxy.py | 12 +- tests/app/test_openid_listener.py | 4 +- 16 files changed, 1052 insertions(+), 2327 deletions(-) create mode 100644 changelog.d/6964.misc create mode 100644 synapse/app/generic_worker.py (limited to 'tests') diff --git a/changelog.d/6964.misc b/changelog.d/6964.misc new file mode 100644 index 0000000000..ec5c004bbe --- /dev/null +++ b/changelog.d/6964.misc @@ -0,0 +1 @@ +Merge worker apps together. diff --git a/synapse/app/appservice.py b/synapse/app/appservice.py index 2217d4a4fb..add43147b3 100644 --- a/synapse/app/appservice.py +++ b/synapse/app/appservice.py @@ -13,161 +13,11 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. -import logging -import sys - -from twisted.internet import defer, reactor -from twisted.web.resource import NoResource - -import synapse -from synapse import events -from synapse.app import _base -from synapse.config._base import ConfigError -from synapse.config.homeserver import HomeServerConfig -from synapse.config.logger import setup_logging -from synapse.http.site import SynapseSite -from synapse.logging.context import LoggingContext, run_in_background -from synapse.metrics import METRICS_PREFIX, MetricsResource, RegistryProxy -from synapse.replication.slave.storage.appservice import SlavedApplicationServiceStore -from synapse.replication.slave.storage.directory import DirectoryStore -from synapse.replication.slave.storage.events import SlavedEventStore -from synapse.replication.slave.storage.registration import SlavedRegistrationStore -from synapse.replication.tcp.client import ReplicationClientHandler -from synapse.server import HomeServer -from synapse.util.httpresourcetree import create_resource_tree -from synapse.util.manhole import manhole -from synapse.util.versionstring import get_version_string - -logger = logging.getLogger("synapse.app.appservice") - - -class AppserviceSlaveStore( - DirectoryStore, - SlavedEventStore, - SlavedApplicationServiceStore, - SlavedRegistrationStore, -): - pass - - -class AppserviceServer(HomeServer): - DATASTORE_CLASS = AppserviceSlaveStore - - def _listen_http(self, listener_config): - port = listener_config["port"] - bind_addresses = listener_config["bind_addresses"] - site_tag = listener_config.get("tag", port) - resources = {} - for res in listener_config["resources"]: - for name in res["names"]: - if name == "metrics": - resources[METRICS_PREFIX] = MetricsResource(RegistryProxy) - - root_resource = create_resource_tree(resources, NoResource()) - - _base.listen_tcp( - bind_addresses, - port, - SynapseSite( - "synapse.access.http.%s" % (site_tag,), - site_tag, - listener_config, - root_resource, - self.version_string, - ), - ) - - logger.info("Synapse appservice now listening on port %d", port) - - def start_listening(self, listeners): - for listener in listeners: - if listener["type"] == "http": - self._listen_http(listener) - elif listener["type"] == "manhole": - _base.listen_tcp( - listener["bind_addresses"], - listener["port"], - manhole( - username="matrix", password="rabbithole", globals={"hs": self} - ), - ) - elif listener["type"] == "metrics": - if not self.get_config().enable_metrics: - logger.warning( - ( - "Metrics listener configured, but " - "enable_metrics is not True!" - ) - ) - else: - _base.listen_metrics(listener["bind_addresses"], listener["port"]) - else: - logger.warning("Unrecognized listener type: %s", listener["type"]) - - self.get_tcp_replication().start_replication(self) - - def build_tcp_replication(self): - return ASReplicationHandler(self) +import sys -class ASReplicationHandler(ReplicationClientHandler): - def __init__(self, hs): - super(ASReplicationHandler, self).__init__(hs.get_datastore()) - self.appservice_handler = hs.get_application_service_handler() - - async def on_rdata(self, stream_name, token, rows): - await super(ASReplicationHandler, self).on_rdata(stream_name, token, rows) - - if stream_name == "events": - max_stream_id = self.store.get_room_max_stream_ordering() - run_in_background(self._notify_app_services, max_stream_id) - - @defer.inlineCallbacks - def _notify_app_services(self, room_stream_id): - try: - yield self.appservice_handler.notify_interested_services(room_stream_id) - except Exception: - logger.exception("Error notifying application services of event") - - -def start(config_options): - try: - config = HomeServerConfig.load_config("Synapse appservice", config_options) - except ConfigError as e: - sys.stderr.write("\n" + str(e) + "\n") - sys.exit(1) - - assert config.worker_app == "synapse.app.appservice" - - events.USE_FROZEN_DICTS = config.use_frozen_dicts - - if config.notify_appservices: - sys.stderr.write( - "\nThe appservices must be disabled in the main synapse process" - "\nbefore they can be run in a separate worker." - "\nPlease add ``notify_appservices: false`` to the main config" - "\n" - ) - sys.exit(1) - - # Force the pushers to start since they will be disabled in the main config - config.notify_appservices = True - - ps = AppserviceServer( - config.server_name, - config=config, - version_string="Synapse/" + get_version_string(synapse), - ) - - setup_logging(ps, config, use_worker_options=True) - - ps.setup() - reactor.addSystemEventTrigger( - "before", "startup", _base.start, ps, config.worker_listeners - ) - - _base.start_worker_reactor("synapse-appservice", config) - +from synapse.app.generic_worker import start +from synapse.util.logcontext import LoggingContext if __name__ == "__main__": with LoggingContext("main"): diff --git a/synapse/app/client_reader.py b/synapse/app/client_reader.py index 7fa91a3b11..add43147b3 100644 --- a/synapse/app/client_reader.py +++ b/synapse/app/client_reader.py @@ -13,195 +13,11 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. -import logging -import sys - -from twisted.internet import reactor -from twisted.web.resource import NoResource - -import synapse -from synapse import events -from synapse.app import _base -from synapse.config._base import ConfigError -from synapse.config.homeserver import HomeServerConfig -from synapse.config.logger import setup_logging -from synapse.http.server import JsonResource -from synapse.http.site import SynapseSite -from synapse.logging.context import LoggingContext -from synapse.metrics import METRICS_PREFIX, MetricsResource, RegistryProxy -from synapse.replication.slave.storage._base import BaseSlavedStore -from synapse.replication.slave.storage.account_data import SlavedAccountDataStore -from synapse.replication.slave.storage.appservice import SlavedApplicationServiceStore -from synapse.replication.slave.storage.client_ips import SlavedClientIpStore -from synapse.replication.slave.storage.deviceinbox import SlavedDeviceInboxStore -from synapse.replication.slave.storage.devices import SlavedDeviceStore -from synapse.replication.slave.storage.directory import DirectoryStore -from synapse.replication.slave.storage.events import SlavedEventStore -from synapse.replication.slave.storage.groups import SlavedGroupServerStore -from synapse.replication.slave.storage.keys import SlavedKeyStore -from synapse.replication.slave.storage.profile import SlavedProfileStore -from synapse.replication.slave.storage.push_rule import SlavedPushRuleStore -from synapse.replication.slave.storage.receipts import SlavedReceiptsStore -from synapse.replication.slave.storage.registration import SlavedRegistrationStore -from synapse.replication.slave.storage.room import RoomStore -from synapse.replication.slave.storage.transactions import SlavedTransactionStore -from synapse.replication.tcp.client import ReplicationClientHandler -from synapse.rest.client.v1.login import LoginRestServlet -from synapse.rest.client.v1.push_rule import PushRuleRestServlet -from synapse.rest.client.v1.room import ( - JoinedRoomMemberListRestServlet, - PublicRoomListRestServlet, - RoomEventContextServlet, - RoomMemberListRestServlet, - RoomMessageListRestServlet, - RoomStateRestServlet, -) -from synapse.rest.client.v1.voip import VoipRestServlet -from synapse.rest.client.v2_alpha import groups -from synapse.rest.client.v2_alpha.account import ThreepidRestServlet -from synapse.rest.client.v2_alpha.keys import KeyChangesServlet, KeyQueryServlet -from synapse.rest.client.v2_alpha.register import RegisterRestServlet -from synapse.rest.client.versions import VersionsRestServlet -from synapse.server import HomeServer -from synapse.storage.data_stores.main.monthly_active_users import ( - MonthlyActiveUsersWorkerStore, -) -from synapse.util.httpresourcetree import create_resource_tree -from synapse.util.manhole import manhole -from synapse.util.versionstring import get_version_string - -logger = logging.getLogger("synapse.app.client_reader") - - -class ClientReaderSlavedStore( - SlavedDeviceInboxStore, - SlavedDeviceStore, - SlavedReceiptsStore, - SlavedPushRuleStore, - SlavedGroupServerStore, - SlavedAccountDataStore, - SlavedEventStore, - SlavedKeyStore, - RoomStore, - DirectoryStore, - SlavedApplicationServiceStore, - SlavedRegistrationStore, - SlavedTransactionStore, - SlavedProfileStore, - SlavedClientIpStore, - MonthlyActiveUsersWorkerStore, - BaseSlavedStore, -): - pass - - -class ClientReaderServer(HomeServer): - DATASTORE_CLASS = ClientReaderSlavedStore - - def _listen_http(self, listener_config): - port = listener_config["port"] - bind_addresses = listener_config["bind_addresses"] - site_tag = listener_config.get("tag", port) - resources = {} - for res in listener_config["resources"]: - for name in res["names"]: - if name == "metrics": - resources[METRICS_PREFIX] = MetricsResource(RegistryProxy) - elif name == "client": - resource = JsonResource(self, canonical_json=False) - - PublicRoomListRestServlet(self).register(resource) - RoomMemberListRestServlet(self).register(resource) - JoinedRoomMemberListRestServlet(self).register(resource) - RoomStateRestServlet(self).register(resource) - RoomEventContextServlet(self).register(resource) - RoomMessageListRestServlet(self).register(resource) - RegisterRestServlet(self).register(resource) - LoginRestServlet(self).register(resource) - ThreepidRestServlet(self).register(resource) - KeyQueryServlet(self).register(resource) - KeyChangesServlet(self).register(resource) - VoipRestServlet(self).register(resource) - PushRuleRestServlet(self).register(resource) - VersionsRestServlet(self).register(resource) - - groups.register_servlets(self, resource) - - resources.update({"/_matrix/client": resource}) - - root_resource = create_resource_tree(resources, NoResource()) - _base.listen_tcp( - bind_addresses, - port, - SynapseSite( - "synapse.access.http.%s" % (site_tag,), - site_tag, - listener_config, - root_resource, - self.version_string, - ), - ) - - logger.info("Synapse client reader now listening on port %d", port) - - def start_listening(self, listeners): - for listener in listeners: - if listener["type"] == "http": - self._listen_http(listener) - elif listener["type"] == "manhole": - _base.listen_tcp( - listener["bind_addresses"], - listener["port"], - manhole( - username="matrix", password="rabbithole", globals={"hs": self} - ), - ) - elif listener["type"] == "metrics": - if not self.get_config().enable_metrics: - logger.warning( - ( - "Metrics listener configured, but " - "enable_metrics is not True!" - ) - ) - else: - _base.listen_metrics(listener["bind_addresses"], listener["port"]) - else: - logger.warning("Unrecognized listener type: %s", listener["type"]) - - self.get_tcp_replication().start_replication(self) - - def build_tcp_replication(self): - return ReplicationClientHandler(self.get_datastore()) - - -def start(config_options): - try: - config = HomeServerConfig.load_config("Synapse client reader", config_options) - except ConfigError as e: - sys.stderr.write("\n" + str(e) + "\n") - sys.exit(1) - - assert config.worker_app == "synapse.app.client_reader" - - events.USE_FROZEN_DICTS = config.use_frozen_dicts - - ss = ClientReaderServer( - config.server_name, - config=config, - version_string="Synapse/" + get_version_string(synapse), - ) - - setup_logging(ss, config, use_worker_options=True) - - ss.setup() - reactor.addSystemEventTrigger( - "before", "startup", _base.start, ss, config.worker_listeners - ) - - _base.start_worker_reactor("synapse-client-reader", config) +import sys +from synapse.app.generic_worker import start +from synapse.util.logcontext import LoggingContext if __name__ == "__main__": with LoggingContext("main"): diff --git a/synapse/app/event_creator.py b/synapse/app/event_creator.py index 58e5b354f6..e9c098c4e7 100644 --- a/synapse/app/event_creator.py +++ b/synapse/app/event_creator.py @@ -13,191 +13,11 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. -import logging -import sys - -from twisted.internet import reactor -from twisted.web.resource import NoResource - -import synapse -from synapse import events -from synapse.app import _base -from synapse.config._base import ConfigError -from synapse.config.homeserver import HomeServerConfig -from synapse.config.logger import setup_logging -from synapse.http.server import JsonResource -from synapse.http.site import SynapseSite -from synapse.logging.context import LoggingContext -from synapse.metrics import METRICS_PREFIX, MetricsResource, RegistryProxy -from synapse.replication.slave.storage._base import BaseSlavedStore -from synapse.replication.slave.storage.account_data import SlavedAccountDataStore -from synapse.replication.slave.storage.appservice import SlavedApplicationServiceStore -from synapse.replication.slave.storage.client_ips import SlavedClientIpStore -from synapse.replication.slave.storage.devices import SlavedDeviceStore -from synapse.replication.slave.storage.directory import DirectoryStore -from synapse.replication.slave.storage.events import SlavedEventStore -from synapse.replication.slave.storage.profile import SlavedProfileStore -from synapse.replication.slave.storage.push_rule import SlavedPushRuleStore -from synapse.replication.slave.storage.pushers import SlavedPusherStore -from synapse.replication.slave.storage.receipts import SlavedReceiptsStore -from synapse.replication.slave.storage.registration import SlavedRegistrationStore -from synapse.replication.slave.storage.room import RoomStore -from synapse.replication.slave.storage.transactions import SlavedTransactionStore -from synapse.replication.tcp.client import ReplicationClientHandler -from synapse.rest.client.v1.profile import ( - ProfileAvatarURLRestServlet, - ProfileDisplaynameRestServlet, - ProfileRestServlet, -) -from synapse.rest.client.v1.room import ( - JoinRoomAliasServlet, - RoomMembershipRestServlet, - RoomSendEventRestServlet, - RoomStateEventRestServlet, -) -from synapse.server import HomeServer -from synapse.storage.data_stores.main.monthly_active_users import ( - MonthlyActiveUsersWorkerStore, -) -from synapse.storage.data_stores.main.user_directory import UserDirectoryStore -from synapse.util.httpresourcetree import create_resource_tree -from synapse.util.manhole import manhole -from synapse.util.versionstring import get_version_string - -logger = logging.getLogger("synapse.app.event_creator") - - -class EventCreatorSlavedStore( - # FIXME(#3714): We need to add UserDirectoryStore as we write directly - # rather than going via the correct worker. - UserDirectoryStore, - DirectoryStore, - SlavedTransactionStore, - SlavedProfileStore, - SlavedAccountDataStore, - SlavedPusherStore, - SlavedReceiptsStore, - SlavedPushRuleStore, - SlavedDeviceStore, - SlavedClientIpStore, - SlavedApplicationServiceStore, - SlavedEventStore, - SlavedRegistrationStore, - RoomStore, - MonthlyActiveUsersWorkerStore, - BaseSlavedStore, -): - pass - - -class EventCreatorServer(HomeServer): - DATASTORE_CLASS = EventCreatorSlavedStore - - def _listen_http(self, listener_config): - port = listener_config["port"] - bind_addresses = listener_config["bind_addresses"] - site_tag = listener_config.get("tag", port) - resources = {} - for res in listener_config["resources"]: - for name in res["names"]: - if name == "metrics": - resources[METRICS_PREFIX] = MetricsResource(RegistryProxy) - elif name == "client": - resource = JsonResource(self, canonical_json=False) - RoomSendEventRestServlet(self).register(resource) - RoomMembershipRestServlet(self).register(resource) - RoomStateEventRestServlet(self).register(resource) - JoinRoomAliasServlet(self).register(resource) - ProfileAvatarURLRestServlet(self).register(resource) - ProfileDisplaynameRestServlet(self).register(resource) - ProfileRestServlet(self).register(resource) - resources.update( - { - "/_matrix/client/r0": resource, - "/_matrix/client/unstable": resource, - "/_matrix/client/v2_alpha": resource, - "/_matrix/client/api/v1": resource, - } - ) - - root_resource = create_resource_tree(resources, NoResource()) - - _base.listen_tcp( - bind_addresses, - port, - SynapseSite( - "synapse.access.http.%s" % (site_tag,), - site_tag, - listener_config, - root_resource, - self.version_string, - ), - ) - - logger.info("Synapse event creator now listening on port %d", port) - - def start_listening(self, listeners): - for listener in listeners: - if listener["type"] == "http": - self._listen_http(listener) - elif listener["type"] == "manhole": - _base.listen_tcp( - listener["bind_addresses"], - listener["port"], - manhole( - username="matrix", password="rabbithole", globals={"hs": self} - ), - ) - elif listener["type"] == "metrics": - if not self.get_config().enable_metrics: - logger.warning( - ( - "Metrics listener configured, but " - "enable_metrics is not True!" - ) - ) - else: - _base.listen_metrics(listener["bind_addresses"], listener["port"]) - else: - logger.warning("Unrecognized listener type: %s", listener["type"]) - self.get_tcp_replication().start_replication(self) - - def build_tcp_replication(self): - return ReplicationClientHandler(self.get_datastore()) - - -def start(config_options): - try: - config = HomeServerConfig.load_config("Synapse event creator", config_options) - except ConfigError as e: - sys.stderr.write("\n" + str(e) + "\n") - sys.exit(1) - - assert config.worker_app == "synapse.app.event_creator" - - assert config.worker_replication_http_port is not None - - # This should only be done on the user directory worker or the master - config.update_user_directory = False - - events.USE_FROZEN_DICTS = config.use_frozen_dicts - - ss = EventCreatorServer( - config.server_name, - config=config, - version_string="Synapse/" + get_version_string(synapse), - ) - - setup_logging(ss, config, use_worker_options=True) - - ss.setup() - reactor.addSystemEventTrigger( - "before", "startup", _base.start, ss, config.worker_listeners - ) - - _base.start_worker_reactor("synapse-event-creator", config) +import sys +from synapse.app.generic_worker import start +from synapse.util.logcontext import LoggingContext if __name__ == "__main__": with LoggingContext("main"): diff --git a/synapse/app/federation_reader.py b/synapse/app/federation_reader.py index d055d11b23..add43147b3 100644 --- a/synapse/app/federation_reader.py +++ b/synapse/app/federation_reader.py @@ -13,177 +13,11 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. -import logging -import sys - -from twisted.internet import reactor -from twisted.web.resource import NoResource - -import synapse -from synapse import events -from synapse.api.urls import FEDERATION_PREFIX, SERVER_KEY_V2_PREFIX -from synapse.app import _base -from synapse.config._base import ConfigError -from synapse.config.homeserver import HomeServerConfig -from synapse.config.logger import setup_logging -from synapse.federation.transport.server import TransportLayerServer -from synapse.http.site import SynapseSite -from synapse.logging.context import LoggingContext -from synapse.metrics import METRICS_PREFIX, MetricsResource, RegistryProxy -from synapse.replication.slave.storage._base import BaseSlavedStore -from synapse.replication.slave.storage.account_data import SlavedAccountDataStore -from synapse.replication.slave.storage.appservice import SlavedApplicationServiceStore -from synapse.replication.slave.storage.devices import SlavedDeviceStore -from synapse.replication.slave.storage.directory import DirectoryStore -from synapse.replication.slave.storage.events import SlavedEventStore -from synapse.replication.slave.storage.groups import SlavedGroupServerStore -from synapse.replication.slave.storage.keys import SlavedKeyStore -from synapse.replication.slave.storage.profile import SlavedProfileStore -from synapse.replication.slave.storage.push_rule import SlavedPushRuleStore -from synapse.replication.slave.storage.pushers import SlavedPusherStore -from synapse.replication.slave.storage.receipts import SlavedReceiptsStore -from synapse.replication.slave.storage.registration import SlavedRegistrationStore -from synapse.replication.slave.storage.room import RoomStore -from synapse.replication.slave.storage.transactions import SlavedTransactionStore -from synapse.replication.tcp.client import ReplicationClientHandler -from synapse.rest.key.v2 import KeyApiV2Resource -from synapse.server import HomeServer -from synapse.storage.data_stores.main.monthly_active_users import ( - MonthlyActiveUsersWorkerStore, -) -from synapse.util.httpresourcetree import create_resource_tree -from synapse.util.manhole import manhole -from synapse.util.versionstring import get_version_string - -logger = logging.getLogger("synapse.app.federation_reader") - - -class FederationReaderSlavedStore( - SlavedAccountDataStore, - SlavedProfileStore, - SlavedApplicationServiceStore, - SlavedPusherStore, - SlavedPushRuleStore, - SlavedReceiptsStore, - SlavedEventStore, - SlavedKeyStore, - SlavedRegistrationStore, - SlavedGroupServerStore, - SlavedDeviceStore, - RoomStore, - DirectoryStore, - SlavedTransactionStore, - MonthlyActiveUsersWorkerStore, - BaseSlavedStore, -): - pass - - -class FederationReaderServer(HomeServer): - DATASTORE_CLASS = FederationReaderSlavedStore - - def _listen_http(self, listener_config): - port = listener_config["port"] - bind_addresses = listener_config["bind_addresses"] - site_tag = listener_config.get("tag", port) - resources = {} - for res in listener_config["resources"]: - for name in res["names"]: - if name == "metrics": - resources[METRICS_PREFIX] = MetricsResource(RegistryProxy) - elif name == "federation": - resources.update({FEDERATION_PREFIX: TransportLayerServer(self)}) - if name == "openid" and "federation" not in res["names"]: - # Only load the openid resource separately if federation resource - # is not specified since federation resource includes openid - # resource. - resources.update( - { - FEDERATION_PREFIX: TransportLayerServer( - self, servlet_groups=["openid"] - ) - } - ) - - if name in ["keys", "federation"]: - resources[SERVER_KEY_V2_PREFIX] = KeyApiV2Resource(self) - - root_resource = create_resource_tree(resources, NoResource()) - - _base.listen_tcp( - bind_addresses, - port, - SynapseSite( - "synapse.access.http.%s" % (site_tag,), - site_tag, - listener_config, - root_resource, - self.version_string, - ), - reactor=self.get_reactor(), - ) - logger.info("Synapse federation reader now listening on port %d", port) - - def start_listening(self, listeners): - for listener in listeners: - if listener["type"] == "http": - self._listen_http(listener) - elif listener["type"] == "manhole": - _base.listen_tcp( - listener["bind_addresses"], - listener["port"], - manhole( - username="matrix", password="rabbithole", globals={"hs": self} - ), - ) - elif listener["type"] == "metrics": - if not self.get_config().enable_metrics: - logger.warning( - ( - "Metrics listener configured, but " - "enable_metrics is not True!" - ) - ) - else: - _base.listen_metrics(listener["bind_addresses"], listener["port"]) - else: - logger.warning("Unrecognized listener type: %s", listener["type"]) - - self.get_tcp_replication().start_replication(self) - - def build_tcp_replication(self): - return ReplicationClientHandler(self.get_datastore()) - - -def start(config_options): - try: - config = HomeServerConfig.load_config( - "Synapse federation reader", config_options - ) - except ConfigError as e: - sys.stderr.write("\n" + str(e) + "\n") - sys.exit(1) - - assert config.worker_app == "synapse.app.federation_reader" - - events.USE_FROZEN_DICTS = config.use_frozen_dicts - - ss = FederationReaderServer( - config.server_name, - config=config, - version_string="Synapse/" + get_version_string(synapse), - ) - - setup_logging(ss, config, use_worker_options=True) - - ss.setup() - reactor.addSystemEventTrigger( - "before", "startup", _base.start, ss, config.worker_listeners - ) - - _base.start_worker_reactor("synapse-federation-reader", config) +import sys +from synapse.app.generic_worker import start +from synapse.util.logcontext import LoggingContext if __name__ == "__main__": with LoggingContext("main"): diff --git a/synapse/app/federation_sender.py b/synapse/app/federation_sender.py index b7fcf80ddc..add43147b3 100644 --- a/synapse/app/federation_sender.py +++ b/synapse/app/federation_sender.py @@ -13,308 +13,11 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. -import logging -import sys - -from twisted.internet import defer, reactor -from twisted.web.resource import NoResource - -import synapse -from synapse import events -from synapse.app import _base -from synapse.config._base import ConfigError -from synapse.config.homeserver import HomeServerConfig -from synapse.config.logger import setup_logging -from synapse.federation import send_queue -from synapse.http.site import SynapseSite -from synapse.logging.context import LoggingContext, run_in_background -from synapse.metrics import METRICS_PREFIX, MetricsResource, RegistryProxy -from synapse.metrics.background_process_metrics import run_as_background_process -from synapse.replication.slave.storage.deviceinbox import SlavedDeviceInboxStore -from synapse.replication.slave.storage.devices import SlavedDeviceStore -from synapse.replication.slave.storage.events import SlavedEventStore -from synapse.replication.slave.storage.presence import SlavedPresenceStore -from synapse.replication.slave.storage.receipts import SlavedReceiptsStore -from synapse.replication.slave.storage.registration import SlavedRegistrationStore -from synapse.replication.slave.storage.transactions import SlavedTransactionStore -from synapse.replication.tcp.client import ReplicationClientHandler -from synapse.replication.tcp.streams._base import ( - DeviceListsStream, - ReceiptsStream, - ToDeviceStream, -) -from synapse.server import HomeServer -from synapse.storage.database import Database -from synapse.types import ReadReceipt -from synapse.util.async_helpers import Linearizer -from synapse.util.httpresourcetree import create_resource_tree -from synapse.util.manhole import manhole -from synapse.util.versionstring import get_version_string - -logger = logging.getLogger("synapse.app.federation_sender") - - -class FederationSenderSlaveStore( - SlavedDeviceInboxStore, - SlavedTransactionStore, - SlavedReceiptsStore, - SlavedEventStore, - SlavedRegistrationStore, - SlavedDeviceStore, - SlavedPresenceStore, -): - def __init__(self, database: Database, db_conn, hs): - super(FederationSenderSlaveStore, self).__init__(database, db_conn, hs) - - # We pull out the current federation stream position now so that we - # always have a known value for the federation position in memory so - # that we don't have to bounce via a deferred once when we start the - # replication streams. - self.federation_out_pos_startup = self._get_federation_out_pos(db_conn) - - def _get_federation_out_pos(self, db_conn): - sql = "SELECT stream_id FROM federation_stream_position WHERE type = ?" - sql = self.database_engine.convert_param_style(sql) - - txn = db_conn.cursor() - txn.execute(sql, ("federation",)) - rows = txn.fetchall() - txn.close() - - return rows[0][0] if rows else -1 - - -class FederationSenderServer(HomeServer): - DATASTORE_CLASS = FederationSenderSlaveStore - - def _listen_http(self, listener_config): - port = listener_config["port"] - bind_addresses = listener_config["bind_addresses"] - site_tag = listener_config.get("tag", port) - resources = {} - for res in listener_config["resources"]: - for name in res["names"]: - if name == "metrics": - resources[METRICS_PREFIX] = MetricsResource(RegistryProxy) - - root_resource = create_resource_tree(resources, NoResource()) - - _base.listen_tcp( - bind_addresses, - port, - SynapseSite( - "synapse.access.http.%s" % (site_tag,), - site_tag, - listener_config, - root_resource, - self.version_string, - ), - ) - - logger.info("Synapse federation_sender now listening on port %d", port) - - def start_listening(self, listeners): - for listener in listeners: - if listener["type"] == "http": - self._listen_http(listener) - elif listener["type"] == "manhole": - _base.listen_tcp( - listener["bind_addresses"], - listener["port"], - manhole( - username="matrix", password="rabbithole", globals={"hs": self} - ), - ) - elif listener["type"] == "metrics": - if not self.get_config().enable_metrics: - logger.warning( - ( - "Metrics listener configured, but " - "enable_metrics is not True!" - ) - ) - else: - _base.listen_metrics(listener["bind_addresses"], listener["port"]) - else: - logger.warning("Unrecognized listener type: %s", listener["type"]) - - self.get_tcp_replication().start_replication(self) - - def build_tcp_replication(self): - return FederationSenderReplicationHandler(self) - - -class FederationSenderReplicationHandler(ReplicationClientHandler): - def __init__(self, hs): - super(FederationSenderReplicationHandler, self).__init__(hs.get_datastore()) - self.send_handler = FederationSenderHandler(hs, self) - - async def on_rdata(self, stream_name, token, rows): - await super(FederationSenderReplicationHandler, self).on_rdata( - stream_name, token, rows - ) - self.send_handler.process_replication_rows(stream_name, token, rows) - - def get_streams_to_replicate(self): - args = super( - FederationSenderReplicationHandler, self - ).get_streams_to_replicate() - args.update(self.send_handler.stream_positions()) - return args - - def on_remote_server_up(self, server: str): - """Called when get a new REMOTE_SERVER_UP command.""" - - # Let's wake up the transaction queue for the server in case we have - # pending stuff to send to it. - self.send_handler.wake_destination(server) - - -def start(config_options): - try: - config = HomeServerConfig.load_config( - "Synapse federation sender", config_options - ) - except ConfigError as e: - sys.stderr.write("\n" + str(e) + "\n") - sys.exit(1) - assert config.worker_app == "synapse.app.federation_sender" - - events.USE_FROZEN_DICTS = config.use_frozen_dicts - - if config.send_federation: - sys.stderr.write( - "\nThe send_federation must be disabled in the main synapse process" - "\nbefore they can be run in a separate worker." - "\nPlease add ``send_federation: false`` to the main config" - "\n" - ) - sys.exit(1) - - # Force the pushers to start since they will be disabled in the main config - config.send_federation = True - - ss = FederationSenderServer( - config.server_name, - config=config, - version_string="Synapse/" + get_version_string(synapse), - ) - - setup_logging(ss, config, use_worker_options=True) - - ss.setup() - reactor.addSystemEventTrigger( - "before", "startup", _base.start, ss, config.worker_listeners - ) - - _base.start_worker_reactor("synapse-federation-sender", config) - - -class FederationSenderHandler(object): - """Processes the replication stream and forwards the appropriate entries - to the federation sender. - """ - - def __init__(self, hs: FederationSenderServer, replication_client): - self.store = hs.get_datastore() - self._is_mine_id = hs.is_mine_id - self.federation_sender = hs.get_federation_sender() - self.replication_client = replication_client - - self.federation_position = self.store.federation_out_pos_startup - self._fed_position_linearizer = Linearizer(name="_fed_position_linearizer") - - self._last_ack = self.federation_position - - self._room_serials = {} - self._room_typing = {} - - def on_start(self): - # There may be some events that are persisted but haven't been sent, - # so send them now. - self.federation_sender.notify_new_events( - self.store.get_room_max_stream_ordering() - ) - - def wake_destination(self, server: str): - self.federation_sender.wake_destination(server) - - def stream_positions(self): - return {"federation": self.federation_position} - - def process_replication_rows(self, stream_name, token, rows): - # The federation stream contains things that we want to send out, e.g. - # presence, typing, etc. - if stream_name == "federation": - send_queue.process_rows_for_federation(self.federation_sender, rows) - run_in_background(self.update_token, token) - - # We also need to poke the federation sender when new events happen - elif stream_name == "events": - self.federation_sender.notify_new_events(token) - - # ... and when new receipts happen - elif stream_name == ReceiptsStream.NAME: - run_as_background_process( - "process_receipts_for_federation", self._on_new_receipts, rows - ) - - # ... as well as device updates and messages - elif stream_name == DeviceListsStream.NAME: - hosts = {row.destination for row in rows} - for host in hosts: - self.federation_sender.send_device_messages(host) - - elif stream_name == ToDeviceStream.NAME: - # The to_device stream includes stuff to be pushed to both local - # clients and remote servers, so we ignore entities that start with - # '@' (since they'll be local users rather than destinations). - hosts = {row.entity for row in rows if not row.entity.startswith("@")} - for host in hosts: - self.federation_sender.send_device_messages(host) - - @defer.inlineCallbacks - def _on_new_receipts(self, rows): - """ - Args: - rows (iterable[synapse.replication.tcp.streams.ReceiptsStreamRow]): - new receipts to be processed - """ - for receipt in rows: - # we only want to send on receipts for our own users - if not self._is_mine_id(receipt.user_id): - continue - receipt_info = ReadReceipt( - receipt.room_id, - receipt.receipt_type, - receipt.user_id, - [receipt.event_id], - receipt.data, - ) - yield self.federation_sender.send_read_receipt(receipt_info) - - @defer.inlineCallbacks - def update_token(self, token): - try: - self.federation_position = token - - # We linearize here to ensure we don't have races updating the token - with (yield self._fed_position_linearizer.queue(None)): - if self._last_ack < self.federation_position: - yield self.store.update_federation_out_pos( - "federation", self.federation_position - ) - - # We ACK this token over replication so that the master can drop - # its in memory queues - self.replication_client.send_federation_ack( - self.federation_position - ) - self._last_ack = self.federation_position - except Exception: - logger.exception("Error updating federation stream position") +import sys +from synapse.app.generic_worker import start +from synapse.util.logcontext import LoggingContext if __name__ == "__main__": with LoggingContext("main"): diff --git a/synapse/app/frontend_proxy.py b/synapse/app/frontend_proxy.py index 30e435eead..add43147b3 100644 --- a/synapse/app/frontend_proxy.py +++ b/synapse/app/frontend_proxy.py @@ -13,241 +13,11 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. -import logging -import sys - -from twisted.internet import defer, reactor -from twisted.web.resource import NoResource - -import synapse -from synapse import events -from synapse.api.errors import HttpResponseException, SynapseError -from synapse.app import _base -from synapse.config._base import ConfigError -from synapse.config.homeserver import HomeServerConfig -from synapse.config.logger import setup_logging -from synapse.http.server import JsonResource -from synapse.http.servlet import RestServlet, parse_json_object_from_request -from synapse.http.site import SynapseSite -from synapse.logging.context import LoggingContext -from synapse.metrics import METRICS_PREFIX, MetricsResource, RegistryProxy -from synapse.replication.slave.storage._base import BaseSlavedStore -from synapse.replication.slave.storage.appservice import SlavedApplicationServiceStore -from synapse.replication.slave.storage.client_ips import SlavedClientIpStore -from synapse.replication.slave.storage.devices import SlavedDeviceStore -from synapse.replication.slave.storage.registration import SlavedRegistrationStore -from synapse.replication.tcp.client import ReplicationClientHandler -from synapse.rest.client.v2_alpha._base import client_patterns -from synapse.server import HomeServer -from synapse.util.httpresourcetree import create_resource_tree -from synapse.util.manhole import manhole -from synapse.util.versionstring import get_version_string - -logger = logging.getLogger("synapse.app.frontend_proxy") - - -class PresenceStatusStubServlet(RestServlet): - PATTERNS = client_patterns("/presence/(?P[^/]*)/status") - - def __init__(self, hs): - super(PresenceStatusStubServlet, self).__init__() - self.http_client = hs.get_simple_http_client() - self.auth = hs.get_auth() - self.main_uri = hs.config.worker_main_http_uri - - @defer.inlineCallbacks - def on_GET(self, request, user_id): - # Pass through the auth headers, if any, in case the access token - # is there. - auth_headers = request.requestHeaders.getRawHeaders("Authorization", []) - headers = {"Authorization": auth_headers} - - try: - result = yield self.http_client.get_json( - self.main_uri + request.uri.decode("ascii"), headers=headers - ) - except HttpResponseException as e: - raise e.to_synapse_error() - - return 200, result - - @defer.inlineCallbacks - def on_PUT(self, request, user_id): - yield self.auth.get_user_by_req(request) - return 200, {} - - -class KeyUploadServlet(RestServlet): - PATTERNS = client_patterns("/keys/upload(/(?P[^/]+))?$") - - def __init__(self, hs): - """ - Args: - hs (synapse.server.HomeServer): server - """ - super(KeyUploadServlet, self).__init__() - self.auth = hs.get_auth() - self.store = hs.get_datastore() - self.http_client = hs.get_simple_http_client() - self.main_uri = hs.config.worker_main_http_uri - - @defer.inlineCallbacks - def on_POST(self, request, device_id): - requester = yield self.auth.get_user_by_req(request, allow_guest=True) - user_id = requester.user.to_string() - body = parse_json_object_from_request(request) - - if device_id is not None: - # passing the device_id here is deprecated; however, we allow it - # for now for compatibility with older clients. - if requester.device_id is not None and device_id != requester.device_id: - logger.warning( - "Client uploading keys for a different device " - "(logged in as %s, uploading for %s)", - requester.device_id, - device_id, - ) - else: - device_id = requester.device_id - - if device_id is None: - raise SynapseError( - 400, "To upload keys, you must pass device_id when authenticating" - ) - - if body: - # They're actually trying to upload something, proxy to main synapse. - # Pass through the auth headers, if any, in case the access token - # is there. - auth_headers = request.requestHeaders.getRawHeaders(b"Authorization", []) - headers = {"Authorization": auth_headers} - result = yield self.http_client.post_json_get_json( - self.main_uri + request.uri.decode("ascii"), body, headers=headers - ) - - return 200, result - else: - # Just interested in counts. - result = yield self.store.count_e2e_one_time_keys(user_id, device_id) - return 200, {"one_time_key_counts": result} - - -class FrontendProxySlavedStore( - SlavedDeviceStore, - SlavedClientIpStore, - SlavedApplicationServiceStore, - SlavedRegistrationStore, - BaseSlavedStore, -): - pass +import sys -class FrontendProxyServer(HomeServer): - DATASTORE_CLASS = FrontendProxySlavedStore - - def _listen_http(self, listener_config): - port = listener_config["port"] - bind_addresses = listener_config["bind_addresses"] - site_tag = listener_config.get("tag", port) - resources = {} - for res in listener_config["resources"]: - for name in res["names"]: - if name == "metrics": - resources[METRICS_PREFIX] = MetricsResource(RegistryProxy) - elif name == "client": - resource = JsonResource(self, canonical_json=False) - KeyUploadServlet(self).register(resource) - - # If presence is disabled, use the stub servlet that does - # not allow sending presence - if not self.config.use_presence: - PresenceStatusStubServlet(self).register(resource) - - resources.update( - { - "/_matrix/client/r0": resource, - "/_matrix/client/unstable": resource, - "/_matrix/client/v2_alpha": resource, - "/_matrix/client/api/v1": resource, - } - ) - - root_resource = create_resource_tree(resources, NoResource()) - - _base.listen_tcp( - bind_addresses, - port, - SynapseSite( - "synapse.access.http.%s" % (site_tag,), - site_tag, - listener_config, - root_resource, - self.version_string, - ), - reactor=self.get_reactor(), - ) - - logger.info("Synapse client reader now listening on port %d", port) - - def start_listening(self, listeners): - for listener in listeners: - if listener["type"] == "http": - self._listen_http(listener) - elif listener["type"] == "manhole": - _base.listen_tcp( - listener["bind_addresses"], - listener["port"], - manhole( - username="matrix", password="rabbithole", globals={"hs": self} - ), - ) - elif listener["type"] == "metrics": - if not self.get_config().enable_metrics: - logger.warning( - ( - "Metrics listener configured, but " - "enable_metrics is not True!" - ) - ) - else: - _base.listen_metrics(listener["bind_addresses"], listener["port"]) - else: - logger.warning("Unrecognized listener type: %s", listener["type"]) - - self.get_tcp_replication().start_replication(self) - - def build_tcp_replication(self): - return ReplicationClientHandler(self.get_datastore()) - - -def start(config_options): - try: - config = HomeServerConfig.load_config("Synapse frontend proxy", config_options) - except ConfigError as e: - sys.stderr.write("\n" + str(e) + "\n") - sys.exit(1) - - assert config.worker_app == "synapse.app.frontend_proxy" - - assert config.worker_main_http_uri is not None - - events.USE_FROZEN_DICTS = config.use_frozen_dicts - - ss = FrontendProxyServer( - config.server_name, - config=config, - version_string="Synapse/" + get_version_string(synapse), - ) - - setup_logging(ss, config, use_worker_options=True) - - ss.setup() - reactor.addSystemEventTrigger( - "before", "startup", _base.start, ss, config.worker_listeners - ) - - _base.start_worker_reactor("synapse-frontend-proxy", config) - +from synapse.app.generic_worker import start +from synapse.util.logcontext import LoggingContext if __name__ == "__main__": with LoggingContext("main"): diff --git a/synapse/app/generic_worker.py b/synapse/app/generic_worker.py new file mode 100644 index 0000000000..30efd39092 --- /dev/null +++ b/synapse/app/generic_worker.py @@ -0,0 +1,917 @@ +#!/usr/bin/env python +# -*- coding: utf-8 -*- +# Copyright 2016 OpenMarket Ltd +# Copyright 2020 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import contextlib +import logging +import sys + +from twisted.internet import defer, reactor +from twisted.web.resource import NoResource + +import synapse +import synapse.events +from synapse.api.constants import EventTypes +from synapse.api.errors import HttpResponseException, SynapseError +from synapse.api.urls import ( + CLIENT_API_PREFIX, + FEDERATION_PREFIX, + LEGACY_MEDIA_PREFIX, + MEDIA_PREFIX, + SERVER_KEY_V2_PREFIX, +) +from synapse.app import _base +from synapse.config._base import ConfigError +from synapse.config.homeserver import HomeServerConfig +from synapse.config.logger import setup_logging +from synapse.federation import send_queue +from synapse.federation.transport.server import TransportLayerServer +from synapse.handlers.presence import PresenceHandler, get_interested_parties +from synapse.http.server import JsonResource +from synapse.http.servlet import RestServlet, parse_json_object_from_request +from synapse.http.site import SynapseSite +from synapse.logging.context import LoggingContext, run_in_background +from synapse.metrics import METRICS_PREFIX, MetricsResource, RegistryProxy +from synapse.metrics.background_process_metrics import run_as_background_process +from synapse.replication.slave.storage._base import BaseSlavedStore, __func__ +from synapse.replication.slave.storage.account_data import SlavedAccountDataStore +from synapse.replication.slave.storage.appservice import SlavedApplicationServiceStore +from synapse.replication.slave.storage.client_ips import SlavedClientIpStore +from synapse.replication.slave.storage.deviceinbox import SlavedDeviceInboxStore +from synapse.replication.slave.storage.devices import SlavedDeviceStore +from synapse.replication.slave.storage.directory import DirectoryStore +from synapse.replication.slave.storage.events import SlavedEventStore +from synapse.replication.slave.storage.filtering import SlavedFilteringStore +from synapse.replication.slave.storage.groups import SlavedGroupServerStore +from synapse.replication.slave.storage.keys import SlavedKeyStore +from synapse.replication.slave.storage.presence import SlavedPresenceStore +from synapse.replication.slave.storage.profile import SlavedProfileStore +from synapse.replication.slave.storage.push_rule import SlavedPushRuleStore +from synapse.replication.slave.storage.pushers import SlavedPusherStore +from synapse.replication.slave.storage.receipts import SlavedReceiptsStore +from synapse.replication.slave.storage.registration import SlavedRegistrationStore +from synapse.replication.slave.storage.room import RoomStore +from synapse.replication.slave.storage.transactions import SlavedTransactionStore +from synapse.replication.tcp.client import ReplicationClientHandler +from synapse.replication.tcp.streams._base import ( + DeviceListsStream, + ReceiptsStream, + ToDeviceStream, +) +from synapse.replication.tcp.streams.events import EventsStreamEventRow, EventsStreamRow +from synapse.rest.admin import register_servlets_for_media_repo +from synapse.rest.client.v1 import events +from synapse.rest.client.v1.initial_sync import InitialSyncRestServlet +from synapse.rest.client.v1.login import LoginRestServlet +from synapse.rest.client.v1.profile import ( + ProfileAvatarURLRestServlet, + ProfileDisplaynameRestServlet, + ProfileRestServlet, +) +from synapse.rest.client.v1.push_rule import PushRuleRestServlet +from synapse.rest.client.v1.room import ( + JoinedRoomMemberListRestServlet, + JoinRoomAliasServlet, + PublicRoomListRestServlet, + RoomEventContextServlet, + RoomInitialSyncRestServlet, + RoomMemberListRestServlet, + RoomMembershipRestServlet, + RoomMessageListRestServlet, + RoomSendEventRestServlet, + RoomStateEventRestServlet, + RoomStateRestServlet, +) +from synapse.rest.client.v1.voip import VoipRestServlet +from synapse.rest.client.v2_alpha import groups, sync, user_directory +from synapse.rest.client.v2_alpha._base import client_patterns +from synapse.rest.client.v2_alpha.account import ThreepidRestServlet +from synapse.rest.client.v2_alpha.keys import KeyChangesServlet, KeyQueryServlet +from synapse.rest.client.v2_alpha.register import RegisterRestServlet +from synapse.rest.client.versions import VersionsRestServlet +from synapse.rest.key.v2 import KeyApiV2Resource +from synapse.server import HomeServer +from synapse.storage.data_stores.main.media_repository import MediaRepositoryStore +from synapse.storage.data_stores.main.monthly_active_users import ( + MonthlyActiveUsersWorkerStore, +) +from synapse.storage.data_stores.main.presence import UserPresenceState +from synapse.storage.data_stores.main.user_directory import UserDirectoryStore +from synapse.types import ReadReceipt +from synapse.util.async_helpers import Linearizer +from synapse.util.httpresourcetree import create_resource_tree +from synapse.util.manhole import manhole +from synapse.util.stringutils import random_string +from synapse.util.versionstring import get_version_string + +logger = logging.getLogger("synapse.app.generic_worker") + + +class PresenceStatusStubServlet(RestServlet): + """If presence is disabled this servlet can be used to stub out setting + presence status, while proxying the getters to the master instance. + """ + + PATTERNS = client_patterns("/presence/(?P[^/]*)/status") + + def __init__(self, hs): + super(PresenceStatusStubServlet, self).__init__() + self.http_client = hs.get_simple_http_client() + self.auth = hs.get_auth() + self.main_uri = hs.config.worker_main_http_uri + + async def on_GET(self, request, user_id): + # Pass through the auth headers, if any, in case the access token + # is there. + auth_headers = request.requestHeaders.getRawHeaders("Authorization", []) + headers = {"Authorization": auth_headers} + + try: + result = await self.http_client.get_json( + self.main_uri + request.uri.decode("ascii"), headers=headers + ) + except HttpResponseException as e: + raise e.to_synapse_error() + + return 200, result + + async def on_PUT(self, request, user_id): + await self.auth.get_user_by_req(request) + return 200, {} + + +class KeyUploadServlet(RestServlet): + """An implementation of the `KeyUploadServlet` that responds to read only + requests, but otherwise proxies through to the master instance. + """ + + PATTERNS = client_patterns("/keys/upload(/(?P[^/]+))?$") + + def __init__(self, hs): + """ + Args: + hs (synapse.server.HomeServer): server + """ + super(KeyUploadServlet, self).__init__() + self.auth = hs.get_auth() + self.store = hs.get_datastore() + self.http_client = hs.get_simple_http_client() + self.main_uri = hs.config.worker_main_http_uri + + async def on_POST(self, request, device_id): + requester = await self.auth.get_user_by_req(request, allow_guest=True) + user_id = requester.user.to_string() + body = parse_json_object_from_request(request) + + if device_id is not None: + # passing the device_id here is deprecated; however, we allow it + # for now for compatibility with older clients. + if requester.device_id is not None and device_id != requester.device_id: + logger.warning( + "Client uploading keys for a different device " + "(logged in as %s, uploading for %s)", + requester.device_id, + device_id, + ) + else: + device_id = requester.device_id + + if device_id is None: + raise SynapseError( + 400, "To upload keys, you must pass device_id when authenticating" + ) + + if body: + # They're actually trying to upload something, proxy to main synapse. + # Pass through the auth headers, if any, in case the access token + # is there. + auth_headers = request.requestHeaders.getRawHeaders(b"Authorization", []) + headers = {"Authorization": auth_headers} + result = await self.http_client.post_json_get_json( + self.main_uri + request.uri.decode("ascii"), body, headers=headers + ) + + return 200, result + else: + # Just interested in counts. + result = await self.store.count_e2e_one_time_keys(user_id, device_id) + return 200, {"one_time_key_counts": result} + + +UPDATE_SYNCING_USERS_MS = 10 * 1000 + + +class GenericWorkerPresence(object): + def __init__(self, hs): + self.hs = hs + self.is_mine_id = hs.is_mine_id + self.http_client = hs.get_simple_http_client() + self.store = hs.get_datastore() + self.user_to_num_current_syncs = {} + self.clock = hs.get_clock() + self.notifier = hs.get_notifier() + + active_presence = self.store.take_presence_startup_info() + self.user_to_current_state = {state.user_id: state for state in active_presence} + + # user_id -> last_sync_ms. Lists the users that have stopped syncing + # but we haven't notified the master of that yet + self.users_going_offline = {} + + self._send_stop_syncing_loop = self.clock.looping_call( + self.send_stop_syncing, UPDATE_SYNCING_USERS_MS + ) + + self.process_id = random_string(16) + logger.info("Presence process_id is %r", self.process_id) + + def send_user_sync(self, user_id, is_syncing, last_sync_ms): + if self.hs.config.use_presence: + self.hs.get_tcp_replication().send_user_sync( + user_id, is_syncing, last_sync_ms + ) + + def mark_as_coming_online(self, user_id): + """A user has started syncing. Send a UserSync to the master, unless they + had recently stopped syncing. + + Args: + user_id (str) + """ + going_offline = self.users_going_offline.pop(user_id, None) + if not going_offline: + # Safe to skip because we haven't yet told the master they were offline + self.send_user_sync(user_id, True, self.clock.time_msec()) + + def mark_as_going_offline(self, user_id): + """A user has stopped syncing. We wait before notifying the master as + its likely they'll come back soon. This allows us to avoid sending + a stopped syncing immediately followed by a started syncing notification + to the master + + Args: + user_id (str) + """ + self.users_going_offline[user_id] = self.clock.time_msec() + + def send_stop_syncing(self): + """Check if there are any users who have stopped syncing a while ago + and haven't come back yet. If there are poke the master about them. + """ + now = self.clock.time_msec() + for user_id, last_sync_ms in list(self.users_going_offline.items()): + if now - last_sync_ms > UPDATE_SYNCING_USERS_MS: + self.users_going_offline.pop(user_id, None) + self.send_user_sync(user_id, False, last_sync_ms) + + def set_state(self, user, state, ignore_status_msg=False): + # TODO Hows this supposed to work? + return defer.succeed(None) + + get_states = __func__(PresenceHandler.get_states) + get_state = __func__(PresenceHandler.get_state) + current_state_for_users = __func__(PresenceHandler.current_state_for_users) + + def user_syncing(self, user_id, affect_presence): + if affect_presence: + curr_sync = self.user_to_num_current_syncs.get(user_id, 0) + self.user_to_num_current_syncs[user_id] = curr_sync + 1 + + # If we went from no in flight sync to some, notify replication + if self.user_to_num_current_syncs[user_id] == 1: + self.mark_as_coming_online(user_id) + + def _end(): + # We check that the user_id is in user_to_num_current_syncs because + # user_to_num_current_syncs may have been cleared if we are + # shutting down. + if affect_presence and user_id in self.user_to_num_current_syncs: + self.user_to_num_current_syncs[user_id] -= 1 + + # If we went from one in flight sync to non, notify replication + if self.user_to_num_current_syncs[user_id] == 0: + self.mark_as_going_offline(user_id) + + @contextlib.contextmanager + def _user_syncing(): + try: + yield + finally: + _end() + + return defer.succeed(_user_syncing()) + + @defer.inlineCallbacks + def notify_from_replication(self, states, stream_id): + parties = yield get_interested_parties(self.store, states) + room_ids_to_states, users_to_states = parties + + self.notifier.on_new_event( + "presence_key", + stream_id, + rooms=room_ids_to_states.keys(), + users=users_to_states.keys(), + ) + + @defer.inlineCallbacks + def process_replication_rows(self, token, rows): + states = [ + UserPresenceState( + row.user_id, + row.state, + row.last_active_ts, + row.last_federation_update_ts, + row.last_user_sync_ts, + row.status_msg, + row.currently_active, + ) + for row in rows + ] + + for state in states: + self.user_to_current_state[state.user_id] = state + + stream_id = token + yield self.notify_from_replication(states, stream_id) + + def get_currently_syncing_users(self): + if self.hs.config.use_presence: + return [ + user_id + for user_id, count in self.user_to_num_current_syncs.items() + if count > 0 + ] + else: + return set() + + +class GenericWorkerTyping(object): + def __init__(self, hs): + self._latest_room_serial = 0 + self._reset() + + def _reset(self): + """ + Reset the typing handler's data caches. + """ + # map room IDs to serial numbers + self._room_serials = {} + # map room IDs to sets of users currently typing + self._room_typing = {} + + def stream_positions(self): + # We must update this typing token from the response of the previous + # sync. In particular, the stream id may "reset" back to zero/a low + # value which we *must* use for the next replication request. + return {"typing": self._latest_room_serial} + + def process_replication_rows(self, token, rows): + if self._latest_room_serial > token: + # The master has gone backwards. To prevent inconsistent data, just + # clear everything. + self._reset() + + # Set the latest serial token to whatever the server gave us. + self._latest_room_serial = token + + for row in rows: + self._room_serials[row.room_id] = token + self._room_typing[row.room_id] = row.user_ids + + +class GenericWorkerSlavedStore( + # FIXME(#3714): We need to add UserDirectoryStore as we write directly + # rather than going via the correct worker. + UserDirectoryStore, + SlavedDeviceInboxStore, + SlavedDeviceStore, + SlavedReceiptsStore, + SlavedPushRuleStore, + SlavedGroupServerStore, + SlavedAccountDataStore, + SlavedPusherStore, + SlavedEventStore, + SlavedKeyStore, + RoomStore, + DirectoryStore, + SlavedApplicationServiceStore, + SlavedRegistrationStore, + SlavedTransactionStore, + SlavedProfileStore, + SlavedClientIpStore, + SlavedPresenceStore, + SlavedFilteringStore, + MonthlyActiveUsersWorkerStore, + MediaRepositoryStore, + BaseSlavedStore, +): + def __init__(self, database, db_conn, hs): + super(GenericWorkerSlavedStore, self).__init__(database, db_conn, hs) + + # We pull out the current federation stream position now so that we + # always have a known value for the federation position in memory so + # that we don't have to bounce via a deferred once when we start the + # replication streams. + self.federation_out_pos_startup = self._get_federation_out_pos(db_conn) + + def _get_federation_out_pos(self, db_conn): + sql = "SELECT stream_id FROM federation_stream_position WHERE type = ?" + sql = self.database_engine.convert_param_style(sql) + + txn = db_conn.cursor() + txn.execute(sql, ("federation",)) + rows = txn.fetchall() + txn.close() + + return rows[0][0] if rows else -1 + + +class GenericWorkerServer(HomeServer): + DATASTORE_CLASS = GenericWorkerSlavedStore + + def _listen_http(self, listener_config): + port = listener_config["port"] + bind_addresses = listener_config["bind_addresses"] + site_tag = listener_config.get("tag", port) + resources = {} + for res in listener_config["resources"]: + for name in res["names"]: + if name == "metrics": + resources[METRICS_PREFIX] = MetricsResource(RegistryProxy) + elif name == "client": + resource = JsonResource(self, canonical_json=False) + + PublicRoomListRestServlet(self).register(resource) + RoomMemberListRestServlet(self).register(resource) + JoinedRoomMemberListRestServlet(self).register(resource) + RoomStateRestServlet(self).register(resource) + RoomEventContextServlet(self).register(resource) + RoomMessageListRestServlet(self).register(resource) + RegisterRestServlet(self).register(resource) + LoginRestServlet(self).register(resource) + ThreepidRestServlet(self).register(resource) + KeyQueryServlet(self).register(resource) + KeyChangesServlet(self).register(resource) + VoipRestServlet(self).register(resource) + PushRuleRestServlet(self).register(resource) + VersionsRestServlet(self).register(resource) + RoomSendEventRestServlet(self).register(resource) + RoomMembershipRestServlet(self).register(resource) + RoomStateEventRestServlet(self).register(resource) + JoinRoomAliasServlet(self).register(resource) + ProfileAvatarURLRestServlet(self).register(resource) + ProfileDisplaynameRestServlet(self).register(resource) + ProfileRestServlet(self).register(resource) + KeyUploadServlet(self).register(resource) + + sync.register_servlets(self, resource) + events.register_servlets(self, resource) + InitialSyncRestServlet(self).register(resource) + RoomInitialSyncRestServlet(self).register(resource) + + user_directory.register_servlets(self, resource) + + # If presence is disabled, use the stub servlet that does + # not allow sending presence + if not self.config.use_presence: + PresenceStatusStubServlet(self).register(resource) + + groups.register_servlets(self, resource) + + resources.update({CLIENT_API_PREFIX: resource}) + elif name == "federation": + resources.update({FEDERATION_PREFIX: TransportLayerServer(self)}) + elif name == "media": + media_repo = self.get_media_repository_resource() + + # We need to serve the admin servlets for media on the + # worker. + admin_resource = JsonResource(self, canonical_json=False) + register_servlets_for_media_repo(self, admin_resource) + + resources.update( + { + MEDIA_PREFIX: media_repo, + LEGACY_MEDIA_PREFIX: media_repo, + "/_synapse/admin": admin_resource, + } + ) + + if name == "openid" and "federation" not in res["names"]: + # Only load the openid resource separately if federation resource + # is not specified since federation resource includes openid + # resource. + resources.update( + { + FEDERATION_PREFIX: TransportLayerServer( + self, servlet_groups=["openid"] + ) + } + ) + + if name in ["keys", "federation"]: + resources[SERVER_KEY_V2_PREFIX] = KeyApiV2Resource(self) + + root_resource = create_resource_tree(resources, NoResource()) + + _base.listen_tcp( + bind_addresses, + port, + SynapseSite( + "synapse.access.http.%s" % (site_tag,), + site_tag, + listener_config, + root_resource, + self.version_string, + ), + reactor=self.get_reactor(), + ) + + logger.info("Synapse worker now listening on port %d", port) + + def start_listening(self, listeners): + for listener in listeners: + if listener["type"] == "http": + self._listen_http(listener) + elif listener["type"] == "manhole": + _base.listen_tcp( + listener["bind_addresses"], + listener["port"], + manhole( + username="matrix", password="rabbithole", globals={"hs": self} + ), + ) + elif listener["type"] == "metrics": + if not self.get_config().enable_metrics: + logger.warning( + ( + "Metrics listener configured, but " + "enable_metrics is not True!" + ) + ) + else: + _base.listen_metrics(listener["bind_addresses"], listener["port"]) + else: + logger.warning("Unrecognized listener type: %s", listener["type"]) + + self.get_tcp_replication().start_replication(self) + + def remove_pusher(self, app_id, push_key, user_id): + self.get_tcp_replication().send_remove_pusher(app_id, push_key, user_id) + + def build_tcp_replication(self): + return GenericWorkerReplicationHandler(self) + + def build_presence_handler(self): + return GenericWorkerPresence(self) + + def build_typing_handler(self): + return GenericWorkerTyping(self) + + +class GenericWorkerReplicationHandler(ReplicationClientHandler): + def __init__(self, hs): + super(GenericWorkerReplicationHandler, self).__init__(hs.get_datastore()) + + self.store = hs.get_datastore() + self.typing_handler = hs.get_typing_handler() + # NB this is a SynchrotronPresence, not a normal PresenceHandler + self.presence_handler = hs.get_presence_handler() + self.notifier = hs.get_notifier() + + self.notify_pushers = hs.config.start_pushers + self.pusher_pool = hs.get_pusherpool() + + if hs.config.send_federation: + self.send_handler = FederationSenderHandler(hs, self) + else: + self.send_handler = None + + async def on_rdata(self, stream_name, token, rows): + await super(GenericWorkerReplicationHandler, self).on_rdata( + stream_name, token, rows + ) + run_in_background(self.process_and_notify, stream_name, token, rows) + + def get_streams_to_replicate(self): + args = super(GenericWorkerReplicationHandler, self).get_streams_to_replicate() + args.update(self.typing_handler.stream_positions()) + if self.send_handler: + args.update(self.send_handler.stream_positions()) + return args + + def get_currently_syncing_users(self): + return self.presence_handler.get_currently_syncing_users() + + async def process_and_notify(self, stream_name, token, rows): + try: + if self.send_handler: + self.send_handler.process_replication_rows(stream_name, token, rows) + + if stream_name == "events": + # We shouldn't get multiple rows per token for events stream, so + # we don't need to optimise this for multiple rows. + for row in rows: + if row.type != EventsStreamEventRow.TypeId: + continue + assert isinstance(row, EventsStreamRow) + + event = await self.store.get_event( + row.data.event_id, allow_rejected=True + ) + if event.rejected_reason: + continue + + extra_users = () + if event.type == EventTypes.Member: + extra_users = (event.state_key,) + max_token = self.store.get_room_max_stream_ordering() + self.notifier.on_new_room_event( + event, token, max_token, extra_users + ) + + await self.pusher_pool.on_new_notifications(token, token) + elif stream_name == "push_rules": + self.notifier.on_new_event( + "push_rules_key", token, users=[row.user_id for row in rows] + ) + elif stream_name in ("account_data", "tag_account_data"): + self.notifier.on_new_event( + "account_data_key", token, users=[row.user_id for row in rows] + ) + elif stream_name == "receipts": + self.notifier.on_new_event( + "receipt_key", token, rooms=[row.room_id for row in rows] + ) + await self.pusher_pool.on_new_receipts( + token, token, {row.room_id for row in rows} + ) + elif stream_name == "typing": + self.typing_handler.process_replication_rows(token, rows) + self.notifier.on_new_event( + "typing_key", token, rooms=[row.room_id for row in rows] + ) + elif stream_name == "to_device": + entities = [row.entity for row in rows if row.entity.startswith("@")] + if entities: + self.notifier.on_new_event("to_device_key", token, users=entities) + elif stream_name == "device_lists": + all_room_ids = set() + for row in rows: + room_ids = await self.store.get_rooms_for_user(row.user_id) + all_room_ids.update(room_ids) + self.notifier.on_new_event("device_list_key", token, rooms=all_room_ids) + elif stream_name == "presence": + await self.presence_handler.process_replication_rows(token, rows) + elif stream_name == "receipts": + self.notifier.on_new_event( + "groups_key", token, users=[row.user_id for row in rows] + ) + elif stream_name == "pushers": + for row in rows: + if row.deleted: + self.stop_pusher(row.user_id, row.app_id, row.pushkey) + else: + await self.start_pusher(row.user_id, row.app_id, row.pushkey) + except Exception: + logger.exception("Error processing replication") + + def stop_pusher(self, user_id, app_id, pushkey): + if not self.notify_pushers: + return + + key = "%s:%s" % (app_id, pushkey) + pushers_for_user = self.pusher_pool.pushers.get(user_id, {}) + pusher = pushers_for_user.pop(key, None) + if pusher is None: + return + logger.info("Stopping pusher %r / %r", user_id, key) + pusher.on_stop() + + async def start_pusher(self, user_id, app_id, pushkey): + if not self.notify_pushers: + return + + key = "%s:%s" % (app_id, pushkey) + logger.info("Starting pusher %r / %r", user_id, key) + return await self.pusher_pool.start_pusher_by_id(app_id, pushkey, user_id) + + def on_remote_server_up(self, server: str): + """Called when get a new REMOTE_SERVER_UP command.""" + + # Let's wake up the transaction queue for the server in case we have + # pending stuff to send to it. + if self.send_handler: + self.send_handler.wake_destination(server) + + +class FederationSenderHandler(object): + """Processes the replication stream and forwards the appropriate entries + to the federation sender. + """ + + def __init__(self, hs: GenericWorkerServer, replication_client): + self.store = hs.get_datastore() + self._is_mine_id = hs.is_mine_id + self.federation_sender = hs.get_federation_sender() + self.replication_client = replication_client + + self.federation_position = self.store.federation_out_pos_startup + self._fed_position_linearizer = Linearizer(name="_fed_position_linearizer") + + self._last_ack = self.federation_position + + self._room_serials = {} + self._room_typing = {} + + def on_start(self): + # There may be some events that are persisted but haven't been sent, + # so send them now. + self.federation_sender.notify_new_events( + self.store.get_room_max_stream_ordering() + ) + + def wake_destination(self, server: str): + self.federation_sender.wake_destination(server) + + def stream_positions(self): + return {"federation": self.federation_position} + + def process_replication_rows(self, stream_name, token, rows): + # The federation stream contains things that we want to send out, e.g. + # presence, typing, etc. + if stream_name == "federation": + send_queue.process_rows_for_federation(self.federation_sender, rows) + run_in_background(self.update_token, token) + + # We also need to poke the federation sender when new events happen + elif stream_name == "events": + self.federation_sender.notify_new_events(token) + + # ... and when new receipts happen + elif stream_name == ReceiptsStream.NAME: + run_as_background_process( + "process_receipts_for_federation", self._on_new_receipts, rows + ) + + # ... as well as device updates and messages + elif stream_name == DeviceListsStream.NAME: + hosts = {row.destination for row in rows} + for host in hosts: + self.federation_sender.send_device_messages(host) + + elif stream_name == ToDeviceStream.NAME: + # The to_device stream includes stuff to be pushed to both local + # clients and remote servers, so we ignore entities that start with + # '@' (since they'll be local users rather than destinations). + hosts = {row.entity for row in rows if not row.entity.startswith("@")} + for host in hosts: + self.federation_sender.send_device_messages(host) + + async def _on_new_receipts(self, rows): + """ + Args: + rows (iterable[synapse.replication.tcp.streams.ReceiptsStreamRow]): + new receipts to be processed + """ + for receipt in rows: + # we only want to send on receipts for our own users + if not self._is_mine_id(receipt.user_id): + continue + receipt_info = ReadReceipt( + receipt.room_id, + receipt.receipt_type, + receipt.user_id, + [receipt.event_id], + receipt.data, + ) + await self.federation_sender.send_read_receipt(receipt_info) + + async def update_token(self, token): + try: + self.federation_position = token + + # We linearize here to ensure we don't have races updating the token + with (await self._fed_position_linearizer.queue(None)): + if self._last_ack < self.federation_position: + await self.store.update_federation_out_pos( + "federation", self.federation_position + ) + + # We ACK this token over replication so that the master can drop + # its in memory queues + self.replication_client.send_federation_ack( + self.federation_position + ) + self._last_ack = self.federation_position + except Exception: + logger.exception("Error updating federation stream position") + + +def start(config_options): + try: + config = HomeServerConfig.load_config("Synapse worker", config_options) + except ConfigError as e: + sys.stderr.write("\n" + str(e) + "\n") + sys.exit(1) + + # For backwards compatibility let any of the old app names. + assert config.worker_app in ( + "synapse.app.appservice", + "synapse.app.client_reader", + "synapse.app.event_creator", + "synapse.app.federation_reader", + "synapse.app.federation_sender", + "synapse.app.frontend_proxy", + "synapse.app.generic_worker", + "synapse.app.media_repository", + "synapse.app.pusher", + "synapse.app.synchrotron", + "synapse.app.user_dir", + ) + + if config.worker_app == "synapse.app.appservice": + if config.notify_appservices: + sys.stderr.write( + "\nThe appservices must be disabled in the main synapse process" + "\nbefore they can be run in a separate worker." + "\nPlease add ``notify_appservices: false`` to the main config" + "\n" + ) + sys.exit(1) + + # Force the appservice to start since they will be disabled in the main config + config.notify_appservices = True + + if config.worker_app == "synapse.app.pusher": + if config.start_pushers: + sys.stderr.write( + "\nThe pushers must be disabled in the main synapse process" + "\nbefore they can be run in a separate worker." + "\nPlease add ``start_pushers: false`` to the main config" + "\n" + ) + sys.exit(1) + + # Force the pushers to start since they will be disabled in the main config + config.start_pushers = True + + if config.worker_app == "synapse.app.user_dir": + if config.update_user_directory: + sys.stderr.write( + "\nThe update_user_directory must be disabled in the main synapse process" + "\nbefore they can be run in a separate worker." + "\nPlease add ``update_user_directory: false`` to the main config" + "\n" + ) + sys.exit(1) + + # Force the pushers to start since they will be disabled in the main config + config.update_user_directory = True + + if config.worker_app == "synapse.app.federation_sender": + if config.send_federation: + sys.stderr.write( + "\nThe send_federation must be disabled in the main synapse process" + "\nbefore they can be run in a separate worker." + "\nPlease add ``send_federation: false`` to the main config" + "\n" + ) + sys.exit(1) + + # Force the pushers to start since they will be disabled in the main config + config.send_federation = True + + synapse.events.USE_FROZEN_DICTS = config.use_frozen_dicts + + ss = GenericWorkerServer( + config.server_name, + config=config, + version_string="Synapse/" + get_version_string(synapse), + ) + + setup_logging(ss, config, use_worker_options=True) + + ss.setup() + reactor.addSystemEventTrigger( + "before", "startup", _base.start, ss, config.worker_listeners + ) + + _base.start_worker_reactor("synapse-generic-worker", config) + + +if __name__ == "__main__": + with LoggingContext("main"): + start(sys.argv[1:]) diff --git a/synapse/app/media_repository.py b/synapse/app/media_repository.py index 5b5832214a..add43147b3 100644 --- a/synapse/app/media_repository.py +++ b/synapse/app/media_repository.py @@ -13,162 +13,11 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. -import logging -import sys - -from twisted.internet import reactor -from twisted.web.resource import NoResource - -import synapse -from synapse import events -from synapse.api.urls import LEGACY_MEDIA_PREFIX, MEDIA_PREFIX -from synapse.app import _base -from synapse.config._base import ConfigError -from synapse.config.homeserver import HomeServerConfig -from synapse.config.logger import setup_logging -from synapse.http.server import JsonResource -from synapse.http.site import SynapseSite -from synapse.logging.context import LoggingContext -from synapse.metrics import METRICS_PREFIX, MetricsResource, RegistryProxy -from synapse.replication.slave.storage._base import BaseSlavedStore -from synapse.replication.slave.storage.appservice import SlavedApplicationServiceStore -from synapse.replication.slave.storage.client_ips import SlavedClientIpStore -from synapse.replication.slave.storage.registration import SlavedRegistrationStore -from synapse.replication.slave.storage.room import RoomStore -from synapse.replication.slave.storage.transactions import SlavedTransactionStore -from synapse.replication.tcp.client import ReplicationClientHandler -from synapse.rest.admin import register_servlets_for_media_repo -from synapse.server import HomeServer -from synapse.storage.data_stores.main.media_repository import MediaRepositoryStore -from synapse.util.httpresourcetree import create_resource_tree -from synapse.util.manhole import manhole -from synapse.util.versionstring import get_version_string - -logger = logging.getLogger("synapse.app.media_repository") - - -class MediaRepositorySlavedStore( - RoomStore, - SlavedApplicationServiceStore, - SlavedRegistrationStore, - SlavedClientIpStore, - SlavedTransactionStore, - BaseSlavedStore, - MediaRepositoryStore, -): - pass - - -class MediaRepositoryServer(HomeServer): - DATASTORE_CLASS = MediaRepositorySlavedStore - - def _listen_http(self, listener_config): - port = listener_config["port"] - bind_addresses = listener_config["bind_addresses"] - site_tag = listener_config.get("tag", port) - resources = {} - for res in listener_config["resources"]: - for name in res["names"]: - if name == "metrics": - resources[METRICS_PREFIX] = MetricsResource(RegistryProxy) - elif name == "media": - media_repo = self.get_media_repository_resource() - - # We need to serve the admin servlets for media on the - # worker. - admin_resource = JsonResource(self, canonical_json=False) - register_servlets_for_media_repo(self, admin_resource) - - resources.update( - { - MEDIA_PREFIX: media_repo, - LEGACY_MEDIA_PREFIX: media_repo, - "/_synapse/admin": admin_resource, - } - ) - - root_resource = create_resource_tree(resources, NoResource()) - - _base.listen_tcp( - bind_addresses, - port, - SynapseSite( - "synapse.access.http.%s" % (site_tag,), - site_tag, - listener_config, - root_resource, - self.version_string, - ), - ) - logger.info("Synapse media repository now listening on port %d", port) - - def start_listening(self, listeners): - for listener in listeners: - if listener["type"] == "http": - self._listen_http(listener) - elif listener["type"] == "manhole": - _base.listen_tcp( - listener["bind_addresses"], - listener["port"], - manhole( - username="matrix", password="rabbithole", globals={"hs": self} - ), - ) - elif listener["type"] == "metrics": - if not self.get_config().enable_metrics: - logger.warning( - ( - "Metrics listener configured, but " - "enable_metrics is not True!" - ) - ) - else: - _base.listen_metrics(listener["bind_addresses"], listener["port"]) - else: - logger.warning("Unrecognized listener type: %s", listener["type"]) - - self.get_tcp_replication().start_replication(self) - - def build_tcp_replication(self): - return ReplicationClientHandler(self.get_datastore()) - - -def start(config_options): - try: - config = HomeServerConfig.load_config( - "Synapse media repository", config_options - ) - except ConfigError as e: - sys.stderr.write("\n" + str(e) + "\n") - sys.exit(1) - - assert config.worker_app == "synapse.app.media_repository" - - if config.enable_media_repo: - _base.quit_with_error( - "enable_media_repo must be disabled in the main synapse process\n" - "before the media repo can be run in a separate worker.\n" - "Please add ``enable_media_repo: false`` to the main config\n" - ) - - events.USE_FROZEN_DICTS = config.use_frozen_dicts - - ss = MediaRepositoryServer( - config.server_name, - config=config, - version_string="Synapse/" + get_version_string(synapse), - ) - - setup_logging(ss, config, use_worker_options=True) - - ss.setup() - reactor.addSystemEventTrigger( - "before", "startup", _base.start, ss, config.worker_listeners - ) - - _base.start_worker_reactor("synapse-media-repository", config) +import sys +from synapse.app.generic_worker import start +from synapse.util.logcontext import LoggingContext if __name__ == "__main__": with LoggingContext("main"): diff --git a/synapse/app/pusher.py b/synapse/app/pusher.py index 84e9f8d5e2..add43147b3 100644 --- a/synapse/app/pusher.py +++ b/synapse/app/pusher.py @@ -13,213 +13,12 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. -import logging -import sys - -from twisted.internet import defer, reactor -from twisted.web.resource import NoResource - -import synapse -from synapse import events -from synapse.app import _base -from synapse.config._base import ConfigError -from synapse.config.homeserver import HomeServerConfig -from synapse.config.logger import setup_logging -from synapse.http.site import SynapseSite -from synapse.logging.context import LoggingContext, run_in_background -from synapse.metrics import METRICS_PREFIX, MetricsResource, RegistryProxy -from synapse.replication.slave.storage._base import __func__ -from synapse.replication.slave.storage.account_data import SlavedAccountDataStore -from synapse.replication.slave.storage.events import SlavedEventStore -from synapse.replication.slave.storage.pushers import SlavedPusherStore -from synapse.replication.slave.storage.receipts import SlavedReceiptsStore -from synapse.replication.slave.storage.room import RoomStore -from synapse.replication.tcp.client import ReplicationClientHandler -from synapse.server import HomeServer -from synapse.storage import DataStore -from synapse.util.httpresourcetree import create_resource_tree -from synapse.util.manhole import manhole -from synapse.util.versionstring import get_version_string - -logger = logging.getLogger("synapse.app.pusher") - - -class PusherSlaveStore( - SlavedEventStore, - SlavedPusherStore, - SlavedReceiptsStore, - SlavedAccountDataStore, - RoomStore, -): - update_pusher_last_stream_ordering_and_success = __func__( - DataStore.update_pusher_last_stream_ordering_and_success - ) - - update_pusher_failing_since = __func__(DataStore.update_pusher_failing_since) - - update_pusher_last_stream_ordering = __func__( - DataStore.update_pusher_last_stream_ordering - ) - - get_throttle_params_by_room = __func__(DataStore.get_throttle_params_by_room) - - set_throttle_params = __func__(DataStore.set_throttle_params) - - get_time_of_last_push_action_before = __func__( - DataStore.get_time_of_last_push_action_before - ) - - get_profile_displayname = __func__(DataStore.get_profile_displayname) - - -class PusherServer(HomeServer): - DATASTORE_CLASS = PusherSlaveStore - - def remove_pusher(self, app_id, push_key, user_id): - self.get_tcp_replication().send_remove_pusher(app_id, push_key, user_id) - - def _listen_http(self, listener_config): - port = listener_config["port"] - bind_addresses = listener_config["bind_addresses"] - site_tag = listener_config.get("tag", port) - resources = {} - for res in listener_config["resources"]: - for name in res["names"]: - if name == "metrics": - resources[METRICS_PREFIX] = MetricsResource(RegistryProxy) - - root_resource = create_resource_tree(resources, NoResource()) - - _base.listen_tcp( - bind_addresses, - port, - SynapseSite( - "synapse.access.http.%s" % (site_tag,), - site_tag, - listener_config, - root_resource, - self.version_string, - ), - ) - - logger.info("Synapse pusher now listening on port %d", port) - - def start_listening(self, listeners): - for listener in listeners: - if listener["type"] == "http": - self._listen_http(listener) - elif listener["type"] == "manhole": - _base.listen_tcp( - listener["bind_addresses"], - listener["port"], - manhole( - username="matrix", password="rabbithole", globals={"hs": self} - ), - ) - elif listener["type"] == "metrics": - if not self.get_config().enable_metrics: - logger.warning( - ( - "Metrics listener configured, but " - "enable_metrics is not True!" - ) - ) - else: - _base.listen_metrics(listener["bind_addresses"], listener["port"]) - else: - logger.warning("Unrecognized listener type: %s", listener["type"]) - - self.get_tcp_replication().start_replication(self) - def build_tcp_replication(self): - return PusherReplicationHandler(self) - - -class PusherReplicationHandler(ReplicationClientHandler): - def __init__(self, hs): - super(PusherReplicationHandler, self).__init__(hs.get_datastore()) - - self.pusher_pool = hs.get_pusherpool() - - async def on_rdata(self, stream_name, token, rows): - await super(PusherReplicationHandler, self).on_rdata(stream_name, token, rows) - run_in_background(self.poke_pushers, stream_name, token, rows) - - @defer.inlineCallbacks - def poke_pushers(self, stream_name, token, rows): - try: - if stream_name == "pushers": - for row in rows: - if row.deleted: - yield self.stop_pusher(row.user_id, row.app_id, row.pushkey) - else: - yield self.start_pusher(row.user_id, row.app_id, row.pushkey) - elif stream_name == "events": - yield self.pusher_pool.on_new_notifications(token, token) - elif stream_name == "receipts": - yield self.pusher_pool.on_new_receipts( - token, token, {row.room_id for row in rows} - ) - except Exception: - logger.exception("Error poking pushers") - - def stop_pusher(self, user_id, app_id, pushkey): - key = "%s:%s" % (app_id, pushkey) - pushers_for_user = self.pusher_pool.pushers.get(user_id, {}) - pusher = pushers_for_user.pop(key, None) - if pusher is None: - return - logger.info("Stopping pusher %r / %r", user_id, key) - pusher.on_stop() - - def start_pusher(self, user_id, app_id, pushkey): - key = "%s:%s" % (app_id, pushkey) - logger.info("Starting pusher %r / %r", user_id, key) - return self.pusher_pool.start_pusher_by_id(app_id, pushkey, user_id) - - -def start(config_options): - try: - config = HomeServerConfig.load_config("Synapse pusher", config_options) - except ConfigError as e: - sys.stderr.write("\n" + str(e) + "\n") - sys.exit(1) - - assert config.worker_app == "synapse.app.pusher" - - events.USE_FROZEN_DICTS = config.use_frozen_dicts - - if config.start_pushers: - sys.stderr.write( - "\nThe pushers must be disabled in the main synapse process" - "\nbefore they can be run in a separate worker." - "\nPlease add ``start_pushers: false`` to the main config" - "\n" - ) - sys.exit(1) - - # Force the pushers to start since they will be disabled in the main config - config.start_pushers = True - - ps = PusherServer( - config.server_name, - config=config, - version_string="Synapse/" + get_version_string(synapse), - ) - - setup_logging(ps, config, use_worker_options=True) - - ps.setup() - - def start(): - _base.start(ps, config.worker_listeners) - ps.get_pusherpool().start() - - reactor.addSystemEventTrigger("before", "startup", start) - - _base.start_worker_reactor("synapse-pusher", config) +import sys +from synapse.app.generic_worker import start +from synapse.util.logcontext import LoggingContext if __name__ == "__main__": with LoggingContext("main"): - ps = start(sys.argv[1:]) + start(sys.argv[1:]) diff --git a/synapse/app/synchrotron.py b/synapse/app/synchrotron.py index 8982c0676e..add43147b3 100644 --- a/synapse/app/synchrotron.py +++ b/synapse/app/synchrotron.py @@ -13,454 +13,11 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. -import contextlib -import logging -import sys - -from six import iteritems - -from twisted.internet import defer, reactor -from twisted.web.resource import NoResource - -import synapse -from synapse.api.constants import EventTypes -from synapse.app import _base -from synapse.config._base import ConfigError -from synapse.config.homeserver import HomeServerConfig -from synapse.config.logger import setup_logging -from synapse.handlers.presence import PresenceHandler, get_interested_parties -from synapse.http.server import JsonResource -from synapse.http.site import SynapseSite -from synapse.logging.context import LoggingContext, run_in_background -from synapse.metrics import METRICS_PREFIX, MetricsResource, RegistryProxy -from synapse.replication.slave.storage._base import BaseSlavedStore, __func__ -from synapse.replication.slave.storage.account_data import SlavedAccountDataStore -from synapse.replication.slave.storage.appservice import SlavedApplicationServiceStore -from synapse.replication.slave.storage.client_ips import SlavedClientIpStore -from synapse.replication.slave.storage.deviceinbox import SlavedDeviceInboxStore -from synapse.replication.slave.storage.devices import SlavedDeviceStore -from synapse.replication.slave.storage.events import SlavedEventStore -from synapse.replication.slave.storage.filtering import SlavedFilteringStore -from synapse.replication.slave.storage.groups import SlavedGroupServerStore -from synapse.replication.slave.storage.presence import SlavedPresenceStore -from synapse.replication.slave.storage.push_rule import SlavedPushRuleStore -from synapse.replication.slave.storage.receipts import SlavedReceiptsStore -from synapse.replication.slave.storage.registration import SlavedRegistrationStore -from synapse.replication.slave.storage.room import RoomStore -from synapse.replication.tcp.client import ReplicationClientHandler -from synapse.replication.tcp.streams.events import EventsStreamEventRow, EventsStreamRow -from synapse.rest.client.v1 import events -from synapse.rest.client.v1.initial_sync import InitialSyncRestServlet -from synapse.rest.client.v1.room import RoomInitialSyncRestServlet -from synapse.rest.client.v2_alpha import sync -from synapse.server import HomeServer -from synapse.storage.data_stores.main.monthly_active_users import ( - MonthlyActiveUsersWorkerStore, -) -from synapse.storage.data_stores.main.presence import UserPresenceState -from synapse.util.httpresourcetree import create_resource_tree -from synapse.util.manhole import manhole -from synapse.util.stringutils import random_string -from synapse.util.versionstring import get_version_string - -logger = logging.getLogger("synapse.app.synchrotron") - - -class SynchrotronSlavedStore( - SlavedReceiptsStore, - SlavedAccountDataStore, - SlavedApplicationServiceStore, - SlavedRegistrationStore, - SlavedFilteringStore, - SlavedPresenceStore, - SlavedGroupServerStore, - SlavedDeviceInboxStore, - SlavedDeviceStore, - SlavedPushRuleStore, - SlavedEventStore, - SlavedClientIpStore, - RoomStore, - MonthlyActiveUsersWorkerStore, - BaseSlavedStore, -): - pass - - -UPDATE_SYNCING_USERS_MS = 10 * 1000 - - -class SynchrotronPresence(object): - def __init__(self, hs): - self.hs = hs - self.is_mine_id = hs.is_mine_id - self.http_client = hs.get_simple_http_client() - self.store = hs.get_datastore() - self.user_to_num_current_syncs = {} - self.clock = hs.get_clock() - self.notifier = hs.get_notifier() - - active_presence = self.store.take_presence_startup_info() - self.user_to_current_state = {state.user_id: state for state in active_presence} - - # user_id -> last_sync_ms. Lists the users that have stopped syncing - # but we haven't notified the master of that yet - self.users_going_offline = {} - - self._send_stop_syncing_loop = self.clock.looping_call( - self.send_stop_syncing, 10 * 1000 - ) - - self.process_id = random_string(16) - logger.info("Presence process_id is %r", self.process_id) - - def send_user_sync(self, user_id, is_syncing, last_sync_ms): - if self.hs.config.use_presence: - self.hs.get_tcp_replication().send_user_sync( - user_id, is_syncing, last_sync_ms - ) - - def mark_as_coming_online(self, user_id): - """A user has started syncing. Send a UserSync to the master, unless they - had recently stopped syncing. - - Args: - user_id (str) - """ - going_offline = self.users_going_offline.pop(user_id, None) - if not going_offline: - # Safe to skip because we haven't yet told the master they were offline - self.send_user_sync(user_id, True, self.clock.time_msec()) - - def mark_as_going_offline(self, user_id): - """A user has stopped syncing. We wait before notifying the master as - its likely they'll come back soon. This allows us to avoid sending - a stopped syncing immediately followed by a started syncing notification - to the master - - Args: - user_id (str) - """ - self.users_going_offline[user_id] = self.clock.time_msec() - - def send_stop_syncing(self): - """Check if there are any users who have stopped syncing a while ago - and haven't come back yet. If there are poke the master about them. - """ - now = self.clock.time_msec() - for user_id, last_sync_ms in list(self.users_going_offline.items()): - if now - last_sync_ms > 10 * 1000: - self.users_going_offline.pop(user_id, None) - self.send_user_sync(user_id, False, last_sync_ms) - - def set_state(self, user, state, ignore_status_msg=False): - # TODO Hows this supposed to work? - return defer.succeed(None) - - get_states = __func__(PresenceHandler.get_states) - get_state = __func__(PresenceHandler.get_state) - current_state_for_users = __func__(PresenceHandler.current_state_for_users) - - def user_syncing(self, user_id, affect_presence): - if affect_presence: - curr_sync = self.user_to_num_current_syncs.get(user_id, 0) - self.user_to_num_current_syncs[user_id] = curr_sync + 1 - - # If we went from no in flight sync to some, notify replication - if self.user_to_num_current_syncs[user_id] == 1: - self.mark_as_coming_online(user_id) - - def _end(): - # We check that the user_id is in user_to_num_current_syncs because - # user_to_num_current_syncs may have been cleared if we are - # shutting down. - if affect_presence and user_id in self.user_to_num_current_syncs: - self.user_to_num_current_syncs[user_id] -= 1 - - # If we went from one in flight sync to non, notify replication - if self.user_to_num_current_syncs[user_id] == 0: - self.mark_as_going_offline(user_id) - - @contextlib.contextmanager - def _user_syncing(): - try: - yield - finally: - _end() - - return defer.succeed(_user_syncing()) - - @defer.inlineCallbacks - def notify_from_replication(self, states, stream_id): - parties = yield get_interested_parties(self.store, states) - room_ids_to_states, users_to_states = parties - - self.notifier.on_new_event( - "presence_key", - stream_id, - rooms=room_ids_to_states.keys(), - users=users_to_states.keys(), - ) - - @defer.inlineCallbacks - def process_replication_rows(self, token, rows): - states = [ - UserPresenceState( - row.user_id, - row.state, - row.last_active_ts, - row.last_federation_update_ts, - row.last_user_sync_ts, - row.status_msg, - row.currently_active, - ) - for row in rows - ] - - for state in states: - self.user_to_current_state[state.user_id] = state - - stream_id = token - yield self.notify_from_replication(states, stream_id) - - def get_currently_syncing_users(self): - if self.hs.config.use_presence: - return [ - user_id - for user_id, count in iteritems(self.user_to_num_current_syncs) - if count > 0 - ] - else: - return set() - -class SynchrotronTyping(object): - def __init__(self, hs): - self._latest_room_serial = 0 - self._reset() - - def _reset(self): - """ - Reset the typing handler's data caches. - """ - # map room IDs to serial numbers - self._room_serials = {} - # map room IDs to sets of users currently typing - self._room_typing = {} - - def stream_positions(self): - # We must update this typing token from the response of the previous - # sync. In particular, the stream id may "reset" back to zero/a low - # value which we *must* use for the next replication request. - return {"typing": self._latest_room_serial} - - def process_replication_rows(self, token, rows): - if self._latest_room_serial > token: - # The master has gone backwards. To prevent inconsistent data, just - # clear everything. - self._reset() - - # Set the latest serial token to whatever the server gave us. - self._latest_room_serial = token - - for row in rows: - self._room_serials[row.room_id] = token - self._room_typing[row.room_id] = row.user_ids - - -class SynchrotronApplicationService(object): - def notify_interested_services(self, event): - pass - - -class SynchrotronServer(HomeServer): - DATASTORE_CLASS = SynchrotronSlavedStore - - def _listen_http(self, listener_config): - port = listener_config["port"] - bind_addresses = listener_config["bind_addresses"] - site_tag = listener_config.get("tag", port) - resources = {} - for res in listener_config["resources"]: - for name in res["names"]: - if name == "metrics": - resources[METRICS_PREFIX] = MetricsResource(RegistryProxy) - elif name == "client": - resource = JsonResource(self, canonical_json=False) - sync.register_servlets(self, resource) - events.register_servlets(self, resource) - InitialSyncRestServlet(self).register(resource) - RoomInitialSyncRestServlet(self).register(resource) - resources.update( - { - "/_matrix/client/r0": resource, - "/_matrix/client/unstable": resource, - "/_matrix/client/v2_alpha": resource, - "/_matrix/client/api/v1": resource, - } - ) - - root_resource = create_resource_tree(resources, NoResource()) - - _base.listen_tcp( - bind_addresses, - port, - SynapseSite( - "synapse.access.http.%s" % (site_tag,), - site_tag, - listener_config, - root_resource, - self.version_string, - ), - ) - - logger.info("Synapse synchrotron now listening on port %d", port) - - def start_listening(self, listeners): - for listener in listeners: - if listener["type"] == "http": - self._listen_http(listener) - elif listener["type"] == "manhole": - _base.listen_tcp( - listener["bind_addresses"], - listener["port"], - manhole( - username="matrix", password="rabbithole", globals={"hs": self} - ), - ) - elif listener["type"] == "metrics": - if not self.get_config().enable_metrics: - logger.warning( - ( - "Metrics listener configured, but " - "enable_metrics is not True!" - ) - ) - else: - _base.listen_metrics(listener["bind_addresses"], listener["port"]) - else: - logger.warning("Unrecognized listener type: %s", listener["type"]) - - self.get_tcp_replication().start_replication(self) - - def build_tcp_replication(self): - return SyncReplicationHandler(self) - - def build_presence_handler(self): - return SynchrotronPresence(self) - - def build_typing_handler(self): - return SynchrotronTyping(self) - - -class SyncReplicationHandler(ReplicationClientHandler): - def __init__(self, hs): - super(SyncReplicationHandler, self).__init__(hs.get_datastore()) - - self.store = hs.get_datastore() - self.typing_handler = hs.get_typing_handler() - # NB this is a SynchrotronPresence, not a normal PresenceHandler - self.presence_handler = hs.get_presence_handler() - self.notifier = hs.get_notifier() - - async def on_rdata(self, stream_name, token, rows): - await super(SyncReplicationHandler, self).on_rdata(stream_name, token, rows) - run_in_background(self.process_and_notify, stream_name, token, rows) - - def get_streams_to_replicate(self): - args = super(SyncReplicationHandler, self).get_streams_to_replicate() - args.update(self.typing_handler.stream_positions()) - return args - - def get_currently_syncing_users(self): - return self.presence_handler.get_currently_syncing_users() - - async def process_and_notify(self, stream_name, token, rows): - try: - if stream_name == "events": - # We shouldn't get multiple rows per token for events stream, so - # we don't need to optimise this for multiple rows. - for row in rows: - if row.type != EventsStreamEventRow.TypeId: - continue - assert isinstance(row, EventsStreamRow) - - event = await self.store.get_event( - row.data.event_id, allow_rejected=True - ) - if event.rejected_reason: - continue - - extra_users = () - if event.type == EventTypes.Member: - extra_users = (event.state_key,) - max_token = self.store.get_room_max_stream_ordering() - self.notifier.on_new_room_event( - event, token, max_token, extra_users - ) - elif stream_name == "push_rules": - self.notifier.on_new_event( - "push_rules_key", token, users=[row.user_id for row in rows] - ) - elif stream_name in ("account_data", "tag_account_data"): - self.notifier.on_new_event( - "account_data_key", token, users=[row.user_id for row in rows] - ) - elif stream_name == "receipts": - self.notifier.on_new_event( - "receipt_key", token, rooms=[row.room_id for row in rows] - ) - elif stream_name == "typing": - self.typing_handler.process_replication_rows(token, rows) - self.notifier.on_new_event( - "typing_key", token, rooms=[row.room_id for row in rows] - ) - elif stream_name == "to_device": - entities = [row.entity for row in rows if row.entity.startswith("@")] - if entities: - self.notifier.on_new_event("to_device_key", token, users=entities) - elif stream_name == "device_lists": - all_room_ids = set() - for row in rows: - room_ids = await self.store.get_rooms_for_user(row.user_id) - all_room_ids.update(room_ids) - self.notifier.on_new_event("device_list_key", token, rooms=all_room_ids) - elif stream_name == "presence": - await self.presence_handler.process_replication_rows(token, rows) - elif stream_name == "receipts": - self.notifier.on_new_event( - "groups_key", token, users=[row.user_id for row in rows] - ) - except Exception: - logger.exception("Error processing replication") - - -def start(config_options): - try: - config = HomeServerConfig.load_config("Synapse synchrotron", config_options) - except ConfigError as e: - sys.stderr.write("\n" + str(e) + "\n") - sys.exit(1) - - assert config.worker_app == "synapse.app.synchrotron" - - synapse.events.USE_FROZEN_DICTS = config.use_frozen_dicts - - ss = SynchrotronServer( - config.server_name, - config=config, - version_string="Synapse/" + get_version_string(synapse), - application_service_handler=SynchrotronApplicationService(), - ) - - setup_logging(ss, config, use_worker_options=True) - - ss.setup() - reactor.addSystemEventTrigger( - "before", "startup", _base.start, ss, config.worker_listeners - ) - - _base.start_worker_reactor("synapse-synchrotron", config) +import sys +from synapse.app.generic_worker import start +from synapse.util.logcontext import LoggingContext if __name__ == "__main__": with LoggingContext("main"): diff --git a/synapse/app/user_dir.py b/synapse/app/user_dir.py index ba536d6f04..503d44f687 100644 --- a/synapse/app/user_dir.py +++ b/synapse/app/user_dir.py @@ -14,217 +14,10 @@ # See the License for the specific language governing permissions and # limitations under the License. -import logging import sys -from twisted.internet import defer, reactor -from twisted.web.resource import NoResource - -import synapse -from synapse import events -from synapse.app import _base -from synapse.config._base import ConfigError -from synapse.config.homeserver import HomeServerConfig -from synapse.config.logger import setup_logging -from synapse.http.server import JsonResource -from synapse.http.site import SynapseSite -from synapse.logging.context import LoggingContext, run_in_background -from synapse.metrics import METRICS_PREFIX, MetricsResource, RegistryProxy -from synapse.replication.slave.storage._base import BaseSlavedStore -from synapse.replication.slave.storage.appservice import SlavedApplicationServiceStore -from synapse.replication.slave.storage.client_ips import SlavedClientIpStore -from synapse.replication.slave.storage.events import SlavedEventStore -from synapse.replication.slave.storage.registration import SlavedRegistrationStore -from synapse.replication.tcp.client import ReplicationClientHandler -from synapse.replication.tcp.streams.events import ( - EventsStream, - EventsStreamCurrentStateRow, -) -from synapse.rest.client.v2_alpha import user_directory -from synapse.server import HomeServer -from synapse.storage.data_stores.main.user_directory import UserDirectoryStore -from synapse.storage.database import Database -from synapse.util.caches.stream_change_cache import StreamChangeCache -from synapse.util.httpresourcetree import create_resource_tree -from synapse.util.manhole import manhole -from synapse.util.versionstring import get_version_string - -logger = logging.getLogger("synapse.app.user_dir") - - -class UserDirectorySlaveStore( - SlavedEventStore, - SlavedApplicationServiceStore, - SlavedRegistrationStore, - SlavedClientIpStore, - UserDirectoryStore, - BaseSlavedStore, -): - def __init__(self, database: Database, db_conn, hs): - super(UserDirectorySlaveStore, self).__init__(database, db_conn, hs) - - events_max = self._stream_id_gen.get_current_token() - curr_state_delta_prefill, min_curr_state_delta_id = self.db.get_cache_dict( - db_conn, - "current_state_delta_stream", - entity_column="room_id", - stream_column="stream_id", - max_value=events_max, # As we share the stream id with events token - limit=1000, - ) - self._curr_state_delta_stream_cache = StreamChangeCache( - "_curr_state_delta_stream_cache", - min_curr_state_delta_id, - prefilled_cache=curr_state_delta_prefill, - ) - - def stream_positions(self): - result = super(UserDirectorySlaveStore, self).stream_positions() - return result - - def process_replication_rows(self, stream_name, token, rows): - if stream_name == EventsStream.NAME: - self._stream_id_gen.advance(token) - for row in rows: - if row.type != EventsStreamCurrentStateRow.TypeId: - continue - self._curr_state_delta_stream_cache.entity_has_changed( - row.data.room_id, token - ) - return super(UserDirectorySlaveStore, self).process_replication_rows( - stream_name, token, rows - ) - - -class UserDirectoryServer(HomeServer): - DATASTORE_CLASS = UserDirectorySlaveStore - - def _listen_http(self, listener_config): - port = listener_config["port"] - bind_addresses = listener_config["bind_addresses"] - site_tag = listener_config.get("tag", port) - resources = {} - for res in listener_config["resources"]: - for name in res["names"]: - if name == "metrics": - resources[METRICS_PREFIX] = MetricsResource(RegistryProxy) - elif name == "client": - resource = JsonResource(self, canonical_json=False) - user_directory.register_servlets(self, resource) - resources.update( - { - "/_matrix/client/r0": resource, - "/_matrix/client/unstable": resource, - "/_matrix/client/v2_alpha": resource, - "/_matrix/client/api/v1": resource, - } - ) - - root_resource = create_resource_tree(resources, NoResource()) - - _base.listen_tcp( - bind_addresses, - port, - SynapseSite( - "synapse.access.http.%s" % (site_tag,), - site_tag, - listener_config, - root_resource, - self.version_string, - ), - ) - - logger.info("Synapse user_dir now listening on port %d", port) - - def start_listening(self, listeners): - for listener in listeners: - if listener["type"] == "http": - self._listen_http(listener) - elif listener["type"] == "manhole": - _base.listen_tcp( - listener["bind_addresses"], - listener["port"], - manhole( - username="matrix", password="rabbithole", globals={"hs": self} - ), - ) - elif listener["type"] == "metrics": - if not self.get_config().enable_metrics: - logger.warning( - ( - "Metrics listener configured, but " - "enable_metrics is not True!" - ) - ) - else: - _base.listen_metrics(listener["bind_addresses"], listener["port"]) - else: - logger.warning("Unrecognized listener type: %s", listener["type"]) - - self.get_tcp_replication().start_replication(self) - - def build_tcp_replication(self): - return UserDirectoryReplicationHandler(self) - - -class UserDirectoryReplicationHandler(ReplicationClientHandler): - def __init__(self, hs): - super(UserDirectoryReplicationHandler, self).__init__(hs.get_datastore()) - self.user_directory = hs.get_user_directory_handler() - - async def on_rdata(self, stream_name, token, rows): - await super(UserDirectoryReplicationHandler, self).on_rdata( - stream_name, token, rows - ) - if stream_name == EventsStream.NAME: - run_in_background(self._notify_directory) - - @defer.inlineCallbacks - def _notify_directory(self): - try: - yield self.user_directory.notify_new_event() - except Exception: - logger.exception("Error notifiying user directory of state update") - - -def start(config_options): - try: - config = HomeServerConfig.load_config("Synapse user directory", config_options) - except ConfigError as e: - sys.stderr.write("\n" + str(e) + "\n") - sys.exit(1) - - assert config.worker_app == "synapse.app.user_dir" - - events.USE_FROZEN_DICTS = config.use_frozen_dicts - - if config.update_user_directory: - sys.stderr.write( - "\nThe update_user_directory must be disabled in the main synapse process" - "\nbefore they can be run in a separate worker." - "\nPlease add ``update_user_directory: false`` to the main config" - "\n" - ) - sys.exit(1) - - # Force the pushers to start since they will be disabled in the main config - config.update_user_directory = True - - ss = UserDirectoryServer( - config.server_name, - config=config, - version_string="Synapse/" + get_version_string(synapse), - ) - - setup_logging(ss, config, use_worker_options=True) - - ss.setup() - reactor.addSystemEventTrigger( - "before", "startup", _base.start, ss, config.worker_listeners - ) - - _base.start_worker_reactor("synapse-user-dir", config) - +from synapse.app.generic_worker import start +from synapse.util.logcontext import LoggingContext if __name__ == "__main__": with LoggingContext("main"): diff --git a/synapse/replication/slave/storage/events.py b/synapse/replication/slave/storage/events.py index 3aa6cb8b96..e73342c657 100644 --- a/synapse/replication/slave/storage/events.py +++ b/synapse/replication/slave/storage/events.py @@ -32,6 +32,7 @@ from synapse.storage.data_stores.main.state import StateGroupWorkerStore from synapse.storage.data_stores.main.stream import StreamWorkerStore from synapse.storage.data_stores.main.user_erasure_store import UserErasureWorkerStore from synapse.storage.database import Database +from synapse.util.caches.stream_change_cache import StreamChangeCache from ._base import BaseSlavedStore from ._slaved_id_tracker import SlavedIdTracker @@ -68,6 +69,21 @@ class SlavedEventStore( super(SlavedEventStore, self).__init__(database, db_conn, hs) + events_max = self._stream_id_gen.get_current_token() + curr_state_delta_prefill, min_curr_state_delta_id = self.db.get_cache_dict( + db_conn, + "current_state_delta_stream", + entity_column="room_id", + stream_column="stream_id", + max_value=events_max, # As we share the stream id with events token + limit=1000, + ) + self._curr_state_delta_stream_cache = StreamChangeCache( + "_curr_state_delta_stream_cache", + min_curr_state_delta_id, + prefilled_cache=curr_state_delta_prefill, + ) + # Cached functions can't be accessed through a class instance so we need # to reach inside the __dict__ to extract them. @@ -120,6 +136,10 @@ class SlavedEventStore( backfilled=False, ) elif row.type == EventsStreamCurrentStateRow.TypeId: + self._curr_state_delta_stream_cache.entity_has_changed( + row.data.room_id, token + ) + if data.type == EventTypes.Member: self.get_rooms_for_user_with_stream_ordering.invalidate( (data.state_key,) diff --git a/synapse/storage/data_stores/main/pusher.py b/synapse/storage/data_stores/main/pusher.py index 6b03233262..547b9d69cb 100644 --- a/synapse/storage/data_stores/main/pusher.py +++ b/synapse/storage/data_stores/main/pusher.py @@ -197,6 +197,84 @@ class PusherWorkerStore(SQLBaseStore): return result + @defer.inlineCallbacks + def update_pusher_last_stream_ordering( + self, app_id, pushkey, user_id, last_stream_ordering + ): + yield self.db.simple_update_one( + "pushers", + {"app_id": app_id, "pushkey": pushkey, "user_name": user_id}, + {"last_stream_ordering": last_stream_ordering}, + desc="update_pusher_last_stream_ordering", + ) + + @defer.inlineCallbacks + def update_pusher_last_stream_ordering_and_success( + self, app_id, pushkey, user_id, last_stream_ordering, last_success + ): + """Update the last stream ordering position we've processed up to for + the given pusher. + + Args: + app_id (str) + pushkey (str) + last_stream_ordering (int) + last_success (int) + + Returns: + Deferred[bool]: True if the pusher still exists; False if it has been deleted. + """ + updated = yield self.db.simple_update( + table="pushers", + keyvalues={"app_id": app_id, "pushkey": pushkey, "user_name": user_id}, + updatevalues={ + "last_stream_ordering": last_stream_ordering, + "last_success": last_success, + }, + desc="update_pusher_last_stream_ordering_and_success", + ) + + return bool(updated) + + @defer.inlineCallbacks + def update_pusher_failing_since(self, app_id, pushkey, user_id, failing_since): + yield self.db.simple_update( + table="pushers", + keyvalues={"app_id": app_id, "pushkey": pushkey, "user_name": user_id}, + updatevalues={"failing_since": failing_since}, + desc="update_pusher_failing_since", + ) + + @defer.inlineCallbacks + def get_throttle_params_by_room(self, pusher_id): + res = yield self.db.simple_select_list( + "pusher_throttle", + {"pusher": pusher_id}, + ["room_id", "last_sent_ts", "throttle_ms"], + desc="get_throttle_params_by_room", + ) + + params_by_room = {} + for row in res: + params_by_room[row["room_id"]] = { + "last_sent_ts": row["last_sent_ts"], + "throttle_ms": row["throttle_ms"], + } + + return params_by_room + + @defer.inlineCallbacks + def set_throttle_params(self, pusher_id, room_id, params): + # no need to lock because `pusher_throttle` has a primary key on + # (pusher, room_id) so simple_upsert will retry + yield self.db.simple_upsert( + "pusher_throttle", + {"pusher": pusher_id, "room_id": room_id}, + params, + desc="set_throttle_params", + lock=False, + ) + class PusherStore(PusherWorkerStore): def get_pushers_stream_token(self): @@ -282,81 +360,3 @@ class PusherStore(PusherWorkerStore): with self._pushers_id_gen.get_next() as stream_id: yield self.db.runInteraction("delete_pusher", delete_pusher_txn, stream_id) - - @defer.inlineCallbacks - def update_pusher_last_stream_ordering( - self, app_id, pushkey, user_id, last_stream_ordering - ): - yield self.db.simple_update_one( - "pushers", - {"app_id": app_id, "pushkey": pushkey, "user_name": user_id}, - {"last_stream_ordering": last_stream_ordering}, - desc="update_pusher_last_stream_ordering", - ) - - @defer.inlineCallbacks - def update_pusher_last_stream_ordering_and_success( - self, app_id, pushkey, user_id, last_stream_ordering, last_success - ): - """Update the last stream ordering position we've processed up to for - the given pusher. - - Args: - app_id (str) - pushkey (str) - last_stream_ordering (int) - last_success (int) - - Returns: - Deferred[bool]: True if the pusher still exists; False if it has been deleted. - """ - updated = yield self.db.simple_update( - table="pushers", - keyvalues={"app_id": app_id, "pushkey": pushkey, "user_name": user_id}, - updatevalues={ - "last_stream_ordering": last_stream_ordering, - "last_success": last_success, - }, - desc="update_pusher_last_stream_ordering_and_success", - ) - - return bool(updated) - - @defer.inlineCallbacks - def update_pusher_failing_since(self, app_id, pushkey, user_id, failing_since): - yield self.db.simple_update( - table="pushers", - keyvalues={"app_id": app_id, "pushkey": pushkey, "user_name": user_id}, - updatevalues={"failing_since": failing_since}, - desc="update_pusher_failing_since", - ) - - @defer.inlineCallbacks - def get_throttle_params_by_room(self, pusher_id): - res = yield self.db.simple_select_list( - "pusher_throttle", - {"pusher": pusher_id}, - ["room_id", "last_sent_ts", "throttle_ms"], - desc="get_throttle_params_by_room", - ) - - params_by_room = {} - for row in res: - params_by_room[row["room_id"]] = { - "last_sent_ts": row["last_sent_ts"], - "throttle_ms": row["throttle_ms"], - } - - return params_by_room - - @defer.inlineCallbacks - def set_throttle_params(self, pusher_id, room_id, params): - # no need to lock because `pusher_throttle` has a primary key on - # (pusher, room_id) so simple_upsert will retry - yield self.db.simple_upsert( - "pusher_throttle", - {"pusher": pusher_id, "room_id": room_id}, - params, - desc="set_throttle_params", - lock=False, - ) diff --git a/tests/app/test_frontend_proxy.py b/tests/app/test_frontend_proxy.py index 8bdbc608a9..160e55aca9 100644 --- a/tests/app/test_frontend_proxy.py +++ b/tests/app/test_frontend_proxy.py @@ -13,7 +13,7 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.app.frontend_proxy import FrontendProxyServer +from synapse.app.generic_worker import GenericWorkerServer from tests.unittest import HomeserverTestCase @@ -22,7 +22,7 @@ class FrontendProxyTests(HomeserverTestCase): def make_homeserver(self, reactor, clock): hs = self.setup_test_homeserver( - http_client=None, homeserverToUse=FrontendProxyServer + http_client=None, homeserverToUse=GenericWorkerServer ) return hs @@ -46,9 +46,7 @@ class FrontendProxyTests(HomeserverTestCase): # Grab the resource from the site that was told to listen self.assertEqual(len(self.reactor.tcpServers), 1) site = self.reactor.tcpServers[0][1] - self.resource = ( - site.resource.children[b"_matrix"].children[b"client"].children[b"r0"] - ) + self.resource = site.resource.children[b"_matrix"].children[b"client"] request, channel = self.make_request("PUT", "presence/a/status") self.render(request) @@ -76,9 +74,7 @@ class FrontendProxyTests(HomeserverTestCase): # Grab the resource from the site that was told to listen self.assertEqual(len(self.reactor.tcpServers), 1) site = self.reactor.tcpServers[0][1] - self.resource = ( - site.resource.children[b"_matrix"].children[b"client"].children[b"r0"] - ) + self.resource = site.resource.children[b"_matrix"].children[b"client"] request, channel = self.make_request("PUT", "presence/a/status") self.render(request) diff --git a/tests/app/test_openid_listener.py b/tests/app/test_openid_listener.py index 48792d1480..1fe048048b 100644 --- a/tests/app/test_openid_listener.py +++ b/tests/app/test_openid_listener.py @@ -16,7 +16,7 @@ from mock import Mock, patch from parameterized import parameterized -from synapse.app.federation_reader import FederationReaderServer +from synapse.app.generic_worker import GenericWorkerServer from synapse.app.homeserver import SynapseHomeServer from tests.unittest import HomeserverTestCase @@ -25,7 +25,7 @@ from tests.unittest import HomeserverTestCase class FederationReaderOpenIDListenerTests(HomeserverTestCase): def make_homeserver(self, reactor, clock): hs = self.setup_test_homeserver( - http_client=None, homeserverToUse=FederationReaderServer + http_client=None, homeserverToUse=GenericWorkerServer ) return hs -- cgit 1.4.1 From 8c75b621bfe03725cc8da071516ebc66d3872760 Mon Sep 17 00:00:00 2001 From: Andrew Morgan <1342360+anoadragon453@users.noreply.github.com> Date: Wed, 26 Feb 2020 12:22:55 +0000 Subject: Ensure 'deactivated' parameter is a boolean on user admin API, Fix error handling of call to deactivate user (#6990) --- changelog.d/6990.bugfix | 1 + synapse/rest/admin/users.py | 11 +++++--- synapse/rest/client/v1/login.py | 1 + tests/rest/admin/test_user.py | 59 +++++++++++++++++++++++++++++++++++++++++ 4 files changed, 68 insertions(+), 4 deletions(-) create mode 100644 changelog.d/6990.bugfix (limited to 'tests') diff --git a/changelog.d/6990.bugfix b/changelog.d/6990.bugfix new file mode 100644 index 0000000000..8c1c48f4d4 --- /dev/null +++ b/changelog.d/6990.bugfix @@ -0,0 +1 @@ +Prevent user from setting 'deactivated' to anything other than a bool on the v2 PUT /users Admin API. \ No newline at end of file diff --git a/synapse/rest/admin/users.py b/synapse/rest/admin/users.py index 2107b5dc56..c5b461a236 100644 --- a/synapse/rest/admin/users.py +++ b/synapse/rest/admin/users.py @@ -228,13 +228,16 @@ class UserRestServletV2(RestServlet): ) if "deactivated" in body: - deactivate = bool(body["deactivated"]) + deactivate = body["deactivated"] + if not isinstance(deactivate, bool): + raise SynapseError( + 400, "'deactivated' parameter is not of type boolean" + ) + if deactivate and not user["deactivated"]: - result = await self.deactivate_account_handler.deactivate_account( + await self.deactivate_account_handler.deactivate_account( target_user.to_string(), False ) - if not result: - raise SynapseError(500, "Could not deactivate user") user = await self.admin_handler.get_user(target_user) return 200, user diff --git a/synapse/rest/client/v1/login.py b/synapse/rest/client/v1/login.py index 1294e080dc..2c99536678 100644 --- a/synapse/rest/client/v1/login.py +++ b/synapse/rest/client/v1/login.py @@ -599,6 +599,7 @@ class SSOAuthHandler(object): redirect_url = self._add_login_token_to_redirect_url( client_redirect_url, login_token ) + # Load page request.redirect(redirect_url) finish_request(request) diff --git a/tests/rest/admin/test_user.py b/tests/rest/admin/test_user.py index 490ce8f55d..cbe4a6a51f 100644 --- a/tests/rest/admin/test_user.py +++ b/tests/rest/admin/test_user.py @@ -507,3 +507,62 @@ class UserRestTestCase(unittest.HomeserverTestCase): self.assertEqual(1, channel.json_body["admin"]) self.assertEqual(0, channel.json_body["is_guest"]) self.assertEqual(1, channel.json_body["deactivated"]) + + def test_accidental_deactivation_prevention(self): + """ + Ensure an account can't accidentally be deactivated by using a str value + for the deactivated body parameter + """ + self.hs.config.registration_shared_secret = None + + # Create user + body = json.dumps({"password": "abc123"}) + + request, channel = self.make_request( + "PUT", + self.url, + access_token=self.admin_user_tok, + content=body.encode(encoding="utf_8"), + ) + self.render(request) + + self.assertEqual(201, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@bob:test", channel.json_body["name"]) + self.assertEqual("bob", channel.json_body["displayname"]) + + # Get user + request, channel = self.make_request( + "GET", self.url, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@bob:test", channel.json_body["name"]) + self.assertEqual("bob", channel.json_body["displayname"]) + self.assertEqual(0, channel.json_body["deactivated"]) + + # Change password (and use a str for deactivate instead of a bool) + body = json.dumps({"password": "abc123", "deactivated": "false"}) # oops! + + request, channel = self.make_request( + "PUT", + self.url, + access_token=self.admin_user_tok, + content=body.encode(encoding="utf_8"), + ) + self.render(request) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + + # Check user is not deactivated + request, channel = self.make_request( + "GET", self.url, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@bob:test", channel.json_body["name"]) + self.assertEqual("bob", channel.json_body["displayname"]) + + # Ensure they're still alive + self.assertEqual(0, channel.json_body["deactivated"]) -- cgit 1.4.1 From 1f773eec912e4908ab60f7823f5c0a024261af4d Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 26 Feb 2020 15:33:26 +0000 Subject: Port PresenceHandler to async/await (#6991) --- changelog.d/6991.misc | 1 + synapse/handlers/message.py | 5 +- synapse/handlers/presence.py | 192 ++++++++++++++++-------------------- synapse/replication/tcp/resource.py | 6 +- synapse/server.pyi | 5 + tests/handlers/test_presence.py | 18 ++-- tox.ini | 1 + 7 files changed, 113 insertions(+), 115 deletions(-) create mode 100644 changelog.d/6991.misc (limited to 'tests') diff --git a/changelog.d/6991.misc b/changelog.d/6991.misc new file mode 100644 index 0000000000..5130f4e8af --- /dev/null +++ b/changelog.d/6991.misc @@ -0,0 +1 @@ +Port `synapse.handlers.presence` to async/await. diff --git a/synapse/handlers/message.py b/synapse/handlers/message.py index d6be280952..a0103addd3 100644 --- a/synapse/handlers/message.py +++ b/synapse/handlers/message.py @@ -1016,11 +1016,10 @@ class EventCreationHandler(object): # matters as sometimes presence code can take a while. run_in_background(self._bump_active_time, requester.user) - @defer.inlineCallbacks - def _bump_active_time(self, user): + async def _bump_active_time(self, user): try: presence = self.hs.get_presence_handler() - yield presence.bump_presence_active_time(user) + await presence.bump_presence_active_time(user) except Exception: logger.exception("Error bumping presence active time") diff --git a/synapse/handlers/presence.py b/synapse/handlers/presence.py index 0d6cf2b008..5526015ddb 100644 --- a/synapse/handlers/presence.py +++ b/synapse/handlers/presence.py @@ -24,11 +24,12 @@ The methods that define policy are: import logging from contextlib import contextmanager -from typing import Dict, Set +from typing import Dict, List, Set from six import iteritems, itervalues from prometheus_client import Counter +from typing_extensions import ContextManager from twisted.internet import defer @@ -42,10 +43,14 @@ from synapse.metrics.background_process_metrics import run_as_background_process from synapse.storage.presence import UserPresenceState from synapse.types import UserID, get_domain_from_id from synapse.util.async_helpers import Linearizer -from synapse.util.caches.descriptors import cachedInlineCallbacks +from synapse.util.caches.descriptors import cached from synapse.util.metrics import Measure from synapse.util.wheel_timer import WheelTimer +MYPY = False +if MYPY: + import synapse.server + logger = logging.getLogger(__name__) @@ -97,7 +102,6 @@ assert LAST_ACTIVE_GRANULARITY < IDLE_TIMER class PresenceHandler(object): def __init__(self, hs: "synapse.server.HomeServer"): self.hs = hs - self.is_mine = hs.is_mine self.is_mine_id = hs.is_mine_id self.server_name = hs.hostname self.clock = hs.get_clock() @@ -150,7 +154,7 @@ class PresenceHandler(object): # Set of users who have presence in the `user_to_current_state` that # have not yet been persisted - self.unpersisted_users_changes = set() + self.unpersisted_users_changes = set() # type: Set[str] hs.get_reactor().addSystemEventTrigger( "before", @@ -160,12 +164,11 @@ class PresenceHandler(object): self._on_shutdown, ) - self.serial_to_user = {} self._next_serial = 1 # Keeps track of the number of *ongoing* syncs on this process. While # this is non zero a user will never go offline. - self.user_to_num_current_syncs = {} + self.user_to_num_current_syncs = {} # type: Dict[str, int] # Keeps track of the number of *ongoing* syncs on other processes. # While any sync is ongoing on another process the user will never @@ -213,8 +216,7 @@ class PresenceHandler(object): self._event_pos = self.store.get_current_events_token() self._event_processing = False - @defer.inlineCallbacks - def _on_shutdown(self): + async def _on_shutdown(self): """Gets called when shutting down. This lets us persist any updates that we haven't yet persisted, e.g. updates that only changes some internal timers. This allows changes to persist across startup without having to @@ -235,7 +237,7 @@ class PresenceHandler(object): if self.unpersisted_users_changes: - yield self.store.update_presence( + await self.store.update_presence( [ self.user_to_current_state[user_id] for user_id in self.unpersisted_users_changes @@ -243,8 +245,7 @@ class PresenceHandler(object): ) logger.info("Finished _on_shutdown") - @defer.inlineCallbacks - def _persist_unpersisted_changes(self): + async def _persist_unpersisted_changes(self): """We periodically persist the unpersisted changes, as otherwise they may stack up and slow down shutdown times. """ @@ -253,12 +254,11 @@ class PresenceHandler(object): if unpersisted: logger.info("Persisting %d unpersisted presence updates", len(unpersisted)) - yield self.store.update_presence( + await self.store.update_presence( [self.user_to_current_state[user_id] for user_id in unpersisted] ) - @defer.inlineCallbacks - def _update_states(self, new_states): + async def _update_states(self, new_states): """Updates presence of users. Sets the appropriate timeouts. Pokes the notifier and federation if and only if the changed presence state should be sent to clients/servers. @@ -267,7 +267,7 @@ class PresenceHandler(object): with Measure(self.clock, "presence_update_states"): - # NOTE: We purposefully don't yield between now and when we've + # NOTE: We purposefully don't await between now and when we've # calculated what we want to do with the new states, to avoid races. to_notify = {} # Changes we want to notify everyone about @@ -311,7 +311,7 @@ class PresenceHandler(object): if to_notify: notified_presence_counter.inc(len(to_notify)) - yield self._persist_and_notify(list(to_notify.values())) + await self._persist_and_notify(list(to_notify.values())) self.unpersisted_users_changes |= {s.user_id for s in new_states} self.unpersisted_users_changes -= set(to_notify.keys()) @@ -326,7 +326,7 @@ class PresenceHandler(object): self._push_to_remotes(to_federation_ping.values()) - def _handle_timeouts(self): + async def _handle_timeouts(self): """Checks the presence of users that have timed out and updates as appropriate. """ @@ -368,10 +368,9 @@ class PresenceHandler(object): now=now, ) - return self._update_states(changes) + return await self._update_states(changes) - @defer.inlineCallbacks - def bump_presence_active_time(self, user): + async def bump_presence_active_time(self, user): """We've seen the user do something that indicates they're interacting with the app. """ @@ -383,16 +382,17 @@ class PresenceHandler(object): bump_active_time_counter.inc() - prev_state = yield self.current_state_for_user(user_id) + prev_state = await self.current_state_for_user(user_id) new_fields = {"last_active_ts": self.clock.time_msec()} if prev_state.state == PresenceState.UNAVAILABLE: new_fields["state"] = PresenceState.ONLINE - yield self._update_states([prev_state.copy_and_replace(**new_fields)]) + await self._update_states([prev_state.copy_and_replace(**new_fields)]) - @defer.inlineCallbacks - def user_syncing(self, user_id, affect_presence=True): + async def user_syncing( + self, user_id: str, affect_presence: bool = True + ) -> ContextManager[None]: """Returns a context manager that should surround any stream requests from the user. @@ -415,11 +415,11 @@ class PresenceHandler(object): curr_sync = self.user_to_num_current_syncs.get(user_id, 0) self.user_to_num_current_syncs[user_id] = curr_sync + 1 - prev_state = yield self.current_state_for_user(user_id) + prev_state = await self.current_state_for_user(user_id) if prev_state.state == PresenceState.OFFLINE: # If they're currently offline then bring them online, otherwise # just update the last sync times. - yield self._update_states( + await self._update_states( [ prev_state.copy_and_replace( state=PresenceState.ONLINE, @@ -429,7 +429,7 @@ class PresenceHandler(object): ] ) else: - yield self._update_states( + await self._update_states( [ prev_state.copy_and_replace( last_user_sync_ts=self.clock.time_msec() @@ -437,13 +437,12 @@ class PresenceHandler(object): ] ) - @defer.inlineCallbacks - def _end(): + async def _end(): try: self.user_to_num_current_syncs[user_id] -= 1 - prev_state = yield self.current_state_for_user(user_id) - yield self._update_states( + prev_state = await self.current_state_for_user(user_id) + await self._update_states( [ prev_state.copy_and_replace( last_user_sync_ts=self.clock.time_msec() @@ -480,8 +479,7 @@ class PresenceHandler(object): else: return set() - @defer.inlineCallbacks - def update_external_syncs_row( + async def update_external_syncs_row( self, process_id, user_id, is_syncing, sync_time_msec ): """Update the syncing users for an external process as a delta. @@ -494,8 +492,8 @@ class PresenceHandler(object): is_syncing (bool): Whether or not the user is now syncing sync_time_msec(int): Time in ms when the user was last syncing """ - with (yield self.external_sync_linearizer.queue(process_id)): - prev_state = yield self.current_state_for_user(user_id) + with (await self.external_sync_linearizer.queue(process_id)): + prev_state = await self.current_state_for_user(user_id) process_presence = self.external_process_to_current_syncs.setdefault( process_id, set() @@ -525,25 +523,24 @@ class PresenceHandler(object): process_presence.discard(user_id) if updates: - yield self._update_states(updates) + await self._update_states(updates) self.external_process_last_updated_ms[process_id] = self.clock.time_msec() - @defer.inlineCallbacks - def update_external_syncs_clear(self, process_id): + async def update_external_syncs_clear(self, process_id): """Marks all users that had been marked as syncing by a given process as offline. Used when the process has stopped/disappeared. """ - with (yield self.external_sync_linearizer.queue(process_id)): + with (await self.external_sync_linearizer.queue(process_id)): process_presence = self.external_process_to_current_syncs.pop( process_id, set() ) - prev_states = yield self.current_state_for_users(process_presence) + prev_states = await self.current_state_for_users(process_presence) time_now_ms = self.clock.time_msec() - yield self._update_states( + await self._update_states( [ prev_state.copy_and_replace(last_user_sync_ts=time_now_ms) for prev_state in itervalues(prev_states) @@ -551,15 +548,13 @@ class PresenceHandler(object): ) self.external_process_last_updated_ms.pop(process_id, None) - @defer.inlineCallbacks - def current_state_for_user(self, user_id): + async def current_state_for_user(self, user_id): """Get the current presence state for a user. """ - res = yield self.current_state_for_users([user_id]) + res = await self.current_state_for_users([user_id]) return res[user_id] - @defer.inlineCallbacks - def current_state_for_users(self, user_ids): + async def current_state_for_users(self, user_ids): """Get the current presence state for multiple users. Returns: @@ -574,7 +569,7 @@ class PresenceHandler(object): if missing: # There are things not in our in memory cache. Lets pull them out of # the database. - res = yield self.store.get_presence_for_users(missing) + res = await self.store.get_presence_for_users(missing) states.update(res) missing = [user_id for user_id, state in iteritems(states) if not state] @@ -587,14 +582,13 @@ class PresenceHandler(object): return states - @defer.inlineCallbacks - def _persist_and_notify(self, states): + async def _persist_and_notify(self, states): """Persist states in the database, poke the notifier and send to interested remote servers """ - stream_id, max_token = yield self.store.update_presence(states) + stream_id, max_token = await self.store.update_presence(states) - parties = yield get_interested_parties(self.store, states) + parties = await get_interested_parties(self.store, states) room_ids_to_states, users_to_states = parties self.notifier.on_new_event( @@ -606,9 +600,8 @@ class PresenceHandler(object): self._push_to_remotes(states) - @defer.inlineCallbacks - def notify_for_states(self, state, stream_id): - parties = yield get_interested_parties(self.store, [state]) + async def notify_for_states(self, state, stream_id): + parties = await get_interested_parties(self.store, [state]) room_ids_to_states, users_to_states = parties self.notifier.on_new_event( @@ -626,8 +619,7 @@ class PresenceHandler(object): """ self.federation.send_presence(states) - @defer.inlineCallbacks - def incoming_presence(self, origin, content): + async def incoming_presence(self, origin, content): """Called when we receive a `m.presence` EDU from a remote server. """ now = self.clock.time_msec() @@ -670,21 +662,19 @@ class PresenceHandler(object): new_fields["status_msg"] = push.get("status_msg", None) new_fields["currently_active"] = push.get("currently_active", False) - prev_state = yield self.current_state_for_user(user_id) + prev_state = await self.current_state_for_user(user_id) updates.append(prev_state.copy_and_replace(**new_fields)) if updates: federation_presence_counter.inc(len(updates)) - yield self._update_states(updates) + await self._update_states(updates) - @defer.inlineCallbacks - def get_state(self, target_user, as_event=False): - results = yield self.get_states([target_user.to_string()], as_event=as_event) + async def get_state(self, target_user, as_event=False): + results = await self.get_states([target_user.to_string()], as_event=as_event) return results[0] - @defer.inlineCallbacks - def get_states(self, target_user_ids, as_event=False): + async def get_states(self, target_user_ids, as_event=False): """Get the presence state for users. Args: @@ -695,7 +685,7 @@ class PresenceHandler(object): list """ - updates = yield self.current_state_for_users(target_user_ids) + updates = await self.current_state_for_users(target_user_ids) updates = list(updates.values()) for user_id in set(target_user_ids) - {u.user_id for u in updates}: @@ -713,8 +703,7 @@ class PresenceHandler(object): else: return updates - @defer.inlineCallbacks - def set_state(self, target_user, state, ignore_status_msg=False): + async def set_state(self, target_user, state, ignore_status_msg=False): """Set the presence state of the user. """ status_msg = state.get("status_msg", None) @@ -730,7 +719,7 @@ class PresenceHandler(object): user_id = target_user.to_string() - prev_state = yield self.current_state_for_user(user_id) + prev_state = await self.current_state_for_user(user_id) new_fields = {"state": presence} @@ -741,16 +730,15 @@ class PresenceHandler(object): if presence == PresenceState.ONLINE: new_fields["last_active_ts"] = self.clock.time_msec() - yield self._update_states([prev_state.copy_and_replace(**new_fields)]) + await self._update_states([prev_state.copy_and_replace(**new_fields)]) - @defer.inlineCallbacks - def is_visible(self, observed_user, observer_user): + async def is_visible(self, observed_user, observer_user): """Returns whether a user can see another user's presence. """ - observer_room_ids = yield self.store.get_rooms_for_user( + observer_room_ids = await self.store.get_rooms_for_user( observer_user.to_string() ) - observed_room_ids = yield self.store.get_rooms_for_user( + observed_room_ids = await self.store.get_rooms_for_user( observed_user.to_string() ) @@ -759,8 +747,7 @@ class PresenceHandler(object): return False - @defer.inlineCallbacks - def get_all_presence_updates(self, last_id, current_id): + async def get_all_presence_updates(self, last_id, current_id): """ Gets a list of presence update rows from between the given stream ids. Each row has: @@ -775,7 +762,7 @@ class PresenceHandler(object): """ # TODO(markjh): replicate the unpersisted changes. # This could use the in-memory stores for recent changes. - rows = yield self.store.get_all_presence_updates(last_id, current_id) + rows = await self.store.get_all_presence_updates(last_id, current_id) return rows def notify_new_event(self): @@ -786,20 +773,18 @@ class PresenceHandler(object): if self._event_processing: return - @defer.inlineCallbacks - def _process_presence(): + async def _process_presence(): assert not self._event_processing self._event_processing = True try: - yield self._unsafe_process() + await self._unsafe_process() finally: self._event_processing = False run_as_background_process("presence.notify_new_event", _process_presence) - @defer.inlineCallbacks - def _unsafe_process(self): + async def _unsafe_process(self): # Loop round handling deltas until we're up to date while True: with Measure(self.clock, "presence_delta"): @@ -812,10 +797,10 @@ class PresenceHandler(object): self._event_pos, room_max_stream_ordering, ) - max_pos, deltas = yield self.store.get_current_state_deltas( + max_pos, deltas = await self.store.get_current_state_deltas( self._event_pos, room_max_stream_ordering ) - yield self._handle_state_delta(deltas) + await self._handle_state_delta(deltas) self._event_pos = max_pos @@ -824,8 +809,7 @@ class PresenceHandler(object): max_pos ) - @defer.inlineCallbacks - def _handle_state_delta(self, deltas): + async def _handle_state_delta(self, deltas): """Process current state deltas to find new joins that need to be handled. """ @@ -846,13 +830,13 @@ class PresenceHandler(object): # joins. continue - event = yield self.store.get_event(event_id, allow_none=True) + event = await self.store.get_event(event_id, allow_none=True) if not event or event.content.get("membership") != Membership.JOIN: # We only care about joins continue if prev_event_id: - prev_event = yield self.store.get_event(prev_event_id, allow_none=True) + prev_event = await self.store.get_event(prev_event_id, allow_none=True) if ( prev_event and prev_event.content.get("membership") == Membership.JOIN @@ -860,10 +844,9 @@ class PresenceHandler(object): # Ignore changes to join events. continue - yield self._on_user_joined_room(room_id, state_key) + await self._on_user_joined_room(room_id, state_key) - @defer.inlineCallbacks - def _on_user_joined_room(self, room_id, user_id): + async def _on_user_joined_room(self, room_id, user_id): """Called when we detect a user joining the room via the current state delta stream. @@ -882,8 +865,8 @@ class PresenceHandler(object): # TODO: We should be able to filter the hosts down to those that # haven't previously seen the user - state = yield self.current_state_for_user(user_id) - hosts = yield self.state.get_current_hosts_in_room(room_id) + state = await self.current_state_for_user(user_id) + hosts = await self.state.get_current_hosts_in_room(room_id) # Filter out ourselves. hosts = {host for host in hosts if host != self.server_name} @@ -903,10 +886,10 @@ class PresenceHandler(object): # TODO: Check that this is actually a new server joining the # room. - user_ids = yield self.state.get_current_users_in_room(room_id) + user_ids = await self.state.get_current_users_in_room(room_id) user_ids = list(filter(self.is_mine_id, user_ids)) - states = yield self.current_state_for_users(user_ids) + states = await self.current_state_for_users(user_ids) # Filter out old presence, i.e. offline presence states where # the user hasn't been active for a week. We can change this @@ -996,9 +979,8 @@ class PresenceEventSource(object): self.store = hs.get_datastore() self.state = hs.get_state_handler() - @defer.inlineCallbacks @log_function - def get_new_events( + async def get_new_events( self, user, from_key, @@ -1045,7 +1027,7 @@ class PresenceEventSource(object): presence = self.get_presence_handler() stream_change_cache = self.store.presence_stream_cache - users_interested_in = yield self._get_interested_in(user, explicit_room_id) + users_interested_in = await self._get_interested_in(user, explicit_room_id) user_ids_changed = set() changed = None @@ -1071,7 +1053,7 @@ class PresenceEventSource(object): else: user_ids_changed = users_interested_in - updates = yield presence.current_state_for_users(user_ids_changed) + updates = await presence.current_state_for_users(user_ids_changed) if include_offline: return (list(updates.values()), max_token) @@ -1084,11 +1066,11 @@ class PresenceEventSource(object): def get_current_key(self): return self.store.get_current_presence_token() - def get_pagination_rows(self, user, pagination_config, key): - return self.get_new_events(user, from_key=None, include_offline=False) + async def get_pagination_rows(self, user, pagination_config, key): + return await self.get_new_events(user, from_key=None, include_offline=False) - @cachedInlineCallbacks(num_args=2, cache_context=True) - def _get_interested_in(self, user, explicit_room_id, cache_context): + @cached(num_args=2, cache_context=True) + async def _get_interested_in(self, user, explicit_room_id, cache_context): """Returns the set of users that the given user should see presence updates for """ @@ -1096,13 +1078,13 @@ class PresenceEventSource(object): users_interested_in = set() users_interested_in.add(user_id) # So that we receive our own presence - users_who_share_room = yield self.store.get_users_who_share_room_with_user( + users_who_share_room = await self.store.get_users_who_share_room_with_user( user_id, on_invalidate=cache_context.invalidate ) users_interested_in.update(users_who_share_room) if explicit_room_id: - user_ids = yield self.store.get_users_in_room( + user_ids = await self.store.get_users_in_room( explicit_room_id, on_invalidate=cache_context.invalidate ) users_interested_in.update(user_ids) @@ -1277,8 +1259,8 @@ def get_interested_parties(store, states): 2-tuple: `(room_ids_to_states, users_to_states)`, with each item being a dict of `entity_name` -> `[UserPresenceState]` """ - room_ids_to_states = {} - users_to_states = {} + room_ids_to_states = {} # type: Dict[str, List[UserPresenceState]] + users_to_states = {} # type: Dict[str, List[UserPresenceState]] for state in states: room_ids = yield store.get_rooms_for_user(state.user_id) for room_id in room_ids: diff --git a/synapse/replication/tcp/resource.py b/synapse/replication/tcp/resource.py index ce60ae2e07..ce9d1fae12 100644 --- a/synapse/replication/tcp/resource.py +++ b/synapse/replication/tcp/resource.py @@ -323,7 +323,11 @@ class ReplicationStreamer(object): # We need to tell the presence handler that the connection has been # lost so that it can handle any ongoing syncs on that connection. - self.presence_handler.update_external_syncs_clear(connection.conn_id) + run_as_background_process( + "update_external_syncs_clear", + self.presence_handler.update_external_syncs_clear, + connection.conn_id, + ) def _batch_updates(updates): diff --git a/synapse/server.pyi b/synapse/server.pyi index 40eabfe5d9..3844f0e12f 100644 --- a/synapse/server.pyi +++ b/synapse/server.pyi @@ -3,6 +3,7 @@ import twisted.internet import synapse.api.auth import synapse.config.homeserver import synapse.crypto.keyring +import synapse.federation.federation_server import synapse.federation.sender import synapse.federation.transport.client import synapse.handlers @@ -107,5 +108,9 @@ class HomeServer(object): self, ) -> synapse.replication.tcp.client.ReplicationClientHandler: pass + def get_federation_registry( + self, + ) -> synapse.federation.federation_server.FederationHandlerRegistry: + pass def is_mine_id(self, domain_id: str) -> bool: pass diff --git a/tests/handlers/test_presence.py b/tests/handlers/test_presence.py index 64915bafcd..05ea40a7de 100644 --- a/tests/handlers/test_presence.py +++ b/tests/handlers/test_presence.py @@ -494,8 +494,10 @@ class PresenceJoinTestCase(unittest.HomeserverTestCase): self.helper.join(room_id, "@test2:server") # Mark test2 as online, test will be offline with a last_active of 0 - self.presence_handler.set_state( - UserID.from_string("@test2:server"), {"presence": PresenceState.ONLINE} + self.get_success( + self.presence_handler.set_state( + UserID.from_string("@test2:server"), {"presence": PresenceState.ONLINE} + ) ) self.reactor.pump([0]) # Wait for presence updates to be handled @@ -543,14 +545,18 @@ class PresenceJoinTestCase(unittest.HomeserverTestCase): room_id = self.helper.create_room_as(self.user_id) # Mark test as online - self.presence_handler.set_state( - UserID.from_string("@test:server"), {"presence": PresenceState.ONLINE} + self.get_success( + self.presence_handler.set_state( + UserID.from_string("@test:server"), {"presence": PresenceState.ONLINE} + ) ) # Mark test2 as online, test will be offline with a last_active of 0. # Note we don't join them to the room yet - self.presence_handler.set_state( - UserID.from_string("@test2:server"), {"presence": PresenceState.ONLINE} + self.get_success( + self.presence_handler.set_state( + UserID.from_string("@test2:server"), {"presence": PresenceState.ONLINE} + ) ) # Add servers to the room diff --git a/tox.ini b/tox.ini index b715ea0bff..4ccfde01b5 100644 --- a/tox.ini +++ b/tox.ini @@ -183,6 +183,7 @@ commands = mypy \ synapse/events/spamcheck.py \ synapse/federation/sender \ synapse/federation/transport \ + synapse/handlers/presence.py \ synapse/handlers/sync.py \ synapse/handlers/ui_auth \ synapse/logging/ \ -- cgit 1.4.1 From 3e99528f2bfaa686c4708fb8efcddce935b2397d Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Wed, 26 Feb 2020 16:58:33 +0000 Subject: Store room version on invite (#6983) When we get an invite over federation, store the room version in the rooms table. The general idea here is that, when we pull the invite out again, we'll want to know what room_version it belongs to (so that we can later redact it if need be). So we need to store it somewhere... --- changelog.d/6983.misc | 1 + synapse/handlers/federation.py | 12 +++++++++++ synapse/replication/http/_base.py | 2 +- synapse/replication/http/federation.py | 36 +++++++++++++++++++++++++++++++- synapse/storage/data_stores/main/room.py | 20 ++++++++++++++++++ tests/app/test_openid_listener.py | 8 +++++++ tests/handlers/test_typing.py | 1 + 7 files changed, 78 insertions(+), 2 deletions(-) create mode 100644 changelog.d/6983.misc (limited to 'tests') diff --git a/changelog.d/6983.misc b/changelog.d/6983.misc new file mode 100644 index 0000000000..08aa80bcd9 --- /dev/null +++ b/changelog.d/6983.misc @@ -0,0 +1 @@ +Refactoring work in preparation for changing the event redaction algorithm. diff --git a/synapse/handlers/federation.py b/synapse/handlers/federation.py index c2e6ee266d..38ab6a8fc3 100644 --- a/synapse/handlers/federation.py +++ b/synapse/handlers/federation.py @@ -60,6 +60,7 @@ from synapse.replication.http.devices import ReplicationUserDevicesResyncRestSer from synapse.replication.http.federation import ( ReplicationCleanRoomRestServlet, ReplicationFederationSendEventsRestServlet, + ReplicationStoreRoomOnInviteRestServlet, ) from synapse.replication.http.membership import ReplicationUserJoinedLeftRoomRestServlet from synapse.state import StateResolutionStore, resolve_events_with_store @@ -160,8 +161,12 @@ class FederationHandler(BaseHandler): self._user_device_resync = ReplicationUserDevicesResyncRestServlet.make_client( hs ) + self._maybe_store_room_on_invite = ReplicationStoreRoomOnInviteRestServlet.make_client( + hs + ) else: self._device_list_updater = hs.get_device_handler().device_list_updater + self._maybe_store_room_on_invite = self.store.maybe_store_room_on_invite # When joining a room we need to queue any events for that room up self.room_queues = {} @@ -1537,6 +1542,13 @@ class FederationHandler(BaseHandler): if event.state_key == self._server_notices_mxid: raise SynapseError(http_client.FORBIDDEN, "Cannot invite this user") + # keep a record of the room version, if we don't yet know it. + # (this may get overwritten if we later get a different room version in a + # join dance). + await self._maybe_store_room_on_invite( + room_id=event.room_id, room_version=room_version + ) + event.internal_metadata.outlier = True event.internal_metadata.out_of_band_membership = True diff --git a/synapse/replication/http/_base.py b/synapse/replication/http/_base.py index 444eb7b7f4..1be1ccbdf3 100644 --- a/synapse/replication/http/_base.py +++ b/synapse/replication/http/_base.py @@ -44,7 +44,7 @@ class ReplicationEndpoint(object): """Helper base class for defining new replication HTTP endpoints. This creates an endpoint under `/_synapse/replication/:NAME/:PATH_ARGS..` - (with an `/:txn_id` prefix for cached requests.), where NAME is a name, + (with a `/:txn_id` suffix for cached requests), where NAME is a name, PATH_ARGS are a tuple of parameters to be encoded in the URL. For example, if `NAME` is "send_event" and `PATH_ARGS` is `("event_id",)`, diff --git a/synapse/replication/http/federation.py b/synapse/replication/http/federation.py index 49a3251372..8794720101 100644 --- a/synapse/replication/http/federation.py +++ b/synapse/replication/http/federation.py @@ -17,6 +17,7 @@ import logging from twisted.internet import defer +from synapse.api.room_versions import KNOWN_ROOM_VERSIONS from synapse.events import event_type_from_format_version from synapse.events.snapshot import EventContext from synapse.http.servlet import parse_json_object_from_request @@ -211,7 +212,7 @@ class ReplicationCleanRoomRestServlet(ReplicationEndpoint): Request format: - POST /_synapse/replication/fed_query/:fed_cleanup_room/:txn_id + POST /_synapse/replication/fed_cleanup_room/:room_id/:txn_id {} """ @@ -238,8 +239,41 @@ class ReplicationCleanRoomRestServlet(ReplicationEndpoint): return 200, {} +class ReplicationStoreRoomOnInviteRestServlet(ReplicationEndpoint): + """Called to clean up any data in DB for a given room, ready for the + server to join the room. + + Request format: + + POST /_synapse/replication/store_room_on_invite/:room_id/:txn_id + + { + "room_version": "1", + } + """ + + NAME = "store_room_on_invite" + PATH_ARGS = ("room_id",) + + def __init__(self, hs): + super().__init__(hs) + + self.store = hs.get_datastore() + + @staticmethod + def _serialize_payload(room_id, room_version): + return {"room_version": room_version.identifier} + + async def _handle_request(self, request, room_id): + content = parse_json_object_from_request(request) + room_version = KNOWN_ROOM_VERSIONS[content["room_version"]] + await self.store.maybe_store_room_on_invite(room_id, room_version) + return 200, {} + + def register_servlets(hs, http_server): ReplicationFederationSendEventsRestServlet(hs).register(http_server) ReplicationFederationSendEduRestServlet(hs).register(http_server) ReplicationGetQueryRestServlet(hs).register(http_server) ReplicationCleanRoomRestServlet(hs).register(http_server) + ReplicationStoreRoomOnInviteRestServlet(hs).register(http_server) diff --git a/synapse/storage/data_stores/main/room.py b/synapse/storage/data_stores/main/room.py index 70137dfbe4..e6c10c6316 100644 --- a/synapse/storage/data_stores/main/room.py +++ b/synapse/storage/data_stores/main/room.py @@ -1020,6 +1020,26 @@ class RoomStore(RoomBackgroundUpdateStore, RoomWorkerStore, SearchStore): logger.error("store_room with room_id=%s failed: %s", room_id, e) raise StoreError(500, "Problem creating room.") + async def maybe_store_room_on_invite(self, room_id: str, room_version: RoomVersion): + """ + When we receive an invite over federation, store the version of the room if we + don't already know the room version. + """ + await self.db.simple_upsert( + desc="maybe_store_room_on_invite", + table="rooms", + keyvalues={"room_id": room_id}, + values={}, + insertion_values={ + "room_version": room_version.identifier, + "is_public": False, + "creator": "", + }, + # rooms has a unique constraint on room_id, so no need to lock when doing an + # emulated upsert. + lock=False, + ) + @defer.inlineCallbacks def set_room_is_public(self, room_id, is_public): def set_room_is_public_txn(txn, next_id): diff --git a/tests/app/test_openid_listener.py b/tests/app/test_openid_listener.py index 1fe048048b..89fcc3889a 100644 --- a/tests/app/test_openid_listener.py +++ b/tests/app/test_openid_listener.py @@ -29,6 +29,14 @@ class FederationReaderOpenIDListenerTests(HomeserverTestCase): ) return hs + def default_config(self, name="test"): + conf = super().default_config(name) + # we're using FederationReaderServer, which uses a SlavedStore, so we + # have to tell the FederationHandler not to try to access stuff that is only + # in the primary store. + conf["worker_app"] = "yes" + return conf + @parameterized.expand( [ (["federation"], "auth_fail"), diff --git a/tests/handlers/test_typing.py b/tests/handlers/test_typing.py index 07b204666e..51e2b37218 100644 --- a/tests/handlers/test_typing.py +++ b/tests/handlers/test_typing.py @@ -74,6 +74,7 @@ class TypingNotificationsTestCase(unittest.HomeserverTestCase): "set_received_txn_response", "get_destination_retry_timings", "get_devices_by_remote", + "maybe_store_room_on_invite", # Bits that user_directory needs "get_user_directory_stream_pos", "get_current_state_deltas", -- cgit 1.4.1 From cab4a52535097c5836fe67c5e09e8350d7ccf03c Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Thu, 27 Feb 2020 13:08:43 +0000 Subject: set worker_app for frontend proxy test (#7003) to stop the federationhandler trying to do master stuff --- changelog.d/7003.misc | 1 + tests/app/test_frontend_proxy.py | 5 +++++ 2 files changed, 6 insertions(+) create mode 100644 changelog.d/7003.misc (limited to 'tests') diff --git a/changelog.d/7003.misc b/changelog.d/7003.misc new file mode 100644 index 0000000000..08aa80bcd9 --- /dev/null +++ b/changelog.d/7003.misc @@ -0,0 +1 @@ +Refactoring work in preparation for changing the event redaction algorithm. diff --git a/tests/app/test_frontend_proxy.py b/tests/app/test_frontend_proxy.py index 160e55aca9..d3feafa1b7 100644 --- a/tests/app/test_frontend_proxy.py +++ b/tests/app/test_frontend_proxy.py @@ -27,6 +27,11 @@ class FrontendProxyTests(HomeserverTestCase): return hs + def default_config(self, name="test"): + c = super().default_config(name) + c["worker_app"] = "synapse.app.frontend_proxy" + return c + def test_listen_http_with_presence_enabled(self): """ When presence is on, the stub servlet will not register. -- cgit 1.4.1 From 9b06d8f8a62dc5c423aa9a694e0759eaf1c3c77e Mon Sep 17 00:00:00 2001 From: Dirk Klimpel <5740567+dklimpel@users.noreply.github.com> Date: Fri, 28 Feb 2020 10:58:05 +0100 Subject: Fixed set a user as an admin with the new API (#6928) Fix #6910 --- changelog.d/6910.bugfix | 1 + synapse/rest/admin/users.py | 6 +- synapse/storage/data_stores/main/registration.py | 16 +- tests/rest/admin/test_user.py | 218 +++++++++++++++++++---- 4 files changed, 199 insertions(+), 42 deletions(-) create mode 100644 changelog.d/6910.bugfix (limited to 'tests') diff --git a/changelog.d/6910.bugfix b/changelog.d/6910.bugfix new file mode 100644 index 0000000000..707f1ff7b5 --- /dev/null +++ b/changelog.d/6910.bugfix @@ -0,0 +1 @@ +Fixed set a user as an admin with the admin API `PUT /_synapse/admin/v2/users/`. Contributed by @dklimpel. diff --git a/synapse/rest/admin/users.py b/synapse/rest/admin/users.py index c5b461a236..80f959248d 100644 --- a/synapse/rest/admin/users.py +++ b/synapse/rest/admin/users.py @@ -211,9 +211,7 @@ class UserRestServletV2(RestServlet): if target_user == auth_user and not set_admin_to: raise SynapseError(400, "You may not demote yourself.") - await self.admin_handler.set_user_server_admin( - target_user, set_admin_to - ) + await self.store.set_server_admin(target_user, set_admin_to) if "password" in body: if ( @@ -651,6 +649,6 @@ class UserAdminServlet(RestServlet): if target_user == auth_user and not set_admin_to: raise SynapseError(400, "You may not demote yourself.") - await self.store.set_user_server_admin(target_user, set_admin_to) + await self.store.set_server_admin(target_user, set_admin_to) return 200, {} diff --git a/synapse/storage/data_stores/main/registration.py b/synapse/storage/data_stores/main/registration.py index 49306642ed..3e53c8568a 100644 --- a/synapse/storage/data_stores/main/registration.py +++ b/synapse/storage/data_stores/main/registration.py @@ -301,12 +301,16 @@ class RegistrationWorkerStore(SQLBaseStore): admin (bool): true iff the user is to be a server admin, false otherwise. """ - return self.db.simple_update_one( - table="users", - keyvalues={"name": user.to_string()}, - updatevalues={"admin": 1 if admin else 0}, - desc="set_server_admin", - ) + + def set_server_admin_txn(txn): + self.db.simple_update_one_txn( + txn, "users", {"name": user.to_string()}, {"admin": 1 if admin else 0} + ) + self._invalidate_cache_and_stream( + txn, self.get_user_by_id, (user.to_string(),) + ) + + return self.db.runInteraction("set_server_admin", set_server_admin_txn) def _query_for_auth(self, txn, token): sql = ( diff --git a/tests/rest/admin/test_user.py b/tests/rest/admin/test_user.py index cbe4a6a51f..6416fb5d2a 100644 --- a/tests/rest/admin/test_user.py +++ b/tests/rest/admin/test_user.py @@ -16,6 +16,7 @@ import hashlib import hmac import json +import urllib.parse from mock import Mock @@ -371,22 +372,24 @@ class UserRestTestCase(unittest.HomeserverTestCase): def prepare(self, reactor, clock, hs): self.store = hs.get_datastore() - self.url = "/_synapse/admin/v2/users/@bob:test" - self.admin_user = self.register_user("admin", "pass", admin=True) self.admin_user_tok = self.login("admin", "pass") self.other_user = self.register_user("user", "pass") self.other_user_token = self.login("user", "pass") + self.url_other_user = "/_synapse/admin/v2/users/%s" % urllib.parse.quote( + self.other_user + ) def test_requester_is_no_admin(self): """ If the user is not a server admin, an error is returned. """ self.hs.config.registration_shared_secret = None + url = "/_synapse/admin/v2/users/@bob:test" request, channel = self.make_request( - "GET", self.url, access_token=self.other_user_token, + "GET", url, access_token=self.other_user_token, ) self.render(request) @@ -394,7 +397,7 @@ class UserRestTestCase(unittest.HomeserverTestCase): self.assertEqual("You are not a server admin", channel.json_body["error"]) request, channel = self.make_request( - "PUT", self.url, access_token=self.other_user_token, content=b"{}", + "PUT", url, access_token=self.other_user_token, content=b"{}", ) self.render(request) @@ -417,24 +420,73 @@ class UserRestTestCase(unittest.HomeserverTestCase): self.assertEqual(404, channel.code, msg=channel.json_body) self.assertEqual("M_NOT_FOUND", channel.json_body["errcode"]) - def test_requester_is_admin(self): + def test_create_server_admin(self): """ - If the user is a server admin, a new user is created. + Check that a new admin user is created successfully. """ self.hs.config.registration_shared_secret = None + url = "/_synapse/admin/v2/users/@bob:test" + # Create user (server admin) body = json.dumps( { "password": "abc123", "admin": True, + "displayname": "Bob's name", "threepids": [{"medium": "email", "address": "bob@bob.bob"}], } ) + request, channel = self.make_request( + "PUT", + url, + access_token=self.admin_user_tok, + content=body.encode(encoding="utf_8"), + ) + self.render(request) + + self.assertEqual(201, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@bob:test", channel.json_body["name"]) + self.assertEqual("Bob's name", channel.json_body["displayname"]) + self.assertEqual("email", channel.json_body["threepids"][0]["medium"]) + self.assertEqual("bob@bob.bob", channel.json_body["threepids"][0]["address"]) + self.assertEqual(True, channel.json_body["admin"]) + + # Get user + request, channel = self.make_request( + "GET", url, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@bob:test", channel.json_body["name"]) + self.assertEqual("Bob's name", channel.json_body["displayname"]) + self.assertEqual("email", channel.json_body["threepids"][0]["medium"]) + self.assertEqual("bob@bob.bob", channel.json_body["threepids"][0]["address"]) + self.assertEqual(True, channel.json_body["admin"]) + self.assertEqual(False, channel.json_body["is_guest"]) + self.assertEqual(False, channel.json_body["deactivated"]) + + def test_create_user(self): + """ + Check that a new regular user is created successfully. + """ + self.hs.config.registration_shared_secret = None + url = "/_synapse/admin/v2/users/@bob:test" + # Create user + body = json.dumps( + { + "password": "abc123", + "admin": False, + "displayname": "Bob's name", + "threepids": [{"medium": "email", "address": "bob@bob.bob"}], + } + ) + request, channel = self.make_request( "PUT", - self.url, + url, access_token=self.admin_user_tok, content=body.encode(encoding="utf_8"), ) @@ -442,29 +494,38 @@ class UserRestTestCase(unittest.HomeserverTestCase): self.assertEqual(201, int(channel.result["code"]), msg=channel.result["body"]) self.assertEqual("@bob:test", channel.json_body["name"]) - self.assertEqual("bob", channel.json_body["displayname"]) + self.assertEqual("Bob's name", channel.json_body["displayname"]) self.assertEqual("email", channel.json_body["threepids"][0]["medium"]) self.assertEqual("bob@bob.bob", channel.json_body["threepids"][0]["address"]) + self.assertEqual(False, channel.json_body["admin"]) # Get user request, channel = self.make_request( - "GET", self.url, access_token=self.admin_user_tok, + "GET", url, access_token=self.admin_user_tok, ) self.render(request) self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) self.assertEqual("@bob:test", channel.json_body["name"]) - self.assertEqual("bob", channel.json_body["displayname"]) - self.assertEqual(1, channel.json_body["admin"]) - self.assertEqual(0, channel.json_body["is_guest"]) - self.assertEqual(0, channel.json_body["deactivated"]) + self.assertEqual("Bob's name", channel.json_body["displayname"]) + self.assertEqual("email", channel.json_body["threepids"][0]["medium"]) + self.assertEqual("bob@bob.bob", channel.json_body["threepids"][0]["address"]) + self.assertEqual(False, channel.json_body["admin"]) + self.assertEqual(False, channel.json_body["is_guest"]) + self.assertEqual(False, channel.json_body["deactivated"]) + + def test_set_password(self): + """ + Test setting a new password for another user. + """ + self.hs.config.registration_shared_secret = None # Change password body = json.dumps({"password": "hahaha"}) request, channel = self.make_request( "PUT", - self.url, + self.url_other_user, access_token=self.admin_user_tok, content=body.encode(encoding="utf_8"), ) @@ -472,41 +533,133 @@ class UserRestTestCase(unittest.HomeserverTestCase): self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + def test_set_displayname(self): + """ + Test setting the displayname of another user. + """ + self.hs.config.registration_shared_secret = None + # Modify user + body = json.dumps({"displayname": "foobar"}) + + request, channel = self.make_request( + "PUT", + self.url_other_user, + access_token=self.admin_user_tok, + content=body.encode(encoding="utf_8"), + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@user:test", channel.json_body["name"]) + self.assertEqual("foobar", channel.json_body["displayname"]) + + # Get user + request, channel = self.make_request( + "GET", self.url_other_user, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@user:test", channel.json_body["name"]) + self.assertEqual("foobar", channel.json_body["displayname"]) + + def test_set_threepid(self): + """ + Test setting threepid for an other user. + """ + self.hs.config.registration_shared_secret = None + + # Delete old and add new threepid to user body = json.dumps( - { - "displayname": "foobar", - "deactivated": True, - "threepids": [{"medium": "email", "address": "bob2@bob.bob"}], - } + {"threepids": [{"medium": "email", "address": "bob3@bob.bob"}]} ) request, channel = self.make_request( "PUT", - self.url, + self.url_other_user, access_token=self.admin_user_tok, content=body.encode(encoding="utf_8"), ) self.render(request) self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual("@bob:test", channel.json_body["name"]) - self.assertEqual("foobar", channel.json_body["displayname"]) + self.assertEqual("@user:test", channel.json_body["name"]) + self.assertEqual("email", channel.json_body["threepids"][0]["medium"]) + self.assertEqual("bob3@bob.bob", channel.json_body["threepids"][0]["address"]) + + # Get user + request, channel = self.make_request( + "GET", self.url_other_user, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@user:test", channel.json_body["name"]) + self.assertEqual("email", channel.json_body["threepids"][0]["medium"]) + self.assertEqual("bob3@bob.bob", channel.json_body["threepids"][0]["address"]) + + def test_deactivate_user(self): + """ + Test deactivating another user. + """ + + # Deactivate user + body = json.dumps({"deactivated": True}) + + request, channel = self.make_request( + "PUT", + self.url_other_user, + access_token=self.admin_user_tok, + content=body.encode(encoding="utf_8"), + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@user:test", channel.json_body["name"]) self.assertEqual(True, channel.json_body["deactivated"]) # the user is deactivated, the threepid will be deleted # Get user request, channel = self.make_request( - "GET", self.url, access_token=self.admin_user_tok, + "GET", self.url_other_user, access_token=self.admin_user_tok, ) self.render(request) self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual("@bob:test", channel.json_body["name"]) - self.assertEqual("foobar", channel.json_body["displayname"]) - self.assertEqual(1, channel.json_body["admin"]) - self.assertEqual(0, channel.json_body["is_guest"]) - self.assertEqual(1, channel.json_body["deactivated"]) + self.assertEqual("@user:test", channel.json_body["name"]) + self.assertEqual(True, channel.json_body["deactivated"]) + + def test_set_user_as_admin(self): + """ + Test setting the admin flag on a user. + """ + self.hs.config.registration_shared_secret = None + + # Set a user as an admin + body = json.dumps({"admin": True}) + + request, channel = self.make_request( + "PUT", + self.url_other_user, + access_token=self.admin_user_tok, + content=body.encode(encoding="utf_8"), + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@user:test", channel.json_body["name"]) + self.assertEqual(True, channel.json_body["admin"]) + + # Get user + request, channel = self.make_request( + "GET", self.url_other_user, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@user:test", channel.json_body["name"]) + self.assertEqual(True, channel.json_body["admin"]) def test_accidental_deactivation_prevention(self): """ @@ -514,13 +667,14 @@ class UserRestTestCase(unittest.HomeserverTestCase): for the deactivated body parameter """ self.hs.config.registration_shared_secret = None + url = "/_synapse/admin/v2/users/@bob:test" # Create user body = json.dumps({"password": "abc123"}) request, channel = self.make_request( "PUT", - self.url, + url, access_token=self.admin_user_tok, content=body.encode(encoding="utf_8"), ) @@ -532,7 +686,7 @@ class UserRestTestCase(unittest.HomeserverTestCase): # Get user request, channel = self.make_request( - "GET", self.url, access_token=self.admin_user_tok, + "GET", url, access_token=self.admin_user_tok, ) self.render(request) @@ -546,7 +700,7 @@ class UserRestTestCase(unittest.HomeserverTestCase): request, channel = self.make_request( "PUT", - self.url, + url, access_token=self.admin_user_tok, content=body.encode(encoding="utf_8"), ) @@ -556,7 +710,7 @@ class UserRestTestCase(unittest.HomeserverTestCase): # Check user is not deactivated request, channel = self.make_request( - "GET", self.url, access_token=self.admin_user_tok, + "GET", url, access_token=self.admin_user_tok, ) self.render(request) -- cgit 1.4.1 From f70f44abc73689a66d0a05dc703ca38241092174 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Fri, 28 Feb 2020 11:45:35 +0000 Subject: Remove handling of multiple rows per ID --- synapse/storage/data_stores/main/devices.py | 35 +--------------------- tests/storage/test_devices.py | 45 ----------------------------- 2 files changed, 1 insertion(+), 79 deletions(-) (limited to 'tests') diff --git a/synapse/storage/data_stores/main/devices.py b/synapse/storage/data_stores/main/devices.py index 768afe7a6c..06e1d9f033 100644 --- a/synapse/storage/data_stores/main/devices.py +++ b/synapse/storage/data_stores/main/devices.py @@ -112,23 +112,13 @@ class DeviceWorkerStore(SQLBaseStore): if not has_changed: return now_stream_id, [] - # We retrieve n+1 devices from the list of outbound pokes where n is - # our outbound device update limit. We then check if the very last - # device has the same stream_id as the second-to-last device. If so, - # then we ignore all devices with that stream_id and only send the - # devices with a lower stream_id. - # - # If when culling the list we end up with no devices afterwards, we - # consider the device update to be too large, and simply skip the - # stream_id; the rationale being that such a large device list update - # is likely an error. updates = yield self.db.runInteraction( "get_device_updates_by_remote", self._get_device_updates_by_remote_txn, destination, from_stream_id, now_stream_id, - limit + 1, + limit, ) # Return an empty list if there are no updates @@ -166,14 +156,6 @@ class DeviceWorkerStore(SQLBaseStore): "device_id": verify_key.version, } - # if we have exceeded the limit, we need to exclude any results with the - # same stream_id as the last row. - if len(updates) > limit: - stream_id_cutoff = updates[-1][2] - now_stream_id = stream_id_cutoff - 1 - else: - stream_id_cutoff = None - # Perform the equivalent of a GROUP BY # # Iterate through the updates list and copy non-duplicate @@ -192,10 +174,6 @@ class DeviceWorkerStore(SQLBaseStore): query_map = {} cross_signing_keys_by_user = {} for user_id, device_id, update_stream_id, update_context in updates: - if stream_id_cutoff is not None and update_stream_id >= stream_id_cutoff: - # Stop processing updates - break - if ( user_id in master_key_by_user and device_id == master_key_by_user[user_id]["device_id"] @@ -218,17 +196,6 @@ class DeviceWorkerStore(SQLBaseStore): if update_stream_id > previous_update_stream_id: query_map[key] = (update_stream_id, update_context) - # If we didn't find any updates with a stream_id lower than the cutoff, it - # means that there are more than limit updates all of which have the same - # steam_id. - - # That should only happen if a client is spamming the server with new - # devices, in which case E2E isn't going to work well anyway. We'll just - # skip that stream_id and return an empty list, and continue with the next - # stream_id next time. - if not query_map and not cross_signing_keys_by_user: - return stream_id_cutoff, [] - results = yield self._get_device_update_edus_by_remote( destination, from_stream_id, query_map ) diff --git a/tests/storage/test_devices.py b/tests/storage/test_devices.py index 6f8d990959..c2539b353a 100644 --- a/tests/storage/test_devices.py +++ b/tests/storage/test_devices.py @@ -88,51 +88,6 @@ class DeviceStoreTestCase(tests.unittest.TestCase): # Check original device_ids are contained within these updates self._check_devices_in_updates(device_ids, device_updates) - @defer.inlineCallbacks - def test_get_device_updates_by_remote_limited(self): - # Test breaking the update limit in 1, 101, and 1 device_id segments - - # first add one device - device_ids1 = ["device_id0"] - yield self.store.add_device_change_to_streams( - "user_id", device_ids1, ["someotherhost"] - ) - - # then add 101 - device_ids2 = ["device_id" + str(i + 1) for i in range(101)] - yield self.store.add_device_change_to_streams( - "user_id", device_ids2, ["someotherhost"] - ) - - # then one more - device_ids3 = ["newdevice"] - yield self.store.add_device_change_to_streams( - "user_id", device_ids3, ["someotherhost"] - ) - - # - # now read them back. - # - - # first we should get a single update - now_stream_id, device_updates = yield self.store.get_device_updates_by_remote( - "someotherhost", -1, limit=100 - ) - self._check_devices_in_updates(device_ids1, device_updates) - - # Then we should get an empty list back as the 101 devices broke the limit - now_stream_id, device_updates = yield self.store.get_device_updates_by_remote( - "someotherhost", now_stream_id, limit=100 - ) - self.assertEqual(len(device_updates), 0) - - # The 101 devices should've been cleared, so we should now just get one device - # update - now_stream_id, device_updates = yield self.store.get_device_updates_by_remote( - "someotherhost", now_stream_id, limit=100 - ) - self._check_devices_in_updates(device_ids3, device_updates) - def _check_devices_in_updates(self, expected_device_ids, device_updates): """Check that an specific device ids exist in a list of device update EDUs""" self.assertEqual(len(device_updates), len(expected_device_ids)) -- cgit 1.4.1 From bbeee33d63c43cb80118c0dccf8abd9d4ac1b8f3 Mon Sep 17 00:00:00 2001 From: Dirk Klimpel <5740567+dklimpel@users.noreply.github.com> Date: Fri, 28 Feb 2020 10:58:05 +0100 Subject: Fixed set a user as an admin with the new API (#6928) Fix #6910 --- changelog.d/6910.bugfix | 1 + synapse/rest/admin/users.py | 6 +- synapse/storage/data_stores/main/registration.py | 16 +- tests/rest/admin/test_user.py | 209 ++++++++++++++++++++--- 4 files changed, 194 insertions(+), 38 deletions(-) create mode 100644 changelog.d/6910.bugfix (limited to 'tests') diff --git a/changelog.d/6910.bugfix b/changelog.d/6910.bugfix new file mode 100644 index 0000000000..707f1ff7b5 --- /dev/null +++ b/changelog.d/6910.bugfix @@ -0,0 +1 @@ +Fixed set a user as an admin with the admin API `PUT /_synapse/admin/v2/users/`. Contributed by @dklimpel. diff --git a/synapse/rest/admin/users.py b/synapse/rest/admin/users.py index 2107b5dc56..064908fbb0 100644 --- a/synapse/rest/admin/users.py +++ b/synapse/rest/admin/users.py @@ -211,9 +211,7 @@ class UserRestServletV2(RestServlet): if target_user == auth_user and not set_admin_to: raise SynapseError(400, "You may not demote yourself.") - await self.admin_handler.set_user_server_admin( - target_user, set_admin_to - ) + await self.store.set_server_admin(target_user, set_admin_to) if "password" in body: if ( @@ -648,6 +646,6 @@ class UserAdminServlet(RestServlet): if target_user == auth_user and not set_admin_to: raise SynapseError(400, "You may not demote yourself.") - await self.store.set_user_server_admin(target_user, set_admin_to) + await self.store.set_server_admin(target_user, set_admin_to) return 200, {} diff --git a/synapse/storage/data_stores/main/registration.py b/synapse/storage/data_stores/main/registration.py index 49306642ed..3e53c8568a 100644 --- a/synapse/storage/data_stores/main/registration.py +++ b/synapse/storage/data_stores/main/registration.py @@ -301,12 +301,16 @@ class RegistrationWorkerStore(SQLBaseStore): admin (bool): true iff the user is to be a server admin, false otherwise. """ - return self.db.simple_update_one( - table="users", - keyvalues={"name": user.to_string()}, - updatevalues={"admin": 1 if admin else 0}, - desc="set_server_admin", - ) + + def set_server_admin_txn(txn): + self.db.simple_update_one_txn( + txn, "users", {"name": user.to_string()}, {"admin": 1 if admin else 0} + ) + self._invalidate_cache_and_stream( + txn, self.get_user_by_id, (user.to_string(),) + ) + + return self.db.runInteraction("set_server_admin", set_server_admin_txn) def _query_for_auth(self, txn, token): sql = ( diff --git a/tests/rest/admin/test_user.py b/tests/rest/admin/test_user.py index 490ce8f55d..70688c2494 100644 --- a/tests/rest/admin/test_user.py +++ b/tests/rest/admin/test_user.py @@ -16,6 +16,7 @@ import hashlib import hmac import json +import urllib.parse from mock import Mock @@ -371,22 +372,24 @@ class UserRestTestCase(unittest.HomeserverTestCase): def prepare(self, reactor, clock, hs): self.store = hs.get_datastore() - self.url = "/_synapse/admin/v2/users/@bob:test" - self.admin_user = self.register_user("admin", "pass", admin=True) self.admin_user_tok = self.login("admin", "pass") self.other_user = self.register_user("user", "pass") self.other_user_token = self.login("user", "pass") + self.url_other_user = "/_synapse/admin/v2/users/%s" % urllib.parse.quote( + self.other_user + ) def test_requester_is_no_admin(self): """ If the user is not a server admin, an error is returned. """ self.hs.config.registration_shared_secret = None + url = "/_synapse/admin/v2/users/@bob:test" request, channel = self.make_request( - "GET", self.url, access_token=self.other_user_token, + "GET", url, access_token=self.other_user_token, ) self.render(request) @@ -394,7 +397,7 @@ class UserRestTestCase(unittest.HomeserverTestCase): self.assertEqual("You are not a server admin", channel.json_body["error"]) request, channel = self.make_request( - "PUT", self.url, access_token=self.other_user_token, content=b"{}", + "PUT", url, access_token=self.other_user_token, content=b"{}", ) self.render(request) @@ -417,24 +420,73 @@ class UserRestTestCase(unittest.HomeserverTestCase): self.assertEqual(404, channel.code, msg=channel.json_body) self.assertEqual("M_NOT_FOUND", channel.json_body["errcode"]) - def test_requester_is_admin(self): + def test_create_server_admin(self): """ - If the user is a server admin, a new user is created. + Check that a new admin user is created successfully. """ self.hs.config.registration_shared_secret = None + url = "/_synapse/admin/v2/users/@bob:test" + # Create user (server admin) body = json.dumps( { "password": "abc123", "admin": True, + "displayname": "Bob's name", "threepids": [{"medium": "email", "address": "bob@bob.bob"}], } ) + request, channel = self.make_request( + "PUT", + url, + access_token=self.admin_user_tok, + content=body.encode(encoding="utf_8"), + ) + self.render(request) + + self.assertEqual(201, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@bob:test", channel.json_body["name"]) + self.assertEqual("Bob's name", channel.json_body["displayname"]) + self.assertEqual("email", channel.json_body["threepids"][0]["medium"]) + self.assertEqual("bob@bob.bob", channel.json_body["threepids"][0]["address"]) + self.assertEqual(True, channel.json_body["admin"]) + + # Get user + request, channel = self.make_request( + "GET", url, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@bob:test", channel.json_body["name"]) + self.assertEqual("Bob's name", channel.json_body["displayname"]) + self.assertEqual("email", channel.json_body["threepids"][0]["medium"]) + self.assertEqual("bob@bob.bob", channel.json_body["threepids"][0]["address"]) + self.assertEqual(True, channel.json_body["admin"]) + self.assertEqual(False, channel.json_body["is_guest"]) + self.assertEqual(False, channel.json_body["deactivated"]) + + def test_create_user(self): + """ + Check that a new regular user is created successfully. + """ + self.hs.config.registration_shared_secret = None + url = "/_synapse/admin/v2/users/@bob:test" + # Create user + body = json.dumps( + { + "password": "abc123", + "admin": False, + "displayname": "Bob's name", + "threepids": [{"medium": "email", "address": "bob@bob.bob"}], + } + ) + request, channel = self.make_request( "PUT", - self.url, + url, access_token=self.admin_user_tok, content=body.encode(encoding="utf_8"), ) @@ -442,29 +494,38 @@ class UserRestTestCase(unittest.HomeserverTestCase): self.assertEqual(201, int(channel.result["code"]), msg=channel.result["body"]) self.assertEqual("@bob:test", channel.json_body["name"]) - self.assertEqual("bob", channel.json_body["displayname"]) + self.assertEqual("Bob's name", channel.json_body["displayname"]) self.assertEqual("email", channel.json_body["threepids"][0]["medium"]) self.assertEqual("bob@bob.bob", channel.json_body["threepids"][0]["address"]) + self.assertEqual(False, channel.json_body["admin"]) # Get user request, channel = self.make_request( - "GET", self.url, access_token=self.admin_user_tok, + "GET", url, access_token=self.admin_user_tok, ) self.render(request) self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) self.assertEqual("@bob:test", channel.json_body["name"]) - self.assertEqual("bob", channel.json_body["displayname"]) - self.assertEqual(1, channel.json_body["admin"]) - self.assertEqual(0, channel.json_body["is_guest"]) - self.assertEqual(0, channel.json_body["deactivated"]) + self.assertEqual("Bob's name", channel.json_body["displayname"]) + self.assertEqual("email", channel.json_body["threepids"][0]["medium"]) + self.assertEqual("bob@bob.bob", channel.json_body["threepids"][0]["address"]) + self.assertEqual(False, channel.json_body["admin"]) + self.assertEqual(False, channel.json_body["is_guest"]) + self.assertEqual(False, channel.json_body["deactivated"]) + + def test_set_password(self): + """ + Test setting a new password for another user. + """ + self.hs.config.registration_shared_secret = None # Change password body = json.dumps({"password": "hahaha"}) request, channel = self.make_request( "PUT", - self.url, + self.url_other_user, access_token=self.admin_user_tok, content=body.encode(encoding="utf_8"), ) @@ -472,38 +533,130 @@ class UserRestTestCase(unittest.HomeserverTestCase): self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + def test_set_displayname(self): + """ + Test setting the displayname of another user. + """ + self.hs.config.registration_shared_secret = None + # Modify user + body = json.dumps({"displayname": "foobar"}) + + request, channel = self.make_request( + "PUT", + self.url_other_user, + access_token=self.admin_user_tok, + content=body.encode(encoding="utf_8"), + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@user:test", channel.json_body["name"]) + self.assertEqual("foobar", channel.json_body["displayname"]) + + # Get user + request, channel = self.make_request( + "GET", self.url_other_user, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@user:test", channel.json_body["name"]) + self.assertEqual("foobar", channel.json_body["displayname"]) + + def test_set_threepid(self): + """ + Test setting threepid for an other user. + """ + self.hs.config.registration_shared_secret = None + + # Delete old and add new threepid to user body = json.dumps( - { - "displayname": "foobar", - "deactivated": True, - "threepids": [{"medium": "email", "address": "bob2@bob.bob"}], - } + {"threepids": [{"medium": "email", "address": "bob3@bob.bob"}]} ) request, channel = self.make_request( "PUT", - self.url, + self.url_other_user, access_token=self.admin_user_tok, content=body.encode(encoding="utf_8"), ) self.render(request) self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual("@bob:test", channel.json_body["name"]) - self.assertEqual("foobar", channel.json_body["displayname"]) + self.assertEqual("@user:test", channel.json_body["name"]) + self.assertEqual("email", channel.json_body["threepids"][0]["medium"]) + self.assertEqual("bob3@bob.bob", channel.json_body["threepids"][0]["address"]) + + # Get user + request, channel = self.make_request( + "GET", self.url_other_user, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@user:test", channel.json_body["name"]) + self.assertEqual("email", channel.json_body["threepids"][0]["medium"]) + self.assertEqual("bob3@bob.bob", channel.json_body["threepids"][0]["address"]) + + def test_deactivate_user(self): + """ + Test deactivating another user. + """ + + # Deactivate user + body = json.dumps({"deactivated": True}) + + request, channel = self.make_request( + "PUT", + self.url_other_user, + access_token=self.admin_user_tok, + content=body.encode(encoding="utf_8"), + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@user:test", channel.json_body["name"]) self.assertEqual(True, channel.json_body["deactivated"]) # the user is deactivated, the threepid will be deleted # Get user request, channel = self.make_request( - "GET", self.url, access_token=self.admin_user_tok, + "GET", self.url_other_user, access_token=self.admin_user_tok, ) self.render(request) self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual("@bob:test", channel.json_body["name"]) - self.assertEqual("foobar", channel.json_body["displayname"]) - self.assertEqual(1, channel.json_body["admin"]) - self.assertEqual(0, channel.json_body["is_guest"]) - self.assertEqual(1, channel.json_body["deactivated"]) + self.assertEqual("@user:test", channel.json_body["name"]) + self.assertEqual(True, channel.json_body["deactivated"]) + + def test_set_user_as_admin(self): + """ + Test setting the admin flag on a user. + """ + self.hs.config.registration_shared_secret = None + + # Set a user as an admin + body = json.dumps({"admin": True}) + + request, channel = self.make_request( + "PUT", + self.url_other_user, + access_token=self.admin_user_tok, + content=body.encode(encoding="utf_8"), + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@user:test", channel.json_body["name"]) + self.assertEqual(True, channel.json_body["admin"]) + + # Get user + request, channel = self.make_request( + "GET", self.url_other_user, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@user:test", channel.json_body["name"]) + self.assertEqual(True, channel.json_body["admin"]) -- cgit 1.4.1 From b2bd54a2e31d9a248f73fadb184ae9b4cbdb49f9 Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Mon, 2 Mar 2020 16:36:32 +0000 Subject: Add a confirmation step to the SSO login flow --- docs/sample_config.yaml | 34 ++++++++++ synapse/config/_base.pyi | 2 + synapse/config/homeserver.py | 2 + synapse/config/sso.py | 74 +++++++++++++++++++++ synapse/res/templates/sso_redirect_confirm.html | 14 ++++ synapse/rest/client/v1/login.py | 40 ++++++++++-- tests/rest/client/v1/test_login.py | 85 +++++++++++++++++++++++++ 7 files changed, 245 insertions(+), 6 deletions(-) create mode 100644 synapse/config/sso.py create mode 100644 synapse/res/templates/sso_redirect_confirm.html (limited to 'tests') diff --git a/docs/sample_config.yaml b/docs/sample_config.yaml index 8a036071e1..bbb8a4d934 100644 --- a/docs/sample_config.yaml +++ b/docs/sample_config.yaml @@ -1360,6 +1360,40 @@ saml2_config: # # name: value +# Additional settings to use with single-sign on systems such as SAML2 and CAS. +# +sso: + # Directory in which Synapse will try to find the template files below. + # If not set, default templates from within the Synapse package will be used. + # + # DO NOT UNCOMMENT THIS SETTING unless you want to customise the templates. + # If you *do* uncomment it, you will need to make sure that all the templates + # below are in the directory. + # + # Synapse will look for the following templates in this directory: + # + # * HTML page for confirmation of redirect during authentication: + # 'sso_redirect_confirm.html'. + # + # When rendering, this template is given three variables: + # * redirect_url: the URL the user is about to be redirected to. Needs + # manual escaping (see + # https://jinja.palletsprojects.com/en/2.11.x/templates/#html-escaping). + # + # * display_url: the same as `redirect_url`, but with the query + # parameters stripped. The intention is to have a + # human-readable URL to show to users, not to use it as + # the final address to redirect to. Needs manual escaping + # (see https://jinja.palletsprojects.com/en/2.11.x/templates/#html-escaping). + # + # * server_name: the homeserver's name. + # + # You can see the default templates at: + # https://github.com/matrix-org/synapse/tree/master/synapse/res/templates + # + #template_dir: "res/templates" + + # The JWT needs to contain a globally unique "sub" (subject) claim. # #jwt_config: diff --git a/synapse/config/_base.pyi b/synapse/config/_base.pyi index 86bc965ee4..3053fc9d27 100644 --- a/synapse/config/_base.pyi +++ b/synapse/config/_base.pyi @@ -24,6 +24,7 @@ from synapse.config import ( server, server_notices_config, spam_checker, + sso, stats, third_party_event_rules, tls, @@ -57,6 +58,7 @@ class RootConfig: key: key.KeyConfig saml2: saml2_config.SAML2Config cas: cas.CasConfig + sso: sso.SSOConfig jwt: jwt_config.JWTConfig password: password.PasswordConfig email: emailconfig.EmailConfig diff --git a/synapse/config/homeserver.py b/synapse/config/homeserver.py index 6e348671c7..b4bca08b20 100644 --- a/synapse/config/homeserver.py +++ b/synapse/config/homeserver.py @@ -38,6 +38,7 @@ from .saml2_config import SAML2Config from .server import ServerConfig from .server_notices_config import ServerNoticesConfig from .spam_checker import SpamCheckerConfig +from .sso import SSOConfig from .stats import StatsConfig from .third_party_event_rules import ThirdPartyRulesConfig from .tls import TlsConfig @@ -65,6 +66,7 @@ class HomeServerConfig(RootConfig): KeyConfig, SAML2Config, CasConfig, + SSOConfig, JWTConfig, PasswordConfig, EmailConfig, diff --git a/synapse/config/sso.py b/synapse/config/sso.py new file mode 100644 index 0000000000..f426b65b4f --- /dev/null +++ b/synapse/config/sso.py @@ -0,0 +1,74 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +from typing import Any, Dict + +import pkg_resources + +from ._base import Config, ConfigError + + +class SSOConfig(Config): + """SSO Configuration + """ + + section = "sso" + + def read_config(self, config, **kwargs): + sso_config = config.get("sso") or {} # type: Dict[str, Any] + + # Pick a template directory in order of: + # * The sso-specific template_dir + # * /path/to/synapse/install/res/templates + template_dir = sso_config.get("template_dir") + if not template_dir: + template_dir = pkg_resources.resource_filename("synapse", "res/templates",) + + self.sso_redirect_confirm_template_dir = template_dir + + def generate_config_section(self, **kwargs): + return """\ + # Additional settings to use with single-sign on systems such as SAML2 and CAS. + # + sso: + # Directory in which Synapse will try to find the template files below. + # If not set, default templates from within the Synapse package will be used. + # + # DO NOT UNCOMMENT THIS SETTING unless you want to customise the templates. + # If you *do* uncomment it, you will need to make sure that all the templates + # below are in the directory. + # + # Synapse will look for the following templates in this directory: + # + # * HTML page for a confirmation step before redirecting back to the client + # with the login token: 'sso_redirect_confirm.html'. + # + # When rendering, this template is given three variables: + # * redirect_url: the URL the user is about to be redirected to. Needs + # manual escaping (see + # https://jinja.palletsprojects.com/en/2.11.x/templates/#html-escaping). + # + # * display_url: the same as `redirect_url`, but with the query + # parameters stripped. The intention is to have a + # human-readable URL to show to users, not to use it as + # the final address to redirect to. Needs manual escaping + # (see https://jinja.palletsprojects.com/en/2.11.x/templates/#html-escaping). + # + # * server_name: the homeserver's name. + # + # You can see the default templates at: + # https://github.com/matrix-org/synapse/tree/master/synapse/res/templates + # + #template_dir: "res/templates" + """ diff --git a/synapse/res/templates/sso_redirect_confirm.html b/synapse/res/templates/sso_redirect_confirm.html new file mode 100644 index 0000000000..20a15e1e74 --- /dev/null +++ b/synapse/res/templates/sso_redirect_confirm.html @@ -0,0 +1,14 @@ + + + + + SSO redirect confirmation + + +

The application at {{ display_url | e }} is requesting full access to your {{ server_name }} Matrix account.

+

If you don't recognise this address, you should ignore this and close this tab.

+

+ I trust this address +

+ + \ No newline at end of file diff --git a/synapse/rest/client/v1/login.py b/synapse/rest/client/v1/login.py index 1294e080dc..1acfd01d8e 100644 --- a/synapse/rest/client/v1/login.py +++ b/synapse/rest/client/v1/login.py @@ -29,6 +29,7 @@ from synapse.http.servlet import ( parse_string, ) from synapse.http.site import SynapseRequest +from synapse.push.mailer import load_jinja2_templates from synapse.rest.client.v2_alpha._base import client_patterns from synapse.rest.well_known import WellKnownBuilder from synapse.types import UserID, map_username_to_mxid_localpart @@ -548,6 +549,13 @@ class SSOAuthHandler(object): self._registration_handler = hs.get_registration_handler() self._macaroon_gen = hs.get_macaroon_generator() + # Load the redirect page HTML template + self._template = load_jinja2_templates( + hs.config.sso_redirect_confirm_template_dir, ["sso_redirect_confirm.html"], + )[0] + + self._server_name = hs.config.server_name + async def on_successful_auth( self, username, request, client_redirect_url, user_display_name=None ): @@ -592,21 +600,41 @@ class SSOAuthHandler(object): request: client_redirect_url: """ - + # Create a login token login_token = self._macaroon_gen.generate_short_term_login_token( registered_user_id ) - redirect_url = self._add_login_token_to_redirect_url( - client_redirect_url, login_token + + # Remove the query parameters from the redirect URL to get a shorter version of + # it. This is only to display a human-readable URL in the template, but not the + # URL we redirect users to. + redirect_url_no_params = client_redirect_url.split("?")[0] + + # Append the login token to the original redirect URL (i.e. with its query + # parameters kept intact) to build the URL to which the template needs to + # redirect the users once they have clicked on the confirmation link. + redirect_url = self._add_query_param_to_url( + client_redirect_url, "loginToken", login_token + ) + + # Serve the redirect confirmation page + html = self._template.render( + display_url=redirect_url_no_params, + redirect_url=redirect_url, + server_name=self._server_name, ) - request.redirect(redirect_url) + + request.setResponseCode(200) + request.setHeader(b"Content-Type", b"text/html; charset=utf-8") + request.setHeader(b"Content-Length", b"%d" % (len(html),)) + request.write(html.encode("utf8")) finish_request(request) @staticmethod - def _add_login_token_to_redirect_url(url, token): + def _add_query_param_to_url(url, param_name, param): url_parts = list(urllib.parse.urlparse(url)) query = dict(urllib.parse.parse_qsl(url_parts[4])) - query.update({"loginToken": token}) + query.update({param_name: param}) url_parts[4] = urllib.parse.urlencode(query) return urllib.parse.urlunparse(url_parts) diff --git a/tests/rest/client/v1/test_login.py b/tests/rest/client/v1/test_login.py index eae5411325..2b8ad5c753 100644 --- a/tests/rest/client/v1/test_login.py +++ b/tests/rest/client/v1/test_login.py @@ -1,4 +1,7 @@ import json +import urllib.parse + +from mock import Mock import synapse.rest.admin from synapse.rest.client.v1 import login @@ -252,3 +255,85 @@ class LoginRestServletTestCase(unittest.HomeserverTestCase): ) self.render(request) self.assertEquals(channel.code, 200, channel.result) + + +class CASRedirectConfirmTestCase(unittest.HomeserverTestCase): + + servlets = [ + login.register_servlets, + ] + + def make_homeserver(self, reactor, clock): + self.base_url = "https://matrix.goodserver.com/" + self.redirect_path = "_synapse/client/login/sso/redirect/confirm" + + config = self.default_config() + config["enable_registration"] = True + config["cas_config"] = { + "enabled": True, + "server_url": "https://fake.test", + "service_url": "https://matrix.goodserver.com:8448", + } + config["public_baseurl"] = self.base_url + + async def get_raw(uri, args): + """Return an example response payload from a call to the `/proxyValidate` + endpoint of a CAS server, copied from + https://apereo.github.io/cas/5.0.x/protocol/CAS-Protocol-V2-Specification.html#26-proxyvalidate-cas-20 + + This needs to be returned by an async function (as opposed to set as the + mock's return value) because the corresponding Synapse code awaits on it. + """ + return """ + + + username + PGTIOU-84678-8a9d... + + https://proxy2/pgtUrl + https://proxy1/pgtUrl + + + + """ + + mocked_http_client = Mock(spec=["get_raw"]) + mocked_http_client.get_raw.side_effect = get_raw + + self.hs = self.setup_test_homeserver( + config=config, proxied_http_client=mocked_http_client, + ) + + return self.hs + + def test_cas_redirect_confirm(self): + """Tests that the SSO login flow serves a confirmation page before redirecting a + user to the redirect URL. + """ + base_url = "/login/cas/ticket?redirectUrl" + redirect_url = "https://dodgy-site.com/" + + url_parts = list(urllib.parse.urlparse(base_url)) + query = dict(urllib.parse.parse_qsl(url_parts[4])) + query.update({"redirectUrl": redirect_url}) + query.update({"ticket": "ticket"}) + url_parts[4] = urllib.parse.urlencode(query) + cas_ticket_url = urllib.parse.urlunparse(url_parts) + + # Get Synapse to call the fake CAS and serve the template. + request, channel = self.make_request("GET", cas_ticket_url) + self.render(request) + + # Test that the response is HTML. + content_type_header_value = "" + for header in channel.result.get("headers", []): + if header[0] == b"Content-Type": + content_type_header_value = header[1].decode("utf8") + + self.assertTrue(content_type_header_value.startswith("text/html")) + + # Test that the body isn't empty. + self.assertTrue(len(channel.result["body"]) > 0) + + # And that it contains our redirect link + self.assertIn(redirect_url, channel.result["body"].decode("UTF-8")) -- cgit 1.4.1 From b68041df3dcbcf3ca04c500d1712aa22a3c2580c Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Mon, 2 Mar 2020 17:05:09 +0000 Subject: Add a whitelist for the SSO confirmation step. --- docs/sample_config.yaml | 22 +++++++++++++++++++--- synapse/config/sso.py | 18 ++++++++++++++++++ synapse/rest/client/v1/login.py | 26 ++++++++++++++++++-------- tests/rest/client/v1/test_login.py | 32 +++++++++++++++++++++++++++++--- 4 files changed, 84 insertions(+), 14 deletions(-) (limited to 'tests') diff --git a/docs/sample_config.yaml b/docs/sample_config.yaml index bbb8a4d934..f719ec696f 100644 --- a/docs/sample_config.yaml +++ b/docs/sample_config.yaml @@ -1363,6 +1363,22 @@ saml2_config: # Additional settings to use with single-sign on systems such as SAML2 and CAS. # sso: + # A list of client URLs which are whitelisted so that the user does not + # have to confirm giving access to their account to the URL. Any client + # whose URL starts with an entry in the following list will not be subject + # to an additional confirmation step after the SSO login is completed. + # + # WARNING: An entry such as "https://my.client" is insecure, because it + # will also match "https://my.client.evil.site", exposing your users to + # phishing attacks from evil.site. To avoid this, include a slash after the + # hostname: "https://my.client/". + # + # By default, this list is empty. + # + #client_whitelist: + # - https://riot.im/develop + # - https://my.custom.client/ + # Directory in which Synapse will try to find the template files below. # If not set, default templates from within the Synapse package will be used. # @@ -1372,8 +1388,8 @@ sso: # # Synapse will look for the following templates in this directory: # - # * HTML page for confirmation of redirect during authentication: - # 'sso_redirect_confirm.html'. + # * HTML page for a confirmation step before redirecting back to the client + # with the login token: 'sso_redirect_confirm.html'. # # When rendering, this template is given three variables: # * redirect_url: the URL the user is about to be redirected to. Needs @@ -1381,7 +1397,7 @@ sso: # https://jinja.palletsprojects.com/en/2.11.x/templates/#html-escaping). # # * display_url: the same as `redirect_url`, but with the query - # parameters stripped. The intention is to have a + # parameters stripped. The intention is to have a # human-readable URL to show to users, not to use it as # the final address to redirect to. Needs manual escaping # (see https://jinja.palletsprojects.com/en/2.11.x/templates/#html-escaping). diff --git a/synapse/config/sso.py b/synapse/config/sso.py index f426b65b4f..56299bd4e4 100644 --- a/synapse/config/sso.py +++ b/synapse/config/sso.py @@ -37,11 +37,29 @@ class SSOConfig(Config): self.sso_redirect_confirm_template_dir = template_dir + self.sso_client_whitelist = sso_config.get("client_whitelist") or [] + def generate_config_section(self, **kwargs): return """\ # Additional settings to use with single-sign on systems such as SAML2 and CAS. # sso: + # A list of client URLs which are whitelisted so that the user does not + # have to confirm giving access to their account to the URL. Any client + # whose URL starts with an entry in the following list will not be subject + # to an additional confirmation step after the SSO login is completed. + # + # WARNING: An entry such as "https://my.client" is insecure, because it + # will also match "https://my.client.evil.site", exposing your users to + # phishing attacks from evil.site. To avoid this, include a slash after the + # hostname: "https://my.client/". + # + # By default, this list is empty. + # + #client_whitelist: + # - https://riot.im/develop + # - https://my.custom.client/ + # Directory in which Synapse will try to find the template files below. # If not set, default templates from within the Synapse package will be used. # diff --git a/synapse/rest/client/v1/login.py b/synapse/rest/client/v1/login.py index 1acfd01d8e..b2bc7537db 100644 --- a/synapse/rest/client/v1/login.py +++ b/synapse/rest/client/v1/login.py @@ -556,6 +556,9 @@ class SSOAuthHandler(object): self._server_name = hs.config.server_name + # cast to tuple for use with str.startswith + self._whitelisted_sso_clients = tuple(hs.config.sso_client_whitelist) + async def on_successful_auth( self, username, request, client_redirect_url, user_display_name=None ): @@ -605,11 +608,6 @@ class SSOAuthHandler(object): registered_user_id ) - # Remove the query parameters from the redirect URL to get a shorter version of - # it. This is only to display a human-readable URL in the template, but not the - # URL we redirect users to. - redirect_url_no_params = client_redirect_url.split("?")[0] - # Append the login token to the original redirect URL (i.e. with its query # parameters kept intact) to build the URL to which the template needs to # redirect the users once they have clicked on the confirmation link. @@ -617,17 +615,29 @@ class SSOAuthHandler(object): client_redirect_url, "loginToken", login_token ) - # Serve the redirect confirmation page + # if the client is whitelisted, we can redirect straight to it + if client_redirect_url.startswith(self._whitelisted_sso_clients): + request.redirect(redirect_url) + finish_request(request) + return + + # Otherwise, serve the redirect confirmation page. + + # Remove the query parameters from the redirect URL to get a shorter version of + # it. This is only to display a human-readable URL in the template, but not the + # URL we redirect users to. + redirect_url_no_params = client_redirect_url.split("?")[0] + html = self._template.render( display_url=redirect_url_no_params, redirect_url=redirect_url, server_name=self._server_name, - ) + ).encode("utf-8") request.setResponseCode(200) request.setHeader(b"Content-Type", b"text/html; charset=utf-8") request.setHeader(b"Content-Length", b"%d" % (len(html),)) - request.write(html.encode("utf8")) + request.write(html) finish_request(request) @staticmethod diff --git a/tests/rest/client/v1/test_login.py b/tests/rest/client/v1/test_login.py index 2b8ad5c753..da2c9bfa1e 100644 --- a/tests/rest/client/v1/test_login.py +++ b/tests/rest/client/v1/test_login.py @@ -268,13 +268,11 @@ class CASRedirectConfirmTestCase(unittest.HomeserverTestCase): self.redirect_path = "_synapse/client/login/sso/redirect/confirm" config = self.default_config() - config["enable_registration"] = True config["cas_config"] = { "enabled": True, "server_url": "https://fake.test", "service_url": "https://matrix.goodserver.com:8448", } - config["public_baseurl"] = self.base_url async def get_raw(uri, args): """Return an example response payload from a call to the `/proxyValidate` @@ -310,7 +308,7 @@ class CASRedirectConfirmTestCase(unittest.HomeserverTestCase): """Tests that the SSO login flow serves a confirmation page before redirecting a user to the redirect URL. """ - base_url = "/login/cas/ticket?redirectUrl" + base_url = "/_matrix/client/r0/login/cas/ticket?redirectUrl" redirect_url = "https://dodgy-site.com/" url_parts = list(urllib.parse.urlparse(base_url)) @@ -325,6 +323,7 @@ class CASRedirectConfirmTestCase(unittest.HomeserverTestCase): self.render(request) # Test that the response is HTML. + self.assertEqual(channel.code, 200) content_type_header_value = "" for header in channel.result.get("headers", []): if header[0] == b"Content-Type": @@ -337,3 +336,30 @@ class CASRedirectConfirmTestCase(unittest.HomeserverTestCase): # And that it contains our redirect link self.assertIn(redirect_url, channel.result["body"].decode("UTF-8")) + + @override_config( + { + "sso": { + "client_whitelist": [ + "https://legit-site.com/", + "https://other-site.com/", + ] + } + } + ) + def test_cas_redirect_whitelisted(self): + """Tests that the SSO login flow serves a redirect to a whitelisted url + """ + redirect_url = "https://legit-site.com/" + cas_ticket_url = ( + "/_matrix/client/r0/login/cas/ticket?redirectUrl=%s&ticket=ticket" + % (urllib.parse.quote(redirect_url)) + ) + + # Get Synapse to call the fake CAS and serve the template. + request, channel = self.make_request("GET", cas_ticket_url) + self.render(request) + + self.assertEqual(channel.code, 302) + location_headers = channel.headers.getRawHeaders("Location") + self.assertEqual(location_headers[0][: len(redirect_url)], redirect_url) -- cgit 1.4.1 From 7dcbc33a1be04c46b930699c03c15bc759f4b22c Mon Sep 17 00:00:00 2001 From: Patrick Cloke Date: Tue, 3 Mar 2020 07:12:45 -0500 Subject: Validate the alt_aliases property of canonical alias events (#6971) --- changelog.d/6971.feature | 1 + synapse/api/errors.py | 1 + synapse/handlers/directory.py | 14 ++-- synapse/handlers/message.py | 47 ++++++++++- synapse/types.py | 15 ++-- tests/handlers/test_directory.py | 66 +++++++-------- tests/rest/client/v1/test_rooms.py | 160 +++++++++++++++++++++++++++++++++++++ tests/test_types.py | 2 +- 8 files changed, 254 insertions(+), 52 deletions(-) create mode 100644 changelog.d/6971.feature (limited to 'tests') diff --git a/changelog.d/6971.feature b/changelog.d/6971.feature new file mode 100644 index 0000000000..ccf02a61df --- /dev/null +++ b/changelog.d/6971.feature @@ -0,0 +1 @@ +Validate the alt_aliases property of canonical alias events. diff --git a/synapse/api/errors.py b/synapse/api/errors.py index 0c20601600..616942b057 100644 --- a/synapse/api/errors.py +++ b/synapse/api/errors.py @@ -66,6 +66,7 @@ class Codes(object): EXPIRED_ACCOUNT = "ORG_MATRIX_EXPIRED_ACCOUNT" INVALID_SIGNATURE = "M_INVALID_SIGNATURE" USER_DEACTIVATED = "M_USER_DEACTIVATED" + BAD_ALIAS = "M_BAD_ALIAS" class CodeMessageException(RuntimeError): diff --git a/synapse/handlers/directory.py b/synapse/handlers/directory.py index 0b23ca919a..61eb49059b 100644 --- a/synapse/handlers/directory.py +++ b/synapse/handlers/directory.py @@ -13,8 +13,6 @@ # See the License for the specific language governing permissions and # limitations under the License. - -import collections import logging import string from typing import List @@ -307,15 +305,17 @@ class DirectoryHandler(BaseHandler): send_update = True content.pop("alias", "") - # Filter alt_aliases for the removed alias. - alt_aliases = content.pop("alt_aliases", None) - # If the aliases are not a list (or not found) do not attempt to modify - # the list. - if isinstance(alt_aliases, collections.Sequence): + # Filter the alt_aliases property for the removed alias. Note that the + # value is not modified if alt_aliases is of an unexpected form. + alt_aliases = content.get("alt_aliases") + if isinstance(alt_aliases, (list, tuple)) and alias_str in alt_aliases: send_update = True alt_aliases = [alias for alias in alt_aliases if alias != alias_str] + if alt_aliases: content["alt_aliases"] = alt_aliases + else: + del content["alt_aliases"] if send_update: yield self.event_creation_handler.create_and_send_nonmember_event( diff --git a/synapse/handlers/message.py b/synapse/handlers/message.py index a0103addd3..0c84c6cec4 100644 --- a/synapse/handlers/message.py +++ b/synapse/handlers/message.py @@ -888,19 +888,60 @@ class EventCreationHandler(object): yield self.base_handler.maybe_kick_guest_users(event, context) if event.type == EventTypes.CanonicalAlias: - # Check the alias is acually valid (at this time at least) + # Validate a newly added alias or newly added alt_aliases. + + original_alias = None + original_alt_aliases = set() + + original_event_id = event.unsigned.get("replaces_state") + if original_event_id: + original_event = yield self.store.get_event(original_event_id) + + if original_event: + original_alias = original_event.content.get("alias", None) + original_alt_aliases = original_event.content.get("alt_aliases", []) + + # Check the alias is currently valid (if it has changed). room_alias_str = event.content.get("alias", None) - if room_alias_str: + directory_handler = self.hs.get_handlers().directory_handler + if room_alias_str and room_alias_str != original_alias: room_alias = RoomAlias.from_string(room_alias_str) - directory_handler = self.hs.get_handlers().directory_handler mapping = yield directory_handler.get_association(room_alias) if mapping["room_id"] != event.room_id: raise SynapseError( 400, "Room alias %s does not point to the room" % (room_alias_str,), + Codes.BAD_ALIAS, ) + # Check that alt_aliases is the proper form. + alt_aliases = event.content.get("alt_aliases", []) + if not isinstance(alt_aliases, (list, tuple)): + raise SynapseError( + 400, "The alt_aliases property must be a list.", Codes.INVALID_PARAM + ) + + # If the old version of alt_aliases is of an unknown form, + # completely replace it. + if not isinstance(original_alt_aliases, (list, tuple)): + original_alt_aliases = [] + + # Check that each alias is currently valid. + new_alt_aliases = set(alt_aliases) - set(original_alt_aliases) + if new_alt_aliases: + for alias_str in new_alt_aliases: + room_alias = RoomAlias.from_string(alias_str) + mapping = yield directory_handler.get_association(room_alias) + + if mapping["room_id"] != event.room_id: + raise SynapseError( + 400, + "Room alias %s does not point to the room" + % (room_alias_str,), + Codes.BAD_ALIAS, + ) + federation_handler = self.hs.get_handlers().federation_handler if event.type == EventTypes.Member: diff --git a/synapse/types.py b/synapse/types.py index f3cd465735..acf60baddc 100644 --- a/synapse/types.py +++ b/synapse/types.py @@ -23,7 +23,7 @@ import attr from signedjson.key import decode_verify_key_bytes from unpaddedbase64 import decode_base64 -from synapse.api.errors import SynapseError +from synapse.api.errors import Codes, SynapseError # define a version of typing.Collection that works on python 3.5 if sys.version_info[:3] >= (3, 6, 0): @@ -166,11 +166,13 @@ class DomainSpecificString(namedtuple("DomainSpecificString", ("localpart", "dom return self @classmethod - def from_string(cls, s): + def from_string(cls, s: str): """Parse the string given by 's' into a structure object.""" if len(s) < 1 or s[0:1] != cls.SIGIL: raise SynapseError( - 400, "Expected %s string to start with '%s'" % (cls.__name__, cls.SIGIL) + 400, + "Expected %s string to start with '%s'" % (cls.__name__, cls.SIGIL), + Codes.INVALID_PARAM, ) parts = s[1:].split(":", 1) @@ -179,6 +181,7 @@ class DomainSpecificString(namedtuple("DomainSpecificString", ("localpart", "dom 400, "Expected %s of the form '%slocalname:domain'" % (cls.__name__, cls.SIGIL), + Codes.INVALID_PARAM, ) domain = parts[1] @@ -235,11 +238,13 @@ class GroupID(DomainSpecificString): def from_string(cls, s): group_id = super(GroupID, cls).from_string(s) if not group_id.localpart: - raise SynapseError(400, "Group ID cannot be empty") + raise SynapseError(400, "Group ID cannot be empty", Codes.INVALID_PARAM) if contains_invalid_mxid_characters(group_id.localpart): raise SynapseError( - 400, "Group ID can only contain characters a-z, 0-9, or '=_-./'" + 400, + "Group ID can only contain characters a-z, 0-9, or '=_-./'", + Codes.INVALID_PARAM, ) return group_id diff --git a/tests/handlers/test_directory.py b/tests/handlers/test_directory.py index 27b916aed4..3397cfa485 100644 --- a/tests/handlers/test_directory.py +++ b/tests/handlers/test_directory.py @@ -88,6 +88,7 @@ class DirectoryTestCase(unittest.HomeserverTestCase): ) def test_delete_alias_not_allowed(self): + """Removing an alias should be denied if a user does not have the proper permissions.""" room_id = "!8765qwer:test" self.get_success( self.store.create_room_alias_association(self.my_room, room_id, ["test"]) @@ -101,6 +102,7 @@ class DirectoryTestCase(unittest.HomeserverTestCase): ) def test_delete_alias(self): + """Removing an alias should work when a user does has the proper permissions.""" room_id = "!8765qwer:test" user_id = "@user:test" self.get_success( @@ -159,30 +161,42 @@ class CanonicalAliasTestCase(unittest.HomeserverTestCase): ) self.test_alias = "#test:test" - self.room_alias = RoomAlias.from_string(self.test_alias) + self.room_alias = self._add_alias(self.test_alias) + + def _add_alias(self, alias: str) -> RoomAlias: + """Add an alias to the test room.""" + room_alias = RoomAlias.from_string(alias) # Create a new alias to this room. self.get_success( self.store.create_room_alias_association( - self.room_alias, self.room_id, ["test"], self.admin_user + room_alias, self.room_id, ["test"], self.admin_user ) ) + return room_alias - def test_remove_alias(self): - """Removing an alias that is the canonical alias should remove it there too.""" - # Set this new alias as the canonical alias for this room + def _set_canonical_alias(self, content): + """Configure the canonical alias state on the room.""" self.helper.send_state( - self.room_id, - "m.room.canonical_alias", - {"alias": self.test_alias, "alt_aliases": [self.test_alias]}, - tok=self.admin_user_tok, + self.room_id, "m.room.canonical_alias", content, tok=self.admin_user_tok, ) - data = self.get_success( + def _get_canonical_alias(self): + """Get the canonical alias state of the room.""" + return self.get_success( self.state_handler.get_current_state( self.room_id, EventTypes.CanonicalAlias, "" ) ) + + def test_remove_alias(self): + """Removing an alias that is the canonical alias should remove it there too.""" + # Set this new alias as the canonical alias for this room + self._set_canonical_alias( + {"alias": self.test_alias, "alt_aliases": [self.test_alias]} + ) + + data = self._get_canonical_alias() self.assertEqual(data["content"]["alias"], self.test_alias) self.assertEqual(data["content"]["alt_aliases"], [self.test_alias]) @@ -193,11 +207,7 @@ class CanonicalAliasTestCase(unittest.HomeserverTestCase): ) ) - data = self.get_success( - self.state_handler.get_current_state( - self.room_id, EventTypes.CanonicalAlias, "" - ) - ) + data = self._get_canonical_alias() self.assertNotIn("alias", data["content"]) self.assertNotIn("alt_aliases", data["content"]) @@ -205,29 +215,17 @@ class CanonicalAliasTestCase(unittest.HomeserverTestCase): """Removing an alias listed as in alt_aliases should remove it there too.""" # Create a second alias. other_test_alias = "#test2:test" - other_room_alias = RoomAlias.from_string(other_test_alias) - self.get_success( - self.store.create_room_alias_association( - other_room_alias, self.room_id, ["test"], self.admin_user - ) - ) + other_room_alias = self._add_alias(other_test_alias) # Set the alias as the canonical alias for this room. - self.helper.send_state( - self.room_id, - "m.room.canonical_alias", + self._set_canonical_alias( { "alias": self.test_alias, "alt_aliases": [self.test_alias, other_test_alias], - }, - tok=self.admin_user_tok, + } ) - data = self.get_success( - self.state_handler.get_current_state( - self.room_id, EventTypes.CanonicalAlias, "" - ) - ) + data = self._get_canonical_alias() self.assertEqual(data["content"]["alias"], self.test_alias) self.assertEqual( data["content"]["alt_aliases"], [self.test_alias, other_test_alias] @@ -240,11 +238,7 @@ class CanonicalAliasTestCase(unittest.HomeserverTestCase): ) ) - data = self.get_success( - self.state_handler.get_current_state( - self.room_id, EventTypes.CanonicalAlias, "" - ) - ) + data = self._get_canonical_alias() self.assertEqual(data["content"]["alias"], self.test_alias) self.assertEqual(data["content"]["alt_aliases"], [self.test_alias]) diff --git a/tests/rest/client/v1/test_rooms.py b/tests/rest/client/v1/test_rooms.py index 2f3df5f88f..7dd86d0c27 100644 --- a/tests/rest/client/v1/test_rooms.py +++ b/tests/rest/client/v1/test_rooms.py @@ -1821,3 +1821,163 @@ class RoomAliasListTestCase(unittest.HomeserverTestCase): ) self.render(request) self.assertEqual(channel.code, expected_code, channel.result) + + +class RoomCanonicalAliasTestCase(unittest.HomeserverTestCase): + servlets = [ + synapse.rest.admin.register_servlets_for_client_rest_resource, + directory.register_servlets, + login.register_servlets, + room.register_servlets, + ] + + def prepare(self, reactor, clock, homeserver): + self.room_owner = self.register_user("room_owner", "test") + self.room_owner_tok = self.login("room_owner", "test") + + self.room_id = self.helper.create_room_as( + self.room_owner, tok=self.room_owner_tok + ) + + self.alias = "#alias:test" + self._set_alias_via_directory(self.alias) + + def _set_alias_via_directory(self, alias: str, expected_code: int = 200): + url = "/_matrix/client/r0/directory/room/" + alias + data = {"room_id": self.room_id} + request_data = json.dumps(data) + + request, channel = self.make_request( + "PUT", url, request_data, access_token=self.room_owner_tok + ) + self.render(request) + self.assertEqual(channel.code, expected_code, channel.result) + + def _get_canonical_alias(self, expected_code: int = 200) -> JsonDict: + """Calls the endpoint under test. returns the json response object.""" + request, channel = self.make_request( + "GET", + "rooms/%s/state/m.room.canonical_alias" % (self.room_id,), + access_token=self.room_owner_tok, + ) + self.render(request) + self.assertEqual(channel.code, expected_code, channel.result) + res = channel.json_body + self.assertIsInstance(res, dict) + return res + + def _set_canonical_alias(self, content: str, expected_code: int = 200) -> JsonDict: + """Calls the endpoint under test. returns the json response object.""" + request, channel = self.make_request( + "PUT", + "rooms/%s/state/m.room.canonical_alias" % (self.room_id,), + json.dumps(content), + access_token=self.room_owner_tok, + ) + self.render(request) + self.assertEqual(channel.code, expected_code, channel.result) + res = channel.json_body + self.assertIsInstance(res, dict) + return res + + def test_canonical_alias(self): + """Test a basic alias message.""" + # There is no canonical alias to start with. + self._get_canonical_alias(expected_code=404) + + # Create an alias. + self._set_canonical_alias({"alias": self.alias}) + + # Canonical alias now exists! + res = self._get_canonical_alias() + self.assertEqual(res, {"alias": self.alias}) + + # Now remove the alias. + self._set_canonical_alias({}) + + # There is an alias event, but it is empty. + res = self._get_canonical_alias() + self.assertEqual(res, {}) + + def test_alt_aliases(self): + """Test a canonical alias message with alt_aliases.""" + # Create an alias. + self._set_canonical_alias({"alt_aliases": [self.alias]}) + + # Canonical alias now exists! + res = self._get_canonical_alias() + self.assertEqual(res, {"alt_aliases": [self.alias]}) + + # Now remove the alt_aliases. + self._set_canonical_alias({}) + + # There is an alias event, but it is empty. + res = self._get_canonical_alias() + self.assertEqual(res, {}) + + def test_alias_alt_aliases(self): + """Test a canonical alias message with an alias and alt_aliases.""" + # Create an alias. + self._set_canonical_alias({"alias": self.alias, "alt_aliases": [self.alias]}) + + # Canonical alias now exists! + res = self._get_canonical_alias() + self.assertEqual(res, {"alias": self.alias, "alt_aliases": [self.alias]}) + + # Now remove the alias and alt_aliases. + self._set_canonical_alias({}) + + # There is an alias event, but it is empty. + res = self._get_canonical_alias() + self.assertEqual(res, {}) + + def test_partial_modify(self): + """Test removing only the alt_aliases.""" + # Create an alias. + self._set_canonical_alias({"alias": self.alias, "alt_aliases": [self.alias]}) + + # Canonical alias now exists! + res = self._get_canonical_alias() + self.assertEqual(res, {"alias": self.alias, "alt_aliases": [self.alias]}) + + # Now remove the alt_aliases. + self._set_canonical_alias({"alias": self.alias}) + + # There is an alias event, but it is empty. + res = self._get_canonical_alias() + self.assertEqual(res, {"alias": self.alias}) + + def test_add_alias(self): + """Test removing only the alt_aliases.""" + # Create an additional alias. + second_alias = "#second:test" + self._set_alias_via_directory(second_alias) + + # Add the canonical alias. + self._set_canonical_alias({"alias": self.alias, "alt_aliases": [self.alias]}) + + # Then add the second alias. + self._set_canonical_alias( + {"alias": self.alias, "alt_aliases": [self.alias, second_alias]} + ) + + # Canonical alias now exists! + res = self._get_canonical_alias() + self.assertEqual( + res, {"alias": self.alias, "alt_aliases": [self.alias, second_alias]} + ) + + def test_bad_data(self): + """Invalid data for alt_aliases should cause errors.""" + self._set_canonical_alias({"alt_aliases": "@bad:test"}, expected_code=400) + self._set_canonical_alias({"alt_aliases": None}, expected_code=400) + self._set_canonical_alias({"alt_aliases": 0}, expected_code=400) + self._set_canonical_alias({"alt_aliases": 1}, expected_code=400) + self._set_canonical_alias({"alt_aliases": False}, expected_code=400) + self._set_canonical_alias({"alt_aliases": True}, expected_code=400) + self._set_canonical_alias({"alt_aliases": {}}, expected_code=400) + + def test_bad_alias(self): + """An alias which does not point to the room raises a SynapseError.""" + self._set_canonical_alias({"alias": "@unknown:test"}, expected_code=400) + self._set_canonical_alias({"alt_aliases": ["@unknown:test"]}, expected_code=400) diff --git a/tests/test_types.py b/tests/test_types.py index 8d97c751ea..480bea1bdc 100644 --- a/tests/test_types.py +++ b/tests/test_types.py @@ -75,7 +75,7 @@ class GroupIDTestCase(unittest.TestCase): self.fail("Parsing '%s' should raise exception" % id_string) except SynapseError as exc: self.assertEqual(400, exc.code) - self.assertEqual("M_UNKNOWN", exc.errcode) + self.assertEqual("M_INVALID_PARAM", exc.errcode) class MapUsernameTestCase(unittest.TestCase): -- cgit 1.4.1 From 8ef8fb2c1c7c4aeb80fce4deea477b37754ce539 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Wed, 4 Mar 2020 13:11:04 +0000 Subject: Read the room version from database when fetching events (#6874) This is a precursor to giving EventBase objects the knowledge of which room version they belong to. --- changelog.d/6874.misc | 1 + synapse/storage/data_stores/main/events_worker.py | 84 ++++++++++++++++++----- tests/replication/slave/storage/test_events.py | 10 +++ 3 files changed, 79 insertions(+), 16 deletions(-) create mode 100644 changelog.d/6874.misc (limited to 'tests') diff --git a/changelog.d/6874.misc b/changelog.d/6874.misc new file mode 100644 index 0000000000..08aa80bcd9 --- /dev/null +++ b/changelog.d/6874.misc @@ -0,0 +1 @@ +Refactoring work in preparation for changing the event redaction algorithm. diff --git a/synapse/storage/data_stores/main/events_worker.py b/synapse/storage/data_stores/main/events_worker.py index 47a3a26072..ca237c6f12 100644 --- a/synapse/storage/data_stores/main/events_worker.py +++ b/synapse/storage/data_stores/main/events_worker.py @@ -28,9 +28,12 @@ from twisted.internet import defer from synapse.api.constants import EventTypes from synapse.api.errors import NotFoundError -from synapse.api.room_versions import EventFormatVersions -from synapse.events import FrozenEvent, event_type_from_format_version # noqa: F401 -from synapse.events.snapshot import EventContext # noqa: F401 +from synapse.api.room_versions import ( + KNOWN_ROOM_VERSIONS, + EventFormatVersions, + RoomVersions, +) +from synapse.events import make_event_from_dict from synapse.events.utils import prune_event from synapse.logging.context import LoggingContext, PreserveLoggingContext from synapse.metrics.background_process_metrics import run_as_background_process @@ -580,8 +583,49 @@ class EventsWorkerStore(SQLBaseStore): # of a event format version, so it must be a V1 event. format_version = EventFormatVersions.V1 - original_ev = event_type_from_format_version(format_version)( + room_version_id = row["room_version_id"] + + if not room_version_id: + # this should only happen for out-of-band membership events + if not internal_metadata.get("out_of_band_membership"): + logger.warning( + "Room %s for event %s is unknown", d["room_id"], event_id + ) + continue + + # take a wild stab at the room version based on the event format + if format_version == EventFormatVersions.V1: + room_version = RoomVersions.V1 + elif format_version == EventFormatVersions.V2: + room_version = RoomVersions.V3 + else: + room_version = RoomVersions.V5 + else: + room_version = KNOWN_ROOM_VERSIONS.get(room_version_id) + if not room_version: + logger.error( + "Event %s in room %s has unknown room version %s", + event_id, + d["room_id"], + room_version_id, + ) + continue + + if room_version.event_format != format_version: + logger.error( + "Event %s in room %s with version %s has wrong format: " + "expected %s, was %s", + event_id, + d["room_id"], + room_version_id, + room_version.event_format, + format_version, + ) + continue + + original_ev = make_event_from_dict( event_dict=d, + room_version=room_version, internal_metadata_dict=internal_metadata, rejected_reason=rejected_reason, ) @@ -661,6 +705,12 @@ class EventsWorkerStore(SQLBaseStore): of EventFormatVersions. 'None' means the event predates EventFormatVersions (so the event is format V1). + * room_version_id (str|None): The version of the room which contains the event. + Hopefully one of RoomVersions. + + Due to historical reasons, there may be a few events in the database which + do not have an associated room; in this case None will be returned here. + * rejected_reason (str|None): if the event was rejected, the reason why. @@ -676,17 +726,18 @@ class EventsWorkerStore(SQLBaseStore): """ event_dict = {} for evs in batch_iter(event_ids, 200): - sql = ( - "SELECT " - " e.event_id, " - " e.internal_metadata," - " e.json," - " e.format_version, " - " rej.reason " - " FROM event_json as e" - " LEFT JOIN rejections as rej USING (event_id)" - " WHERE " - ) + sql = """\ + SELECT + e.event_id, + e.internal_metadata, + e.json, + e.format_version, + r.room_version, + rej.reason + FROM event_json as e + LEFT JOIN rooms r USING (room_id) + LEFT JOIN rejections as rej USING (event_id) + WHERE """ clause, args = make_in_list_sql_clause( txn.database_engine, "e.event_id", evs @@ -701,7 +752,8 @@ class EventsWorkerStore(SQLBaseStore): "internal_metadata": row[1], "json": row[2], "format_version": row[3], - "rejected_reason": row[4], + "room_version_id": row[4], + "rejected_reason": row[5], "redactions": [], } diff --git a/tests/replication/slave/storage/test_events.py b/tests/replication/slave/storage/test_events.py index d31210fbe4..f0561b30e3 100644 --- a/tests/replication/slave/storage/test_events.py +++ b/tests/replication/slave/storage/test_events.py @@ -15,6 +15,7 @@ import logging from canonicaljson import encode_canonical_json +from synapse.api.room_versions import RoomVersions from synapse.events import FrozenEvent, _EventInternalMetadata, make_event_from_dict from synapse.events.snapshot import EventContext from synapse.handlers.room import RoomEventSource @@ -58,6 +59,15 @@ class SlavedEventStoreTestCase(BaseSlavedStoreTestCase): self.unpatches = [patch__eq__(_EventInternalMetadata), patch__eq__(FrozenEvent)] return super(SlavedEventStoreTestCase, self).setUp() + def prepare(self, *args, **kwargs): + super().prepare(*args, **kwargs) + + self.get_success( + self.master_store.store_room( + ROOM_ID, USER_ID, is_public=False, room_version=RoomVersions.V1, + ) + ) + def tearDown(self): [unpatch() for unpatch in self.unpatches] -- cgit 1.4.1 From 13892776ef7e0b1af2f82c9ca53f7bbd1c60d66f Mon Sep 17 00:00:00 2001 From: Patrick Cloke Date: Wed, 4 Mar 2020 11:30:46 -0500 Subject: Allow deleting an alias if the user has sufficient power level (#6986) --- changelog.d/6986.feature | 1 + synapse/api/auth.py | 9 +-- synapse/handlers/directory.py | 107 ++++++++++++++++++++++---------- tests/handlers/test_directory.py | 128 +++++++++++++++++++++++++++++++-------- tox.ini | 1 + 5 files changed, 182 insertions(+), 64 deletions(-) create mode 100644 changelog.d/6986.feature (limited to 'tests') diff --git a/changelog.d/6986.feature b/changelog.d/6986.feature new file mode 100644 index 0000000000..16dea8bd7f --- /dev/null +++ b/changelog.d/6986.feature @@ -0,0 +1 @@ +Users with a power level sufficient to modify the canonical alias of a room can now delete room aliases. diff --git a/synapse/api/auth.py b/synapse/api/auth.py index 5ca18b4301..c1ade1333b 100644 --- a/synapse/api/auth.py +++ b/synapse/api/auth.py @@ -539,7 +539,7 @@ class Auth(object): @defer.inlineCallbacks def check_can_change_room_list(self, room_id: str, user: UserID): - """Check if the user is allowed to edit the room's entry in the + """Determine whether the user is allowed to edit the room's entry in the published room list. Args: @@ -570,12 +570,7 @@ class Auth(object): ) user_level = event_auth.get_user_power_level(user_id, auth_events) - if user_level < send_level: - raise AuthError( - 403, - "This server requires you to be a moderator in the room to" - " edit its room list entry", - ) + return user_level >= send_level @staticmethod def has_access_token(request): diff --git a/synapse/handlers/directory.py b/synapse/handlers/directory.py index 61eb49059b..1d842c369b 100644 --- a/synapse/handlers/directory.py +++ b/synapse/handlers/directory.py @@ -15,7 +15,7 @@ import logging import string -from typing import List +from typing import Iterable, List, Optional from twisted.internet import defer @@ -28,6 +28,7 @@ from synapse.api.errors import ( StoreError, SynapseError, ) +from synapse.appservice import ApplicationService from synapse.types import Requester, RoomAlias, UserID, get_domain_from_id from ._base import BaseHandler @@ -55,7 +56,13 @@ class DirectoryHandler(BaseHandler): self.spam_checker = hs.get_spam_checker() @defer.inlineCallbacks - def _create_association(self, room_alias, room_id, servers=None, creator=None): + def _create_association( + self, + room_alias: RoomAlias, + room_id: str, + servers: Optional[Iterable[str]] = None, + creator: Optional[str] = None, + ): # general association creation for both human users and app services for wchar in string.whitespace: @@ -81,17 +88,21 @@ class DirectoryHandler(BaseHandler): @defer.inlineCallbacks def create_association( - self, requester, room_alias, room_id, servers=None, check_membership=True, + self, + requester: Requester, + room_alias: RoomAlias, + room_id: str, + servers: Optional[List[str]] = None, + check_membership: bool = True, ): """Attempt to create a new alias Args: - requester (Requester) - room_alias (RoomAlias) - room_id (str) - servers (list[str]|None): List of servers that others servers - should try and join via - check_membership (bool): Whether to check if the user is in the room + requester + room_alias + room_id + servers: Iterable of servers that others servers should try and join via + check_membership: Whether to check if the user is in the room before the alias can be set (if the server's config requires it). Returns: @@ -145,15 +156,15 @@ class DirectoryHandler(BaseHandler): yield self._create_association(room_alias, room_id, servers, creator=user_id) @defer.inlineCallbacks - def delete_association(self, requester, room_alias): + def delete_association(self, requester: Requester, room_alias: RoomAlias): """Remove an alias from the directory (this is only meant for human users; AS users should call delete_appservice_association) Args: - requester (Requester): - room_alias (RoomAlias): + requester + room_alias Returns: Deferred[unicode]: room id that the alias used to point to @@ -189,16 +200,16 @@ class DirectoryHandler(BaseHandler): room_id = yield self._delete_association(room_alias) try: - yield self._update_canonical_alias( - requester, requester.user.to_string(), room_id, room_alias - ) + yield self._update_canonical_alias(requester, user_id, room_id, room_alias) except AuthError as e: logger.info("Failed to update alias events: %s", e) return room_id @defer.inlineCallbacks - def delete_appservice_association(self, service, room_alias): + def delete_appservice_association( + self, service: ApplicationService, room_alias: RoomAlias + ): if not service.is_interested_in_alias(room_alias.to_string()): raise SynapseError( 400, @@ -208,7 +219,7 @@ class DirectoryHandler(BaseHandler): yield self._delete_association(room_alias) @defer.inlineCallbacks - def _delete_association(self, room_alias): + def _delete_association(self, room_alias: RoomAlias): if not self.hs.is_mine(room_alias): raise SynapseError(400, "Room alias must be local") @@ -217,7 +228,7 @@ class DirectoryHandler(BaseHandler): return room_id @defer.inlineCallbacks - def get_association(self, room_alias): + def get_association(self, room_alias: RoomAlias): room_id = None if self.hs.is_mine(room_alias): result = yield self.get_association_from_room_alias(room_alias) @@ -282,7 +293,9 @@ class DirectoryHandler(BaseHandler): ) @defer.inlineCallbacks - def _update_canonical_alias(self, requester, user_id, room_id, room_alias): + def _update_canonical_alias( + self, requester: Requester, user_id: str, room_id: str, room_alias: RoomAlias + ): """ Send an updated canonical alias event if the removed alias was set as the canonical alias or listed in the alt_aliases field. @@ -331,7 +344,7 @@ class DirectoryHandler(BaseHandler): ) @defer.inlineCallbacks - def get_association_from_room_alias(self, room_alias): + def get_association_from_room_alias(self, room_alias: RoomAlias): result = yield self.store.get_association_from_room_alias(room_alias) if not result: # Query AS to see if it exists @@ -339,7 +352,7 @@ class DirectoryHandler(BaseHandler): result = yield as_handler.query_room_alias_exists(room_alias) return result - def can_modify_alias(self, alias, user_id=None): + def can_modify_alias(self, alias: RoomAlias, user_id: Optional[str] = None): # Any application service "interested" in an alias they are regexing on # can modify the alias. # Users can only modify the alias if ALL the interested services have @@ -360,22 +373,42 @@ class DirectoryHandler(BaseHandler): return defer.succeed(True) @defer.inlineCallbacks - def _user_can_delete_alias(self, alias, user_id): + def _user_can_delete_alias(self, alias: RoomAlias, user_id: str): + """Determine whether a user can delete an alias. + + One of the following must be true: + + 1. The user created the alias. + 2. The user is a server administrator. + 3. The user has a power-level sufficient to send a canonical alias event + for the current room. + + """ creator = yield self.store.get_room_alias_creator(alias.to_string()) if creator is not None and creator == user_id: return True - is_admin = yield self.auth.is_server_admin(UserID.from_string(user_id)) - return is_admin + # Resolve the alias to the corresponding room. + room_mapping = yield self.get_association(alias) + room_id = room_mapping["room_id"] + if not room_id: + return False + + res = yield self.auth.check_can_change_room_list( + room_id, UserID.from_string(user_id) + ) + return res @defer.inlineCallbacks - def edit_published_room_list(self, requester, room_id, visibility): + def edit_published_room_list( + self, requester: Requester, room_id: str, visibility: str + ): """Edit the entry of the room in the published room list. requester - room_id (str) - visibility (str): "public" or "private" + room_id + visibility: "public" or "private" """ user_id = requester.user.to_string() @@ -400,7 +433,15 @@ class DirectoryHandler(BaseHandler): if room is None: raise SynapseError(400, "Unknown room") - yield self.auth.check_can_change_room_list(room_id, requester.user) + can_change_room_list = yield self.auth.check_can_change_room_list( + room_id, requester.user + ) + if not can_change_room_list: + raise AuthError( + 403, + "This server requires you to be a moderator in the room to" + " edit its room list entry", + ) making_public = visibility == "public" if making_public: @@ -421,16 +462,16 @@ class DirectoryHandler(BaseHandler): @defer.inlineCallbacks def edit_published_appservice_room_list( - self, appservice_id, network_id, room_id, visibility + self, appservice_id: str, network_id: str, room_id: str, visibility: str ): """Add or remove a room from the appservice/network specific public room list. Args: - appservice_id (str): ID of the appservice that owns the list - network_id (str): The ID of the network the list is associated with - room_id (str) - visibility (str): either "public" or "private" + appservice_id: ID of the appservice that owns the list + network_id: The ID of the network the list is associated with + room_id + visibility: either "public" or "private" """ if visibility not in ["public", "private"]: raise SynapseError(400, "Invalid visibility setting") diff --git a/tests/handlers/test_directory.py b/tests/handlers/test_directory.py index 3397cfa485..5e40adba52 100644 --- a/tests/handlers/test_directory.py +++ b/tests/handlers/test_directory.py @@ -18,6 +18,7 @@ from mock import Mock from twisted.internet import defer +import synapse import synapse.api.errors from synapse.api.constants import EventTypes from synapse.config.room_directory import RoomDirectoryConfig @@ -87,52 +88,131 @@ class DirectoryTestCase(unittest.HomeserverTestCase): ignore_backoff=True, ) - def test_delete_alias_not_allowed(self): - """Removing an alias should be denied if a user does not have the proper permissions.""" - room_id = "!8765qwer:test" + def test_incoming_fed_query(self): + self.get_success( + self.store.create_room_alias_association( + self.your_room, "!8765asdf:test", ["test"] + ) + ) + + response = self.get_success( + self.handler.on_directory_query({"room_alias": "#your-room:test"}) + ) + + self.assertEquals({"room_id": "!8765asdf:test", "servers": ["test"]}, response) + + +class TestDeleteAlias(unittest.HomeserverTestCase): + servlets = [ + synapse.rest.admin.register_servlets, + login.register_servlets, + room.register_servlets, + directory.register_servlets, + ] + + def prepare(self, reactor, clock, hs): + self.store = hs.get_datastore() + self.handler = hs.get_handlers().directory_handler + self.state_handler = hs.get_state_handler() + + # Create user + self.admin_user = self.register_user("admin", "pass", admin=True) + self.admin_user_tok = self.login("admin", "pass") + + # Create a test room + self.room_id = self.helper.create_room_as( + self.admin_user, tok=self.admin_user_tok + ) + + self.test_alias = "#test:test" + self.room_alias = RoomAlias.from_string(self.test_alias) + + # Create a test user. + self.test_user = self.register_user("user", "pass", admin=False) + self.test_user_tok = self.login("user", "pass") + self.helper.join(room=self.room_id, user=self.test_user, tok=self.test_user_tok) + + def _create_alias(self, user): + # Create a new alias to this room. self.get_success( - self.store.create_room_alias_association(self.my_room, room_id, ["test"]) + self.store.create_room_alias_association( + self.room_alias, self.room_id, ["test"], user + ) ) + def test_delete_alias_not_allowed(self): + """A user that doesn't meet the expected guidelines cannot delete an alias.""" + self._create_alias(self.admin_user) self.get_failure( self.handler.delete_association( - create_requester("@user:test"), self.my_room + create_requester(self.test_user), self.room_alias ), synapse.api.errors.AuthError, ) - def test_delete_alias(self): - """Removing an alias should work when a user does has the proper permissions.""" - room_id = "!8765qwer:test" - user_id = "@user:test" - self.get_success( - self.store.create_room_alias_association( - self.my_room, room_id, ["test"], user_id + def test_delete_alias_creator(self): + """An alias creator can delete their own alias.""" + # Create an alias from a different user. + self._create_alias(self.test_user) + + # Delete the user's alias. + result = self.get_success( + self.handler.delete_association( + create_requester(self.test_user), self.room_alias ) ) + self.assertEquals(self.room_id, result) + # Confirm the alias is gone. + self.get_failure( + self.handler.get_association(self.room_alias), + synapse.api.errors.SynapseError, + ) + + def test_delete_alias_admin(self): + """A server admin can delete an alias created by another user.""" + # Create an alias from a different user. + self._create_alias(self.test_user) + + # Delete the user's alias as the admin. result = self.get_success( - self.handler.delete_association(create_requester(user_id), self.my_room) + self.handler.delete_association( + create_requester(self.admin_user), self.room_alias + ) ) - self.assertEquals(room_id, result) + self.assertEquals(self.room_id, result) - # The alias should not be found. + # Confirm the alias is gone. self.get_failure( - self.handler.get_association(self.my_room), synapse.api.errors.SynapseError + self.handler.get_association(self.room_alias), + synapse.api.errors.SynapseError, ) - def test_incoming_fed_query(self): - self.get_success( - self.store.create_room_alias_association( - self.your_room, "!8765asdf:test", ["test"] - ) + def test_delete_alias_sufficient_power(self): + """A user with a sufficient power level should be able to delete an alias.""" + self._create_alias(self.admin_user) + + # Increase the user's power level. + self.helper.send_state( + self.room_id, + "m.room.power_levels", + {"users": {self.test_user: 100}}, + tok=self.admin_user_tok, ) - response = self.get_success( - self.handler.on_directory_query({"room_alias": "#your-room:test"}) + # They can now delete the alias. + result = self.get_success( + self.handler.delete_association( + create_requester(self.test_user), self.room_alias + ) ) + self.assertEquals(self.room_id, result) - self.assertEquals({"room_id": "!8765asdf:test", "servers": ["test"]}, response) + # Confirm the alias is gone. + self.get_failure( + self.handler.get_association(self.room_alias), + synapse.api.errors.SynapseError, + ) class CanonicalAliasTestCase(unittest.HomeserverTestCase): diff --git a/tox.ini b/tox.ini index 097ebb8774..7622aa19f1 100644 --- a/tox.ini +++ b/tox.ini @@ -185,6 +185,7 @@ commands = mypy \ synapse/federation/federation_client.py \ synapse/federation/sender \ synapse/federation/transport \ + synapse/handlers/directory.py \ synapse/handlers/presence.py \ synapse/handlers/sync.py \ synapse/handlers/ui_auth \ -- cgit 1.4.1 From 1d66dce83e58827aae12080552edeaeb357b1997 Mon Sep 17 00:00:00 2001 From: Neil Johnson Date: Fri, 6 Mar 2020 18:14:19 +0000 Subject: Break down monthly active users by appservice_id (#7030) * Break down monthly active users by appservice_id and emit via prometheus. Co-authored-by: Brendan Abolivier --- changelog.d/7030.feature | 1 + synapse/app/homeserver.py | 13 +++++++ .../data_stores/main/monthly_active_users.py | 32 ++++++++++++++++- tests/storage/test_monthly_active_users.py | 42 ++++++++++++++++++++++ 4 files changed, 87 insertions(+), 1 deletion(-) create mode 100644 changelog.d/7030.feature (limited to 'tests') diff --git a/changelog.d/7030.feature b/changelog.d/7030.feature new file mode 100644 index 0000000000..fcfdb8d8a1 --- /dev/null +++ b/changelog.d/7030.feature @@ -0,0 +1 @@ +Break down monthly active users by `appservice_id` and emit via Prometheus. diff --git a/synapse/app/homeserver.py b/synapse/app/homeserver.py index c2a334a2b0..e0fdddfdc9 100644 --- a/synapse/app/homeserver.py +++ b/synapse/app/homeserver.py @@ -298,6 +298,11 @@ class SynapseHomeServer(HomeServer): # Gauges to expose monthly active user control metrics current_mau_gauge = Gauge("synapse_admin_mau:current", "Current MAU") +current_mau_by_service_gauge = Gauge( + "synapse_admin_mau_current_mau_by_service", + "Current MAU by service", + ["app_service"], +) max_mau_gauge = Gauge("synapse_admin_mau:max", "MAU Limit") registered_reserved_users_mau_gauge = Gauge( "synapse_admin_mau:registered_reserved_users", @@ -585,12 +590,20 @@ def run(hs): @defer.inlineCallbacks def generate_monthly_active_users(): current_mau_count = 0 + current_mau_count_by_service = {} reserved_users = () store = hs.get_datastore() if hs.config.limit_usage_by_mau or hs.config.mau_stats_only: current_mau_count = yield store.get_monthly_active_count() + current_mau_count_by_service = ( + yield store.get_monthly_active_count_by_service() + ) reserved_users = yield store.get_registered_reserved_users() current_mau_gauge.set(float(current_mau_count)) + + for app_service, count in current_mau_count_by_service.items(): + current_mau_by_service_gauge.labels(app_service).set(float(count)) + registered_reserved_users_mau_gauge.set(float(len(reserved_users))) max_mau_gauge.set(float(hs.config.max_mau_value)) diff --git a/synapse/storage/data_stores/main/monthly_active_users.py b/synapse/storage/data_stores/main/monthly_active_users.py index 1507a14e09..925bc5691b 100644 --- a/synapse/storage/data_stores/main/monthly_active_users.py +++ b/synapse/storage/data_stores/main/monthly_active_users.py @@ -43,13 +43,40 @@ class MonthlyActiveUsersWorkerStore(SQLBaseStore): def _count_users(txn): sql = "SELECT COALESCE(count(*), 0) FROM monthly_active_users" - txn.execute(sql) (count,) = txn.fetchone() return count return self.db.runInteraction("count_users", _count_users) + @cached(num_args=0) + def get_monthly_active_count_by_service(self): + """Generates current count of monthly active users broken down by service. + A service is typically an appservice but also includes native matrix users. + Since the `monthly_active_users` table is populated from the `user_ips` table + `config.track_appservice_user_ips` must be set to `true` for this + method to return anything other than native matrix users. + + Returns: + Deferred[dict]: dict that includes a mapping between app_service_id + and the number of occurrences. + + """ + + def _count_users_by_service(txn): + sql = """ + SELECT COALESCE(appservice_id, 'native'), COALESCE(count(*), 0) + FROM monthly_active_users + LEFT JOIN users ON monthly_active_users.user_id=users.name + GROUP BY appservice_id; + """ + + txn.execute(sql) + result = txn.fetchall() + return dict(result) + + return self.db.runInteraction("count_users_by_service", _count_users_by_service) + @defer.inlineCallbacks def get_registered_reserved_users(self): """Of the reserved threepids defined in config, which are associated @@ -291,6 +318,9 @@ class MonthlyActiveUsersStore(MonthlyActiveUsersWorkerStore): ) self._invalidate_cache_and_stream(txn, self.get_monthly_active_count, ()) + self._invalidate_cache_and_stream( + txn, self.get_monthly_active_count_by_service, () + ) self._invalidate_cache_and_stream( txn, self.user_last_seen_monthly_active, (user_id,) ) diff --git a/tests/storage/test_monthly_active_users.py b/tests/storage/test_monthly_active_users.py index 3c78faab45..bc53bf0951 100644 --- a/tests/storage/test_monthly_active_users.py +++ b/tests/storage/test_monthly_active_users.py @@ -303,3 +303,45 @@ class MonthlyActiveUsersTestCase(unittest.HomeserverTestCase): self.pump() self.store.upsert_monthly_active_user.assert_not_called() + + def test_get_monthly_active_count_by_service(self): + appservice1_user1 = "@appservice1_user1:example.com" + appservice1_user2 = "@appservice1_user2:example.com" + + appservice2_user1 = "@appservice2_user1:example.com" + native_user1 = "@native_user1:example.com" + + service1 = "service1" + service2 = "service2" + native = "native" + + self.store.register_user( + user_id=appservice1_user1, password_hash=None, appservice_id=service1 + ) + self.store.register_user( + user_id=appservice1_user2, password_hash=None, appservice_id=service1 + ) + self.store.register_user( + user_id=appservice2_user1, password_hash=None, appservice_id=service2 + ) + self.store.register_user(user_id=native_user1, password_hash=None) + self.pump() + + count = self.store.get_monthly_active_count_by_service() + self.assertEqual({}, self.get_success(count)) + + self.store.upsert_monthly_active_user(native_user1) + self.store.upsert_monthly_active_user(appservice1_user1) + self.store.upsert_monthly_active_user(appservice1_user2) + self.store.upsert_monthly_active_user(appservice2_user1) + self.pump() + + count = self.store.get_monthly_active_count() + self.assertEqual(4, self.get_success(count)) + + count = self.store.get_monthly_active_count_by_service() + result = self.get_success(count) + + self.assertEqual(2, result[service1]) + self.assertEqual(1, result[service2]) + self.assertEqual(1, result[native]) -- cgit 1.4.1 From 1f5f3ae8b1c5db96d36ac7c104f13553bc4283da Mon Sep 17 00:00:00 2001 From: dklimpel <5740567+dklimpel@users.noreply.github.com> Date: Sun, 8 Mar 2020 14:49:33 +0100 Subject: Add options to disable setting profile info for prevent changes. --- synapse/config/registration.py | 11 +++++++++++ synapse/handlers/profile.py | 10 ++++++++++ tests/handlers/test_profile.py | 33 ++++++++++++++++++++++++++++++++- 3 files changed, 53 insertions(+), 1 deletion(-) (limited to 'tests') diff --git a/synapse/config/registration.py b/synapse/config/registration.py index 9bb3beedbc..d9f452dcea 100644 --- a/synapse/config/registration.py +++ b/synapse/config/registration.py @@ -129,6 +129,9 @@ class RegistrationConfig(Config): raise ConfigError("Invalid auto_join_rooms entry %s" % (room_alias,)) self.autocreate_auto_join_rooms = config.get("autocreate_auto_join_rooms", True) + self.disable_set_displayname = config.get("disable_set_displayname", False) + self.disable_set_avatar_url = config.get("disable_set_avatar_url", False) + self.disable_msisdn_registration = config.get( "disable_msisdn_registration", False ) @@ -330,6 +333,14 @@ class RegistrationConfig(Config): #email: https://example.com # Delegate email sending to example.com #msisdn: http://localhost:8090 # Delegate SMS sending to this local process + # If enabled, don't let users set their own display names/avatars + # other than for the very first time (unless they are a server admin). + # Useful when provisioning users based on the contents of a 3rd party + # directory and to avoid ambiguities. + # + # disable_set_displayname: False + # disable_set_avatar_url: False + # Users who register on this homeserver will automatically be joined # to these rooms # diff --git a/synapse/handlers/profile.py b/synapse/handlers/profile.py index 50ce0c585b..fb7e84f3b8 100644 --- a/synapse/handlers/profile.py +++ b/synapse/handlers/profile.py @@ -157,6 +157,11 @@ class BaseProfileHandler(BaseHandler): if not by_admin and target_user != requester.user: raise AuthError(400, "Cannot set another user's displayname") + if not by_admin and self.hs.config.disable_set_displayname: + profile = yield self.store.get_profileinfo(target_user.localpart) + if profile.display_name: + raise SynapseError(400, "Changing displayname is disabled on this server") + if len(new_displayname) > MAX_DISPLAYNAME_LEN: raise SynapseError( 400, "Displayname is too long (max %i)" % (MAX_DISPLAYNAME_LEN,) @@ -218,6 +223,11 @@ class BaseProfileHandler(BaseHandler): if not by_admin and target_user != requester.user: raise AuthError(400, "Cannot set another user's avatar_url") + if not by_admin and self.hs.config.disable_set_avatar_url: + profile = yield self.store.get_profileinfo(target_user.localpart) + if profile.avatar_url: + raise SynapseError(400, "Changing avatar url is disabled on this server") + if len(new_avatar_url) > MAX_AVATAR_URL_LEN: raise SynapseError( 400, "Avatar URL is too long (max %i)" % (MAX_AVATAR_URL_LEN,) diff --git a/tests/handlers/test_profile.py b/tests/handlers/test_profile.py index d60c124eec..b85520c688 100644 --- a/tests/handlers/test_profile.py +++ b/tests/handlers/test_profile.py @@ -19,7 +19,7 @@ from mock import Mock, NonCallableMock from twisted.internet import defer import synapse.types -from synapse.api.errors import AuthError +from synapse.api.errors import AuthError, SynapseError from synapse.handlers.profile import MasterProfileHandler from synapse.types import UserID @@ -70,6 +70,7 @@ class ProfileTestCase(unittest.TestCase): yield self.store.create_profile(self.frank.localpart) self.handler = hs.get_profile_handler() + self.config = hs.config @defer.inlineCallbacks def test_get_my_name(self): @@ -90,6 +91,19 @@ class ProfileTestCase(unittest.TestCase): "Frank Jr.", ) + @defer.inlineCallbacks + def test_set_my_name_if_disabled(self): + self.config.disable_set_displayname = True + + # Set first displayname is allowed, if displayname is null + self.store.set_profile_displayname(self.frank.localpart, "Frank") + + d = self.handler.set_displayname( + self.frank, synapse.types.create_requester(self.frank), "Frank Jr." + ) + + yield self.assertFailure(d, SynapseError) + @defer.inlineCallbacks def test_set_my_name_noauth(self): d = self.handler.set_displayname( @@ -147,3 +161,20 @@ class ProfileTestCase(unittest.TestCase): (yield self.store.get_profile_avatar_url(self.frank.localpart)), "http://my.server/pic.gif", ) + + @defer.inlineCallbacks + def test_set_my_avatar_if_disabled(self): + self.config.disable_set_avatar_url = True + + # Set first time avatar is allowed, if displayname is null + self.store.set_profile_avatar_url( + self.frank.localpart, "http://my.server/me.png" + ) + + d = self.handler.set_avatar_url( + self.frank, + synapse.types.create_requester(self.frank), + "http://my.server/pic.gif", + ) + + yield self.assertFailure(d, SynapseError) -- cgit 1.4.1 From 06eb5cae08272c401a586991fc81f788825f910b Mon Sep 17 00:00:00 2001 From: Patrick Cloke Date: Mon, 9 Mar 2020 08:58:25 -0400 Subject: Remove special auth and redaction rules for aliases events in experimental room ver. (#7037) --- changelog.d/7037.feature | 1 + synapse/api/room_versions.py | 9 +-- synapse/crypto/event_signing.py | 2 +- synapse/event_auth.py | 8 +-- synapse/events/utils.py | 12 ++-- synapse/storage/data_stores/main/events.py | 10 +++- tests/events/test_utils.py | 35 ++++++++++- tests/test_event_auth.py | 93 +++++++++++++++++++++++++++++- 8 files changed, 148 insertions(+), 22 deletions(-) create mode 100644 changelog.d/7037.feature (limited to 'tests') diff --git a/changelog.d/7037.feature b/changelog.d/7037.feature new file mode 100644 index 0000000000..4bc1b3b19f --- /dev/null +++ b/changelog.d/7037.feature @@ -0,0 +1 @@ +Implement updated authorization rules and redaction rules for aliases events, from [MSC2261](https://github.com/matrix-org/matrix-doc/pull/2261) and [MSC2432](https://github.com/matrix-org/matrix-doc/pull/2432). diff --git a/synapse/api/room_versions.py b/synapse/api/room_versions.py index cf7ee60d3a..871179749a 100644 --- a/synapse/api/room_versions.py +++ b/synapse/api/room_versions.py @@ -57,7 +57,7 @@ class RoomVersion(object): state_res = attr.ib() # int; one of the StateResolutionVersions enforce_key_validity = attr.ib() # bool - # bool: before MSC2260, anyone was allowed to send an aliases event + # bool: before MSC2261/MSC2432, m.room.aliases had special auth rules and redaction rules special_case_aliases_auth = attr.ib(type=bool, default=False) @@ -102,12 +102,13 @@ class RoomVersions(object): enforce_key_validity=True, special_case_aliases_auth=True, ) - MSC2260_DEV = RoomVersion( - "org.matrix.msc2260", + MSC2432_DEV = RoomVersion( + "org.matrix.msc2432", RoomDisposition.UNSTABLE, EventFormatVersions.V3, StateResolutionVersions.V2, enforce_key_validity=True, + special_case_aliases_auth=False, ) @@ -119,6 +120,6 @@ KNOWN_ROOM_VERSIONS = { RoomVersions.V3, RoomVersions.V4, RoomVersions.V5, - RoomVersions.MSC2260_DEV, + RoomVersions.MSC2432_DEV, ) } # type: Dict[str, RoomVersion] diff --git a/synapse/crypto/event_signing.py b/synapse/crypto/event_signing.py index 5f733c1cf5..0422c43fab 100644 --- a/synapse/crypto/event_signing.py +++ b/synapse/crypto/event_signing.py @@ -140,7 +140,7 @@ def compute_event_signature( Returns: a dictionary in the same format of an event's signatures field. """ - redact_json = prune_event_dict(event_dict) + redact_json = prune_event_dict(room_version, event_dict) redact_json.pop("age_ts", None) redact_json.pop("unsigned", None) if logger.isEnabledFor(logging.DEBUG): diff --git a/synapse/event_auth.py b/synapse/event_auth.py index 472f165044..46beb5334f 100644 --- a/synapse/event_auth.py +++ b/synapse/event_auth.py @@ -137,7 +137,7 @@ def check( raise AuthError(403, "This room has been marked as unfederatable.") # 4. If type is m.room.aliases - if event.type == EventTypes.Aliases: + if event.type == EventTypes.Aliases and room_version_obj.special_case_aliases_auth: # 4a. If event has no state_key, reject if not event.is_state(): raise AuthError(403, "Alias event must be a state event") @@ -152,10 +152,8 @@ def check( ) # 4c. Otherwise, allow. - # This is removed by https://github.com/matrix-org/matrix-doc/pull/2260 - if room_version_obj.special_case_aliases_auth: - logger.debug("Allowing! %s", event) - return + logger.debug("Allowing! %s", event) + return if logger.isEnabledFor(logging.DEBUG): logger.debug("Auth events: %s", [a.event_id for a in auth_events.values()]) diff --git a/synapse/events/utils.py b/synapse/events/utils.py index bc6f98ae3b..b75b097e5e 100644 --- a/synapse/events/utils.py +++ b/synapse/events/utils.py @@ -23,6 +23,7 @@ from frozendict import frozendict from twisted.internet import defer from synapse.api.constants import EventTypes, RelationTypes +from synapse.api.room_versions import RoomVersion from synapse.util.async_helpers import yieldable_gather_results from . import EventBase @@ -43,7 +44,7 @@ def prune_event(event: EventBase) -> EventBase: the user has specified, but we do want to keep necessary information like type, state_key etc. """ - pruned_event_dict = prune_event_dict(event.get_dict()) + pruned_event_dict = prune_event_dict(event.room_version, event.get_dict()) from . import make_event_from_dict @@ -57,15 +58,12 @@ def prune_event(event: EventBase) -> EventBase: return pruned_event -def prune_event_dict(event_dict): +def prune_event_dict(room_version: RoomVersion, event_dict: dict) -> dict: """Redacts the event_dict in the same way as `prune_event`, except it operates on dicts rather than event objects - Args: - event_dict (dict) - Returns: - dict: A copy of the pruned event dict + A copy of the pruned event dict """ allowed_keys = [ @@ -112,7 +110,7 @@ def prune_event_dict(event_dict): "kick", "redact", ) - elif event_type == EventTypes.Aliases: + elif event_type == EventTypes.Aliases and room_version.special_case_aliases_auth: add_fields("aliases") elif event_type == EventTypes.RoomHistoryVisibility: add_fields("history_visibility") diff --git a/synapse/storage/data_stores/main/events.py b/synapse/storage/data_stores/main/events.py index 8ae23df00a..d593ef47b8 100644 --- a/synapse/storage/data_stores/main/events.py +++ b/synapse/storage/data_stores/main/events.py @@ -1168,7 +1168,11 @@ class EventsStore( and original_event.internal_metadata.is_redacted() ): # Redaction was allowed - pruned_json = encode_json(prune_event_dict(original_event.get_dict())) + pruned_json = encode_json( + prune_event_dict( + original_event.room_version, original_event.get_dict() + ) + ) else: # Redaction wasn't allowed pruned_json = None @@ -1929,7 +1933,9 @@ class EventsStore( return # Prune the event's dict then convert it to JSON. - pruned_json = encode_json(prune_event_dict(event.get_dict())) + pruned_json = encode_json( + prune_event_dict(event.room_version, event.get_dict()) + ) # Update the event_json table to replace the event's JSON with the pruned # JSON. diff --git a/tests/events/test_utils.py b/tests/events/test_utils.py index 45d55b9e94..ab5f5ac549 100644 --- a/tests/events/test_utils.py +++ b/tests/events/test_utils.py @@ -13,6 +13,7 @@ # See the License for the specific language governing permissions and # limitations under the License. +from synapse.api.room_versions import RoomVersions from synapse.events import make_event_from_dict from synapse.events.utils import ( copy_power_levels_contents, @@ -36,9 +37,9 @@ class PruneEventTestCase(unittest.TestCase): """ Asserts that a new event constructed with `evdict` will look like `matchdict` when it is redacted. """ - def run_test(self, evdict, matchdict): + def run_test(self, evdict, matchdict, **kwargs): self.assertEquals( - prune_event(make_event_from_dict(evdict)).get_dict(), matchdict + prune_event(make_event_from_dict(evdict, **kwargs)).get_dict(), matchdict ) def test_minimal(self): @@ -128,6 +129,36 @@ class PruneEventTestCase(unittest.TestCase): }, ) + def test_alias_event(self): + """Alias events have special behavior up through room version 6.""" + self.run_test( + { + "type": "m.room.aliases", + "event_id": "$test:domain", + "content": {"aliases": ["test"]}, + }, + { + "type": "m.room.aliases", + "event_id": "$test:domain", + "content": {"aliases": ["test"]}, + "signatures": {}, + "unsigned": {}, + }, + ) + + def test_msc2432_alias_event(self): + """After MSC2432, alias events have no special behavior.""" + self.run_test( + {"type": "m.room.aliases", "content": {"aliases": ["test"]}}, + { + "type": "m.room.aliases", + "content": {}, + "signatures": {}, + "unsigned": {}, + }, + room_version=RoomVersions.MSC2432_DEV, + ) + class SerializeEventTestCase(unittest.TestCase): def serialize(self, ev, fields): diff --git a/tests/test_event_auth.py b/tests/test_event_auth.py index bfa5d6f510..6c2351cf55 100644 --- a/tests/test_event_auth.py +++ b/tests/test_event_auth.py @@ -19,6 +19,7 @@ from synapse import event_auth from synapse.api.errors import AuthError from synapse.api.room_versions import RoomVersions from synapse.events import make_event_from_dict +from synapse.types import get_domain_from_id class EventAuthTestCase(unittest.TestCase): @@ -51,7 +52,7 @@ class EventAuthTestCase(unittest.TestCase): _random_state_event(joiner), auth_events, do_sig_check=False, - ), + ) def test_state_default_level(self): """ @@ -87,6 +88,83 @@ class EventAuthTestCase(unittest.TestCase): RoomVersions.V1, _random_state_event(king), auth_events, do_sig_check=False, ) + def test_alias_event(self): + """Alias events have special behavior up through room version 6.""" + creator = "@creator:example.com" + other = "@other:example.com" + auth_events = { + ("m.room.create", ""): _create_event(creator), + ("m.room.member", creator): _join_event(creator), + } + + # creator should be able to send aliases + event_auth.check( + RoomVersions.V1, _alias_event(creator), auth_events, do_sig_check=False, + ) + + # Reject an event with no state key. + with self.assertRaises(AuthError): + event_auth.check( + RoomVersions.V1, + _alias_event(creator, state_key=""), + auth_events, + do_sig_check=False, + ) + + # If the domain of the sender does not match the state key, reject. + with self.assertRaises(AuthError): + event_auth.check( + RoomVersions.V1, + _alias_event(creator, state_key="test.com"), + auth_events, + do_sig_check=False, + ) + + # Note that the member does *not* need to be in the room. + event_auth.check( + RoomVersions.V1, _alias_event(other), auth_events, do_sig_check=False, + ) + + def test_msc2432_alias_event(self): + """After MSC2432, alias events have no special behavior.""" + creator = "@creator:example.com" + other = "@other:example.com" + auth_events = { + ("m.room.create", ""): _create_event(creator), + ("m.room.member", creator): _join_event(creator), + } + + # creator should be able to send aliases + event_auth.check( + RoomVersions.MSC2432_DEV, + _alias_event(creator), + auth_events, + do_sig_check=False, + ) + + # No particular checks are done on the state key. + event_auth.check( + RoomVersions.MSC2432_DEV, + _alias_event(creator, state_key=""), + auth_events, + do_sig_check=False, + ) + event_auth.check( + RoomVersions.MSC2432_DEV, + _alias_event(creator, state_key="test.com"), + auth_events, + do_sig_check=False, + ) + + # Per standard auth rules, the member must be in the room. + with self.assertRaises(AuthError): + event_auth.check( + RoomVersions.MSC2432_DEV, + _alias_event(other), + auth_events, + do_sig_check=False, + ) + # helpers for making events @@ -131,6 +209,19 @@ def _power_levels_event(sender, content): ) +def _alias_event(sender, **kwargs): + data = { + "room_id": TEST_ROOM_ID, + "event_id": _get_event_id(), + "type": "m.room.aliases", + "sender": sender, + "state_key": get_domain_from_id(sender), + "content": {"aliases": []}, + } + data.update(**kwargs) + return make_event_from_dict(data) + + def _random_state_event(sender): return make_event_from_dict( { -- cgit 1.4.1 From 04f4b5f6f87fbba0b2f1a4f011c496de3021c81a Mon Sep 17 00:00:00 2001 From: dklimpel <5740567+dklimpel@users.noreply.github.com> Date: Mon, 9 Mar 2020 19:51:31 +0100 Subject: add tests --- tests/handlers/test_profile.py | 6 +- tests/rest/client/v2_alpha/test_account.py | 308 +++++++++++++++++++++++++++++ 2 files changed, 311 insertions(+), 3 deletions(-) (limited to 'tests') diff --git a/tests/handlers/test_profile.py b/tests/handlers/test_profile.py index b85520c688..98b508c3d4 100644 --- a/tests/handlers/test_profile.py +++ b/tests/handlers/test_profile.py @@ -70,7 +70,7 @@ class ProfileTestCase(unittest.TestCase): yield self.store.create_profile(self.frank.localpart) self.handler = hs.get_profile_handler() - self.config = hs.config + self.hs = hs @defer.inlineCallbacks def test_get_my_name(self): @@ -93,7 +93,7 @@ class ProfileTestCase(unittest.TestCase): @defer.inlineCallbacks def test_set_my_name_if_disabled(self): - self.config.disable_set_displayname = True + self.hs.config.disable_set_displayname = True # Set first displayname is allowed, if displayname is null self.store.set_profile_displayname(self.frank.localpart, "Frank") @@ -164,7 +164,7 @@ class ProfileTestCase(unittest.TestCase): @defer.inlineCallbacks def test_set_my_avatar_if_disabled(self): - self.config.disable_set_avatar_url = True + self.hs.config.disable_set_avatar_url = True # Set first time avatar is allowed, if displayname is null self.store.set_profile_avatar_url( diff --git a/tests/rest/client/v2_alpha/test_account.py b/tests/rest/client/v2_alpha/test_account.py index c3facc00eb..ac9f200de3 100644 --- a/tests/rest/client/v2_alpha/test_account.py +++ b/tests/rest/client/v2_alpha/test_account.py @@ -325,3 +325,311 @@ class DeactivateTestCase(unittest.HomeserverTestCase): ) self.render(request) self.assertEqual(request.code, 200) + + +class ThreepidEmailRestTestCase(unittest.HomeserverTestCase): + + servlets = [ + account.register_servlets, + login.register_servlets, + synapse.rest.admin.register_servlets_for_client_rest_resource, + ] + + def make_homeserver(self, reactor, clock): + config = self.default_config() + + # Email config. + self.email_attempts = [] + + def sendmail(smtphost, from_addr, to_addrs, msg, **kwargs): + self.email_attempts.append(msg) + return + + config["email"] = { + "enable_notifs": False, + "template_dir": os.path.abspath( + pkg_resources.resource_filename("synapse", "res/templates") + ), + "smtp_host": "127.0.0.1", + "smtp_port": 20, + "require_transport_security": False, + "smtp_user": None, + "smtp_pass": None, + "notif_from": "test@example.com", + } + config["public_baseurl"] = "https://example.com" + + self.hs = self.setup_test_homeserver(config=config, sendmail=sendmail) + return self.hs + + def prepare(self, reactor, clock, hs): + self.store = hs.get_datastore() + + self.user_id = self.register_user("kermit", "test") + self.user_id_tok = self.login("kermit", "test") + self.email = "test@example.com" + self.url_3pid = b"account/3pid" + + def test_add_email(self): + """Test add mail to profile + """ + client_secret = "foobar" + session_id = self._request_token(self.email, client_secret) + + self.assertEquals(len(self.email_attempts), 1) + link = self._get_link_from_email() + + self._validate_token(link) + + request, channel = self.make_request( + "POST", + b"/_matrix/client/unstable/account/3pid/add", + { + "client_secret": client_secret, + "sid": session_id, + "auth": { + "type": "m.login.password", + "user": self.user_id, + "password": "test", + }, + }, + access_token=self.user_id_tok, + ) + + self.render(request) + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + + # Get user + request, channel = self.make_request( + "GET", self.url_3pid, access_token=self.user_id_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("email", channel.json_body["threepids"][0]["medium"]) + self.assertEqual(self.email, channel.json_body["threepids"][0]["address"]) + + def test_add_email_if_disabled(self): + """Test add mail to profile if disabled + """ + self.hs.config.disable_3pid_changes = True + + client_secret = "foobar" + session_id = self._request_token(self.email, client_secret) + + self.assertEquals(len(self.email_attempts), 1) + link = self._get_link_from_email() + + self._validate_token(link) + + request, channel = self.make_request( + "POST", + b"/_matrix/client/unstable/account/3pid/add", + { + "client_secret": client_secret, + "sid": session_id, + "auth": { + "type": "m.login.password", + "user": self.user_id, + "password": "test", + }, + }, + access_token=self.user_id_tok, + ) + self.render(request) + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("3PID changes disabled on this server", channel.json_body["error"]) + + # Get user + request, channel = self.make_request( + "GET", self.url_3pid, access_token=self.user_id_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertFalse(channel.json_body["threepids"]) + + def test_delete_email(self): + """Test delete mail from profile + """ + # Add a threepid + self.get_success( + self.store.user_add_threepid( + user_id=self.user_id, + medium="email", + address=self.email, + validated_at=0, + added_at=0, + ) + ) + + request, channel = self.make_request( + "POST", + b"account/3pid/delete", + { + "medium": "email", + "address": self.email + }, + access_token=self.user_id_tok, + ) + self.render(request) + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + + # Get user + request, channel = self.make_request( + "GET", self.url_3pid, access_token=self.user_id_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertFalse(channel.json_body["threepids"]) + + def test_delete_email_if_disabled(self): + """Test delete mail from profile if disabled + """ + self.hs.config.disable_3pid_changes = True + + # Add a threepid + self.get_success( + self.store.user_add_threepid( + user_id=self.user_id, + medium="email", + address=self.email, + validated_at=0, + added_at=0, + ) + ) + + request, channel = self.make_request( + "POST", + b"account/3pid/delete", + { + "medium": "email", + "address": self.email + }, + access_token=self.user_id_tok, + ) + self.render(request) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("3PID changes disabled on this server", channel.json_body["error"]) + + # Get user + request, channel = self.make_request( + "GET", self.url_3pid, access_token=self.user_id_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("email", channel.json_body["threepids"][0]["medium"]) + self.assertEqual(self.email, channel.json_body["threepids"][0]["address"]) + + def test_cant_add_email_without_clicking_link(self): + """Test that we do actually need to click the link in the email + """ + client_secret = "foobar" + session_id = self._request_token(self.email, client_secret) + + self.assertEquals(len(self.email_attempts), 1) + + # Attempt to add email without clicking the link + request, channel = self.make_request( + "POST", + b"/_matrix/client/unstable/account/3pid/add", + { + "client_secret": client_secret, + "sid": session_id, + "auth": { + "type": "m.login.password", + "user": self.user_id, + "password": "test", + }, + }, + access_token=self.user_id_tok, + ) + self.render(request) + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("No validated 3pid session found", channel.json_body["error"]) + + # Get user + request, channel = self.make_request( + "GET", self.url_3pid, access_token=self.user_id_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertFalse(channel.json_body["threepids"]) + + def test_no_valid_token(self): + """Test that we do actually need to request a token and can't just + make a session up. + """ + client_secret = "foobar" + session_id = "weasle" + + # Attempt to add email without even requesting an email + request, channel = self.make_request( + "POST", + b"/_matrix/client/unstable/account/3pid/add", + { + "client_secret": client_secret, + "sid": session_id, + "auth": { + "type": "m.login.password", + "user": self.user_id, + "password": "test", + }, + }, + access_token=self.user_id_tok, + ) + self.render(request) + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("No validated 3pid session found", channel.json_body["error"]) + + # Get user + request, channel = self.make_request( + "GET", self.url_3pid, access_token=self.user_id_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertFalse(channel.json_body["threepids"]) + + def _request_token(self, email, client_secret): + request, channel = self.make_request( + "POST", + b"account/3pid/email/requestToken", + {"client_secret": client_secret, "email": email, "send_attempt": 1}, + ) + self.render(request) + self.assertEquals(200, channel.code, channel.result) + + return channel.json_body["sid"] + + def _validate_token(self, link): + # Remove the host + path = link.replace("https://example.com", "") + + request, channel = self.make_request("GET", path, shorthand=False) + self.render(request) + self.assertEquals(200, channel.code, channel.result) + + def _get_link_from_email(self): + assert self.email_attempts, "No emails have been sent" + + raw_msg = self.email_attempts[-1].decode("UTF-8") + mail = Parser().parsestr(raw_msg) + + text = None + for part in mail.walk(): + if part.get_content_type() == "text/plain": + text = part.get_payload(decode=True).decode("UTF-8") + break + + if not text: + self.fail("Could not find text portion of email to parse") + + match = re.search(r"https://example.com\S+", text) + assert match, "Could not find link in email" + + return match.group(0) -- cgit 1.4.1 From 50ea178c201588b5e6b3f93e1af56aef0b4e8368 Mon Sep 17 00:00:00 2001 From: dklimpel <5740567+dklimpel@users.noreply.github.com> Date: Mon, 9 Mar 2020 19:57:04 +0100 Subject: lint --- tests/rest/client/v2_alpha/test_account.py | 18 ++++++++---------- 1 file changed, 8 insertions(+), 10 deletions(-) (limited to 'tests') diff --git a/tests/rest/client/v2_alpha/test_account.py b/tests/rest/client/v2_alpha/test_account.py index ac9f200de3..e178a53335 100644 --- a/tests/rest/client/v2_alpha/test_account.py +++ b/tests/rest/client/v2_alpha/test_account.py @@ -438,7 +438,9 @@ class ThreepidEmailRestTestCase(unittest.HomeserverTestCase): ) self.render(request) self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual("3PID changes disabled on this server", channel.json_body["error"]) + self.assertEqual( + "3PID changes disabled on this server", channel.json_body["error"] + ) # Get user request, channel = self.make_request( @@ -466,10 +468,7 @@ class ThreepidEmailRestTestCase(unittest.HomeserverTestCase): request, channel = self.make_request( "POST", b"account/3pid/delete", - { - "medium": "email", - "address": self.email - }, + {"medium": "email", "address": self.email}, access_token=self.user_id_tok, ) self.render(request) @@ -503,16 +502,15 @@ class ThreepidEmailRestTestCase(unittest.HomeserverTestCase): request, channel = self.make_request( "POST", b"account/3pid/delete", - { - "medium": "email", - "address": self.email - }, + {"medium": "email", "address": self.email}, access_token=self.user_id_tok, ) self.render(request) self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual("3PID changes disabled on this server", channel.json_body["error"]) + self.assertEqual( + "3PID changes disabled on this server", channel.json_body["error"] + ) # Get user request, channel = self.make_request( -- cgit 1.4.1 From 7e5f40e7716813f0d32e2efcb32df3c263fbfc63 Mon Sep 17 00:00:00 2001 From: dklimpel <5740567+dklimpel@users.noreply.github.com> Date: Mon, 9 Mar 2020 21:00:36 +0100 Subject: fix tests --- tests/handlers/test_profile.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'tests') diff --git a/tests/handlers/test_profile.py b/tests/handlers/test_profile.py index 98b508c3d4..f8c0da5ced 100644 --- a/tests/handlers/test_profile.py +++ b/tests/handlers/test_profile.py @@ -96,7 +96,7 @@ class ProfileTestCase(unittest.TestCase): self.hs.config.disable_set_displayname = True # Set first displayname is allowed, if displayname is null - self.store.set_profile_displayname(self.frank.localpart, "Frank") + yield self.store.set_profile_displayname(self.frank.localpart, "Frank") d = self.handler.set_displayname( self.frank, synapse.types.create_requester(self.frank), "Frank Jr." @@ -167,7 +167,7 @@ class ProfileTestCase(unittest.TestCase): self.hs.config.disable_set_avatar_url = True # Set first time avatar is allowed, if displayname is null - self.store.set_profile_avatar_url( + yield self.store.set_profile_avatar_url( self.frank.localpart, "http://my.server/me.png" ) -- cgit 1.4.1 From 885134529ffd95dd118d3228e69f0e3553f5a6a7 Mon Sep 17 00:00:00 2001 From: dklimpel <5740567+dklimpel@users.noreply.github.com> Date: Mon, 9 Mar 2020 22:09:29 +0100 Subject: updates after review --- changelog.d/7053.feature | 2 +- docs/sample_config.yaml | 10 +++++----- synapse/config/registration.py | 16 ++++++++-------- synapse/handlers/profile.py | 8 ++++---- synapse/rest/client/v2_alpha/account.py | 18 ++++++++++++------ tests/handlers/test_profile.py | 6 +++--- tests/rest/client/v2_alpha/test_account.py | 17 +++++++---------- 7 files changed, 40 insertions(+), 37 deletions(-) (limited to 'tests') diff --git a/changelog.d/7053.feature b/changelog.d/7053.feature index 79955b9780..00f47b2a14 100644 --- a/changelog.d/7053.feature +++ b/changelog.d/7053.feature @@ -1 +1 @@ -Add options to disable setting profile info for prevent changes. \ No newline at end of file +Add options to prevent users from changing their profile or associated 3PIDs. \ No newline at end of file diff --git a/docs/sample_config.yaml b/docs/sample_config.yaml index d3ecffac7d..8333800a10 100644 --- a/docs/sample_config.yaml +++ b/docs/sample_config.yaml @@ -1057,18 +1057,18 @@ account_threepid_delegates: #email: https://example.com # Delegate email sending to example.com #msisdn: http://localhost:8090 # Delegate SMS sending to this local process -# If enabled, don't let users set their own display names/avatars +# If disabled, don't let users set their own display names/avatars # other than for the very first time (unless they are a server admin). # Useful when provisioning users based on the contents of a 3rd party # directory and to avoid ambiguities. # -#disable_set_displayname: false -#disable_set_avatar_url: false +#enable_set_displayname: true +#enable_set_avatar_url: true -# If true, stop users from trying to change the 3PIDs associated with +# If false, stop users from trying to change the 3PIDs associated with # their accounts. # -#disable_3pid_changes: false +#enable_3pid_changes: true # Users who register on this homeserver will automatically be joined # to these rooms diff --git a/synapse/config/registration.py b/synapse/config/registration.py index 1abc0a79af..d4897ec9b6 100644 --- a/synapse/config/registration.py +++ b/synapse/config/registration.py @@ -129,9 +129,9 @@ class RegistrationConfig(Config): raise ConfigError("Invalid auto_join_rooms entry %s" % (room_alias,)) self.autocreate_auto_join_rooms = config.get("autocreate_auto_join_rooms", True) - self.disable_set_displayname = config.get("disable_set_displayname", False) - self.disable_set_avatar_url = config.get("disable_set_avatar_url", False) - self.disable_3pid_changes = config.get("disable_3pid_changes", False) + self.enable_set_displayname = config.get("enable_set_displayname", True) + self.enable_set_avatar_url = config.get("enable_set_avatar_url", True) + self.enable_3pid_changes = config.get("enable_3pid_changes", True) self.disable_msisdn_registration = config.get( "disable_msisdn_registration", False @@ -334,18 +334,18 @@ class RegistrationConfig(Config): #email: https://example.com # Delegate email sending to example.com #msisdn: http://localhost:8090 # Delegate SMS sending to this local process - # If enabled, don't let users set their own display names/avatars + # If disabled, don't let users set their own display names/avatars # other than for the very first time (unless they are a server admin). # Useful when provisioning users based on the contents of a 3rd party # directory and to avoid ambiguities. # - #disable_set_displayname: false - #disable_set_avatar_url: false + #enable_set_displayname: true + #enable_set_avatar_url: true - # If true, stop users from trying to change the 3PIDs associated with + # If false, stop users from trying to change the 3PIDs associated with # their accounts. # - #disable_3pid_changes: false + #enable_3pid_changes: true # Users who register on this homeserver will automatically be joined # to these rooms diff --git a/synapse/handlers/profile.py b/synapse/handlers/profile.py index b049dd8e26..eb85dba015 100644 --- a/synapse/handlers/profile.py +++ b/synapse/handlers/profile.py @@ -157,11 +157,11 @@ class BaseProfileHandler(BaseHandler): if not by_admin and target_user != requester.user: raise AuthError(400, "Cannot set another user's displayname") - if not by_admin and self.hs.config.disable_set_displayname: + if not by_admin and not self.hs.config.enable_set_displayname: profile = yield self.store.get_profileinfo(target_user.localpart) if profile.display_name: raise SynapseError( - 400, "Changing displayname is disabled on this server" + 400, "Changing display name is disabled on this server", Codes.FORBIDDEN ) if len(new_displayname) > MAX_DISPLAYNAME_LEN: @@ -225,11 +225,11 @@ class BaseProfileHandler(BaseHandler): if not by_admin and target_user != requester.user: raise AuthError(400, "Cannot set another user's avatar_url") - if not by_admin and self.hs.config.disable_set_avatar_url: + if not by_admin and not self.hs.config.enable_set_avatar_url: profile = yield self.store.get_profileinfo(target_user.localpart) if profile.avatar_url: raise SynapseError( - 400, "Changing avatar url is disabled on this server" + 400, "Changing avatar is disabled on this server", Codes.FORBIDDEN ) if len(new_avatar_url) > MAX_AVATAR_URL_LEN: diff --git a/synapse/rest/client/v2_alpha/account.py b/synapse/rest/client/v2_alpha/account.py index 97bddf36d9..e40136f2f3 100644 --- a/synapse/rest/client/v2_alpha/account.py +++ b/synapse/rest/client/v2_alpha/account.py @@ -599,8 +599,10 @@ class ThreepidRestServlet(RestServlet): return 200, {"threepids": threepids} async def on_POST(self, request): - if self.hs.config.disable_3pid_changes: - raise SynapseError(400, "3PID changes disabled on this server") + if not self.hs.config.enable_3pid_changes: + raise SynapseError( + 400, "3PID changes are disabled on this server", Codes.FORBIDDEN + ) requester = await self.auth.get_user_by_req(request) user_id = requester.user.to_string() @@ -646,8 +648,10 @@ class ThreepidAddRestServlet(RestServlet): @interactive_auth_handler async def on_POST(self, request): - if self.hs.config.disable_3pid_changes: - raise SynapseError(400, "3PID changes disabled on this server") + if not self.hs.config.enable_3pid_changes: + raise SynapseError( + 400, "3PID changes are disabled on this server", Codes.FORBIDDEN + ) requester = await self.auth.get_user_by_req(request) user_id = requester.user.to_string() @@ -749,8 +753,10 @@ class ThreepidDeleteRestServlet(RestServlet): self.auth_handler = hs.get_auth_handler() async def on_POST(self, request): - if self.hs.config.disable_3pid_changes: - raise SynapseError(400, "3PID changes disabled on this server") + if not self.hs.config.enable_3pid_changes: + raise SynapseError( + 400, "3PID changes are disabled on this server", Codes.FORBIDDEN + ) body = parse_json_object_from_request(request) assert_params_in_dict(body, ["medium", "address"]) diff --git a/tests/handlers/test_profile.py b/tests/handlers/test_profile.py index f8c0da5ced..e600b9777b 100644 --- a/tests/handlers/test_profile.py +++ b/tests/handlers/test_profile.py @@ -93,7 +93,7 @@ class ProfileTestCase(unittest.TestCase): @defer.inlineCallbacks def test_set_my_name_if_disabled(self): - self.hs.config.disable_set_displayname = True + self.hs.config.enable_set_displayname = False # Set first displayname is allowed, if displayname is null yield self.store.set_profile_displayname(self.frank.localpart, "Frank") @@ -164,9 +164,9 @@ class ProfileTestCase(unittest.TestCase): @defer.inlineCallbacks def test_set_my_avatar_if_disabled(self): - self.hs.config.disable_set_avatar_url = True + self.hs.config.enable_set_avatar_url = False - # Set first time avatar is allowed, if displayname is null + # Set first time avatar is allowed, if avatar is null yield self.store.set_profile_avatar_url( self.frank.localpart, "http://my.server/me.png" ) diff --git a/tests/rest/client/v2_alpha/test_account.py b/tests/rest/client/v2_alpha/test_account.py index e178a53335..34e40a36d0 100644 --- a/tests/rest/client/v2_alpha/test_account.py +++ b/tests/rest/client/v2_alpha/test_account.py @@ -24,6 +24,7 @@ import pkg_resources import synapse.rest.admin from synapse.api.constants import LoginType, Membership +from synapse.api.errors import Codes from synapse.rest.client.v1 import login, room from synapse.rest.client.v2_alpha import account, register @@ -412,7 +413,7 @@ class ThreepidEmailRestTestCase(unittest.HomeserverTestCase): def test_add_email_if_disabled(self): """Test add mail to profile if disabled """ - self.hs.config.disable_3pid_changes = True + self.hs.config.enable_3pid_changes = True client_secret = "foobar" session_id = self._request_token(self.email, client_secret) @@ -438,9 +439,7 @@ class ThreepidEmailRestTestCase(unittest.HomeserverTestCase): ) self.render(request) self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual( - "3PID changes disabled on this server", channel.json_body["error"] - ) + self.assertEqual(Codes.FORBIDDEN, channel.json_body["errcode"]) # Get user request, channel = self.make_request( @@ -486,7 +485,7 @@ class ThreepidEmailRestTestCase(unittest.HomeserverTestCase): def test_delete_email_if_disabled(self): """Test delete mail from profile if disabled """ - self.hs.config.disable_3pid_changes = True + self.hs.config.enable_3pid_changes = True # Add a threepid self.get_success( @@ -508,9 +507,7 @@ class ThreepidEmailRestTestCase(unittest.HomeserverTestCase): self.render(request) self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual( - "3PID changes disabled on this server", channel.json_body["error"] - ) + self.assertEqual(Codes.FORBIDDEN, channel.json_body["errcode"]) # Get user request, channel = self.make_request( @@ -547,7 +544,7 @@ class ThreepidEmailRestTestCase(unittest.HomeserverTestCase): ) self.render(request) self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual("No validated 3pid session found", channel.json_body["error"]) + self.assertEqual(Codes.THREEPID_AUTH_FAILED, channel.json_body["errcode"]) # Get user request, channel = self.make_request( @@ -582,7 +579,7 @@ class ThreepidEmailRestTestCase(unittest.HomeserverTestCase): ) self.render(request) self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual("No validated 3pid session found", channel.json_body["error"]) + self.assertEqual(Codes.THREEPID_AUTH_FAILED, channel.json_body["errcode"]) # Get user request, channel = self.make_request( -- cgit 1.4.1 From 39f6595b4ab108cb451072ae251a91117002191c Mon Sep 17 00:00:00 2001 From: dklimpel <5740567+dklimpel@users.noreply.github.com> Date: Mon, 9 Mar 2020 22:13:20 +0100 Subject: lint, fix tests --- synapse/handlers/profile.py | 4 +++- tests/rest/client/v2_alpha/test_account.py | 4 ++-- 2 files changed, 5 insertions(+), 3 deletions(-) (limited to 'tests') diff --git a/synapse/handlers/profile.py b/synapse/handlers/profile.py index eb85dba015..6aa1c0f5e0 100644 --- a/synapse/handlers/profile.py +++ b/synapse/handlers/profile.py @@ -161,7 +161,9 @@ class BaseProfileHandler(BaseHandler): profile = yield self.store.get_profileinfo(target_user.localpart) if profile.display_name: raise SynapseError( - 400, "Changing display name is disabled on this server", Codes.FORBIDDEN + 400, + "Changing display name is disabled on this server", + Codes.FORBIDDEN, ) if len(new_displayname) > MAX_DISPLAYNAME_LEN: diff --git a/tests/rest/client/v2_alpha/test_account.py b/tests/rest/client/v2_alpha/test_account.py index 34e40a36d0..99cc9163f3 100644 --- a/tests/rest/client/v2_alpha/test_account.py +++ b/tests/rest/client/v2_alpha/test_account.py @@ -413,7 +413,7 @@ class ThreepidEmailRestTestCase(unittest.HomeserverTestCase): def test_add_email_if_disabled(self): """Test add mail to profile if disabled """ - self.hs.config.enable_3pid_changes = True + self.hs.config.enable_3pid_changes = False client_secret = "foobar" session_id = self._request_token(self.email, client_secret) @@ -485,7 +485,7 @@ class ThreepidEmailRestTestCase(unittest.HomeserverTestCase): def test_delete_email_if_disabled(self): """Test delete mail from profile if disabled """ - self.hs.config.enable_3pid_changes = True + self.hs.config.enable_3pid_changes = False # Add a threepid self.get_success( -- cgit 1.4.1 From 6a35046363a6f5d41199256c80eef4ea7e385986 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Tue, 17 Mar 2020 11:25:01 +0000 Subject: Revert "Add options to disable setting profile info for prevent changes. (#7053)" This reverts commit 54dd28621b070ca67de9f773fe9a89e1f4dc19da, reversing changes made to 6640460d054e8f4444046a34bdf638921b31c01e. --- changelog.d/7053.feature | 1 - docs/sample_config.yaml | 13 -- synapse/config/registration.py | 17 -- synapse/handlers/profile.py | 16 -- synapse/rest/client/v2_alpha/account.py | 16 -- tests/handlers/test_profile.py | 33 +--- tests/rest/client/v2_alpha/test_account.py | 303 ----------------------------- 7 files changed, 1 insertion(+), 398 deletions(-) delete mode 100644 changelog.d/7053.feature (limited to 'tests') diff --git a/changelog.d/7053.feature b/changelog.d/7053.feature deleted file mode 100644 index 00f47b2a14..0000000000 --- a/changelog.d/7053.feature +++ /dev/null @@ -1 +0,0 @@ -Add options to prevent users from changing their profile or associated 3PIDs. \ No newline at end of file diff --git a/docs/sample_config.yaml b/docs/sample_config.yaml index 91eff4c8ad..2ff0dd05a2 100644 --- a/docs/sample_config.yaml +++ b/docs/sample_config.yaml @@ -1057,19 +1057,6 @@ account_threepid_delegates: #email: https://example.com # Delegate email sending to example.com #msisdn: http://localhost:8090 # Delegate SMS sending to this local process -# If disabled, don't let users set their own display names/avatars -# (unless they are a server admin) other than for the very first time. -# Useful when provisioning users based on the contents of a 3rd party -# directory and to avoid ambiguities. -# -#enable_set_displayname: true -#enable_set_avatar_url: true - -# If false, stop users from trying to change the 3PIDs associated with -# their accounts. -# -#enable_3pid_changes: true - # Users who register on this homeserver will automatically be joined # to these rooms # diff --git a/synapse/config/registration.py b/synapse/config/registration.py index ee737eb40d..9bb3beedbc 100644 --- a/synapse/config/registration.py +++ b/synapse/config/registration.py @@ -129,10 +129,6 @@ class RegistrationConfig(Config): raise ConfigError("Invalid auto_join_rooms entry %s" % (room_alias,)) self.autocreate_auto_join_rooms = config.get("autocreate_auto_join_rooms", True) - self.enable_set_displayname = config.get("enable_set_displayname", True) - self.enable_set_avatar_url = config.get("enable_set_avatar_url", True) - self.enable_3pid_changes = config.get("enable_3pid_changes", True) - self.disable_msisdn_registration = config.get( "disable_msisdn_registration", False ) @@ -334,19 +330,6 @@ class RegistrationConfig(Config): #email: https://example.com # Delegate email sending to example.com #msisdn: http://localhost:8090 # Delegate SMS sending to this local process - # If disabled, don't let users set their own display names/avatars - # (unless they are a server admin) other than for the very first time. - # Useful when provisioning users based on the contents of a 3rd party - # directory and to avoid ambiguities. - # - #enable_set_displayname: true - #enable_set_avatar_url: true - - # If false, stop users from trying to change the 3PIDs associated with - # their accounts. - # - #enable_3pid_changes: true - # Users who register on this homeserver will automatically be joined # to these rooms # diff --git a/synapse/handlers/profile.py b/synapse/handlers/profile.py index 6aa1c0f5e0..50ce0c585b 100644 --- a/synapse/handlers/profile.py +++ b/synapse/handlers/profile.py @@ -157,15 +157,6 @@ class BaseProfileHandler(BaseHandler): if not by_admin and target_user != requester.user: raise AuthError(400, "Cannot set another user's displayname") - if not by_admin and not self.hs.config.enable_set_displayname: - profile = yield self.store.get_profileinfo(target_user.localpart) - if profile.display_name: - raise SynapseError( - 400, - "Changing display name is disabled on this server", - Codes.FORBIDDEN, - ) - if len(new_displayname) > MAX_DISPLAYNAME_LEN: raise SynapseError( 400, "Displayname is too long (max %i)" % (MAX_DISPLAYNAME_LEN,) @@ -227,13 +218,6 @@ class BaseProfileHandler(BaseHandler): if not by_admin and target_user != requester.user: raise AuthError(400, "Cannot set another user's avatar_url") - if not by_admin and not self.hs.config.enable_set_avatar_url: - profile = yield self.store.get_profileinfo(target_user.localpart) - if profile.avatar_url: - raise SynapseError( - 400, "Changing avatar is disabled on this server", Codes.FORBIDDEN - ) - if len(new_avatar_url) > MAX_AVATAR_URL_LEN: raise SynapseError( 400, "Avatar URL is too long (max %i)" % (MAX_AVATAR_URL_LEN,) diff --git a/synapse/rest/client/v2_alpha/account.py b/synapse/rest/client/v2_alpha/account.py index e40136f2f3..dc837d6c75 100644 --- a/synapse/rest/client/v2_alpha/account.py +++ b/synapse/rest/client/v2_alpha/account.py @@ -599,11 +599,6 @@ class ThreepidRestServlet(RestServlet): return 200, {"threepids": threepids} async def on_POST(self, request): - if not self.hs.config.enable_3pid_changes: - raise SynapseError( - 400, "3PID changes are disabled on this server", Codes.FORBIDDEN - ) - requester = await self.auth.get_user_by_req(request) user_id = requester.user.to_string() body = parse_json_object_from_request(request) @@ -648,11 +643,6 @@ class ThreepidAddRestServlet(RestServlet): @interactive_auth_handler async def on_POST(self, request): - if not self.hs.config.enable_3pid_changes: - raise SynapseError( - 400, "3PID changes are disabled on this server", Codes.FORBIDDEN - ) - requester = await self.auth.get_user_by_req(request) user_id = requester.user.to_string() body = parse_json_object_from_request(request) @@ -748,16 +738,10 @@ class ThreepidDeleteRestServlet(RestServlet): def __init__(self, hs): super(ThreepidDeleteRestServlet, self).__init__() - self.hs = hs self.auth = hs.get_auth() self.auth_handler = hs.get_auth_handler() async def on_POST(self, request): - if not self.hs.config.enable_3pid_changes: - raise SynapseError( - 400, "3PID changes are disabled on this server", Codes.FORBIDDEN - ) - body = parse_json_object_from_request(request) assert_params_in_dict(body, ["medium", "address"]) diff --git a/tests/handlers/test_profile.py b/tests/handlers/test_profile.py index e600b9777b..d60c124eec 100644 --- a/tests/handlers/test_profile.py +++ b/tests/handlers/test_profile.py @@ -19,7 +19,7 @@ from mock import Mock, NonCallableMock from twisted.internet import defer import synapse.types -from synapse.api.errors import AuthError, SynapseError +from synapse.api.errors import AuthError from synapse.handlers.profile import MasterProfileHandler from synapse.types import UserID @@ -70,7 +70,6 @@ class ProfileTestCase(unittest.TestCase): yield self.store.create_profile(self.frank.localpart) self.handler = hs.get_profile_handler() - self.hs = hs @defer.inlineCallbacks def test_get_my_name(self): @@ -91,19 +90,6 @@ class ProfileTestCase(unittest.TestCase): "Frank Jr.", ) - @defer.inlineCallbacks - def test_set_my_name_if_disabled(self): - self.hs.config.enable_set_displayname = False - - # Set first displayname is allowed, if displayname is null - yield self.store.set_profile_displayname(self.frank.localpart, "Frank") - - d = self.handler.set_displayname( - self.frank, synapse.types.create_requester(self.frank), "Frank Jr." - ) - - yield self.assertFailure(d, SynapseError) - @defer.inlineCallbacks def test_set_my_name_noauth(self): d = self.handler.set_displayname( @@ -161,20 +147,3 @@ class ProfileTestCase(unittest.TestCase): (yield self.store.get_profile_avatar_url(self.frank.localpart)), "http://my.server/pic.gif", ) - - @defer.inlineCallbacks - def test_set_my_avatar_if_disabled(self): - self.hs.config.enable_set_avatar_url = False - - # Set first time avatar is allowed, if avatar is null - yield self.store.set_profile_avatar_url( - self.frank.localpart, "http://my.server/me.png" - ) - - d = self.handler.set_avatar_url( - self.frank, - synapse.types.create_requester(self.frank), - "http://my.server/pic.gif", - ) - - yield self.assertFailure(d, SynapseError) diff --git a/tests/rest/client/v2_alpha/test_account.py b/tests/rest/client/v2_alpha/test_account.py index 99cc9163f3..c3facc00eb 100644 --- a/tests/rest/client/v2_alpha/test_account.py +++ b/tests/rest/client/v2_alpha/test_account.py @@ -24,7 +24,6 @@ import pkg_resources import synapse.rest.admin from synapse.api.constants import LoginType, Membership -from synapse.api.errors import Codes from synapse.rest.client.v1 import login, room from synapse.rest.client.v2_alpha import account, register @@ -326,305 +325,3 @@ class DeactivateTestCase(unittest.HomeserverTestCase): ) self.render(request) self.assertEqual(request.code, 200) - - -class ThreepidEmailRestTestCase(unittest.HomeserverTestCase): - - servlets = [ - account.register_servlets, - login.register_servlets, - synapse.rest.admin.register_servlets_for_client_rest_resource, - ] - - def make_homeserver(self, reactor, clock): - config = self.default_config() - - # Email config. - self.email_attempts = [] - - def sendmail(smtphost, from_addr, to_addrs, msg, **kwargs): - self.email_attempts.append(msg) - return - - config["email"] = { - "enable_notifs": False, - "template_dir": os.path.abspath( - pkg_resources.resource_filename("synapse", "res/templates") - ), - "smtp_host": "127.0.0.1", - "smtp_port": 20, - "require_transport_security": False, - "smtp_user": None, - "smtp_pass": None, - "notif_from": "test@example.com", - } - config["public_baseurl"] = "https://example.com" - - self.hs = self.setup_test_homeserver(config=config, sendmail=sendmail) - return self.hs - - def prepare(self, reactor, clock, hs): - self.store = hs.get_datastore() - - self.user_id = self.register_user("kermit", "test") - self.user_id_tok = self.login("kermit", "test") - self.email = "test@example.com" - self.url_3pid = b"account/3pid" - - def test_add_email(self): - """Test add mail to profile - """ - client_secret = "foobar" - session_id = self._request_token(self.email, client_secret) - - self.assertEquals(len(self.email_attempts), 1) - link = self._get_link_from_email() - - self._validate_token(link) - - request, channel = self.make_request( - "POST", - b"/_matrix/client/unstable/account/3pid/add", - { - "client_secret": client_secret, - "sid": session_id, - "auth": { - "type": "m.login.password", - "user": self.user_id, - "password": "test", - }, - }, - access_token=self.user_id_tok, - ) - - self.render(request) - self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) - - # Get user - request, channel = self.make_request( - "GET", self.url_3pid, access_token=self.user_id_tok, - ) - self.render(request) - - self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual("email", channel.json_body["threepids"][0]["medium"]) - self.assertEqual(self.email, channel.json_body["threepids"][0]["address"]) - - def test_add_email_if_disabled(self): - """Test add mail to profile if disabled - """ - self.hs.config.enable_3pid_changes = False - - client_secret = "foobar" - session_id = self._request_token(self.email, client_secret) - - self.assertEquals(len(self.email_attempts), 1) - link = self._get_link_from_email() - - self._validate_token(link) - - request, channel = self.make_request( - "POST", - b"/_matrix/client/unstable/account/3pid/add", - { - "client_secret": client_secret, - "sid": session_id, - "auth": { - "type": "m.login.password", - "user": self.user_id, - "password": "test", - }, - }, - access_token=self.user_id_tok, - ) - self.render(request) - self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual(Codes.FORBIDDEN, channel.json_body["errcode"]) - - # Get user - request, channel = self.make_request( - "GET", self.url_3pid, access_token=self.user_id_tok, - ) - self.render(request) - - self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) - self.assertFalse(channel.json_body["threepids"]) - - def test_delete_email(self): - """Test delete mail from profile - """ - # Add a threepid - self.get_success( - self.store.user_add_threepid( - user_id=self.user_id, - medium="email", - address=self.email, - validated_at=0, - added_at=0, - ) - ) - - request, channel = self.make_request( - "POST", - b"account/3pid/delete", - {"medium": "email", "address": self.email}, - access_token=self.user_id_tok, - ) - self.render(request) - self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) - - # Get user - request, channel = self.make_request( - "GET", self.url_3pid, access_token=self.user_id_tok, - ) - self.render(request) - - self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) - self.assertFalse(channel.json_body["threepids"]) - - def test_delete_email_if_disabled(self): - """Test delete mail from profile if disabled - """ - self.hs.config.enable_3pid_changes = False - - # Add a threepid - self.get_success( - self.store.user_add_threepid( - user_id=self.user_id, - medium="email", - address=self.email, - validated_at=0, - added_at=0, - ) - ) - - request, channel = self.make_request( - "POST", - b"account/3pid/delete", - {"medium": "email", "address": self.email}, - access_token=self.user_id_tok, - ) - self.render(request) - - self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual(Codes.FORBIDDEN, channel.json_body["errcode"]) - - # Get user - request, channel = self.make_request( - "GET", self.url_3pid, access_token=self.user_id_tok, - ) - self.render(request) - - self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual("email", channel.json_body["threepids"][0]["medium"]) - self.assertEqual(self.email, channel.json_body["threepids"][0]["address"]) - - def test_cant_add_email_without_clicking_link(self): - """Test that we do actually need to click the link in the email - """ - client_secret = "foobar" - session_id = self._request_token(self.email, client_secret) - - self.assertEquals(len(self.email_attempts), 1) - - # Attempt to add email without clicking the link - request, channel = self.make_request( - "POST", - b"/_matrix/client/unstable/account/3pid/add", - { - "client_secret": client_secret, - "sid": session_id, - "auth": { - "type": "m.login.password", - "user": self.user_id, - "password": "test", - }, - }, - access_token=self.user_id_tok, - ) - self.render(request) - self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual(Codes.THREEPID_AUTH_FAILED, channel.json_body["errcode"]) - - # Get user - request, channel = self.make_request( - "GET", self.url_3pid, access_token=self.user_id_tok, - ) - self.render(request) - - self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) - self.assertFalse(channel.json_body["threepids"]) - - def test_no_valid_token(self): - """Test that we do actually need to request a token and can't just - make a session up. - """ - client_secret = "foobar" - session_id = "weasle" - - # Attempt to add email without even requesting an email - request, channel = self.make_request( - "POST", - b"/_matrix/client/unstable/account/3pid/add", - { - "client_secret": client_secret, - "sid": session_id, - "auth": { - "type": "m.login.password", - "user": self.user_id, - "password": "test", - }, - }, - access_token=self.user_id_tok, - ) - self.render(request) - self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) - self.assertEqual(Codes.THREEPID_AUTH_FAILED, channel.json_body["errcode"]) - - # Get user - request, channel = self.make_request( - "GET", self.url_3pid, access_token=self.user_id_tok, - ) - self.render(request) - - self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) - self.assertFalse(channel.json_body["threepids"]) - - def _request_token(self, email, client_secret): - request, channel = self.make_request( - "POST", - b"account/3pid/email/requestToken", - {"client_secret": client_secret, "email": email, "send_attempt": 1}, - ) - self.render(request) - self.assertEquals(200, channel.code, channel.result) - - return channel.json_body["sid"] - - def _validate_token(self, link): - # Remove the host - path = link.replace("https://example.com", "") - - request, channel = self.make_request("GET", path, shorthand=False) - self.render(request) - self.assertEquals(200, channel.code, channel.result) - - def _get_link_from_email(self): - assert self.email_attempts, "No emails have been sent" - - raw_msg = self.email_attempts[-1].decode("UTF-8") - mail = Parser().parsestr(raw_msg) - - text = None - for part in mail.walk(): - if part.get_content_type() == "text/plain": - text = part.get_payload(decode=True).decode("UTF-8") - break - - if not text: - self.fail("Could not find text portion of email to parse") - - match = re.search(r"https://example.com\S+", text) - assert match, "Could not find link in email" - - return match.group(0) -- cgit 1.4.1 From 60724c46b7dc5300243fd97d5a485564b3e00afe Mon Sep 17 00:00:00 2001 From: Patrick Cloke Date: Tue, 17 Mar 2020 07:37:04 -0400 Subject: Remove special casing of `m.room.aliases` events (#7034) --- changelog.d/7034.removal | 1 + synapse/handlers/room.py | 16 +------------ synapse/rest/client/v1/room.py | 12 ---------- tests/rest/admin/test_admin.py | 7 ++++++ tests/rest/client/v1/test_directory.py | 41 +++++++++++++++++++++------------- 5 files changed, 35 insertions(+), 42 deletions(-) create mode 100644 changelog.d/7034.removal (limited to 'tests') diff --git a/changelog.d/7034.removal b/changelog.d/7034.removal new file mode 100644 index 0000000000..be8d20e14f --- /dev/null +++ b/changelog.d/7034.removal @@ -0,0 +1 @@ +Remove special handling of aliases events from [MSC2260](https://github.com/matrix-org/matrix-doc/pull/2260) added in v1.10.0rc1. diff --git a/synapse/handlers/room.py b/synapse/handlers/room.py index 8ee870f0bb..f580ab2e9f 100644 --- a/synapse/handlers/room.py +++ b/synapse/handlers/room.py @@ -292,16 +292,6 @@ class RoomCreationHandler(BaseHandler): except AuthError as e: logger.warning("Unable to update PLs in old room: %s", e) - new_pl_content = copy_power_levels_contents(old_room_pl_state.content) - - # pre-msc2260 rooms may not have the right setting for aliases. If no other - # value is set, set it now. - events_default = new_pl_content.get("events_default", 0) - new_pl_content.setdefault("events", {}).setdefault( - EventTypes.Aliases, events_default - ) - - logger.debug("Setting correct PLs in new room to %s", new_pl_content) yield self.event_creation_handler.create_and_send_nonmember_event( requester, { @@ -309,7 +299,7 @@ class RoomCreationHandler(BaseHandler): "state_key": "", "room_id": new_room_id, "sender": requester.user.to_string(), - "content": new_pl_content, + "content": old_room_pl_state.content, }, ratelimit=False, ) @@ -814,10 +804,6 @@ class RoomCreationHandler(BaseHandler): EventTypes.RoomHistoryVisibility: 100, EventTypes.CanonicalAlias: 50, EventTypes.RoomAvatar: 50, - # MSC2260: Allow everybody to send alias events by default - # This will be reudundant on pre-MSC2260 rooms, since the - # aliases event is special-cased. - EventTypes.Aliases: 0, EventTypes.Tombstone: 100, EventTypes.ServerACL: 100, }, diff --git a/synapse/rest/client/v1/room.py b/synapse/rest/client/v1/room.py index 64f51406fb..bffd43de5f 100644 --- a/synapse/rest/client/v1/room.py +++ b/synapse/rest/client/v1/room.py @@ -189,12 +189,6 @@ class RoomStateEventRestServlet(TransactionRestServlet): content = parse_json_object_from_request(request) - if event_type == EventTypes.Aliases: - # MSC2260 - raise SynapseError( - 400, "Cannot send m.room.aliases events via /rooms/{room_id}/state" - ) - event_dict = { "type": event_type, "content": content, @@ -242,12 +236,6 @@ class RoomSendEventRestServlet(TransactionRestServlet): requester = await self.auth.get_user_by_req(request, allow_guest=True) content = parse_json_object_from_request(request) - if event_type == EventTypes.Aliases: - # MSC2260 - raise SynapseError( - 400, "Cannot send m.room.aliases events via /rooms/{room_id}/send" - ) - event_dict = { "type": event_type, "content": content, diff --git a/tests/rest/admin/test_admin.py b/tests/rest/admin/test_admin.py index e5984aaad8..0342aed416 100644 --- a/tests/rest/admin/test_admin.py +++ b/tests/rest/admin/test_admin.py @@ -868,6 +868,13 @@ class RoomTestCase(unittest.HomeserverTestCase): self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) # Set this new alias as the canonical alias for this room + self.helper.send_state( + room_id, + "m.room.aliases", + {"aliases": [test_alias]}, + tok=self.admin_user_tok, + state_key="test", + ) self.helper.send_state( room_id, "m.room.canonical_alias", diff --git a/tests/rest/client/v1/test_directory.py b/tests/rest/client/v1/test_directory.py index 914cf54927..633b7dbda0 100644 --- a/tests/rest/client/v1/test_directory.py +++ b/tests/rest/client/v1/test_directory.py @@ -51,30 +51,26 @@ class DirectoryTestCase(unittest.HomeserverTestCase): self.user = self.register_user("user", "test") self.user_tok = self.login("user", "test") - def test_cannot_set_alias_via_state_event(self): - self.ensure_user_joined_room() - url = "/_matrix/client/r0/rooms/%s/state/m.room.aliases/%s" % ( - self.room_id, - self.hs.hostname, - ) - - data = {"aliases": [self.random_alias(5)]} - request_data = json.dumps(data) - - request, channel = self.make_request( - "PUT", url, request_data, access_token=self.user_tok - ) - self.render(request) - self.assertEqual(channel.code, 400, channel.result) + def test_state_event_not_in_room(self): + self.ensure_user_left_room() + self.set_alias_via_state_event(403) def test_directory_endpoint_not_in_room(self): self.ensure_user_left_room() self.set_alias_via_directory(403) + def test_state_event_in_room_too_long(self): + self.ensure_user_joined_room() + self.set_alias_via_state_event(400, alias_length=256) + def test_directory_in_room_too_long(self): self.ensure_user_joined_room() self.set_alias_via_directory(400, alias_length=256) + def test_state_event_in_room(self): + self.ensure_user_joined_room() + self.set_alias_via_state_event(200) + def test_directory_in_room(self): self.ensure_user_joined_room() self.set_alias_via_directory(200) @@ -106,6 +102,21 @@ class DirectoryTestCase(unittest.HomeserverTestCase): self.render(request) self.assertEqual(channel.code, 200, channel.result) + def set_alias_via_state_event(self, expected_code, alias_length=5): + url = "/_matrix/client/r0/rooms/%s/state/m.room.aliases/%s" % ( + self.room_id, + self.hs.hostname, + ) + + data = {"aliases": [self.random_alias(alias_length)]} + request_data = json.dumps(data) + + request, channel = self.make_request( + "PUT", url, request_data, access_token=self.user_tok + ) + self.render(request) + self.assertEqual(channel.code, expected_code, channel.result) + def set_alias_via_directory(self, expected_code, alias_length=5): url = "/_matrix/client/r0/directory/room/%s" % self.random_alias(alias_length) data = {"room_id": self.room_id} -- cgit 1.4.1 From c37db0211e36cd298426ff8811e547b0acd10bf4 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Tue, 17 Mar 2020 22:32:25 +0100 Subject: Share SSL contexts for non-federation requests (#7094) Extends #5794 etc to the SimpleHttpClient so that it also applies to non-federation requests. Fixes #7092. --- changelog.d/7094.misc | 1 + synapse/crypto/context_factory.py | 68 ++++++++++++++-------- synapse/http/client.py | 3 - synapse/http/federation/matrix_federation_agent.py | 2 +- synapse/server.py | 6 +- tests/config/test_tls.py | 29 +++++---- .../federation/test_matrix_federation_agent.py | 6 +- 7 files changed, 71 insertions(+), 44 deletions(-) create mode 100644 changelog.d/7094.misc (limited to 'tests') diff --git a/changelog.d/7094.misc b/changelog.d/7094.misc new file mode 100644 index 0000000000..aa093ee3c0 --- /dev/null +++ b/changelog.d/7094.misc @@ -0,0 +1 @@ +Improve performance when making HTTPS requests to sygnal, sydent, etc, by sharing the SSL context object between connections. diff --git a/synapse/crypto/context_factory.py b/synapse/crypto/context_factory.py index e93f0b3705..a5a2a7815d 100644 --- a/synapse/crypto/context_factory.py +++ b/synapse/crypto/context_factory.py @@ -75,7 +75,7 @@ class ServerContextFactory(ContextFactory): @implementer(IPolicyForHTTPS) -class ClientTLSOptionsFactory(object): +class FederationPolicyForHTTPS(object): """Factory for Twisted SSLClientConnectionCreators that are used to make connections to remote servers for federation. @@ -103,15 +103,15 @@ class ClientTLSOptionsFactory(object): # let us do). minTLS = _TLS_VERSION_MAP[config.federation_client_minimum_tls_version] - self._verify_ssl = CertificateOptions( + _verify_ssl = CertificateOptions( trustRoot=trust_root, insecurelyLowerMinimumTo=minTLS ) - self._verify_ssl_context = self._verify_ssl.getContext() - self._verify_ssl_context.set_info_callback(self._context_info_cb) + self._verify_ssl_context = _verify_ssl.getContext() + self._verify_ssl_context.set_info_callback(_context_info_cb) - self._no_verify_ssl = CertificateOptions(insecurelyLowerMinimumTo=minTLS) - self._no_verify_ssl_context = self._no_verify_ssl.getContext() - self._no_verify_ssl_context.set_info_callback(self._context_info_cb) + _no_verify_ssl = CertificateOptions(insecurelyLowerMinimumTo=minTLS) + self._no_verify_ssl_context = _no_verify_ssl.getContext() + self._no_verify_ssl_context.set_info_callback(_context_info_cb) def get_options(self, host: bytes): @@ -136,23 +136,6 @@ class ClientTLSOptionsFactory(object): return SSLClientConnectionCreator(host, ssl_context, should_verify) - @staticmethod - def _context_info_cb(ssl_connection, where, ret): - """The 'information callback' for our openssl context object.""" - # we assume that the app_data on the connection object has been set to - # a TLSMemoryBIOProtocol object. (This is done by SSLClientConnectionCreator) - tls_protocol = ssl_connection.get_app_data() - try: - # ... we further assume that SSLClientConnectionCreator has set the - # '_synapse_tls_verifier' attribute to a ConnectionVerifier object. - tls_protocol._synapse_tls_verifier.verify_context_info_cb( - ssl_connection, where - ) - except: # noqa: E722, taken from the twisted implementation - logger.exception("Error during info_callback") - f = Failure() - tls_protocol.failVerification(f) - def creatorForNetloc(self, hostname, port): """Implements the IPolicyForHTTPS interace so that this can be passed directly to agents. @@ -160,6 +143,43 @@ class ClientTLSOptionsFactory(object): return self.get_options(hostname) +@implementer(IPolicyForHTTPS) +class RegularPolicyForHTTPS(object): + """Factory for Twisted SSLClientConnectionCreators that are used to make connections + to remote servers, for other than federation. + + Always uses the same OpenSSL context object, which uses the default OpenSSL CA + trust root. + """ + + def __init__(self): + trust_root = platformTrust() + self._ssl_context = CertificateOptions(trustRoot=trust_root).getContext() + self._ssl_context.set_info_callback(_context_info_cb) + + def creatorForNetloc(self, hostname, port): + return SSLClientConnectionCreator(hostname, self._ssl_context, True) + + +def _context_info_cb(ssl_connection, where, ret): + """The 'information callback' for our openssl context objects. + + Note: Once this is set as the info callback on a Context object, the Context should + only be used with the SSLClientConnectionCreator. + """ + # we assume that the app_data on the connection object has been set to + # a TLSMemoryBIOProtocol object. (This is done by SSLClientConnectionCreator) + tls_protocol = ssl_connection.get_app_data() + try: + # ... we further assume that SSLClientConnectionCreator has set the + # '_synapse_tls_verifier' attribute to a ConnectionVerifier object. + tls_protocol._synapse_tls_verifier.verify_context_info_cb(ssl_connection, where) + except: # noqa: E722, taken from the twisted implementation + logger.exception("Error during info_callback") + f = Failure() + tls_protocol.failVerification(f) + + @implementer(IOpenSSLClientConnectionCreator) class SSLClientConnectionCreator(object): """Creates openssl connection objects for client connections. diff --git a/synapse/http/client.py b/synapse/http/client.py index d4c285445e..3797545824 100644 --- a/synapse/http/client.py +++ b/synapse/http/client.py @@ -244,9 +244,6 @@ class SimpleHttpClient(object): pool.maxPersistentPerHost = max((100 * CACHE_SIZE_FACTOR, 5)) pool.cachedConnectionTimeout = 2 * 60 - # The default context factory in Twisted 14.0.0 (which we require) is - # BrowserLikePolicyForHTTPS which will do regular cert validation - # 'like a browser' self.agent = ProxyAgent( self.reactor, connectTimeout=15, diff --git a/synapse/http/federation/matrix_federation_agent.py b/synapse/http/federation/matrix_federation_agent.py index 647d26dc56..f5f917f5ae 100644 --- a/synapse/http/federation/matrix_federation_agent.py +++ b/synapse/http/federation/matrix_federation_agent.py @@ -45,7 +45,7 @@ class MatrixFederationAgent(object): Args: reactor (IReactor): twisted reactor to use for underlying requests - tls_client_options_factory (ClientTLSOptionsFactory|None): + tls_client_options_factory (FederationPolicyForHTTPS|None): factory to use for fetching client tls options, or none to disable TLS. _srv_resolver (SrvResolver|None): diff --git a/synapse/server.py b/synapse/server.py index fd2f69e928..1b980371de 100644 --- a/synapse/server.py +++ b/synapse/server.py @@ -26,7 +26,6 @@ import logging import os from twisted.mail.smtp import sendmail -from twisted.web.client import BrowserLikePolicyForHTTPS from synapse.api.auth import Auth from synapse.api.filtering import Filtering @@ -35,6 +34,7 @@ from synapse.appservice.api import ApplicationServiceApi from synapse.appservice.scheduler import ApplicationServiceScheduler from synapse.config.homeserver import HomeServerConfig from synapse.crypto import context_factory +from synapse.crypto.context_factory import RegularPolicyForHTTPS from synapse.crypto.keyring import Keyring from synapse.events.builder import EventBuilderFactory from synapse.events.spamcheck import SpamChecker @@ -310,7 +310,7 @@ class HomeServer(object): return ( InsecureInterceptableContextFactory() if self.config.use_insecure_ssl_client_just_for_testing_do_not_use - else BrowserLikePolicyForHTTPS() + else RegularPolicyForHTTPS() ) def build_simple_http_client(self): @@ -420,7 +420,7 @@ class HomeServer(object): return PusherPool(self) def build_http_client(self): - tls_client_options_factory = context_factory.ClientTLSOptionsFactory( + tls_client_options_factory = context_factory.FederationPolicyForHTTPS( self.config ) return MatrixFederationHttpClient(self, tls_client_options_factory) diff --git a/tests/config/test_tls.py b/tests/config/test_tls.py index 1be6ff563b..ec32d4b1ca 100644 --- a/tests/config/test_tls.py +++ b/tests/config/test_tls.py @@ -23,7 +23,7 @@ from OpenSSL import SSL from synapse.config._base import Config, RootConfig from synapse.config.tls import ConfigError, TlsConfig -from synapse.crypto.context_factory import ClientTLSOptionsFactory +from synapse.crypto.context_factory import FederationPolicyForHTTPS from tests.unittest import TestCase @@ -180,12 +180,13 @@ s4niecZKPBizL6aucT59CsunNmmb5Glq8rlAcU+1ZTZZzGYqVYhF6axB9Qg= t = TestConfig() t.read_config(config, config_dir_path="", data_dir_path="") - cf = ClientTLSOptionsFactory(t) + cf = FederationPolicyForHTTPS(t) + options = _get_ssl_context_options(cf._verify_ssl_context) # The context has had NO_TLSv1_1 and NO_TLSv1_0 set, but not NO_TLSv1_2 - self.assertNotEqual(cf._verify_ssl._options & SSL.OP_NO_TLSv1, 0) - self.assertNotEqual(cf._verify_ssl._options & SSL.OP_NO_TLSv1_1, 0) - self.assertEqual(cf._verify_ssl._options & SSL.OP_NO_TLSv1_2, 0) + self.assertNotEqual(options & SSL.OP_NO_TLSv1, 0) + self.assertNotEqual(options & SSL.OP_NO_TLSv1_1, 0) + self.assertEqual(options & SSL.OP_NO_TLSv1_2, 0) def test_tls_client_minimum_set_passed_through_1_0(self): """ @@ -195,12 +196,13 @@ s4niecZKPBizL6aucT59CsunNmmb5Glq8rlAcU+1ZTZZzGYqVYhF6axB9Qg= t = TestConfig() t.read_config(config, config_dir_path="", data_dir_path="") - cf = ClientTLSOptionsFactory(t) + cf = FederationPolicyForHTTPS(t) + options = _get_ssl_context_options(cf._verify_ssl_context) # The context has not had any of the NO_TLS set. - self.assertEqual(cf._verify_ssl._options & SSL.OP_NO_TLSv1, 0) - self.assertEqual(cf._verify_ssl._options & SSL.OP_NO_TLSv1_1, 0) - self.assertEqual(cf._verify_ssl._options & SSL.OP_NO_TLSv1_2, 0) + self.assertEqual(options & SSL.OP_NO_TLSv1, 0) + self.assertEqual(options & SSL.OP_NO_TLSv1_1, 0) + self.assertEqual(options & SSL.OP_NO_TLSv1_2, 0) def test_acme_disabled_in_generated_config_no_acme_domain_provied(self): """ @@ -273,7 +275,7 @@ s4niecZKPBizL6aucT59CsunNmmb5Glq8rlAcU+1ZTZZzGYqVYhF6axB9Qg= t = TestConfig() t.read_config(config, config_dir_path="", data_dir_path="") - cf = ClientTLSOptionsFactory(t) + cf = FederationPolicyForHTTPS(t) # Not in the whitelist opts = cf.get_options(b"notexample.com") @@ -282,3 +284,10 @@ s4niecZKPBizL6aucT59CsunNmmb5Glq8rlAcU+1ZTZZzGYqVYhF6axB9Qg= # Caught by the wildcard opts = cf.get_options(idna.encode("テスト.ドメイン.テスト")) self.assertFalse(opts._verifier._verify_certs) + + +def _get_ssl_context_options(ssl_context: SSL.Context) -> int: + """get the options bits from an openssl context object""" + # the OpenSSL.SSL.Context wrapper doesn't expose get_options, so we have to + # use the low-level interface + return SSL._lib.SSL_CTX_get_options(ssl_context._context) diff --git a/tests/http/federation/test_matrix_federation_agent.py b/tests/http/federation/test_matrix_federation_agent.py index cfcd98ff7d..fdc1d918ff 100644 --- a/tests/http/federation/test_matrix_federation_agent.py +++ b/tests/http/federation/test_matrix_federation_agent.py @@ -31,7 +31,7 @@ from twisted.web.http_headers import Headers from twisted.web.iweb import IPolicyForHTTPS from synapse.config.homeserver import HomeServerConfig -from synapse.crypto.context_factory import ClientTLSOptionsFactory +from synapse.crypto.context_factory import FederationPolicyForHTTPS from synapse.http.federation.matrix_federation_agent import MatrixFederationAgent from synapse.http.federation.srv_resolver import Server from synapse.http.federation.well_known_resolver import ( @@ -79,7 +79,7 @@ class MatrixFederationAgentTests(unittest.TestCase): self._config = config = HomeServerConfig() config.parse_config_dict(config_dict, "", "") - self.tls_factory = ClientTLSOptionsFactory(config) + self.tls_factory = FederationPolicyForHTTPS(config) self.well_known_cache = TTLCache("test_cache", timer=self.reactor.seconds) self.had_well_known_cache = TTLCache("test_cache", timer=self.reactor.seconds) @@ -715,7 +715,7 @@ class MatrixFederationAgentTests(unittest.TestCase): config = default_config("test", parse=True) # Build a new agent and WellKnownResolver with a different tls factory - tls_factory = ClientTLSOptionsFactory(config) + tls_factory = FederationPolicyForHTTPS(config) agent = MatrixFederationAgent( reactor=self.reactor, tls_client_options_factory=tls_factory, -- cgit 1.4.1 From 4a17a647a9508b70de35130fd82e3e21474270a9 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 18 Mar 2020 16:46:41 +0000 Subject: Improve get auth chain difference algorithm. (#7095) It was originally implemented by pulling the full auth chain of all state sets out of the database and doing set comparison. However, that can take a lot work if the state and auth chains are large. Instead, lets try and fetch the auth chains at the same time and calculate the difference on the fly, allowing us to bail early if all the auth chains converge. Assuming that the auth chains do converge more often than not, this should improve performance. Hopefully. --- changelog.d/7095.misc | 1 + synapse/state/__init__.py | 28 ++-- synapse/state/v2.py | 32 +---- .../storage/data_stores/main/event_federation.py | 150 +++++++++++++++++++- tests/state/test_v2.py | 13 +- tests/storage/test_event_federation.py | 157 ++++++++++++++++++--- 6 files changed, 310 insertions(+), 71 deletions(-) create mode 100644 changelog.d/7095.misc (limited to 'tests') diff --git a/changelog.d/7095.misc b/changelog.d/7095.misc new file mode 100644 index 0000000000..44fc9f616f --- /dev/null +++ b/changelog.d/7095.misc @@ -0,0 +1 @@ +Attempt to improve performance of state res v2 algorithm. diff --git a/synapse/state/__init__.py b/synapse/state/__init__.py index df7a4f6a89..4afefc6b1d 100644 --- a/synapse/state/__init__.py +++ b/synapse/state/__init__.py @@ -662,28 +662,16 @@ class StateResolutionStore(object): allow_rejected=allow_rejected, ) - def get_auth_chain(self, event_ids: List[str], ignore_events: Set[str]): - """Gets the full auth chain for a set of events (including rejected - events). - - Includes the given event IDs in the result. - - Note that: - 1. All events must be state events. - 2. For v1 rooms this may not have the full auth chain in the - presence of rejected events - - Args: - event_ids: The event IDs of the events to fetch the auth chain for. - Must be state events. - ignore_events: Set of events to exclude from the returned auth - chain. + def get_auth_chain_difference(self, state_sets: List[Set[str]]): + """Given sets of state events figure out the auth chain difference (as + per state res v2 algorithm). + This equivalent to fetching the full auth chain for each set of state + and returning the events that don't appear in each and every auth + chain. Returns: - Deferred[list[str]]: List of event IDs of the auth chain. + Deferred[Set[str]]: Set of event IDs. """ - return self.store.get_auth_chain_ids( - event_ids, include_given=True, ignore_events=ignore_events, - ) + return self.store.get_auth_chain_difference(state_sets) diff --git a/synapse/state/v2.py b/synapse/state/v2.py index 0ffe6d8c14..18484e2fa6 100644 --- a/synapse/state/v2.py +++ b/synapse/state/v2.py @@ -227,36 +227,12 @@ def _get_auth_chain_difference(state_sets, event_map, state_res_store): Returns: Deferred[set[str]]: Set of event IDs """ - common = set(itervalues(state_sets[0])).intersection( - *(itervalues(s) for s in state_sets[1:]) - ) - - auth_sets = [] - for state_set in state_sets: - auth_ids = { - eid - for key, eid in iteritems(state_set) - if ( - key[0] in (EventTypes.Member, EventTypes.ThirdPartyInvite) - or key - in ( - (EventTypes.PowerLevels, ""), - (EventTypes.Create, ""), - (EventTypes.JoinRules, ""), - ) - ) - and eid not in common - } - auth_chain = yield state_res_store.get_auth_chain(auth_ids, common) - auth_ids.update(auth_chain) - - auth_sets.append(auth_ids) - - intersection = set(auth_sets[0]).intersection(*auth_sets[1:]) - union = set().union(*auth_sets) + difference = yield state_res_store.get_auth_chain_difference( + [set(state_set.values()) for state_set in state_sets] + ) - return union - intersection + return difference def _seperate(state_sets): diff --git a/synapse/storage/data_stores/main/event_federation.py b/synapse/storage/data_stores/main/event_federation.py index 49a7b8b433..62d4e9f599 100644 --- a/synapse/storage/data_stores/main/event_federation.py +++ b/synapse/storage/data_stores/main/event_federation.py @@ -14,7 +14,7 @@ # limitations under the License. import itertools import logging -from typing import List, Optional, Set +from typing import Dict, List, Optional, Set, Tuple from six.moves.queue import Empty, PriorityQueue @@ -103,6 +103,154 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBas return list(results) + def get_auth_chain_difference(self, state_sets: List[Set[str]]): + """Given sets of state events figure out the auth chain difference (as + per state res v2 algorithm). + + This equivalent to fetching the full auth chain for each set of state + and returning the events that don't appear in each and every auth + chain. + + Returns: + Deferred[Set[str]] + """ + + return self.db.runInteraction( + "get_auth_chain_difference", + self._get_auth_chain_difference_txn, + state_sets, + ) + + def _get_auth_chain_difference_txn( + self, txn, state_sets: List[Set[str]] + ) -> Set[str]: + + # Algorithm Description + # ~~~~~~~~~~~~~~~~~~~~~ + # + # The idea here is to basically walk the auth graph of each state set in + # tandem, keeping track of which auth events are reachable by each state + # set. If we reach an auth event we've already visited (via a different + # state set) then we mark that auth event and all ancestors as reachable + # by the state set. This requires that we keep track of the auth chains + # in memory. + # + # Doing it in a such a way means that we can stop early if all auth + # events we're currently walking are reachable by all state sets. + # + # *Note*: We can't stop walking an event's auth chain if it is reachable + # by all state sets. This is because other auth chains we're walking + # might be reachable only via the original auth chain. For example, + # given the following auth chain: + # + # A -> C -> D -> E + # / / + # B -´---------´ + # + # and state sets {A} and {B} then walking the auth chains of A and B + # would immediately show that C is reachable by both. However, if we + # stopped at C then we'd only reach E via the auth chain of B and so E + # would errornously get included in the returned difference. + # + # The other thing that we do is limit the number of auth chains we walk + # at once, due to practical limits (i.e. we can only query the database + # with a limited set of parameters). We pick the auth chains we walk + # each iteration based on their depth, in the hope that events with a + # lower depth are likely reachable by those with higher depths. + # + # We could use any ordering that we believe would give a rough + # topological ordering, e.g. origin server timestamp. If the ordering + # chosen is not topological then the algorithm still produces the right + # result, but perhaps a bit more inefficiently. This is why it is safe + # to use "depth" here. + + initial_events = set(state_sets[0]).union(*state_sets[1:]) + + # Dict from events in auth chains to which sets *cannot* reach them. + # I.e. if the set is empty then all sets can reach the event. + event_to_missing_sets = { + event_id: {i for i, a in enumerate(state_sets) if event_id not in a} + for event_id in initial_events + } + + # We need to get the depth of the initial events for sorting purposes. + sql = """ + SELECT depth, event_id FROM events + WHERE %s + ORDER BY depth ASC + """ + clause, args = make_in_list_sql_clause( + txn.database_engine, "event_id", initial_events + ) + txn.execute(sql % (clause,), args) + + # The sorted list of events whose auth chains we should walk. + search = txn.fetchall() # type: List[Tuple[int, str]] + + # Map from event to its auth events + event_to_auth_events = {} # type: Dict[str, Set[str]] + + base_sql = """ + SELECT a.event_id, auth_id, depth + FROM event_auth AS a + INNER JOIN events AS e ON (e.event_id = a.auth_id) + WHERE + """ + + while search: + # Check whether all our current walks are reachable by all state + # sets. If so we can bail. + if all(not event_to_missing_sets[eid] for _, eid in search): + break + + # Fetch the auth events and their depths of the N last events we're + # currently walking + search, chunk = search[:-100], search[-100:] + clause, args = make_in_list_sql_clause( + txn.database_engine, "a.event_id", [e_id for _, e_id in chunk] + ) + txn.execute(base_sql + clause, args) + + for event_id, auth_event_id, auth_event_depth in txn: + event_to_auth_events.setdefault(event_id, set()).add(auth_event_id) + + sets = event_to_missing_sets.get(auth_event_id) + if sets is None: + # First time we're seeing this event, so we add it to the + # queue of things to fetch. + search.append((auth_event_depth, auth_event_id)) + + # Assume that this event is unreachable from any of the + # state sets until proven otherwise + sets = event_to_missing_sets[auth_event_id] = set( + range(len(state_sets)) + ) + else: + # We've previously seen this event, so look up its auth + # events and recursively mark all ancestors as reachable + # by the current event's state set. + a_ids = event_to_auth_events.get(auth_event_id) + while a_ids: + new_aids = set() + for a_id in a_ids: + event_to_missing_sets[a_id].intersection_update( + event_to_missing_sets[event_id] + ) + + b = event_to_auth_events.get(a_id) + if b: + new_aids.update(b) + + a_ids = new_aids + + # Mark that the auth event is reachable by the approriate sets. + sets.intersection_update(event_to_missing_sets[event_id]) + + search.sort() + + # Return all events where not all sets can reach them. + return {eid for eid, n in event_to_missing_sets.items() if n} + def get_oldest_events_in_room(self, room_id): return self.db.runInteraction( "get_oldest_events_in_room", self._get_oldest_events_in_room_txn, room_id diff --git a/tests/state/test_v2.py b/tests/state/test_v2.py index 5059ade850..a44960203e 100644 --- a/tests/state/test_v2.py +++ b/tests/state/test_v2.py @@ -603,7 +603,7 @@ class TestStateResolutionStore(object): return {eid: self.event_map[eid] for eid in event_ids if eid in self.event_map} - def get_auth_chain(self, event_ids, ignore_events): + def _get_auth_chain(self, event_ids): """Gets the full auth chain for a set of events (including rejected events). @@ -617,9 +617,6 @@ class TestStateResolutionStore(object): Args: event_ids (list): The event IDs of the events to fetch the auth chain for. Must be state events. - ignore_events: Set of events to exclude from the returned auth - chain. - Returns: Deferred[list[str]]: List of event IDs of the auth chain. """ @@ -629,7 +626,7 @@ class TestStateResolutionStore(object): stack = list(event_ids) while stack: event_id = stack.pop() - if event_id in result or event_id in ignore_events: + if event_id in result: continue result.add(event_id) @@ -639,3 +636,9 @@ class TestStateResolutionStore(object): stack.append(aid) return list(result) + + def get_auth_chain_difference(self, auth_sets): + chains = [frozenset(self._get_auth_chain(a)) for a in auth_sets] + + common = set(chains[0]).intersection(*chains[1:]) + return set(chains[0]).union(*chains[1:]) - common diff --git a/tests/storage/test_event_federation.py b/tests/storage/test_event_federation.py index a331517f4d..3aeec0dc0f 100644 --- a/tests/storage/test_event_federation.py +++ b/tests/storage/test_event_federation.py @@ -13,19 +13,14 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.internet import defer - import tests.unittest import tests.utils -class EventFederationWorkerStoreTestCase(tests.unittest.TestCase): - @defer.inlineCallbacks - def setUp(self): - hs = yield tests.utils.setup_test_homeserver(self.addCleanup) +class EventFederationWorkerStoreTestCase(tests.unittest.HomeserverTestCase): + def prepare(self, reactor, clock, hs): self.store = hs.get_datastore() - @defer.inlineCallbacks def test_get_prev_events_for_room(self): room_id = "@ROOM:local" @@ -61,15 +56,14 @@ class EventFederationWorkerStoreTestCase(tests.unittest.TestCase): ) for i in range(0, 20): - yield self.store.db.runInteraction("insert", insert_event, i) + self.get_success(self.store.db.runInteraction("insert", insert_event, i)) # this should get the last ten - r = yield self.store.get_prev_events_for_room(room_id) + r = self.get_success(self.store.get_prev_events_for_room(room_id)) self.assertEqual(10, len(r)) for i in range(0, 10): self.assertEqual("$event_%i:local" % (19 - i), r[i]) - @defer.inlineCallbacks def test_get_rooms_with_many_extremities(self): room1 = "#room1" room2 = "#room2" @@ -86,25 +80,154 @@ class EventFederationWorkerStoreTestCase(tests.unittest.TestCase): ) for i in range(0, 20): - yield self.store.db.runInteraction("insert", insert_event, i, room1) - yield self.store.db.runInteraction("insert", insert_event, i, room2) - yield self.store.db.runInteraction("insert", insert_event, i, room3) + self.get_success( + self.store.db.runInteraction("insert", insert_event, i, room1) + ) + self.get_success( + self.store.db.runInteraction("insert", insert_event, i, room2) + ) + self.get_success( + self.store.db.runInteraction("insert", insert_event, i, room3) + ) # Test simple case - r = yield self.store.get_rooms_with_many_extremities(5, 5, []) + r = self.get_success(self.store.get_rooms_with_many_extremities(5, 5, [])) self.assertEqual(len(r), 3) # Does filter work? - r = yield self.store.get_rooms_with_many_extremities(5, 5, [room1]) + r = self.get_success(self.store.get_rooms_with_many_extremities(5, 5, [room1])) self.assertTrue(room2 in r) self.assertTrue(room3 in r) self.assertEqual(len(r), 2) - r = yield self.store.get_rooms_with_many_extremities(5, 5, [room1, room2]) + r = self.get_success( + self.store.get_rooms_with_many_extremities(5, 5, [room1, room2]) + ) self.assertEqual(r, [room3]) # Does filter and limit work? - r = yield self.store.get_rooms_with_many_extremities(5, 1, [room1]) + r = self.get_success(self.store.get_rooms_with_many_extremities(5, 1, [room1])) self.assertTrue(r == [room2] or r == [room3]) + + def test_auth_difference(self): + room_id = "@ROOM:local" + + # The silly auth graph we use to test the auth difference algorithm, + # where the top are the most recent events. + # + # A B + # \ / + # D E + # \ | + # ` F C + # | /| + # G ´ | + # | \ | + # H I + # | | + # K J + + auth_graph = { + "a": ["e"], + "b": ["e"], + "c": ["g", "i"], + "d": ["f"], + "e": ["f"], + "f": ["g"], + "g": ["h", "i"], + "h": ["k"], + "i": ["j"], + "k": [], + "j": [], + } + + depth_map = { + "a": 7, + "b": 7, + "c": 4, + "d": 6, + "e": 6, + "f": 5, + "g": 3, + "h": 2, + "i": 2, + "k": 1, + "j": 1, + } + + # We rudely fiddle with the appropriate tables directly, as that's much + # easier than constructing events properly. + + def insert_event(txn, event_id, stream_ordering): + + depth = depth_map[event_id] + + self.store.db.simple_insert_txn( + txn, + table="events", + values={ + "event_id": event_id, + "room_id": room_id, + "depth": depth, + "topological_ordering": depth, + "type": "m.test", + "processed": True, + "outlier": False, + "stream_ordering": stream_ordering, + }, + ) + + self.store.db.simple_insert_many_txn( + txn, + table="event_auth", + values=[ + {"event_id": event_id, "room_id": room_id, "auth_id": a} + for a in auth_graph[event_id] + ], + ) + + next_stream_ordering = 0 + for event_id in auth_graph: + next_stream_ordering += 1 + self.get_success( + self.store.db.runInteraction( + "insert", insert_event, event_id, next_stream_ordering + ) + ) + + # Now actually test that various combinations give the right result: + + difference = self.get_success( + self.store.get_auth_chain_difference([{"a"}, {"b"}]) + ) + self.assertSetEqual(difference, {"a", "b"}) + + difference = self.get_success( + self.store.get_auth_chain_difference([{"a"}, {"b"}, {"c"}]) + ) + self.assertSetEqual(difference, {"a", "b", "c", "e", "f"}) + + difference = self.get_success( + self.store.get_auth_chain_difference([{"a", "c"}, {"b"}]) + ) + self.assertSetEqual(difference, {"a", "b", "c"}) + + difference = self.get_success( + self.store.get_auth_chain_difference([{"a"}, {"b"}, {"d"}]) + ) + self.assertSetEqual(difference, {"a", "b", "d", "e"}) + + difference = self.get_success( + self.store.get_auth_chain_difference([{"a"}, {"b"}, {"c"}, {"d"}]) + ) + self.assertSetEqual(difference, {"a", "b", "c", "d", "e", "f"}) + + difference = self.get_success( + self.store.get_auth_chain_difference([{"a"}, {"b"}, {"e"}]) + ) + self.assertSetEqual(difference, {"a", "b"}) + + difference = self.get_success(self.store.get_auth_chain_difference([{"a"}])) + self.assertSetEqual(difference, set()) -- cgit 1.4.1 From c165c1233b8ef244fadca97c7d465fdcf473d077 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Fri, 20 Mar 2020 16:24:22 +0100 Subject: Improve database configuration docs (#6988) Attempts to clarify the sample config for databases, and add some stuff about tcp keepalives to `postgres.md`. --- changelog.d/6988.doc | 1 + docs/postgres.md | 42 ++++++++++++++----- docs/sample_config.yaml | 43 +++++++++++++++++--- synapse/config/_base.py | 2 - synapse/config/database.py | 93 +++++++++++++++++++++++++++---------------- tests/config/test_database.py | 22 +--------- 6 files changed, 132 insertions(+), 71 deletions(-) create mode 100644 changelog.d/6988.doc (limited to 'tests') diff --git a/changelog.d/6988.doc b/changelog.d/6988.doc new file mode 100644 index 0000000000..b6f71bb966 --- /dev/null +++ b/changelog.d/6988.doc @@ -0,0 +1 @@ +Improve the documentation for database configuration. diff --git a/docs/postgres.md b/docs/postgres.md index e0793ecee8..16a630c3d1 100644 --- a/docs/postgres.md +++ b/docs/postgres.md @@ -105,19 +105,41 @@ of free memory the database host has available. When you are ready to start using PostgreSQL, edit the `database` section in your config file to match the following lines: - database: - name: psycopg2 - args: - user: - password: - database: - host: - cp_min: 5 - cp_max: 10 +```yaml +database: + name: psycopg2 + args: + user: + password: + database: + host: + cp_min: 5 + cp_max: 10 +``` All key, values in `args` are passed to the `psycopg2.connect(..)` function, except keys beginning with `cp_`, which are consumed by the -twisted adbapi connection pool. +twisted adbapi connection pool. See the [libpq +documentation](https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-PARAMKEYWORDS) +for a list of options which can be passed. + +You should consider tuning the `args.keepalives_*` options if there is any danger of +the connection between your homeserver and database dropping, otherwise Synapse +may block for an extended period while it waits for a response from the +database server. Example values might be: + +```yaml +# seconds of inactivity after which TCP should send a keepalive message to the server +keepalives_idle: 10 + +# the number of seconds after which a TCP keepalive message that is not +# acknowledged by the server should be retransmitted +keepalives_interval: 10 + +# the number of TCP keepalives that can be lost before the client's connection +# to the server is considered dead +keepalives_count: 3 +``` ## Porting from SQLite diff --git a/docs/sample_config.yaml b/docs/sample_config.yaml index 2ff0dd05a2..276e43b732 100644 --- a/docs/sample_config.yaml +++ b/docs/sample_config.yaml @@ -578,13 +578,46 @@ acme: ## Database ## +# The 'database' setting defines the database that synapse uses to store all of +# its data. +# +# 'name' gives the database engine to use: either 'sqlite3' (for SQLite) or +# 'psycopg2' (for PostgreSQL). +# +# 'args' gives options which are passed through to the database engine, +# except for options starting 'cp_', which are used to configure the Twisted +# connection pool. For a reference to valid arguments, see: +# * for sqlite: https://docs.python.org/3/library/sqlite3.html#sqlite3.connect +# * for postgres: https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-PARAMKEYWORDS +# * for the connection pool: https://twistedmatrix.com/documents/current/api/twisted.enterprise.adbapi.ConnectionPool.html#__init__ +# +# +# Example SQLite configuration: +# +#database: +# name: sqlite3 +# args: +# database: /path/to/homeserver.db +# +# +# Example Postgres configuration: +# +#database: +# name: psycopg2 +# args: +# user: synapse +# password: secretpassword +# database: synapse +# host: localhost +# cp_min: 5 +# cp_max: 10 +# +# For more information on using Synapse with Postgres, see `docs/postgres.md`. +# database: - # The database engine name - name: "sqlite3" - # Arguments to pass to the engine + name: sqlite3 args: - # Path to the database - database: "DATADIR/homeserver.db" + database: DATADIR/homeserver.db # Number of events to cache in memory. # diff --git a/synapse/config/_base.py b/synapse/config/_base.py index ba846042c4..efe2af5504 100644 --- a/synapse/config/_base.py +++ b/synapse/config/_base.py @@ -294,7 +294,6 @@ class RootConfig(object): report_stats=None, open_private_ports=False, listeners=None, - database_conf=None, tls_certificate_path=None, tls_private_key_path=None, acme_domain=None, @@ -367,7 +366,6 @@ class RootConfig(object): report_stats=report_stats, open_private_ports=open_private_ports, listeners=listeners, - database_conf=database_conf, tls_certificate_path=tls_certificate_path, tls_private_key_path=tls_private_key_path, acme_domain=acme_domain, diff --git a/synapse/config/database.py b/synapse/config/database.py index 219b32f670..b8ab2f86ac 100644 --- a/synapse/config/database.py +++ b/synapse/config/database.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2020 The Matrix.org Foundation C.I.C. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -14,14 +15,60 @@ # limitations under the License. import logging import os -from textwrap import indent - -import yaml from synapse.config._base import Config, ConfigError logger = logging.getLogger(__name__) +DEFAULT_CONFIG = """\ +## Database ## + +# The 'database' setting defines the database that synapse uses to store all of +# its data. +# +# 'name' gives the database engine to use: either 'sqlite3' (for SQLite) or +# 'psycopg2' (for PostgreSQL). +# +# 'args' gives options which are passed through to the database engine, +# except for options starting 'cp_', which are used to configure the Twisted +# connection pool. For a reference to valid arguments, see: +# * for sqlite: https://docs.python.org/3/library/sqlite3.html#sqlite3.connect +# * for postgres: https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-PARAMKEYWORDS +# * for the connection pool: https://twistedmatrix.com/documents/current/api/twisted.enterprise.adbapi.ConnectionPool.html#__init__ +# +# +# Example SQLite configuration: +# +#database: +# name: sqlite3 +# args: +# database: /path/to/homeserver.db +# +# +# Example Postgres configuration: +# +#database: +# name: psycopg2 +# args: +# user: synapse +# password: secretpassword +# database: synapse +# host: localhost +# cp_min: 5 +# cp_max: 10 +# +# For more information on using Synapse with Postgres, see `docs/postgres.md`. +# +database: + name: sqlite3 + args: + database: %(database_path)s + +# Number of events to cache in memory. +# +#event_cache_size: 10K +""" + class DatabaseConnectionConfig: """Contains the connection config for a particular database. @@ -36,10 +83,12 @@ class DatabaseConnectionConfig: """ def __init__(self, name: str, db_config: dict): - if db_config["name"] not in ("sqlite3", "psycopg2"): - raise ConfigError("Unsupported database type %r" % (db_config["name"],)) + db_engine = db_config.get("name", "sqlite3") - if db_config["name"] == "sqlite3": + if db_engine not in ("sqlite3", "psycopg2"): + raise ConfigError("Unsupported database type %r" % (db_engine,)) + + if db_engine == "sqlite3": db_config.setdefault("args", {}).update( {"cp_min": 1, "cp_max": 1, "check_same_thread": False} ) @@ -97,34 +146,10 @@ class DatabaseConfig(Config): self.set_databasepath(config.get("database_path")) - def generate_config_section(self, data_dir_path, database_conf, **kwargs): - if not database_conf: - database_path = os.path.join(data_dir_path, "homeserver.db") - database_conf = ( - """# The database engine name - name: "sqlite3" - # Arguments to pass to the engine - args: - # Path to the database - database: "%(database_path)s" - """ - % locals() - ) - else: - database_conf = indent(yaml.dump(database_conf), " " * 10).lstrip() - - return ( - """\ - ## Database ## - - database: - %(database_conf)s - # Number of events to cache in memory. - # - #event_cache_size: 10K - """ - % locals() - ) + def generate_config_section(self, data_dir_path, **kwargs): + return DEFAULT_CONFIG % { + "database_path": os.path.join(data_dir_path, "homeserver.db") + } def read_arguments(self, args): self.set_databasepath(args.database_path) diff --git a/tests/config/test_database.py b/tests/config/test_database.py index 151d3006ac..f675bde68e 100644 --- a/tests/config/test_database.py +++ b/tests/config/test_database.py @@ -21,9 +21,9 @@ from tests import unittest class DatabaseConfigTestCase(unittest.TestCase): - def test_database_configured_correctly_no_database_conf_param(self): + def test_database_configured_correctly(self): conf = yaml.safe_load( - DatabaseConfig().generate_config_section("/data_dir_path", None) + DatabaseConfig().generate_config_section(data_dir_path="/data_dir_path") ) expected_database_conf = { @@ -32,21 +32,3 @@ class DatabaseConfigTestCase(unittest.TestCase): } self.assertEqual(conf["database"], expected_database_conf) - - def test_database_configured_correctly_database_conf_param(self): - - database_conf = { - "name": "my super fast datastore", - "args": { - "user": "matrix", - "password": "synapse_database_password", - "host": "synapse_database_host", - "database": "matrix", - }, - } - - conf = yaml.safe_load( - DatabaseConfig().generate_config_section("/data_dir_path", database_conf) - ) - - self.assertEqual(conf["database"], database_conf) -- cgit 1.4.1 From a564b92d37625855940fe599c730a9958c33f973 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Mon, 23 Mar 2020 13:59:11 +0000 Subject: Convert `*StreamRow` classes to inner classes (#7116) This just helps keep the rows closer to their streams, so that it's easier to see what the format of each stream is. --- changelog.d/7116.misc | 1 + synapse/app/generic_worker.py | 2 +- synapse/federation/send_queue.py | 2 +- synapse/replication/tcp/streams/_base.py | 181 +++++++++++++------------ synapse/replication/tcp/streams/federation.py | 16 +-- tests/replication/tcp/streams/test_receipts.py | 4 +- 6 files changed, 106 insertions(+), 100 deletions(-) create mode 100644 changelog.d/7116.misc (limited to 'tests') diff --git a/changelog.d/7116.misc b/changelog.d/7116.misc new file mode 100644 index 0000000000..89d90bd49e --- /dev/null +++ b/changelog.d/7116.misc @@ -0,0 +1 @@ +Convert `*StreamRow` classes to inner classes. diff --git a/synapse/app/generic_worker.py b/synapse/app/generic_worker.py index 136babe6ce..c8fd8909a4 100644 --- a/synapse/app/generic_worker.py +++ b/synapse/app/generic_worker.py @@ -804,7 +804,7 @@ class FederationSenderHandler(object): async def _on_new_receipts(self, rows): """ Args: - rows (iterable[synapse.replication.tcp.streams.ReceiptsStreamRow]): + rows (Iterable[synapse.replication.tcp.streams.ReceiptsStream.ReceiptsStreamRow]): new receipts to be processed """ for receipt in rows: diff --git a/synapse/federation/send_queue.py b/synapse/federation/send_queue.py index 876fb0e245..e1700ca8aa 100644 --- a/synapse/federation/send_queue.py +++ b/synapse/federation/send_queue.py @@ -477,7 +477,7 @@ def process_rows_for_federation(transaction_queue, rows): Args: transaction_queue (FederationSender) - rows (list(synapse.replication.tcp.streams.FederationStreamRow)) + rows (list(synapse.replication.tcp.streams.federation.FederationStream.FederationStreamRow)) """ # The federation stream contains a bunch of different types of diff --git a/synapse/replication/tcp/streams/_base.py b/synapse/replication/tcp/streams/_base.py index abf5c6c6a8..32d9514883 100644 --- a/synapse/replication/tcp/streams/_base.py +++ b/synapse/replication/tcp/streams/_base.py @@ -28,94 +28,6 @@ logger = logging.getLogger(__name__) MAX_EVENTS_BEHIND = 500000 -BackfillStreamRow = namedtuple( - "BackfillStreamRow", - ( - "event_id", # str - "room_id", # str - "type", # str - "state_key", # str, optional - "redacts", # str, optional - "relates_to", # str, optional - ), -) -PresenceStreamRow = namedtuple( - "PresenceStreamRow", - ( - "user_id", # str - "state", # str - "last_active_ts", # int - "last_federation_update_ts", # int - "last_user_sync_ts", # int - "status_msg", # str - "currently_active", # bool - ), -) -TypingStreamRow = namedtuple( - "TypingStreamRow", ("room_id", "user_ids") # str # list(str) -) -ReceiptsStreamRow = namedtuple( - "ReceiptsStreamRow", - ( - "room_id", # str - "receipt_type", # str - "user_id", # str - "event_id", # str - "data", # dict - ), -) -PushRulesStreamRow = namedtuple("PushRulesStreamRow", ("user_id",)) # str -PushersStreamRow = namedtuple( - "PushersStreamRow", - ("user_id", "app_id", "pushkey", "deleted"), # str # str # str # bool -) - - -@attr.s -class CachesStreamRow: - """Stream to inform workers they should invalidate their cache. - - Attributes: - cache_func: Name of the cached function. - keys: The entry in the cache to invalidate. If None then will - invalidate all. - invalidation_ts: Timestamp of when the invalidation took place. - """ - - cache_func = attr.ib(type=str) - keys = attr.ib(type=Optional[List[Any]]) - invalidation_ts = attr.ib(type=int) - - -PublicRoomsStreamRow = namedtuple( - "PublicRoomsStreamRow", - ( - "room_id", # str - "visibility", # str - "appservice_id", # str, optional - "network_id", # str, optional - ), -) - - -@attr.s -class DeviceListsStreamRow: - entity = attr.ib(type=str) - - -ToDeviceStreamRow = namedtuple("ToDeviceStreamRow", ("entity",)) # str -TagAccountDataStreamRow = namedtuple( - "TagAccountDataStreamRow", ("user_id", "room_id", "data") # str # str # dict -) -AccountDataStreamRow = namedtuple( - "AccountDataStream", ("user_id", "room_id", "data_type") # str # str # str -) -GroupsStreamRow = namedtuple( - "GroupsStreamRow", - ("group_id", "user_id", "type", "content"), # str # str # str # dict -) -UserSignatureStreamRow = namedtuple("UserSignatureStreamRow", ("user_id")) # str - class Stream(object): """Base class for the streams. @@ -234,6 +146,18 @@ class BackfillStream(Stream): or it went from being an outlier to not. """ + BackfillStreamRow = namedtuple( + "BackfillStreamRow", + ( + "event_id", # str + "room_id", # str + "type", # str + "state_key", # str, optional + "redacts", # str, optional + "relates_to", # str, optional + ), + ) + NAME = "backfill" ROW_TYPE = BackfillStreamRow @@ -246,6 +170,19 @@ class BackfillStream(Stream): class PresenceStream(Stream): + PresenceStreamRow = namedtuple( + "PresenceStreamRow", + ( + "user_id", # str + "state", # str + "last_active_ts", # int + "last_federation_update_ts", # int + "last_user_sync_ts", # int + "status_msg", # str + "currently_active", # bool + ), + ) + NAME = "presence" ROW_TYPE = PresenceStreamRow @@ -260,6 +197,10 @@ class PresenceStream(Stream): class TypingStream(Stream): + TypingStreamRow = namedtuple( + "TypingStreamRow", ("room_id", "user_ids") # str # list(str) + ) + NAME = "typing" ROW_TYPE = TypingStreamRow @@ -273,6 +214,17 @@ class TypingStream(Stream): class ReceiptsStream(Stream): + ReceiptsStreamRow = namedtuple( + "ReceiptsStreamRow", + ( + "room_id", # str + "receipt_type", # str + "user_id", # str + "event_id", # str + "data", # dict + ), + ) + NAME = "receipts" ROW_TYPE = ReceiptsStreamRow @@ -289,6 +241,8 @@ class PushRulesStream(Stream): """A user has changed their push rules """ + PushRulesStreamRow = namedtuple("PushRulesStreamRow", ("user_id",)) # str + NAME = "push_rules" ROW_TYPE = PushRulesStreamRow @@ -309,6 +263,11 @@ class PushersStream(Stream): """A user has added/changed/removed a pusher """ + PushersStreamRow = namedtuple( + "PushersStreamRow", + ("user_id", "app_id", "pushkey", "deleted"), # str # str # str # bool + ) + NAME = "pushers" ROW_TYPE = PushersStreamRow @@ -326,6 +285,21 @@ class CachesStream(Stream): the cache on the workers """ + @attr.s + class CachesStreamRow: + """Stream to inform workers they should invalidate their cache. + + Attributes: + cache_func: Name of the cached function. + keys: The entry in the cache to invalidate. If None then will + invalidate all. + invalidation_ts: Timestamp of when the invalidation took place. + """ + + cache_func = attr.ib(type=str) + keys = attr.ib(type=Optional[List[Any]]) + invalidation_ts = attr.ib(type=int) + NAME = "caches" ROW_TYPE = CachesStreamRow @@ -342,6 +316,16 @@ class PublicRoomsStream(Stream): """The public rooms list changed """ + PublicRoomsStreamRow = namedtuple( + "PublicRoomsStreamRow", + ( + "room_id", # str + "visibility", # str + "appservice_id", # str, optional + "network_id", # str, optional + ), + ) + NAME = "public_rooms" ROW_TYPE = PublicRoomsStreamRow @@ -359,6 +343,10 @@ class DeviceListsStream(Stream): told about a device update. """ + @attr.s + class DeviceListsStreamRow: + entity = attr.ib(type=str) + NAME = "device_lists" ROW_TYPE = DeviceListsStreamRow @@ -375,6 +363,8 @@ class ToDeviceStream(Stream): """New to_device messages for a client """ + ToDeviceStreamRow = namedtuple("ToDeviceStreamRow", ("entity",)) # str + NAME = "to_device" ROW_TYPE = ToDeviceStreamRow @@ -391,6 +381,10 @@ class TagAccountDataStream(Stream): """Someone added/removed a tag for a room """ + TagAccountDataStreamRow = namedtuple( + "TagAccountDataStreamRow", ("user_id", "room_id", "data") # str # str # dict + ) + NAME = "tag_account_data" ROW_TYPE = TagAccountDataStreamRow @@ -407,6 +401,10 @@ class AccountDataStream(Stream): """Global or per room account data was changed """ + AccountDataStreamRow = namedtuple( + "AccountDataStream", ("user_id", "room_id", "data_type") # str # str # str + ) + NAME = "account_data" ROW_TYPE = AccountDataStreamRow @@ -432,6 +430,11 @@ class AccountDataStream(Stream): class GroupServerStream(Stream): + GroupsStreamRow = namedtuple( + "GroupsStreamRow", + ("group_id", "user_id", "type", "content"), # str # str # str # dict + ) + NAME = "groups" ROW_TYPE = GroupsStreamRow @@ -448,6 +451,8 @@ class UserSignatureStream(Stream): """A user has signed their own device with their user-signing key """ + UserSignatureStreamRow = namedtuple("UserSignatureStreamRow", ("user_id")) # str + NAME = "user_signature" ROW_TYPE = UserSignatureStreamRow diff --git a/synapse/replication/tcp/streams/federation.py b/synapse/replication/tcp/streams/federation.py index 615f3dc9ac..f5f9336430 100644 --- a/synapse/replication/tcp/streams/federation.py +++ b/synapse/replication/tcp/streams/federation.py @@ -17,20 +17,20 @@ from collections import namedtuple from ._base import Stream -FederationStreamRow = namedtuple( - "FederationStreamRow", - ( - "type", # str, the type of data as defined in the BaseFederationRows - "data", # dict, serialization of a federation.send_queue.BaseFederationRow - ), -) - class FederationStream(Stream): """Data to be sent over federation. Only available when master has federation sending disabled. """ + FederationStreamRow = namedtuple( + "FederationStreamRow", + ( + "type", # str, the type of data as defined in the BaseFederationRows + "data", # dict, serialization of a federation.send_queue.BaseFederationRow + ), + ) + NAME = "federation" ROW_TYPE = FederationStreamRow diff --git a/tests/replication/tcp/streams/test_receipts.py b/tests/replication/tcp/streams/test_receipts.py index d5a99f6caa..fa2493cad6 100644 --- a/tests/replication/tcp/streams/test_receipts.py +++ b/tests/replication/tcp/streams/test_receipts.py @@ -12,7 +12,7 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. -from synapse.replication.tcp.streams._base import ReceiptsStreamRow +from synapse.replication.tcp.streams._base import ReceiptsStream from tests.replication.tcp.streams._base import BaseStreamTestCase @@ -38,7 +38,7 @@ class ReceiptsStreamTestCase(BaseStreamTestCase): rdata_rows = self.test_handler.received_rdata_rows self.assertEqual(1, len(rdata_rows)) self.assertEqual(rdata_rows[0][0], "receipts") - row = rdata_rows[0][2] # type: ReceiptsStreamRow + row = rdata_rows[0][2] # type: ReceiptsStream.ReceiptsStreamRow self.assertEqual(ROOM_ID, row.room_id) self.assertEqual("m.read", row.receipt_type) self.assertEqual(USER_ID, row.user_id) -- cgit 1.4.1 From 39230d217104f3cd7aba9065dc478f935ce1e614 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Tue, 24 Mar 2020 14:45:33 +0000 Subject: Clean up some LoggingContext stuff (#7120) * Pull Sentinel out of LoggingContext ... and drop a few unnecessary references to it * Factor out LoggingContext.current_context move `current_context` and `set_context` out to top-level functions. Mostly this means that I can more easily trace what's actually referring to LoggingContext, but I think it's generally neater. * move copy-to-parent into `stop` this really just makes `start` and `stop` more symetric. It also means that it behaves correctly if you manually `set_log_context` rather than using the context manager. * Replace `LoggingContext.alive` with `finished` Turn `alive` into `finished` and make it a bit better defined. --- changelog.d/7120.misc | 1 + docs/log_contexts.md | 5 +- synapse/crypto/keyring.py | 4 +- synapse/federation/federation_base.py | 4 +- synapse/handlers/sync.py | 4 +- synapse/http/request_metrics.py | 6 +- synapse/logging/_structured.py | 4 +- synapse/logging/context.py | 234 +++++++++++---------- synapse/logging/scopecontextmanager.py | 13 +- synapse/storage/data_stores/main/events_worker.py | 4 +- synapse/storage/database.py | 11 +- synapse/util/metrics.py | 4 +- synapse/util/patch_inline_callbacks.py | 36 ++-- tests/crypto/test_keyring.py | 7 +- .../federation/test_matrix_federation_agent.py | 6 +- tests/http/federation/test_srv_resolver.py | 6 +- tests/http/test_fedclient.py | 6 +- tests/rest/client/test_transactions.py | 16 +- tests/unittest.py | 12 +- tests/util/caches/test_descriptors.py | 22 +- tests/util/test_async_utils.py | 15 +- tests/util/test_linearizer.py | 6 +- tests/util/test_logcontext.py | 22 +- tests/utils.py | 6 +- 24 files changed, 232 insertions(+), 222 deletions(-) create mode 100644 changelog.d/7120.misc (limited to 'tests') diff --git a/changelog.d/7120.misc b/changelog.d/7120.misc new file mode 100644 index 0000000000..731f4dcb52 --- /dev/null +++ b/changelog.d/7120.misc @@ -0,0 +1 @@ +Clean up some LoggingContext code. diff --git a/docs/log_contexts.md b/docs/log_contexts.md index 5331e8c88b..fe30ca2791 100644 --- a/docs/log_contexts.md +++ b/docs/log_contexts.md @@ -29,14 +29,13 @@ from synapse.logging import context # omitted from future snippets def handle_request(request_id): request_context = context.LoggingContext() - calling_context = context.LoggingContext.current_context() - context.LoggingContext.set_current_context(request_context) + calling_context = context.set_current_context(request_context) try: request_context.request = request_id do_request_handling() logger.debug("finished") finally: - context.LoggingContext.set_current_context(calling_context) + context.set_current_context(calling_context) def do_request_handling(): logger.debug("phew") # this will be logged against request_id diff --git a/synapse/crypto/keyring.py b/synapse/crypto/keyring.py index 983f0ead8c..a9f4025bfe 100644 --- a/synapse/crypto/keyring.py +++ b/synapse/crypto/keyring.py @@ -43,8 +43,8 @@ from synapse.api.errors import ( SynapseError, ) from synapse.logging.context import ( - LoggingContext, PreserveLoggingContext, + current_context, make_deferred_yieldable, preserve_fn, run_in_background, @@ -236,7 +236,7 @@ class Keyring(object): """ try: - ctx = LoggingContext.current_context() + ctx = current_context() # map from server name to a set of outstanding request ids server_to_request_ids = {} diff --git a/synapse/federation/federation_base.py b/synapse/federation/federation_base.py index b0b0eba41e..4b115aac04 100644 --- a/synapse/federation/federation_base.py +++ b/synapse/federation/federation_base.py @@ -32,8 +32,8 @@ from synapse.events import EventBase, make_event_from_dict from synapse.events.utils import prune_event from synapse.http.servlet import assert_params_in_dict from synapse.logging.context import ( - LoggingContext, PreserveLoggingContext, + current_context, make_deferred_yieldable, ) from synapse.types import JsonDict, get_domain_from_id @@ -78,7 +78,7 @@ class FederationBase(object): """ deferreds = _check_sigs_on_pdus(self.keyring, room_version, pdus) - ctx = LoggingContext.current_context() + ctx = current_context() def callback(_, pdu: EventBase): with PreserveLoggingContext(ctx): diff --git a/synapse/handlers/sync.py b/synapse/handlers/sync.py index 669dbc8a48..5746fdea14 100644 --- a/synapse/handlers/sync.py +++ b/synapse/handlers/sync.py @@ -26,7 +26,7 @@ from prometheus_client import Counter from synapse.api.constants import EventTypes, Membership from synapse.api.filtering import FilterCollection from synapse.events import EventBase -from synapse.logging.context import LoggingContext +from synapse.logging.context import current_context from synapse.push.clientformat import format_push_rules_for_user from synapse.storage.roommember import MemberSummary from synapse.storage.state import StateFilter @@ -301,7 +301,7 @@ class SyncHandler(object): else: sync_type = "incremental_sync" - context = LoggingContext.current_context() + context = current_context() if context: context.tag = sync_type diff --git a/synapse/http/request_metrics.py b/synapse/http/request_metrics.py index 58f9cc61c8..b58ae3d9db 100644 --- a/synapse/http/request_metrics.py +++ b/synapse/http/request_metrics.py @@ -19,7 +19,7 @@ import threading from prometheus_client.core import Counter, Histogram -from synapse.logging.context import LoggingContext +from synapse.logging.context import current_context from synapse.metrics import LaterGauge logger = logging.getLogger(__name__) @@ -148,7 +148,7 @@ LaterGauge( class RequestMetrics(object): def start(self, time_sec, name, method): self.start = time_sec - self.start_context = LoggingContext.current_context() + self.start_context = current_context() self.name = name self.method = method @@ -163,7 +163,7 @@ class RequestMetrics(object): with _in_flight_requests_lock: _in_flight_requests.discard(self) - context = LoggingContext.current_context() + context = current_context() tag = "" if context: diff --git a/synapse/logging/_structured.py b/synapse/logging/_structured.py index ffa7b20ca8..7372450b45 100644 --- a/synapse/logging/_structured.py +++ b/synapse/logging/_structured.py @@ -42,7 +42,7 @@ from synapse.logging._terse_json import ( TerseJSONToConsoleLogObserver, TerseJSONToTCPLogObserver, ) -from synapse.logging.context import LoggingContext +from synapse.logging.context import current_context def stdlib_log_level_to_twisted(level: str) -> LogLevel: @@ -86,7 +86,7 @@ class LogContextObserver(object): ].startswith("Timing out client"): return - context = LoggingContext.current_context() + context = current_context() # Copy the context information to the log event. if context is not None: diff --git a/synapse/logging/context.py b/synapse/logging/context.py index 860b99a4c6..a8eafb1c7c 100644 --- a/synapse/logging/context.py +++ b/synapse/logging/context.py @@ -175,7 +175,54 @@ class ContextResourceUsage(object): return res -LoggingContextOrSentinel = Union["LoggingContext", "LoggingContext.Sentinel"] +LoggingContextOrSentinel = Union["LoggingContext", "_Sentinel"] + + +class _Sentinel(object): + """Sentinel to represent the root context""" + + __slots__ = ["previous_context", "finished", "request", "scope", "tag"] + + def __init__(self) -> None: + # Minimal set for compatibility with LoggingContext + self.previous_context = None + self.finished = False + self.request = None + self.scope = None + self.tag = None + + def __str__(self): + return "sentinel" + + def copy_to(self, record): + pass + + def copy_to_twisted_log_entry(self, record): + record["request"] = None + record["scope"] = None + + def start(self): + pass + + def stop(self): + pass + + def add_database_transaction(self, duration_sec): + pass + + def add_database_scheduled(self, sched_sec): + pass + + def record_event_fetch(self, event_count): + pass + + def __nonzero__(self): + return False + + __bool__ = __nonzero__ # python3 + + +SENTINEL_CONTEXT = _Sentinel() class LoggingContext(object): @@ -199,76 +246,33 @@ class LoggingContext(object): "_resource_usage", "usage_start", "main_thread", - "alive", + "finished", "request", "tag", "scope", ] - thread_local = threading.local() - - class Sentinel(object): - """Sentinel to represent the root context""" - - __slots__ = ["previous_context", "alive", "request", "scope", "tag"] - - def __init__(self) -> None: - # Minimal set for compatibility with LoggingContext - self.previous_context = None - self.alive = None - self.request = None - self.scope = None - self.tag = None - - def __str__(self): - return "sentinel" - - def copy_to(self, record): - pass - - def copy_to_twisted_log_entry(self, record): - record["request"] = None - record["scope"] = None - - def start(self): - pass - - def stop(self): - pass - - def add_database_transaction(self, duration_sec): - pass - - def add_database_scheduled(self, sched_sec): - pass - - def record_event_fetch(self, event_count): - pass - - def __nonzero__(self): - return False - - __bool__ = __nonzero__ # python3 - - sentinel = Sentinel() - def __init__(self, name=None, parent_context=None, request=None) -> None: - self.previous_context = LoggingContext.current_context() + self.previous_context = current_context() self.name = name # track the resources used by this context so far self._resource_usage = ContextResourceUsage() - # If alive has the thread resource usage when the logcontext last - # became active. + # The thread resource usage when the logcontext became active. None + # if the context is not currently active. self.usage_start = None self.main_thread = get_thread_id() self.request = None self.tag = "" - self.alive = True self.scope = None # type: Optional[_LogContextScope] + # keep track of whether we have hit the __exit__ block for this context + # (suggesting that the the thing that created the context thinks it should + # be finished, and that re-activating it would suggest an error). + self.finished = False + self.parent_context = parent_context if self.parent_context is not None: @@ -283,44 +287,15 @@ class LoggingContext(object): return str(self.request) return "%s@%x" % (self.name, id(self)) - @classmethod - def current_context(cls) -> LoggingContextOrSentinel: - """Get the current logging context from thread local storage - - Returns: - LoggingContext: the current logging context - """ - return getattr(cls.thread_local, "current_context", cls.sentinel) - - @classmethod - def set_current_context( - cls, context: LoggingContextOrSentinel - ) -> LoggingContextOrSentinel: - """Set the current logging context in thread local storage - Args: - context(LoggingContext): The context to activate. - Returns: - The context that was previously active - """ - current = cls.current_context() - - if current is not context: - current.stop() - cls.thread_local.current_context = context - context.start() - return current - def __enter__(self) -> "LoggingContext": """Enters this logging context into thread local storage""" - old_context = self.set_current_context(self) + old_context = set_current_context(self) if self.previous_context != old_context: logger.warning( "Expected previous context %r, found %r", self.previous_context, old_context, ) - self.alive = True - return self def __exit__(self, type, value, traceback) -> None: @@ -329,24 +304,19 @@ class LoggingContext(object): Returns: None to avoid suppressing any exceptions that were thrown. """ - current = self.set_current_context(self.previous_context) + current = set_current_context(self.previous_context) if current is not self: - if current is self.sentinel: + if current is SENTINEL_CONTEXT: logger.warning("Expected logging context %s was lost", self) else: logger.warning( "Expected logging context %s but found %s", self, current ) - self.alive = False - - # if we have a parent, pass our CPU usage stats on - if self.parent_context is not None and hasattr( - self.parent_context, "_resource_usage" - ): - self.parent_context._resource_usage += self._resource_usage - # reset them in case we get entered again - self._resource_usage.reset() + # the fact that we are here suggests that the caller thinks that everything + # is done and dusted for this logcontext, and further activity will not get + # recorded against the correct metrics. + self.finished = True def copy_to(self, record) -> None: """Copy logging fields from this context to a log record or @@ -371,9 +341,14 @@ class LoggingContext(object): logger.warning("Started logcontext %s on different thread", self) return + if self.finished: + logger.warning("Re-starting finished log context %s", self) + # If we haven't already started record the thread resource usage so # far - if not self.usage_start: + if self.usage_start: + logger.warning("Re-starting already-active log context %s", self) + else: self.usage_start = get_thread_resource_usage() def stop(self) -> None: @@ -396,6 +371,15 @@ class LoggingContext(object): self.usage_start = None + # if we have a parent, pass our CPU usage stats on + if self.parent_context is not None and hasattr( + self.parent_context, "_resource_usage" + ): + self.parent_context._resource_usage += self._resource_usage + + # reset them in case we get entered again + self._resource_usage.reset() + def get_resource_usage(self) -> ContextResourceUsage: """Get resources used by this logcontext so far. @@ -409,7 +393,7 @@ class LoggingContext(object): # If we are on the correct thread and we're currently running then we # can include resource usage so far. is_main_thread = get_thread_id() == self.main_thread - if self.alive and self.usage_start and is_main_thread: + if self.usage_start and is_main_thread: utime_delta, stime_delta = self._get_cputime() res.ru_utime += utime_delta res.ru_stime += stime_delta @@ -492,7 +476,7 @@ class LoggingContextFilter(logging.Filter): Returns: True to include the record in the log output. """ - context = LoggingContext.current_context() + context = current_context() for key, value in self.defaults.items(): setattr(record, key, value) @@ -512,27 +496,24 @@ class PreserveLoggingContext(object): __slots__ = ["current_context", "new_context", "has_parent"] - def __init__(self, new_context: Optional[LoggingContextOrSentinel] = None) -> None: - if new_context is None: - self.new_context = LoggingContext.sentinel # type: LoggingContextOrSentinel - else: - self.new_context = new_context + def __init__( + self, new_context: LoggingContextOrSentinel = SENTINEL_CONTEXT + ) -> None: + self.new_context = new_context def __enter__(self) -> None: """Captures the current logging context""" - self.current_context = LoggingContext.set_current_context(self.new_context) + self.current_context = set_current_context(self.new_context) if self.current_context: self.has_parent = self.current_context.previous_context is not None - if not self.current_context.alive: - logger.debug("Entering dead context: %s", self.current_context) def __exit__(self, type, value, traceback) -> None: """Restores the current logging context""" - context = LoggingContext.set_current_context(self.current_context) + context = set_current_context(self.current_context) if context != self.new_context: - if context is LoggingContext.sentinel: + if not context: logger.warning("Expected logging context %s was lost", self.new_context) else: logger.warning( @@ -541,9 +522,30 @@ class PreserveLoggingContext(object): context, ) - if self.current_context is not LoggingContext.sentinel: - if not self.current_context.alive: - logger.debug("Restoring dead context: %s", self.current_context) + +_thread_local = threading.local() +_thread_local.current_context = SENTINEL_CONTEXT + + +def current_context() -> LoggingContextOrSentinel: + """Get the current logging context from thread local storage""" + return getattr(_thread_local, "current_context", SENTINEL_CONTEXT) + + +def set_current_context(context: LoggingContextOrSentinel) -> LoggingContextOrSentinel: + """Set the current logging context in thread local storage + Args: + context(LoggingContext): The context to activate. + Returns: + The context that was previously active + """ + current = current_context() + + if current is not context: + current.stop() + _thread_local.current_context = context + context.start() + return current def nested_logging_context( @@ -572,7 +574,7 @@ def nested_logging_context( if parent_context is not None: context = parent_context # type: LoggingContextOrSentinel else: - context = LoggingContext.current_context() + context = current_context() return LoggingContext( parent_context=context, request=str(context.request) + "-" + suffix ) @@ -604,7 +606,7 @@ def run_in_background(f, *args, **kwargs): CRITICAL error about an unhandled error will be logged without much indication about where it came from. """ - current = LoggingContext.current_context() + current = current_context() try: res = f(*args, **kwargs) except: # noqa: E722 @@ -625,7 +627,7 @@ def run_in_background(f, *args, **kwargs): # The function may have reset the context before returning, so # we need to restore it now. - ctx = LoggingContext.set_current_context(current) + ctx = set_current_context(current) # The original context will be restored when the deferred # completes, but there is nothing waiting for it, so it will @@ -674,7 +676,7 @@ def make_deferred_yieldable(deferred): # ok, we can't be sure that a yield won't block, so let's reset the # logcontext, and add a callback to the deferred to restore it. - prev_context = LoggingContext.set_current_context(LoggingContext.sentinel) + prev_context = set_current_context(SENTINEL_CONTEXT) deferred.addBoth(_set_context_cb, prev_context) return deferred @@ -684,7 +686,7 @@ ResultT = TypeVar("ResultT") def _set_context_cb(result: ResultT, context: LoggingContext) -> ResultT: """A callback function which just sets the logging context""" - LoggingContext.set_current_context(context) + set_current_context(context) return result @@ -752,7 +754,7 @@ def defer_to_threadpool(reactor, threadpool, f, *args, **kwargs): Deferred: A Deferred which fires a callback with the result of `f`, or an errback if `f` throws an exception. """ - logcontext = LoggingContext.current_context() + logcontext = current_context() def g(): with LoggingContext(parent_context=logcontext): diff --git a/synapse/logging/scopecontextmanager.py b/synapse/logging/scopecontextmanager.py index 4eed4f2338..dc3ab00cbb 100644 --- a/synapse/logging/scopecontextmanager.py +++ b/synapse/logging/scopecontextmanager.py @@ -19,7 +19,7 @@ from opentracing import Scope, ScopeManager import twisted -from synapse.logging.context import LoggingContext, nested_logging_context +from synapse.logging.context import current_context, nested_logging_context logger = logging.getLogger(__name__) @@ -49,11 +49,8 @@ class LogContextScopeManager(ScopeManager): (Scope) : the Scope that is active, or None if not available. """ - ctx = LoggingContext.current_context() - if ctx is LoggingContext.sentinel: - return None - else: - return ctx.scope + ctx = current_context() + return ctx.scope def activate(self, span, finish_on_close): """ @@ -70,9 +67,9 @@ class LogContextScopeManager(ScopeManager): """ enter_logcontext = False - ctx = LoggingContext.current_context() + ctx = current_context() - if ctx is LoggingContext.sentinel: + if not ctx: # We don't want this scope to affect. logger.error("Tried to activate scope outside of loggingcontext") return Scope(None, span) diff --git a/synapse/storage/data_stores/main/events_worker.py b/synapse/storage/data_stores/main/events_worker.py index ca237c6f12..3013f49d32 100644 --- a/synapse/storage/data_stores/main/events_worker.py +++ b/synapse/storage/data_stores/main/events_worker.py @@ -35,7 +35,7 @@ from synapse.api.room_versions import ( ) from synapse.events import make_event_from_dict from synapse.events.utils import prune_event -from synapse.logging.context import LoggingContext, PreserveLoggingContext +from synapse.logging.context import PreserveLoggingContext, current_context from synapse.metrics.background_process_metrics import run_as_background_process from synapse.storage._base import SQLBaseStore, make_in_list_sql_clause from synapse.storage.database import Database @@ -409,7 +409,7 @@ class EventsWorkerStore(SQLBaseStore): missing_events_ids = [e for e in event_ids if e not in event_entry_map] if missing_events_ids: - log_ctx = LoggingContext.current_context() + log_ctx = current_context() log_ctx.record_event_fetch(len(missing_events_ids)) # Note that _get_events_from_db is also responsible for turning db rows diff --git a/synapse/storage/database.py b/synapse/storage/database.py index e61595336c..715c0346dd 100644 --- a/synapse/storage/database.py +++ b/synapse/storage/database.py @@ -32,6 +32,7 @@ from synapse.config.database import DatabaseConnectionConfig from synapse.logging.context import ( LoggingContext, LoggingContextOrSentinel, + current_context, make_deferred_yieldable, ) from synapse.metrics.background_process_metrics import run_as_background_process @@ -483,7 +484,7 @@ class Database(object): end = monotonic_time() duration = end - start - LoggingContext.current_context().add_database_transaction(duration) + current_context().add_database_transaction(duration) transaction_logger.debug("[TXN END] {%s} %f sec", name, duration) @@ -510,7 +511,7 @@ class Database(object): after_callbacks = [] # type: List[_CallbackListEntry] exception_callbacks = [] # type: List[_CallbackListEntry] - if LoggingContext.current_context() == LoggingContext.sentinel: + if not current_context(): logger.warning("Starting db txn '%s' from sentinel context", desc) try: @@ -547,10 +548,8 @@ class Database(object): Returns: Deferred: The result of func """ - parent_context = ( - LoggingContext.current_context() - ) # type: Optional[LoggingContextOrSentinel] - if parent_context == LoggingContext.sentinel: + parent_context = current_context() # type: Optional[LoggingContextOrSentinel] + if not parent_context: logger.warning( "Starting db connection from sentinel context: metrics will be lost" ) diff --git a/synapse/util/metrics.py b/synapse/util/metrics.py index 7b18455469..ec61e14423 100644 --- a/synapse/util/metrics.py +++ b/synapse/util/metrics.py @@ -21,7 +21,7 @@ from prometheus_client import Counter from twisted.internet import defer -from synapse.logging.context import LoggingContext +from synapse.logging.context import LoggingContext, current_context from synapse.metrics import InFlightGauge logger = logging.getLogger(__name__) @@ -106,7 +106,7 @@ class Measure(object): raise RuntimeError("Measure() objects cannot be re-used") self.start = self.clock.time() - parent_context = LoggingContext.current_context() + parent_context = current_context() self._logging_context = LoggingContext( "Measure[%s]" % (self.name,), parent_context ) diff --git a/synapse/util/patch_inline_callbacks.py b/synapse/util/patch_inline_callbacks.py index 3925927f9f..fdff195771 100644 --- a/synapse/util/patch_inline_callbacks.py +++ b/synapse/util/patch_inline_callbacks.py @@ -32,7 +32,7 @@ def do_patch(): Patch defer.inlineCallbacks so that it checks the state of the logcontext on exit """ - from synapse.logging.context import LoggingContext + from synapse.logging.context import current_context global _already_patched @@ -43,35 +43,35 @@ def do_patch(): def new_inline_callbacks(f): @functools.wraps(f) def wrapped(*args, **kwargs): - start_context = LoggingContext.current_context() + start_context = current_context() changes = [] # type: List[str] orig = orig_inline_callbacks(_check_yield_points(f, changes)) try: res = orig(*args, **kwargs) except Exception: - if LoggingContext.current_context() != start_context: + if current_context() != start_context: for err in changes: print(err, file=sys.stderr) err = "%s changed context from %s to %s on exception" % ( f, start_context, - LoggingContext.current_context(), + current_context(), ) print(err, file=sys.stderr) raise Exception(err) raise if not isinstance(res, Deferred) or res.called: - if LoggingContext.current_context() != start_context: + if current_context() != start_context: for err in changes: print(err, file=sys.stderr) err = "Completed %s changed context from %s to %s" % ( f, start_context, - LoggingContext.current_context(), + current_context(), ) # print the error to stderr because otherwise all we # see in travis-ci is the 500 error @@ -79,23 +79,23 @@ def do_patch(): raise Exception(err) return res - if LoggingContext.current_context() != LoggingContext.sentinel: + if current_context(): err = ( "%s returned incomplete deferred in non-sentinel context " "%s (start was %s)" - ) % (f, LoggingContext.current_context(), start_context) + ) % (f, current_context(), start_context) print(err, file=sys.stderr) raise Exception(err) def check_ctx(r): - if LoggingContext.current_context() != start_context: + if current_context() != start_context: for err in changes: print(err, file=sys.stderr) err = "%s completion of %s changed context from %s to %s" % ( "Failure" if isinstance(r, Failure) else "Success", f, start_context, - LoggingContext.current_context(), + current_context(), ) print(err, file=sys.stderr) raise Exception(err) @@ -127,7 +127,7 @@ def _check_yield_points(f: Callable, changes: List[str]): function """ - from synapse.logging.context import LoggingContext + from synapse.logging.context import current_context @functools.wraps(f) def check_yield_points_inner(*args, **kwargs): @@ -136,7 +136,7 @@ def _check_yield_points(f: Callable, changes: List[str]): last_yield_line_no = gen.gi_frame.f_lineno result = None # type: Any while True: - expected_context = LoggingContext.current_context() + expected_context = current_context() try: isFailure = isinstance(result, Failure) @@ -145,7 +145,7 @@ def _check_yield_points(f: Callable, changes: List[str]): else: d = gen.send(result) except (StopIteration, defer._DefGen_Return) as e: - if LoggingContext.current_context() != expected_context: + if current_context() != expected_context: # This happens when the context is lost sometime *after* the # final yield and returning. E.g. we forgot to yield on a # function that returns a deferred. @@ -159,7 +159,7 @@ def _check_yield_points(f: Callable, changes: List[str]): % ( f.__qualname__, expected_context, - LoggingContext.current_context(), + current_context(), f.__code__.co_filename, last_yield_line_no, ) @@ -173,13 +173,13 @@ def _check_yield_points(f: Callable, changes: List[str]): # This happens if we yield on a deferred that doesn't follow # the log context rules without wrapping in a `make_deferred_yieldable`. # We raise here as this should never happen. - if LoggingContext.current_context() is not LoggingContext.sentinel: + if current_context(): err = ( "%s yielded with context %s rather than sentinel," " yielded on line %d in %s" % ( frame.f_code.co_name, - LoggingContext.current_context(), + current_context(), frame.f_lineno, frame.f_code.co_filename, ) @@ -191,7 +191,7 @@ def _check_yield_points(f: Callable, changes: List[str]): except Exception as e: result = Failure(e) - if LoggingContext.current_context() != expected_context: + if current_context() != expected_context: # This happens because the context is lost sometime *after* the # previous yield and *after* the current yield. E.g. the @@ -206,7 +206,7 @@ def _check_yield_points(f: Callable, changes: List[str]): % ( frame.f_code.co_name, expected_context, - LoggingContext.current_context(), + current_context(), last_yield_line_no, frame.f_lineno, frame.f_code.co_filename, diff --git a/tests/crypto/test_keyring.py b/tests/crypto/test_keyring.py index 34d5895f18..70c8e72303 100644 --- a/tests/crypto/test_keyring.py +++ b/tests/crypto/test_keyring.py @@ -34,6 +34,7 @@ from synapse.crypto.keyring import ( from synapse.logging.context import ( LoggingContext, PreserveLoggingContext, + current_context, make_deferred_yieldable, ) from synapse.storage.keys import FetchKeyResult @@ -83,9 +84,7 @@ class KeyringTestCase(unittest.HomeserverTestCase): ) def check_context(self, _, expected): - self.assertEquals( - getattr(LoggingContext.current_context(), "request", None), expected - ) + self.assertEquals(getattr(current_context(), "request", None), expected) def test_verify_json_objects_for_server_awaits_previous_requests(self): key1 = signedjson.key.generate_signing_key(1) @@ -105,7 +104,7 @@ class KeyringTestCase(unittest.HomeserverTestCase): @defer.inlineCallbacks def get_perspectives(**kwargs): - self.assertEquals(LoggingContext.current_context().request, "11") + self.assertEquals(current_context().request, "11") with PreserveLoggingContext(): yield persp_deferred return persp_resp diff --git a/tests/http/federation/test_matrix_federation_agent.py b/tests/http/federation/test_matrix_federation_agent.py index fdc1d918ff..562397cdda 100644 --- a/tests/http/federation/test_matrix_federation_agent.py +++ b/tests/http/federation/test_matrix_federation_agent.py @@ -38,7 +38,7 @@ from synapse.http.federation.well_known_resolver import ( WellKnownResolver, _cache_period_from_headers, ) -from synapse.logging.context import LoggingContext +from synapse.logging.context import SENTINEL_CONTEXT, LoggingContext, current_context from synapse.util.caches.ttlcache import TTLCache from tests import unittest @@ -155,7 +155,7 @@ class MatrixFederationAgentTests(unittest.TestCase): self.assertNoResult(fetch_d) # should have reset logcontext to the sentinel - _check_logcontext(LoggingContext.sentinel) + _check_logcontext(SENTINEL_CONTEXT) try: fetch_res = yield fetch_d @@ -1197,7 +1197,7 @@ class TestCachePeriodFromHeaders(unittest.TestCase): def _check_logcontext(context): - current = LoggingContext.current_context() + current = current_context() if current is not context: raise AssertionError("Expected logcontext %s but was %s" % (context, current)) diff --git a/tests/http/federation/test_srv_resolver.py b/tests/http/federation/test_srv_resolver.py index df034ab237..babc201643 100644 --- a/tests/http/federation/test_srv_resolver.py +++ b/tests/http/federation/test_srv_resolver.py @@ -22,7 +22,7 @@ from twisted.internet.error import ConnectError from twisted.names import dns, error from synapse.http.federation.srv_resolver import SrvResolver -from synapse.logging.context import LoggingContext +from synapse.logging.context import SENTINEL_CONTEXT, LoggingContext, current_context from tests import unittest from tests.utils import MockClock @@ -54,12 +54,12 @@ class SrvResolverTestCase(unittest.TestCase): self.assertNoResult(resolve_d) # should have reset to the sentinel context - self.assertIs(LoggingContext.current_context(), LoggingContext.sentinel) + self.assertIs(current_context(), SENTINEL_CONTEXT) result = yield resolve_d # should have restored our context - self.assertIs(LoggingContext.current_context(), ctx) + self.assertIs(current_context(), ctx) return result diff --git a/tests/http/test_fedclient.py b/tests/http/test_fedclient.py index 2b01f40a42..fff4f0cbf4 100644 --- a/tests/http/test_fedclient.py +++ b/tests/http/test_fedclient.py @@ -29,14 +29,14 @@ from synapse.http.matrixfederationclient import ( MatrixFederationHttpClient, MatrixFederationRequest, ) -from synapse.logging.context import LoggingContext +from synapse.logging.context import SENTINEL_CONTEXT, LoggingContext, current_context from tests.server import FakeTransport from tests.unittest import HomeserverTestCase def check_logcontext(context): - current = LoggingContext.current_context() + current = current_context() if current is not context: raise AssertionError("Expected logcontext %s but was %s" % (context, current)) @@ -64,7 +64,7 @@ class FederationClientTests(HomeserverTestCase): self.assertNoResult(fetch_d) # should have reset logcontext to the sentinel - check_logcontext(LoggingContext.sentinel) + check_logcontext(SENTINEL_CONTEXT) try: fetch_res = yield fetch_d diff --git a/tests/rest/client/test_transactions.py b/tests/rest/client/test_transactions.py index a3d7e3c046..171632e195 100644 --- a/tests/rest/client/test_transactions.py +++ b/tests/rest/client/test_transactions.py @@ -2,7 +2,7 @@ from mock import Mock, call from twisted.internet import defer, reactor -from synapse.logging.context import LoggingContext +from synapse.logging.context import SENTINEL_CONTEXT, LoggingContext, current_context from synapse.rest.client.transactions import CLEANUP_PERIOD_MS, HttpTransactionCache from synapse.util import Clock @@ -52,14 +52,14 @@ class HttpTransactionCacheTestCase(unittest.TestCase): def test(): with LoggingContext("c") as c1: res = yield self.cache.fetch_or_execute(self.mock_key, cb) - self.assertIs(LoggingContext.current_context(), c1) + self.assertIs(current_context(), c1) self.assertEqual(res, "yay") # run the test twice in parallel d = defer.gatherResults([test(), test()]) - self.assertIs(LoggingContext.current_context(), LoggingContext.sentinel) + self.assertIs(current_context(), SENTINEL_CONTEXT) yield d - self.assertIs(LoggingContext.current_context(), LoggingContext.sentinel) + self.assertIs(current_context(), SENTINEL_CONTEXT) @defer.inlineCallbacks def test_does_not_cache_exceptions(self): @@ -81,11 +81,11 @@ class HttpTransactionCacheTestCase(unittest.TestCase): yield self.cache.fetch_or_execute(self.mock_key, cb) except Exception as e: self.assertEqual(e.args[0], "boo") - self.assertIs(LoggingContext.current_context(), test_context) + self.assertIs(current_context(), test_context) res = yield self.cache.fetch_or_execute(self.mock_key, cb) self.assertEqual(res, self.mock_http_response) - self.assertIs(LoggingContext.current_context(), test_context) + self.assertIs(current_context(), test_context) @defer.inlineCallbacks def test_does_not_cache_failures(self): @@ -107,11 +107,11 @@ class HttpTransactionCacheTestCase(unittest.TestCase): yield self.cache.fetch_or_execute(self.mock_key, cb) except Exception as e: self.assertEqual(e.args[0], "boo") - self.assertIs(LoggingContext.current_context(), test_context) + self.assertIs(current_context(), test_context) res = yield self.cache.fetch_or_execute(self.mock_key, cb) self.assertEqual(res, self.mock_http_response) - self.assertIs(LoggingContext.current_context(), test_context) + self.assertIs(current_context(), test_context) @defer.inlineCallbacks def test_cleans_up(self): diff --git a/tests/unittest.py b/tests/unittest.py index 8816a4d152..439174dbfc 100644 --- a/tests/unittest.py +++ b/tests/unittest.py @@ -38,7 +38,11 @@ from synapse.config.ratelimiting import FederationRateLimitConfig from synapse.federation.transport import server as federation_server from synapse.http.server import JsonResource from synapse.http.site import SynapseRequest, SynapseSite -from synapse.logging.context import LoggingContext +from synapse.logging.context import ( + SENTINEL_CONTEXT, + current_context, + set_current_context, +) from synapse.server import HomeServer from synapse.types import Requester, UserID, create_requester from synapse.util.ratelimitutils import FederationRateLimiter @@ -97,10 +101,10 @@ class TestCase(unittest.TestCase): def setUp(orig): # if we're not starting in the sentinel logcontext, then to be honest # all future bets are off. - if LoggingContext.current_context() is not LoggingContext.sentinel: + if current_context(): self.fail( "Test starting with non-sentinel logging context %s" - % (LoggingContext.current_context(),) + % (current_context(),) ) old_level = logging.getLogger().level @@ -122,7 +126,7 @@ class TestCase(unittest.TestCase): # force a GC to workaround problems with deferreds leaking logcontexts when # they are GCed (see the logcontext docs) gc.collect() - LoggingContext.set_current_context(LoggingContext.sentinel) + set_current_context(SENTINEL_CONTEXT) return ret diff --git a/tests/util/caches/test_descriptors.py b/tests/util/caches/test_descriptors.py index 39e360fe24..4d2b9e0d64 100644 --- a/tests/util/caches/test_descriptors.py +++ b/tests/util/caches/test_descriptors.py @@ -22,8 +22,10 @@ from twisted.internet import defer, reactor from synapse.api.errors import SynapseError from synapse.logging.context import ( + SENTINEL_CONTEXT, LoggingContext, PreserveLoggingContext, + current_context, make_deferred_yieldable, ) from synapse.util.caches import descriptors @@ -194,7 +196,7 @@ class DescriptorTestCase(unittest.TestCase): with LoggingContext() as c1: c1.name = "c1" r = yield obj.fn(1) - self.assertEqual(LoggingContext.current_context(), c1) + self.assertEqual(current_context(), c1) return r def check_result(r): @@ -204,12 +206,12 @@ class DescriptorTestCase(unittest.TestCase): # set off a deferred which will do a cache lookup d1 = do_lookup() - self.assertEqual(LoggingContext.current_context(), LoggingContext.sentinel) + self.assertEqual(current_context(), SENTINEL_CONTEXT) d1.addCallback(check_result) # and another d2 = do_lookup() - self.assertEqual(LoggingContext.current_context(), LoggingContext.sentinel) + self.assertEqual(current_context(), SENTINEL_CONTEXT) d2.addCallback(check_result) # let the lookup complete @@ -239,14 +241,14 @@ class DescriptorTestCase(unittest.TestCase): try: d = obj.fn(1) self.assertEqual( - LoggingContext.current_context(), LoggingContext.sentinel + current_context(), SENTINEL_CONTEXT, ) yield d self.fail("No exception thrown") except SynapseError: pass - self.assertEqual(LoggingContext.current_context(), c1) + self.assertEqual(current_context(), c1) # the cache should now be empty self.assertEqual(len(obj.fn.cache.cache), 0) @@ -255,7 +257,7 @@ class DescriptorTestCase(unittest.TestCase): # set off a deferred which will do a cache lookup d1 = do_lookup() - self.assertEqual(LoggingContext.current_context(), LoggingContext.sentinel) + self.assertEqual(current_context(), SENTINEL_CONTEXT) return d1 @@ -366,10 +368,10 @@ class CachedListDescriptorTestCase(unittest.TestCase): @descriptors.cachedList("fn", "args1", inlineCallbacks=True) def list_fn(self, args1, arg2): - assert LoggingContext.current_context().request == "c1" + assert current_context().request == "c1" # we want this to behave like an asynchronous function yield run_on_reactor() - assert LoggingContext.current_context().request == "c1" + assert current_context().request == "c1" return self.mock(args1, arg2) with LoggingContext() as c1: @@ -377,9 +379,9 @@ class CachedListDescriptorTestCase(unittest.TestCase): obj = Cls() obj.mock.return_value = {10: "fish", 20: "chips"} d1 = obj.list_fn([10, 20], 2) - self.assertEqual(LoggingContext.current_context(), LoggingContext.sentinel) + self.assertEqual(current_context(), SENTINEL_CONTEXT) r = yield d1 - self.assertEqual(LoggingContext.current_context(), c1) + self.assertEqual(current_context(), c1) obj.mock.assert_called_once_with([10, 20], 2) self.assertEqual(r, {10: "fish", 20: "chips"}) obj.mock.reset_mock() diff --git a/tests/util/test_async_utils.py b/tests/util/test_async_utils.py index f60918069a..17fd86d02d 100644 --- a/tests/util/test_async_utils.py +++ b/tests/util/test_async_utils.py @@ -16,7 +16,12 @@ from twisted.internet import defer from twisted.internet.defer import CancelledError, Deferred from twisted.internet.task import Clock -from synapse.logging.context import LoggingContext, PreserveLoggingContext +from synapse.logging.context import ( + SENTINEL_CONTEXT, + LoggingContext, + PreserveLoggingContext, + current_context, +) from synapse.util.async_helpers import timeout_deferred from tests.unittest import TestCase @@ -79,10 +84,10 @@ class TimeoutDeferredTest(TestCase): # the errbacks should be run in the test logcontext def errback(res, deferred_name): self.assertIs( - LoggingContext.current_context(), + current_context(), context_one, "errback %s run in unexpected logcontext %s" - % (deferred_name, LoggingContext.current_context()), + % (deferred_name, current_context()), ) return res @@ -90,7 +95,7 @@ class TimeoutDeferredTest(TestCase): original_deferred.addErrback(errback, "orig") timing_out_d = timeout_deferred(original_deferred, 1.0, self.clock) self.assertNoResult(timing_out_d) - self.assertIs(LoggingContext.current_context(), LoggingContext.sentinel) + self.assertIs(current_context(), SENTINEL_CONTEXT) timing_out_d.addErrback(errback, "timingout") self.clock.pump((1.0,)) @@ -99,4 +104,4 @@ class TimeoutDeferredTest(TestCase): blocking_was_cancelled[0], "non-completing deferred was not cancelled" ) self.failureResultOf(timing_out_d, defer.TimeoutError) - self.assertIs(LoggingContext.current_context(), context_one) + self.assertIs(current_context(), context_one) diff --git a/tests/util/test_linearizer.py b/tests/util/test_linearizer.py index 0ec8ef90ce..852ef23185 100644 --- a/tests/util/test_linearizer.py +++ b/tests/util/test_linearizer.py @@ -19,7 +19,7 @@ from six.moves import range from twisted.internet import defer, reactor from twisted.internet.defer import CancelledError -from synapse.logging.context import LoggingContext +from synapse.logging.context import LoggingContext, current_context from synapse.util import Clock from synapse.util.async_helpers import Linearizer @@ -54,11 +54,11 @@ class LinearizerTestCase(unittest.TestCase): def func(i, sleep=False): with LoggingContext("func(%s)" % i) as lc: with (yield linearizer.queue("")): - self.assertEqual(LoggingContext.current_context(), lc) + self.assertEqual(current_context(), lc) if sleep: yield Clock(reactor).sleep(0) - self.assertEqual(LoggingContext.current_context(), lc) + self.assertEqual(current_context(), lc) func(0, sleep=True) for i in range(1, 100): diff --git a/tests/util/test_logcontext.py b/tests/util/test_logcontext.py index 281b32c4b8..95301c013c 100644 --- a/tests/util/test_logcontext.py +++ b/tests/util/test_logcontext.py @@ -2,8 +2,10 @@ import twisted.python.failure from twisted.internet import defer, reactor from synapse.logging.context import ( + SENTINEL_CONTEXT, LoggingContext, PreserveLoggingContext, + current_context, make_deferred_yieldable, nested_logging_context, run_in_background, @@ -15,7 +17,7 @@ from .. import unittest class LoggingContextTestCase(unittest.TestCase): def _check_test_key(self, value): - self.assertEquals(LoggingContext.current_context().request, value) + self.assertEquals(current_context().request, value) def test_with_context(self): with LoggingContext() as context_one: @@ -41,7 +43,7 @@ class LoggingContextTestCase(unittest.TestCase): self._check_test_key("one") def _test_run_in_background(self, function): - sentinel_context = LoggingContext.current_context() + sentinel_context = current_context() callback_completed = [False] @@ -71,7 +73,7 @@ class LoggingContextTestCase(unittest.TestCase): # make sure that the context was reset before it got thrown back # into the reactor try: - self.assertIs(LoggingContext.current_context(), sentinel_context) + self.assertIs(current_context(), sentinel_context) d2.callback(None) except BaseException: d2.errback(twisted.python.failure.Failure()) @@ -108,7 +110,7 @@ class LoggingContextTestCase(unittest.TestCase): async def testfunc(): self._check_test_key("one") d = Clock(reactor).sleep(0) - self.assertIs(LoggingContext.current_context(), LoggingContext.sentinel) + self.assertIs(current_context(), SENTINEL_CONTEXT) await d self._check_test_key("one") @@ -129,14 +131,14 @@ class LoggingContextTestCase(unittest.TestCase): reactor.callLater(0, d.callback, None) return d - sentinel_context = LoggingContext.current_context() + sentinel_context = current_context() with LoggingContext() as context_one: context_one.request = "one" d1 = make_deferred_yieldable(blocking_function()) # make sure that the context was reset by make_deferred_yieldable - self.assertIs(LoggingContext.current_context(), sentinel_context) + self.assertIs(current_context(), sentinel_context) yield d1 @@ -145,14 +147,14 @@ class LoggingContextTestCase(unittest.TestCase): @defer.inlineCallbacks def test_make_deferred_yieldable_with_chained_deferreds(self): - sentinel_context = LoggingContext.current_context() + sentinel_context = current_context() with LoggingContext() as context_one: context_one.request = "one" d1 = make_deferred_yieldable(_chained_deferred_function()) # make sure that the context was reset by make_deferred_yieldable - self.assertIs(LoggingContext.current_context(), sentinel_context) + self.assertIs(current_context(), sentinel_context) yield d1 @@ -189,14 +191,14 @@ class LoggingContextTestCase(unittest.TestCase): reactor.callLater(0, d.callback, None) await d - sentinel_context = LoggingContext.current_context() + sentinel_context = current_context() with LoggingContext() as context_one: context_one.request = "one" d1 = make_deferred_yieldable(blocking_function()) # make sure that the context was reset by make_deferred_yieldable - self.assertIs(LoggingContext.current_context(), sentinel_context) + self.assertIs(current_context(), sentinel_context) yield d1 diff --git a/tests/utils.py b/tests/utils.py index 513f358f4f..968d109f77 100644 --- a/tests/utils.py +++ b/tests/utils.py @@ -35,7 +35,7 @@ from synapse.config.homeserver import HomeServerConfig from synapse.config.server import DEFAULT_ROOM_VERSION from synapse.federation.transport import server as federation_server from synapse.http.server import HttpServer -from synapse.logging.context import LoggingContext +from synapse.logging.context import current_context, set_current_context from synapse.server import HomeServer from synapse.storage import DataStore from synapse.storage.engines import PostgresEngine, create_engine @@ -493,10 +493,10 @@ class MockClock(object): return self.time() * 1000 def call_later(self, delay, callback, *args, **kwargs): - current_context = LoggingContext.current_context() + ctx = current_context() def wrapped_callback(): - LoggingContext.thread_local.current_context = current_context + set_current_context(ctx) callback(*args, **kwargs) t = [self.now + delay, wrapped_callback, False] -- cgit 1.4.1 From 28d9d6e8a9d6a6d5162de41cada1b6d6d4b0f941 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Tue, 24 Mar 2020 18:33:49 +0000 Subject: Remove spurious "name" parameter to `default_config` this is never set to anything other than "test", and is a source of unnecessary boilerplate. --- tests/app/test_frontend_proxy.py | 4 ++-- tests/app/test_openid_listener.py | 4 ++-- tests/federation/test_complexity.py | 4 ++-- tests/handlers/test_register.py | 2 +- tests/rest/client/v2_alpha/test_register.py | 4 ++-- tests/rest/key/v2/test_remote_key_resource.py | 4 ++-- tests/server_notices/test_resource_limits_server_notices.py | 2 +- tests/test_terms_auth.py | 4 ++-- tests/unittest.py | 7 ++----- 9 files changed, 16 insertions(+), 19 deletions(-) (limited to 'tests') diff --git a/tests/app/test_frontend_proxy.py b/tests/app/test_frontend_proxy.py index d3feafa1b7..be20a89682 100644 --- a/tests/app/test_frontend_proxy.py +++ b/tests/app/test_frontend_proxy.py @@ -27,8 +27,8 @@ class FrontendProxyTests(HomeserverTestCase): return hs - def default_config(self, name="test"): - c = super().default_config(name) + def default_config(self): + c = super().default_config() c["worker_app"] = "synapse.app.frontend_proxy" return c diff --git a/tests/app/test_openid_listener.py b/tests/app/test_openid_listener.py index 89fcc3889a..7364f9f1ec 100644 --- a/tests/app/test_openid_listener.py +++ b/tests/app/test_openid_listener.py @@ -29,8 +29,8 @@ class FederationReaderOpenIDListenerTests(HomeserverTestCase): ) return hs - def default_config(self, name="test"): - conf = super().default_config(name) + def default_config(self): + conf = super().default_config() # we're using FederationReaderServer, which uses a SlavedStore, so we # have to tell the FederationHandler not to try to access stuff that is only # in the primary store. diff --git a/tests/federation/test_complexity.py b/tests/federation/test_complexity.py index 24fa8dbb45..94980733c4 100644 --- a/tests/federation/test_complexity.py +++ b/tests/federation/test_complexity.py @@ -33,8 +33,8 @@ class RoomComplexityTests(unittest.FederatingHomeserverTestCase): login.register_servlets, ] - def default_config(self, name="test"): - config = super().default_config(name=name) + def default_config(self): + config = super().default_config() config["limit_remote_rooms"] = {"enabled": True, "complexity": 0.05} return config diff --git a/tests/handlers/test_register.py b/tests/handlers/test_register.py index e2915eb7b1..e7b638dbfe 100644 --- a/tests/handlers/test_register.py +++ b/tests/handlers/test_register.py @@ -34,7 +34,7 @@ class RegistrationTestCase(unittest.HomeserverTestCase): """ Tests the RegistrationHandler. """ def make_homeserver(self, reactor, clock): - hs_config = self.default_config("test") + hs_config = self.default_config() # some of the tests rely on us having a user consent version hs_config["user_consent"] = { diff --git a/tests/rest/client/v2_alpha/test_register.py b/tests/rest/client/v2_alpha/test_register.py index d0c997e385..b6ed06e02d 100644 --- a/tests/rest/client/v2_alpha/test_register.py +++ b/tests/rest/client/v2_alpha/test_register.py @@ -36,8 +36,8 @@ class RegisterRestServletTestCase(unittest.HomeserverTestCase): servlets = [register.register_servlets] url = b"/_matrix/client/r0/register" - def default_config(self, name="test"): - config = super().default_config(name) + def default_config(self): + config = super().default_config() config["allow_guest_access"] = True return config diff --git a/tests/rest/key/v2/test_remote_key_resource.py b/tests/rest/key/v2/test_remote_key_resource.py index 6776a56cad..99eb477149 100644 --- a/tests/rest/key/v2/test_remote_key_resource.py +++ b/tests/rest/key/v2/test_remote_key_resource.py @@ -143,8 +143,8 @@ class EndToEndPerspectivesTests(BaseRemoteKeyResourceTestCase): endpoint, to check that the two implementations are compatible. """ - def default_config(self, *args, **kwargs): - config = super().default_config(*args, **kwargs) + def default_config(self): + config = super().default_config() # replace the signing key with our own self.hs_signing_key = signedjson.key.generate_signing_key("kssk") diff --git a/tests/server_notices/test_resource_limits_server_notices.py b/tests/server_notices/test_resource_limits_server_notices.py index eb540e34f6..0d27b92a86 100644 --- a/tests/server_notices/test_resource_limits_server_notices.py +++ b/tests/server_notices/test_resource_limits_server_notices.py @@ -28,7 +28,7 @@ from tests import unittest class TestResourceLimitsServerNotices(unittest.HomeserverTestCase): def make_homeserver(self, reactor, clock): - hs_config = self.default_config("test") + hs_config = self.default_config() hs_config["server_notices"] = { "system_mxid_localpart": "server", "system_mxid_display_name": "test display name", diff --git a/tests/test_terms_auth.py b/tests/test_terms_auth.py index 5ec5d2b358..81d796f3f3 100644 --- a/tests/test_terms_auth.py +++ b/tests/test_terms_auth.py @@ -28,8 +28,8 @@ from tests import unittest class TermsTestCase(unittest.HomeserverTestCase): servlets = [register_servlets] - def default_config(self, name="test"): - config = super().default_config(name) + def default_config(self): + config = super().default_config() config.update( { "public_baseurl": "https://example.org/", diff --git a/tests/unittest.py b/tests/unittest.py index 8816a4d152..23b59bea22 100644 --- a/tests/unittest.py +++ b/tests/unittest.py @@ -311,14 +311,11 @@ class HomeserverTestCase(TestCase): return resource - def default_config(self, name="test"): + def default_config(self): """ Get a default HomeServer config dict. - - Args: - name (str): The homeserver name/domain. """ - config = default_config(name) + config = default_config("test") # apply any additional config which was specified via the override_config # decorator. -- cgit 1.4.1 From 4cff617df1ba6f241fee6957cc44859f57edcc0e Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 25 Mar 2020 14:54:01 +0000 Subject: Move catchup of replication streams to worker. (#7024) This changes the replication protocol so that the server does not send down `RDATA` for rows that happened before the client connected. Instead, the server will send a `POSITION` and clients then query the database (or master out of band) to get up to date. --- changelog.d/7024.misc | 1 + docs/tcp_replication.md | 46 ++--- synapse/app/generic_worker.py | 3 + synapse/federation/sender/__init__.py | 9 + synapse/replication/http/__init__.py | 2 + synapse/replication/http/streams.py | 78 ++++++++ synapse/replication/slave/storage/_base.py | 14 +- synapse/replication/slave/storage/pushers.py | 3 + synapse/replication/tcp/client.py | 3 +- synapse/replication/tcp/commands.py | 34 +--- synapse/replication/tcp/protocol.py | 206 ++++++++-------------- synapse/replication/tcp/resource.py | 19 +- synapse/replication/tcp/streams/__init__.py | 8 +- synapse/replication/tcp/streams/_base.py | 160 +++++++++++------ synapse/replication/tcp/streams/events.py | 5 +- synapse/replication/tcp/streams/federation.py | 19 +- synapse/server.py | 5 + synapse/storage/data_stores/main/cache.py | 44 ++--- synapse/storage/data_stores/main/deviceinbox.py | 88 ++++----- synapse/storage/data_stores/main/events.py | 114 ------------ synapse/storage/data_stores/main/events_worker.py | 114 ++++++++++++ synapse/storage/data_stores/main/room.py | 40 ++--- tests/replication/tcp/streams/_base.py | 55 ++++-- tests/replication/tcp/streams/test_receipts.py | 52 +++++- 24 files changed, 635 insertions(+), 487 deletions(-) create mode 100644 changelog.d/7024.misc create mode 100644 synapse/replication/http/streams.py (limited to 'tests') diff --git a/changelog.d/7024.misc b/changelog.d/7024.misc new file mode 100644 index 0000000000..676f285377 --- /dev/null +++ b/changelog.d/7024.misc @@ -0,0 +1 @@ +Move catchup of replication streams logic to worker. diff --git a/docs/tcp_replication.md b/docs/tcp_replication.md index e3a4634b14..d4f7d9ec18 100644 --- a/docs/tcp_replication.md +++ b/docs/tcp_replication.md @@ -14,16 +14,16 @@ example flow would be (where '>' indicates master to worker and '<' worker to master flows): > SERVER example.com - < REPLICATE events 53 + < REPLICATE + > POSITION events 53 > RDATA events 54 ["$foo1:bar.com", ...] > RDATA events 55 ["$foo4:bar.com", ...] -The example shows the server accepting a new connection and sending its -identity with the `SERVER` command, followed by the client asking to -subscribe to the `events` stream from the token `53`. The server then -periodically sends `RDATA` commands which have the format -`RDATA `, where the format of `` is -defined by the individual streams. +The example shows the server accepting a new connection and sending its identity +with the `SERVER` command, followed by the client server to respond with the +position of all streams. The server then periodically sends `RDATA` commands +which have the format `RDATA `, where the format of +`` is defined by the individual streams. Error reporting happens by either the client or server sending an ERROR command, and usually the connection will be closed. @@ -32,9 +32,6 @@ Since the protocol is a simple line based, its possible to manually connect to the server using a tool like netcat. A few things should be noted when manually using the protocol: -- When subscribing to a stream using `REPLICATE`, the special token - `NOW` can be used to get all future updates. The special stream name - `ALL` can be used with `NOW` to subscribe to all available streams. - The federation stream is only available if federation sending has been disabled on the main process. - The server will only time connections out that have sent a `PING` @@ -91,9 +88,7 @@ The client: - Sends a `NAME` command, allowing the server to associate a human friendly name with the connection. This is optional. - Sends a `PING` as above -- For each stream the client wishes to subscribe to it sends a - `REPLICATE` with the `stream_name` and token it wants to subscribe - from. +- Sends a `REPLICATE` to get the current position of all streams. - On receipt of a `SERVER` command, checks that the server name matches the expected server name. @@ -140,9 +135,7 @@ the wire: > PING 1490197665618 < NAME synapse.app.appservice < PING 1490197665618 - < REPLICATE events 1 - < REPLICATE backfill 1 - < REPLICATE caches 1 + < REPLICATE > POSITION events 1 > POSITION backfill 1 > POSITION caches 1 @@ -181,9 +174,9 @@ client (C): #### POSITION (S) - The position of the stream has been updated. Sent to the client - after all missing updates for a stream have been sent to the client - and they're now up to date. + On receipt of a POSITION command clients should check if they have missed any + updates, and if so then fetch them out of band. Sent in response to a + REPLICATE command (but can happen at any time). #### ERROR (S, C) @@ -199,20 +192,7 @@ client (C): #### REPLICATE (C) -Asks the server to replicate a given stream. The syntax is: - -``` - REPLICATE -``` - -Where `` may be either: - * a numeric stream_id to stream updates since (exclusive) - * `NOW` to stream all subsequent updates. - -The `` is the name of a replication stream to subscribe -to (see [here](../synapse/replication/tcp/streams/_base.py) for a list -of streams). It can also be `ALL` to subscribe to all known streams, -in which case the `` must be set to `NOW`. +Asks the server for the current position of all streams. #### USER_SYNC (C) diff --git a/synapse/app/generic_worker.py b/synapse/app/generic_worker.py index bd1733573b..fba7ad9551 100644 --- a/synapse/app/generic_worker.py +++ b/synapse/app/generic_worker.py @@ -401,6 +401,9 @@ class GenericWorkerTyping(object): self._room_serials[row.room_id] = token self._room_typing[row.room_id] = row.user_ids + def get_current_token(self) -> int: + return self._latest_room_serial + class GenericWorkerSlavedStore( # FIXME(#3714): We need to add UserDirectoryStore as we write directly diff --git a/synapse/federation/sender/__init__.py b/synapse/federation/sender/__init__.py index 233cb33daf..a477578e44 100644 --- a/synapse/federation/sender/__init__.py +++ b/synapse/federation/sender/__init__.py @@ -499,4 +499,13 @@ class FederationSender(object): self._get_per_destination_queue(destination).attempt_new_transaction() def get_current_token(self) -> int: + # Dummy implementation for case where federation sender isn't offloaded + # to a worker. return 0 + + async def get_replication_rows( + self, from_token, to_token, limit, federation_ack=None + ): + # Dummy implementation for case where federation sender isn't offloaded + # to a worker. + return [] diff --git a/synapse/replication/http/__init__.py b/synapse/replication/http/__init__.py index 28dbc6fcba..4613b2538c 100644 --- a/synapse/replication/http/__init__.py +++ b/synapse/replication/http/__init__.py @@ -21,6 +21,7 @@ from synapse.replication.http import ( membership, register, send_event, + streams, ) REPLICATION_PREFIX = "/_synapse/replication" @@ -38,3 +39,4 @@ class ReplicationRestResource(JsonResource): login.register_servlets(hs, self) register.register_servlets(hs, self) devices.register_servlets(hs, self) + streams.register_servlets(hs, self) diff --git a/synapse/replication/http/streams.py b/synapse/replication/http/streams.py new file mode 100644 index 0000000000..ffd4c61993 --- /dev/null +++ b/synapse/replication/http/streams.py @@ -0,0 +1,78 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging + +from synapse.api.errors import SynapseError +from synapse.http.servlet import parse_integer +from synapse.replication.http._base import ReplicationEndpoint + +logger = logging.getLogger(__name__) + + +class ReplicationGetStreamUpdates(ReplicationEndpoint): + """Fetches stream updates from a server. Used for streams not persisted to + the database, e.g. typing notifications. + + The API looks like: + + GET /_synapse/replication/get_repl_stream_updates/events?from_token=0&to_token=10&limit=100 + + 200 OK + + { + updates: [ ... ], + upto_token: 10, + limited: False, + } + + """ + + NAME = "get_repl_stream_updates" + PATH_ARGS = ("stream_name",) + METHOD = "GET" + + def __init__(self, hs): + super().__init__(hs) + + # We pull the streams from the replication steamer (if we try and make + # them ourselves we end up in an import loop). + self.streams = hs.get_replication_streamer().get_streams() + + @staticmethod + def _serialize_payload(stream_name, from_token, upto_token, limit): + return {"from_token": from_token, "upto_token": upto_token, "limit": limit} + + async def _handle_request(self, request, stream_name): + stream = self.streams.get(stream_name) + if stream is None: + raise SynapseError(400, "Unknown stream") + + from_token = parse_integer(request, "from_token", required=True) + upto_token = parse_integer(request, "upto_token", required=True) + limit = parse_integer(request, "limit", required=True) + + updates, upto_token, limited = await stream.get_updates_since( + from_token, upto_token, limit + ) + + return ( + 200, + {"updates": updates, "upto_token": upto_token, "limited": limited}, + ) + + +def register_servlets(hs, http_server): + ReplicationGetStreamUpdates(hs).register(http_server) diff --git a/synapse/replication/slave/storage/_base.py b/synapse/replication/slave/storage/_base.py index f45cbd37a0..751c799d94 100644 --- a/synapse/replication/slave/storage/_base.py +++ b/synapse/replication/slave/storage/_base.py @@ -18,8 +18,10 @@ from typing import Dict, Optional import six -from synapse.storage._base import SQLBaseStore -from synapse.storage.data_stores.main.cache import CURRENT_STATE_CACHE_NAME +from synapse.storage.data_stores.main.cache import ( + CURRENT_STATE_CACHE_NAME, + CacheInvalidationWorkerStore, +) from synapse.storage.database import Database from synapse.storage.engines import PostgresEngine @@ -35,7 +37,7 @@ def __func__(inp): return inp.__func__ -class BaseSlavedStore(SQLBaseStore): +class BaseSlavedStore(CacheInvalidationWorkerStore): def __init__(self, database: Database, db_conn, hs): super(BaseSlavedStore, self).__init__(database, db_conn, hs) if isinstance(self.database_engine, PostgresEngine): @@ -60,6 +62,12 @@ class BaseSlavedStore(SQLBaseStore): pos["caches"] = self._cache_id_gen.get_current_token() return pos + def get_cache_stream_token(self): + if self._cache_id_gen: + return self._cache_id_gen.get_current_token() + else: + return 0 + def process_replication_rows(self, stream_name, token, rows): if stream_name == "caches": if self._cache_id_gen: diff --git a/synapse/replication/slave/storage/pushers.py b/synapse/replication/slave/storage/pushers.py index f22c2d44a3..bce8a3d115 100644 --- a/synapse/replication/slave/storage/pushers.py +++ b/synapse/replication/slave/storage/pushers.py @@ -33,6 +33,9 @@ class SlavedPusherStore(PusherWorkerStore, BaseSlavedStore): result["pushers"] = self._pushers_id_gen.get_current_token() return result + def get_pushers_stream_token(self): + return self._pushers_id_gen.get_current_token() + def process_replication_rows(self, stream_name, token, rows): if stream_name == "pushers": self._pushers_id_gen.advance(token) diff --git a/synapse/replication/tcp/client.py b/synapse/replication/tcp/client.py index 02ab5b66ea..7e7ad0f798 100644 --- a/synapse/replication/tcp/client.py +++ b/synapse/replication/tcp/client.py @@ -55,6 +55,7 @@ class ReplicationClientFactory(ReconnectingClientFactory): self.client_name = client_name self.handler = handler self.server_name = hs.config.server_name + self.hs = hs self._clock = hs.get_clock() # As self.clock is defined in super class hs.get_reactor().addSystemEventTrigger("before", "shutdown", self.stopTrying) @@ -65,7 +66,7 @@ class ReplicationClientFactory(ReconnectingClientFactory): def buildProtocol(self, addr): logger.info("Connected to replication: %r", addr) return ClientReplicationStreamProtocol( - self.client_name, self.server_name, self._clock, self.handler + self.hs, self.client_name, self.server_name, self._clock, self.handler, ) def clientConnectionLost(self, connector, reason): diff --git a/synapse/replication/tcp/commands.py b/synapse/replication/tcp/commands.py index 451671412d..5a6b734094 100644 --- a/synapse/replication/tcp/commands.py +++ b/synapse/replication/tcp/commands.py @@ -136,8 +136,8 @@ class PositionCommand(Command): """Sent by the server to tell the client the stream postition without needing to send an RDATA. - Sent to the client after all missing updates for a stream have been sent - to the client and they're now up to date. + On receipt of a POSITION command clients should check if they have missed + any updates, and if so then fetch them out of band. """ NAME = "POSITION" @@ -179,42 +179,24 @@ class NameCommand(Command): class ReplicateCommand(Command): - """Sent by the client to subscribe to the stream. + """Sent by the client to subscribe to streams. Format:: - REPLICATE - - Where may be either: - * a numeric stream_id to stream updates from - * "NOW" to stream all subsequent updates. - - The can be "ALL" to subscribe to all known streams, in which - case the must be set to "NOW", i.e.:: - - REPLICATE ALL NOW + REPLICATE """ NAME = "REPLICATE" - def __init__(self, stream_name, token): - self.stream_name = stream_name - self.token = token + def __init__(self): + pass @classmethod def from_line(cls, line): - stream_name, token = line.split(" ", 1) - if token in ("NOW", "now"): - token = "NOW" - else: - token = int(token) - return cls(stream_name, token) + return cls() def to_line(self): - return " ".join((self.stream_name, str(self.token))) - - def get_logcontext_id(self): - return "REPLICATE-" + self.stream_name + return "" class UserSyncCommand(Command): diff --git a/synapse/replication/tcp/protocol.py b/synapse/replication/tcp/protocol.py index bc1482a9bb..f81d2e2442 100644 --- a/synapse/replication/tcp/protocol.py +++ b/synapse/replication/tcp/protocol.py @@ -35,9 +35,7 @@ indicate which side is sending, these are *not* included on the wire:: > PING 1490197665618 < NAME synapse.app.appservice < PING 1490197665618 - < REPLICATE events 1 - < REPLICATE backfill 1 - < REPLICATE caches 1 + < REPLICATE > POSITION events 1 > POSITION backfill 1 > POSITION caches 1 @@ -53,17 +51,15 @@ import fcntl import logging import struct from collections import defaultdict -from typing import Any, DefaultDict, Dict, List, Set, Tuple +from typing import Any, DefaultDict, Dict, List, Set -from six import iteritems, iterkeys +from six import iteritems from prometheus_client import Counter -from twisted.internet import defer from twisted.protocols.basic import LineOnlyReceiver from twisted.python.failure import Failure -from synapse.logging.context import make_deferred_yieldable, run_in_background from synapse.metrics import LaterGauge from synapse.metrics.background_process_metrics import run_as_background_process from synapse.replication.tcp.commands import ( @@ -82,11 +78,16 @@ from synapse.replication.tcp.commands import ( SyncCommand, UserSyncCommand, ) -from synapse.replication.tcp.streams import STREAMS_MAP +from synapse.replication.tcp.streams import STREAMS_MAP, Stream from synapse.types import Collection from synapse.util import Clock from synapse.util.stringutils import random_string +MYPY = False +if MYPY: + from synapse.server import HomeServer + + connection_close_counter = Counter( "synapse_replication_tcp_protocol_close_reason", "", ["reason_type"] ) @@ -411,16 +412,6 @@ class ServerReplicationStreamProtocol(BaseReplicationStreamProtocol): self.server_name = server_name self.streamer = streamer - # The streams the client has subscribed to and is up to date with - self.replication_streams = set() # type: Set[str] - - # The streams the client is currently subscribing to. - self.connecting_streams = set() # type: Set[str] - - # Map from stream name to list of updates to send once we've finished - # subscribing the client to the stream. - self.pending_rdata = {} # type: Dict[str, List[Tuple[int, Any]]] - def connectionMade(self): self.send_command(ServerCommand(self.server_name)) BaseReplicationStreamProtocol.connectionMade(self) @@ -436,21 +427,10 @@ class ServerReplicationStreamProtocol(BaseReplicationStreamProtocol): ) async def on_REPLICATE(self, cmd): - stream_name = cmd.stream_name - token = cmd.token - - if stream_name == "ALL": - # Subscribe to all streams we're publishing to. - deferreds = [ - run_in_background(self.subscribe_to_stream, stream, token) - for stream in iterkeys(self.streamer.streams_by_name) - ] - - await make_deferred_yieldable( - defer.gatherResults(deferreds, consumeErrors=True) - ) - else: - await self.subscribe_to_stream(stream_name, token) + # Subscribe to all streams we're publishing to. + for stream_name in self.streamer.streams_by_name: + current_token = self.streamer.get_stream_token(stream_name) + self.send_command(PositionCommand(stream_name, current_token)) async def on_FEDERATION_ACK(self, cmd): self.streamer.federation_ack(cmd.token) @@ -474,87 +454,12 @@ class ServerReplicationStreamProtocol(BaseReplicationStreamProtocol): cmd.last_seen, ) - async def subscribe_to_stream(self, stream_name, token): - """Subscribe the remote to a stream. - - This invloves checking if they've missed anything and sending those - updates down if they have. During that time new updates for the stream - are queued and sent once we've sent down any missed updates. - """ - self.replication_streams.discard(stream_name) - self.connecting_streams.add(stream_name) - - try: - # Get missing updates - updates, current_token = await self.streamer.get_stream_updates( - stream_name, token - ) - - # Send all the missing updates - for update in updates: - token, row = update[0], update[1] - self.send_command(RdataCommand(stream_name, token, row)) - - # We send a POSITION command to ensure that they have an up to - # date token (especially useful if we didn't send any updates - # above) - self.send_command(PositionCommand(stream_name, current_token)) - - # Now we can send any updates that came in while we were subscribing - pending_rdata = self.pending_rdata.pop(stream_name, []) - updates = [] - for token, update in pending_rdata: - # If the token is null, it is part of a batch update. Batches - # are multiple updates that share a single token. To denote - # this, the token is set to None for all tokens in the batch - # except for the last. If we find a None token, we keep looking - # through tokens until we find one that is not None and then - # process all previous updates in the batch as if they had the - # final token. - if token is None: - # Store this update as part of a batch - updates.append(update) - continue - - if token <= current_token: - # This update or batch of updates is older than - # current_token, dismiss it - updates = [] - continue - - updates.append(update) - - # Send all updates that are part of this batch with the - # found token - for update in updates: - self.send_command(RdataCommand(stream_name, token, update)) - - # Clear stored updates - updates = [] - - # They're now fully subscribed - self.replication_streams.add(stream_name) - except Exception as e: - logger.exception("[%s] Failed to handle REPLICATE command", self.id()) - self.send_error("failed to handle replicate: %r", e) - finally: - self.connecting_streams.discard(stream_name) - def stream_update(self, stream_name, token, data): """Called when a new update is available to stream to clients. We need to check if the client is interested in the stream or not """ - if stream_name in self.replication_streams: - # The client is subscribed to the stream - self.send_command(RdataCommand(stream_name, token, data)) - elif stream_name in self.connecting_streams: - # The client is being subscribed to the stream - logger.debug("[%s] Queuing RDATA %r %r", self.id(), stream_name, token) - self.pending_rdata.setdefault(stream_name, []).append((token, data)) - else: - # The client isn't subscribed - logger.debug("[%s] Dropping RDATA %r %r", self.id(), stream_name, token) + self.send_command(RdataCommand(stream_name, token, data)) def send_sync(self, data): self.send_command(SyncCommand(data)) @@ -638,6 +543,7 @@ class ClientReplicationStreamProtocol(BaseReplicationStreamProtocol): def __init__( self, + hs: "HomeServer", client_name: str, server_name: str, clock: Clock, @@ -649,22 +555,25 @@ class ClientReplicationStreamProtocol(BaseReplicationStreamProtocol): self.server_name = server_name self.handler = handler + self.streams = { + stream.NAME: stream(hs) for stream in STREAMS_MAP.values() + } # type: Dict[str, Stream] + # Set of stream names that have been subscribe to, but haven't yet # caught up with. This is used to track when the client has been fully # connected to the remote. - self.streams_connecting = set() # type: Set[str] + self.streams_connecting = set(STREAMS_MAP) # type: Set[str] # Map of stream to batched updates. See RdataCommand for info on how # batching works. - self.pending_batches = {} # type: Dict[str, Any] + self.pending_batches = {} # type: Dict[str, List[Any]] def connectionMade(self): self.send_command(NameCommand(self.client_name)) BaseReplicationStreamProtocol.connectionMade(self) # Once we've connected subscribe to the necessary streams - for stream_name, token in iteritems(self.handler.get_streams_to_replicate()): - self.replicate(stream_name, token) + self.replicate() # Tell the server if we have any users currently syncing (should only # happen on synchrotrons) @@ -676,10 +585,6 @@ class ClientReplicationStreamProtocol(BaseReplicationStreamProtocol): # We've now finished connecting to so inform the client handler self.handler.update_connection(self) - # This will happen if we don't actually subscribe to any streams - if not self.streams_connecting: - self.handler.finished_connecting() - async def on_SERVER(self, cmd): if cmd.data != self.server_name: logger.error("[%s] Connected to wrong remote: %r", self.id(), cmd.data) @@ -697,7 +602,7 @@ class ClientReplicationStreamProtocol(BaseReplicationStreamProtocol): ) raise - if cmd.token is None: + if cmd.token is None or stream_name in self.streams_connecting: # I.e. this is part of a batch of updates for this stream. Batch # until we get an update for the stream with a non None token self.pending_batches.setdefault(stream_name, []).append(row) @@ -707,14 +612,55 @@ class ClientReplicationStreamProtocol(BaseReplicationStreamProtocol): rows.append(row) await self.handler.on_rdata(stream_name, cmd.token, rows) - async def on_POSITION(self, cmd): - # When we get a `POSITION` command it means we've finished getting - # missing updates for the given stream, and are now up to date. + async def on_POSITION(self, cmd: PositionCommand): + stream = self.streams.get(cmd.stream_name) + if not stream: + logger.error("Got POSITION for unknown stream: %s", cmd.stream_name) + return + + # Find where we previously streamed up to. + current_token = self.handler.get_streams_to_replicate().get(cmd.stream_name) + if current_token is None: + logger.warning( + "Got POSITION for stream we're not subscribed to: %s", cmd.stream_name + ) + return + + # Fetch all updates between then and now. + limited = True + while limited: + updates, current_token, limited = await stream.get_updates_since( + current_token, cmd.token + ) + + # Check if the connection was closed underneath us, if so we bail + # rather than risk having concurrent catch ups going on. + if self.state == ConnectionStates.CLOSED: + return + + if updates: + await self.handler.on_rdata( + cmd.stream_name, + current_token, + [stream.parse_row(update[1]) for update in updates], + ) + + # We've now caught up to position sent to us, notify handler. + await self.handler.on_position(cmd.stream_name, cmd.token) + self.streams_connecting.discard(cmd.stream_name) if not self.streams_connecting: self.handler.finished_connecting() - await self.handler.on_position(cmd.stream_name, cmd.token) + # Check if the connection was closed underneath us, if so we bail + # rather than risk having concurrent catch ups going on. + if self.state == ConnectionStates.CLOSED: + return + + # Handle any RDATA that came in while we were catching up. + rows = self.pending_batches.pop(cmd.stream_name, []) + if rows: + await self.handler.on_rdata(cmd.stream_name, rows[-1].token, rows) async def on_SYNC(self, cmd): self.handler.on_sync(cmd.data) @@ -722,22 +668,12 @@ class ClientReplicationStreamProtocol(BaseReplicationStreamProtocol): async def on_REMOTE_SERVER_UP(self, cmd: RemoteServerUpCommand): self.handler.on_remote_server_up(cmd.data) - def replicate(self, stream_name, token): + def replicate(self): """Send the subscription request to the server """ - if stream_name not in STREAMS_MAP: - raise Exception("Invalid stream name %r" % (stream_name,)) - - logger.info( - "[%s] Subscribing to replication stream: %r from %r", - self.id(), - stream_name, - token, - ) - - self.streams_connecting.add(stream_name) + logger.info("[%s] Subscribing to replication streams", self.id()) - self.send_command(ReplicateCommand(stream_name, token)) + self.send_command(ReplicateCommand()) def on_connection_closed(self): BaseReplicationStreamProtocol.on_connection_closed(self) diff --git a/synapse/replication/tcp/resource.py b/synapse/replication/tcp/resource.py index 6e2ebaf614..4374e99e32 100644 --- a/synapse/replication/tcp/resource.py +++ b/synapse/replication/tcp/resource.py @@ -17,7 +17,7 @@ import logging import random -from typing import Any, List +from typing import Any, Dict, List from six import itervalues @@ -30,7 +30,7 @@ from synapse.metrics.background_process_metrics import run_as_background_process from synapse.util.metrics import Measure, measure_func from .protocol import ServerReplicationStreamProtocol -from .streams import STREAMS_MAP +from .streams import STREAMS_MAP, Stream from .streams.federation import FederationStream stream_updates_counter = Counter( @@ -52,7 +52,7 @@ class ReplicationStreamProtocolFactory(Factory): """ def __init__(self, hs): - self.streamer = ReplicationStreamer(hs) + self.streamer = hs.get_replication_streamer() self.clock = hs.get_clock() self.server_name = hs.config.server_name @@ -133,6 +133,11 @@ class ReplicationStreamer(object): for conn in self.connections: conn.send_error("server shutting down") + def get_streams(self) -> Dict[str, Stream]: + """Get a mapp from stream name to stream instance. + """ + return self.streams_by_name + def on_notifier_poke(self): """Checks if there is actually any new data and sends it to the connections if there are. @@ -190,7 +195,8 @@ class ReplicationStreamer(object): stream.current_token(), ) try: - updates, current_token = await stream.get_updates() + updates, current_token, limited = await stream.get_updates() + self.pending_updates |= limited except Exception: logger.info("Failed to handle stream %s", stream.NAME) raise @@ -226,8 +232,7 @@ class ReplicationStreamer(object): self.pending_updates = False self.is_looping = False - @measure_func("repl.get_stream_updates") - async def get_stream_updates(self, stream_name, token): + def get_stream_token(self, stream_name): """For a given stream get all updates since token. This is called when a client first subscribes to a stream. """ @@ -235,7 +240,7 @@ class ReplicationStreamer(object): if not stream: raise Exception("unknown stream %s", stream_name) - return await stream.get_updates_since(token) + return stream.current_token() @measure_func("repl.federation_ack") def federation_ack(self, token): diff --git a/synapse/replication/tcp/streams/__init__.py b/synapse/replication/tcp/streams/__init__.py index 29199f5b46..37bcd3de66 100644 --- a/synapse/replication/tcp/streams/__init__.py +++ b/synapse/replication/tcp/streams/__init__.py @@ -24,6 +24,9 @@ Each stream is defined by the following information: current_token: The function that returns the current token for the stream update_function: The function that returns a list of updates between two tokens """ + +from typing import Dict, Type + from synapse.replication.tcp.streams._base import ( AccountDataStream, BackfillStream, @@ -35,6 +38,7 @@ from synapse.replication.tcp.streams._base import ( PushersStream, PushRulesStream, ReceiptsStream, + Stream, TagAccountDataStream, ToDeviceStream, TypingStream, @@ -63,10 +67,12 @@ STREAMS_MAP = { GroupServerStream, UserSignatureStream, ) -} +} # type: Dict[str, Type[Stream]] + __all__ = [ "STREAMS_MAP", + "Stream", "BackfillStream", "PresenceStream", "TypingStream", diff --git a/synapse/replication/tcp/streams/_base.py b/synapse/replication/tcp/streams/_base.py index 32d9514883..c14dff6c64 100644 --- a/synapse/replication/tcp/streams/_base.py +++ b/synapse/replication/tcp/streams/_base.py @@ -14,13 +14,13 @@ # See the License for the specific language governing permissions and # limitations under the License. -import itertools import logging from collections import namedtuple -from typing import Any, List, Optional, Tuple +from typing import Any, Awaitable, Callable, List, Optional, Tuple import attr +from synapse.replication.http.streams import ReplicationGetStreamUpdates from synapse.types import JsonDict logger = logging.getLogger(__name__) @@ -29,6 +29,15 @@ logger = logging.getLogger(__name__) MAX_EVENTS_BEHIND = 500000 +# Some type aliases to make things a bit easier. + +# A stream position token +Token = int + +# A pair of position in stream and args used to create an instance of `ROW_TYPE`. +StreamRow = Tuple[Token, tuple] + + class Stream(object): """Base class for the streams. @@ -56,6 +65,7 @@ class Stream(object): return cls.ROW_TYPE(*row) def __init__(self, hs): + # The token from which we last asked for updates self.last_token = self.current_token() @@ -65,61 +75,46 @@ class Stream(object): """ self.last_token = self.current_token() - async def get_updates(self): + async def get_updates(self) -> Tuple[List[Tuple[Token, JsonDict]], Token, bool]: """Gets all updates since the last time this function was called (or since the stream was constructed if it hadn't been called before). Returns: - Deferred[Tuple[List[Tuple[int, Any]], int]: - Resolves to a pair ``(updates, current_token)``, where ``updates`` is a - list of ``(token, row)`` entries. ``row`` will be json-serialised and - sent over the replication steam. + A triplet `(updates, new_last_token, limited)`, where `updates` is + a list of `(token, row)` entries, `new_last_token` is the new + position in stream, and `limited` is whether there are more updates + to fetch. """ - updates, current_token = await self.get_updates_since(self.last_token) + current_token = self.current_token() + updates, current_token, limited = await self.get_updates_since( + self.last_token, current_token + ) self.last_token = current_token - return updates, current_token + return updates, current_token, limited async def get_updates_since( - self, from_token: int - ) -> Tuple[List[Tuple[int, JsonDict]], int]: + self, from_token: Token, upto_token: Token, limit: int = 100 + ) -> Tuple[List[Tuple[Token, JsonDict]], Token, bool]: """Like get_updates except allows specifying from when we should stream updates Returns: - Resolves to a pair `(updates, new_last_token)`, where `updates` is - a list of `(token, row)` entries and `new_last_token` is the new - position in stream. + A triplet `(updates, new_last_token, limited)`, where `updates` is + a list of `(token, row)` entries, `new_last_token` is the new + position in stream, and `limited` is whether there are more updates + to fetch. """ - if from_token in ("NOW", "now"): - return [], self.current_token() - - current_token = self.current_token() - from_token = int(from_token) - if from_token == current_token: - return [], current_token + if from_token == upto_token: + return [], upto_token, False - rows = await self.update_function( - from_token, current_token, limit=MAX_EVENTS_BEHIND + 1 + updates, upto_token, limited = await self.update_function( + from_token, upto_token, limit=limit, ) - - # never turn more than MAX_EVENTS_BEHIND + 1 into updates. - rows = itertools.islice(rows, MAX_EVENTS_BEHIND + 1) - - updates = [(row[0], row[1:]) for row in rows] - - # check we didn't get more rows than the limit. - # doing it like this allows the update_function to be a generator. - if len(updates) >= MAX_EVENTS_BEHIND: - raise Exception("stream %s has fallen behind" % (self.NAME)) - - # The update function didn't hit the limit, so we must have got all - # the updates to `current_token`, and can return that as our new - # stream position. - return updates, current_token + return updates, upto_token, limited def current_token(self): """Gets the current token of the underlying streams. Should be provided @@ -141,6 +136,48 @@ class Stream(object): raise NotImplementedError() +def db_query_to_update_function( + query_function: Callable[[Token, Token, int], Awaitable[List[tuple]]] +) -> Callable[[Token, Token, int], Awaitable[Tuple[List[StreamRow], Token, bool]]]: + """Wraps a db query function which returns a list of rows to make it + suitable for use as an `update_function` for the Stream class + """ + + async def update_function(from_token, upto_token, limit): + rows = await query_function(from_token, upto_token, limit) + updates = [(row[0], row[1:]) for row in rows] + limited = False + if len(updates) == limit: + upto_token = rows[-1][0] + limited = True + + return updates, upto_token, limited + + return update_function + + +def make_http_update_function( + hs, stream_name: str +) -> Callable[[Token, Token, Token], Awaitable[Tuple[List[StreamRow], Token, bool]]]: + """Makes a suitable function for use as an `update_function` that queries + the master process for updates. + """ + + client = ReplicationGetStreamUpdates.make_client(hs) + + async def update_function( + from_token: int, upto_token: int, limit: int + ) -> Tuple[List[Tuple[int, tuple]], int, bool]: + return await client( + stream_name=stream_name, + from_token=from_token, + upto_token=upto_token, + limit=limit, + ) + + return update_function + + class BackfillStream(Stream): """We fetched some old events and either we had never seen that event before or it went from being an outlier to not. @@ -164,7 +201,7 @@ class BackfillStream(Stream): def __init__(self, hs): store = hs.get_datastore() self.current_token = store.get_current_backfill_token # type: ignore - self.update_function = store.get_all_new_backfill_event_rows # type: ignore + self.update_function = db_query_to_update_function(store.get_all_new_backfill_event_rows) # type: ignore super(BackfillStream, self).__init__(hs) @@ -190,8 +227,15 @@ class PresenceStream(Stream): store = hs.get_datastore() presence_handler = hs.get_presence_handler() + self._is_worker = hs.config.worker_app is not None + self.current_token = store.get_current_presence_token # type: ignore - self.update_function = presence_handler.get_all_presence_updates # type: ignore + + if hs.config.worker_app is None: + self.update_function = db_query_to_update_function(presence_handler.get_all_presence_updates) # type: ignore + else: + # Query master process + self.update_function = make_http_update_function(hs, self.NAME) # type: ignore super(PresenceStream, self).__init__(hs) @@ -208,7 +252,12 @@ class TypingStream(Stream): typing_handler = hs.get_typing_handler() self.current_token = typing_handler.get_current_token # type: ignore - self.update_function = typing_handler.get_all_typing_updates # type: ignore + + if hs.config.worker_app is None: + self.update_function = db_query_to_update_function(typing_handler.get_all_typing_updates) # type: ignore + else: + # Query master process + self.update_function = make_http_update_function(hs, self.NAME) # type: ignore super(TypingStream, self).__init__(hs) @@ -232,7 +281,7 @@ class ReceiptsStream(Stream): store = hs.get_datastore() self.current_token = store.get_max_receipt_stream_id # type: ignore - self.update_function = store.get_all_updated_receipts # type: ignore + self.update_function = db_query_to_update_function(store.get_all_updated_receipts) # type: ignore super(ReceiptsStream, self).__init__(hs) @@ -256,7 +305,13 @@ class PushRulesStream(Stream): async def update_function(self, from_token, to_token, limit): rows = await self.store.get_all_push_rule_updates(from_token, to_token, limit) - return [(row[0], row[2]) for row in rows] + + limited = False + if len(rows) == limit: + to_token = rows[-1][0] + limited = True + + return [(row[0], (row[2],)) for row in rows], to_token, limited class PushersStream(Stream): @@ -275,7 +330,7 @@ class PushersStream(Stream): store = hs.get_datastore() self.current_token = store.get_pushers_stream_token # type: ignore - self.update_function = store.get_all_updated_pushers_rows # type: ignore + self.update_function = db_query_to_update_function(store.get_all_updated_pushers_rows) # type: ignore super(PushersStream, self).__init__(hs) @@ -307,7 +362,7 @@ class CachesStream(Stream): store = hs.get_datastore() self.current_token = store.get_cache_stream_token # type: ignore - self.update_function = store.get_all_updated_caches # type: ignore + self.update_function = db_query_to_update_function(store.get_all_updated_caches) # type: ignore super(CachesStream, self).__init__(hs) @@ -333,7 +388,7 @@ class PublicRoomsStream(Stream): store = hs.get_datastore() self.current_token = store.get_current_public_room_stream_id # type: ignore - self.update_function = store.get_all_new_public_rooms # type: ignore + self.update_function = db_query_to_update_function(store.get_all_new_public_rooms) # type: ignore super(PublicRoomsStream, self).__init__(hs) @@ -354,7 +409,7 @@ class DeviceListsStream(Stream): store = hs.get_datastore() self.current_token = store.get_device_stream_token # type: ignore - self.update_function = store.get_all_device_list_changes_for_remotes # type: ignore + self.update_function = db_query_to_update_function(store.get_all_device_list_changes_for_remotes) # type: ignore super(DeviceListsStream, self).__init__(hs) @@ -372,7 +427,7 @@ class ToDeviceStream(Stream): store = hs.get_datastore() self.current_token = store.get_to_device_stream_token # type: ignore - self.update_function = store.get_all_new_device_messages # type: ignore + self.update_function = db_query_to_update_function(store.get_all_new_device_messages) # type: ignore super(ToDeviceStream, self).__init__(hs) @@ -392,7 +447,7 @@ class TagAccountDataStream(Stream): store = hs.get_datastore() self.current_token = store.get_max_account_data_stream_id # type: ignore - self.update_function = store.get_all_updated_tags # type: ignore + self.update_function = db_query_to_update_function(store.get_all_updated_tags) # type: ignore super(TagAccountDataStream, self).__init__(hs) @@ -412,10 +467,11 @@ class AccountDataStream(Stream): self.store = hs.get_datastore() self.current_token = self.store.get_max_account_data_stream_id # type: ignore + self.update_function = db_query_to_update_function(self._update_function) # type: ignore super(AccountDataStream, self).__init__(hs) - async def update_function(self, from_token, to_token, limit): + async def _update_function(self, from_token, to_token, limit): global_results, room_results = await self.store.get_all_updated_account_data( from_token, from_token, to_token, limit ) @@ -442,7 +498,7 @@ class GroupServerStream(Stream): store = hs.get_datastore() self.current_token = store.get_group_stream_token # type: ignore - self.update_function = store.get_all_groups_changes # type: ignore + self.update_function = db_query_to_update_function(store.get_all_groups_changes) # type: ignore super(GroupServerStream, self).__init__(hs) @@ -460,6 +516,6 @@ class UserSignatureStream(Stream): store = hs.get_datastore() self.current_token = store.get_device_stream_token # type: ignore - self.update_function = store.get_all_user_signature_changes_for_remotes # type: ignore + self.update_function = db_query_to_update_function(store.get_all_user_signature_changes_for_remotes) # type: ignore super(UserSignatureStream, self).__init__(hs) diff --git a/synapse/replication/tcp/streams/events.py b/synapse/replication/tcp/streams/events.py index b3afabb8cd..c6a595629f 100644 --- a/synapse/replication/tcp/streams/events.py +++ b/synapse/replication/tcp/streams/events.py @@ -19,7 +19,7 @@ from typing import Tuple, Type import attr -from ._base import Stream +from ._base import Stream, db_query_to_update_function """Handling of the 'events' replication stream @@ -117,10 +117,11 @@ class EventsStream(Stream): def __init__(self, hs): self._store = hs.get_datastore() self.current_token = self._store.get_current_events_token # type: ignore + self.update_function = db_query_to_update_function(self._update_function) # type: ignore super(EventsStream, self).__init__(hs) - async def update_function(self, from_token, current_token, limit=None): + async def _update_function(self, from_token, current_token, limit=None): event_rows = await self._store.get_all_new_forward_event_rows( from_token, current_token, limit ) diff --git a/synapse/replication/tcp/streams/federation.py b/synapse/replication/tcp/streams/federation.py index f5f9336430..48c1d45718 100644 --- a/synapse/replication/tcp/streams/federation.py +++ b/synapse/replication/tcp/streams/federation.py @@ -15,7 +15,9 @@ # limitations under the License. from collections import namedtuple -from ._base import Stream +from twisted.internet import defer + +from synapse.replication.tcp.streams._base import Stream, db_query_to_update_function class FederationStream(Stream): @@ -33,11 +35,18 @@ class FederationStream(Stream): NAME = "federation" ROW_TYPE = FederationStreamRow + _QUERY_MASTER = True def __init__(self, hs): - federation_sender = hs.get_federation_sender() - - self.current_token = federation_sender.get_current_token # type: ignore - self.update_function = federation_sender.get_replication_rows # type: ignore + # Not all synapse instances will have a federation sender instance, + # whether that's a `FederationSender` or a `FederationRemoteSendQueue`, + # so we stub the stream out when that is the case. + if hs.config.worker_app is None or hs.should_send_federation(): + federation_sender = hs.get_federation_sender() + self.current_token = federation_sender.get_current_token # type: ignore + self.update_function = db_query_to_update_function(federation_sender.get_replication_rows) # type: ignore + else: + self.current_token = lambda: 0 # type: ignore + self.update_function = lambda from_token, upto_token, limit: defer.succeed(([], upto_token, bool)) # type: ignore super(FederationStream, self).__init__(hs) diff --git a/synapse/server.py b/synapse/server.py index 1b980371de..9426eb1672 100644 --- a/synapse/server.py +++ b/synapse/server.py @@ -85,6 +85,7 @@ from synapse.http.matrixfederationclient import MatrixFederationHttpClient from synapse.notifier import Notifier from synapse.push.action_generator import ActionGenerator from synapse.push.pusherpool import PusherPool +from synapse.replication.tcp.resource import ReplicationStreamer from synapse.rest.media.v1.media_repository import ( MediaRepository, MediaRepositoryResource, @@ -199,6 +200,7 @@ class HomeServer(object): "saml_handler", "event_client_serializer", "storage", + "replication_streamer", ] REQUIRED_ON_MASTER_STARTUP = ["user_directory_handler", "stats_handler"] @@ -536,6 +538,9 @@ class HomeServer(object): def build_storage(self) -> Storage: return Storage(self, self.datastores) + def build_replication_streamer(self) -> ReplicationStreamer: + return ReplicationStreamer(self) + def remove_pusher(self, app_id, push_key, user_id): return self.get_pusherpool().remove_pusher(app_id, push_key, user_id) diff --git a/synapse/storage/data_stores/main/cache.py b/synapse/storage/data_stores/main/cache.py index d4c44dcc75..4dc5da3fe8 100644 --- a/synapse/storage/data_stores/main/cache.py +++ b/synapse/storage/data_stores/main/cache.py @@ -32,7 +32,29 @@ logger = logging.getLogger(__name__) CURRENT_STATE_CACHE_NAME = "cs_cache_fake" -class CacheInvalidationStore(SQLBaseStore): +class CacheInvalidationWorkerStore(SQLBaseStore): + def get_all_updated_caches(self, last_id, current_id, limit): + if last_id == current_id: + return defer.succeed([]) + + def get_all_updated_caches_txn(txn): + # We purposefully don't bound by the current token, as we want to + # send across cache invalidations as quickly as possible. Cache + # invalidations are idempotent, so duplicates are fine. + sql = ( + "SELECT stream_id, cache_func, keys, invalidation_ts" + " FROM cache_invalidation_stream" + " WHERE stream_id > ? ORDER BY stream_id ASC LIMIT ?" + ) + txn.execute(sql, (last_id, limit)) + return txn.fetchall() + + return self.db.runInteraction( + "get_all_updated_caches", get_all_updated_caches_txn + ) + + +class CacheInvalidationStore(CacheInvalidationWorkerStore): async def invalidate_cache_and_stream(self, cache_name: str, keys: Tuple[Any, ...]): """Invalidates the cache and adds it to the cache stream so slaves will know to invalidate their caches. @@ -145,26 +167,6 @@ class CacheInvalidationStore(SQLBaseStore): }, ) - def get_all_updated_caches(self, last_id, current_id, limit): - if last_id == current_id: - return defer.succeed([]) - - def get_all_updated_caches_txn(txn): - # We purposefully don't bound by the current token, as we want to - # send across cache invalidations as quickly as possible. Cache - # invalidations are idempotent, so duplicates are fine. - sql = ( - "SELECT stream_id, cache_func, keys, invalidation_ts" - " FROM cache_invalidation_stream" - " WHERE stream_id > ? ORDER BY stream_id ASC LIMIT ?" - ) - txn.execute(sql, (last_id, limit)) - return txn.fetchall() - - return self.db.runInteraction( - "get_all_updated_caches", get_all_updated_caches_txn - ) - def get_cache_stream_token(self): if self._cache_id_gen: return self._cache_id_gen.get_current_token() diff --git a/synapse/storage/data_stores/main/deviceinbox.py b/synapse/storage/data_stores/main/deviceinbox.py index 0613b49f4a..9a1178fb39 100644 --- a/synapse/storage/data_stores/main/deviceinbox.py +++ b/synapse/storage/data_stores/main/deviceinbox.py @@ -207,6 +207,50 @@ class DeviceInboxWorkerStore(SQLBaseStore): "delete_device_msgs_for_remote", delete_messages_for_remote_destination_txn ) + def get_all_new_device_messages(self, last_pos, current_pos, limit): + """ + Args: + last_pos(int): + current_pos(int): + limit(int): + Returns: + A deferred list of rows from the device inbox + """ + if last_pos == current_pos: + return defer.succeed([]) + + def get_all_new_device_messages_txn(txn): + # We limit like this as we might have multiple rows per stream_id, and + # we want to make sure we always get all entries for any stream_id + # we return. + upper_pos = min(current_pos, last_pos + limit) + sql = ( + "SELECT max(stream_id), user_id" + " FROM device_inbox" + " WHERE ? < stream_id AND stream_id <= ?" + " GROUP BY user_id" + ) + txn.execute(sql, (last_pos, upper_pos)) + rows = txn.fetchall() + + sql = ( + "SELECT max(stream_id), destination" + " FROM device_federation_outbox" + " WHERE ? < stream_id AND stream_id <= ?" + " GROUP BY destination" + ) + txn.execute(sql, (last_pos, upper_pos)) + rows.extend(txn) + + # Order by ascending stream ordering + rows.sort() + + return rows + + return self.db.runInteraction( + "get_all_new_device_messages", get_all_new_device_messages_txn + ) + class DeviceInboxBackgroundUpdateStore(SQLBaseStore): DEVICE_INBOX_STREAM_ID = "device_inbox_stream_drop" @@ -411,47 +455,3 @@ class DeviceInboxStore(DeviceInboxWorkerStore, DeviceInboxBackgroundUpdateStore) rows.append((user_id, device_id, stream_id, message_json)) txn.executemany(sql, rows) - - def get_all_new_device_messages(self, last_pos, current_pos, limit): - """ - Args: - last_pos(int): - current_pos(int): - limit(int): - Returns: - A deferred list of rows from the device inbox - """ - if last_pos == current_pos: - return defer.succeed([]) - - def get_all_new_device_messages_txn(txn): - # We limit like this as we might have multiple rows per stream_id, and - # we want to make sure we always get all entries for any stream_id - # we return. - upper_pos = min(current_pos, last_pos + limit) - sql = ( - "SELECT max(stream_id), user_id" - " FROM device_inbox" - " WHERE ? < stream_id AND stream_id <= ?" - " GROUP BY user_id" - ) - txn.execute(sql, (last_pos, upper_pos)) - rows = txn.fetchall() - - sql = ( - "SELECT max(stream_id), destination" - " FROM device_federation_outbox" - " WHERE ? < stream_id AND stream_id <= ?" - " GROUP BY destination" - ) - txn.execute(sql, (last_pos, upper_pos)) - rows.extend(txn) - - # Order by ascending stream ordering - rows.sort() - - return rows - - return self.db.runInteraction( - "get_all_new_device_messages", get_all_new_device_messages_txn - ) diff --git a/synapse/storage/data_stores/main/events.py b/synapse/storage/data_stores/main/events.py index d593ef47b8..e71c23541d 100644 --- a/synapse/storage/data_stores/main/events.py +++ b/synapse/storage/data_stores/main/events.py @@ -1267,104 +1267,6 @@ class EventsStore( ret = yield self.db.runInteraction("count_daily_active_rooms", _count) return ret - def get_current_backfill_token(self): - """The current minimum token that backfilled events have reached""" - return -self._backfill_id_gen.get_current_token() - - def get_current_events_token(self): - """The current maximum token that events have reached""" - return self._stream_id_gen.get_current_token() - - def get_all_new_forward_event_rows(self, last_id, current_id, limit): - if last_id == current_id: - return defer.succeed([]) - - def get_all_new_forward_event_rows(txn): - sql = ( - "SELECT e.stream_ordering, e.event_id, e.room_id, e.type," - " state_key, redacts, relates_to_id" - " FROM events AS e" - " LEFT JOIN redactions USING (event_id)" - " LEFT JOIN state_events USING (event_id)" - " LEFT JOIN event_relations USING (event_id)" - " WHERE ? < stream_ordering AND stream_ordering <= ?" - " ORDER BY stream_ordering ASC" - " LIMIT ?" - ) - txn.execute(sql, (last_id, current_id, limit)) - new_event_updates = txn.fetchall() - - if len(new_event_updates) == limit: - upper_bound = new_event_updates[-1][0] - else: - upper_bound = current_id - - sql = ( - "SELECT event_stream_ordering, e.event_id, e.room_id, e.type," - " state_key, redacts, relates_to_id" - " FROM events AS e" - " INNER JOIN ex_outlier_stream USING (event_id)" - " LEFT JOIN redactions USING (event_id)" - " LEFT JOIN state_events USING (event_id)" - " LEFT JOIN event_relations USING (event_id)" - " WHERE ? < event_stream_ordering" - " AND event_stream_ordering <= ?" - " ORDER BY event_stream_ordering DESC" - ) - txn.execute(sql, (last_id, upper_bound)) - new_event_updates.extend(txn) - - return new_event_updates - - return self.db.runInteraction( - "get_all_new_forward_event_rows", get_all_new_forward_event_rows - ) - - def get_all_new_backfill_event_rows(self, last_id, current_id, limit): - if last_id == current_id: - return defer.succeed([]) - - def get_all_new_backfill_event_rows(txn): - sql = ( - "SELECT -e.stream_ordering, e.event_id, e.room_id, e.type," - " state_key, redacts, relates_to_id" - " FROM events AS e" - " LEFT JOIN redactions USING (event_id)" - " LEFT JOIN state_events USING (event_id)" - " LEFT JOIN event_relations USING (event_id)" - " WHERE ? > stream_ordering AND stream_ordering >= ?" - " ORDER BY stream_ordering ASC" - " LIMIT ?" - ) - txn.execute(sql, (-last_id, -current_id, limit)) - new_event_updates = txn.fetchall() - - if len(new_event_updates) == limit: - upper_bound = new_event_updates[-1][0] - else: - upper_bound = current_id - - sql = ( - "SELECT -event_stream_ordering, e.event_id, e.room_id, e.type," - " state_key, redacts, relates_to_id" - " FROM events AS e" - " INNER JOIN ex_outlier_stream USING (event_id)" - " LEFT JOIN redactions USING (event_id)" - " LEFT JOIN state_events USING (event_id)" - " LEFT JOIN event_relations USING (event_id)" - " WHERE ? > event_stream_ordering" - " AND event_stream_ordering >= ?" - " ORDER BY event_stream_ordering DESC" - ) - txn.execute(sql, (-last_id, -upper_bound)) - new_event_updates.extend(txn.fetchall()) - - return new_event_updates - - return self.db.runInteraction( - "get_all_new_backfill_event_rows", get_all_new_backfill_event_rows - ) - @cached(num_args=5, max_entries=10) def get_all_new_events( self, @@ -1850,22 +1752,6 @@ class EventsStore( return (int(res["topological_ordering"]), int(res["stream_ordering"])) - def get_all_updated_current_state_deltas(self, from_token, to_token, limit): - def get_all_updated_current_state_deltas_txn(txn): - sql = """ - SELECT stream_id, room_id, type, state_key, event_id - FROM current_state_delta_stream - WHERE ? < stream_id AND stream_id <= ? - ORDER BY stream_id ASC LIMIT ? - """ - txn.execute(sql, (from_token, to_token, limit)) - return txn.fetchall() - - return self.db.runInteraction( - "get_all_updated_current_state_deltas", - get_all_updated_current_state_deltas_txn, - ) - def insert_labels_for_event_txn( self, txn, event_id, labels, room_id, topological_ordering ): diff --git a/synapse/storage/data_stores/main/events_worker.py b/synapse/storage/data_stores/main/events_worker.py index 3013f49d32..16ea8948b1 100644 --- a/synapse/storage/data_stores/main/events_worker.py +++ b/synapse/storage/data_stores/main/events_worker.py @@ -963,3 +963,117 @@ class EventsWorkerStore(SQLBaseStore): complexity_v1 = round(state_events / 500, 2) return {"v1": complexity_v1} + + def get_current_backfill_token(self): + """The current minimum token that backfilled events have reached""" + return -self._backfill_id_gen.get_current_token() + + def get_current_events_token(self): + """The current maximum token that events have reached""" + return self._stream_id_gen.get_current_token() + + def get_all_new_forward_event_rows(self, last_id, current_id, limit): + if last_id == current_id: + return defer.succeed([]) + + def get_all_new_forward_event_rows(txn): + sql = ( + "SELECT e.stream_ordering, e.event_id, e.room_id, e.type," + " state_key, redacts, relates_to_id" + " FROM events AS e" + " LEFT JOIN redactions USING (event_id)" + " LEFT JOIN state_events USING (event_id)" + " LEFT JOIN event_relations USING (event_id)" + " WHERE ? < stream_ordering AND stream_ordering <= ?" + " ORDER BY stream_ordering ASC" + " LIMIT ?" + ) + txn.execute(sql, (last_id, current_id, limit)) + new_event_updates = txn.fetchall() + + if len(new_event_updates) == limit: + upper_bound = new_event_updates[-1][0] + else: + upper_bound = current_id + + sql = ( + "SELECT event_stream_ordering, e.event_id, e.room_id, e.type," + " state_key, redacts, relates_to_id" + " FROM events AS e" + " INNER JOIN ex_outlier_stream USING (event_id)" + " LEFT JOIN redactions USING (event_id)" + " LEFT JOIN state_events USING (event_id)" + " LEFT JOIN event_relations USING (event_id)" + " WHERE ? < event_stream_ordering" + " AND event_stream_ordering <= ?" + " ORDER BY event_stream_ordering DESC" + ) + txn.execute(sql, (last_id, upper_bound)) + new_event_updates.extend(txn) + + return new_event_updates + + return self.db.runInteraction( + "get_all_new_forward_event_rows", get_all_new_forward_event_rows + ) + + def get_all_new_backfill_event_rows(self, last_id, current_id, limit): + if last_id == current_id: + return defer.succeed([]) + + def get_all_new_backfill_event_rows(txn): + sql = ( + "SELECT -e.stream_ordering, e.event_id, e.room_id, e.type," + " state_key, redacts, relates_to_id" + " FROM events AS e" + " LEFT JOIN redactions USING (event_id)" + " LEFT JOIN state_events USING (event_id)" + " LEFT JOIN event_relations USING (event_id)" + " WHERE ? > stream_ordering AND stream_ordering >= ?" + " ORDER BY stream_ordering ASC" + " LIMIT ?" + ) + txn.execute(sql, (-last_id, -current_id, limit)) + new_event_updates = txn.fetchall() + + if len(new_event_updates) == limit: + upper_bound = new_event_updates[-1][0] + else: + upper_bound = current_id + + sql = ( + "SELECT -event_stream_ordering, e.event_id, e.room_id, e.type," + " state_key, redacts, relates_to_id" + " FROM events AS e" + " INNER JOIN ex_outlier_stream USING (event_id)" + " LEFT JOIN redactions USING (event_id)" + " LEFT JOIN state_events USING (event_id)" + " LEFT JOIN event_relations USING (event_id)" + " WHERE ? > event_stream_ordering" + " AND event_stream_ordering >= ?" + " ORDER BY event_stream_ordering DESC" + ) + txn.execute(sql, (-last_id, -upper_bound)) + new_event_updates.extend(txn.fetchall()) + + return new_event_updates + + return self.db.runInteraction( + "get_all_new_backfill_event_rows", get_all_new_backfill_event_rows + ) + + def get_all_updated_current_state_deltas(self, from_token, to_token, limit): + def get_all_updated_current_state_deltas_txn(txn): + sql = """ + SELECT stream_id, room_id, type, state_key, event_id + FROM current_state_delta_stream + WHERE ? < stream_id AND stream_id <= ? + ORDER BY stream_id ASC LIMIT ? + """ + txn.execute(sql, (from_token, to_token, limit)) + return txn.fetchall() + + return self.db.runInteraction( + "get_all_updated_current_state_deltas", + get_all_updated_current_state_deltas_txn, + ) diff --git a/synapse/storage/data_stores/main/room.py b/synapse/storage/data_stores/main/room.py index e6c10c6316..aaebe427d3 100644 --- a/synapse/storage/data_stores/main/room.py +++ b/synapse/storage/data_stores/main/room.py @@ -732,6 +732,26 @@ class RoomWorkerStore(SQLBaseStore): return total_media_quarantined + def get_all_new_public_rooms(self, prev_id, current_id, limit): + def get_all_new_public_rooms(txn): + sql = """ + SELECT stream_id, room_id, visibility, appservice_id, network_id + FROM public_room_list_stream + WHERE stream_id > ? AND stream_id <= ? + ORDER BY stream_id ASC + LIMIT ? + """ + + txn.execute(sql, (prev_id, current_id, limit)) + return txn.fetchall() + + if prev_id == current_id: + return defer.succeed([]) + + return self.db.runInteraction( + "get_all_new_public_rooms", get_all_new_public_rooms + ) + class RoomBackgroundUpdateStore(SQLBaseStore): REMOVE_TOMESTONED_ROOMS_BG_UPDATE = "remove_tombstoned_rooms_from_directory" @@ -1249,26 +1269,6 @@ class RoomStore(RoomBackgroundUpdateStore, RoomWorkerStore, SearchStore): def get_current_public_room_stream_id(self): return self._public_room_id_gen.get_current_token() - def get_all_new_public_rooms(self, prev_id, current_id, limit): - def get_all_new_public_rooms(txn): - sql = """ - SELECT stream_id, room_id, visibility, appservice_id, network_id - FROM public_room_list_stream - WHERE stream_id > ? AND stream_id <= ? - ORDER BY stream_id ASC - LIMIT ? - """ - - txn.execute(sql, (prev_id, current_id, limit)) - return txn.fetchall() - - if prev_id == current_id: - return defer.succeed([]) - - return self.db.runInteraction( - "get_all_new_public_rooms", get_all_new_public_rooms - ) - @defer.inlineCallbacks def block_room(self, room_id, user_id): """Marks the room as blocked. Can be called multiple times. diff --git a/tests/replication/tcp/streams/_base.py b/tests/replication/tcp/streams/_base.py index e96ad4ca4e..a755fe2879 100644 --- a/tests/replication/tcp/streams/_base.py +++ b/tests/replication/tcp/streams/_base.py @@ -12,6 +12,7 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. + from mock import Mock from synapse.replication.tcp.commands import ReplicateCommand @@ -29,19 +30,37 @@ class BaseStreamTestCase(unittest.HomeserverTestCase): # build a replication server server_factory = ReplicationStreamProtocolFactory(self.hs) self.streamer = server_factory.streamer - server = server_factory.buildProtocol(None) + self.server = server_factory.buildProtocol(None) - # build a replication client, with a dummy handler - handler_factory = Mock() - self.test_handler = TestReplicationClientHandler() - self.test_handler.factory = handler_factory + self.test_handler = Mock(wraps=TestReplicationClientHandler()) self.client = ClientReplicationStreamProtocol( - "client", "test", clock, self.test_handler + hs, "client", "test", clock, self.test_handler, ) - # wire them together - self.client.makeConnection(FakeTransport(server, reactor)) - server.makeConnection(FakeTransport(self.client, reactor)) + self._client_transport = None + self._server_transport = None + + def reconnect(self): + if self._client_transport: + self.client.close() + + if self._server_transport: + self.server.close() + + self._client_transport = FakeTransport(self.server, self.reactor) + self.client.makeConnection(self._client_transport) + + self._server_transport = FakeTransport(self.client, self.reactor) + self.server.makeConnection(self._server_transport) + + def disconnect(self): + if self._client_transport: + self._client_transport = None + self.client.close() + + if self._server_transport: + self._server_transport = None + self.server.close() def replicate(self): """Tell the master side of replication that something has happened, and then @@ -50,19 +69,24 @@ class BaseStreamTestCase(unittest.HomeserverTestCase): self.streamer.on_notifier_poke() self.pump(0.1) - def replicate_stream(self, stream, token="NOW"): + def replicate_stream(self): """Make the client end a REPLICATE command to set up a subscription to a stream""" - self.client.send_command(ReplicateCommand(stream, token)) + self.client.send_command(ReplicateCommand()) class TestReplicationClientHandler(object): """Drop-in for ReplicationClientHandler which just collects RDATA rows""" def __init__(self): - self.received_rdata_rows = [] + self.streams = set() + self._received_rdata_rows = [] def get_streams_to_replicate(self): - return {} + positions = {s: 0 for s in self.streams} + for stream, token, _ in self._received_rdata_rows: + if stream in self.streams: + positions[stream] = max(token, positions.get(stream, 0)) + return positions def get_currently_syncing_users(self): return [] @@ -73,6 +97,9 @@ class TestReplicationClientHandler(object): def finished_connecting(self): pass + async def on_position(self, stream_name, token): + """Called when we get new position data.""" + async def on_rdata(self, stream_name, token, rows): for r in rows: - self.received_rdata_rows.append((stream_name, token, r)) + self._received_rdata_rows.append((stream_name, token, r)) diff --git a/tests/replication/tcp/streams/test_receipts.py b/tests/replication/tcp/streams/test_receipts.py index fa2493cad6..0ec0825a0e 100644 --- a/tests/replication/tcp/streams/test_receipts.py +++ b/tests/replication/tcp/streams/test_receipts.py @@ -17,30 +17,64 @@ from synapse.replication.tcp.streams._base import ReceiptsStream from tests.replication.tcp.streams._base import BaseStreamTestCase USER_ID = "@feeling:blue" -ROOM_ID = "!room:blue" -EVENT_ID = "$event:blue" class ReceiptsStreamTestCase(BaseStreamTestCase): def test_receipt(self): + self.reconnect() + # make the client subscribe to the receipts stream - self.replicate_stream("receipts", "NOW") + self.replicate_stream() + self.test_handler.streams.add("receipts") # tell the master to send a new receipt self.get_success( self.hs.get_datastore().insert_receipt( - ROOM_ID, "m.read", USER_ID, [EVENT_ID], {"a": 1} + "!room:blue", "m.read", USER_ID, ["$event:blue"], {"a": 1} ) ) self.replicate() # there should be one RDATA command - rdata_rows = self.test_handler.received_rdata_rows + self.test_handler.on_rdata.assert_called_once() + stream_name, token, rdata_rows = self.test_handler.on_rdata.call_args[0] + self.assertEqual(stream_name, "receipts") self.assertEqual(1, len(rdata_rows)) - self.assertEqual(rdata_rows[0][0], "receipts") - row = rdata_rows[0][2] # type: ReceiptsStream.ReceiptsStreamRow - self.assertEqual(ROOM_ID, row.room_id) + row = rdata_rows[0] # type: ReceiptsStream.ReceiptsStreamRow + self.assertEqual("!room:blue", row.room_id) self.assertEqual("m.read", row.receipt_type) self.assertEqual(USER_ID, row.user_id) - self.assertEqual(EVENT_ID, row.event_id) + self.assertEqual("$event:blue", row.event_id) self.assertEqual({"a": 1}, row.data) + + # Now let's disconnect and insert some data. + self.disconnect() + + self.test_handler.on_rdata.reset_mock() + + self.get_success( + self.hs.get_datastore().insert_receipt( + "!room2:blue", "m.read", USER_ID, ["$event2:foo"], {"a": 2} + ) + ) + self.replicate() + + # Nothing should have happened as we are disconnected + self.test_handler.on_rdata.assert_not_called() + + self.reconnect() + self.pump(0.1) + + # We should now have caught up and get the missing data + self.test_handler.on_rdata.assert_called_once() + stream_name, token, rdata_rows = self.test_handler.on_rdata.call_args[0] + self.assertEqual(stream_name, "receipts") + self.assertEqual(token, 3) + self.assertEqual(1, len(rdata_rows)) + + row = rdata_rows[0] # type: ReceiptsStream.ReceiptsStreamRow + self.assertEqual("!room2:blue", row.room_id) + self.assertEqual("m.read", row.receipt_type) + self.assertEqual(USER_ID, row.user_id) + self.assertEqual("$event2:foo", row.event_id) + self.assertEqual({"a": 2}, row.data) -- cgit 1.4.1 From 1c1242acba9694a3a4b1eb3b14ec0bac11ee4ff8 Mon Sep 17 00:00:00 2001 From: Patrick Cloke Date: Thu, 26 Mar 2020 07:39:34 -0400 Subject: Validate that the session is not modified during UI-Auth (#7068) --- changelog.d/7068.bugfix | 1 + synapse/handlers/auth.py | 37 +++++++++++++++-- synapse/rest/client/v2_alpha/account.py | 11 ++++-- synapse/rest/client/v2_alpha/devices.py | 4 +- synapse/rest/client/v2_alpha/keys.py | 2 +- synapse/rest/client/v2_alpha/register.py | 5 ++- tests/rest/client/v2_alpha/test_auth.py | 68 +++++++++++++++++++++++++++++++- tests/test_terms_auth.py | 3 +- 8 files changed, 117 insertions(+), 14 deletions(-) create mode 100644 changelog.d/7068.bugfix (limited to 'tests') diff --git a/changelog.d/7068.bugfix b/changelog.d/7068.bugfix new file mode 100644 index 0000000000..d1693a7f22 --- /dev/null +++ b/changelog.d/7068.bugfix @@ -0,0 +1 @@ +Ensure that a user inteactive authentication session is tied to a single request. diff --git a/synapse/handlers/auth.py b/synapse/handlers/auth.py index 7860f9625e..2ce1425dfa 100644 --- a/synapse/handlers/auth.py +++ b/synapse/handlers/auth.py @@ -125,7 +125,11 @@ class AuthHandler(BaseHandler): @defer.inlineCallbacks def validate_user_via_ui_auth( - self, requester: Requester, request_body: Dict[str, Any], clientip: str + self, + requester: Requester, + request: SynapseRequest, + request_body: Dict[str, Any], + clientip: str, ): """ Checks that the user is who they claim to be, via a UI auth. @@ -137,6 +141,8 @@ class AuthHandler(BaseHandler): Args: requester: The user, as given by the access token + request: The request sent by the client. + request_body: The body of the request sent by the client clientip: The IP address of the client. @@ -172,7 +178,9 @@ class AuthHandler(BaseHandler): flows = [[login_type] for login_type in self._supported_login_types] try: - result, params, _ = yield self.check_auth(flows, request_body, clientip) + result, params, _ = yield self.check_auth( + flows, request, request_body, clientip + ) except LoginError: # Update the ratelimite to say we failed (`can_do_action` doesn't raise). self._failed_uia_attempts_ratelimiter.can_do_action( @@ -211,7 +219,11 @@ class AuthHandler(BaseHandler): @defer.inlineCallbacks def check_auth( - self, flows: List[List[str]], clientdict: Dict[str, Any], clientip: str + self, + flows: List[List[str]], + request: SynapseRequest, + clientdict: Dict[str, Any], + clientip: str, ): """ Takes a dictionary sent by the client in the login / registration @@ -231,6 +243,8 @@ class AuthHandler(BaseHandler): strings representing auth-types. At least one full flow must be completed in order for auth to be successful. + request: The request sent by the client. + clientdict: The dictionary from the client root level, not the 'auth' key: this method prompts for auth if none is sent. @@ -270,13 +284,27 @@ class AuthHandler(BaseHandler): # email auth link on there). It's probably too open to abuse # because it lets unauthenticated clients store arbitrary objects # on a homeserver. - # Revisit: Assumimg the REST APIs do sensible validation, the data + # Revisit: Assuming the REST APIs do sensible validation, the data # isn't arbintrary. session["clientdict"] = clientdict self._save_session(session) elif "clientdict" in session: clientdict = session["clientdict"] + # Ensure that the queried operation does not vary between stages of + # the UI authentication session. This is done by generating a stable + # comparator based on the URI, method, and body (minus the auth dict) + # and storing it during the initial query. Subsequent queries ensure + # that this comparator has not changed. + comparator = (request.uri, request.method, clientdict) + if "ui_auth" not in session: + session["ui_auth"] = comparator + elif session["ui_auth"] != comparator: + raise SynapseError( + 403, + "Requested operation has changed during the UI authentication session.", + ) + if not authdict: raise InteractiveAuthIncompleteError( self._auth_dict_for_flows(flows, session) @@ -322,6 +350,7 @@ class AuthHandler(BaseHandler): creds, list(clientdict), ) + return creds, clientdict, session["id"] ret = self._auth_dict_for_flows(flows, session) diff --git a/synapse/rest/client/v2_alpha/account.py b/synapse/rest/client/v2_alpha/account.py index 631cc74cb4..b1249b664c 100644 --- a/synapse/rest/client/v2_alpha/account.py +++ b/synapse/rest/client/v2_alpha/account.py @@ -234,13 +234,16 @@ class PasswordRestServlet(RestServlet): if self.auth.has_access_token(request): requester = await self.auth.get_user_by_req(request) params = await self.auth_handler.validate_user_via_ui_auth( - requester, body, self.hs.get_ip_from_request(request) + requester, request, body, self.hs.get_ip_from_request(request), ) user_id = requester.user.to_string() else: requester = None result, params, _ = await self.auth_handler.check_auth( - [[LoginType.EMAIL_IDENTITY]], body, self.hs.get_ip_from_request(request) + [[LoginType.EMAIL_IDENTITY]], + request, + body, + self.hs.get_ip_from_request(request), ) if LoginType.EMAIL_IDENTITY in result: @@ -308,7 +311,7 @@ class DeactivateAccountRestServlet(RestServlet): return 200, {} await self.auth_handler.validate_user_via_ui_auth( - requester, body, self.hs.get_ip_from_request(request) + requester, request, body, self.hs.get_ip_from_request(request), ) result = await self._deactivate_account_handler.deactivate_account( requester.user.to_string(), erase, id_server=body.get("id_server") @@ -656,7 +659,7 @@ class ThreepidAddRestServlet(RestServlet): assert_valid_client_secret(client_secret) await self.auth_handler.validate_user_via_ui_auth( - requester, body, self.hs.get_ip_from_request(request) + requester, request, body, self.hs.get_ip_from_request(request), ) validation_session = await self.identity_handler.validate_threepid_session( diff --git a/synapse/rest/client/v2_alpha/devices.py b/synapse/rest/client/v2_alpha/devices.py index 94ff73f384..119d979052 100644 --- a/synapse/rest/client/v2_alpha/devices.py +++ b/synapse/rest/client/v2_alpha/devices.py @@ -81,7 +81,7 @@ class DeleteDevicesRestServlet(RestServlet): assert_params_in_dict(body, ["devices"]) await self.auth_handler.validate_user_via_ui_auth( - requester, body, self.hs.get_ip_from_request(request) + requester, request, body, self.hs.get_ip_from_request(request), ) await self.device_handler.delete_devices( @@ -127,7 +127,7 @@ class DeviceRestServlet(RestServlet): raise await self.auth_handler.validate_user_via_ui_auth( - requester, body, self.hs.get_ip_from_request(request) + requester, request, body, self.hs.get_ip_from_request(request), ) await self.device_handler.delete_device(requester.user.to_string(), device_id) diff --git a/synapse/rest/client/v2_alpha/keys.py b/synapse/rest/client/v2_alpha/keys.py index f7ed4daf90..5eb7ef35a4 100644 --- a/synapse/rest/client/v2_alpha/keys.py +++ b/synapse/rest/client/v2_alpha/keys.py @@ -263,7 +263,7 @@ class SigningKeyUploadServlet(RestServlet): body = parse_json_object_from_request(request) await self.auth_handler.validate_user_via_ui_auth( - requester, body, self.hs.get_ip_from_request(request) + requester, request, body, self.hs.get_ip_from_request(request), ) result = await self.e2e_keys_handler.upload_signing_keys_for_user(user_id, body) diff --git a/synapse/rest/client/v2_alpha/register.py b/synapse/rest/client/v2_alpha/register.py index a09189b1b4..6963d79310 100644 --- a/synapse/rest/client/v2_alpha/register.py +++ b/synapse/rest/client/v2_alpha/register.py @@ -499,7 +499,10 @@ class RegisterRestServlet(RestServlet): ) auth_result, params, session_id = await self.auth_handler.check_auth( - self._registration_flows, body, self.hs.get_ip_from_request(request) + self._registration_flows, + request, + body, + self.hs.get_ip_from_request(request), ) # Check that we're not trying to register a denied 3pid. diff --git a/tests/rest/client/v2_alpha/test_auth.py b/tests/rest/client/v2_alpha/test_auth.py index b6df1396ad..624bf5ada2 100644 --- a/tests/rest/client/v2_alpha/test_auth.py +++ b/tests/rest/client/v2_alpha/test_auth.py @@ -104,7 +104,7 @@ class FallbackAuthTests(unittest.HomeserverTestCase): ) self.render(request) - # Now we should have fufilled a complete auth flow, including + # Now we should have fulfilled a complete auth flow, including # the recaptcha fallback step, we can then send a # request to the register API with the session in the authdict. request, channel = self.make_request( @@ -115,3 +115,69 @@ class FallbackAuthTests(unittest.HomeserverTestCase): # We're given a registered user. self.assertEqual(channel.json_body["user_id"], "@user:test") + + def test_cannot_change_operation(self): + """ + The initial requested operation cannot be modified during the user interactive authentication session. + """ + + # Make the initial request to register. (Later on a different password + # will be used.) + request, channel = self.make_request( + "POST", + "register", + {"username": "user", "type": "m.login.password", "password": "bar"}, + ) + self.render(request) + + # Returns a 401 as per the spec + self.assertEqual(request.code, 401) + # Grab the session + session = channel.json_body["session"] + # Assert our configured public key is being given + self.assertEqual( + channel.json_body["params"]["m.login.recaptcha"]["public_key"], "brokencake" + ) + + request, channel = self.make_request( + "GET", "auth/m.login.recaptcha/fallback/web?session=" + session + ) + self.render(request) + self.assertEqual(request.code, 200) + + request, channel = self.make_request( + "POST", + "auth/m.login.recaptcha/fallback/web?session=" + + session + + "&g-recaptcha-response=a", + ) + self.render(request) + self.assertEqual(request.code, 200) + + # The recaptcha handler is called with the response given + attempts = self.recaptcha_checker.recaptcha_attempts + self.assertEqual(len(attempts), 1) + self.assertEqual(attempts[0][0]["response"], "a") + + # also complete the dummy auth + request, channel = self.make_request( + "POST", "register", {"auth": {"session": session, "type": "m.login.dummy"}} + ) + self.render(request) + + # Now we should have fulfilled a complete auth flow, including + # the recaptcha fallback step. Make the initial request again, but + # with a different password. This causes the request to fail since the + # operaiton was modified during the ui auth session. + request, channel = self.make_request( + "POST", + "register", + { + "username": "user", + "type": "m.login.password", + "password": "foo", # Note this doesn't match the original request. + "auth": {"session": session}, + }, + ) + self.render(request) + self.assertEqual(channel.code, 403) diff --git a/tests/test_terms_auth.py b/tests/test_terms_auth.py index 5ec5d2b358..a3f98a1412 100644 --- a/tests/test_terms_auth.py +++ b/tests/test_terms_auth.py @@ -53,7 +53,8 @@ class TermsTestCase(unittest.HomeserverTestCase): def test_ui_auth(self): # Do a UI auth request - request, channel = self.make_request(b"POST", self.url, b"{}") + request_data = json.dumps({"username": "kermit", "password": "monkey"}) + request, channel = self.make_request(b"POST", self.url, request_data) self.render(request) self.assertEquals(channel.result["code"], b"401", channel.result) -- cgit 1.4.1 From e8e2ddb60ae11db488f159901d918cb159695912 Mon Sep 17 00:00:00 2001 From: Dirk Klimpel <5740567+dklimpel@users.noreply.github.com> Date: Thu, 26 Mar 2020 17:51:13 +0100 Subject: Allow server admins to define and enforce a password policy (MSC2000). (#7118) --- changelog.d/7118.feature | 1 + docs/sample_config.yaml | 35 ++++ synapse/api/errors.py | 21 +++ synapse/config/password.py | 39 +++++ synapse/handlers/password_policy.py | 93 +++++++++++ synapse/handlers/set_password.py | 2 + synapse/rest/__init__.py | 2 + synapse/rest/client/v2_alpha/password_policy.py | 58 +++++++ synapse/rest/client/v2_alpha/register.py | 2 + synapse/server.py | 5 + tests/rest/client/v2_alpha/test_password_policy.py | 179 +++++++++++++++++++++ 11 files changed, 437 insertions(+) create mode 100644 changelog.d/7118.feature create mode 100644 synapse/handlers/password_policy.py create mode 100644 synapse/rest/client/v2_alpha/password_policy.py create mode 100644 tests/rest/client/v2_alpha/test_password_policy.py (limited to 'tests') diff --git a/changelog.d/7118.feature b/changelog.d/7118.feature new file mode 100644 index 0000000000..5cbfd98160 --- /dev/null +++ b/changelog.d/7118.feature @@ -0,0 +1 @@ +Allow server admins to define and enforce a password policy (MSC2000). \ No newline at end of file diff --git a/docs/sample_config.yaml b/docs/sample_config.yaml index 2ef83646b3..1a1d061759 100644 --- a/docs/sample_config.yaml +++ b/docs/sample_config.yaml @@ -1482,6 +1482,41 @@ password_config: # #pepper: "EVEN_MORE_SECRET" + # Define and enforce a password policy. Each parameter is optional. + # This is an implementation of MSC2000. + # + policy: + # Whether to enforce the password policy. + # Defaults to 'false'. + # + #enabled: true + + # Minimum accepted length for a password. + # Defaults to 0. + # + #minimum_length: 15 + + # Whether a password must contain at least one digit. + # Defaults to 'false'. + # + #require_digit: true + + # Whether a password must contain at least one symbol. + # A symbol is any character that's not a number or a letter. + # Defaults to 'false'. + # + #require_symbol: true + + # Whether a password must contain at least one lowercase letter. + # Defaults to 'false'. + # + #require_lowercase: true + + # Whether a password must contain at least one lowercase letter. + # Defaults to 'false'. + # + #require_uppercase: true + # Configuration for sending emails from Synapse. # diff --git a/synapse/api/errors.py b/synapse/api/errors.py index 616942b057..11da016ac5 100644 --- a/synapse/api/errors.py +++ b/synapse/api/errors.py @@ -64,6 +64,13 @@ class Codes(object): INCOMPATIBLE_ROOM_VERSION = "M_INCOMPATIBLE_ROOM_VERSION" WRONG_ROOM_KEYS_VERSION = "M_WRONG_ROOM_KEYS_VERSION" EXPIRED_ACCOUNT = "ORG_MATRIX_EXPIRED_ACCOUNT" + PASSWORD_TOO_SHORT = "M_PASSWORD_TOO_SHORT" + PASSWORD_NO_DIGIT = "M_PASSWORD_NO_DIGIT" + PASSWORD_NO_UPPERCASE = "M_PASSWORD_NO_UPPERCASE" + PASSWORD_NO_LOWERCASE = "M_PASSWORD_NO_LOWERCASE" + PASSWORD_NO_SYMBOL = "M_PASSWORD_NO_SYMBOL" + PASSWORD_IN_DICTIONARY = "M_PASSWORD_IN_DICTIONARY" + WEAK_PASSWORD = "M_WEAK_PASSWORD" INVALID_SIGNATURE = "M_INVALID_SIGNATURE" USER_DEACTIVATED = "M_USER_DEACTIVATED" BAD_ALIAS = "M_BAD_ALIAS" @@ -439,6 +446,20 @@ class IncompatibleRoomVersionError(SynapseError): return cs_error(self.msg, self.errcode, room_version=self._room_version) +class PasswordRefusedError(SynapseError): + """A password has been refused, either during password reset/change or registration. + """ + + def __init__( + self, + msg="This password doesn't comply with the server's policy", + errcode=Codes.WEAK_PASSWORD, + ): + super(PasswordRefusedError, self).__init__( + code=400, msg=msg, errcode=errcode, + ) + + class RequestSendFailed(RuntimeError): """Sending a HTTP request over federation failed due to not being able to talk to the remote server for some reason. diff --git a/synapse/config/password.py b/synapse/config/password.py index 2a634ac751..9c0ea8c30a 100644 --- a/synapse/config/password.py +++ b/synapse/config/password.py @@ -31,6 +31,10 @@ class PasswordConfig(Config): self.password_localdb_enabled = password_config.get("localdb_enabled", True) self.password_pepper = password_config.get("pepper", "") + # Password policy + self.password_policy = password_config.get("policy") or {} + self.password_policy_enabled = self.password_policy.get("enabled", False) + def generate_config_section(self, config_dir_path, server_name, **kwargs): return """\ password_config: @@ -48,4 +52,39 @@ class PasswordConfig(Config): # DO NOT CHANGE THIS AFTER INITIAL SETUP! # #pepper: "EVEN_MORE_SECRET" + + # Define and enforce a password policy. Each parameter is optional. + # This is an implementation of MSC2000. + # + policy: + # Whether to enforce the password policy. + # Defaults to 'false'. + # + #enabled: true + + # Minimum accepted length for a password. + # Defaults to 0. + # + #minimum_length: 15 + + # Whether a password must contain at least one digit. + # Defaults to 'false'. + # + #require_digit: true + + # Whether a password must contain at least one symbol. + # A symbol is any character that's not a number or a letter. + # Defaults to 'false'. + # + #require_symbol: true + + # Whether a password must contain at least one lowercase letter. + # Defaults to 'false'. + # + #require_lowercase: true + + # Whether a password must contain at least one lowercase letter. + # Defaults to 'false'. + # + #require_uppercase: true """ diff --git a/synapse/handlers/password_policy.py b/synapse/handlers/password_policy.py new file mode 100644 index 0000000000..d06b110269 --- /dev/null +++ b/synapse/handlers/password_policy.py @@ -0,0 +1,93 @@ +# -*- coding: utf-8 -*- +# Copyright 2019 New Vector Ltd +# Copyright 2019 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging +import re + +from synapse.api.errors import Codes, PasswordRefusedError + +logger = logging.getLogger(__name__) + + +class PasswordPolicyHandler(object): + def __init__(self, hs): + self.policy = hs.config.password_policy + self.enabled = hs.config.password_policy_enabled + + # Regexps for the spec'd policy parameters. + self.regexp_digit = re.compile("[0-9]") + self.regexp_symbol = re.compile("[^a-zA-Z0-9]") + self.regexp_uppercase = re.compile("[A-Z]") + self.regexp_lowercase = re.compile("[a-z]") + + def validate_password(self, password): + """Checks whether a given password complies with the server's policy. + + Args: + password (str): The password to check against the server's policy. + + Raises: + PasswordRefusedError: The password doesn't comply with the server's policy. + """ + + if not self.enabled: + return + + minimum_accepted_length = self.policy.get("minimum_length", 0) + if len(password) < minimum_accepted_length: + raise PasswordRefusedError( + msg=( + "The password must be at least %d characters long" + % minimum_accepted_length + ), + errcode=Codes.PASSWORD_TOO_SHORT, + ) + + if ( + self.policy.get("require_digit", False) + and self.regexp_digit.search(password) is None + ): + raise PasswordRefusedError( + msg="The password must include at least one digit", + errcode=Codes.PASSWORD_NO_DIGIT, + ) + + if ( + self.policy.get("require_symbol", False) + and self.regexp_symbol.search(password) is None + ): + raise PasswordRefusedError( + msg="The password must include at least one symbol", + errcode=Codes.PASSWORD_NO_SYMBOL, + ) + + if ( + self.policy.get("require_uppercase", False) + and self.regexp_uppercase.search(password) is None + ): + raise PasswordRefusedError( + msg="The password must include at least one uppercase letter", + errcode=Codes.PASSWORD_NO_UPPERCASE, + ) + + if ( + self.policy.get("require_lowercase", False) + and self.regexp_lowercase.search(password) is None + ): + raise PasswordRefusedError( + msg="The password must include at least one lowercase letter", + errcode=Codes.PASSWORD_NO_LOWERCASE, + ) diff --git a/synapse/handlers/set_password.py b/synapse/handlers/set_password.py index 12657ca698..7d1263caf2 100644 --- a/synapse/handlers/set_password.py +++ b/synapse/handlers/set_password.py @@ -32,6 +32,7 @@ class SetPasswordHandler(BaseHandler): super(SetPasswordHandler, self).__init__(hs) self._auth_handler = hs.get_auth_handler() self._device_handler = hs.get_device_handler() + self._password_policy_handler = hs.get_password_policy_handler() @defer.inlineCallbacks def set_password( @@ -44,6 +45,7 @@ class SetPasswordHandler(BaseHandler): if not self.hs.config.password_localdb_enabled: raise SynapseError(403, "Password change disabled", errcode=Codes.FORBIDDEN) + self._password_policy_handler.validate_password(new_password) password_hash = yield self._auth_handler.hash(new_password) try: diff --git a/synapse/rest/__init__.py b/synapse/rest/__init__.py index 4a1fc2ec2b..46e458e95b 100644 --- a/synapse/rest/__init__.py +++ b/synapse/rest/__init__.py @@ -41,6 +41,7 @@ from synapse.rest.client.v2_alpha import ( keys, notifications, openid, + password_policy, read_marker, receipts, register, @@ -118,6 +119,7 @@ class ClientRestResource(JsonResource): capabilities.register_servlets(hs, client_resource) account_validity.register_servlets(hs, client_resource) relations.register_servlets(hs, client_resource) + password_policy.register_servlets(hs, client_resource) # moving to /_synapse/admin synapse.rest.admin.register_servlets_for_client_rest_resource( diff --git a/synapse/rest/client/v2_alpha/password_policy.py b/synapse/rest/client/v2_alpha/password_policy.py new file mode 100644 index 0000000000..968403cca4 --- /dev/null +++ b/synapse/rest/client/v2_alpha/password_policy.py @@ -0,0 +1,58 @@ +# -*- coding: utf-8 -*- +# Copyright 2019 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging + +from synapse.http.servlet import RestServlet + +from ._base import client_patterns + +logger = logging.getLogger(__name__) + + +class PasswordPolicyServlet(RestServlet): + PATTERNS = client_patterns("/password_policy$") + + def __init__(self, hs): + """ + Args: + hs (synapse.server.HomeServer): server + """ + super(PasswordPolicyServlet, self).__init__() + + self.policy = hs.config.password_policy + self.enabled = hs.config.password_policy_enabled + + def on_GET(self, request): + if not self.enabled or not self.policy: + return (200, {}) + + policy = {} + + for param in [ + "minimum_length", + "require_digit", + "require_symbol", + "require_lowercase", + "require_uppercase", + ]: + if param in self.policy: + policy["m.%s" % param] = self.policy[param] + + return (200, policy) + + +def register_servlets(hs, http_server): + PasswordPolicyServlet(hs).register(http_server) diff --git a/synapse/rest/client/v2_alpha/register.py b/synapse/rest/client/v2_alpha/register.py index 6963d79310..66fc8ec179 100644 --- a/synapse/rest/client/v2_alpha/register.py +++ b/synapse/rest/client/v2_alpha/register.py @@ -373,6 +373,7 @@ class RegisterRestServlet(RestServlet): self.room_member_handler = hs.get_room_member_handler() self.macaroon_gen = hs.get_macaroon_generator() self.ratelimiter = hs.get_registration_ratelimiter() + self.password_policy_handler = hs.get_password_policy_handler() self.clock = hs.get_clock() self._registration_flows = _calculate_registration_flows( @@ -420,6 +421,7 @@ class RegisterRestServlet(RestServlet): or len(body["password"]) > 512 ): raise SynapseError(400, "Invalid password") + self.password_policy_handler.validate_password(body["password"]) desired_username = None if "username" in body: diff --git a/synapse/server.py b/synapse/server.py index 9426eb1672..d0d80e8ac5 100644 --- a/synapse/server.py +++ b/synapse/server.py @@ -66,6 +66,7 @@ from synapse.handlers.groups_local import GroupsLocalHandler, GroupsLocalWorkerH from synapse.handlers.initial_sync import InitialSyncHandler from synapse.handlers.message import EventCreationHandler, MessageHandler from synapse.handlers.pagination import PaginationHandler +from synapse.handlers.password_policy import PasswordPolicyHandler from synapse.handlers.presence import PresenceHandler from synapse.handlers.profile import BaseProfileHandler, MasterProfileHandler from synapse.handlers.read_marker import ReadMarkerHandler @@ -199,6 +200,7 @@ class HomeServer(object): "account_validity_handler", "saml_handler", "event_client_serializer", + "password_policy_handler", "storage", "replication_streamer", ] @@ -535,6 +537,9 @@ class HomeServer(object): def build_event_client_serializer(self): return EventClientSerializer(self) + def build_password_policy_handler(self): + return PasswordPolicyHandler(self) + def build_storage(self) -> Storage: return Storage(self, self.datastores) diff --git a/tests/rest/client/v2_alpha/test_password_policy.py b/tests/rest/client/v2_alpha/test_password_policy.py new file mode 100644 index 0000000000..c57072f50c --- /dev/null +++ b/tests/rest/client/v2_alpha/test_password_policy.py @@ -0,0 +1,179 @@ +# -*- coding: utf-8 -*- +# Copyright 2019 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import json + +from synapse.api.constants import LoginType +from synapse.api.errors import Codes +from synapse.rest import admin +from synapse.rest.client.v1 import login +from synapse.rest.client.v2_alpha import account, password_policy, register + +from tests import unittest + + +class PasswordPolicyTestCase(unittest.HomeserverTestCase): + """Tests the password policy feature and its compliance with MSC2000. + + When validating a password, Synapse does the necessary checks in this order: + + 1. Password is long enough + 2. Password contains digit(s) + 3. Password contains symbol(s) + 4. Password contains uppercase letter(s) + 5. Password contains lowercase letter(s) + + For each test below that checks whether a password triggers the right error code, + that test provides a password good enough to pass the previous tests, but not the + one it is currently testing (nor any test that comes afterward). + """ + + servlets = [ + admin.register_servlets_for_client_rest_resource, + login.register_servlets, + register.register_servlets, + password_policy.register_servlets, + account.register_servlets, + ] + + def make_homeserver(self, reactor, clock): + self.register_url = "/_matrix/client/r0/register" + self.policy = { + "enabled": True, + "minimum_length": 10, + "require_digit": True, + "require_symbol": True, + "require_lowercase": True, + "require_uppercase": True, + } + + config = self.default_config() + config["password_config"] = { + "policy": self.policy, + } + + hs = self.setup_test_homeserver(config=config) + return hs + + def test_get_policy(self): + """Tests if the /password_policy endpoint returns the configured policy.""" + + request, channel = self.make_request( + "GET", "/_matrix/client/r0/password_policy" + ) + self.render(request) + + self.assertEqual(channel.code, 200, channel.result) + self.assertEqual( + channel.json_body, + { + "m.minimum_length": 10, + "m.require_digit": True, + "m.require_symbol": True, + "m.require_lowercase": True, + "m.require_uppercase": True, + }, + channel.result, + ) + + def test_password_too_short(self): + request_data = json.dumps({"username": "kermit", "password": "shorty"}) + request, channel = self.make_request("POST", self.register_url, request_data) + self.render(request) + + self.assertEqual(channel.code, 400, channel.result) + self.assertEqual( + channel.json_body["errcode"], Codes.PASSWORD_TOO_SHORT, channel.result, + ) + + def test_password_no_digit(self): + request_data = json.dumps({"username": "kermit", "password": "longerpassword"}) + request, channel = self.make_request("POST", self.register_url, request_data) + self.render(request) + + self.assertEqual(channel.code, 400, channel.result) + self.assertEqual( + channel.json_body["errcode"], Codes.PASSWORD_NO_DIGIT, channel.result, + ) + + def test_password_no_symbol(self): + request_data = json.dumps({"username": "kermit", "password": "l0ngerpassword"}) + request, channel = self.make_request("POST", self.register_url, request_data) + self.render(request) + + self.assertEqual(channel.code, 400, channel.result) + self.assertEqual( + channel.json_body["errcode"], Codes.PASSWORD_NO_SYMBOL, channel.result, + ) + + def test_password_no_uppercase(self): + request_data = json.dumps({"username": "kermit", "password": "l0ngerpassword!"}) + request, channel = self.make_request("POST", self.register_url, request_data) + self.render(request) + + self.assertEqual(channel.code, 400, channel.result) + self.assertEqual( + channel.json_body["errcode"], Codes.PASSWORD_NO_UPPERCASE, channel.result, + ) + + def test_password_no_lowercase(self): + request_data = json.dumps({"username": "kermit", "password": "L0NGERPASSWORD!"}) + request, channel = self.make_request("POST", self.register_url, request_data) + self.render(request) + + self.assertEqual(channel.code, 400, channel.result) + self.assertEqual( + channel.json_body["errcode"], Codes.PASSWORD_NO_LOWERCASE, channel.result, + ) + + def test_password_compliant(self): + request_data = json.dumps({"username": "kermit", "password": "L0ngerpassword!"}) + request, channel = self.make_request("POST", self.register_url, request_data) + self.render(request) + + # Getting a 401 here means the password has passed validation and the server has + # responded with a list of registration flows. + self.assertEqual(channel.code, 401, channel.result) + + def test_password_change(self): + """This doesn't test every possible use case, only that hitting /account/password + triggers the password validation code. + """ + compliant_password = "C0mpl!antpassword" + not_compliant_password = "notcompliantpassword" + + user_id = self.register_user("kermit", compliant_password) + tok = self.login("kermit", compliant_password) + + request_data = json.dumps( + { + "new_password": not_compliant_password, + "auth": { + "password": compliant_password, + "type": LoginType.PASSWORD, + "user": user_id, + }, + } + ) + request, channel = self.make_request( + "POST", + "/_matrix/client/r0/account/password", + request_data, + access_token=tok, + ) + self.render(request) + + self.assertEqual(channel.code, 400, channel.result) + self.assertEqual(channel.json_body["errcode"], Codes.PASSWORD_NO_DIGIT) -- cgit 1.4.1 From 665630fcaab8f09e83ff77f35d5244a718e20701 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Fri, 27 Mar 2020 11:39:43 +0000 Subject: Add tests for outbound device pokes --- changelog.d/7157.misc | 1 + tests/federation/test_federation_sender.py | 303 ++++++++++++++++++++++++++++- tests/unittest.py | 1 + 3 files changed, 302 insertions(+), 3 deletions(-) create mode 100644 changelog.d/7157.misc (limited to 'tests') diff --git a/changelog.d/7157.misc b/changelog.d/7157.misc new file mode 100644 index 0000000000..0eb1128c7a --- /dev/null +++ b/changelog.d/7157.misc @@ -0,0 +1 @@ +Add tests for outbound device pokes. diff --git a/tests/federation/test_federation_sender.py b/tests/federation/test_federation_sender.py index d456267b87..7763b12159 100644 --- a/tests/federation/test_federation_sender.py +++ b/tests/federation/test_federation_sender.py @@ -12,19 +12,25 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. +from typing import Optional from mock import Mock +from signedjson import key, sign +from signedjson.types import BaseKey, SigningKey + from twisted.internet import defer -from synapse.types import ReadReceipt +from synapse.rest import admin +from synapse.rest.client.v1 import login +from synapse.types import JsonDict, ReadReceipt from tests.unittest import HomeserverTestCase, override_config -class FederationSenderTestCases(HomeserverTestCase): +class FederationSenderReceiptsTestCases(HomeserverTestCase): def make_homeserver(self, reactor, clock): - return super(FederationSenderTestCases, self).setup_test_homeserver( + return self.setup_test_homeserver( state_handler=Mock(spec=["get_current_hosts_in_room"]), federation_transport_client=Mock(spec=["send_transaction"]), ) @@ -147,3 +153,294 @@ class FederationSenderTestCases(HomeserverTestCase): } ], ) + + +class FederationSenderDevicesTestCases(HomeserverTestCase): + servlets = [ + admin.register_servlets, + login.register_servlets, + ] + + def make_homeserver(self, reactor, clock): + return self.setup_test_homeserver( + state_handler=Mock(spec=["get_current_hosts_in_room"]), + federation_transport_client=Mock(spec=["send_transaction"]), + ) + + def default_config(self): + c = super().default_config() + c["send_federation"] = True + return c + + def prepare(self, reactor, clock, hs): + # stub out get_current_hosts_in_room + mock_state_handler = hs.get_state_handler() + mock_state_handler.get_current_hosts_in_room.return_value = ["test", "host2"] + + # stub out get_users_who_share_room_with_user so that it claims that + # `@user2:host2` is in the room + def get_users_who_share_room_with_user(user_id): + return defer.succeed({"@user2:host2"}) + + hs.get_datastore().get_users_who_share_room_with_user = ( + get_users_who_share_room_with_user + ) + + # whenever send_transaction is called, record the edu data + self.edus = [] + self.hs.get_federation_transport_client().send_transaction.side_effect = ( + self.record_transaction + ) + + def record_transaction(self, txn, json_cb): + data = json_cb() + self.edus.extend(data["edus"]) + return defer.succeed({}) + + def test_send_device_updates(self): + """Basic case: each device update should result in an EDU""" + # create a device + u1 = self.register_user("user", "pass") + self.login(u1, "pass", device_id="D1") + + # expect one edu + self.assertEqual(len(self.edus), 1) + stream_id = self.check_device_update_edu(self.edus.pop(0), u1, "D1", None) + + # a second call should produce no new device EDUs + self.hs.get_federation_sender().send_device_messages("host2") + self.pump() + self.assertEqual(self.edus, []) + + # a second device + self.login("user", "pass", device_id="D2") + + self.assertEqual(len(self.edus), 1) + self.check_device_update_edu(self.edus.pop(0), u1, "D2", stream_id) + + def test_upload_signatures(self): + """Uploading signatures on some devices should produce updates for that user""" + + e2e_handler = self.hs.get_e2e_keys_handler() + + # register two devices + u1 = self.register_user("user", "pass") + self.login(u1, "pass", device_id="D1") + self.login(u1, "pass", device_id="D2") + + # expect two edus + self.assertEqual(len(self.edus), 2) + stream_id = None + stream_id = self.check_device_update_edu(self.edus.pop(0), u1, "D1", stream_id) + stream_id = self.check_device_update_edu(self.edus.pop(0), u1, "D2", stream_id) + + # upload signing keys for each device + device1_signing_key = self.generate_and_upload_device_signing_key(u1, "D1") + device2_signing_key = self.generate_and_upload_device_signing_key(u1, "D2") + + # expect two more edus + self.assertEqual(len(self.edus), 2) + stream_id = self.check_device_update_edu(self.edus.pop(0), u1, "D1", stream_id) + stream_id = self.check_device_update_edu(self.edus.pop(0), u1, "D2", stream_id) + + # upload master key and self-signing key + master_signing_key = generate_self_id_key() + master_key = { + "user_id": u1, + "usage": ["master"], + "keys": {key_id(master_signing_key): encode_pubkey(master_signing_key)}, + } + + # private key: HvQBbU+hc2Zr+JP1sE0XwBe1pfZZEYtJNPJLZJtS+F8 + selfsigning_signing_key = generate_self_id_key() + selfsigning_key = { + "user_id": u1, + "usage": ["self_signing"], + "keys": { + key_id(selfsigning_signing_key): encode_pubkey(selfsigning_signing_key) + }, + } + sign.sign_json(selfsigning_key, u1, master_signing_key) + + cross_signing_keys = { + "master_key": master_key, + "self_signing_key": selfsigning_key, + } + + self.get_success( + e2e_handler.upload_signing_keys_for_user(u1, cross_signing_keys) + ) + + # expect signing key update edu + self.assertEqual(len(self.edus), 1) + self.assertEqual(self.edus.pop(0)["edu_type"], "org.matrix.signing_key_update") + + # sign the devices + d1_json = build_device_dict(u1, "D1", device1_signing_key) + sign.sign_json(d1_json, u1, selfsigning_signing_key) + d2_json = build_device_dict(u1, "D2", device2_signing_key) + sign.sign_json(d2_json, u1, selfsigning_signing_key) + + ret = self.get_success( + e2e_handler.upload_signatures_for_device_keys( + u1, {u1: {"D1": d1_json, "D2": d2_json}}, + ) + ) + self.assertEqual(ret["failures"], {}) + + # expect two edus, in one or two transactions. We don't know what order the + # devices will be updated. + self.assertEqual(len(self.edus), 2) + stream_id = None # FIXME: there is a discontinuity in the stream IDs: see #7142 + for edu in self.edus: + self.assertEqual(edu["edu_type"], "m.device_list_update") + c = edu["content"] + if stream_id is not None: + self.assertEqual(c["prev_id"], [stream_id]) + stream_id = c["stream_id"] + devices = {edu["content"]["device_id"] for edu in self.edus} + self.assertEqual({"D1", "D2"}, devices) + + def test_delete_devices(self): + """If devices are deleted, that should result in EDUs too""" + + # create devices + u1 = self.register_user("user", "pass") + self.login("user", "pass", device_id="D1") + self.login("user", "pass", device_id="D2") + self.login("user", "pass", device_id="D3") + + # expect three edus + self.assertEqual(len(self.edus), 3) + stream_id = self.check_device_update_edu(self.edus.pop(0), u1, "D1", None) + stream_id = self.check_device_update_edu(self.edus.pop(0), u1, "D2", stream_id) + stream_id = self.check_device_update_edu(self.edus.pop(0), u1, "D3", stream_id) + + # delete them again + self.get_success( + self.hs.get_device_handler().delete_devices(u1, ["D1", "D2", "D3"]) + ) + + # expect three edus, in an unknown order + self.assertEqual(len(self.edus), 3) + for edu in self.edus: + self.assertEqual(edu["edu_type"], "m.device_list_update") + c = edu["content"] + self.assertGreaterEqual( + c.items(), + {"user_id": u1, "prev_id": [stream_id], "deleted": True}.items(), + ) + stream_id = c["stream_id"] + devices = {edu["content"]["device_id"] for edu in self.edus} + self.assertEqual({"D1", "D2", "D3"}, devices) + + def test_unreachable_server(self): + """If the destination server is unreachable, all the updates should get sent on + recovery + """ + mock_send_txn = self.hs.get_federation_transport_client().send_transaction + mock_send_txn.side_effect = lambda t, cb: defer.fail("fail") + + # create devices + u1 = self.register_user("user", "pass") + self.login("user", "pass", device_id="D1") + self.login("user", "pass", device_id="D2") + self.login("user", "pass", device_id="D3") + + # delete them again + self.get_success( + self.hs.get_device_handler().delete_devices(u1, ["D1", "D2", "D3"]) + ) + + self.assertGreaterEqual(mock_send_txn.call_count, 4) + + # recover the server + mock_send_txn.side_effect = self.record_transaction + self.hs.get_federation_sender().send_device_messages("host2") + self.pump() + + # for each device, there should be a single update + self.assertEqual(len(self.edus), 3) + stream_id = None + for edu in self.edus: + self.assertEqual(edu["edu_type"], "m.device_list_update") + c = edu["content"] + self.assertEqual(c["prev_id"], [stream_id] if stream_id is not None else []) + stream_id = c["stream_id"] + devices = {edu["content"]["device_id"] for edu in self.edus} + self.assertEqual({"D1", "D2", "D3"}, devices) + + def check_device_update_edu( + self, + edu: JsonDict, + user_id: str, + device_id: str, + prev_stream_id: Optional[int], + ) -> int: + """Check that the given EDU is an update for the given device + Returns the stream_id. + """ + self.assertEqual(edu["edu_type"], "m.device_list_update") + content = edu["content"] + + expected = { + "user_id": user_id, + "device_id": device_id, + "prev_id": [prev_stream_id] if prev_stream_id is not None else [], + } + + self.assertLessEqual(expected.items(), content.items()) + return content["stream_id"] + + def check_signing_key_update_txn(self, txn: JsonDict,) -> None: + """Check that the txn has an EDU with a signing key update. + """ + edus = txn["edus"] + self.assertEqual(len(edus), 1) + + def generate_and_upload_device_signing_key( + self, user_id: str, device_id: str + ) -> SigningKey: + """Generate a signing keypair for the given device, and upload it""" + sk = key.generate_signing_key(device_id) + + device_dict = build_device_dict(user_id, device_id, sk) + + self.get_success( + self.hs.get_e2e_keys_handler().upload_keys_for_user( + user_id, device_id, {"device_keys": device_dict}, + ) + ) + return sk + + +def generate_self_id_key() -> SigningKey: + """generate a signing key whose version is its public key + + ... as used by the cross-signing-keys. + """ + k = key.generate_signing_key("x") + k.version = encode_pubkey(k) + return k + + +def key_id(k: BaseKey) -> str: + return "%s:%s" % (k.alg, k.version) + + +def encode_pubkey(sk: SigningKey) -> str: + """Encode the public key corresponding to the given signing key as base64""" + return key.encode_verify_key_base64(key.get_verify_key(sk)) + + +def build_device_dict(user_id: str, device_id: str, sk: SigningKey): + """Build a dict representing the given device""" + return { + "user_id": user_id, + "device_id": device_id, + "algorithms": ["m.olm.curve25519-aes-sha256", "m.megolm.v1.aes-sha"], + "keys": { + "curve25519:" + device_id: "curve25519+key", + key_id(sk): encode_pubkey(sk), + }, + } diff --git a/tests/unittest.py b/tests/unittest.py index 23b59bea22..3d57b77a5d 100644 --- a/tests/unittest.py +++ b/tests/unittest.py @@ -490,6 +490,7 @@ class HomeserverTestCase(TestCase): "password": password, "admin": admin, "mac": want_mac, + "inhibit_login": True, } ) request, channel = self.make_request( -- cgit 1.4.1 From 8327eb9280cbcb492e05652a96be9f1cd1c0e7c4 Mon Sep 17 00:00:00 2001 From: Dirk Klimpel <5740567+dklimpel@users.noreply.github.com> Date: Fri, 27 Mar 2020 20:15:23 +0100 Subject: Add options to prevent users from changing their profile. (#7096) --- changelog.d/7096.feature | 1 + docs/sample_config.yaml | 23 +++ synapse/config/registration.py | 27 +++ synapse/handlers/profile.py | 16 ++ synapse/rest/client/v2_alpha/account.py | 16 ++ tests/handlers/test_profile.py | 65 ++++++- tests/rest/client/v2_alpha/test_account.py | 302 +++++++++++++++++++++++++++++ 7 files changed, 449 insertions(+), 1 deletion(-) create mode 100644 changelog.d/7096.feature (limited to 'tests') diff --git a/changelog.d/7096.feature b/changelog.d/7096.feature new file mode 100644 index 0000000000..00f47b2a14 --- /dev/null +++ b/changelog.d/7096.feature @@ -0,0 +1 @@ +Add options to prevent users from changing their profile or associated 3PIDs. \ No newline at end of file diff --git a/docs/sample_config.yaml b/docs/sample_config.yaml index 1a1d061759..545226f753 100644 --- a/docs/sample_config.yaml +++ b/docs/sample_config.yaml @@ -1086,6 +1086,29 @@ account_threepid_delegates: #email: https://example.com # Delegate email sending to example.com #msisdn: http://localhost:8090 # Delegate SMS sending to this local process +# Whether users are allowed to change their displayname after it has +# been initially set. Useful when provisioning users based on the +# contents of a third-party directory. +# +# Does not apply to server administrators. Defaults to 'true' +# +#enable_set_displayname: false + +# Whether users are allowed to change their avatar after it has been +# initially set. Useful when provisioning users based on the contents +# of a third-party directory. +# +# Does not apply to server administrators. Defaults to 'true' +# +#enable_set_avatar_url: false + +# Whether users can change the 3PIDs associated with their accounts +# (email address and msisdn). +# +# Defaults to 'true' +# +#enable_3pid_changes: false + # Users who register on this homeserver will automatically be joined # to these rooms # diff --git a/synapse/config/registration.py b/synapse/config/registration.py index 9bb3beedbc..e7ea3a01cb 100644 --- a/synapse/config/registration.py +++ b/synapse/config/registration.py @@ -129,6 +129,10 @@ class RegistrationConfig(Config): raise ConfigError("Invalid auto_join_rooms entry %s" % (room_alias,)) self.autocreate_auto_join_rooms = config.get("autocreate_auto_join_rooms", True) + self.enable_set_displayname = config.get("enable_set_displayname", True) + self.enable_set_avatar_url = config.get("enable_set_avatar_url", True) + self.enable_3pid_changes = config.get("enable_3pid_changes", True) + self.disable_msisdn_registration = config.get( "disable_msisdn_registration", False ) @@ -330,6 +334,29 @@ class RegistrationConfig(Config): #email: https://example.com # Delegate email sending to example.com #msisdn: http://localhost:8090 # Delegate SMS sending to this local process + # Whether users are allowed to change their displayname after it has + # been initially set. Useful when provisioning users based on the + # contents of a third-party directory. + # + # Does not apply to server administrators. Defaults to 'true' + # + #enable_set_displayname: false + + # Whether users are allowed to change their avatar after it has been + # initially set. Useful when provisioning users based on the contents + # of a third-party directory. + # + # Does not apply to server administrators. Defaults to 'true' + # + #enable_set_avatar_url: false + + # Whether users can change the 3PIDs associated with their accounts + # (email address and msisdn). + # + # Defaults to 'true' + # + #enable_3pid_changes: false + # Users who register on this homeserver will automatically be joined # to these rooms # diff --git a/synapse/handlers/profile.py b/synapse/handlers/profile.py index 50ce0c585b..6aa1c0f5e0 100644 --- a/synapse/handlers/profile.py +++ b/synapse/handlers/profile.py @@ -157,6 +157,15 @@ class BaseProfileHandler(BaseHandler): if not by_admin and target_user != requester.user: raise AuthError(400, "Cannot set another user's displayname") + if not by_admin and not self.hs.config.enable_set_displayname: + profile = yield self.store.get_profileinfo(target_user.localpart) + if profile.display_name: + raise SynapseError( + 400, + "Changing display name is disabled on this server", + Codes.FORBIDDEN, + ) + if len(new_displayname) > MAX_DISPLAYNAME_LEN: raise SynapseError( 400, "Displayname is too long (max %i)" % (MAX_DISPLAYNAME_LEN,) @@ -218,6 +227,13 @@ class BaseProfileHandler(BaseHandler): if not by_admin and target_user != requester.user: raise AuthError(400, "Cannot set another user's avatar_url") + if not by_admin and not self.hs.config.enable_set_avatar_url: + profile = yield self.store.get_profileinfo(target_user.localpart) + if profile.avatar_url: + raise SynapseError( + 400, "Changing avatar is disabled on this server", Codes.FORBIDDEN + ) + if len(new_avatar_url) > MAX_AVATAR_URL_LEN: raise SynapseError( 400, "Avatar URL is too long (max %i)" % (MAX_AVATAR_URL_LEN,) diff --git a/synapse/rest/client/v2_alpha/account.py b/synapse/rest/client/v2_alpha/account.py index b1249b664c..f80b5e40ea 100644 --- a/synapse/rest/client/v2_alpha/account.py +++ b/synapse/rest/client/v2_alpha/account.py @@ -605,6 +605,11 @@ class ThreepidRestServlet(RestServlet): return 200, {"threepids": threepids} async def on_POST(self, request): + if not self.hs.config.enable_3pid_changes: + raise SynapseError( + 400, "3PID changes are disabled on this server", Codes.FORBIDDEN + ) + requester = await self.auth.get_user_by_req(request) user_id = requester.user.to_string() body = parse_json_object_from_request(request) @@ -649,6 +654,11 @@ class ThreepidAddRestServlet(RestServlet): @interactive_auth_handler async def on_POST(self, request): + if not self.hs.config.enable_3pid_changes: + raise SynapseError( + 400, "3PID changes are disabled on this server", Codes.FORBIDDEN + ) + requester = await self.auth.get_user_by_req(request) user_id = requester.user.to_string() body = parse_json_object_from_request(request) @@ -744,10 +754,16 @@ class ThreepidDeleteRestServlet(RestServlet): def __init__(self, hs): super(ThreepidDeleteRestServlet, self).__init__() + self.hs = hs self.auth = hs.get_auth() self.auth_handler = hs.get_auth_handler() async def on_POST(self, request): + if not self.hs.config.enable_3pid_changes: + raise SynapseError( + 400, "3PID changes are disabled on this server", Codes.FORBIDDEN + ) + body = parse_json_object_from_request(request) assert_params_in_dict(body, ["medium", "address"]) diff --git a/tests/handlers/test_profile.py b/tests/handlers/test_profile.py index d60c124eec..be665262c6 100644 --- a/tests/handlers/test_profile.py +++ b/tests/handlers/test_profile.py @@ -19,7 +19,7 @@ from mock import Mock, NonCallableMock from twisted.internet import defer import synapse.types -from synapse.api.errors import AuthError +from synapse.api.errors import AuthError, SynapseError from synapse.handlers.profile import MasterProfileHandler from synapse.types import UserID @@ -70,6 +70,7 @@ class ProfileTestCase(unittest.TestCase): yield self.store.create_profile(self.frank.localpart) self.handler = hs.get_profile_handler() + self.hs = hs @defer.inlineCallbacks def test_get_my_name(self): @@ -90,6 +91,33 @@ class ProfileTestCase(unittest.TestCase): "Frank Jr.", ) + # Set displayname again + yield self.handler.set_displayname( + self.frank, synapse.types.create_requester(self.frank), "Frank" + ) + + self.assertEquals( + (yield self.store.get_profile_displayname(self.frank.localpart)), "Frank", + ) + + @defer.inlineCallbacks + def test_set_my_name_if_disabled(self): + self.hs.config.enable_set_displayname = False + + # Setting displayname for the first time is allowed + yield self.store.set_profile_displayname(self.frank.localpart, "Frank") + + self.assertEquals( + (yield self.store.get_profile_displayname(self.frank.localpart)), "Frank", + ) + + # Setting displayname a second time is forbidden + d = self.handler.set_displayname( + self.frank, synapse.types.create_requester(self.frank), "Frank Jr." + ) + + yield self.assertFailure(d, SynapseError) + @defer.inlineCallbacks def test_set_my_name_noauth(self): d = self.handler.set_displayname( @@ -147,3 +175,38 @@ class ProfileTestCase(unittest.TestCase): (yield self.store.get_profile_avatar_url(self.frank.localpart)), "http://my.server/pic.gif", ) + + # Set avatar again + yield self.handler.set_avatar_url( + self.frank, + synapse.types.create_requester(self.frank), + "http://my.server/me.png", + ) + + self.assertEquals( + (yield self.store.get_profile_avatar_url(self.frank.localpart)), + "http://my.server/me.png", + ) + + @defer.inlineCallbacks + def test_set_my_avatar_if_disabled(self): + self.hs.config.enable_set_avatar_url = False + + # Setting displayname for the first time is allowed + yield self.store.set_profile_avatar_url( + self.frank.localpart, "http://my.server/me.png" + ) + + self.assertEquals( + (yield self.store.get_profile_avatar_url(self.frank.localpart)), + "http://my.server/me.png", + ) + + # Set avatar a second time is forbidden + d = self.handler.set_avatar_url( + self.frank, + synapse.types.create_requester(self.frank), + "http://my.server/pic.gif", + ) + + yield self.assertFailure(d, SynapseError) diff --git a/tests/rest/client/v2_alpha/test_account.py b/tests/rest/client/v2_alpha/test_account.py index c3facc00eb..45a9d445f8 100644 --- a/tests/rest/client/v2_alpha/test_account.py +++ b/tests/rest/client/v2_alpha/test_account.py @@ -24,6 +24,7 @@ import pkg_resources import synapse.rest.admin from synapse.api.constants import LoginType, Membership +from synapse.api.errors import Codes from synapse.rest.client.v1 import login, room from synapse.rest.client.v2_alpha import account, register @@ -325,3 +326,304 @@ class DeactivateTestCase(unittest.HomeserverTestCase): ) self.render(request) self.assertEqual(request.code, 200) + + +class ThreepidEmailRestTestCase(unittest.HomeserverTestCase): + + servlets = [ + account.register_servlets, + login.register_servlets, + synapse.rest.admin.register_servlets_for_client_rest_resource, + ] + + def make_homeserver(self, reactor, clock): + config = self.default_config() + + # Email config. + self.email_attempts = [] + + def sendmail(smtphost, from_addr, to_addrs, msg, **kwargs): + self.email_attempts.append(msg) + + config["email"] = { + "enable_notifs": False, + "template_dir": os.path.abspath( + pkg_resources.resource_filename("synapse", "res/templates") + ), + "smtp_host": "127.0.0.1", + "smtp_port": 20, + "require_transport_security": False, + "smtp_user": None, + "smtp_pass": None, + "notif_from": "test@example.com", + } + config["public_baseurl"] = "https://example.com" + + self.hs = self.setup_test_homeserver(config=config, sendmail=sendmail) + return self.hs + + def prepare(self, reactor, clock, hs): + self.store = hs.get_datastore() + + self.user_id = self.register_user("kermit", "test") + self.user_id_tok = self.login("kermit", "test") + self.email = "test@example.com" + self.url_3pid = b"account/3pid" + + def test_add_email(self): + """Test adding an email to profile + """ + client_secret = "foobar" + session_id = self._request_token(self.email, client_secret) + + self.assertEquals(len(self.email_attempts), 1) + link = self._get_link_from_email() + + self._validate_token(link) + + request, channel = self.make_request( + "POST", + b"/_matrix/client/unstable/account/3pid/add", + { + "client_secret": client_secret, + "sid": session_id, + "auth": { + "type": "m.login.password", + "user": self.user_id, + "password": "test", + }, + }, + access_token=self.user_id_tok, + ) + + self.render(request) + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + + # Get user + request, channel = self.make_request( + "GET", self.url_3pid, access_token=self.user_id_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("email", channel.json_body["threepids"][0]["medium"]) + self.assertEqual(self.email, channel.json_body["threepids"][0]["address"]) + + def test_add_email_if_disabled(self): + """Test adding email to profile when doing so is disallowed + """ + self.hs.config.enable_3pid_changes = False + + client_secret = "foobar" + session_id = self._request_token(self.email, client_secret) + + self.assertEquals(len(self.email_attempts), 1) + link = self._get_link_from_email() + + self._validate_token(link) + + request, channel = self.make_request( + "POST", + b"/_matrix/client/unstable/account/3pid/add", + { + "client_secret": client_secret, + "sid": session_id, + "auth": { + "type": "m.login.password", + "user": self.user_id, + "password": "test", + }, + }, + access_token=self.user_id_tok, + ) + self.render(request) + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual(Codes.FORBIDDEN, channel.json_body["errcode"]) + + # Get user + request, channel = self.make_request( + "GET", self.url_3pid, access_token=self.user_id_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertFalse(channel.json_body["threepids"]) + + def test_delete_email(self): + """Test deleting an email from profile + """ + # Add a threepid + self.get_success( + self.store.user_add_threepid( + user_id=self.user_id, + medium="email", + address=self.email, + validated_at=0, + added_at=0, + ) + ) + + request, channel = self.make_request( + "POST", + b"account/3pid/delete", + {"medium": "email", "address": self.email}, + access_token=self.user_id_tok, + ) + self.render(request) + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + + # Get user + request, channel = self.make_request( + "GET", self.url_3pid, access_token=self.user_id_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertFalse(channel.json_body["threepids"]) + + def test_delete_email_if_disabled(self): + """Test deleting an email from profile when disallowed + """ + self.hs.config.enable_3pid_changes = False + + # Add a threepid + self.get_success( + self.store.user_add_threepid( + user_id=self.user_id, + medium="email", + address=self.email, + validated_at=0, + added_at=0, + ) + ) + + request, channel = self.make_request( + "POST", + b"account/3pid/delete", + {"medium": "email", "address": self.email}, + access_token=self.user_id_tok, + ) + self.render(request) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual(Codes.FORBIDDEN, channel.json_body["errcode"]) + + # Get user + request, channel = self.make_request( + "GET", self.url_3pid, access_token=self.user_id_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("email", channel.json_body["threepids"][0]["medium"]) + self.assertEqual(self.email, channel.json_body["threepids"][0]["address"]) + + def test_cant_add_email_without_clicking_link(self): + """Test that we do actually need to click the link in the email + """ + client_secret = "foobar" + session_id = self._request_token(self.email, client_secret) + + self.assertEquals(len(self.email_attempts), 1) + + # Attempt to add email without clicking the link + request, channel = self.make_request( + "POST", + b"/_matrix/client/unstable/account/3pid/add", + { + "client_secret": client_secret, + "sid": session_id, + "auth": { + "type": "m.login.password", + "user": self.user_id, + "password": "test", + }, + }, + access_token=self.user_id_tok, + ) + self.render(request) + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual(Codes.THREEPID_AUTH_FAILED, channel.json_body["errcode"]) + + # Get user + request, channel = self.make_request( + "GET", self.url_3pid, access_token=self.user_id_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertFalse(channel.json_body["threepids"]) + + def test_no_valid_token(self): + """Test that we do actually need to request a token and can't just + make a session up. + """ + client_secret = "foobar" + session_id = "weasle" + + # Attempt to add email without even requesting an email + request, channel = self.make_request( + "POST", + b"/_matrix/client/unstable/account/3pid/add", + { + "client_secret": client_secret, + "sid": session_id, + "auth": { + "type": "m.login.password", + "user": self.user_id, + "password": "test", + }, + }, + access_token=self.user_id_tok, + ) + self.render(request) + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual(Codes.THREEPID_AUTH_FAILED, channel.json_body["errcode"]) + + # Get user + request, channel = self.make_request( + "GET", self.url_3pid, access_token=self.user_id_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertFalse(channel.json_body["threepids"]) + + def _request_token(self, email, client_secret): + request, channel = self.make_request( + "POST", + b"account/3pid/email/requestToken", + {"client_secret": client_secret, "email": email, "send_attempt": 1}, + ) + self.render(request) + self.assertEquals(200, channel.code, channel.result) + + return channel.json_body["sid"] + + def _validate_token(self, link): + # Remove the host + path = link.replace("https://example.com", "") + + request, channel = self.make_request("GET", path, shorthand=False) + self.render(request) + self.assertEquals(200, channel.code, channel.result) + + def _get_link_from_email(self): + assert self.email_attempts, "No emails have been sent" + + raw_msg = self.email_attempts[-1].decode("UTF-8") + mail = Parser().parsestr(raw_msg) + + text = None + for part in mail.walk(): + if part.get_content_type() == "text/plain": + text = part.get_payload(decode=True).decode("UTF-8") + break + + if not text: + self.fail("Could not find text portion of email to parse") + + match = re.search(r"https://example.com\S+", text) + assert match, "Could not find link in email" + + return match.group(0) -- cgit 1.4.1 From fb69690761762092c8e44d509d4f72408c4c67e0 Mon Sep 17 00:00:00 2001 From: Dirk Klimpel <5740567+dklimpel@users.noreply.github.com> Date: Fri, 27 Mar 2020 20:16:43 +0100 Subject: Admin API to join users to a room. (#7051) --- changelog.d/7051.feature | 1 + docs/admin_api/room_membership.md | 34 +++++ synapse/rest/admin/__init__.py | 7 +- synapse/rest/admin/rooms.py | 79 ++++++++++- tests/rest/admin/test_room.py | 288 ++++++++++++++++++++++++++++++++++++++ 5 files changed, 405 insertions(+), 4 deletions(-) create mode 100644 changelog.d/7051.feature create mode 100644 docs/admin_api/room_membership.md create mode 100644 tests/rest/admin/test_room.py (limited to 'tests') diff --git a/changelog.d/7051.feature b/changelog.d/7051.feature new file mode 100644 index 0000000000..3e36a3f65e --- /dev/null +++ b/changelog.d/7051.feature @@ -0,0 +1 @@ +Admin API `POST /_synapse/admin/v1/join/` to join users to a room like `auto_join_rooms` for creation of users. \ No newline at end of file diff --git a/docs/admin_api/room_membership.md b/docs/admin_api/room_membership.md new file mode 100644 index 0000000000..16736d3d37 --- /dev/null +++ b/docs/admin_api/room_membership.md @@ -0,0 +1,34 @@ +# Edit Room Membership API + +This API allows an administrator to join an user account with a given `user_id` +to a room with a given `room_id_or_alias`. You can only modify the membership of +local users. The server administrator must be in the room and have permission to +invite users. + +## Parameters + +The following parameters are available: + +* `user_id` - Fully qualified user: for example, `@user:server.com`. +* `room_id_or_alias` - The room identifier or alias to join: for example, + `!636q39766251:server.com`. + +## Usage + +``` +POST /_synapse/admin/v1/join/ + +{ + "user_id": "@user:server.com" +} +``` + +Including an `access_token` of a server admin. + +Response: + +``` +{ + "room_id": "!636q39766251:server.com" +} +``` diff --git a/synapse/rest/admin/__init__.py b/synapse/rest/admin/__init__.py index 42cc2b062a..ed70d448a1 100644 --- a/synapse/rest/admin/__init__.py +++ b/synapse/rest/admin/__init__.py @@ -29,7 +29,11 @@ from synapse.rest.admin._base import ( from synapse.rest.admin.groups import DeleteGroupAdminRestServlet from synapse.rest.admin.media import ListMediaInRoom, register_servlets_for_media_repo from synapse.rest.admin.purge_room_servlet import PurgeRoomServlet -from synapse.rest.admin.rooms import ListRoomRestServlet, ShutdownRoomRestServlet +from synapse.rest.admin.rooms import ( + JoinRoomAliasServlet, + ListRoomRestServlet, + ShutdownRoomRestServlet, +) from synapse.rest.admin.server_notice_servlet import SendServerNoticeServlet from synapse.rest.admin.users import ( AccountValidityRenewServlet, @@ -189,6 +193,7 @@ def register_servlets(hs, http_server): """ register_servlets_for_client_rest_resource(hs, http_server) ListRoomRestServlet(hs).register(http_server) + JoinRoomAliasServlet(hs).register(http_server) PurgeRoomServlet(hs).register(http_server) SendServerNoticeServlet(hs).register(http_server) VersionServlet(hs).register(http_server) diff --git a/synapse/rest/admin/rooms.py b/synapse/rest/admin/rooms.py index f9b8c0a4f0..659b8a10ee 100644 --- a/synapse/rest/admin/rooms.py +++ b/synapse/rest/admin/rooms.py @@ -13,9 +13,10 @@ # See the License for the specific language governing permissions and # limitations under the License. import logging +from typing import List, Optional -from synapse.api.constants import Membership -from synapse.api.errors import Codes, SynapseError +from synapse.api.constants import EventTypes, JoinRules, Membership +from synapse.api.errors import Codes, NotFoundError, SynapseError from synapse.http.servlet import ( RestServlet, assert_params_in_dict, @@ -29,7 +30,7 @@ from synapse.rest.admin._base import ( historical_admin_path_patterns, ) from synapse.storage.data_stores.main.room import RoomSortOrder -from synapse.types import create_requester +from synapse.types import RoomAlias, RoomID, UserID, create_requester from synapse.util.async_helpers import maybe_awaitable logger = logging.getLogger(__name__) @@ -237,3 +238,75 @@ class ListRoomRestServlet(RestServlet): response["prev_batch"] = 0 return 200, response + + +class JoinRoomAliasServlet(RestServlet): + + PATTERNS = admin_patterns("/join/(?P[^/]*)") + + def __init__(self, hs): + self.hs = hs + self.auth = hs.get_auth() + self.room_member_handler = hs.get_room_member_handler() + self.admin_handler = hs.get_handlers().admin_handler + self.state_handler = hs.get_state_handler() + + async def on_POST(self, request, room_identifier): + requester = await self.auth.get_user_by_req(request) + await assert_user_is_admin(self.auth, requester.user) + + content = parse_json_object_from_request(request) + + assert_params_in_dict(content, ["user_id"]) + target_user = UserID.from_string(content["user_id"]) + + if not self.hs.is_mine(target_user): + raise SynapseError(400, "This endpoint can only be used with local users") + + if not await self.admin_handler.get_user(target_user): + raise NotFoundError("User not found") + + if RoomID.is_valid(room_identifier): + room_id = room_identifier + try: + remote_room_hosts = [ + x.decode("ascii") for x in request.args[b"server_name"] + ] # type: Optional[List[str]] + except Exception: + remote_room_hosts = None + elif RoomAlias.is_valid(room_identifier): + handler = self.room_member_handler + room_alias = RoomAlias.from_string(room_identifier) + room_id, remote_room_hosts = await handler.lookup_room_alias(room_alias) + room_id = room_id.to_string() + else: + raise SynapseError( + 400, "%s was not legal room ID or room alias" % (room_identifier,) + ) + + fake_requester = create_requester(target_user) + + # send invite if room has "JoinRules.INVITE" + room_state = await self.state_handler.get_current_state(room_id) + join_rules_event = room_state.get((EventTypes.JoinRules, "")) + if join_rules_event: + if not (join_rules_event.content.get("join_rule") == JoinRules.PUBLIC): + await self.room_member_handler.update_membership( + requester=requester, + target=fake_requester.user, + room_id=room_id, + action="invite", + remote_room_hosts=remote_room_hosts, + ratelimit=False, + ) + + await self.room_member_handler.update_membership( + requester=fake_requester, + target=fake_requester.user, + room_id=room_id, + action="join", + remote_room_hosts=remote_room_hosts, + ratelimit=False, + ) + + return 200, {"room_id": room_id} diff --git a/tests/rest/admin/test_room.py b/tests/rest/admin/test_room.py new file mode 100644 index 0000000000..672cc3eac5 --- /dev/null +++ b/tests/rest/admin/test_room.py @@ -0,0 +1,288 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 Dirk Klimpel +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import json + +import synapse.rest.admin +from synapse.api.errors import Codes +from synapse.rest.client.v1 import login, room + +from tests import unittest + +"""Tests admin REST events for /rooms paths.""" + + +class JoinAliasRoomTestCase(unittest.HomeserverTestCase): + + servlets = [ + synapse.rest.admin.register_servlets, + room.register_servlets, + login.register_servlets, + ] + + def prepare(self, reactor, clock, homeserver): + self.admin_user = self.register_user("admin", "pass", admin=True) + self.admin_user_tok = self.login("admin", "pass") + + self.creator = self.register_user("creator", "test") + self.creator_tok = self.login("creator", "test") + + self.second_user_id = self.register_user("second", "test") + self.second_tok = self.login("second", "test") + + self.public_room_id = self.helper.create_room_as( + self.creator, tok=self.creator_tok, is_public=True + ) + self.url = "/_synapse/admin/v1/join/{}".format(self.public_room_id) + + def test_requester_is_no_admin(self): + """ + If the user is not a server admin, an error 403 is returned. + """ + body = json.dumps({"user_id": self.second_user_id}) + + request, channel = self.make_request( + "POST", + self.url, + content=body.encode(encoding="utf_8"), + access_token=self.second_tok, + ) + self.render(request) + + self.assertEqual(403, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual(Codes.FORBIDDEN, channel.json_body["errcode"]) + + def test_invalid_parameter(self): + """ + If a parameter is missing, return an error + """ + body = json.dumps({"unknown_parameter": "@unknown:test"}) + + request, channel = self.make_request( + "POST", + self.url, + content=body.encode(encoding="utf_8"), + access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual(Codes.MISSING_PARAM, channel.json_body["errcode"]) + + def test_local_user_does_not_exist(self): + """ + Tests that a lookup for a user that does not exist returns a 404 + """ + body = json.dumps({"user_id": "@unknown:test"}) + + request, channel = self.make_request( + "POST", + self.url, + content=body.encode(encoding="utf_8"), + access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(404, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual(Codes.NOT_FOUND, channel.json_body["errcode"]) + + def test_remote_user(self): + """ + Check that only local user can join rooms. + """ + body = json.dumps({"user_id": "@not:exist.bla"}) + + request, channel = self.make_request( + "POST", + self.url, + content=body.encode(encoding="utf_8"), + access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual( + "This endpoint can only be used with local users", + channel.json_body["error"], + ) + + def test_room_does_not_exist(self): + """ + Check that unknown rooms/server return error 404. + """ + body = json.dumps({"user_id": self.second_user_id}) + url = "/_synapse/admin/v1/join/!unknown:test" + + request, channel = self.make_request( + "POST", + url, + content=body.encode(encoding="utf_8"), + access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(404, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("No known servers", channel.json_body["error"]) + + def test_room_is_not_valid(self): + """ + Check that invalid room names, return an error 400. + """ + body = json.dumps({"user_id": self.second_user_id}) + url = "/_synapse/admin/v1/join/invalidroom" + + request, channel = self.make_request( + "POST", + url, + content=body.encode(encoding="utf_8"), + access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual( + "invalidroom was not legal room ID or room alias", + channel.json_body["error"], + ) + + def test_join_public_room(self): + """ + Test joining a local user to a public room with "JoinRules.PUBLIC" + """ + body = json.dumps({"user_id": self.second_user_id}) + + request, channel = self.make_request( + "POST", + self.url, + content=body.encode(encoding="utf_8"), + access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual(self.public_room_id, channel.json_body["room_id"]) + + # Validate if user is a member of the room + + request, channel = self.make_request( + "GET", "/_matrix/client/r0/joined_rooms", access_token=self.second_tok, + ) + self.render(request) + self.assertEquals(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual(self.public_room_id, channel.json_body["joined_rooms"][0]) + + def test_join_private_room_if_not_member(self): + """ + Test joining a local user to a private room with "JoinRules.INVITE" + when server admin is not member of this room. + """ + private_room_id = self.helper.create_room_as( + self.creator, tok=self.creator_tok, is_public=False + ) + url = "/_synapse/admin/v1/join/{}".format(private_room_id) + body = json.dumps({"user_id": self.second_user_id}) + + request, channel = self.make_request( + "POST", + url, + content=body.encode(encoding="utf_8"), + access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(403, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual(Codes.FORBIDDEN, channel.json_body["errcode"]) + + def test_join_private_room_if_member(self): + """ + Test joining a local user to a private room with "JoinRules.INVITE", + when server admin is member of this room. + """ + private_room_id = self.helper.create_room_as( + self.creator, tok=self.creator_tok, is_public=False + ) + self.helper.invite( + room=private_room_id, + src=self.creator, + targ=self.admin_user, + tok=self.creator_tok, + ) + self.helper.join( + room=private_room_id, user=self.admin_user, tok=self.admin_user_tok + ) + + # Validate if server admin is a member of the room + + request, channel = self.make_request( + "GET", "/_matrix/client/r0/joined_rooms", access_token=self.admin_user_tok, + ) + self.render(request) + self.assertEquals(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual(private_room_id, channel.json_body["joined_rooms"][0]) + + # Join user to room. + + url = "/_synapse/admin/v1/join/{}".format(private_room_id) + body = json.dumps({"user_id": self.second_user_id}) + + request, channel = self.make_request( + "POST", + url, + content=body.encode(encoding="utf_8"), + access_token=self.admin_user_tok, + ) + self.render(request) + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual(private_room_id, channel.json_body["room_id"]) + + # Validate if user is a member of the room + + request, channel = self.make_request( + "GET", "/_matrix/client/r0/joined_rooms", access_token=self.second_tok, + ) + self.render(request) + self.assertEquals(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual(private_room_id, channel.json_body["joined_rooms"][0]) + + def test_join_private_room_if_owner(self): + """ + Test joining a local user to a private room with "JoinRules.INVITE", + when server admin is owner of this room. + """ + private_room_id = self.helper.create_room_as( + self.admin_user, tok=self.admin_user_tok, is_public=False + ) + url = "/_synapse/admin/v1/join/{}".format(private_room_id) + body = json.dumps({"user_id": self.second_user_id}) + + request, channel = self.make_request( + "POST", + url, + content=body.encode(encoding="utf_8"), + access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual(private_room_id, channel.json_body["room_id"]) + + # Validate if user is a member of the room + + request, channel = self.make_request( + "GET", "/_matrix/client/r0/joined_rooms", access_token=self.second_tok, + ) + self.render(request) + self.assertEquals(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual(private_room_id, channel.json_body["joined_rooms"][0]) -- cgit 1.4.1 From b7da598a61a1bcea3855edf403bdc5ea32cc9e7a Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Fri, 27 Mar 2020 20:24:52 +0000 Subject: Always whitelist the login fallback for SSO (#7153) That fallback sets the redirect URL to itself (so it can process the login token then return gracefully to the client). This would make it pointless to ask the user for confirmation, since the URL the confirmation page would be showing wouldn't be the client's. --- changelog.d/7153.feature | 1 + docs/sample_config.yaml | 4 ++++ synapse/config/sso.py | 15 +++++++++++++++ tests/rest/client/v1/test_login.py | 9 ++++++++- 4 files changed, 28 insertions(+), 1 deletion(-) create mode 100644 changelog.d/7153.feature (limited to 'tests') diff --git a/changelog.d/7153.feature b/changelog.d/7153.feature new file mode 100644 index 0000000000..414ebe1f69 --- /dev/null +++ b/changelog.d/7153.feature @@ -0,0 +1 @@ +Always whitelist the login fallback in the SSO configuration if `public_baseurl` is set. diff --git a/docs/sample_config.yaml b/docs/sample_config.yaml index 545226f753..743949945a 100644 --- a/docs/sample_config.yaml +++ b/docs/sample_config.yaml @@ -1444,6 +1444,10 @@ sso: # phishing attacks from evil.site. To avoid this, include a slash after the # hostname: "https://my.client/". # + # If public_baseurl is set, then the login fallback page (used by clients + # that don't natively support the required login flows) is whitelisted in + # addition to any URLs in this list. + # # By default, this list is empty. # #client_whitelist: diff --git a/synapse/config/sso.py b/synapse/config/sso.py index 95762689bc..ec3dca9efc 100644 --- a/synapse/config/sso.py +++ b/synapse/config/sso.py @@ -39,6 +39,17 @@ class SSOConfig(Config): self.sso_client_whitelist = sso_config.get("client_whitelist") or [] + # Attempt to also whitelist the server's login fallback, since that fallback sets + # the redirect URL to itself (so it can process the login token then return + # gracefully to the client). This would make it pointless to ask the user for + # confirmation, since the URL the confirmation page would be showing wouldn't be + # the client's. + # public_baseurl is an optional setting, so we only add the fallback's URL to the + # list if it's provided (because we can't figure out what that URL is otherwise). + if self.public_baseurl: + login_fallback_url = self.public_baseurl + "_matrix/static/client/login" + self.sso_client_whitelist.append(login_fallback_url) + def generate_config_section(self, **kwargs): return """\ # Additional settings to use with single-sign on systems such as SAML2 and CAS. @@ -54,6 +65,10 @@ class SSOConfig(Config): # phishing attacks from evil.site. To avoid this, include a slash after the # hostname: "https://my.client/". # + # If public_baseurl is set, then the login fallback page (used by clients + # that don't natively support the required login flows) is whitelisted in + # addition to any URLs in this list. + # # By default, this list is empty. # #client_whitelist: diff --git a/tests/rest/client/v1/test_login.py b/tests/rest/client/v1/test_login.py index da2c9bfa1e..aed8853d6e 100644 --- a/tests/rest/client/v1/test_login.py +++ b/tests/rest/client/v1/test_login.py @@ -350,7 +350,14 @@ class CASRedirectConfirmTestCase(unittest.HomeserverTestCase): def test_cas_redirect_whitelisted(self): """Tests that the SSO login flow serves a redirect to a whitelisted url """ - redirect_url = "https://legit-site.com/" + self._test_redirect("https://legit-site.com/") + + @override_config({"public_baseurl": "https://example.com"}) + def test_cas_redirect_login_fallback(self): + self._test_redirect("https://example.com/_matrix/static/client/login") + + def _test_redirect(self, redirect_url): + """Tests that the SSO login flow serves a redirect for the given redirect URL.""" cas_ticket_url = ( "/_matrix/client/r0/login/cas/ticket?redirectUrl=%s&ticket=ticket" % (urllib.parse.quote(redirect_url)) -- cgit 1.4.1 From 7966a1cde9d4b598faa06620424844f2b35c94af Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Mon, 30 Mar 2020 19:06:52 +0100 Subject: Rewrite prune_old_outbound_device_pokes for efficiency (#7159) make sure we clear out all but one update for the user --- changelog.d/7159.bugfix | 1 + synapse/handlers/federation.py | 25 +------- synapse/storage/data_stores/main/devices.py | 71 ++++++++++++++++++---- synapse/util/stringutils.py | 21 ++++++- tests/federation/test_federation_sender.py | 92 +++++++++++++++++++++++++++++ 5 files changed, 173 insertions(+), 37 deletions(-) create mode 100644 changelog.d/7159.bugfix (limited to 'tests') diff --git a/changelog.d/7159.bugfix b/changelog.d/7159.bugfix new file mode 100644 index 0000000000..1b341b127b --- /dev/null +++ b/changelog.d/7159.bugfix @@ -0,0 +1 @@ +Fix excessive CPU usage by `prune_old_outbound_device_pokes` job. diff --git a/synapse/handlers/federation.py b/synapse/handlers/federation.py index 38ab6a8fc3..c7aa7acf3b 100644 --- a/synapse/handlers/federation.py +++ b/synapse/handlers/federation.py @@ -49,6 +49,7 @@ from synapse.event_auth import auth_types_for_event from synapse.events import EventBase from synapse.events.snapshot import EventContext from synapse.events.validator import EventValidator +from synapse.handlers._base import BaseHandler from synapse.logging.context import ( make_deferred_yieldable, nested_logging_context, @@ -69,10 +70,9 @@ from synapse.types import JsonDict, StateMap, UserID, get_domain_from_id from synapse.util.async_helpers import Linearizer, concurrently_execute from synapse.util.distributor import user_joined_room from synapse.util.retryutils import NotRetryingDestination +from synapse.util.stringutils import shortstr from synapse.visibility import filter_events_for_server -from ._base import BaseHandler - logger = logging.getLogger(__name__) @@ -93,27 +93,6 @@ class _NewEventInfo: auth_events = attr.ib(type=Optional[StateMap[EventBase]], default=None) -def shortstr(iterable, maxitems=5): - """If iterable has maxitems or fewer, return the stringification of a list - containing those items. - - Otherwise, return the stringification of a a list with the first maxitems items, - followed by "...". - - Args: - iterable (Iterable): iterable to truncate - maxitems (int): number of items to return before truncating - - Returns: - unicode - """ - - items = list(itertools.islice(iterable, maxitems + 1)) - if len(items) <= maxitems: - return str(items) - return "[" + ", ".join(repr(r) for r in items[:maxitems]) + ", ...]" - - class FederationHandler(BaseHandler): """Handles events that originated from federation. Responsible for: diff --git a/synapse/storage/data_stores/main/devices.py b/synapse/storage/data_stores/main/devices.py index 2d47cfd131..3140e1b722 100644 --- a/synapse/storage/data_stores/main/devices.py +++ b/synapse/storage/data_stores/main/devices.py @@ -41,6 +41,7 @@ from synapse.util.caches.descriptors import ( cachedList, ) from synapse.util.iterutils import batch_iter +from synapse.util.stringutils import shortstr logger = logging.getLogger(__name__) @@ -1092,18 +1093,47 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): ], ) - def _prune_old_outbound_device_pokes(self): + def _prune_old_outbound_device_pokes(self, prune_age=24 * 60 * 60 * 1000): """Delete old entries out of the device_lists_outbound_pokes to ensure - that we don't fill up due to dead servers. We keep one entry per - (destination, user_id) tuple to ensure that the prev_ids remain correct - if the server does come back. + that we don't fill up due to dead servers. + + Normally, we try to send device updates as a delta since a previous known point: + this is done by setting the prev_id in the m.device_list_update EDU. However, + for that to work, we have to have a complete record of each change to + each device, which can add up to quite a lot of data. + + An alternative mechanism is that, if the remote server sees that it has missed + an entry in the stream_id sequence for a given user, it will request a full + list of that user's devices. Hence, we can reduce the amount of data we have to + store (and transmit in some future transaction), by clearing almost everything + for a given destination out of the database, and having the remote server + resync. + + All we need to do is make sure we keep at least one row for each + (user, destination) pair, to remind us to send a m.device_list_update EDU for + that user when the destination comes back. It doesn't matter which device + we keep. """ - yesterday = self._clock.time_msec() - 24 * 60 * 60 * 1000 + yesterday = self._clock.time_msec() - prune_age def _prune_txn(txn): + # look for (user, destination) pairs which have an update older than + # the cutoff. + # + # For each pair, we also need to know the most recent stream_id, and + # an arbitrary device_id at that stream_id. select_sql = """ - SELECT destination, user_id, max(stream_id) as stream_id - FROM device_lists_outbound_pokes + SELECT + dlop1.destination, + dlop1.user_id, + MAX(dlop1.stream_id) AS stream_id, + (SELECT MIN(dlop2.device_id) AS device_id FROM + device_lists_outbound_pokes dlop2 + WHERE dlop2.destination = dlop1.destination AND + dlop2.user_id=dlop1.user_id AND + dlop2.stream_id=MAX(dlop1.stream_id) + ) + FROM device_lists_outbound_pokes dlop1 GROUP BY destination, user_id HAVING min(ts) < ? AND count(*) > 1 """ @@ -1114,14 +1144,29 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): if not rows: return + logger.info( + "Pruning old outbound device list updates for %i users/destinations: %s", + len(rows), + shortstr((row[0], row[1]) for row in rows), + ) + + # we want to keep the update with the highest stream_id for each user. + # + # there might be more than one update (with different device_ids) with the + # same stream_id, so we also delete all but one rows with the max stream id. delete_sql = """ DELETE FROM device_lists_outbound_pokes - WHERE ts < ? AND destination = ? AND user_id = ? AND stream_id < ? + WHERE destination = ? AND user_id = ? AND ( + stream_id < ? OR + (stream_id = ? AND device_id != ?) + ) """ - - txn.executemany( - delete_sql, ((yesterday, row[0], row[1], row[2]) for row in rows) - ) + count = 0 + for (destination, user_id, stream_id, device_id) in rows: + txn.execute( + delete_sql, (destination, user_id, stream_id, stream_id, device_id) + ) + count += txn.rowcount # Since we've deleted unsent deltas, we need to remove the entry # of last successful sent so that the prev_ids are correctly set. @@ -1131,7 +1176,7 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): """ txn.executemany(sql, ((row[0], row[1]) for row in rows)) - logger.info("Pruned %d device list outbound pokes", txn.rowcount) + logger.info("Pruned %d device list outbound pokes", count) return run_as_background_process( "prune_old_outbound_device_pokes", diff --git a/synapse/util/stringutils.py b/synapse/util/stringutils.py index 2c0dcb5208..6899bcb788 100644 --- a/synapse/util/stringutils.py +++ b/synapse/util/stringutils.py @@ -13,10 +13,11 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. - +import itertools import random import re import string +from collections import Iterable import six from six import PY2, PY3 @@ -126,3 +127,21 @@ def assert_valid_client_secret(client_secret): raise SynapseError( 400, "Invalid client_secret parameter", errcode=Codes.INVALID_PARAM ) + + +def shortstr(iterable: Iterable, maxitems: int = 5) -> str: + """If iterable has maxitems or fewer, return the stringification of a list + containing those items. + + Otherwise, return the stringification of a a list with the first maxitems items, + followed by "...". + + Args: + iterable: iterable to truncate + maxitems: number of items to return before truncating + """ + + items = list(itertools.islice(iterable, maxitems + 1)) + if len(items) <= maxitems: + return str(items) + return "[" + ", ".join(repr(r) for r in items[:maxitems]) + ", ...]" diff --git a/tests/federation/test_federation_sender.py b/tests/federation/test_federation_sender.py index 7763b12159..a5fe5c6880 100644 --- a/tests/federation/test_federation_sender.py +++ b/tests/federation/test_federation_sender.py @@ -370,6 +370,98 @@ class FederationSenderDevicesTestCases(HomeserverTestCase): devices = {edu["content"]["device_id"] for edu in self.edus} self.assertEqual({"D1", "D2", "D3"}, devices) + def test_prune_outbound_device_pokes1(self): + """If a destination is unreachable, and the updates are pruned, we should get + a single update. + + This case tests the behaviour when the server has never been reachable. + """ + mock_send_txn = self.hs.get_federation_transport_client().send_transaction + mock_send_txn.side_effect = lambda t, cb: defer.fail("fail") + + # create devices + u1 = self.register_user("user", "pass") + self.login("user", "pass", device_id="D1") + self.login("user", "pass", device_id="D2") + self.login("user", "pass", device_id="D3") + + # delete them again + self.get_success( + self.hs.get_device_handler().delete_devices(u1, ["D1", "D2", "D3"]) + ) + + self.assertGreaterEqual(mock_send_txn.call_count, 4) + + # run the prune job + self.reactor.advance(10) + self.get_success( + self.hs.get_datastore()._prune_old_outbound_device_pokes(prune_age=1) + ) + + # recover the server + mock_send_txn.side_effect = self.record_transaction + self.hs.get_federation_sender().send_device_messages("host2") + self.pump() + + # there should be a single update for this user. + self.assertEqual(len(self.edus), 1) + edu = self.edus.pop(0) + self.assertEqual(edu["edu_type"], "m.device_list_update") + c = edu["content"] + + # synapse uses an empty prev_id list to indicate "needs a full resync". + self.assertEqual(c["prev_id"], []) + + def test_prune_outbound_device_pokes2(self): + """If a destination is unreachable, and the updates are pruned, we should get + a single update. + + This case tests the behaviour when the server was reachable, but then goes + offline. + """ + + # create first device + u1 = self.register_user("user", "pass") + self.login("user", "pass", device_id="D1") + + # expect the update EDU + self.assertEqual(len(self.edus), 1) + self.check_device_update_edu(self.edus.pop(0), u1, "D1", None) + + # now the server goes offline + mock_send_txn = self.hs.get_federation_transport_client().send_transaction + mock_send_txn.side_effect = lambda t, cb: defer.fail("fail") + + self.login("user", "pass", device_id="D2") + self.login("user", "pass", device_id="D3") + + # delete them again + self.get_success( + self.hs.get_device_handler().delete_devices(u1, ["D1", "D2", "D3"]) + ) + + self.assertGreaterEqual(mock_send_txn.call_count, 3) + + # run the prune job + self.reactor.advance(10) + self.get_success( + self.hs.get_datastore()._prune_old_outbound_device_pokes(prune_age=1) + ) + + # recover the server + mock_send_txn.side_effect = self.record_transaction + self.hs.get_federation_sender().send_device_messages("host2") + self.pump() + + # ... and we should get a single update for this user. + self.assertEqual(len(self.edus), 1) + edu = self.edus.pop(0) + self.assertEqual(edu["edu_type"], "m.device_list_update") + c = edu["content"] + + # synapse uses an empty prev_id list to indicate "needs a full resync". + self.assertEqual(c["prev_id"], []) + def check_device_update_edu( self, edu: JsonDict, -- cgit 1.4.1 From cfe8c8ab8e412b6320e5963ced0670fbc7b00d1b Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Tue, 31 Mar 2020 17:24:06 +0100 Subject: Remove unused `start_background_update` This was only used in a unit test, so let's just inline it in the test. --- synapse/storage/background_updates.py | 21 --------------------- tests/storage/test_background_update.py | 14 +++++++++----- 2 files changed, 9 insertions(+), 26 deletions(-) (limited to 'tests') diff --git a/synapse/storage/background_updates.py b/synapse/storage/background_updates.py index eb1a7e5002..d4e26eab6c 100644 --- a/synapse/storage/background_updates.py +++ b/synapse/storage/background_updates.py @@ -400,27 +400,6 @@ class BackgroundUpdater(object): self.register_background_update_handler(update_name, updater) - def start_background_update(self, update_name, progress): - """Starts a background update running. - - Args: - update_name: The update to set running. - progress: The initial state of the progress of the update. - - Returns: - A deferred that completes once the task has been added to the - queue. - """ - # Clear the background update queue so that we will pick up the new - # task on the next iteration of do_background_update. - self._background_update_queue = [] - progress_json = json.dumps(progress) - - return self.db.simple_insert( - "background_updates", - {"update_name": update_name, "progress_json": progress_json}, - ) - def _end_background_update(self, update_name): """Removes a completed background update task from the queue. diff --git a/tests/storage/test_background_update.py b/tests/storage/test_background_update.py index ae14fb407d..aca41eb215 100644 --- a/tests/storage/test_background_update.py +++ b/tests/storage/test_background_update.py @@ -25,12 +25,20 @@ class BackgroundUpdateTestCase(unittest.HomeserverTestCase): # the target runtime for each bg update target_background_update_duration_ms = 50000 + store = self.hs.get_datastore() + self.get_success( + store.db.simple_insert( + "background_updates", + values={"update_name": "test_update", "progress_json": '{"my_key": 1}'}, + ) + ) + # first step: make a bit of progress @defer.inlineCallbacks def update(progress, count): yield self.clock.sleep((count * duration_ms) / 1000) progress = {"my_key": progress["my_key"] + 1} - yield self.hs.get_datastore().db.runInteraction( + yield store.db.runInteraction( "update_progress", self.updates._background_update_progress_txn, "test_update", @@ -39,10 +47,6 @@ class BackgroundUpdateTestCase(unittest.HomeserverTestCase): return count self.update_handler.side_effect = update - - self.get_success( - self.updates.start_background_update("test_update", {"my_key": 1}) - ) self.update_handler.reset_mock() res = self.get_success( self.updates.do_next_background_update( -- cgit 1.4.1 From 26d17b9bdc0de51d5f1a7526e8ab70e7f7796e4d Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Tue, 31 Mar 2020 17:25:10 +0100 Subject: Make `has_completed_background_updates` async (Almost) everywhere that uses it is happy with an awaitable. --- synapse/storage/background_updates.py | 7 +++---- tests/storage/test_background_update.py | 4 +++- 2 files changed, 6 insertions(+), 5 deletions(-) (limited to 'tests') diff --git a/synapse/storage/background_updates.py b/synapse/storage/background_updates.py index d4e26eab6c..79494bcdf5 100644 --- a/synapse/storage/background_updates.py +++ b/synapse/storage/background_updates.py @@ -119,12 +119,11 @@ class BackgroundUpdater(object): self._all_done = True return None - @defer.inlineCallbacks - def has_completed_background_updates(self): + async def has_completed_background_updates(self) -> bool: """Check if all the background updates have completed Returns: - Deferred[bool]: True if all background updates have completed + True if all background updates have completed """ # if we've previously determined that there is nothing left to do, that # is easy @@ -138,7 +137,7 @@ class BackgroundUpdater(object): # otherwise, check if there are updates to be run. This is important, # as we may be running on a worker which doesn't perform the bg updates # itself, but still wants to wait for them to happen. - updates = yield self.db.simple_select_onecol( + updates = await self.db.simple_select_onecol( "background_updates", keyvalues=None, retcol="1", diff --git a/tests/storage/test_background_update.py b/tests/storage/test_background_update.py index aca41eb215..e578de8acf 100644 --- a/tests/storage/test_background_update.py +++ b/tests/storage/test_background_update.py @@ -11,7 +11,9 @@ class BackgroundUpdateTestCase(unittest.HomeserverTestCase): def prepare(self, reactor, clock, homeserver): self.updates = self.hs.get_datastore().db.updates # type: BackgroundUpdater # the base test class should have run the real bg updates for us - self.assertTrue(self.updates.has_completed_background_updates()) + self.assertTrue( + self.get_success(self.updates.has_completed_background_updates()) + ) self.update_handler = Mock() self.updates.register_background_update_handler( -- cgit 1.4.1 From 51f4d52cb4663a056372d779b78488aeae45f554 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Tue, 31 Mar 2020 17:27:56 +0100 Subject: Set a logging context while running the bg updates This mostly just reduces the amount of "running from sentinel context" spam during unittest setup. --- tests/unittest.py | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-) (limited to 'tests') diff --git a/tests/unittest.py b/tests/unittest.py index d0406ca2fd..27af5228fe 100644 --- a/tests/unittest.py +++ b/tests/unittest.py @@ -40,6 +40,7 @@ from synapse.http.server import JsonResource from synapse.http.site import SynapseRequest, SynapseSite from synapse.logging.context import ( SENTINEL_CONTEXT, + LoggingContext, current_context, set_current_context, ) @@ -419,15 +420,17 @@ class HomeserverTestCase(TestCase): config_obj.parse_config_dict(config, "", "") kwargs["config"] = config_obj + async def run_bg_updates(): + with LoggingContext("run_bg_updates", request="run_bg_updates-1"): + while not await stor.db.updates.has_completed_background_updates(): + await stor.db.updates.do_next_background_update(1) + hs = setup_test_homeserver(self.addCleanup, *args, **kwargs) stor = hs.get_datastore() # Run the database background updates, when running against "master". if hs.__class__.__name__ == "TestHomeServer": - while not self.get_success( - stor.db.updates.has_completed_background_updates() - ): - self.get_success(stor.db.updates.do_next_background_update(1)) + self.get_success(run_bg_updates()) return hs -- cgit 1.4.1 From b4c22342320d7de86c02dfb36415a38c62bec88d Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Tue, 31 Mar 2020 17:31:32 +0100 Subject: Make do_next_background_update return a bool returning a None or an int that we don't use is confusing. --- synapse/storage/background_updates.py | 12 +++++------- tests/storage/test_background_update.py | 6 +++--- 2 files changed, 8 insertions(+), 10 deletions(-) (limited to 'tests') diff --git a/synapse/storage/background_updates.py b/synapse/storage/background_updates.py index 79494bcdf5..4a59132bf3 100644 --- a/synapse/storage/background_updates.py +++ b/synapse/storage/background_updates.py @@ -111,7 +111,7 @@ class BackgroundUpdater(object): except Exception: logger.exception("Error doing update") else: - if result is None: + if result: logger.info( "No more background updates to do." " Unscheduling background update task." @@ -169,9 +169,7 @@ class BackgroundUpdater(object): return not update_exists - async def do_next_background_update( - self, desired_duration_ms: float - ) -> Optional[int]: + async def do_next_background_update(self, desired_duration_ms: float) -> bool: """Does some amount of work on the next queued background update Returns once some amount of work is done. @@ -180,7 +178,7 @@ class BackgroundUpdater(object): desired_duration_ms(float): How long we want to spend updating. Returns: - None if there is no more work to do, otherwise an int + True if there is no more work to do, otherwise False """ if not self._background_update_queue: updates = await self.db.simple_select_list( @@ -195,14 +193,14 @@ class BackgroundUpdater(object): if not self._background_update_queue: # no work left to do - return None + return True # pop from the front, and add back to the back update_name = self._background_update_queue.pop(0) self._background_update_queue.append(update_name) res = await self._do_background_update(update_name, desired_duration_ms) - return res + return False async def _do_background_update( self, update_name: str, desired_duration_ms: float diff --git a/tests/storage/test_background_update.py b/tests/storage/test_background_update.py index e578de8acf..940b166129 100644 --- a/tests/storage/test_background_update.py +++ b/tests/storage/test_background_update.py @@ -56,7 +56,7 @@ class BackgroundUpdateTestCase(unittest.HomeserverTestCase): ), by=0.1, ) - self.assertIsNotNone(res) + self.assertFalse(res) # on the first call, we should get run with the default background update size self.update_handler.assert_called_once_with( @@ -79,7 +79,7 @@ class BackgroundUpdateTestCase(unittest.HomeserverTestCase): result = self.get_success( self.updates.do_next_background_update(target_background_update_duration_ms) ) - self.assertIsNotNone(result) + self.assertFalse(result) self.update_handler.assert_called_once() # third step: we don't expect to be called any more @@ -87,5 +87,5 @@ class BackgroundUpdateTestCase(unittest.HomeserverTestCase): result = self.get_success( self.updates.do_next_background_update(target_background_update_duration_ms) ) - self.assertIsNone(result) + self.assertTrue(result) self.assertFalse(self.update_handler.called) -- cgit 1.4.1 From 468dcc767bf379ba2b4ed4b6d1c6537473175eab Mon Sep 17 00:00:00 2001 From: Patrick Cloke Date: Wed, 1 Apr 2020 08:27:05 -0400 Subject: Allow admins to create aliases when they are not in the room (#7191) --- changelog.d/7191.feature | 1 + synapse/handlers/directory.py | 6 +++- tests/handlers/test_directory.py | 62 ++++++++++++++++++++++++++++++++++++++++ 3 files changed, 68 insertions(+), 1 deletion(-) create mode 100644 changelog.d/7191.feature (limited to 'tests') diff --git a/changelog.d/7191.feature b/changelog.d/7191.feature new file mode 100644 index 0000000000..83d5685bb2 --- /dev/null +++ b/changelog.d/7191.feature @@ -0,0 +1 @@ +Admin users are no longer required to be in a room to create an alias for it. diff --git a/synapse/handlers/directory.py b/synapse/handlers/directory.py index 1d842c369b..53e5f585d9 100644 --- a/synapse/handlers/directory.py +++ b/synapse/handlers/directory.py @@ -127,7 +127,11 @@ class DirectoryHandler(BaseHandler): errcode=Codes.EXCLUSIVE, ) else: - if self.require_membership and check_membership: + # Server admins are not subject to the same constraints as normal + # users when creating an alias (e.g. being in the room). + is_admin = yield self.auth.is_server_admin(requester.user) + + if (self.require_membership and check_membership) and not is_admin: rooms_for_user = yield self.store.get_rooms_for_user(user_id) if room_id not in rooms_for_user: raise AuthError( diff --git a/tests/handlers/test_directory.py b/tests/handlers/test_directory.py index 5e40adba52..00bb776271 100644 --- a/tests/handlers/test_directory.py +++ b/tests/handlers/test_directory.py @@ -102,6 +102,68 @@ class DirectoryTestCase(unittest.HomeserverTestCase): self.assertEquals({"room_id": "!8765asdf:test", "servers": ["test"]}, response) +class TestCreateAlias(unittest.HomeserverTestCase): + servlets = [ + synapse.rest.admin.register_servlets, + login.register_servlets, + room.register_servlets, + directory.register_servlets, + ] + + def prepare(self, reactor, clock, hs): + self.handler = hs.get_handlers().directory_handler + + # Create user + self.admin_user = self.register_user("admin", "pass", admin=True) + self.admin_user_tok = self.login("admin", "pass") + + # Create a test room + self.room_id = self.helper.create_room_as( + self.admin_user, tok=self.admin_user_tok + ) + + self.test_alias = "#test:test" + self.room_alias = RoomAlias.from_string(self.test_alias) + + # Create a test user. + self.test_user = self.register_user("user", "pass", admin=False) + self.test_user_tok = self.login("user", "pass") + self.helper.join(room=self.room_id, user=self.test_user, tok=self.test_user_tok) + + def test_create_alias_joined_room(self): + """A user can create an alias for a room they're in.""" + self.get_success( + self.handler.create_association( + create_requester(self.test_user), self.room_alias, self.room_id, + ) + ) + + def test_create_alias_other_room(self): + """A user cannot create an alias for a room they're NOT in.""" + other_room_id = self.helper.create_room_as( + self.admin_user, tok=self.admin_user_tok + ) + + self.get_failure( + self.handler.create_association( + create_requester(self.test_user), self.room_alias, other_room_id, + ), + synapse.api.errors.SynapseError, + ) + + def test_create_alias_admin(self): + """An admin can create an alias for a room they're NOT in.""" + other_room_id = self.helper.create_room_as( + self.test_user, tok=self.test_user_tok + ) + + self.get_success( + self.handler.create_association( + create_requester(self.admin_user), self.room_alias, other_room_id, + ) + ) + + class TestDeleteAlias(unittest.HomeserverTestCase): servlets = [ synapse.rest.admin.register_servlets, -- cgit 1.4.1 From daa1ac89a0be4dd3cc941da4caeb2ddcbd701eff Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Fri, 3 Apr 2020 10:40:22 +0100 Subject: Fix device list update stream ids going backward (#7158) Occasionally we could get a federation device list update transaction which looked like: ``` [ {'edu_type': 'm.device_list_update', 'content': {'user_id': '@user:test', 'device_id': 'D2', 'prev_id': [], 'stream_id': 12, 'deleted': True}}, {'edu_type': 'm.device_list_update', 'content': {'user_id': '@user:test', 'device_id': 'D1', 'prev_id': [12], 'stream_id': 11, 'deleted': True}}, {'edu_type': 'm.device_list_update', 'content': {'user_id': '@user:test', 'device_id': 'D3', 'prev_id': [11], 'stream_id': 13, 'deleted': True}} ] ``` Having `stream_ids` which are lower than `prev_ids` looks odd. It might work (I'm not actually sure), but in any case it doesn't seem like a reasonable thing to expect other implementations to support. --- changelog.d/7158.misc | 1 + synapse/storage/data_stores/main/devices.py | 10 ++++++++-- tests/federation/test_federation_sender.py | 6 ++++++ 3 files changed, 15 insertions(+), 2 deletions(-) create mode 100644 changelog.d/7158.misc (limited to 'tests') diff --git a/changelog.d/7158.misc b/changelog.d/7158.misc new file mode 100644 index 0000000000..269b8daeb0 --- /dev/null +++ b/changelog.d/7158.misc @@ -0,0 +1 @@ +Fix device list update stream ids going backward. diff --git a/synapse/storage/data_stores/main/devices.py b/synapse/storage/data_stores/main/devices.py index 20995e1b78..dd3561e9b2 100644 --- a/synapse/storage/data_stores/main/devices.py +++ b/synapse/storage/data_stores/main/devices.py @@ -165,7 +165,6 @@ class DeviceWorkerStore(SQLBaseStore): # the max stream_id across each set of duplicate entries # # maps (user_id, device_id) -> (stream_id, opentracing_context) - # as long as their stream_id does not match that of the last row # # opentracing_context contains the opentracing metadata for the request # that created the poke @@ -270,7 +269,14 @@ class DeviceWorkerStore(SQLBaseStore): prev_id = yield self._get_last_device_update_for_remote_user( destination, user_id, from_stream_id ) - for device_id, device in iteritems(user_devices): + + # make sure we go through the devices in stream order + device_ids = sorted( + user_devices.keys(), key=lambda i: query_map[(user_id, i)][0], + ) + + for device_id in device_ids: + device = user_devices[device_id] stream_id, opentracing_context = query_map[(user_id, device_id)] result = { "user_id": user_id, diff --git a/tests/federation/test_federation_sender.py b/tests/federation/test_federation_sender.py index a5fe5c6880..33105576af 100644 --- a/tests/federation/test_federation_sender.py +++ b/tests/federation/test_federation_sender.py @@ -297,6 +297,7 @@ class FederationSenderDevicesTestCases(HomeserverTestCase): c = edu["content"] if stream_id is not None: self.assertEqual(c["prev_id"], [stream_id]) + self.assertGreaterEqual(c["stream_id"], stream_id) stream_id = c["stream_id"] devices = {edu["content"]["device_id"] for edu in self.edus} self.assertEqual({"D1", "D2"}, devices) @@ -330,6 +331,7 @@ class FederationSenderDevicesTestCases(HomeserverTestCase): c.items(), {"user_id": u1, "prev_id": [stream_id], "deleted": True}.items(), ) + self.assertGreaterEqual(c["stream_id"], stream_id) stream_id = c["stream_id"] devices = {edu["content"]["device_id"] for edu in self.edus} self.assertEqual({"D1", "D2", "D3"}, devices) @@ -366,6 +368,8 @@ class FederationSenderDevicesTestCases(HomeserverTestCase): self.assertEqual(edu["edu_type"], "m.device_list_update") c = edu["content"] self.assertEqual(c["prev_id"], [stream_id] if stream_id is not None else []) + if stream_id is not None: + self.assertGreaterEqual(c["stream_id"], stream_id) stream_id = c["stream_id"] devices = {edu["content"]["device_id"] for edu in self.edus} self.assertEqual({"D1", "D2", "D3"}, devices) @@ -482,6 +486,8 @@ class FederationSenderDevicesTestCases(HomeserverTestCase): } self.assertLessEqual(expected.items(), content.items()) + if prev_stream_id is not None: + self.assertGreaterEqual(content["stream_id"], prev_stream_id) return content["stream_id"] def check_signing_key_update_txn(self, txn: JsonDict,) -> None: -- cgit 1.4.1 From d73bf18d13031d9f9c0375b83f2cc5ff6f415251 Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Sat, 4 Apr 2020 17:27:45 +0200 Subject: Server notices: Dissociate room creation/lookup from invite (#7199) Fixes #6815 Before figuring out whether we should alert a user on MAU, we call get_notice_room_for_user to get some info on the existing server notices room for this user. This function, if the room doesn't exist, creates it and invites the user in it. This means that, if we decide later that no server notice is needed, the user gets invited in a room with no message in it. This happens at every restart of the server, since the room ID returned by get_notice_room_for_user is cached. This PR fixes that by moving the inviting bit to a dedicated function, that's only called when the server actually needs to send a notice to the user. A potential issue with this approach is that the room that's created by get_notice_room_for_user doesn't match how that same function looks for an existing room (i.e. it creates a room that doesn't have an invite or a join for the current user in it, so it could lead to a new room being created each time a user syncs), but I'm not sure this is a problem given it's cached until the server restarts, so that function won't run very often. It also renames get_notice_room_for_user into get_or_create_notice_room_for_user to make what it does clearer. --- changelog.d/7199.bugfix | 1 + .../resource_limits_server_notices.py | 4 +- synapse/server_notices/server_notices_manager.py | 51 +++++++-- .../test_resource_limits_server_notices.py | 120 ++++++++++++++++++--- 4 files changed, 154 insertions(+), 22 deletions(-) create mode 100644 changelog.d/7199.bugfix (limited to 'tests') diff --git a/changelog.d/7199.bugfix b/changelog.d/7199.bugfix new file mode 100644 index 0000000000..b234163ea8 --- /dev/null +++ b/changelog.d/7199.bugfix @@ -0,0 +1 @@ +Fix a bug that could cause a user to be invited to a server notices (aka System Alerts) room without any notice being sent. diff --git a/synapse/server_notices/resource_limits_server_notices.py b/synapse/server_notices/resource_limits_server_notices.py index 9fae2e0afe..ce4a828894 100644 --- a/synapse/server_notices/resource_limits_server_notices.py +++ b/synapse/server_notices/resource_limits_server_notices.py @@ -80,7 +80,9 @@ class ResourceLimitsServerNotices(object): # In practice, not sure we can ever get here return - room_id = yield self._server_notices_manager.get_notice_room_for_user(user_id) + room_id = yield self._server_notices_manager.get_or_create_notice_room_for_user( + user_id + ) if not room_id: logger.warning("Failed to get server notices room") diff --git a/synapse/server_notices/server_notices_manager.py b/synapse/server_notices/server_notices_manager.py index f7432c8d2f..bf0943f265 100644 --- a/synapse/server_notices/server_notices_manager.py +++ b/synapse/server_notices/server_notices_manager.py @@ -17,7 +17,7 @@ import logging from twisted.internet import defer from synapse.api.constants import EventTypes, Membership, RoomCreationPreset -from synapse.types import create_requester +from synapse.types import UserID, create_requester from synapse.util.caches.descriptors import cachedInlineCallbacks logger = logging.getLogger(__name__) @@ -36,10 +36,12 @@ class ServerNoticesManager(object): self._store = hs.get_datastore() self._config = hs.config self._room_creation_handler = hs.get_room_creation_handler() + self._room_member_handler = hs.get_room_member_handler() self._event_creation_handler = hs.get_event_creation_handler() self._is_mine_id = hs.is_mine_id self._notifier = hs.get_notifier() + self.server_notices_mxid = self._config.server_notices_mxid def is_enabled(self): """Checks if server notices are enabled on this server. @@ -66,7 +68,8 @@ class ServerNoticesManager(object): Returns: Deferred[FrozenEvent] """ - room_id = yield self.get_notice_room_for_user(user_id) + room_id = yield self.get_or_create_notice_room_for_user(user_id) + yield self.maybe_invite_user_to_room(user_id, room_id) system_mxid = self._config.server_notices_mxid requester = create_requester(system_mxid) @@ -89,10 +92,11 @@ class ServerNoticesManager(object): return res @cachedInlineCallbacks() - def get_notice_room_for_user(self, user_id): + def get_or_create_notice_room_for_user(self, user_id): """Get the room for notices for a given user - If we have not yet created a notice room for this user, create it + If we have not yet created a notice room for this user, create it, but don't + invite the user to it. Args: user_id (str): complete user id for the user we want a room for @@ -108,7 +112,6 @@ class ServerNoticesManager(object): rooms = yield self._store.get_rooms_for_local_user_where_membership_is( user_id, [Membership.INVITE, Membership.JOIN] ) - system_mxid = self._config.server_notices_mxid for room in rooms: # it's worth noting that there is an asymmetry here in that we # expect the user to be invited or joined, but the system user must @@ -116,10 +119,14 @@ class ServerNoticesManager(object): # manages to invite the system user to a room, that doesn't make it # the server notices room. user_ids = yield self._store.get_users_in_room(room.room_id) - if system_mxid in user_ids: + if self.server_notices_mxid in user_ids: # we found a room which our user shares with the system notice # user - logger.info("Using room %s", room.room_id) + logger.info( + "Using existing server notices room %s for user %s", + room.room_id, + user_id, + ) return room.room_id # apparently no existing notice room: create a new one @@ -138,14 +145,13 @@ class ServerNoticesManager(object): "avatar_url": self._config.server_notices_mxid_avatar_url, } - requester = create_requester(system_mxid) + requester = create_requester(self.server_notices_mxid) info = yield self._room_creation_handler.create_room( requester, config={ "preset": RoomCreationPreset.PRIVATE_CHAT, "name": self._config.server_notices_room_name, "power_level_content_override": {"users_default": -10}, - "invite": (user_id,), }, ratelimit=False, creator_join_profile=join_profile, @@ -159,3 +165,30 @@ class ServerNoticesManager(object): logger.info("Created server notices room %s for %s", room_id, user_id) return room_id + + @defer.inlineCallbacks + def maybe_invite_user_to_room(self, user_id: str, room_id: str): + """Invite the given user to the given server room, unless the user has already + joined or been invited to it. + + Args: + user_id: The ID of the user to invite. + room_id: The ID of the room to invite the user to. + """ + requester = create_requester(self.server_notices_mxid) + + # Check whether the user has already joined or been invited to this room. If + # that's the case, there is no need to re-invite them. + joined_rooms = yield self._store.get_rooms_for_local_user_where_membership_is( + user_id, [Membership.INVITE, Membership.JOIN] + ) + for room in joined_rooms: + if room.room_id == room_id: + return + + yield self._room_member_handler.update_membership( + requester=requester, + target=UserID.from_string(user_id), + room_id=room_id, + action="invite", + ) diff --git a/tests/server_notices/test_resource_limits_server_notices.py b/tests/server_notices/test_resource_limits_server_notices.py index 0d27b92a86..93eb053b8c 100644 --- a/tests/server_notices/test_resource_limits_server_notices.py +++ b/tests/server_notices/test_resource_limits_server_notices.py @@ -19,6 +19,9 @@ from twisted.internet import defer from synapse.api.constants import EventTypes, LimitBlockingTypes, ServerNoticeMsgType from synapse.api.errors import ResourceLimitError +from synapse.rest import admin +from synapse.rest.client.v1 import login, room +from synapse.rest.client.v2_alpha import sync from synapse.server_notices.resource_limits_server_notices import ( ResourceLimitsServerNotices, ) @@ -67,7 +70,7 @@ class TestResourceLimitsServerNotices(unittest.HomeserverTestCase): # self.server_notices_mxid_avatar_url = None # self.server_notices_room_name = "Server Notices" - self._rlsn._server_notices_manager.get_notice_room_for_user = Mock( + self._rlsn._server_notices_manager.get_or_create_notice_room_for_user = Mock( returnValue="" ) self._rlsn._store.add_tag_to_room = Mock() @@ -215,6 +218,26 @@ class TestResourceLimitsServerNotices(unittest.HomeserverTestCase): class TestResourceLimitsServerNoticesWithRealRooms(unittest.HomeserverTestCase): + servlets = [ + admin.register_servlets, + login.register_servlets, + room.register_servlets, + sync.register_servlets, + ] + + def default_config(self): + c = super().default_config() + c["server_notices"] = { + "system_mxid_localpart": "server", + "system_mxid_display_name": None, + "system_mxid_avatar_url": None, + "room_name": "Test Server Notice Room", + } + c["limit_usage_by_mau"] = True + c["max_mau_value"] = 5 + c["admin_contact"] = "mailto:user@test.com" + return c + def prepare(self, reactor, clock, hs): self.store = self.hs.get_datastore() self.server_notices_sender = self.hs.get_server_notices_sender() @@ -228,18 +251,8 @@ class TestResourceLimitsServerNoticesWithRealRooms(unittest.HomeserverTestCase): if not isinstance(self._rlsn, ResourceLimitsServerNotices): raise Exception("Failed to find reference to ResourceLimitsServerNotices") - self.hs.config.limit_usage_by_mau = True - self.hs.config.hs_disabled = False - self.hs.config.max_mau_value = 5 - self.hs.config.server_notices_mxid = "@server:test" - self.hs.config.server_notices_mxid_display_name = None - self.hs.config.server_notices_mxid_avatar_url = None - self.hs.config.server_notices_room_name = "Test Server Notice Room" - self.user_id = "@user_id:test" - self.hs.config.admin_contact = "mailto:user@test.com" - def test_server_notice_only_sent_once(self): self.store.get_monthly_active_count = Mock(return_value=1000) @@ -253,7 +266,7 @@ class TestResourceLimitsServerNoticesWithRealRooms(unittest.HomeserverTestCase): # Now lets get the last load of messages in the service notice room and # check that there is only one server notice room_id = self.get_success( - self.server_notices_manager.get_notice_room_for_user(self.user_id) + self.server_notices_manager.get_or_create_notice_room_for_user(self.user_id) ) token = self.get_success(self.event_source.get_current_token()) @@ -273,3 +286,86 @@ class TestResourceLimitsServerNoticesWithRealRooms(unittest.HomeserverTestCase): count += 1 self.assertEqual(count, 1) + + def test_no_invite_without_notice(self): + """Tests that a user doesn't get invited to a server notices room without a + server notice being sent. + + The scenario for this test is a single user on a server where the MAU limit + hasn't been reached (since it's the only user and the limit is 5), so users + shouldn't receive a server notice. + """ + self.register_user("user", "password") + tok = self.login("user", "password") + + request, channel = self.make_request("GET", "/sync?timeout=0", access_token=tok) + self.render(request) + + invites = channel.json_body["rooms"]["invite"] + self.assertEqual(len(invites), 0, invites) + + def test_invite_with_notice(self): + """Tests that, if the MAU limit is hit, the server notices user invites each user + to a room in which it has sent a notice. + """ + user_id, tok, room_id = self._trigger_notice_and_join() + + # Sync again to retrieve the events in the room, so we can check whether this + # room has a notice in it. + request, channel = self.make_request("GET", "/sync?timeout=0", access_token=tok) + self.render(request) + + # Scan the events in the room to search for a message from the server notices + # user. + events = channel.json_body["rooms"]["join"][room_id]["timeline"]["events"] + notice_in_room = False + for event in events: + if ( + event["type"] == EventTypes.Message + and event["sender"] == self.hs.config.server_notices_mxid + ): + notice_in_room = True + + self.assertTrue(notice_in_room, "No server notice in room") + + def _trigger_notice_and_join(self): + """Creates enough active users to hit the MAU limit and trigger a system notice + about it, then joins the system notices room with one of the users created. + + Returns: + user_id (str): The ID of the user that joined the room. + tok (str): The access token of the user that joined the room. + room_id (str): The ID of the room that's been joined. + """ + user_id = None + tok = None + invites = [] + + # Register as many users as the MAU limit allows. + for i in range(self.hs.config.max_mau_value): + localpart = "user%d" % i + user_id = self.register_user(localpart, "password") + tok = self.login(localpart, "password") + + # Sync with the user's token to mark the user as active. + request, channel = self.make_request( + "GET", "/sync?timeout=0", access_token=tok, + ) + self.render(request) + + # Also retrieves the list of invites for this user. We don't care about that + # one except if we're processing the last user, which should have received an + # invite to a room with a server notice about the MAU limit being reached. + # We could also pick another user and sync with it, which would return an + # invite to a system notices room, but it doesn't matter which user we're + # using so we use the last one because it saves us an extra sync. + invites = channel.json_body["rooms"]["invite"] + + # Make sure we have an invite to process. + self.assertEqual(len(invites), 1, invites) + + # Join the room. + room_id = list(invites.keys())[0] + self.helper.join(room=room_id, user=user_id, tok=tok) + + return user_id, tok, room_id -- cgit 1.4.1 From 5016b162fcf0372fe35404c64f80aeaf21461f31 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Mon, 6 Apr 2020 09:58:42 +0100 Subject: Move client command handling out of TCP protocol (#7185) The aim here is to move the command handling out of the TCP protocol classes and to also merge the client and server command handling (so that we can reuse them for redis protocol). This PR simply moves the client paths to the new `ReplicationCommandHandler`, a future PR will move the server paths too. --- changelog.d/7185.misc | 1 + synapse/app/admin_cmd.py | 12 -- synapse/app/generic_worker.py | 9 +- synapse/replication/tcp/__init__.py | 30 ++- synapse/replication/tcp/client.py | 179 +++--------------- synapse/replication/tcp/handler.py | 252 +++++++++++++++++++++++++ synapse/replication/tcp/protocol.py | 197 +++---------------- synapse/server.py | 8 +- synapse/server.pyi | 7 +- tests/replication/slave/storage/_base.py | 15 +- tests/replication/tcp/streams/_base.py | 38 ++-- tests/replication/tcp/streams/test_receipts.py | 1 - 12 files changed, 378 insertions(+), 371 deletions(-) create mode 100644 changelog.d/7185.misc create mode 100644 synapse/replication/tcp/handler.py (limited to 'tests') diff --git a/changelog.d/7185.misc b/changelog.d/7185.misc new file mode 100644 index 0000000000..deb9ca7021 --- /dev/null +++ b/changelog.d/7185.misc @@ -0,0 +1 @@ +Move client command handling out of TCP protocol. diff --git a/synapse/app/admin_cmd.py b/synapse/app/admin_cmd.py index 1c7c6ec0c8..a37818fe9a 100644 --- a/synapse/app/admin_cmd.py +++ b/synapse/app/admin_cmd.py @@ -43,7 +43,6 @@ from synapse.replication.slave.storage.push_rule import SlavedPushRuleStore from synapse.replication.slave.storage.receipts import SlavedReceiptsStore from synapse.replication.slave.storage.registration import SlavedRegistrationStore from synapse.replication.slave.storage.room import RoomStore -from synapse.replication.tcp.client import ReplicationClientHandler from synapse.server import HomeServer from synapse.util.logcontext import LoggingContext from synapse.util.versionstring import get_version_string @@ -79,17 +78,6 @@ class AdminCmdServer(HomeServer): def start_listening(self, listeners): pass - def build_tcp_replication(self): - return AdminCmdReplicationHandler(self) - - -class AdminCmdReplicationHandler(ReplicationClientHandler): - async def on_rdata(self, stream_name, token, rows): - pass - - def get_streams_to_replicate(self): - return {} - @defer.inlineCallbacks def export_data_command(hs, args): diff --git a/synapse/app/generic_worker.py b/synapse/app/generic_worker.py index 174bef360f..dcd0709a02 100644 --- a/synapse/app/generic_worker.py +++ b/synapse/app/generic_worker.py @@ -64,7 +64,7 @@ from synapse.replication.slave.storage.receipts import SlavedReceiptsStore from synapse.replication.slave.storage.registration import SlavedRegistrationStore from synapse.replication.slave.storage.room import RoomStore from synapse.replication.slave.storage.transactions import SlavedTransactionStore -from synapse.replication.tcp.client import ReplicationClientHandler +from synapse.replication.tcp.client import ReplicationDataHandler from synapse.replication.tcp.commands import ClearUserSyncsCommand from synapse.replication.tcp.streams import ( AccountDataStream, @@ -603,7 +603,7 @@ class GenericWorkerServer(HomeServer): def remove_pusher(self, app_id, push_key, user_id): self.get_tcp_replication().send_remove_pusher(app_id, push_key, user_id) - def build_tcp_replication(self): + def build_replication_data_handler(self): return GenericWorkerReplicationHandler(self) def build_presence_handler(self): @@ -613,7 +613,7 @@ class GenericWorkerServer(HomeServer): return GenericWorkerTyping(self) -class GenericWorkerReplicationHandler(ReplicationClientHandler): +class GenericWorkerReplicationHandler(ReplicationDataHandler): def __init__(self, hs): super(GenericWorkerReplicationHandler, self).__init__(hs.get_datastore()) @@ -644,9 +644,6 @@ class GenericWorkerReplicationHandler(ReplicationClientHandler): args.update(self.send_handler.stream_positions()) return args - def get_currently_syncing_users(self): - return self.presence_handler.get_currently_syncing_users() - async def process_and_notify(self, stream_name, token, rows): try: if self.send_handler: diff --git a/synapse/replication/tcp/__init__.py b/synapse/replication/tcp/__init__.py index 81c2ea7ee9..523a1358d4 100644 --- a/synapse/replication/tcp/__init__.py +++ b/synapse/replication/tcp/__init__.py @@ -20,11 +20,31 @@ Further details can be found in docs/tcp_replication.rst Structure of the module: - * client.py - the client classes used for workers to connect to master + * handler.py - the classes used to handle sending/receiving commands to + replication * command.py - the definitions of all the valid commands - * protocol.py - contains bot the client and server protocol implementations, - these should not be used directly - * resource.py - the server classes that accepts and handle client connections - * streams.py - the definitons of all the valid streams + * protocol.py - the TCP protocol classes + * resource.py - handles streaming stream updates to replications + * streams/ - the definitons of all the valid streams + +The general interaction of the classes are: + + +---------------------+ + | ReplicationStreamer | + +---------------------+ + | + v + +---------------------------+ +----------------------+ + | ReplicationCommandHandler |---->|ReplicationDataHandler| + +---------------------------+ +----------------------+ + | ^ + v | + +-------------+ + | Protocols | + | (TCP/redis) | + +-------------+ + +Where the ReplicationDataHandler (or subclasses) handles incoming stream +updates. """ diff --git a/synapse/replication/tcp/client.py b/synapse/replication/tcp/client.py index e86d9805f1..700ae79158 100644 --- a/synapse/replication/tcp/client.py +++ b/synapse/replication/tcp/client.py @@ -16,26 +16,16 @@ """ import logging -from typing import Dict, List, Optional +from typing import TYPE_CHECKING, Dict -from twisted.internet import defer from twisted.internet.protocol import ReconnectingClientFactory from synapse.replication.slave.storage._base import BaseSlavedStore -from synapse.replication.tcp.protocol import ( - AbstractReplicationClientHandler, - ClientReplicationStreamProtocol, -) - -from .commands import ( - Command, - FederationAckCommand, - InvalidateCacheCommand, - RemoteServerUpCommand, - RemovePusherCommand, - UserIpCommand, - UserSyncCommand, -) +from synapse.replication.tcp.protocol import ClientReplicationStreamProtocol + +if TYPE_CHECKING: + from synapse.server import HomeServer + from synapse.replication.tcp.handler import ReplicationCommandHandler logger = logging.getLogger(__name__) @@ -44,16 +34,20 @@ class ReplicationClientFactory(ReconnectingClientFactory): """Factory for building connections to the master. Will reconnect if the connection is lost. - Accepts a handler that will be called when new data is available or data - is required. + Accepts a handler that is passed to `ClientReplicationStreamProtocol`. """ initialDelay = 0.1 maxDelay = 1 # Try at least once every N seconds - def __init__(self, hs, client_name, handler: AbstractReplicationClientHandler): + def __init__( + self, + hs: "HomeServer", + client_name: str, + command_handler: "ReplicationCommandHandler", + ): self.client_name = client_name - self.handler = handler + self.command_handler = command_handler self.server_name = hs.config.server_name self.hs = hs self._clock = hs.get_clock() # As self.clock is defined in super class @@ -66,7 +60,11 @@ class ReplicationClientFactory(ReconnectingClientFactory): def buildProtocol(self, addr): logger.info("Connected to replication: %r", addr) return ClientReplicationStreamProtocol( - self.hs, self.client_name, self.server_name, self._clock, self.handler, + self.hs, + self.client_name, + self.server_name, + self._clock, + self.command_handler, ) def clientConnectionLost(self, connector, reason): @@ -78,41 +76,17 @@ class ReplicationClientFactory(ReconnectingClientFactory): ReconnectingClientFactory.clientConnectionFailed(self, connector, reason) -class ReplicationClientHandler(AbstractReplicationClientHandler): - """A base handler that can be passed to the ReplicationClientFactory. +class ReplicationDataHandler: + """Handles incoming stream updates from replication. - By default proxies incoming replication data to the SlaveStore. + This instance notifies the slave data store about updates. Can be subclassed + to handle updates in additional ways. """ def __init__(self, store: BaseSlavedStore): self.store = store - # The current connection. None if we are currently (re)connecting - self.connection = None - - # Any pending commands to be sent once a new connection has been - # established - self.pending_commands = [] # type: List[Command] - - # Map from string -> deferred, to wake up when receiveing a SYNC with - # the given string. - # Used for tests. - self.awaiting_syncs = {} # type: Dict[str, defer.Deferred] - - # The factory used to create connections. - self.factory = None # type: Optional[ReplicationClientFactory] - - def start_replication(self, hs): - """Helper method to start a replication connection to the remote server - using TCP. - """ - client_name = hs.config.worker_name - self.factory = ReplicationClientFactory(hs, client_name, self) - host = hs.config.worker_replication_host - port = hs.config.worker_replication_port - hs.get_reactor().connectTCP(host, port, self.factory) - - async def on_rdata(self, stream_name, token, rows): + async def on_rdata(self, stream_name: str, token: int, rows: list): """Called to handle a batch of replication data with a given stream token. By default this just pokes the slave store. Can be overridden in subclasses to @@ -124,30 +98,8 @@ class ReplicationClientHandler(AbstractReplicationClientHandler): rows (list): a list of Stream.ROW_TYPE objects as returned by Stream.parse_row. """ - logger.debug("Received rdata %s -> %s", stream_name, token) self.store.process_replication_rows(stream_name, token, rows) - async def on_position(self, stream_name, token): - """Called when we get new position data. By default this just pokes - the slave store. - - Can be overriden in subclasses to handle more. - """ - self.store.process_replication_rows(stream_name, token, []) - - def on_sync(self, data): - """When we received a SYNC we wake up any deferreds that were waiting - for the sync with the given data. - - Used by tests. - """ - d = self.awaiting_syncs.pop(data, None) - if d: - d.callback(data) - - def on_remote_server_up(self, server: str): - """Called when get a new REMOTE_SERVER_UP command.""" - def get_streams_to_replicate(self) -> Dict[str, int]: """Called when a new connection has been established and we need to subscribe to streams. @@ -163,85 +115,10 @@ class ReplicationClientHandler(AbstractReplicationClientHandler): args["account_data"] = user_account_data elif room_account_data: args["account_data"] = room_account_data - return args - def get_currently_syncing_users(self): - """Get the list of currently syncing users (if any). This is called - when a connection has been established and we need to send the - currently syncing users. (Overriden by the synchrotron's only) - """ - return [] - - def send_command(self, cmd): - """Send a command to master (when we get establish a connection if we - don't have one already.) - """ - if self.connection: - self.connection.send_command(cmd) - else: - logger.warning("Queuing command as not connected: %r", cmd.NAME) - self.pending_commands.append(cmd) - - def send_federation_ack(self, token): - """Ack data for the federation stream. This allows the master to drop - data stored purely in memory. - """ - self.send_command(FederationAckCommand(token)) - - def send_user_sync(self, instance_id, user_id, is_syncing, last_sync_ms): - """Poke the master that a user has started/stopped syncing. - """ - self.send_command( - UserSyncCommand(instance_id, user_id, is_syncing, last_sync_ms) - ) - - def send_remove_pusher(self, app_id, push_key, user_id): - """Poke the master to remove a pusher for a user - """ - cmd = RemovePusherCommand(app_id, push_key, user_id) - self.send_command(cmd) - - def send_invalidate_cache(self, cache_func, keys): - """Poke the master to invalidate a cache. - """ - cmd = InvalidateCacheCommand(cache_func.__name__, keys) - self.send_command(cmd) - - def send_user_ip(self, user_id, access_token, ip, user_agent, device_id, last_seen): - """Tell the master that the user made a request. - """ - cmd = UserIpCommand(user_id, access_token, ip, user_agent, device_id, last_seen) - self.send_command(cmd) - - def send_remote_server_up(self, server: str): - self.send_command(RemoteServerUpCommand(server)) - - def await_sync(self, data): - """Returns a deferred that is resolved when we receive a SYNC command - with given data. - - [Not currently] used by tests. - """ - return self.awaiting_syncs.setdefault(data, defer.Deferred()) - - def update_connection(self, connection): - """Called when a connection has been established (or lost with None). - """ - self.connection = connection - if connection: - for cmd in self.pending_commands: - connection.send_command(cmd) - self.pending_commands = [] - - def finished_connecting(self): - """Called when we have successfully subscribed and caught up to all - streams we're interested in. - """ - logger.info("Finished connecting to server") + async def on_position(self, stream_name: str, token: int): + self.store.process_replication_rows(stream_name, token, []) - # We don't reset the delay any earlier as otherwise if there is a - # problem during start up we'll end up tight looping connecting to the - # server. - if self.factory: - self.factory.resetDelay() + def on_remote_server_up(self, server: str): + """Called when get a new REMOTE_SERVER_UP command.""" diff --git a/synapse/replication/tcp/handler.py b/synapse/replication/tcp/handler.py new file mode 100644 index 0000000000..12a1cfd6d1 --- /dev/null +++ b/synapse/replication/tcp/handler.py @@ -0,0 +1,252 @@ +# -*- coding: utf-8 -*- +# Copyright 2017 Vector Creations Ltd +# Copyright 2020 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging +from typing import Any, Callable, Dict, List, Optional, Set + +from prometheus_client import Counter + +from synapse.replication.tcp.client import ReplicationClientFactory +from synapse.replication.tcp.commands import ( + Command, + FederationAckCommand, + InvalidateCacheCommand, + PositionCommand, + RdataCommand, + RemoteServerUpCommand, + RemovePusherCommand, + SyncCommand, + UserIpCommand, + UserSyncCommand, +) +from synapse.replication.tcp.streams import STREAMS_MAP, Stream +from synapse.util.async_helpers import Linearizer + +logger = logging.getLogger(__name__) + + +# number of updates received for each RDATA stream +inbound_rdata_count = Counter( + "synapse_replication_tcp_protocol_inbound_rdata_count", "", ["stream_name"] +) + + +class ReplicationCommandHandler: + """Handles incoming commands from replication as well as sending commands + back out to connections. + """ + + def __init__(self, hs): + self._replication_data_handler = hs.get_replication_data_handler() + self._presence_handler = hs.get_presence_handler() + + # Set of streams that we've caught up with. + self._streams_connected = set() # type: Set[str] + + self._streams = { + stream.NAME: stream(hs) for stream in STREAMS_MAP.values() + } # type: Dict[str, Stream] + + self._position_linearizer = Linearizer("replication_position") + + # Map of stream to batched updates. See RdataCommand for info on how + # batching works. + self._pending_batches = {} # type: Dict[str, List[Any]] + + # The factory used to create connections. + self._factory = None # type: Optional[ReplicationClientFactory] + + # The current connection. None if we are currently (re)connecting + self._connection = None + + def start_replication(self, hs): + """Helper method to start a replication connection to the remote server + using TCP. + """ + client_name = hs.config.worker_name + self._factory = ReplicationClientFactory(hs, client_name, self) + host = hs.config.worker_replication_host + port = hs.config.worker_replication_port + hs.get_reactor().connectTCP(host, port, self._factory) + + async def on_RDATA(self, cmd: RdataCommand): + stream_name = cmd.stream_name + inbound_rdata_count.labels(stream_name).inc() + + try: + row = STREAMS_MAP[stream_name].parse_row(cmd.row) + except Exception: + logger.exception("Failed to parse RDATA: %r %r", stream_name, cmd.row) + raise + + if cmd.token is None or stream_name not in self._streams_connected: + # I.e. either this is part of a batch of updates for this stream (in + # which case batch until we get an update for the stream with a non + # None token) or we're currently connecting so we queue up rows. + self._pending_batches.setdefault(stream_name, []).append(row) + else: + # Check if this is the last of a batch of updates + rows = self._pending_batches.pop(stream_name, []) + rows.append(row) + await self.on_rdata(stream_name, cmd.token, rows) + + async def on_rdata(self, stream_name: str, token: int, rows: list): + """Called to handle a batch of replication data with a given stream token. + + Args: + stream_name: name of the replication stream for this batch of rows + token: stream token for this batch of rows + rows: a list of Stream.ROW_TYPE objects as returned by + Stream.parse_row. + """ + logger.debug("Received rdata %s -> %s", stream_name, token) + await self._replication_data_handler.on_rdata(stream_name, token, rows) + + async def on_POSITION(self, cmd: PositionCommand): + stream = self._streams.get(cmd.stream_name) + if not stream: + logger.error("Got POSITION for unknown stream: %s", cmd.stream_name) + return + + # We protect catching up with a linearizer in case the replication + # connection reconnects under us. + with await self._position_linearizer.queue(cmd.stream_name): + # We're about to go and catch up with the stream, so mark as connecting + # to stop RDATA being handled at the same time by removing stream from + # list of connected streams. We also clear any batched up RDATA from + # before we got the POSITION. + self._streams_connected.discard(cmd.stream_name) + self._pending_batches.clear() + + # Find where we previously streamed up to. + current_token = self._replication_data_handler.get_streams_to_replicate().get( + cmd.stream_name + ) + if current_token is None: + logger.warning( + "Got POSITION for stream we're not subscribed to: %s", + cmd.stream_name, + ) + return + + # Fetch all updates between then and now. + limited = True + while limited: + updates, current_token, limited = await stream.get_updates_since( + current_token, cmd.token + ) + if updates: + await self.on_rdata( + cmd.stream_name, + current_token, + [stream.parse_row(update[1]) for update in updates], + ) + + # We've now caught up to position sent to us, notify handler. + await self._replication_data_handler.on_position(cmd.stream_name, cmd.token) + + # Handle any RDATA that came in while we were catching up. + rows = self._pending_batches.pop(cmd.stream_name, []) + if rows: + await self._replication_data_handler.on_rdata( + cmd.stream_name, rows[-1].token, rows + ) + + self._streams_connected.add(cmd.stream_name) + + async def on_SYNC(self, cmd: SyncCommand): + pass + + async def on_REMOTE_SERVER_UP(self, cmd: RemoteServerUpCommand): + """"Called when get a new REMOTE_SERVER_UP command.""" + self._replication_data_handler.on_remote_server_up(cmd.data) + + def get_currently_syncing_users(self): + """Get the list of currently syncing users (if any). This is called + when a connection has been established and we need to send the + currently syncing users. + """ + return self._presence_handler.get_currently_syncing_users() + + def update_connection(self, connection): + """Called when a connection has been established (or lost with None). + """ + self._connection = connection + + def finished_connecting(self): + """Called when we have successfully subscribed and caught up to all + streams we're interested in. + """ + logger.info("Finished connecting to server") + + # We don't reset the delay any earlier as otherwise if there is a + # problem during start up we'll end up tight looping connecting to the + # server. + if self._factory: + self._factory.resetDelay() + + def send_command(self, cmd: Command): + """Send a command to master (when we get establish a connection if we + don't have one already.) + """ + if self._connection: + self._connection.send_command(cmd) + else: + logger.warning("Dropping command as not connected: %r", cmd.NAME) + + def send_federation_ack(self, token: int): + """Ack data for the federation stream. This allows the master to drop + data stored purely in memory. + """ + self.send_command(FederationAckCommand(token)) + + def send_user_sync( + self, instance_id: str, user_id: str, is_syncing: bool, last_sync_ms: int + ): + """Poke the master that a user has started/stopped syncing. + """ + self.send_command( + UserSyncCommand(instance_id, user_id, is_syncing, last_sync_ms) + ) + + def send_remove_pusher(self, app_id: str, push_key: str, user_id: str): + """Poke the master to remove a pusher for a user + """ + cmd = RemovePusherCommand(app_id, push_key, user_id) + self.send_command(cmd) + + def send_invalidate_cache(self, cache_func: Callable, keys: tuple): + """Poke the master to invalidate a cache. + """ + cmd = InvalidateCacheCommand(cache_func.__name__, keys) + self.send_command(cmd) + + def send_user_ip( + self, + user_id: str, + access_token: str, + ip: str, + user_agent: str, + device_id: str, + last_seen: int, + ): + """Tell the master that the user made a request. + """ + cmd = UserIpCommand(user_id, access_token, ip, user_agent, device_id, last_seen) + self.send_command(cmd) + + def send_remote_server_up(self, server: str): + self.send_command(RemoteServerUpCommand(server)) diff --git a/synapse/replication/tcp/protocol.py b/synapse/replication/tcp/protocol.py index dae246825f..f2a37f568e 100644 --- a/synapse/replication/tcp/protocol.py +++ b/synapse/replication/tcp/protocol.py @@ -46,12 +46,11 @@ indicate which side is sending, these are *not* included on the wire:: > ERROR server stopping * connection closed by server * """ -import abc import fcntl import logging import struct from collections import defaultdict -from typing import Any, DefaultDict, Dict, List, Set +from typing import TYPE_CHECKING, DefaultDict, List from six import iteritems @@ -78,13 +77,12 @@ from synapse.replication.tcp.commands import ( SyncCommand, UserSyncCommand, ) -from synapse.replication.tcp.streams import STREAMS_MAP, Stream from synapse.types import Collection from synapse.util import Clock from synapse.util.stringutils import random_string -MYPY = False -if MYPY: +if TYPE_CHECKING: + from synapse.replication.tcp.handler import ReplicationCommandHandler from synapse.server import HomeServer @@ -475,71 +473,6 @@ class ServerReplicationStreamProtocol(BaseReplicationStreamProtocol): self.streamer.lost_connection(self) -class AbstractReplicationClientHandler(metaclass=abc.ABCMeta): - """ - The interface for the handler that should be passed to - ClientReplicationStreamProtocol - """ - - @abc.abstractmethod - async def on_rdata(self, stream_name, token, rows): - """Called to handle a batch of replication data with a given stream token. - - Args: - stream_name (str): name of the replication stream for this batch of rows - token (int): stream token for this batch of rows - rows (list): a list of Stream.ROW_TYPE objects as returned by - Stream.parse_row. - """ - raise NotImplementedError() - - @abc.abstractmethod - async def on_position(self, stream_name, token): - """Called when we get new position data.""" - raise NotImplementedError() - - @abc.abstractmethod - def on_sync(self, data): - """Called when get a new SYNC command.""" - raise NotImplementedError() - - @abc.abstractmethod - async def on_remote_server_up(self, server: str): - """Called when get a new REMOTE_SERVER_UP command.""" - raise NotImplementedError() - - @abc.abstractmethod - def get_streams_to_replicate(self): - """Called when a new connection has been established and we need to - subscribe to streams. - - Returns: - map from stream name to the most recent update we have for - that stream (ie, the point we want to start replicating from) - """ - raise NotImplementedError() - - @abc.abstractmethod - def get_currently_syncing_users(self): - """Get the list of currently syncing users (if any). This is called - when a connection has been established and we need to send the - currently syncing users.""" - raise NotImplementedError() - - @abc.abstractmethod - def update_connection(self, connection): - """Called when a connection has been established (or lost with None). - """ - raise NotImplementedError() - - @abc.abstractmethod - def finished_connecting(self): - """Called when we have successfully subscribed and caught up to all - streams we're interested in. - """ - raise NotImplementedError() - - class ClientReplicationStreamProtocol(BaseReplicationStreamProtocol): VALID_INBOUND_COMMANDS = VALID_SERVER_COMMANDS VALID_OUTBOUND_COMMANDS = VALID_CLIENT_COMMANDS @@ -550,7 +483,7 @@ class ClientReplicationStreamProtocol(BaseReplicationStreamProtocol): client_name: str, server_name: str, clock: Clock, - handler: AbstractReplicationClientHandler, + command_handler: "ReplicationCommandHandler", ): BaseReplicationStreamProtocol.__init__(self, clock) @@ -558,20 +491,7 @@ class ClientReplicationStreamProtocol(BaseReplicationStreamProtocol): self.client_name = client_name self.server_name = server_name - self.handler = handler - - self.streams = { - stream.NAME: stream(hs) for stream in STREAMS_MAP.values() - } # type: Dict[str, Stream] - - # Set of stream names that have been subscribe to, but haven't yet - # caught up with. This is used to track when the client has been fully - # connected to the remote. - self.streams_connecting = set(STREAMS_MAP) # type: Set[str] - - # Map of stream to batched updates. See RdataCommand for info on how - # batching works. - self.pending_batches = {} # type: Dict[str, List[Any]] + self.handler = command_handler def connectionMade(self): self.send_command(NameCommand(self.client_name)) @@ -589,89 +509,39 @@ class ClientReplicationStreamProtocol(BaseReplicationStreamProtocol): # We've now finished connecting to so inform the client handler self.handler.update_connection(self) + self.handler.finished_connecting() - async def on_SERVER(self, cmd): - if cmd.data != self.server_name: - logger.error("[%s] Connected to wrong remote: %r", self.id(), cmd.data) - self.send_error("Wrong remote") - - async def on_RDATA(self, cmd): - stream_name = cmd.stream_name - inbound_rdata_count.labels(stream_name).inc() - - try: - row = STREAMS_MAP[stream_name].parse_row(cmd.row) - except Exception: - logger.exception( - "[%s] Failed to parse RDATA: %r %r", self.id(), stream_name, cmd.row - ) - raise - - if cmd.token is None or stream_name in self.streams_connecting: - # I.e. this is part of a batch of updates for this stream. Batch - # until we get an update for the stream with a non None token - self.pending_batches.setdefault(stream_name, []).append(row) - else: - # Check if this is the last of a batch of updates - rows = self.pending_batches.pop(stream_name, []) - rows.append(row) - await self.handler.on_rdata(stream_name, cmd.token, rows) - - async def on_POSITION(self, cmd: PositionCommand): - stream = self.streams.get(cmd.stream_name) - if not stream: - logger.error("Got POSITION for unknown stream: %s", cmd.stream_name) - return - - # Find where we previously streamed up to. - current_token = self.handler.get_streams_to_replicate().get(cmd.stream_name) - if current_token is None: - logger.warning( - "Got POSITION for stream we're not subscribed to: %s", cmd.stream_name - ) - return - - # Fetch all updates between then and now. - limited = True - while limited: - updates, current_token, limited = await stream.get_updates_since( - current_token, cmd.token - ) - - # Check if the connection was closed underneath us, if so we bail - # rather than risk having concurrent catch ups going on. - if self.state == ConnectionStates.CLOSED: - return - - if updates: - await self.handler.on_rdata( - cmd.stream_name, - current_token, - [stream.parse_row(update[1]) for update in updates], - ) + async def handle_command(self, cmd: Command): + """Handle a command we have received over the replication stream. - # We've now caught up to position sent to us, notify handler. - await self.handler.on_position(cmd.stream_name, cmd.token) + Delegates to `command_handler.on_`, which must return an + awaitable. - self.streams_connecting.discard(cmd.stream_name) - if not self.streams_connecting: - self.handler.finished_connecting() + Args: + cmd: received command + """ + handled = False - # Check if the connection was closed underneath us, if so we bail - # rather than risk having concurrent catch ups going on. - if self.state == ConnectionStates.CLOSED: - return + # First call any command handlers on this instance. These are for TCP + # specific handling. + cmd_func = getattr(self, "on_%s" % (cmd.NAME,), None) + if cmd_func: + await cmd_func(cmd) + handled = True - # Handle any RDATA that came in while we were catching up. - rows = self.pending_batches.pop(cmd.stream_name, []) - if rows: - await self.handler.on_rdata(cmd.stream_name, rows[-1].token, rows) + # Then call out to the handler. + cmd_func = getattr(self.handler, "on_%s" % (cmd.NAME,), None) + if cmd_func: + await cmd_func(cmd) + handled = True - async def on_SYNC(self, cmd): - self.handler.on_sync(cmd.data) + if not handled: + logger.warning("Unhandled command: %r", cmd) - async def on_REMOTE_SERVER_UP(self, cmd: RemoteServerUpCommand): - self.handler.on_remote_server_up(cmd.data) + async def on_SERVER(self, cmd): + if cmd.data != self.server_name: + logger.error("[%s] Connected to wrong remote: %r", self.id(), cmd.data) + self.send_error("Wrong remote") def replicate(self): """Send the subscription request to the server @@ -768,8 +638,3 @@ tcp_outbound_commands = LaterGauge( for k, count in iteritems(p.outbound_commands_counter) }, ) - -# number of updates received for each RDATA stream -inbound_rdata_count = Counter( - "synapse_replication_tcp_protocol_inbound_rdata_count", "", ["stream_name"] -) diff --git a/synapse/server.py b/synapse/server.py index 9228e1c892..9d273c980c 100644 --- a/synapse/server.py +++ b/synapse/server.py @@ -87,6 +87,8 @@ from synapse.http.matrixfederationclient import MatrixFederationHttpClient from synapse.notifier import Notifier from synapse.push.action_generator import ActionGenerator from synapse.push.pusherpool import PusherPool +from synapse.replication.tcp.client import ReplicationDataHandler +from synapse.replication.tcp.handler import ReplicationCommandHandler from synapse.replication.tcp.resource import ReplicationStreamer from synapse.rest.media.v1.media_repository import ( MediaRepository, @@ -206,6 +208,7 @@ class HomeServer(object): "password_policy_handler", "storage", "replication_streamer", + "replication_data_handler", ] REQUIRED_ON_MASTER_STARTUP = ["user_directory_handler", "stats_handler"] @@ -468,7 +471,7 @@ class HomeServer(object): return ReadMarkerHandler(self) def build_tcp_replication(self): - raise NotImplementedError() + return ReplicationCommandHandler(self) def build_action_generator(self): return ActionGenerator(self) @@ -562,6 +565,9 @@ class HomeServer(object): def build_replication_streamer(self) -> ReplicationStreamer: return ReplicationStreamer(self) + def build_replication_data_handler(self): + return ReplicationDataHandler(self.get_datastore()) + def remove_pusher(self, app_id, push_key, user_id): return self.get_pusherpool().remove_pusher(app_id, push_key, user_id) diff --git a/synapse/server.pyi b/synapse/server.pyi index 9d1dfa71e7..9013e9bac9 100644 --- a/synapse/server.pyi +++ b/synapse/server.pyi @@ -19,6 +19,7 @@ import synapse.handlers.set_password import synapse.http.client import synapse.notifier import synapse.replication.tcp.client +import synapse.replication.tcp.handler import synapse.rest.media.v1.media_repository import synapse.server_notices.server_notices_manager import synapse.server_notices.server_notices_sender @@ -106,7 +107,11 @@ class HomeServer(object): pass def get_tcp_replication( self, - ) -> synapse.replication.tcp.client.ReplicationClientHandler: + ) -> synapse.replication.tcp.handler.ReplicationCommandHandler: + pass + def get_replication_data_handler( + self, + ) -> synapse.replication.tcp.client.ReplicationDataHandler: pass def get_federation_registry( self, diff --git a/tests/replication/slave/storage/_base.py b/tests/replication/slave/storage/_base.py index 2a1e7c7166..8902a5ab69 100644 --- a/tests/replication/slave/storage/_base.py +++ b/tests/replication/slave/storage/_base.py @@ -17,8 +17,9 @@ from mock import Mock, NonCallableMock from synapse.replication.tcp.client import ( ReplicationClientFactory, - ReplicationClientHandler, + ReplicationDataHandler, ) +from synapse.replication.tcp.handler import ReplicationCommandHandler from synapse.replication.tcp.resource import ReplicationStreamProtocolFactory from synapse.storage.database import make_conn @@ -51,15 +52,19 @@ class BaseSlavedStoreTestCase(unittest.HomeserverTestCase): self.event_id = 0 server_factory = ReplicationStreamProtocolFactory(self.hs) - self.streamer = server_factory.streamer + self.streamer = hs.get_replication_streamer() - handler_factory = Mock() - self.replication_handler = ReplicationClientHandler(self.slaved_store) - self.replication_handler.factory = handler_factory + # We now do some gut wrenching so that we have a client that is based + # off of the slave store rather than the main store. + self.replication_handler = ReplicationCommandHandler(self.hs) + self.replication_handler._replication_data_handler = ReplicationDataHandler( + self.slaved_store + ) client_factory = ReplicationClientFactory( self.hs, "client_name", self.replication_handler ) + client_factory.handler = self.replication_handler server = server_factory.buildProtocol(None) client = client_factory.buildProtocol(None) diff --git a/tests/replication/tcp/streams/_base.py b/tests/replication/tcp/streams/_base.py index a755fe2879..32238fe79a 100644 --- a/tests/replication/tcp/streams/_base.py +++ b/tests/replication/tcp/streams/_base.py @@ -15,7 +15,7 @@ from mock import Mock -from synapse.replication.tcp.commands import ReplicateCommand +from synapse.replication.tcp.handler import ReplicationCommandHandler from synapse.replication.tcp.protocol import ClientReplicationStreamProtocol from synapse.replication.tcp.resource import ReplicationStreamProtocolFactory @@ -26,15 +26,20 @@ from tests.server import FakeTransport class BaseStreamTestCase(unittest.HomeserverTestCase): """Base class for tests of the replication streams""" + def make_homeserver(self, reactor, clock): + self.test_handler = Mock(wraps=TestReplicationDataHandler()) + return self.setup_test_homeserver(replication_data_handler=self.test_handler) + def prepare(self, reactor, clock, hs): # build a replication server - server_factory = ReplicationStreamProtocolFactory(self.hs) - self.streamer = server_factory.streamer + server_factory = ReplicationStreamProtocolFactory(hs) + self.streamer = hs.get_replication_streamer() self.server = server_factory.buildProtocol(None) - self.test_handler = Mock(wraps=TestReplicationClientHandler()) + repl_handler = ReplicationCommandHandler(hs) + repl_handler.handler = self.test_handler self.client = ClientReplicationStreamProtocol( - hs, "client", "test", clock, self.test_handler, + hs, "client", "test", clock, repl_handler, ) self._client_transport = None @@ -69,13 +74,9 @@ class BaseStreamTestCase(unittest.HomeserverTestCase): self.streamer.on_notifier_poke() self.pump(0.1) - def replicate_stream(self): - """Make the client end a REPLICATE command to set up a subscription to a stream""" - self.client.send_command(ReplicateCommand()) - -class TestReplicationClientHandler(object): - """Drop-in for ReplicationClientHandler which just collects RDATA rows""" +class TestReplicationDataHandler: + """Drop-in for ReplicationDataHandler which just collects RDATA rows""" def __init__(self): self.streams = set() @@ -88,18 +89,9 @@ class TestReplicationClientHandler(object): positions[stream] = max(token, positions.get(stream, 0)) return positions - def get_currently_syncing_users(self): - return [] - - def update_connection(self, connection): - pass - - def finished_connecting(self): - pass - - async def on_position(self, stream_name, token): - """Called when we get new position data.""" - async def on_rdata(self, stream_name, token, rows): for r in rows: self._received_rdata_rows.append((stream_name, token, r)) + + async def on_position(self, stream_name, token): + pass diff --git a/tests/replication/tcp/streams/test_receipts.py b/tests/replication/tcp/streams/test_receipts.py index 0ec0825a0e..a0206f7363 100644 --- a/tests/replication/tcp/streams/test_receipts.py +++ b/tests/replication/tcp/streams/test_receipts.py @@ -24,7 +24,6 @@ class ReceiptsStreamTestCase(BaseStreamTestCase): self.reconnect() # make the client subscribe to the receipts stream - self.replicate_stream() self.test_handler.streams.add("receipts") # tell the master to send a new receipt -- cgit 1.4.1 From f31e65a749f84f8b3278c91784509d908d4fb342 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Tue, 7 Apr 2020 23:06:39 +0100 Subject: bg update to clear out duplicate outbound_device_list_pokes (#7193) We seem to have some duplicates, which could do with being cleared out. --- changelog.d/7193.misc | 1 + synapse/storage/data_stores/main/client_ips.py | 16 ++--- synapse/storage/data_stores/main/devices.py | 73 ++++++++++++++++++- .../delta/58/02remove_dup_outbound_pokes.sql | 22 ++++++ synapse/storage/database.py | 83 +++++++++++++++++++++- tests/storage/test_database.py | 52 ++++++++++++++ 6 files changed, 234 insertions(+), 13 deletions(-) create mode 100644 changelog.d/7193.misc create mode 100644 synapse/storage/data_stores/main/schema/delta/58/02remove_dup_outbound_pokes.sql create mode 100644 tests/storage/test_database.py (limited to 'tests') diff --git a/changelog.d/7193.misc b/changelog.d/7193.misc new file mode 100644 index 0000000000..383a738e64 --- /dev/null +++ b/changelog.d/7193.misc @@ -0,0 +1 @@ +Add a background database update job to clear out duplicate `device_lists_outbound_pokes`. diff --git a/synapse/storage/data_stores/main/client_ips.py b/synapse/storage/data_stores/main/client_ips.py index e1ccb27142..92bc06919b 100644 --- a/synapse/storage/data_stores/main/client_ips.py +++ b/synapse/storage/data_stores/main/client_ips.py @@ -21,7 +21,7 @@ from twisted.internet import defer from synapse.metrics.background_process_metrics import wrap_as_background_process from synapse.storage._base import SQLBaseStore -from synapse.storage.database import Database +from synapse.storage.database import Database, make_tuple_comparison_clause from synapse.util.caches import CACHE_SIZE_FACTOR from synapse.util.caches.descriptors import Cache @@ -303,16 +303,10 @@ class ClientIpBackgroundUpdateStore(SQLBaseStore): # we'll just end up updating the same device row multiple # times, which is fine. - if self.database_engine.supports_tuple_comparison: - where_clause = "(user_id, device_id) > (?, ?)" - where_args = [last_user_id, last_device_id] - else: - # We explicitly do a `user_id >= ? AND (...)` here to ensure - # that an index is used, as doing `user_id > ? OR (user_id = ? AND ...)` - # makes it hard for query optimiser to tell that it can use the - # index on user_id - where_clause = "user_id >= ? AND (user_id > ? OR device_id > ?)" - where_args = [last_user_id, last_user_id, last_device_id] + where_clause, where_args = make_tuple_comparison_clause( + self.database_engine, + [("user_id", last_user_id), ("device_id", last_device_id)], + ) sql = """ SELECT diff --git a/synapse/storage/data_stores/main/devices.py b/synapse/storage/data_stores/main/devices.py index 4c5bea4a5c..ee3a2ab031 100644 --- a/synapse/storage/data_stores/main/devices.py +++ b/synapse/storage/data_stores/main/devices.py @@ -32,7 +32,11 @@ from synapse.logging.opentracing import ( ) from synapse.metrics.background_process_metrics import run_as_background_process from synapse.storage._base import SQLBaseStore, db_to_json, make_in_list_sql_clause -from synapse.storage.database import Database, LoggingTransaction +from synapse.storage.database import ( + Database, + LoggingTransaction, + make_tuple_comparison_clause, +) from synapse.types import Collection, get_verify_key_from_cross_signing_key from synapse.util.caches.descriptors import ( Cache, @@ -49,6 +53,8 @@ DROP_DEVICE_LIST_STREAMS_NON_UNIQUE_INDEXES = ( "drop_device_list_streams_non_unique_indexes" ) +BG_UPDATE_REMOVE_DUP_OUTBOUND_POKES = "remove_dup_outbound_pokes" + class DeviceWorkerStore(SQLBaseStore): def get_device(self, user_id, device_id): @@ -714,6 +720,11 @@ class DeviceBackgroundUpdateStore(SQLBaseStore): self._drop_device_list_streams_non_unique_indexes, ) + # clear out duplicate device list outbound pokes + self.db.updates.register_background_update_handler( + BG_UPDATE_REMOVE_DUP_OUTBOUND_POKES, self._remove_duplicate_outbound_pokes, + ) + @defer.inlineCallbacks def _drop_device_list_streams_non_unique_indexes(self, progress, batch_size): def f(conn): @@ -728,6 +739,66 @@ class DeviceBackgroundUpdateStore(SQLBaseStore): ) return 1 + async def _remove_duplicate_outbound_pokes(self, progress, batch_size): + # for some reason, we have accumulated duplicate entries in + # device_lists_outbound_pokes, which makes prune_outbound_device_list_pokes less + # efficient. + # + # For each duplicate, we delete all the existing rows and put one back. + + KEY_COLS = ["stream_id", "destination", "user_id", "device_id"] + last_row = progress.get( + "last_row", + {"stream_id": 0, "destination": "", "user_id": "", "device_id": ""}, + ) + + def _txn(txn): + clause, args = make_tuple_comparison_clause( + self.db.engine, [(x, last_row[x]) for x in KEY_COLS] + ) + sql = """ + SELECT stream_id, destination, user_id, device_id, MAX(ts) AS ts + FROM device_lists_outbound_pokes + WHERE %s + GROUP BY %s + HAVING count(*) > 1 + ORDER BY %s + LIMIT ? + """ % ( + clause, # WHERE + ",".join(KEY_COLS), # GROUP BY + ",".join(KEY_COLS), # ORDER BY + ) + txn.execute(sql, args + [batch_size]) + rows = self.db.cursor_to_dict(txn) + + row = None + for row in rows: + self.db.simple_delete_txn( + txn, "device_lists_outbound_pokes", {x: row[x] for x in KEY_COLS}, + ) + + row["sent"] = False + self.db.simple_insert_txn( + txn, "device_lists_outbound_pokes", row, + ) + + if row: + self.db.updates._background_update_progress_txn( + txn, BG_UPDATE_REMOVE_DUP_OUTBOUND_POKES, {"last_row": row}, + ) + + return len(rows) + + rows = await self.db.runInteraction(BG_UPDATE_REMOVE_DUP_OUTBOUND_POKES, _txn) + + if not rows: + await self.db.updates._end_background_update( + BG_UPDATE_REMOVE_DUP_OUTBOUND_POKES + ) + + return rows + class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore): def __init__(self, database: Database, db_conn, hs): diff --git a/synapse/storage/data_stores/main/schema/delta/58/02remove_dup_outbound_pokes.sql b/synapse/storage/data_stores/main/schema/delta/58/02remove_dup_outbound_pokes.sql new file mode 100644 index 0000000000..fdc39e9ba5 --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/58/02remove_dup_outbound_pokes.sql @@ -0,0 +1,22 @@ +/* Copyright 2020 The Matrix.org Foundation C.I.C + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + /* for some reason, we have accumulated duplicate entries in + * device_lists_outbound_pokes, which makes prune_outbound_device_list_pokes less + * efficient. + */ + +INSERT INTO background_updates (ordering, update_name, progress_json) + VALUES (5800, 'remove_dup_outbound_pokes', '{}'); diff --git a/synapse/storage/database.py b/synapse/storage/database.py index 715c0346dd..a7cd97b0b0 100644 --- a/synapse/storage/database.py +++ b/synapse/storage/database.py @@ -17,7 +17,17 @@ import logging import time from time import monotonic as monotonic_time -from typing import Any, Callable, Dict, Iterable, Iterator, List, Optional, Tuple +from typing import ( + Any, + Callable, + Dict, + Iterable, + Iterator, + List, + Optional, + Tuple, + TypeVar, +) from six import iteritems, iterkeys, itervalues from six.moves import intern, range @@ -1557,3 +1567,74 @@ def make_in_list_sql_clause( return "%s = ANY(?)" % (column,), [list(iterable)] else: return "%s IN (%s)" % (column, ",".join("?" for _ in iterable)), list(iterable) + + +KV = TypeVar("KV") + + +def make_tuple_comparison_clause( + database_engine: BaseDatabaseEngine, keys: List[Tuple[str, KV]] +) -> Tuple[str, List[KV]]: + """Returns a tuple comparison SQL clause + + Depending what the SQL engine supports, builds a SQL clause that looks like either + "(a, b) > (?, ?)", or "(a > ?) OR (a == ? AND b > ?)". + + Args: + database_engine + keys: A set of (column, value) pairs to be compared. + + Returns: + A tuple of SQL query and the args + """ + if database_engine.supports_tuple_comparison: + return ( + "(%s) > (%s)" % (",".join(k[0] for k in keys), ",".join("?" for _ in keys)), + [k[1] for k in keys], + ) + + # we want to build a clause + # (a > ?) OR + # (a == ? AND b > ?) OR + # (a == ? AND b == ? AND c > ?) + # ... + # (a == ? AND b == ? AND ... AND z > ?) + # + # or, equivalently: + # + # (a > ? OR (a == ? AND + # (b > ? OR (b == ? AND + # ... + # (y > ? OR (y == ? AND + # z > ? + # )) + # ... + # )) + # )) + # + # which itself is equivalent to (and apparently easier for the query optimiser): + # + # (a >= ? AND (a > ? OR + # (b >= ? AND (b > ? OR + # ... + # (y >= ? AND (y > ? OR + # z > ? + # )) + # ... + # )) + # )) + # + # + + clause = "" + args = [] # type: List[KV] + for k, v in keys[:-1]: + clause = clause + "(%s >= ? AND (%s > ? OR " % (k, k) + args.extend([v, v]) + + (k, v) = keys[-1] + clause += "%s > ?" % (k,) + args.append(v) + + clause += "))" * (len(keys) - 1) + return clause, args diff --git a/tests/storage/test_database.py b/tests/storage/test_database.py new file mode 100644 index 0000000000..5a77c84962 --- /dev/null +++ b/tests/storage/test_database.py @@ -0,0 +1,52 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from synapse.storage.database import make_tuple_comparison_clause +from synapse.storage.engines import BaseDatabaseEngine + +from tests import unittest + + +def _stub_db_engine(**kwargs) -> BaseDatabaseEngine: + # returns a DatabaseEngine, circumventing the abc mechanism + # any kwargs are set as attributes on the class before instantiating it + t = type( + "TestBaseDatabaseEngine", + (BaseDatabaseEngine,), + dict(BaseDatabaseEngine.__dict__), + ) + # defeat the abc mechanism + t.__abstractmethods__ = set() + for k, v in kwargs.items(): + setattr(t, k, v) + return t(None, None) + + +class TupleComparisonClauseTestCase(unittest.TestCase): + def test_native_tuple_comparison(self): + db_engine = _stub_db_engine(supports_tuple_comparison=True) + clause, args = make_tuple_comparison_clause(db_engine, [("a", 1), ("b", 2)]) + self.assertEqual(clause, "(a,b) > (?,?)") + self.assertEqual(args, [1, 2]) + + def test_emulated_tuple_comparison(self): + db_engine = _stub_db_engine(supports_tuple_comparison=False) + clause, args = make_tuple_comparison_clause( + db_engine, [("a", 1), ("b", 2), ("c", 3)] + ) + self.assertEqual( + clause, "(a >= ? AND (a > ? OR (b >= ? AND (b > ? OR c > ?))))" + ) + self.assertEqual(args, [1, 1, 2, 2, 3]) -- cgit 1.4.1 From b85d7652ff084fee997e0bb44ecd46c2789abbdd Mon Sep 17 00:00:00 2001 From: Patrick Cloke Date: Thu, 9 Apr 2020 13:28:13 -0400 Subject: Do not allow a deactivated user to login via SSO. (#7240) --- changelog.d/7240.bugfix | 1 + synapse/config/sso.py | 7 ++++ synapse/handlers/auth.py | 34 +++++++++++++++--- synapse/handlers/cas_handler.py | 2 +- synapse/handlers/saml_handler.py | 2 +- synapse/module_api/__init__.py | 22 +++++++++++- synapse/res/templates/sso_account_deactivated.html | 10 ++++++ tests/rest/client/v1/test_login.py | 42 ++++++++++++++++++++-- 8 files changed, 110 insertions(+), 10 deletions(-) create mode 100644 changelog.d/7240.bugfix create mode 100644 synapse/res/templates/sso_account_deactivated.html (limited to 'tests') diff --git a/changelog.d/7240.bugfix b/changelog.d/7240.bugfix new file mode 100644 index 0000000000..83b18d3e11 --- /dev/null +++ b/changelog.d/7240.bugfix @@ -0,0 +1 @@ +Do not allow a deactivated user to login via SSO. diff --git a/synapse/config/sso.py b/synapse/config/sso.py index ec3dca9efc..686678a3b7 100644 --- a/synapse/config/sso.py +++ b/synapse/config/sso.py @@ -12,6 +12,7 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. +import os from typing import Any, Dict import pkg_resources @@ -36,6 +37,12 @@ class SSOConfig(Config): template_dir = pkg_resources.resource_filename("synapse", "res/templates",) self.sso_redirect_confirm_template_dir = template_dir + self.sso_account_deactivated_template = self.read_file( + os.path.join( + self.sso_redirect_confirm_template_dir, "sso_account_deactivated.html" + ), + "sso_account_deactivated_template", + ) self.sso_client_whitelist = sso_config.get("client_whitelist") or [] diff --git a/synapse/handlers/auth.py b/synapse/handlers/auth.py index 892adb00b9..fbfbd44a2e 100644 --- a/synapse/handlers/auth.py +++ b/synapse/handlers/auth.py @@ -161,6 +161,9 @@ class AuthHandler(BaseHandler): self._sso_auth_confirm_template = load_jinja2_templates( hs.config.sso_redirect_confirm_template_dir, ["sso_auth_confirm.html"], )[0] + self._sso_account_deactivated_template = ( + hs.config.sso_account_deactivated_template + ) self._server_name = hs.config.server_name @@ -644,9 +647,6 @@ class AuthHandler(BaseHandler): Returns: defer.Deferred: (unicode) canonical_user_id, or None if zero or multiple matches - - Raises: - UserDeactivatedError if a user is found but is deactivated. """ res = yield self._find_user_id_and_pwd_hash(user_id) if res is not None: @@ -1099,7 +1099,7 @@ class AuthHandler(BaseHandler): request.write(html_bytes) finish_request(request) - def complete_sso_login( + async def complete_sso_login( self, registered_user_id: str, request: SynapseRequest, @@ -1113,6 +1113,32 @@ class AuthHandler(BaseHandler): client_redirect_url: The URL to which to redirect the user at the end of the process. """ + # If the account has been deactivated, do not proceed with the login + # flow. + deactivated = await self.store.get_user_deactivated_status(registered_user_id) + if deactivated: + html = self._sso_account_deactivated_template.encode("utf-8") + + request.setResponseCode(403) + request.setHeader(b"Content-Type", b"text/html; charset=utf-8") + request.setHeader(b"Content-Length", b"%d" % (len(html),)) + request.write(html) + finish_request(request) + return + + self._complete_sso_login(registered_user_id, request, client_redirect_url) + + def _complete_sso_login( + self, + registered_user_id: str, + request: SynapseRequest, + client_redirect_url: str, + ): + """ + The synchronous portion of complete_sso_login. + + This exists purely for backwards compatibility of synapse.module_api.ModuleApi. + """ # Create a login token login_token = self.macaroon_gen.generate_short_term_login_token( registered_user_id diff --git a/synapse/handlers/cas_handler.py b/synapse/handlers/cas_handler.py index d977badf35..5cb3f9d133 100644 --- a/synapse/handlers/cas_handler.py +++ b/synapse/handlers/cas_handler.py @@ -216,6 +216,6 @@ class CasHandler: localpart=localpart, default_display_name=user_display_name ) - self._auth_handler.complete_sso_login( + await self._auth_handler.complete_sso_login( registered_user_id, request, client_redirect_url ) diff --git a/synapse/handlers/saml_handler.py b/synapse/handlers/saml_handler.py index 4741c82f61..7c9454b504 100644 --- a/synapse/handlers/saml_handler.py +++ b/synapse/handlers/saml_handler.py @@ -154,7 +154,7 @@ class SamlHandler: ) else: - self._auth_handler.complete_sso_login(user_id, request, relay_state) + await self._auth_handler.complete_sso_login(user_id, request, relay_state) async def _map_saml_response_to_user( self, resp_bytes: str, client_redirect_url: str diff --git a/synapse/module_api/__init__.py b/synapse/module_api/__init__.py index c7fffd72f2..afc3598e11 100644 --- a/synapse/module_api/__init__.py +++ b/synapse/module_api/__init__.py @@ -220,6 +220,26 @@ class ModuleApi(object): want their access token sent to `client_redirect_url`, or redirect them to that URL with a token directly if the URL matches with one of the whitelisted clients. + This is deprecated in favor of complete_sso_login_async. + + Args: + registered_user_id: The MXID that has been registered as a previous step of + of this SSO login. + request: The request to respond to. + client_redirect_url: The URL to which to offer to redirect the user (or to + redirect them directly if whitelisted). + """ + self._auth_handler._complete_sso_login( + registered_user_id, request, client_redirect_url, + ) + + async def complete_sso_login_async( + self, registered_user_id: str, request: SynapseRequest, client_redirect_url: str + ): + """Complete a SSO login by redirecting the user to a page to confirm whether they + want their access token sent to `client_redirect_url`, or redirect them to that + URL with a token directly if the URL matches with one of the whitelisted clients. + Args: registered_user_id: The MXID that has been registered as a previous step of of this SSO login. @@ -227,6 +247,6 @@ class ModuleApi(object): client_redirect_url: The URL to which to offer to redirect the user (or to redirect them directly if whitelisted). """ - self._auth_handler.complete_sso_login( + await self._auth_handler.complete_sso_login( registered_user_id, request, client_redirect_url, ) diff --git a/synapse/res/templates/sso_account_deactivated.html b/synapse/res/templates/sso_account_deactivated.html new file mode 100644 index 0000000000..4eb8db9fb4 --- /dev/null +++ b/synapse/res/templates/sso_account_deactivated.html @@ -0,0 +1,10 @@ + + + + + SSO account deactivated + + +

This account has been deactivated.

+ + diff --git a/tests/rest/client/v1/test_login.py b/tests/rest/client/v1/test_login.py index aed8853d6e..1856c7ffd5 100644 --- a/tests/rest/client/v1/test_login.py +++ b/tests/rest/client/v1/test_login.py @@ -257,7 +257,7 @@ class LoginRestServletTestCase(unittest.HomeserverTestCase): self.assertEquals(channel.code, 200, channel.result) -class CASRedirectConfirmTestCase(unittest.HomeserverTestCase): +class CASTestCase(unittest.HomeserverTestCase): servlets = [ login.register_servlets, @@ -274,6 +274,9 @@ class CASRedirectConfirmTestCase(unittest.HomeserverTestCase): "service_url": "https://matrix.goodserver.com:8448", } + cas_user_id = "username" + self.user_id = "@%s:test" % cas_user_id + async def get_raw(uri, args): """Return an example response payload from a call to the `/proxyValidate` endpoint of a CAS server, copied from @@ -282,10 +285,11 @@ class CASRedirectConfirmTestCase(unittest.HomeserverTestCase): This needs to be returned by an async function (as opposed to set as the mock's return value) because the corresponding Synapse code awaits on it. """ - return """ + return ( + """ - username + %s PGTIOU-84678-8a9d... https://proxy2/pgtUrl @@ -294,6 +298,8 @@ class CASRedirectConfirmTestCase(unittest.HomeserverTestCase): """ + % cas_user_id + ) mocked_http_client = Mock(spec=["get_raw"]) mocked_http_client.get_raw.side_effect = get_raw @@ -304,6 +310,9 @@ class CASRedirectConfirmTestCase(unittest.HomeserverTestCase): return self.hs + def prepare(self, reactor, clock, hs): + self.deactivate_account_handler = hs.get_deactivate_account_handler() + def test_cas_redirect_confirm(self): """Tests that the SSO login flow serves a confirmation page before redirecting a user to the redirect URL. @@ -370,3 +379,30 @@ class CASRedirectConfirmTestCase(unittest.HomeserverTestCase): self.assertEqual(channel.code, 302) location_headers = channel.headers.getRawHeaders("Location") self.assertEqual(location_headers[0][: len(redirect_url)], redirect_url) + + @override_config({"sso": {"client_whitelist": ["https://legit-site.com/"]}}) + def test_deactivated_user(self): + """Logging in as a deactivated account should error.""" + redirect_url = "https://legit-site.com/" + + # First login (to create the user). + self._test_redirect(redirect_url) + + # Deactivate the account. + self.get_success( + self.deactivate_account_handler.deactivate_account(self.user_id, False) + ) + + # Request the CAS ticket. + cas_ticket_url = ( + "/_matrix/client/r0/login/cas/ticket?redirectUrl=%s&ticket=ticket" + % (urllib.parse.quote(redirect_url)) + ) + + # Get Synapse to call the fake CAS and serve the template. + request, channel = self.make_request("GET", cas_ticket_url) + self.render(request) + + # Because the user is deactivated they are served an error template. + self.assertEqual(channel.code, 403) + self.assertIn(b"SSO account deactivated", channel.result["body"]) -- cgit 1.4.1 From ac978ab3da48ca1d0c6f40d9857d769e72e960dd Mon Sep 17 00:00:00 2001 From: Andrew Morgan <1342360+anoadragon453@users.noreply.github.com> Date: Thu, 9 Apr 2020 18:45:38 +0100 Subject: Default PL100 to enable encryption in a room (#7230) --- changelog.d/7230.feature | 1 + synapse/handlers/room.py | 1 + tests/rest/client/test_power_levels.py | 205 +++++++++++++++++++++++++++++++++ tests/rest/client/v1/utils.py | 96 ++++++++++++++- 4 files changed, 299 insertions(+), 4 deletions(-) create mode 100644 changelog.d/7230.feature create mode 100644 tests/rest/client/test_power_levels.py (limited to 'tests') diff --git a/changelog.d/7230.feature b/changelog.d/7230.feature new file mode 100644 index 0000000000..aab777648f --- /dev/null +++ b/changelog.d/7230.feature @@ -0,0 +1 @@ +Require admin privileges to enable room encryption by default. This does not affect existing rooms. diff --git a/synapse/handlers/room.py b/synapse/handlers/room.py index f580ab2e9f..df3e0cff67 100644 --- a/synapse/handlers/room.py +++ b/synapse/handlers/room.py @@ -806,6 +806,7 @@ class RoomCreationHandler(BaseHandler): EventTypes.RoomAvatar: 50, EventTypes.Tombstone: 100, EventTypes.ServerACL: 100, + EventTypes.RoomEncryption: 100, }, "events_default": 0, "state_default": 50, diff --git a/tests/rest/client/test_power_levels.py b/tests/rest/client/test_power_levels.py new file mode 100644 index 0000000000..913ea3c98e --- /dev/null +++ b/tests/rest/client/test_power_levels.py @@ -0,0 +1,205 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from synapse.rest import admin +from synapse.rest.client.v1 import login, room +from synapse.rest.client.v2_alpha import sync + +from tests.unittest import HomeserverTestCase + + +class PowerLevelsTestCase(HomeserverTestCase): + """Tests that power levels are enforced in various situations""" + + servlets = [ + admin.register_servlets, + room.register_servlets, + login.register_servlets, + sync.register_servlets, + ] + + def make_homeserver(self, reactor, clock): + config = self.default_config() + + return self.setup_test_homeserver(config=config) + + def prepare(self, reactor, clock, hs): + # register a room admin, moderator and regular user + self.admin_user_id = self.register_user("admin", "pass") + self.admin_access_token = self.login("admin", "pass") + self.mod_user_id = self.register_user("mod", "pass") + self.mod_access_token = self.login("mod", "pass") + self.user_user_id = self.register_user("user", "pass") + self.user_access_token = self.login("user", "pass") + + # Create a room + self.room_id = self.helper.create_room_as( + self.admin_user_id, tok=self.admin_access_token + ) + + # Invite the other users + self.helper.invite( + room=self.room_id, + src=self.admin_user_id, + tok=self.admin_access_token, + targ=self.mod_user_id, + ) + self.helper.invite( + room=self.room_id, + src=self.admin_user_id, + tok=self.admin_access_token, + targ=self.user_user_id, + ) + + # Make the other users join the room + self.helper.join( + room=self.room_id, user=self.mod_user_id, tok=self.mod_access_token + ) + self.helper.join( + room=self.room_id, user=self.user_user_id, tok=self.user_access_token + ) + + # Mod the mod + room_power_levels = self.helper.get_state( + self.room_id, "m.room.power_levels", tok=self.admin_access_token, + ) + + # Update existing power levels with mod at PL50 + room_power_levels["users"].update({self.mod_user_id: 50}) + + self.helper.send_state( + self.room_id, + "m.room.power_levels", + room_power_levels, + tok=self.admin_access_token, + ) + + def test_non_admins_cannot_enable_room_encryption(self): + # have the mod try to enable room encryption + self.helper.send_state( + self.room_id, + "m.room.encryption", + {"algorithm": "m.megolm.v1.aes-sha2"}, + tok=self.mod_access_token, + expect_code=403, # expect failure + ) + + # have the user try to enable room encryption + self.helper.send_state( + self.room_id, + "m.room.encryption", + {"algorithm": "m.megolm.v1.aes-sha2"}, + tok=self.user_access_token, + expect_code=403, # expect failure + ) + + def test_non_admins_cannot_send_server_acl(self): + # have the mod try to send a server ACL + self.helper.send_state( + self.room_id, + "m.room.server_acl", + { + "allow": ["*"], + "allow_ip_literals": False, + "deny": ["*.evil.com", "evil.com"], + }, + tok=self.mod_access_token, + expect_code=403, # expect failure + ) + + # have the user try to send a server ACL + self.helper.send_state( + self.room_id, + "m.room.server_acl", + { + "allow": ["*"], + "allow_ip_literals": False, + "deny": ["*.evil.com", "evil.com"], + }, + tok=self.user_access_token, + expect_code=403, # expect failure + ) + + def test_non_admins_cannot_tombstone_room(self): + # Create another room that will serve as our "upgraded room" + self.upgraded_room_id = self.helper.create_room_as( + self.admin_user_id, tok=self.admin_access_token + ) + + # have the mod try to send a tombstone event + self.helper.send_state( + self.room_id, + "m.room.tombstone", + { + "body": "This room has been replaced", + "replacement_room": self.upgraded_room_id, + }, + tok=self.mod_access_token, + expect_code=403, # expect failure + ) + + # have the user try to send a tombstone event + self.helper.send_state( + self.room_id, + "m.room.tombstone", + { + "body": "This room has been replaced", + "replacement_room": self.upgraded_room_id, + }, + tok=self.user_access_token, + expect_code=403, # expect failure + ) + + def test_admins_can_enable_room_encryption(self): + # have the admin try to enable room encryption + self.helper.send_state( + self.room_id, + "m.room.encryption", + {"algorithm": "m.megolm.v1.aes-sha2"}, + tok=self.admin_access_token, + expect_code=200, # expect success + ) + + def test_admins_can_send_server_acl(self): + # have the admin try to send a server ACL + self.helper.send_state( + self.room_id, + "m.room.server_acl", + { + "allow": ["*"], + "allow_ip_literals": False, + "deny": ["*.evil.com", "evil.com"], + }, + tok=self.admin_access_token, + expect_code=200, # expect success + ) + + def test_admins_can_tombstone_room(self): + # Create another room that will serve as our "upgraded room" + self.upgraded_room_id = self.helper.create_room_as( + self.admin_user_id, tok=self.admin_access_token + ) + + # have the admin try to send a tombstone event + self.helper.send_state( + self.room_id, + "m.room.tombstone", + { + "body": "This room has been replaced", + "replacement_room": self.upgraded_room_id, + }, + tok=self.admin_access_token, + expect_code=200, # expect success + ) diff --git a/tests/rest/client/v1/utils.py b/tests/rest/client/v1/utils.py index 873d5ef99c..371637618d 100644 --- a/tests/rest/client/v1/utils.py +++ b/tests/rest/client/v1/utils.py @@ -18,6 +18,7 @@ import json import time +from typing import Any, Dict, Optional import attr @@ -142,7 +143,34 @@ class RestHelper(object): return channel.json_body - def send_state(self, room_id, event_type, body, tok, expect_code=200, state_key=""): + def _read_write_state( + self, + room_id: str, + event_type: str, + body: Optional[Dict[str, Any]], + tok: str, + expect_code: int = 200, + state_key: str = "", + method: str = "GET", + ) -> Dict: + """Read or write some state from a given room + + Args: + room_id: + event_type: The type of state event + body: Body that is sent when making the request. The content of the state event. + If None, the request to the server will have an empty body + tok: The access token to use + expect_code: The HTTP code to expect in the response + state_key: + method: "GET" or "PUT" for reading or writing state, respectively + + Returns: + The response body from the server + + Raises: + AssertionError: if expect_code doesn't match the HTTP code we received + """ path = "/_matrix/client/r0/rooms/%s/state/%s/%s" % ( room_id, event_type, @@ -151,9 +179,13 @@ class RestHelper(object): if tok: path = path + "?access_token=%s" % tok - request, channel = make_request( - self.hs.get_reactor(), "PUT", path, json.dumps(body).encode("utf8") - ) + # Set request body if provided + content = b"" + if body is not None: + content = json.dumps(body).encode("utf8") + + request, channel = make_request(self.hs.get_reactor(), method, path, content) + render(request, self.resource, self.hs.get_reactor()) assert int(channel.result["code"]) == expect_code, ( @@ -163,6 +195,62 @@ class RestHelper(object): return channel.json_body + def get_state( + self, + room_id: str, + event_type: str, + tok: str, + expect_code: int = 200, + state_key: str = "", + ): + """Gets some state from a room + + Args: + room_id: + event_type: The type of state event + tok: The access token to use + expect_code: The HTTP code to expect in the response + state_key: + + Returns: + The response body from the server + + Raises: + AssertionError: if expect_code doesn't match the HTTP code we received + """ + return self._read_write_state( + room_id, event_type, None, tok, expect_code, state_key, method="GET" + ) + + def send_state( + self, + room_id: str, + event_type: str, + body: Dict[str, Any], + tok: str, + expect_code: int = 200, + state_key: str = "", + ): + """Set some state in a room + + Args: + room_id: + event_type: The type of state event + body: Body that is sent when making the request. The content of the state event. + tok: The access token to use + expect_code: The HTTP code to expect in the response + state_key: + + Returns: + The response body from the server + + Raises: + AssertionError: if expect_code doesn't match the HTTP code we received + """ + return self._read_write_state( + room_id, event_type, body, tok, expect_code, state_key, method="PUT" + ) + def upload_media( self, resource: Resource, -- cgit 1.4.1 From f1097e772042bf9bdfc3d5f71d1c37a4e13084da Mon Sep 17 00:00:00 2001 From: Zay11Zay Date: Tue, 14 Apr 2020 15:37:28 -0400 Subject: Fix the parameters of a test fixture (#7243) --- changelog.d/7243.misc | 1 + tests/rest/client/v1/test_events.py | 2 +- 2 files changed, 2 insertions(+), 1 deletion(-) create mode 100644 changelog.d/7243.misc (limited to 'tests') diff --git a/changelog.d/7243.misc b/changelog.d/7243.misc new file mode 100644 index 0000000000..a39c257a54 --- /dev/null +++ b/changelog.d/7243.misc @@ -0,0 +1 @@ +Correct the parameters of a test fixture. Contributed by Isaiah Singletary. diff --git a/tests/rest/client/v1/test_events.py b/tests/rest/client/v1/test_events.py index ffb2de1505..b54b06482b 100644 --- a/tests/rest/client/v1/test_events.py +++ b/tests/rest/client/v1/test_events.py @@ -50,7 +50,7 @@ class EventStreamPermissionsTestCase(unittest.HomeserverTestCase): return hs - def prepare(self, hs, reactor, clock): + def prepare(self, reactor, clock, hs): # register an account self.user_id = self.register_user("sid1", "pass") -- cgit 1.4.1 From a48138784ea20dd8d8a68ce5c3563da6f3fbde43 Mon Sep 17 00:00:00 2001 From: Andrew Morgan <1342360+anoadragon453@users.noreply.github.com> Date: Wed, 15 Apr 2020 13:35:29 +0100 Subject: Allow specifying the value of Accept-Language header for URL previews (#7265) --- changelog.d/7265.feature | 1 + docs/sample_config.yaml | 25 ++++++++++++ synapse/config/repository.py | 29 ++++++++++++++ synapse/rest/media/v1/preview_url_resource.py | 8 +++- tests/rest/media/v1/test_url_preview.py | 55 +++++++++++++++++++++++++++ 5 files changed, 116 insertions(+), 2 deletions(-) create mode 100644 changelog.d/7265.feature (limited to 'tests') diff --git a/changelog.d/7265.feature b/changelog.d/7265.feature new file mode 100644 index 0000000000..345b63e0b7 --- /dev/null +++ b/changelog.d/7265.feature @@ -0,0 +1 @@ +Add a config option for specifying the value of the Accept-Language HTTP header when generating URL previews. \ No newline at end of file diff --git a/docs/sample_config.yaml b/docs/sample_config.yaml index 3417813750..81dccbd997 100644 --- a/docs/sample_config.yaml +++ b/docs/sample_config.yaml @@ -859,6 +859,31 @@ media_store_path: "DATADIR/media_store" # #max_spider_size: 10M +# A list of values for the Accept-Language HTTP header used when +# downloading webpages during URL preview generation. This allows +# Synapse to specify the preferred languages that URL previews should +# be in when communicating with remote servers. +# +# Each value is a IETF language tag; a 2-3 letter identifier for a +# language, optionally followed by subtags separated by '-', specifying +# a country or region variant. +# +# Multiple values can be provided, and a weight can be added to each by +# using quality value syntax (;q=). '*' translates to any language. +# +# Defaults to "en". +# +# Example: +# +# url_preview_accept_language: +# - en-UK +# - en-US;q=0.9 +# - fr;q=0.8 +# - *;q=0.7 +# +url_preview_accept_language: +# - en + ## Captcha ## # See docs/CAPTCHA_SETUP for full details of configuring this. diff --git a/synapse/config/repository.py b/synapse/config/repository.py index 7d2dd27fd0..7193ea1114 100644 --- a/synapse/config/repository.py +++ b/synapse/config/repository.py @@ -192,6 +192,10 @@ class ContentRepositoryConfig(Config): self.url_preview_url_blacklist = config.get("url_preview_url_blacklist", ()) + self.url_preview_accept_language = config.get( + "url_preview_accept_language" + ) or ["en"] + def generate_config_section(self, data_dir_path, **kwargs): media_store = os.path.join(data_dir_path, "media_store") uploads_path = os.path.join(data_dir_path, "uploads") @@ -329,6 +333,31 @@ class ContentRepositoryConfig(Config): # The largest allowed URL preview spidering size in bytes # #max_spider_size: 10M + + # A list of values for the Accept-Language HTTP header used when + # downloading webpages during URL preview generation. This allows + # Synapse to specify the preferred languages that URL previews should + # be in when communicating with remote servers. + # + # Each value is a IETF language tag; a 2-3 letter identifier for a + # language, optionally followed by subtags separated by '-', specifying + # a country or region variant. + # + # Multiple values can be provided, and a weight can be added to each by + # using quality value syntax (;q=). '*' translates to any language. + # + # Defaults to "en". + # + # Example: + # + # url_preview_accept_language: + # - en-UK + # - en-US;q=0.9 + # - fr;q=0.8 + # - *;q=0.7 + # + url_preview_accept_language: + # - en """ % locals() ) diff --git a/synapse/rest/media/v1/preview_url_resource.py b/synapse/rest/media/v1/preview_url_resource.py index c46676f8fc..f68e18ea8a 100644 --- a/synapse/rest/media/v1/preview_url_resource.py +++ b/synapse/rest/media/v1/preview_url_resource.py @@ -86,6 +86,7 @@ class PreviewUrlResource(DirectServeResource): self.media_storage = media_storage self.url_preview_url_blacklist = hs.config.url_preview_url_blacklist + self.url_preview_accept_language = hs.config.url_preview_accept_language # memory cache mapping urls to an ObservableDeferred returning # JSON-encoded OG metadata @@ -315,9 +316,12 @@ class PreviewUrlResource(DirectServeResource): with self.media_storage.store_into_file(file_info) as (f, fname, finish): try: - logger.debug("Trying to get url '%s'", url) + logger.debug("Trying to get preview for url '%s'", url) length, headers, uri, code = await self.client.get_file( - url, output_stream=f, max_size=self.max_spider_size + url, + output_stream=f, + max_size=self.max_spider_size, + headers={"Accept-Language": self.url_preview_accept_language}, ) except SynapseError: # Pass SynapseErrors through directly, so that the servlet diff --git a/tests/rest/media/v1/test_url_preview.py b/tests/rest/media/v1/test_url_preview.py index 852b8ab11c..2826211f32 100644 --- a/tests/rest/media/v1/test_url_preview.py +++ b/tests/rest/media/v1/test_url_preview.py @@ -74,6 +74,12 @@ class URLPreviewTests(unittest.HomeserverTestCase): ) config["url_preview_ip_range_whitelist"] = ("1.1.1.1",) config["url_preview_url_blacklist"] = [] + config["url_preview_accept_language"] = [ + "en-UK", + "en-US;q=0.9", + "fr;q=0.8", + "*;q=0.7", + ] self.storage_path = self.mktemp() self.media_store_path = self.mktemp() @@ -507,3 +513,52 @@ class URLPreviewTests(unittest.HomeserverTestCase): self.pump() self.assertEqual(channel.code, 200) self.assertEqual(channel.json_body, {}) + + def test_accept_language_config_option(self): + """ + Accept-Language header is sent to the remote server + """ + self.lookups["example.com"] = [(IPv4Address, "8.8.8.8")] + + # Build and make a request to the server + request, channel = self.make_request( + "GET", "url_preview?url=http://example.com", shorthand=False + ) + request.render(self.preview_url) + self.pump() + + # Extract Synapse's tcp client + client = self.reactor.tcpClients[0][2].buildProtocol(None) + + # Build a fake remote server to reply with + server = AccumulatingProtocol() + + # Connect the two together + server.makeConnection(FakeTransport(client, self.reactor)) + client.makeConnection(FakeTransport(server, self.reactor)) + + # Tell Synapse that it has received some data from the remote server + client.dataReceived( + b"HTTP/1.0 200 OK\r\nContent-Length: %d\r\nContent-Type: text/html\r\n\r\n" + % (len(self.end_content),) + + self.end_content + ) + + # Move the reactor along until we get a response on our original channel + self.pump() + self.assertEqual(channel.code, 200) + self.assertEqual( + channel.json_body, {"og:title": "~matrix~", "og:description": "hi"} + ) + + # Check that the server received the Accept-Language header as part + # of the request from Synapse + self.assertIn( + ( + b"Accept-Language: en-UK\r\n" + b"Accept-Language: en-US;q=0.9\r\n" + b"Accept-Language: fr;q=0.8\r\n" + b"Accept-Language: *;q=0.7" + ), + server.data, + ) -- cgit 1.4.1 From eed7c5b89eee6951ac17861b1695817470bace36 Mon Sep 17 00:00:00 2001 From: Patrick Cloke Date: Wed, 15 Apr 2020 12:40:18 -0400 Subject: Convert auth handler to async/await (#7261) --- changelog.d/7261.misc | 1 + synapse/handlers/auth.py | 173 ++++++++++++++++++--------------------- synapse/handlers/device.py | 12 ++- synapse/handlers/register.py | 28 +++++-- synapse/handlers/set_password.py | 13 ++- synapse/module_api/__init__.py | 6 +- tests/api/test_auth.py | 64 +++++++++------ tests/handlers/test_auth.py | 80 +++++++++++------- tests/handlers/test_register.py | 4 +- tests/utils.py | 13 ++- 10 files changed, 224 insertions(+), 170 deletions(-) create mode 100644 changelog.d/7261.misc (limited to 'tests') diff --git a/changelog.d/7261.misc b/changelog.d/7261.misc new file mode 100644 index 0000000000..88165f0105 --- /dev/null +++ b/changelog.d/7261.misc @@ -0,0 +1 @@ +Convert auth handler to async/await. diff --git a/synapse/handlers/auth.py b/synapse/handlers/auth.py index fbfbd44a2e..0aae929ecc 100644 --- a/synapse/handlers/auth.py +++ b/synapse/handlers/auth.py @@ -18,14 +18,12 @@ import logging import time import unicodedata import urllib.parse -from typing import Any, Dict, Iterable, List, Optional +from typing import Any, Callable, Dict, Iterable, List, Optional, Tuple, Union import attr import bcrypt # type: ignore[import] import pymacaroons -from twisted.internet import defer - import synapse.util.stringutils as stringutils from synapse.api.constants import LoginType from synapse.api.errors import ( @@ -170,15 +168,14 @@ class AuthHandler(BaseHandler): # cast to tuple for use with str.startswith self._whitelisted_sso_clients = tuple(hs.config.sso_client_whitelist) - @defer.inlineCallbacks - def validate_user_via_ui_auth( + async def validate_user_via_ui_auth( self, requester: Requester, request: SynapseRequest, request_body: Dict[str, Any], clientip: str, description: str, - ): + ) -> dict: """ Checks that the user is who they claim to be, via a UI auth. @@ -199,7 +196,7 @@ class AuthHandler(BaseHandler): describes the operation happening on their account. Returns: - defer.Deferred[dict]: the parameters for this request (which may + The parameters for this request (which may have been given only in a previous call). Raises: @@ -229,7 +226,7 @@ class AuthHandler(BaseHandler): flows = [[login_type] for login_type in self._supported_ui_auth_types] try: - result, params, _ = yield self.check_auth( + result, params, _ = await self.check_auth( flows, request, request_body, clientip, description ) except LoginError: @@ -268,15 +265,14 @@ class AuthHandler(BaseHandler): """ return self.checkers.keys() - @defer.inlineCallbacks - def check_auth( + async def check_auth( self, flows: List[List[str]], request: SynapseRequest, clientdict: Dict[str, Any], clientip: str, description: str, - ): + ) -> Tuple[dict, dict, str]: """ Takes a dictionary sent by the client in the login / registration protocol and handles the User-Interactive Auth flow. @@ -306,8 +302,7 @@ class AuthHandler(BaseHandler): describes the operation happening on their account. Returns: - defer.Deferred[dict, dict, str]: a deferred tuple of - (creds, params, session_id). + A tuple of (creds, params, session_id). 'creds' contains the authenticated credentials of each stage. @@ -380,7 +375,7 @@ class AuthHandler(BaseHandler): if "type" in authdict: login_type = authdict["type"] # type: str try: - result = yield self._check_auth_dict(authdict, clientip) + result = await self._check_auth_dict(authdict, clientip) if result: creds[login_type] = result self._save_session(session) @@ -419,8 +414,9 @@ class AuthHandler(BaseHandler): ret.update(errordict) raise InteractiveAuthIncompleteError(ret) - @defer.inlineCallbacks - def add_oob_auth(self, stagetype: str, authdict: Dict[str, Any], clientip: str): + async def add_oob_auth( + self, stagetype: str, authdict: Dict[str, Any], clientip: str + ) -> bool: """ Adds the result of out-of-band authentication into an existing auth session. Currently used for adding the result of fallback auth. @@ -435,7 +431,7 @@ class AuthHandler(BaseHandler): sess["creds"] = {} creds = sess["creds"] - result = yield self.checkers[stagetype].check_auth(authdict, clientip) + result = await self.checkers[stagetype].check_auth(authdict, clientip) if result: creds[stagetype] = result self._save_session(sess) @@ -489,8 +485,9 @@ class AuthHandler(BaseHandler): sess = self._get_session_info(session_id) return sess.setdefault("serverdict", {}).get(key, default) - @defer.inlineCallbacks - def _check_auth_dict(self, authdict: Dict[str, Any], clientip: str): + async def _check_auth_dict( + self, authdict: Dict[str, Any], clientip: str + ) -> Union[Dict[str, Any], str]: """Attempt to validate the auth dict provided by a client Args: @@ -498,7 +495,7 @@ class AuthHandler(BaseHandler): clientip: IP address of the client Returns: - Deferred: result of the stage verification. + Result of the stage verification. Raises: StoreError if there was a problem accessing the database @@ -508,7 +505,7 @@ class AuthHandler(BaseHandler): login_type = authdict["type"] checker = self.checkers.get(login_type) if checker is not None: - res = yield checker.check_auth(authdict, clientip=clientip) + res = await checker.check_auth(authdict, clientip=clientip) return res # build a v1-login-style dict out of the authdict and fall back to the @@ -518,7 +515,7 @@ class AuthHandler(BaseHandler): if user_id is None: raise SynapseError(400, "", Codes.MISSING_PARAM) - (canonical_id, callback) = yield self.validate_login(user_id, authdict) + (canonical_id, callback) = await self.validate_login(user_id, authdict) return canonical_id def _get_params_recaptcha(self) -> dict: @@ -584,8 +581,7 @@ class AuthHandler(BaseHandler): return self.sessions[session_id] - @defer.inlineCallbacks - def get_access_token_for_user_id( + async def get_access_token_for_user_id( self, user_id: str, device_id: Optional[str], valid_until_ms: Optional[int] ): """ @@ -615,10 +611,10 @@ class AuthHandler(BaseHandler): ) logger.info("Logging in user %s on device %s%s", user_id, device_id, fmt_expiry) - yield self.auth.check_auth_blocking(user_id) + await self.auth.check_auth_blocking(user_id) access_token = self.macaroon_gen.generate_access_token(user_id) - yield self.store.add_access_token_to_user( + await self.store.add_access_token_to_user( user_id, access_token, device_id, valid_until_ms ) @@ -628,15 +624,14 @@ class AuthHandler(BaseHandler): # device, so we double-check it here. if device_id is not None: try: - yield self.store.get_device(user_id, device_id) + await self.store.get_device(user_id, device_id) except StoreError: - yield self.store.delete_access_token(access_token) + await self.store.delete_access_token(access_token) raise StoreError(400, "Login raced against device deletion") return access_token - @defer.inlineCallbacks - def check_user_exists(self, user_id: str): + async def check_user_exists(self, user_id: str) -> Optional[str]: """ Checks to see if a user with the given id exists. Will check case insensitively, but return None if there are multiple inexact matches. @@ -645,25 +640,25 @@ class AuthHandler(BaseHandler): user_id: complete @user:id Returns: - defer.Deferred: (unicode) canonical_user_id, or None if zero or - multiple matches + The canonical_user_id, or None if zero or multiple matches """ - res = yield self._find_user_id_and_pwd_hash(user_id) + res = await self._find_user_id_and_pwd_hash(user_id) if res is not None: return res[0] return None - @defer.inlineCallbacks - def _find_user_id_and_pwd_hash(self, user_id: str): + async def _find_user_id_and_pwd_hash( + self, user_id: str + ) -> Optional[Tuple[str, str]]: """Checks to see if a user with the given id exists. Will check case insensitively, but will return None if there are multiple inexact matches. Returns: - tuple: A 2-tuple of `(canonical_user_id, password_hash)` - None: if there is not exactly one match + A 2-tuple of `(canonical_user_id, password_hash)` or `None` + if there is not exactly one match """ - user_infos = yield self.store.get_users_by_id_case_insensitive(user_id) + user_infos = await self.store.get_users_by_id_case_insensitive(user_id) result = None if not user_infos: @@ -696,8 +691,9 @@ class AuthHandler(BaseHandler): """ return self._supported_login_types - @defer.inlineCallbacks - def validate_login(self, username: str, login_submission: Dict[str, Any]): + async def validate_login( + self, username: str, login_submission: Dict[str, Any] + ) -> Tuple[str, Optional[Callable[[Dict[str, str]], None]]]: """Authenticates the user for the /login API Also used by the user-interactive auth flow to validate @@ -708,7 +704,7 @@ class AuthHandler(BaseHandler): login_submission: the whole of the login submission (including 'type' and other relevant fields) Returns: - Deferred[str, func]: canonical user id, and optional callback + A tuple of the canonical user id, and optional callback to be called once the access token and device id are issued Raises: StoreError if there was a problem accessing the database @@ -737,7 +733,7 @@ class AuthHandler(BaseHandler): for provider in self.password_providers: if hasattr(provider, "check_password") and login_type == LoginType.PASSWORD: known_login_type = True - is_valid = yield provider.check_password(qualified_user_id, password) + is_valid = await provider.check_password(qualified_user_id, password) if is_valid: return qualified_user_id, None @@ -769,7 +765,7 @@ class AuthHandler(BaseHandler): % (login_type, missing_fields), ) - result = yield provider.check_auth(username, login_type, login_dict) + result = await provider.check_auth(username, login_type, login_dict) if result: if isinstance(result, str): result = (result, None) @@ -778,8 +774,8 @@ class AuthHandler(BaseHandler): if login_type == LoginType.PASSWORD and self.hs.config.password_localdb_enabled: known_login_type = True - canonical_user_id = yield self._check_local_password( - qualified_user_id, password + canonical_user_id = await self._check_local_password( + qualified_user_id, password # type: ignore ) if canonical_user_id: @@ -792,8 +788,9 @@ class AuthHandler(BaseHandler): # login, it turns all LoginErrors into a 401 anyway. raise LoginError(403, "Invalid password", errcode=Codes.FORBIDDEN) - @defer.inlineCallbacks - def check_password_provider_3pid(self, medium: str, address: str, password: str): + async def check_password_provider_3pid( + self, medium: str, address: str, password: str + ) -> Tuple[Optional[str], Optional[Callable[[Dict[str, str]], None]]]: """Check if a password provider is able to validate a thirdparty login Args: @@ -802,9 +799,8 @@ class AuthHandler(BaseHandler): password: The password of the user. Returns: - Deferred[(str|None, func|None)]: A tuple of `(user_id, - callback)`. If authentication is successful, `user_id` is a `str` - containing the authenticated, canonical user ID. `callback` is + A tuple of `(user_id, callback)`. If authentication is successful, + `user_id`is the authenticated, canonical user ID. `callback` is then either a function to be later run after the server has completed login/registration, or `None`. If authentication was unsuccessful, `user_id` and `callback` are both `None`. @@ -816,7 +812,7 @@ class AuthHandler(BaseHandler): # success, to a str (which is the user_id) or a tuple of # (user_id, callback_func), where callback_func should be run # after we've finished everything else - result = yield provider.check_3pid_auth(medium, address, password) + result = await provider.check_3pid_auth(medium, address, password) if result: # Check if the return value is a str or a tuple if isinstance(result, str): @@ -826,8 +822,7 @@ class AuthHandler(BaseHandler): return None, None - @defer.inlineCallbacks - def _check_local_password(self, user_id: str, password: str): + async def _check_local_password(self, user_id: str, password: str) -> Optional[str]: """Authenticate a user against the local password database. user_id is checked case insensitively, but will return None if there are @@ -837,28 +832,26 @@ class AuthHandler(BaseHandler): user_id: complete @user:id password: the provided password Returns: - Deferred[unicode] the canonical_user_id, or Deferred[None] if - unknown user/bad password + The canonical_user_id, or None if unknown user/bad password """ - lookupres = yield self._find_user_id_and_pwd_hash(user_id) + lookupres = await self._find_user_id_and_pwd_hash(user_id) if not lookupres: return None (user_id, password_hash) = lookupres # If the password hash is None, the account has likely been deactivated if not password_hash: - deactivated = yield self.store.get_user_deactivated_status(user_id) + deactivated = await self.store.get_user_deactivated_status(user_id) if deactivated: raise UserDeactivatedError("This account has been deactivated") - result = yield self.validate_hash(password, password_hash) + result = await self.validate_hash(password, password_hash) if not result: logger.warning("Failed password login for user %s", user_id) return None return user_id - @defer.inlineCallbacks - def validate_short_term_login_token_and_get_user_id(self, login_token: str): + async def validate_short_term_login_token_and_get_user_id(self, login_token: str): auth_api = self.hs.get_auth() user_id = None try: @@ -868,26 +861,23 @@ class AuthHandler(BaseHandler): except Exception: raise AuthError(403, "Invalid token", errcode=Codes.FORBIDDEN) - yield self.auth.check_auth_blocking(user_id) + await self.auth.check_auth_blocking(user_id) return user_id - @defer.inlineCallbacks - def delete_access_token(self, access_token: str): + async def delete_access_token(self, access_token: str): """Invalidate a single access token Args: access_token: access token to be deleted - Returns: - Deferred """ - user_info = yield self.auth.get_user_by_access_token(access_token) - yield self.store.delete_access_token(access_token) + user_info = await self.auth.get_user_by_access_token(access_token) + await self.store.delete_access_token(access_token) # see if any of our auth providers want to know about this for provider in self.password_providers: if hasattr(provider, "on_logged_out"): - yield provider.on_logged_out( + await provider.on_logged_out( user_id=str(user_info["user"]), device_id=user_info["device_id"], access_token=access_token, @@ -895,12 +885,11 @@ class AuthHandler(BaseHandler): # delete pushers associated with this access token if user_info["token_id"] is not None: - yield self.hs.get_pusherpool().remove_pushers_by_access_token( + await self.hs.get_pusherpool().remove_pushers_by_access_token( str(user_info["user"]), (user_info["token_id"],) ) - @defer.inlineCallbacks - def delete_access_tokens_for_user( + async def delete_access_tokens_for_user( self, user_id: str, except_token_id: Optional[str] = None, @@ -914,10 +903,8 @@ class AuthHandler(BaseHandler): device_id: ID of device the tokens are associated with. If None, tokens associated with any device (or no device) will be deleted - Returns: - Deferred """ - tokens_and_devices = yield self.store.user_delete_access_tokens( + tokens_and_devices = await self.store.user_delete_access_tokens( user_id, except_token_id=except_token_id, device_id=device_id ) @@ -925,17 +912,18 @@ class AuthHandler(BaseHandler): for provider in self.password_providers: if hasattr(provider, "on_logged_out"): for token, token_id, device_id in tokens_and_devices: - yield provider.on_logged_out( + await provider.on_logged_out( user_id=user_id, device_id=device_id, access_token=token ) # delete pushers associated with the access tokens - yield self.hs.get_pusherpool().remove_pushers_by_access_token( + await self.hs.get_pusherpool().remove_pushers_by_access_token( user_id, (token_id for _, token_id, _ in tokens_and_devices) ) - @defer.inlineCallbacks - def add_threepid(self, user_id: str, medium: str, address: str, validated_at: int): + async def add_threepid( + self, user_id: str, medium: str, address: str, validated_at: int + ): # check if medium has a valid value if medium not in ["email", "msisdn"]: raise SynapseError( @@ -956,14 +944,13 @@ class AuthHandler(BaseHandler): if medium == "email": address = address.lower() - yield self.store.user_add_threepid( + await self.store.user_add_threepid( user_id, medium, address, validated_at, self.hs.get_clock().time_msec() ) - @defer.inlineCallbacks - def delete_threepid( + async def delete_threepid( self, user_id: str, medium: str, address: str, id_server: Optional[str] = None - ): + ) -> bool: """Attempts to unbind the 3pid on the identity servers and deletes it from the local database. @@ -976,7 +963,7 @@ class AuthHandler(BaseHandler): identity server specified when binding (if known). Returns: - Deferred[bool]: Returns True if successfully unbound the 3pid on + Returns True if successfully unbound the 3pid on the identity server, False if identity server doesn't support the unbind API. """ @@ -986,11 +973,11 @@ class AuthHandler(BaseHandler): address = address.lower() identity_handler = self.hs.get_handlers().identity_handler - result = yield identity_handler.try_unbind_threepid( + result = await identity_handler.try_unbind_threepid( user_id, {"medium": medium, "address": address, "id_server": id_server} ) - yield self.store.user_delete_threepid(user_id, medium, address) + await self.store.user_delete_threepid(user_id, medium, address) return result def _save_session(self, session: Dict[str, Any]) -> None: @@ -1000,14 +987,14 @@ class AuthHandler(BaseHandler): session["last_used"] = self.hs.get_clock().time_msec() self.sessions[session["id"]] = session - def hash(self, password: str): + async def hash(self, password: str) -> str: """Computes a secure hash of password. Args: password: Password to hash. Returns: - Deferred(unicode): Hashed password. + Hashed password. """ def _do_hash(): @@ -1019,9 +1006,11 @@ class AuthHandler(BaseHandler): bcrypt.gensalt(self.bcrypt_rounds), ).decode("ascii") - return defer_to_thread(self.hs.get_reactor(), _do_hash) + return await defer_to_thread(self.hs.get_reactor(), _do_hash) - def validate_hash(self, password: str, stored_hash: bytes): + async def validate_hash( + self, password: str, stored_hash: Union[bytes, str] + ) -> bool: """Validates that self.hash(password) == stored_hash. Args: @@ -1029,7 +1018,7 @@ class AuthHandler(BaseHandler): stored_hash: Expected hash value. Returns: - Deferred(bool): Whether self.hash(password) == stored_hash. + Whether self.hash(password) == stored_hash. """ def _do_validate_hash(): @@ -1045,9 +1034,9 @@ class AuthHandler(BaseHandler): if not isinstance(stored_hash, bytes): stored_hash = stored_hash.encode("ascii") - return defer_to_thread(self.hs.get_reactor(), _do_validate_hash) + return await defer_to_thread(self.hs.get_reactor(), _do_validate_hash) else: - return defer.succeed(False) + return False def start_sso_ui_auth(self, redirect_url: str, session_id: str) -> str: """ diff --git a/synapse/handlers/device.py b/synapse/handlers/device.py index 993499f446..9bd941b5a0 100644 --- a/synapse/handlers/device.py +++ b/synapse/handlers/device.py @@ -338,8 +338,10 @@ class DeviceHandler(DeviceWorkerHandler): else: raise - yield self._auth_handler.delete_access_tokens_for_user( - user_id, device_id=device_id + yield defer.ensureDeferred( + self._auth_handler.delete_access_tokens_for_user( + user_id, device_id=device_id + ) ) yield self.store.delete_e2e_keys_by_device(user_id=user_id, device_id=device_id) @@ -391,8 +393,10 @@ class DeviceHandler(DeviceWorkerHandler): # Delete access tokens and e2e keys for each device. Not optimised as it is not # considered as part of a critical path. for device_id in device_ids: - yield self._auth_handler.delete_access_tokens_for_user( - user_id, device_id=device_id + yield defer.ensureDeferred( + self._auth_handler.delete_access_tokens_for_user( + user_id, device_id=device_id + ) ) yield self.store.delete_e2e_keys_by_device( user_id=user_id, device_id=device_id diff --git a/synapse/handlers/register.py b/synapse/handlers/register.py index 7ffc194f0c..3a65b46ecd 100644 --- a/synapse/handlers/register.py +++ b/synapse/handlers/register.py @@ -166,7 +166,9 @@ class RegistrationHandler(BaseHandler): yield self.auth.check_auth_blocking(threepid=threepid) password_hash = None if password: - password_hash = yield self._auth_handler.hash(password) + password_hash = yield defer.ensureDeferred( + self._auth_handler.hash(password) + ) if localpart is not None: yield self.check_username(localpart, guest_access_token=guest_access_token) @@ -540,8 +542,10 @@ class RegistrationHandler(BaseHandler): user_id, ["guest = true"] ) else: - access_token = yield self._auth_handler.get_access_token_for_user_id( - user_id, device_id=device_id, valid_until_ms=valid_until_ms + access_token = yield defer.ensureDeferred( + self._auth_handler.get_access_token_for_user_id( + user_id, device_id=device_id, valid_until_ms=valid_until_ms + ) ) return (device_id, access_token) @@ -617,8 +621,13 @@ class RegistrationHandler(BaseHandler): logger.info("Can't add incomplete 3pid") return - yield self._auth_handler.add_threepid( - user_id, threepid["medium"], threepid["address"], threepid["validated_at"] + yield defer.ensureDeferred( + self._auth_handler.add_threepid( + user_id, + threepid["medium"], + threepid["address"], + threepid["validated_at"], + ) ) # And we add an email pusher for them by default, but only @@ -670,6 +679,11 @@ class RegistrationHandler(BaseHandler): return None raise - yield self._auth_handler.add_threepid( - user_id, threepid["medium"], threepid["address"], threepid["validated_at"] + yield defer.ensureDeferred( + self._auth_handler.add_threepid( + user_id, + threepid["medium"], + threepid["address"], + threepid["validated_at"], + ) ) diff --git a/synapse/handlers/set_password.py b/synapse/handlers/set_password.py index 7d1263caf2..63d8f9aa0d 100644 --- a/synapse/handlers/set_password.py +++ b/synapse/handlers/set_password.py @@ -15,8 +15,6 @@ import logging from typing import Optional -from twisted.internet import defer - from synapse.api.errors import Codes, StoreError, SynapseError from synapse.types import Requester @@ -34,8 +32,7 @@ class SetPasswordHandler(BaseHandler): self._device_handler = hs.get_device_handler() self._password_policy_handler = hs.get_password_policy_handler() - @defer.inlineCallbacks - def set_password( + async def set_password( self, user_id: str, new_password: str, @@ -46,10 +43,10 @@ class SetPasswordHandler(BaseHandler): raise SynapseError(403, "Password change disabled", errcode=Codes.FORBIDDEN) self._password_policy_handler.validate_password(new_password) - password_hash = yield self._auth_handler.hash(new_password) + password_hash = await self._auth_handler.hash(new_password) try: - yield self.store.user_set_password_hash(user_id, password_hash) + await self.store.user_set_password_hash(user_id, password_hash) except StoreError as e: if e.code == 404: raise SynapseError(404, "Unknown user", Codes.NOT_FOUND) @@ -61,12 +58,12 @@ class SetPasswordHandler(BaseHandler): except_access_token_id = requester.access_token_id if requester else None # First delete all of their other devices. - yield self._device_handler.delete_all_devices_for_user( + await self._device_handler.delete_all_devices_for_user( user_id, except_device_id=except_device_id ) # and now delete any access tokens which weren't associated with # devices (or were associated with this device). - yield self._auth_handler.delete_access_tokens_for_user( + await self._auth_handler.delete_access_tokens_for_user( user_id, except_token_id=except_access_token_id ) diff --git a/synapse/module_api/__init__.py b/synapse/module_api/__init__.py index afc3598e11..d678c0eb9b 100644 --- a/synapse/module_api/__init__.py +++ b/synapse/module_api/__init__.py @@ -86,7 +86,7 @@ class ModuleApi(object): Deferred[str|None]: Canonical (case-corrected) user_id, or None if the user is not registered. """ - return self._auth_handler.check_user_exists(user_id) + return defer.ensureDeferred(self._auth_handler.check_user_exists(user_id)) @defer.inlineCallbacks def register(self, localpart, displayname=None, emails=[]): @@ -196,7 +196,9 @@ class ModuleApi(object): yield self._hs.get_device_handler().delete_device(user_id, device_id) else: # no associated device. Just delete the access token. - yield self._auth_handler.delete_access_token(access_token) + yield defer.ensureDeferred( + self._auth_handler.delete_access_token(access_token) + ) def run_db_interaction(self, desc, func, *args, **kwargs): """Run a function with a database connection diff --git a/tests/api/test_auth.py b/tests/api/test_auth.py index 6121efcfa9..cc0b10e7f6 100644 --- a/tests/api/test_auth.py +++ b/tests/api/test_auth.py @@ -68,7 +68,7 @@ class AuthTestCase(unittest.TestCase): request = Mock(args={}) request.args[b"access_token"] = [self.test_token] request.requestHeaders.getRawHeaders = mock_getRawHeaders() - requester = yield self.auth.get_user_by_req(request) + requester = yield defer.ensureDeferred(self.auth.get_user_by_req(request)) self.assertEquals(requester.user.to_string(), self.test_user) def test_get_user_by_req_user_bad_token(self): @@ -105,7 +105,7 @@ class AuthTestCase(unittest.TestCase): request.getClientIP.return_value = "127.0.0.1" request.args[b"access_token"] = [self.test_token] request.requestHeaders.getRawHeaders = mock_getRawHeaders() - requester = yield self.auth.get_user_by_req(request) + requester = yield defer.ensureDeferred(self.auth.get_user_by_req(request)) self.assertEquals(requester.user.to_string(), self.test_user) @defer.inlineCallbacks @@ -125,7 +125,7 @@ class AuthTestCase(unittest.TestCase): request.getClientIP.return_value = "192.168.10.10" request.args[b"access_token"] = [self.test_token] request.requestHeaders.getRawHeaders = mock_getRawHeaders() - requester = yield self.auth.get_user_by_req(request) + requester = yield defer.ensureDeferred(self.auth.get_user_by_req(request)) self.assertEquals(requester.user.to_string(), self.test_user) def test_get_user_by_req_appservice_valid_token_bad_ip(self): @@ -188,7 +188,7 @@ class AuthTestCase(unittest.TestCase): request.args[b"access_token"] = [self.test_token] request.args[b"user_id"] = [masquerading_user_id] request.requestHeaders.getRawHeaders = mock_getRawHeaders() - requester = yield self.auth.get_user_by_req(request) + requester = yield defer.ensureDeferred(self.auth.get_user_by_req(request)) self.assertEquals( requester.user.to_string(), masquerading_user_id.decode("utf8") ) @@ -225,7 +225,9 @@ class AuthTestCase(unittest.TestCase): macaroon.add_first_party_caveat("gen = 1") macaroon.add_first_party_caveat("type = access") macaroon.add_first_party_caveat("user_id = %s" % (user_id,)) - user_info = yield self.auth.get_user_by_access_token(macaroon.serialize()) + user_info = yield defer.ensureDeferred( + self.auth.get_user_by_access_token(macaroon.serialize()) + ) user = user_info["user"] self.assertEqual(UserID.from_string(user_id), user) @@ -250,7 +252,9 @@ class AuthTestCase(unittest.TestCase): macaroon.add_first_party_caveat("guest = true") serialized = macaroon.serialize() - user_info = yield self.auth.get_user_by_access_token(serialized) + user_info = yield defer.ensureDeferred( + self.auth.get_user_by_access_token(serialized) + ) user = user_info["user"] is_guest = user_info["is_guest"] self.assertEqual(UserID.from_string(user_id), user) @@ -260,10 +264,13 @@ class AuthTestCase(unittest.TestCase): @defer.inlineCallbacks def test_cannot_use_regular_token_as_guest(self): USER_ID = "@percy:matrix.org" - self.store.add_access_token_to_user = Mock() + self.store.add_access_token_to_user = Mock(return_value=defer.succeed(None)) + self.store.get_device = Mock(return_value=defer.succeed(None)) - token = yield self.hs.handlers.auth_handler.get_access_token_for_user_id( - USER_ID, "DEVICE", valid_until_ms=None + token = yield defer.ensureDeferred( + self.hs.handlers.auth_handler.get_access_token_for_user_id( + USER_ID, "DEVICE", valid_until_ms=None + ) ) self.store.add_access_token_to_user.assert_called_with( USER_ID, token, "DEVICE", None @@ -286,7 +293,9 @@ class AuthTestCase(unittest.TestCase): request = Mock(args={}) request.args[b"access_token"] = [token.encode("ascii")] request.requestHeaders.getRawHeaders = mock_getRawHeaders() - requester = yield self.auth.get_user_by_req(request, allow_guest=True) + requester = yield defer.ensureDeferred( + self.auth.get_user_by_req(request, allow_guest=True) + ) self.assertEqual(UserID.from_string(USER_ID), requester.user) self.assertFalse(requester.is_guest) @@ -301,7 +310,9 @@ class AuthTestCase(unittest.TestCase): request.requestHeaders.getRawHeaders = mock_getRawHeaders() with self.assertRaises(InvalidClientCredentialsError) as cm: - yield self.auth.get_user_by_req(request, allow_guest=True) + yield defer.ensureDeferred( + self.auth.get_user_by_req(request, allow_guest=True) + ) self.assertEqual(401, cm.exception.code) self.assertEqual("Guest access token used for regular user", cm.exception.msg) @@ -316,7 +327,7 @@ class AuthTestCase(unittest.TestCase): small_number_of_users = 1 # Ensure no error thrown - yield self.auth.check_auth_blocking() + yield defer.ensureDeferred(self.auth.check_auth_blocking()) self.hs.config.limit_usage_by_mau = True @@ -325,7 +336,7 @@ class AuthTestCase(unittest.TestCase): ) with self.assertRaises(ResourceLimitError) as e: - yield self.auth.check_auth_blocking() + yield defer.ensureDeferred(self.auth.check_auth_blocking()) self.assertEquals(e.exception.admin_contact, self.hs.config.admin_contact) self.assertEquals(e.exception.errcode, Codes.RESOURCE_LIMIT_EXCEEDED) self.assertEquals(e.exception.code, 403) @@ -334,7 +345,7 @@ class AuthTestCase(unittest.TestCase): self.store.get_monthly_active_count = Mock( return_value=defer.succeed(small_number_of_users) ) - yield self.auth.check_auth_blocking() + yield defer.ensureDeferred(self.auth.check_auth_blocking()) @defer.inlineCallbacks def test_blocking_mau__depending_on_user_type(self): @@ -343,15 +354,19 @@ class AuthTestCase(unittest.TestCase): self.store.get_monthly_active_count = Mock(return_value=defer.succeed(100)) # Support users allowed - yield self.auth.check_auth_blocking(user_type=UserTypes.SUPPORT) + yield defer.ensureDeferred( + self.auth.check_auth_blocking(user_type=UserTypes.SUPPORT) + ) self.store.get_monthly_active_count = Mock(return_value=defer.succeed(100)) # Bots not allowed with self.assertRaises(ResourceLimitError): - yield self.auth.check_auth_blocking(user_type=UserTypes.BOT) + yield defer.ensureDeferred( + self.auth.check_auth_blocking(user_type=UserTypes.BOT) + ) self.store.get_monthly_active_count = Mock(return_value=defer.succeed(100)) # Real users not allowed with self.assertRaises(ResourceLimitError): - yield self.auth.check_auth_blocking() + yield defer.ensureDeferred(self.auth.check_auth_blocking()) @defer.inlineCallbacks def test_reserved_threepid(self): @@ -362,21 +377,22 @@ class AuthTestCase(unittest.TestCase): unknown_threepid = {"medium": "email", "address": "unreserved@server.com"} self.hs.config.mau_limits_reserved_threepids = [threepid] - yield self.store.register_user(user_id="user1", password_hash=None) with self.assertRaises(ResourceLimitError): - yield self.auth.check_auth_blocking() + yield defer.ensureDeferred(self.auth.check_auth_blocking()) with self.assertRaises(ResourceLimitError): - yield self.auth.check_auth_blocking(threepid=unknown_threepid) + yield defer.ensureDeferred( + self.auth.check_auth_blocking(threepid=unknown_threepid) + ) - yield self.auth.check_auth_blocking(threepid=threepid) + yield defer.ensureDeferred(self.auth.check_auth_blocking(threepid=threepid)) @defer.inlineCallbacks def test_hs_disabled(self): self.hs.config.hs_disabled = True self.hs.config.hs_disabled_message = "Reason for being disabled" with self.assertRaises(ResourceLimitError) as e: - yield self.auth.check_auth_blocking() + yield defer.ensureDeferred(self.auth.check_auth_blocking()) self.assertEquals(e.exception.admin_contact, self.hs.config.admin_contact) self.assertEquals(e.exception.errcode, Codes.RESOURCE_LIMIT_EXCEEDED) self.assertEquals(e.exception.code, 403) @@ -393,7 +409,7 @@ class AuthTestCase(unittest.TestCase): self.hs.config.hs_disabled = True self.hs.config.hs_disabled_message = "Reason for being disabled" with self.assertRaises(ResourceLimitError) as e: - yield self.auth.check_auth_blocking() + yield defer.ensureDeferred(self.auth.check_auth_blocking()) self.assertEquals(e.exception.admin_contact, self.hs.config.admin_contact) self.assertEquals(e.exception.errcode, Codes.RESOURCE_LIMIT_EXCEEDED) self.assertEquals(e.exception.code, 403) @@ -404,4 +420,4 @@ class AuthTestCase(unittest.TestCase): user = "@user:server" self.hs.config.server_notices_mxid = user self.hs.config.hs_disabled_message = "Reason for being disabled" - yield self.auth.check_auth_blocking(user) + yield defer.ensureDeferred(self.auth.check_auth_blocking(user)) diff --git a/tests/handlers/test_auth.py b/tests/handlers/test_auth.py index b03103d96f..52c4ac8b11 100644 --- a/tests/handlers/test_auth.py +++ b/tests/handlers/test_auth.py @@ -82,16 +82,16 @@ class AuthTestCase(unittest.TestCase): self.hs.clock.now = 1000 token = self.macaroon_generator.generate_short_term_login_token("a_user", 5000) - user_id = yield self.auth_handler.validate_short_term_login_token_and_get_user_id( - token + user_id = yield defer.ensureDeferred( + self.auth_handler.validate_short_term_login_token_and_get_user_id(token) ) self.assertEqual("a_user", user_id) # when we advance the clock, the token should be rejected self.hs.clock.now = 6000 with self.assertRaises(synapse.api.errors.AuthError): - yield self.auth_handler.validate_short_term_login_token_and_get_user_id( - token + yield defer.ensureDeferred( + self.auth_handler.validate_short_term_login_token_and_get_user_id(token) ) @defer.inlineCallbacks @@ -99,8 +99,10 @@ class AuthTestCase(unittest.TestCase): token = self.macaroon_generator.generate_short_term_login_token("a_user", 5000) macaroon = pymacaroons.Macaroon.deserialize(token) - user_id = yield self.auth_handler.validate_short_term_login_token_and_get_user_id( - macaroon.serialize() + user_id = yield defer.ensureDeferred( + self.auth_handler.validate_short_term_login_token_and_get_user_id( + macaroon.serialize() + ) ) self.assertEqual("a_user", user_id) @@ -109,20 +111,26 @@ class AuthTestCase(unittest.TestCase): macaroon.add_first_party_caveat("user_id = b_user") with self.assertRaises(synapse.api.errors.AuthError): - yield self.auth_handler.validate_short_term_login_token_and_get_user_id( - macaroon.serialize() + yield defer.ensureDeferred( + self.auth_handler.validate_short_term_login_token_and_get_user_id( + macaroon.serialize() + ) ) @defer.inlineCallbacks def test_mau_limits_disabled(self): self.hs.config.limit_usage_by_mau = False # Ensure does not throw exception - yield self.auth_handler.get_access_token_for_user_id( - "user_a", device_id=None, valid_until_ms=None + yield defer.ensureDeferred( + self.auth_handler.get_access_token_for_user_id( + "user_a", device_id=None, valid_until_ms=None + ) ) - yield self.auth_handler.validate_short_term_login_token_and_get_user_id( - self._get_macaroon().serialize() + yield defer.ensureDeferred( + self.auth_handler.validate_short_term_login_token_and_get_user_id( + self._get_macaroon().serialize() + ) ) @defer.inlineCallbacks @@ -133,16 +141,20 @@ class AuthTestCase(unittest.TestCase): ) with self.assertRaises(ResourceLimitError): - yield self.auth_handler.get_access_token_for_user_id( - "user_a", device_id=None, valid_until_ms=None + yield defer.ensureDeferred( + self.auth_handler.get_access_token_for_user_id( + "user_a", device_id=None, valid_until_ms=None + ) ) self.hs.get_datastore().get_monthly_active_count = Mock( return_value=defer.succeed(self.large_number_of_users) ) with self.assertRaises(ResourceLimitError): - yield self.auth_handler.validate_short_term_login_token_and_get_user_id( - self._get_macaroon().serialize() + yield defer.ensureDeferred( + self.auth_handler.validate_short_term_login_token_and_get_user_id( + self._get_macaroon().serialize() + ) ) @defer.inlineCallbacks @@ -154,16 +166,20 @@ class AuthTestCase(unittest.TestCase): return_value=defer.succeed(self.hs.config.max_mau_value) ) with self.assertRaises(ResourceLimitError): - yield self.auth_handler.get_access_token_for_user_id( - "user_a", device_id=None, valid_until_ms=None + yield defer.ensureDeferred( + self.auth_handler.get_access_token_for_user_id( + "user_a", device_id=None, valid_until_ms=None + ) ) self.hs.get_datastore().get_monthly_active_count = Mock( return_value=defer.succeed(self.hs.config.max_mau_value) ) with self.assertRaises(ResourceLimitError): - yield self.auth_handler.validate_short_term_login_token_and_get_user_id( - self._get_macaroon().serialize() + yield defer.ensureDeferred( + self.auth_handler.validate_short_term_login_token_and_get_user_id( + self._get_macaroon().serialize() + ) ) # If in monthly active cohort self.hs.get_datastore().user_last_seen_monthly_active = Mock( @@ -172,8 +188,10 @@ class AuthTestCase(unittest.TestCase): self.hs.get_datastore().get_monthly_active_count = Mock( return_value=defer.succeed(self.hs.config.max_mau_value) ) - yield self.auth_handler.get_access_token_for_user_id( - "user_a", device_id=None, valid_until_ms=None + yield defer.ensureDeferred( + self.auth_handler.get_access_token_for_user_id( + "user_a", device_id=None, valid_until_ms=None + ) ) self.hs.get_datastore().user_last_seen_monthly_active = Mock( return_value=defer.succeed(self.hs.get_clock().time_msec()) @@ -181,8 +199,10 @@ class AuthTestCase(unittest.TestCase): self.hs.get_datastore().get_monthly_active_count = Mock( return_value=defer.succeed(self.hs.config.max_mau_value) ) - yield self.auth_handler.validate_short_term_login_token_and_get_user_id( - self._get_macaroon().serialize() + yield defer.ensureDeferred( + self.auth_handler.validate_short_term_login_token_and_get_user_id( + self._get_macaroon().serialize() + ) ) @defer.inlineCallbacks @@ -193,15 +213,19 @@ class AuthTestCase(unittest.TestCase): return_value=defer.succeed(self.small_number_of_users) ) # Ensure does not raise exception - yield self.auth_handler.get_access_token_for_user_id( - "user_a", device_id=None, valid_until_ms=None + yield defer.ensureDeferred( + self.auth_handler.get_access_token_for_user_id( + "user_a", device_id=None, valid_until_ms=None + ) ) self.hs.get_datastore().get_monthly_active_count = Mock( return_value=defer.succeed(self.small_number_of_users) ) - yield self.auth_handler.validate_short_term_login_token_and_get_user_id( - self._get_macaroon().serialize() + yield defer.ensureDeferred( + self.auth_handler.validate_short_term_login_token_and_get_user_id( + self._get_macaroon().serialize() + ) ) def _get_macaroon(self): diff --git a/tests/handlers/test_register.py b/tests/handlers/test_register.py index e7b638dbfe..f1dc51d6c9 100644 --- a/tests/handlers/test_register.py +++ b/tests/handlers/test_register.py @@ -294,7 +294,9 @@ class RegistrationTestCase(unittest.HomeserverTestCase): create_profile_with_displayname=user.localpart, ) else: - yield self.hs.get_auth_handler().delete_access_tokens_for_user(user_id) + yield defer.ensureDeferred( + self.hs.get_auth_handler().delete_access_tokens_for_user(user_id) + ) yield self.store.add_access_token_to_user( user_id=user_id, token=token, device_id=None, valid_until_ms=None diff --git a/tests/utils.py b/tests/utils.py index 968d109f77..2079e0143d 100644 --- a/tests/utils.py +++ b/tests/utils.py @@ -332,10 +332,15 @@ def setup_test_homeserver( # Need to let the HS build an auth handler and then mess with it # because AuthHandler's constructor requires the HS, so we can't make one # beforehand and pass it in to the HS's constructor (chicken / egg) - hs.get_auth_handler().hash = lambda p: hashlib.md5(p.encode("utf8")).hexdigest() - hs.get_auth_handler().validate_hash = ( - lambda p, h: hashlib.md5(p.encode("utf8")).hexdigest() == h - ) + async def hash(p): + return hashlib.md5(p.encode("utf8")).hexdigest() + + hs.get_auth_handler().hash = hash + + async def validate_hash(p, h): + return hashlib.md5(p.encode("utf8")).hexdigest() == h + + hs.get_auth_handler().validate_hash = validate_hash fed = kargs.get("resource_for_federation", None) if fed: -- cgit 1.4.1 From 01294e6b3a5c6ba909a0e70dad97a9ce7fdb64b9 Mon Sep 17 00:00:00 2001 From: Patrick Cloke Date: Thu, 16 Apr 2020 10:52:55 -0400 Subject: Do not treat display names as globs for push rules. (#7271) --- changelog.d/7271.bugfix | 1 + synapse/push/push_rule_evaluator.py | 69 +++++++++++++++++++--------------- tests/push/test_push_rule_evaluator.py | 65 ++++++++++++++++++++++++++++++++ tox.ini | 1 + 4 files changed, 106 insertions(+), 30 deletions(-) create mode 100644 changelog.d/7271.bugfix create mode 100644 tests/push/test_push_rule_evaluator.py (limited to 'tests') diff --git a/changelog.d/7271.bugfix b/changelog.d/7271.bugfix new file mode 100644 index 0000000000..e8315e4ce4 --- /dev/null +++ b/changelog.d/7271.bugfix @@ -0,0 +1 @@ +Do not treat display names as globs in push rules. diff --git a/synapse/push/push_rule_evaluator.py b/synapse/push/push_rule_evaluator.py index b1587183a8..4cd702b5fa 100644 --- a/synapse/push/push_rule_evaluator.py +++ b/synapse/push/push_rule_evaluator.py @@ -16,9 +16,11 @@ import logging import re +from typing import Pattern from six import string_types +from synapse.events import EventBase from synapse.types import UserID from synapse.util.caches import CACHE_SIZE_FACTOR, register_cache from synapse.util.caches.lrucache import LruCache @@ -56,18 +58,18 @@ def _test_ineq_condition(condition, number): rhs = m.group(2) if not rhs.isdigit(): return False - rhs = int(rhs) + rhs_int = int(rhs) if ineq == "" or ineq == "==": - return number == rhs + return number == rhs_int elif ineq == "<": - return number < rhs + return number < rhs_int elif ineq == ">": - return number > rhs + return number > rhs_int elif ineq == ">=": - return number >= rhs + return number >= rhs_int elif ineq == "<=": - return number <= rhs + return number <= rhs_int else: return False @@ -83,7 +85,13 @@ def tweaks_for_actions(actions): class PushRuleEvaluatorForEvent(object): - def __init__(self, event, room_member_count, sender_power_level, power_levels): + def __init__( + self, + event: EventBase, + room_member_count: int, + sender_power_level: int, + power_levels: dict, + ): self._event = event self._room_member_count = room_member_count self._sender_power_level = sender_power_level @@ -92,7 +100,7 @@ class PushRuleEvaluatorForEvent(object): # Maps strings of e.g. 'content.body' -> event["content"]["body"] self._value_cache = _flatten_dict(event) - def matches(self, condition, user_id, display_name): + def matches(self, condition: dict, user_id: str, display_name: str) -> bool: if condition["kind"] == "event_match": return self._event_match(condition, user_id) elif condition["kind"] == "contains_display_name": @@ -106,7 +114,7 @@ class PushRuleEvaluatorForEvent(object): else: return True - def _event_match(self, condition, user_id): + def _event_match(self, condition: dict, user_id: str) -> bool: pattern = condition.get("pattern", None) if not pattern: @@ -134,7 +142,7 @@ class PushRuleEvaluatorForEvent(object): return _glob_matches(pattern, haystack) - def _contains_display_name(self, display_name): + def _contains_display_name(self, display_name: str) -> bool: if not display_name: return False @@ -142,51 +150,52 @@ class PushRuleEvaluatorForEvent(object): if not body: return False - return _glob_matches(display_name, body, word_boundary=True) + # Similar to _glob_matches, but do not treat display_name as a glob. + r = regex_cache.get((display_name, False, True), None) + if not r: + r = re.escape(display_name) + r = _re_word_boundary(r) + r = re.compile(r, flags=re.IGNORECASE) + regex_cache[(display_name, False, True)] = r + + return r.search(body) - def _get_value(self, dotted_key): + def _get_value(self, dotted_key: str) -> str: return self._value_cache.get(dotted_key, None) -# Caches (glob, word_boundary) -> regex for push. See _glob_matches +# Caches (string, is_glob, word_boundary) -> regex for push. See _glob_matches regex_cache = LruCache(50000 * CACHE_SIZE_FACTOR) register_cache("cache", "regex_push_cache", regex_cache) -def _glob_matches(glob, value, word_boundary=False): +def _glob_matches(glob: str, value: str, word_boundary: bool = False) -> bool: """Tests if value matches glob. Args: - glob (string) - value (string): String to test against glob. - word_boundary (bool): Whether to match against word boundaries or entire + glob + value: String to test against glob. + word_boundary: Whether to match against word boundaries or entire string. Defaults to False. - - Returns: - bool """ try: - r = regex_cache.get((glob, word_boundary), None) + r = regex_cache.get((glob, True, word_boundary), None) if not r: r = _glob_to_re(glob, word_boundary) - regex_cache[(glob, word_boundary)] = r + regex_cache[(glob, True, word_boundary)] = r return r.search(value) except re.error: logger.warning("Failed to parse glob to regex: %r", glob) return False -def _glob_to_re(glob, word_boundary): +def _glob_to_re(glob: str, word_boundary: bool) -> Pattern: """Generates regex for a given glob. Args: - glob (string) - word_boundary (bool): Whether to match against word boundaries or entire - string. Defaults to False. - - Returns: - regex object + glob + word_boundary: Whether to match against word boundaries or entire string. """ if IS_GLOB.search(glob): r = re.escape(glob) @@ -219,7 +228,7 @@ def _glob_to_re(glob, word_boundary): return re.compile(r, flags=re.IGNORECASE) -def _re_word_boundary(r): +def _re_word_boundary(r: str) -> str: """ Adds word boundary characters to the start and end of an expression to require that the match occur as a whole word, diff --git a/tests/push/test_push_rule_evaluator.py b/tests/push/test_push_rule_evaluator.py new file mode 100644 index 0000000000..9ae6a87d7b --- /dev/null +++ b/tests/push/test_push_rule_evaluator.py @@ -0,0 +1,65 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from synapse.api.room_versions import RoomVersions +from synapse.events import FrozenEvent +from synapse.push.push_rule_evaluator import PushRuleEvaluatorForEvent + +from tests import unittest + + +class PushRuleEvaluatorTestCase(unittest.TestCase): + def setUp(self): + event = FrozenEvent( + { + "event_id": "$event_id", + "type": "m.room.history_visibility", + "sender": "@user:test", + "state_key": "", + "room_id": "@room:test", + "content": {"body": "foo bar baz"}, + }, + RoomVersions.V1, + ) + room_member_count = 0 + sender_power_level = 0 + power_levels = {} + self.evaluator = PushRuleEvaluatorForEvent( + event, room_member_count, sender_power_level, power_levels + ) + + def test_display_name(self): + """Check for a matching display name in the body of the event.""" + condition = { + "kind": "contains_display_name", + } + + # Blank names are skipped. + self.assertFalse(self.evaluator.matches(condition, "@user:test", "")) + + # Check a display name that doesn't match. + self.assertFalse(self.evaluator.matches(condition, "@user:test", "not found")) + + # Check a display name which matches. + self.assertTrue(self.evaluator.matches(condition, "@user:test", "foo")) + + # A display name that matches, but not a full word does not result in a match. + self.assertFalse(self.evaluator.matches(condition, "@user:test", "ba")) + + # A display name should not be interpreted as a regular expression. + self.assertFalse(self.evaluator.matches(condition, "@user:test", "ba[rz]")) + + # A display name with spaces should work fine. + self.assertTrue(self.evaluator.matches(condition, "@user:test", "foo bar")) diff --git a/tox.ini b/tox.ini index 763c8463d9..42b2d74891 100644 --- a/tox.ini +++ b/tox.ini @@ -196,6 +196,7 @@ commands = mypy \ synapse/metrics \ synapse/module_api \ synapse/push/pusherpool.py \ + synapse/push/push_rule_evaluator.py \ synapse/replication \ synapse/rest \ synapse/spam_checker_api \ -- cgit 1.4.1 From 51f7eaf908a84fcaf231899e2bf1beae14ae72c0 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 22 Apr 2020 13:07:41 +0100 Subject: Add ability to run replication protocol over redis. (#7040) This is configured via the `redis` config options. --- changelog.d/7040.feature | 1 + stubs/txredisapi.pyi | 40 +++++++ synapse/app/homeserver.py | 6 + synapse/config/homeserver.py | 2 + synapse/config/redis.py | 35 ++++++ synapse/python_dependencies.py | 1 + synapse/replication/tcp/client.py | 2 +- synapse/replication/tcp/commands.py | 18 +++ synapse/replication/tcp/handler.py | 50 +++++++-- synapse/replication/tcp/protocol.py | 38 ++----- synapse/replication/tcp/redis.py | 181 +++++++++++++++++++++++++++++++ tests/replication/slave/storage/_base.py | 4 +- 12 files changed, 342 insertions(+), 36 deletions(-) create mode 100644 changelog.d/7040.feature create mode 100644 stubs/txredisapi.pyi create mode 100644 synapse/config/redis.py create mode 100644 synapse/replication/tcp/redis.py (limited to 'tests') diff --git a/changelog.d/7040.feature b/changelog.d/7040.feature new file mode 100644 index 0000000000..ce6140fdd1 --- /dev/null +++ b/changelog.d/7040.feature @@ -0,0 +1 @@ +Add support for running replication over Redis when using workers. diff --git a/stubs/txredisapi.pyi b/stubs/txredisapi.pyi new file mode 100644 index 0000000000..763d3fb404 --- /dev/null +++ b/stubs/txredisapi.pyi @@ -0,0 +1,40 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +"""Contains *incomplete* type hints for txredisapi. +""" + +from typing import List, Optional, Union + +class RedisProtocol: + def publish(self, channel: str, message: bytes): ... + +class SubscriberProtocol: + def subscribe(self, channels: Union[str, List[str]]): ... + +def lazyConnection( + host: str = ..., + port: int = ..., + dbid: Optional[int] = ..., + reconnect: bool = ..., + charset: str = ..., + password: Optional[str] = ..., + connectTimeout: Optional[int] = ..., + replyTimeout: Optional[int] = ..., + convertNumbers: bool = ..., +) -> RedisProtocol: ... + +class SubscriberFactory: + def buildProtocol(self, addr): ... diff --git a/synapse/app/homeserver.py b/synapse/app/homeserver.py index 49df63acd0..cbd1ea475a 100644 --- a/synapse/app/homeserver.py +++ b/synapse/app/homeserver.py @@ -273,6 +273,12 @@ class SynapseHomeServer(HomeServer): def start_listening(self, listeners): config = self.get_config() + if config.redis_enabled: + # If redis is enabled we connect via the replication command handler + # in the same way as the workers (since we're effectively a client + # rather than a server). + self.get_tcp_replication().start_replication(self) + for listener in listeners: if listener["type"] == "http": self._listening_services.extend(self._listener_http(config, listener)) diff --git a/synapse/config/homeserver.py b/synapse/config/homeserver.py index b4bca08b20..be6c6afa74 100644 --- a/synapse/config/homeserver.py +++ b/synapse/config/homeserver.py @@ -31,6 +31,7 @@ from .password import PasswordConfig from .password_auth_providers import PasswordAuthProviderConfig from .push import PushConfig from .ratelimiting import RatelimitConfig +from .redis import RedisConfig from .registration import RegistrationConfig from .repository import ContentRepositoryConfig from .room_directory import RoomDirectoryConfig @@ -82,4 +83,5 @@ class HomeServerConfig(RootConfig): RoomDirectoryConfig, ThirdPartyRulesConfig, TracerConfig, + RedisConfig, ] diff --git a/synapse/config/redis.py b/synapse/config/redis.py new file mode 100644 index 0000000000..81a27619ec --- /dev/null +++ b/synapse/config/redis.py @@ -0,0 +1,35 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from synapse.config._base import Config +from synapse.python_dependencies import check_requirements + + +class RedisConfig(Config): + section = "redis" + + def read_config(self, config, **kwargs): + redis_config = config.get("redis", {}) + self.redis_enabled = redis_config.get("enabled", False) + + if not self.redis_enabled: + return + + check_requirements("redis") + + self.redis_host = redis_config.get("host", "localhost") + self.redis_port = redis_config.get("port", 6379) + self.redis_dbid = redis_config.get("dbid") + self.redis_password = redis_config.get("password") diff --git a/synapse/python_dependencies.py b/synapse/python_dependencies.py index 8de8cb2c12..733c51b758 100644 --- a/synapse/python_dependencies.py +++ b/synapse/python_dependencies.py @@ -98,6 +98,7 @@ CONDITIONAL_REQUIREMENTS = { "sentry": ["sentry-sdk>=0.7.2"], "opentracing": ["jaeger-client>=4.0.0", "opentracing>=2.2.0"], "jwt": ["pyjwt>=1.6.4"], + "redis": ["txredisapi>=1.4.7"], } ALL_OPTIONAL_REQUIREMENTS = set() # type: Set[str] diff --git a/synapse/replication/tcp/client.py b/synapse/replication/tcp/client.py index 700ae79158..2d07b8b2d0 100644 --- a/synapse/replication/tcp/client.py +++ b/synapse/replication/tcp/client.py @@ -30,7 +30,7 @@ if TYPE_CHECKING: logger = logging.getLogger(__name__) -class ReplicationClientFactory(ReconnectingClientFactory): +class DirectTcpReplicationClientFactory(ReconnectingClientFactory): """Factory for building connections to the master. Will reconnect if the connection is lost. diff --git a/synapse/replication/tcp/commands.py b/synapse/replication/tcp/commands.py index 5ec89d0fb8..5002efe6a0 100644 --- a/synapse/replication/tcp/commands.py +++ b/synapse/replication/tcp/commands.py @@ -454,3 +454,21 @@ VALID_CLIENT_COMMANDS = ( ErrorCommand.NAME, RemoteServerUpCommand.NAME, ) + + +def parse_command_from_line(line: str) -> Command: + """Parses a command from a received line. + + Line should already be stripped of whitespace and be checked if blank. + """ + + idx = line.index(" ") + if idx >= 0: + cmd_name = line[:idx] + rest_of_line = line[idx + 1 :] + else: + cmd_name = line + rest_of_line = "" + + cmd_cls = COMMAND_MAP[cmd_name] + return cmd_cls.from_line(rest_of_line) diff --git a/synapse/replication/tcp/handler.py b/synapse/replication/tcp/handler.py index e32e68e8c4..5b5ee2c13e 100644 --- a/synapse/replication/tcp/handler.py +++ b/synapse/replication/tcp/handler.py @@ -30,8 +30,10 @@ from typing import ( from prometheus_client import Counter +from twisted.internet.protocol import ReconnectingClientFactory + from synapse.metrics import LaterGauge -from synapse.replication.tcp.client import ReplicationClientFactory +from synapse.replication.tcp.client import DirectTcpReplicationClientFactory from synapse.replication.tcp.commands import ( ClearUserSyncsCommand, Command, @@ -92,7 +94,7 @@ class ReplicationCommandHandler: self._pending_batches = {} # type: Dict[str, List[Any]] # The factory used to create connections. - self._factory = None # type: Optional[ReplicationClientFactory] + self._factory = None # type: Optional[ReconnectingClientFactory] # The currently connected connections. self._connections = [] # type: List[AbstractConnection] @@ -119,11 +121,45 @@ class ReplicationCommandHandler: """Helper method to start a replication connection to the remote server using TCP. """ - client_name = hs.config.worker_name - self._factory = ReplicationClientFactory(hs, client_name, self) - host = hs.config.worker_replication_host - port = hs.config.worker_replication_port - hs.get_reactor().connectTCP(host, port, self._factory) + if hs.config.redis.redis_enabled: + from synapse.replication.tcp.redis import ( + RedisDirectTcpReplicationClientFactory, + ) + import txredisapi + + logger.info( + "Connecting to redis (host=%r port=%r DBID=%r)", + hs.config.redis_host, + hs.config.redis_port, + hs.config.redis_dbid, + ) + + # We need two connections to redis, one for the subscription stream and + # one to send commands to (as you can't send further redis commands to a + # connection after SUBSCRIBE is called). + + # First create the connection for sending commands. + outbound_redis_connection = txredisapi.lazyConnection( + host=hs.config.redis_host, + port=hs.config.redis_port, + dbid=hs.config.redis_dbid, + password=hs.config.redis.redis_password, + reconnect=True, + ) + + # Now create the factory/connection for the subscription stream. + self._factory = RedisDirectTcpReplicationClientFactory( + hs, outbound_redis_connection + ) + hs.get_reactor().connectTCP( + hs.config.redis.redis_host, hs.config.redis.redis_port, self._factory, + ) + else: + client_name = hs.config.worker_name + self._factory = DirectTcpReplicationClientFactory(hs, client_name, self) + host = hs.config.worker_replication_host + port = hs.config.worker_replication_port + hs.get_reactor().connectTCP(host, port, self._factory) async def on_REPLICATE(self, cmd: ReplicateCommand): # We only want to announce positions by the writer of the streams. diff --git a/synapse/replication/tcp/protocol.py b/synapse/replication/tcp/protocol.py index 9276ed2965..7240acb0a2 100644 --- a/synapse/replication/tcp/protocol.py +++ b/synapse/replication/tcp/protocol.py @@ -63,7 +63,6 @@ from twisted.python.failure import Failure from synapse.metrics import LaterGauge from synapse.metrics.background_process_metrics import run_as_background_process from synapse.replication.tcp.commands import ( - COMMAND_MAP, VALID_CLIENT_COMMANDS, VALID_SERVER_COMMANDS, Command, @@ -72,6 +71,7 @@ from synapse.replication.tcp.commands import ( PingCommand, ReplicateCommand, ServerCommand, + parse_command_from_line, ) from synapse.types import Collection from synapse.util import Clock @@ -210,38 +210,24 @@ class BaseReplicationStreamProtocol(LineOnlyReceiver): linestr = line.decode("utf-8") - # split at the first " ", handling one-word commands - idx = linestr.index(" ") - if idx >= 0: - cmd_name = linestr[:idx] - rest_of_line = linestr[idx + 1 :] - else: - cmd_name = linestr - rest_of_line = "" + try: + cmd = parse_command_from_line(linestr) + except Exception as e: + logger.exception("[%s] failed to parse line: %r", self.id(), linestr) + self.send_error("failed to parse line: %r (%r):" % (e, linestr)) + return - if cmd_name not in self.VALID_INBOUND_COMMANDS: - logger.error("[%s] invalid command %s", self.id(), cmd_name) - self.send_error("invalid command: %s", cmd_name) + if cmd.NAME not in self.VALID_INBOUND_COMMANDS: + logger.error("[%s] invalid command %s", self.id(), cmd.NAME) + self.send_error("invalid command: %s", cmd.NAME) return self.last_received_command = self.clock.time_msec() - self.inbound_commands_counter[cmd_name] = ( - self.inbound_commands_counter[cmd_name] + 1 + self.inbound_commands_counter[cmd.NAME] = ( + self.inbound_commands_counter[cmd.NAME] + 1 ) - cmd_cls = COMMAND_MAP[cmd_name] - try: - cmd = cmd_cls.from_line(rest_of_line) - except Exception as e: - logger.exception( - "[%s] failed to parse line %r: %r", self.id(), cmd_name, rest_of_line - ) - self.send_error( - "failed to parse line for %r: %r (%r):" % (cmd_name, e, rest_of_line) - ) - return - # Now lets try and call on_ function run_as_background_process( "replication-" + cmd.get_logcontext_id(), self.handle_command, cmd diff --git a/synapse/replication/tcp/redis.py b/synapse/replication/tcp/redis.py new file mode 100644 index 0000000000..4c08425735 --- /dev/null +++ b/synapse/replication/tcp/redis.py @@ -0,0 +1,181 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging +from typing import TYPE_CHECKING + +import txredisapi + +from synapse.logging.context import PreserveLoggingContext +from synapse.metrics.background_process_metrics import run_as_background_process +from synapse.replication.tcp.commands import ( + Command, + ReplicateCommand, + parse_command_from_line, +) +from synapse.replication.tcp.protocol import AbstractConnection + +if TYPE_CHECKING: + from synapse.replication.tcp.handler import ReplicationCommandHandler + from synapse.server import HomeServer + +logger = logging.getLogger(__name__) + + +class RedisSubscriber(txredisapi.SubscriberProtocol, AbstractConnection): + """Connection to redis subscribed to replication stream. + + Parses incoming messages from redis into replication commands, and passes + them to `ReplicationCommandHandler` + + Due to the vagaries of `txredisapi` we don't want to have a custom + constructor, so instead we expect the defined attributes below to be set + immediately after initialisation. + + Attributes: + handler: The command handler to handle incoming commands. + stream_name: The *redis* stream name to subscribe to (not anything to + do with Synapse replication streams). + outbound_redis_connection: The connection to redis to use to send + commands. + """ + + handler = None # type: ReplicationCommandHandler + stream_name = None # type: str + outbound_redis_connection = None # type: txredisapi.RedisProtocol + + def connectionMade(self): + logger.info("Connected to redis instance") + self.subscribe(self.stream_name) + self.send_command(ReplicateCommand()) + + self.handler.new_connection(self) + + def messageReceived(self, pattern: str, channel: str, message: str): + """Received a message from redis. + """ + + if message.strip() == "": + # Ignore blank lines + return + + try: + cmd = parse_command_from_line(message) + except Exception: + logger.exception( + "[%s] failed to parse line: %r", message, + ) + return + + # Now lets try and call on_ function + run_as_background_process( + "replication-" + cmd.get_logcontext_id(), self.handle_command, cmd + ) + + async def handle_command(self, cmd: Command): + """Handle a command we have received over the replication stream. + + By default delegates to on_, which should return an awaitable. + + Args: + cmd: received command + """ + handled = False + + # First call any command handlers on this instance. These are for redis + # specific handling. + cmd_func = getattr(self, "on_%s" % (cmd.NAME,), None) + if cmd_func: + await cmd_func(cmd) + handled = True + + # Then call out to the handler. + cmd_func = getattr(self.handler, "on_%s" % (cmd.NAME,), None) + if cmd_func: + await cmd_func(cmd) + handled = True + + if not handled: + logger.warning("Unhandled command: %r", cmd) + + def connectionLost(self, reason): + logger.info("Lost connection to redis instance") + self.handler.lost_connection(self) + + def send_command(self, cmd: Command): + """Send a command if connection has been established. + + Args: + cmd (Command) + """ + string = "%s %s" % (cmd.NAME, cmd.to_line()) + if "\n" in string: + raise Exception("Unexpected newline in command: %r", string) + + encoded_string = string.encode("utf-8") + + async def _send(): + with PreserveLoggingContext(): + # Note that we use the other connection as we can't send + # commands using the subscription connection. + await self.outbound_redis_connection.publish( + self.stream_name, encoded_string + ) + + run_as_background_process("send-cmd", _send) + + +class RedisDirectTcpReplicationClientFactory(txredisapi.SubscriberFactory): + """This is a reconnecting factory that connects to redis and immediately + subscribes to a stream. + + Args: + hs + outbound_redis_connection: A connection to redis that will be used to + send outbound commands (this is seperate to the redis connection + used to subscribe). + """ + + maxDelay = 5 + continueTrying = True + protocol = RedisSubscriber + + def __init__( + self, hs: "HomeServer", outbound_redis_connection: txredisapi.RedisProtocol + ): + + super().__init__() + + # This sets the password on the RedisFactory base class (as + # SubscriberFactory constructor doesn't pass it through). + self.password = hs.config.redis.redis_password + + self.handler = hs.get_tcp_replication() + self.stream_name = hs.hostname + + self.outbound_redis_connection = outbound_redis_connection + + def buildProtocol(self, addr): + p = super().buildProtocol(addr) # type: RedisSubscriber + + # We do this here rather than add to the constructor of `RedisSubcriber` + # as to do so would involve overriding `buildProtocol` entirely, however + # the base method does some other things than just instantiating the + # protocol. + p.handler = self.handler + p.outbound_redis_connection = self.outbound_redis_connection + p.stream_name = self.stream_name + + return p diff --git a/tests/replication/slave/storage/_base.py b/tests/replication/slave/storage/_base.py index 8902a5ab69..395c7d0306 100644 --- a/tests/replication/slave/storage/_base.py +++ b/tests/replication/slave/storage/_base.py @@ -16,7 +16,7 @@ from mock import Mock, NonCallableMock from synapse.replication.tcp.client import ( - ReplicationClientFactory, + DirectTcpReplicationClientFactory, ReplicationDataHandler, ) from synapse.replication.tcp.handler import ReplicationCommandHandler @@ -61,7 +61,7 @@ class BaseSlavedStoreTestCase(unittest.HomeserverTestCase): self.slaved_store ) - client_factory = ReplicationClientFactory( + client_factory = DirectTcpReplicationClientFactory( self.hs, "client_name", self.replication_handler ) client_factory.handler = self.replication_handler -- cgit 1.4.1 From 6b6685db9f8cf8a55dfe0edc8f2131be8001c360 Mon Sep 17 00:00:00 2001 From: Dirk Klimpel <5740567+dklimpel@users.noreply.github.com> Date: Wed, 22 Apr 2020 14:38:41 +0200 Subject: Extend room admin api with additional attributes (#7225) --- changelog.d/7225.misc | 1 + docs/admin_api/rooms.md | 107 ++++- synapse/rest/admin/rooms.py | 15 +- synapse/storage/data_stores/main/room.py | 78 +++- tests/rest/admin/test_admin.py | 592 +-------------------------- tests/rest/admin/test_room.py | 680 ++++++++++++++++++++++++++++++- 6 files changed, 869 insertions(+), 604 deletions(-) create mode 100644 changelog.d/7225.misc (limited to 'tests') diff --git a/changelog.d/7225.misc b/changelog.d/7225.misc new file mode 100644 index 0000000000..375e2a475f --- /dev/null +++ b/changelog.d/7225.misc @@ -0,0 +1 @@ +Extend room admin api (`GET /_synapse/admin/v1/rooms`) with additional attributes. \ No newline at end of file diff --git a/docs/admin_api/rooms.md b/docs/admin_api/rooms.md index 2db457c1b6..26fe8b8679 100644 --- a/docs/admin_api/rooms.md +++ b/docs/admin_api/rooms.md @@ -11,8 +11,21 @@ The following query parameters are available: * `from` - Offset in the returned list. Defaults to `0`. * `limit` - Maximum amount of rooms to return. Defaults to `100`. * `order_by` - The method in which to sort the returned list of rooms. Valid values are: - - `alphabetical` - Rooms are ordered alphabetically by room name. This is the default. - - `size` - Rooms are ordered by the number of members. Largest to smallest. + - `alphabetical` - Same as `name`. This is deprecated. + - `size` - Same as `joined_members`. This is deprecated. + - `name` - Rooms are ordered alphabetically by room name. This is the default. + - `canonical_alias` - Rooms are ordered alphabetically by main alias address of the room. + - `joined_members` - Rooms are ordered by the number of members. Largest to smallest. + - `joined_local_members` - Rooms are ordered by the number of local members. Largest to smallest. + - `version` - Rooms are ordered by room version. Largest to smallest. + - `creator` - Rooms are ordered alphabetically by creator of the room. + - `encryption` - Rooms are ordered alphabetically by the end-to-end encryption algorithm. + - `federatable` - Rooms are ordered by whether the room is federatable. + - `public` - Rooms are ordered by visibility in room list. + - `join_rules` - Rooms are ordered alphabetically by join rules of the room. + - `guest_access` - Rooms are ordered alphabetically by guest access option of the room. + - `history_visibility` - Rooms are ordered alphabetically by visibility of history of the room. + - `state_events` - Rooms are ordered by number of state events. Largest to smallest. * `dir` - Direction of room order. Either `f` for forwards or `b` for backwards. Setting this value to `b` will reverse the above sort order. Defaults to `f`. * `search_term` - Filter rooms by their room name. Search term can be contained in any @@ -26,6 +39,16 @@ The following fields are possible in the JSON response body: - `name` - The name of the room. - `canonical_alias` - The canonical (main) alias address of the room. - `joined_members` - How many users are currently in the room. + - `joined_local_members` - How many local users are currently in the room. + - `version` - The version of the room as a string. + - `creator` - The `user_id` of the room creator. + - `encryption` - Algorithm of end-to-end encryption of messages. Is `null` if encryption is not active. + - `federatable` - Whether users on other servers can join this room. + - `public` - Whether the room is visible in room directory. + - `join_rules` - The type of rules used for users wishing to join this room. One of: ["public", "knock", "invite", "private"]. + - `guest_access` - Whether guests can join the room. One of: ["can_join", "forbidden"]. + - `history_visibility` - Who can see the room history. One of: ["invited", "joined", "shared", "world_readable"]. + - `state_events` - Total number of state_events of a room. Complexity of the room. * `offset` - The current pagination offset in rooms. This parameter should be used instead of `next_token` for room offset as `next_token` is not intended to be parsed. @@ -60,14 +83,34 @@ Response: "room_id": "!OGEhHVWSdvArJzumhm:matrix.org", "name": "Matrix HQ", "canonical_alias": "#matrix:matrix.org", - "joined_members": 8326 + "joined_members": 8326, + "joined_local_members": 2, + "version": "1", + "creator": "@foo:matrix.org", + "encryption": null, + "federatable": true, + "public": true, + "join_rules": "invite", + "guest_access": null, + "history_visibility": "shared", + "state_events": 93534 }, ... (8 hidden items) ... { "room_id": "!xYvNcQPhnkrdUmYczI:matrix.org", "name": "This Week In Matrix (TWIM)", "canonical_alias": "#twim:matrix.org", - "joined_members": 314 + "joined_members": 314, + "joined_local_members": 20, + "version": "4", + "creator": "@foo:matrix.org", + "encryption": "m.megolm.v1.aes-sha2", + "federatable": true, + "public": false, + "join_rules": "invite", + "guest_access": null, + "history_visibility": "shared", + "state_events": 8345 } ], "offset": 0, @@ -92,7 +135,17 @@ Response: "room_id": "!xYvNcQPhnkrdUmYczI:matrix.org", "name": "This Week In Matrix (TWIM)", "canonical_alias": "#twim:matrix.org", - "joined_members": 314 + "joined_members": 314, + "joined_local_members": 20, + "version": "4", + "creator": "@foo:matrix.org", + "encryption": "m.megolm.v1.aes-sha2", + "federatable": true, + "public": false, + "join_rules": "invite", + "guest_access": null, + "history_visibility": "shared", + "state_events": 8 } ], "offset": 0, @@ -117,14 +170,34 @@ Response: "room_id": "!OGEhHVWSdvArJzumhm:matrix.org", "name": "Matrix HQ", "canonical_alias": "#matrix:matrix.org", - "joined_members": 8326 + "joined_members": 8326, + "joined_local_members": 2, + "version": "1", + "creator": "@foo:matrix.org", + "encryption": null, + "federatable": true, + "public": true, + "join_rules": "invite", + "guest_access": null, + "history_visibility": "shared", + "state_events": 93534 }, ... (98 hidden items) ... { "room_id": "!xYvNcQPhnkrdUmYczI:matrix.org", "name": "This Week In Matrix (TWIM)", "canonical_alias": "#twim:matrix.org", - "joined_members": 314 + "joined_members": 314, + "joined_local_members": 20, + "version": "4", + "creator": "@foo:matrix.org", + "encryption": "m.megolm.v1.aes-sha2", + "federatable": true, + "public": false, + "join_rules": "invite", + "guest_access": null, + "history_visibility": "shared", + "state_events": 8345 } ], "offset": 0, @@ -154,6 +227,16 @@ Response: "name": "Music Theory", "canonical_alias": "#musictheory:matrix.org", "joined_members": 127 + "joined_local_members": 2, + "version": "1", + "creator": "@foo:matrix.org", + "encryption": null, + "federatable": true, + "public": true, + "join_rules": "invite", + "guest_access": null, + "history_visibility": "shared", + "state_events": 93534 }, ... (48 hidden items) ... { @@ -161,6 +244,16 @@ Response: "name": "weechat-matrix", "canonical_alias": "#weechat-matrix:termina.org.uk", "joined_members": 137 + "joined_local_members": 20, + "version": "4", + "creator": "@foo:termina.org.uk", + "encryption": null, + "federatable": true, + "public": true, + "join_rules": "invite", + "guest_access": null, + "history_visibility": "shared", + "state_events": 8345 } ], "offset": 100, diff --git a/synapse/rest/admin/rooms.py b/synapse/rest/admin/rooms.py index 659b8a10ee..d1bdb64111 100644 --- a/synapse/rest/admin/rooms.py +++ b/synapse/rest/admin/rooms.py @@ -183,10 +183,23 @@ class ListRoomRestServlet(RestServlet): # Extract query parameters start = parse_integer(request, "from", default=0) limit = parse_integer(request, "limit", default=100) - order_by = parse_string(request, "order_by", default="alphabetical") + order_by = parse_string(request, "order_by", default=RoomSortOrder.NAME.value) if order_by not in ( RoomSortOrder.ALPHABETICAL.value, RoomSortOrder.SIZE.value, + RoomSortOrder.NAME.value, + RoomSortOrder.CANONICAL_ALIAS.value, + RoomSortOrder.JOINED_MEMBERS.value, + RoomSortOrder.JOINED_LOCAL_MEMBERS.value, + RoomSortOrder.VERSION.value, + RoomSortOrder.CREATOR.value, + RoomSortOrder.ENCRYPTION.value, + RoomSortOrder.FEDERATABLE.value, + RoomSortOrder.PUBLIC.value, + RoomSortOrder.JOIN_RULES.value, + RoomSortOrder.GUEST_ACCESS.value, + RoomSortOrder.HISTORY_VISIBILITY.value, + RoomSortOrder.STATE_EVENTS.value, ): raise SynapseError( 400, diff --git a/synapse/storage/data_stores/main/room.py b/synapse/storage/data_stores/main/room.py index aaebe427d3..147eba1df7 100644 --- a/synapse/storage/data_stores/main/room.py +++ b/synapse/storage/data_stores/main/room.py @@ -52,12 +52,28 @@ class RoomSortOrder(Enum): """ Enum to define the sorting method used when returning rooms with get_rooms_paginate - ALPHABETICAL = sort rooms alphabetically by name - SIZE = sort rooms by membership size, highest to lowest + NAME = sort rooms alphabetically by name + JOINED_MEMBERS = sort rooms by membership size, highest to lowest """ + # ALPHABETICAL and SIZE are deprecated. + # ALPHABETICAL is the same as NAME. ALPHABETICAL = "alphabetical" + # SIZE is the same as JOINED_MEMBERS. SIZE = "size" + NAME = "name" + CANONICAL_ALIAS = "canonical_alias" + JOINED_MEMBERS = "joined_members" + JOINED_LOCAL_MEMBERS = "joined_local_members" + VERSION = "version" + CREATOR = "creator" + ENCRYPTION = "encryption" + FEDERATABLE = "federatable" + PUBLIC = "public" + JOIN_RULES = "join_rules" + GUEST_ACCESS = "guest_access" + HISTORY_VISIBILITY = "history_visibility" + STATE_EVENTS = "state_events" class RoomWorkerStore(SQLBaseStore): @@ -329,12 +345,52 @@ class RoomWorkerStore(SQLBaseStore): # Set ordering if RoomSortOrder(order_by) == RoomSortOrder.SIZE: + # Deprecated in favour of RoomSortOrder.JOINED_MEMBERS order_by_column = "curr.joined_members" order_by_asc = False elif RoomSortOrder(order_by) == RoomSortOrder.ALPHABETICAL: - # Sort alphabetically + # Deprecated in favour of RoomSortOrder.NAME order_by_column = "state.name" order_by_asc = True + elif RoomSortOrder(order_by) == RoomSortOrder.NAME: + order_by_column = "state.name" + order_by_asc = True + elif RoomSortOrder(order_by) == RoomSortOrder.CANONICAL_ALIAS: + order_by_column = "state.canonical_alias" + order_by_asc = True + elif RoomSortOrder(order_by) == RoomSortOrder.JOINED_MEMBERS: + order_by_column = "curr.joined_members" + order_by_asc = False + elif RoomSortOrder(order_by) == RoomSortOrder.JOINED_LOCAL_MEMBERS: + order_by_column = "curr.local_users_in_room" + order_by_asc = False + elif RoomSortOrder(order_by) == RoomSortOrder.VERSION: + order_by_column = "rooms.room_version" + order_by_asc = False + elif RoomSortOrder(order_by) == RoomSortOrder.CREATOR: + order_by_column = "rooms.creator" + order_by_asc = True + elif RoomSortOrder(order_by) == RoomSortOrder.ENCRYPTION: + order_by_column = "state.encryption" + order_by_asc = True + elif RoomSortOrder(order_by) == RoomSortOrder.FEDERATABLE: + order_by_column = "state.is_federatable" + order_by_asc = True + elif RoomSortOrder(order_by) == RoomSortOrder.PUBLIC: + order_by_column = "rooms.is_public" + order_by_asc = True + elif RoomSortOrder(order_by) == RoomSortOrder.JOIN_RULES: + order_by_column = "state.join_rules" + order_by_asc = True + elif RoomSortOrder(order_by) == RoomSortOrder.GUEST_ACCESS: + order_by_column = "state.guest_access" + order_by_asc = True + elif RoomSortOrder(order_by) == RoomSortOrder.HISTORY_VISIBILITY: + order_by_column = "state.history_visibility" + order_by_asc = True + elif RoomSortOrder(order_by) == RoomSortOrder.STATE_EVENTS: + order_by_column = "curr.current_state_events" + order_by_asc = False else: raise StoreError( 500, "Incorrect value for order_by provided: %s" % order_by @@ -349,9 +405,13 @@ class RoomWorkerStore(SQLBaseStore): # for, and another query for getting the total number of events that could be # returned. Thus allowing us to see if there are more events to paginate through info_sql = """ - SELECT state.room_id, state.name, state.canonical_alias, curr.joined_members + SELECT state.room_id, state.name, state.canonical_alias, curr.joined_members, + curr.local_users_in_room, rooms.room_version, rooms.creator, + state.encryption, state.is_federatable, rooms.is_public, state.join_rules, + state.guest_access, state.history_visibility, curr.current_state_events FROM room_stats_state state INNER JOIN room_stats_current curr USING (room_id) + INNER JOIN rooms USING (room_id) %s ORDER BY %s %s LIMIT ? @@ -389,6 +449,16 @@ class RoomWorkerStore(SQLBaseStore): "name": room[1], "canonical_alias": room[2], "joined_members": room[3], + "joined_local_members": room[4], + "version": room[5], + "creator": room[6], + "encryption": room[7], + "federatable": room[8], + "public": room[9], + "join_rules": room[10], + "guest_access": room[11], + "history_visibility": room[12], + "state_events": room[13], } ) diff --git a/tests/rest/admin/test_admin.py b/tests/rest/admin/test_admin.py index 0342aed416..977615ebef 100644 --- a/tests/rest/admin/test_admin.py +++ b/tests/rest/admin/test_admin.py @@ -17,7 +17,6 @@ import json import os import urllib.parse from binascii import unhexlify -from typing import List, Optional from mock import Mock @@ -27,7 +26,7 @@ import synapse.rest.admin from synapse.http.server import JsonResource from synapse.logging.context import make_deferred_yieldable from synapse.rest.admin import VersionServlet -from synapse.rest.client.v1 import directory, events, login, room +from synapse.rest.client.v1 import login, room from synapse.rest.client.v2_alpha import groups from tests import unittest @@ -51,129 +50,6 @@ class VersionTestCase(unittest.HomeserverTestCase): ) -class ShutdownRoomTestCase(unittest.HomeserverTestCase): - servlets = [ - synapse.rest.admin.register_servlets_for_client_rest_resource, - login.register_servlets, - events.register_servlets, - room.register_servlets, - room.register_deprecated_servlets, - ] - - def prepare(self, reactor, clock, hs): - self.event_creation_handler = hs.get_event_creation_handler() - hs.config.user_consent_version = "1" - - consent_uri_builder = Mock() - consent_uri_builder.build_user_consent_uri.return_value = "http://example.com" - self.event_creation_handler._consent_uri_builder = consent_uri_builder - - self.store = hs.get_datastore() - - self.admin_user = self.register_user("admin", "pass", admin=True) - self.admin_user_tok = self.login("admin", "pass") - - self.other_user = self.register_user("user", "pass") - self.other_user_token = self.login("user", "pass") - - # Mark the admin user as having consented - self.get_success(self.store.user_set_consent_version(self.admin_user, "1")) - - def test_shutdown_room_consent(self): - """Test that we can shutdown rooms with local users who have not - yet accepted the privacy policy. This used to fail when we tried to - force part the user from the old room. - """ - self.event_creation_handler._block_events_without_consent_error = None - - room_id = self.helper.create_room_as(self.other_user, tok=self.other_user_token) - - # Assert one user in room - users_in_room = self.get_success(self.store.get_users_in_room(room_id)) - self.assertEqual([self.other_user], users_in_room) - - # Enable require consent to send events - self.event_creation_handler._block_events_without_consent_error = "Error" - - # Assert that the user is getting consent error - self.helper.send( - room_id, body="foo", tok=self.other_user_token, expect_code=403 - ) - - # Test that the admin can still send shutdown - url = "admin/shutdown_room/" + room_id - request, channel = self.make_request( - "POST", - url.encode("ascii"), - json.dumps({"new_room_user_id": self.admin_user}), - access_token=self.admin_user_tok, - ) - self.render(request) - - self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) - - # Assert there is now no longer anyone in the room - users_in_room = self.get_success(self.store.get_users_in_room(room_id)) - self.assertEqual([], users_in_room) - - def test_shutdown_room_block_peek(self): - """Test that a world_readable room can no longer be peeked into after - it has been shut down. - """ - - self.event_creation_handler._block_events_without_consent_error = None - - room_id = self.helper.create_room_as(self.other_user, tok=self.other_user_token) - - # Enable world readable - url = "rooms/%s/state/m.room.history_visibility" % (room_id,) - request, channel = self.make_request( - "PUT", - url.encode("ascii"), - json.dumps({"history_visibility": "world_readable"}), - access_token=self.other_user_token, - ) - self.render(request) - self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) - - # Test that the admin can still send shutdown - url = "admin/shutdown_room/" + room_id - request, channel = self.make_request( - "POST", - url.encode("ascii"), - json.dumps({"new_room_user_id": self.admin_user}), - access_token=self.admin_user_tok, - ) - self.render(request) - - self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) - - # Assert we can no longer peek into the room - self._assert_peek(room_id, expect_code=403) - - def _assert_peek(self, room_id, expect_code): - """Assert that the admin user can (or cannot) peek into the room. - """ - - url = "rooms/%s/initialSync" % (room_id,) - request, channel = self.make_request( - "GET", url.encode("ascii"), access_token=self.admin_user_tok - ) - self.render(request) - self.assertEqual( - expect_code, int(channel.result["code"]), msg=channel.result["body"] - ) - - url = "events?timeout=0&room_id=" + room_id - request, channel = self.make_request( - "GET", url.encode("ascii"), access_token=self.admin_user_tok - ) - self.render(request) - self.assertEqual( - expect_code, int(channel.result["code"]), msg=channel.result["body"] - ) - - class DeleteGroupTestCase(unittest.HomeserverTestCase): servlets = [ synapse.rest.admin.register_servlets_for_client_rest_resource, @@ -273,86 +149,6 @@ class DeleteGroupTestCase(unittest.HomeserverTestCase): return channel.json_body["groups"] -class PurgeRoomTestCase(unittest.HomeserverTestCase): - """Test /purge_room admin API. - """ - - servlets = [ - synapse.rest.admin.register_servlets, - login.register_servlets, - room.register_servlets, - ] - - def prepare(self, reactor, clock, hs): - self.store = hs.get_datastore() - - self.admin_user = self.register_user("admin", "pass", admin=True) - self.admin_user_tok = self.login("admin", "pass") - - def test_purge_room(self): - room_id = self.helper.create_room_as(self.admin_user, tok=self.admin_user_tok) - - # All users have to have left the room. - self.helper.leave(room_id, user=self.admin_user, tok=self.admin_user_tok) - - url = "/_synapse/admin/v1/purge_room" - request, channel = self.make_request( - "POST", - url.encode("ascii"), - {"room_id": room_id}, - access_token=self.admin_user_tok, - ) - self.render(request) - - self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) - - # Test that the following tables have been purged of all rows related to the room. - for table in ( - "current_state_events", - "event_backward_extremities", - "event_forward_extremities", - "event_json", - "event_push_actions", - "event_search", - "events", - "group_rooms", - "public_room_list_stream", - "receipts_graph", - "receipts_linearized", - "room_aliases", - "room_depth", - "room_memberships", - "room_stats_state", - "room_stats_current", - "room_stats_historical", - "room_stats_earliest_token", - "rooms", - "stream_ordering_to_exterm", - "users_in_public_rooms", - "users_who_share_private_rooms", - "appservice_room_list", - "e2e_room_keys", - "event_push_summary", - "pusher_throttle", - "group_summary_rooms", - "local_invites", - "room_account_data", - "room_tags", - # "state_groups", # Current impl leaves orphaned state groups around. - "state_groups_state", - ): - count = self.get_success( - self.store.db.simple_select_one_onecol( - table=table, - keyvalues={"room_id": room_id}, - retcol="COUNT(*)", - desc="test_purge_room", - ) - ) - - self.assertEqual(count, 0, msg="Rows not purged in {}".format(table)) - - class QuarantineMediaTestCase(unittest.HomeserverTestCase): """Test /quarantine_media admin API. """ @@ -691,389 +487,3 @@ class QuarantineMediaTestCase(unittest.HomeserverTestCase): % server_and_media_id_2 ), ) - - -class RoomTestCase(unittest.HomeserverTestCase): - """Test /room admin API. - """ - - servlets = [ - synapse.rest.admin.register_servlets, - login.register_servlets, - room.register_servlets, - directory.register_servlets, - ] - - def prepare(self, reactor, clock, hs): - self.store = hs.get_datastore() - - # Create user - self.admin_user = self.register_user("admin", "pass", admin=True) - self.admin_user_tok = self.login("admin", "pass") - - def test_list_rooms(self): - """Test that we can list rooms""" - # Create 3 test rooms - total_rooms = 3 - room_ids = [] - for x in range(total_rooms): - room_id = self.helper.create_room_as( - self.admin_user, tok=self.admin_user_tok - ) - room_ids.append(room_id) - - # Request the list of rooms - url = "/_synapse/admin/v1/rooms" - request, channel = self.make_request( - "GET", url.encode("ascii"), access_token=self.admin_user_tok, - ) - self.render(request) - - # Check request completed successfully - self.assertEqual(200, int(channel.code), msg=channel.json_body) - - # Check that response json body contains a "rooms" key - self.assertTrue( - "rooms" in channel.json_body, - msg="Response body does not " "contain a 'rooms' key", - ) - - # Check that 3 rooms were returned - self.assertEqual(3, len(channel.json_body["rooms"]), msg=channel.json_body) - - # Check their room_ids match - returned_room_ids = [room["room_id"] for room in channel.json_body["rooms"]] - self.assertEqual(room_ids, returned_room_ids) - - # Check that all fields are available - for r in channel.json_body["rooms"]: - self.assertIn("name", r) - self.assertIn("canonical_alias", r) - self.assertIn("joined_members", r) - - # Check that the correct number of total rooms was returned - self.assertEqual(channel.json_body["total_rooms"], total_rooms) - - # Check that the offset is correct - # Should be 0 as we aren't paginating - self.assertEqual(channel.json_body["offset"], 0) - - # Check that the prev_batch parameter is not present - self.assertNotIn("prev_batch", channel.json_body) - - # We shouldn't receive a next token here as there's no further rooms to show - self.assertNotIn("next_batch", channel.json_body) - - def test_list_rooms_pagination(self): - """Test that we can get a full list of rooms through pagination""" - # Create 5 test rooms - total_rooms = 5 - room_ids = [] - for x in range(total_rooms): - room_id = self.helper.create_room_as( - self.admin_user, tok=self.admin_user_tok - ) - room_ids.append(room_id) - - # Set the name of the rooms so we get a consistent returned ordering - for idx, room_id in enumerate(room_ids): - self.helper.send_state( - room_id, "m.room.name", {"name": str(idx)}, tok=self.admin_user_tok, - ) - - # Request the list of rooms - returned_room_ids = [] - start = 0 - limit = 2 - - run_count = 0 - should_repeat = True - while should_repeat: - run_count += 1 - - url = "/_synapse/admin/v1/rooms?from=%d&limit=%d&order_by=%s" % ( - start, - limit, - "alphabetical", - ) - request, channel = self.make_request( - "GET", url.encode("ascii"), access_token=self.admin_user_tok, - ) - self.render(request) - self.assertEqual( - 200, int(channel.result["code"]), msg=channel.result["body"] - ) - - self.assertTrue("rooms" in channel.json_body) - for r in channel.json_body["rooms"]: - returned_room_ids.append(r["room_id"]) - - # Check that the correct number of total rooms was returned - self.assertEqual(channel.json_body["total_rooms"], total_rooms) - - # Check that the offset is correct - # We're only getting 2 rooms each page, so should be 2 * last run_count - self.assertEqual(channel.json_body["offset"], 2 * (run_count - 1)) - - if run_count > 1: - # Check the value of prev_batch is correct - self.assertEqual(channel.json_body["prev_batch"], 2 * (run_count - 2)) - - if "next_batch" not in channel.json_body: - # We have reached the end of the list - should_repeat = False - else: - # Make another query with an updated start value - start = channel.json_body["next_batch"] - - # We should've queried the endpoint 3 times - self.assertEqual( - run_count, - 3, - msg="Should've queried 3 times for 5 rooms with limit 2 per query", - ) - - # Check that we received all of the room ids - self.assertEqual(room_ids, returned_room_ids) - - url = "/_synapse/admin/v1/rooms?from=%d&limit=%d" % (start, limit) - request, channel = self.make_request( - "GET", url.encode("ascii"), access_token=self.admin_user_tok, - ) - self.render(request) - self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) - - def test_correct_room_attributes(self): - """Test the correct attributes for a room are returned""" - # Create a test room - room_id = self.helper.create_room_as(self.admin_user, tok=self.admin_user_tok) - - test_alias = "#test:test" - test_room_name = "something" - - # Have another user join the room - user_2 = self.register_user("user4", "pass") - user_tok_2 = self.login("user4", "pass") - self.helper.join(room_id, user_2, tok=user_tok_2) - - # Create a new alias to this room - url = "/_matrix/client/r0/directory/room/%s" % (urllib.parse.quote(test_alias),) - request, channel = self.make_request( - "PUT", - url.encode("ascii"), - {"room_id": room_id}, - access_token=self.admin_user_tok, - ) - self.render(request) - self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) - - # Set this new alias as the canonical alias for this room - self.helper.send_state( - room_id, - "m.room.aliases", - {"aliases": [test_alias]}, - tok=self.admin_user_tok, - state_key="test", - ) - self.helper.send_state( - room_id, - "m.room.canonical_alias", - {"alias": test_alias}, - tok=self.admin_user_tok, - ) - - # Set a name for the room - self.helper.send_state( - room_id, "m.room.name", {"name": test_room_name}, tok=self.admin_user_tok, - ) - - # Request the list of rooms - url = "/_synapse/admin/v1/rooms" - request, channel = self.make_request( - "GET", url.encode("ascii"), access_token=self.admin_user_tok, - ) - self.render(request) - self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) - - # Check that rooms were returned - self.assertTrue("rooms" in channel.json_body) - rooms = channel.json_body["rooms"] - - # Check that only one room was returned - self.assertEqual(len(rooms), 1) - - # And that the value of the total_rooms key was correct - self.assertEqual(channel.json_body["total_rooms"], 1) - - # Check that the offset is correct - # We're not paginating, so should be 0 - self.assertEqual(channel.json_body["offset"], 0) - - # Check that there is no `prev_batch` - self.assertNotIn("prev_batch", channel.json_body) - - # Check that there is no `next_batch` - self.assertNotIn("next_batch", channel.json_body) - - # Check that all provided attributes are set - r = rooms[0] - self.assertEqual(room_id, r["room_id"]) - self.assertEqual(test_room_name, r["name"]) - self.assertEqual(test_alias, r["canonical_alias"]) - - def test_room_list_sort_order(self): - """Test room list sort ordering. alphabetical versus number of members, - reversing the order, etc. - """ - # Create 3 test rooms - room_id_1 = self.helper.create_room_as(self.admin_user, tok=self.admin_user_tok) - room_id_2 = self.helper.create_room_as(self.admin_user, tok=self.admin_user_tok) - room_id_3 = self.helper.create_room_as(self.admin_user, tok=self.admin_user_tok) - - # Set room names in alphabetical order. room 1 -> A, 2 -> B, 3 -> C - self.helper.send_state( - room_id_1, "m.room.name", {"name": "A"}, tok=self.admin_user_tok, - ) - self.helper.send_state( - room_id_2, "m.room.name", {"name": "B"}, tok=self.admin_user_tok, - ) - self.helper.send_state( - room_id_3, "m.room.name", {"name": "C"}, tok=self.admin_user_tok, - ) - - # Set room member size in the reverse order. room 1 -> 1 member, 2 -> 2, 3 -> 3 - user_1 = self.register_user("bob1", "pass") - user_1_tok = self.login("bob1", "pass") - self.helper.join(room_id_2, user_1, tok=user_1_tok) - - user_2 = self.register_user("bob2", "pass") - user_2_tok = self.login("bob2", "pass") - self.helper.join(room_id_3, user_2, tok=user_2_tok) - - user_3 = self.register_user("bob3", "pass") - user_3_tok = self.login("bob3", "pass") - self.helper.join(room_id_3, user_3, tok=user_3_tok) - - def _order_test( - order_type: str, expected_room_list: List[str], reverse: bool = False, - ): - """Request the list of rooms in a certain order. Assert that order is what - we expect - - Args: - order_type: The type of ordering to give the server - expected_room_list: The list of room_ids in the order we expect to get - back from the server - """ - # Request the list of rooms in the given order - url = "/_synapse/admin/v1/rooms?order_by=%s" % (order_type,) - if reverse: - url += "&dir=b" - request, channel = self.make_request( - "GET", url.encode("ascii"), access_token=self.admin_user_tok, - ) - self.render(request) - self.assertEqual(200, channel.code, msg=channel.json_body) - - # Check that rooms were returned - self.assertTrue("rooms" in channel.json_body) - rooms = channel.json_body["rooms"] - - # Check for the correct total_rooms value - self.assertEqual(channel.json_body["total_rooms"], 3) - - # Check that the offset is correct - # We're not paginating, so should be 0 - self.assertEqual(channel.json_body["offset"], 0) - - # Check that there is no `prev_batch` - self.assertNotIn("prev_batch", channel.json_body) - - # Check that there is no `next_batch` - self.assertNotIn("next_batch", channel.json_body) - - # Check that rooms were returned in alphabetical order - returned_order = [r["room_id"] for r in rooms] - self.assertListEqual(expected_room_list, returned_order) # order is checked - - # Test different sort orders, with forward and reverse directions - _order_test("alphabetical", [room_id_1, room_id_2, room_id_3]) - _order_test("alphabetical", [room_id_3, room_id_2, room_id_1], reverse=True) - - _order_test("size", [room_id_3, room_id_2, room_id_1]) - _order_test("size", [room_id_1, room_id_2, room_id_3], reverse=True) - - def test_search_term(self): - """Test that searching for a room works correctly""" - # Create two test rooms - room_id_1 = self.helper.create_room_as(self.admin_user, tok=self.admin_user_tok) - room_id_2 = self.helper.create_room_as(self.admin_user, tok=self.admin_user_tok) - - room_name_1 = "something" - room_name_2 = "else" - - # Set the name for each room - self.helper.send_state( - room_id_1, "m.room.name", {"name": room_name_1}, tok=self.admin_user_tok, - ) - self.helper.send_state( - room_id_2, "m.room.name", {"name": room_name_2}, tok=self.admin_user_tok, - ) - - def _search_test( - expected_room_id: Optional[str], - search_term: str, - expected_http_code: int = 200, - ): - """Search for a room and check that the returned room's id is a match - - Args: - expected_room_id: The room_id expected to be returned by the API. Set - to None to expect zero results for the search - search_term: The term to search for room names with - expected_http_code: The expected http code for the request - """ - url = "/_synapse/admin/v1/rooms?search_term=%s" % (search_term,) - request, channel = self.make_request( - "GET", url.encode("ascii"), access_token=self.admin_user_tok, - ) - self.render(request) - self.assertEqual(expected_http_code, channel.code, msg=channel.json_body) - - if expected_http_code != 200: - return - - # Check that rooms were returned - self.assertTrue("rooms" in channel.json_body) - rooms = channel.json_body["rooms"] - - # Check that the expected number of rooms were returned - expected_room_count = 1 if expected_room_id else 0 - self.assertEqual(len(rooms), expected_room_count) - self.assertEqual(channel.json_body["total_rooms"], expected_room_count) - - # Check that the offset is correct - # We're not paginating, so should be 0 - self.assertEqual(channel.json_body["offset"], 0) - - # Check that there is no `prev_batch` - self.assertNotIn("prev_batch", channel.json_body) - - # Check that there is no `next_batch` - self.assertNotIn("next_batch", channel.json_body) - - if expected_room_id: - # Check that the first returned room id is correct - r = rooms[0] - self.assertEqual(expected_room_id, r["room_id"]) - - # Perform search tests - _search_test(room_id_1, "something") - _search_test(room_id_1, "thing") - - _search_test(room_id_2, "else") - _search_test(room_id_2, "se") - - _search_test(None, "foo") - _search_test(None, "bar") - _search_test(None, "", expected_http_code=400) diff --git a/tests/rest/admin/test_room.py b/tests/rest/admin/test_room.py index 672cc3eac5..249c93722f 100644 --- a/tests/rest/admin/test_room.py +++ b/tests/rest/admin/test_room.py @@ -14,16 +14,694 @@ # limitations under the License. import json +import urllib.parse +from typing import List, Optional + +from mock import Mock import synapse.rest.admin from synapse.api.errors import Codes -from synapse.rest.client.v1 import login, room +from synapse.rest.client.v1 import directory, events, login, room from tests import unittest """Tests admin REST events for /rooms paths.""" +class ShutdownRoomTestCase(unittest.HomeserverTestCase): + servlets = [ + synapse.rest.admin.register_servlets_for_client_rest_resource, + login.register_servlets, + events.register_servlets, + room.register_servlets, + room.register_deprecated_servlets, + ] + + def prepare(self, reactor, clock, hs): + self.event_creation_handler = hs.get_event_creation_handler() + hs.config.user_consent_version = "1" + + consent_uri_builder = Mock() + consent_uri_builder.build_user_consent_uri.return_value = "http://example.com" + self.event_creation_handler._consent_uri_builder = consent_uri_builder + + self.store = hs.get_datastore() + + self.admin_user = self.register_user("admin", "pass", admin=True) + self.admin_user_tok = self.login("admin", "pass") + + self.other_user = self.register_user("user", "pass") + self.other_user_token = self.login("user", "pass") + + # Mark the admin user as having consented + self.get_success(self.store.user_set_consent_version(self.admin_user, "1")) + + def test_shutdown_room_consent(self): + """Test that we can shutdown rooms with local users who have not + yet accepted the privacy policy. This used to fail when we tried to + force part the user from the old room. + """ + self.event_creation_handler._block_events_without_consent_error = None + + room_id = self.helper.create_room_as(self.other_user, tok=self.other_user_token) + + # Assert one user in room + users_in_room = self.get_success(self.store.get_users_in_room(room_id)) + self.assertEqual([self.other_user], users_in_room) + + # Enable require consent to send events + self.event_creation_handler._block_events_without_consent_error = "Error" + + # Assert that the user is getting consent error + self.helper.send( + room_id, body="foo", tok=self.other_user_token, expect_code=403 + ) + + # Test that the admin can still send shutdown + url = "admin/shutdown_room/" + room_id + request, channel = self.make_request( + "POST", + url.encode("ascii"), + json.dumps({"new_room_user_id": self.admin_user}), + access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + + # Assert there is now no longer anyone in the room + users_in_room = self.get_success(self.store.get_users_in_room(room_id)) + self.assertEqual([], users_in_room) + + def test_shutdown_room_block_peek(self): + """Test that a world_readable room can no longer be peeked into after + it has been shut down. + """ + + self.event_creation_handler._block_events_without_consent_error = None + + room_id = self.helper.create_room_as(self.other_user, tok=self.other_user_token) + + # Enable world readable + url = "rooms/%s/state/m.room.history_visibility" % (room_id,) + request, channel = self.make_request( + "PUT", + url.encode("ascii"), + json.dumps({"history_visibility": "world_readable"}), + access_token=self.other_user_token, + ) + self.render(request) + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + + # Test that the admin can still send shutdown + url = "admin/shutdown_room/" + room_id + request, channel = self.make_request( + "POST", + url.encode("ascii"), + json.dumps({"new_room_user_id": self.admin_user}), + access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + + # Assert we can no longer peek into the room + self._assert_peek(room_id, expect_code=403) + + def _assert_peek(self, room_id, expect_code): + """Assert that the admin user can (or cannot) peek into the room. + """ + + url = "rooms/%s/initialSync" % (room_id,) + request, channel = self.make_request( + "GET", url.encode("ascii"), access_token=self.admin_user_tok + ) + self.render(request) + self.assertEqual( + expect_code, int(channel.result["code"]), msg=channel.result["body"] + ) + + url = "events?timeout=0&room_id=" + room_id + request, channel = self.make_request( + "GET", url.encode("ascii"), access_token=self.admin_user_tok + ) + self.render(request) + self.assertEqual( + expect_code, int(channel.result["code"]), msg=channel.result["body"] + ) + + +class PurgeRoomTestCase(unittest.HomeserverTestCase): + """Test /purge_room admin API. + """ + + servlets = [ + synapse.rest.admin.register_servlets, + login.register_servlets, + room.register_servlets, + ] + + def prepare(self, reactor, clock, hs): + self.store = hs.get_datastore() + + self.admin_user = self.register_user("admin", "pass", admin=True) + self.admin_user_tok = self.login("admin", "pass") + + def test_purge_room(self): + room_id = self.helper.create_room_as(self.admin_user, tok=self.admin_user_tok) + + # All users have to have left the room. + self.helper.leave(room_id, user=self.admin_user, tok=self.admin_user_tok) + + url = "/_synapse/admin/v1/purge_room" + request, channel = self.make_request( + "POST", + url.encode("ascii"), + {"room_id": room_id}, + access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + + # Test that the following tables have been purged of all rows related to the room. + for table in ( + "current_state_events", + "event_backward_extremities", + "event_forward_extremities", + "event_json", + "event_push_actions", + "event_search", + "events", + "group_rooms", + "public_room_list_stream", + "receipts_graph", + "receipts_linearized", + "room_aliases", + "room_depth", + "room_memberships", + "room_stats_state", + "room_stats_current", + "room_stats_historical", + "room_stats_earliest_token", + "rooms", + "stream_ordering_to_exterm", + "users_in_public_rooms", + "users_who_share_private_rooms", + "appservice_room_list", + "e2e_room_keys", + "event_push_summary", + "pusher_throttle", + "group_summary_rooms", + "local_invites", + "room_account_data", + "room_tags", + # "state_groups", # Current impl leaves orphaned state groups around. + "state_groups_state", + ): + count = self.get_success( + self.store.db.simple_select_one_onecol( + table=table, + keyvalues={"room_id": room_id}, + retcol="COUNT(*)", + desc="test_purge_room", + ) + ) + + self.assertEqual(count, 0, msg="Rows not purged in {}".format(table)) + + +class RoomTestCase(unittest.HomeserverTestCase): + """Test /room admin API. + """ + + servlets = [ + synapse.rest.admin.register_servlets, + login.register_servlets, + room.register_servlets, + directory.register_servlets, + ] + + def prepare(self, reactor, clock, hs): + self.store = hs.get_datastore() + + # Create user + self.admin_user = self.register_user("admin", "pass", admin=True) + self.admin_user_tok = self.login("admin", "pass") + + def test_list_rooms(self): + """Test that we can list rooms""" + # Create 3 test rooms + total_rooms = 3 + room_ids = [] + for x in range(total_rooms): + room_id = self.helper.create_room_as( + self.admin_user, tok=self.admin_user_tok + ) + room_ids.append(room_id) + + # Request the list of rooms + url = "/_synapse/admin/v1/rooms" + request, channel = self.make_request( + "GET", url.encode("ascii"), access_token=self.admin_user_tok, + ) + self.render(request) + + # Check request completed successfully + self.assertEqual(200, int(channel.code), msg=channel.json_body) + + # Check that response json body contains a "rooms" key + self.assertTrue( + "rooms" in channel.json_body, + msg="Response body does not " "contain a 'rooms' key", + ) + + # Check that 3 rooms were returned + self.assertEqual(3, len(channel.json_body["rooms"]), msg=channel.json_body) + + # Check their room_ids match + returned_room_ids = [room["room_id"] for room in channel.json_body["rooms"]] + self.assertEqual(room_ids, returned_room_ids) + + # Check that all fields are available + for r in channel.json_body["rooms"]: + self.assertIn("name", r) + self.assertIn("canonical_alias", r) + self.assertIn("joined_members", r) + self.assertIn("joined_local_members", r) + self.assertIn("version", r) + self.assertIn("creator", r) + self.assertIn("encryption", r) + self.assertIn("federatable", r) + self.assertIn("public", r) + self.assertIn("join_rules", r) + self.assertIn("guest_access", r) + self.assertIn("history_visibility", r) + self.assertIn("state_events", r) + + # Check that the correct number of total rooms was returned + self.assertEqual(channel.json_body["total_rooms"], total_rooms) + + # Check that the offset is correct + # Should be 0 as we aren't paginating + self.assertEqual(channel.json_body["offset"], 0) + + # Check that the prev_batch parameter is not present + self.assertNotIn("prev_batch", channel.json_body) + + # We shouldn't receive a next token here as there's no further rooms to show + self.assertNotIn("next_batch", channel.json_body) + + def test_list_rooms_pagination(self): + """Test that we can get a full list of rooms through pagination""" + # Create 5 test rooms + total_rooms = 5 + room_ids = [] + for x in range(total_rooms): + room_id = self.helper.create_room_as( + self.admin_user, tok=self.admin_user_tok + ) + room_ids.append(room_id) + + # Set the name of the rooms so we get a consistent returned ordering + for idx, room_id in enumerate(room_ids): + self.helper.send_state( + room_id, "m.room.name", {"name": str(idx)}, tok=self.admin_user_tok, + ) + + # Request the list of rooms + returned_room_ids = [] + start = 0 + limit = 2 + + run_count = 0 + should_repeat = True + while should_repeat: + run_count += 1 + + url = "/_synapse/admin/v1/rooms?from=%d&limit=%d&order_by=%s" % ( + start, + limit, + "name", + ) + request, channel = self.make_request( + "GET", url.encode("ascii"), access_token=self.admin_user_tok, + ) + self.render(request) + self.assertEqual( + 200, int(channel.result["code"]), msg=channel.result["body"] + ) + + self.assertTrue("rooms" in channel.json_body) + for r in channel.json_body["rooms"]: + returned_room_ids.append(r["room_id"]) + + # Check that the correct number of total rooms was returned + self.assertEqual(channel.json_body["total_rooms"], total_rooms) + + # Check that the offset is correct + # We're only getting 2 rooms each page, so should be 2 * last run_count + self.assertEqual(channel.json_body["offset"], 2 * (run_count - 1)) + + if run_count > 1: + # Check the value of prev_batch is correct + self.assertEqual(channel.json_body["prev_batch"], 2 * (run_count - 2)) + + if "next_batch" not in channel.json_body: + # We have reached the end of the list + should_repeat = False + else: + # Make another query with an updated start value + start = channel.json_body["next_batch"] + + # We should've queried the endpoint 3 times + self.assertEqual( + run_count, + 3, + msg="Should've queried 3 times for 5 rooms with limit 2 per query", + ) + + # Check that we received all of the room ids + self.assertEqual(room_ids, returned_room_ids) + + url = "/_synapse/admin/v1/rooms?from=%d&limit=%d" % (start, limit) + request, channel = self.make_request( + "GET", url.encode("ascii"), access_token=self.admin_user_tok, + ) + self.render(request) + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + + def test_correct_room_attributes(self): + """Test the correct attributes for a room are returned""" + # Create a test room + room_id = self.helper.create_room_as(self.admin_user, tok=self.admin_user_tok) + + test_alias = "#test:test" + test_room_name = "something" + + # Have another user join the room + user_2 = self.register_user("user4", "pass") + user_tok_2 = self.login("user4", "pass") + self.helper.join(room_id, user_2, tok=user_tok_2) + + # Create a new alias to this room + url = "/_matrix/client/r0/directory/room/%s" % (urllib.parse.quote(test_alias),) + request, channel = self.make_request( + "PUT", + url.encode("ascii"), + {"room_id": room_id}, + access_token=self.admin_user_tok, + ) + self.render(request) + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + + # Set this new alias as the canonical alias for this room + self.helper.send_state( + room_id, + "m.room.aliases", + {"aliases": [test_alias]}, + tok=self.admin_user_tok, + state_key="test", + ) + self.helper.send_state( + room_id, + "m.room.canonical_alias", + {"alias": test_alias}, + tok=self.admin_user_tok, + ) + + # Set a name for the room + self.helper.send_state( + room_id, "m.room.name", {"name": test_room_name}, tok=self.admin_user_tok, + ) + + # Request the list of rooms + url = "/_synapse/admin/v1/rooms" + request, channel = self.make_request( + "GET", url.encode("ascii"), access_token=self.admin_user_tok, + ) + self.render(request) + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + + # Check that rooms were returned + self.assertTrue("rooms" in channel.json_body) + rooms = channel.json_body["rooms"] + + # Check that only one room was returned + self.assertEqual(len(rooms), 1) + + # And that the value of the total_rooms key was correct + self.assertEqual(channel.json_body["total_rooms"], 1) + + # Check that the offset is correct + # We're not paginating, so should be 0 + self.assertEqual(channel.json_body["offset"], 0) + + # Check that there is no `prev_batch` + self.assertNotIn("prev_batch", channel.json_body) + + # Check that there is no `next_batch` + self.assertNotIn("next_batch", channel.json_body) + + # Check that all provided attributes are set + r = rooms[0] + self.assertEqual(room_id, r["room_id"]) + self.assertEqual(test_room_name, r["name"]) + self.assertEqual(test_alias, r["canonical_alias"]) + + def test_room_list_sort_order(self): + """Test room list sort ordering. alphabetical name versus number of members, + reversing the order, etc. + """ + + def _set_canonical_alias(room_id: str, test_alias: str, admin_user_tok: str): + # Create a new alias to this room + url = "/_matrix/client/r0/directory/room/%s" % ( + urllib.parse.quote(test_alias), + ) + request, channel = self.make_request( + "PUT", + url.encode("ascii"), + {"room_id": room_id}, + access_token=admin_user_tok, + ) + self.render(request) + self.assertEqual( + 200, int(channel.result["code"]), msg=channel.result["body"] + ) + + # Set this new alias as the canonical alias for this room + self.helper.send_state( + room_id, + "m.room.aliases", + {"aliases": [test_alias]}, + tok=admin_user_tok, + state_key="test", + ) + self.helper.send_state( + room_id, + "m.room.canonical_alias", + {"alias": test_alias}, + tok=admin_user_tok, + ) + + def _order_test( + order_type: str, expected_room_list: List[str], reverse: bool = False, + ): + """Request the list of rooms in a certain order. Assert that order is what + we expect + + Args: + order_type: The type of ordering to give the server + expected_room_list: The list of room_ids in the order we expect to get + back from the server + """ + # Request the list of rooms in the given order + url = "/_synapse/admin/v1/rooms?order_by=%s" % (order_type,) + if reverse: + url += "&dir=b" + request, channel = self.make_request( + "GET", url.encode("ascii"), access_token=self.admin_user_tok, + ) + self.render(request) + self.assertEqual(200, channel.code, msg=channel.json_body) + + # Check that rooms were returned + self.assertTrue("rooms" in channel.json_body) + rooms = channel.json_body["rooms"] + + # Check for the correct total_rooms value + self.assertEqual(channel.json_body["total_rooms"], 3) + + # Check that the offset is correct + # We're not paginating, so should be 0 + self.assertEqual(channel.json_body["offset"], 0) + + # Check that there is no `prev_batch` + self.assertNotIn("prev_batch", channel.json_body) + + # Check that there is no `next_batch` + self.assertNotIn("next_batch", channel.json_body) + + # Check that rooms were returned in alphabetical order + returned_order = [r["room_id"] for r in rooms] + self.assertListEqual(expected_room_list, returned_order) # order is checked + + # Create 3 test rooms + room_id_1 = self.helper.create_room_as(self.admin_user, tok=self.admin_user_tok) + room_id_2 = self.helper.create_room_as(self.admin_user, tok=self.admin_user_tok) + room_id_3 = self.helper.create_room_as(self.admin_user, tok=self.admin_user_tok) + + # Set room names in alphabetical order. room 1 -> A, 2 -> B, 3 -> C + self.helper.send_state( + room_id_1, "m.room.name", {"name": "A"}, tok=self.admin_user_tok, + ) + self.helper.send_state( + room_id_2, "m.room.name", {"name": "B"}, tok=self.admin_user_tok, + ) + self.helper.send_state( + room_id_3, "m.room.name", {"name": "C"}, tok=self.admin_user_tok, + ) + + # Set room canonical room aliases + _set_canonical_alias(room_id_1, "#A_alias:test", self.admin_user_tok) + _set_canonical_alias(room_id_2, "#B_alias:test", self.admin_user_tok) + _set_canonical_alias(room_id_3, "#C_alias:test", self.admin_user_tok) + + # Set room member size in the reverse order. room 1 -> 1 member, 2 -> 2, 3 -> 3 + user_1 = self.register_user("bob1", "pass") + user_1_tok = self.login("bob1", "pass") + self.helper.join(room_id_2, user_1, tok=user_1_tok) + + user_2 = self.register_user("bob2", "pass") + user_2_tok = self.login("bob2", "pass") + self.helper.join(room_id_3, user_2, tok=user_2_tok) + + user_3 = self.register_user("bob3", "pass") + user_3_tok = self.login("bob3", "pass") + self.helper.join(room_id_3, user_3, tok=user_3_tok) + + # Test different sort orders, with forward and reverse directions + _order_test("name", [room_id_1, room_id_2, room_id_3]) + _order_test("name", [room_id_3, room_id_2, room_id_1], reverse=True) + + _order_test("canonical_alias", [room_id_1, room_id_2, room_id_3]) + _order_test("canonical_alias", [room_id_3, room_id_2, room_id_1], reverse=True) + + _order_test("joined_members", [room_id_3, room_id_2, room_id_1]) + _order_test("joined_members", [room_id_1, room_id_2, room_id_3], reverse=True) + + _order_test("joined_local_members", [room_id_3, room_id_2, room_id_1]) + _order_test( + "joined_local_members", [room_id_1, room_id_2, room_id_3], reverse=True + ) + + _order_test("version", [room_id_1, room_id_2, room_id_3]) + _order_test("version", [room_id_1, room_id_2, room_id_3], reverse=True) + + _order_test("creator", [room_id_1, room_id_2, room_id_3]) + _order_test("creator", [room_id_1, room_id_2, room_id_3], reverse=True) + + _order_test("encryption", [room_id_1, room_id_2, room_id_3]) + _order_test("encryption", [room_id_1, room_id_2, room_id_3], reverse=True) + + _order_test("federatable", [room_id_1, room_id_2, room_id_3]) + _order_test("federatable", [room_id_1, room_id_2, room_id_3], reverse=True) + + _order_test("public", [room_id_1, room_id_2, room_id_3]) + # Different sort order of SQlite and PostreSQL + # _order_test("public", [room_id_3, room_id_2, room_id_1], reverse=True) + + _order_test("join_rules", [room_id_1, room_id_2, room_id_3]) + _order_test("join_rules", [room_id_1, room_id_2, room_id_3], reverse=True) + + _order_test("guest_access", [room_id_1, room_id_2, room_id_3]) + _order_test("guest_access", [room_id_1, room_id_2, room_id_3], reverse=True) + + _order_test("history_visibility", [room_id_1, room_id_2, room_id_3]) + _order_test( + "history_visibility", [room_id_1, room_id_2, room_id_3], reverse=True + ) + + _order_test("state_events", [room_id_3, room_id_2, room_id_1]) + _order_test("state_events", [room_id_1, room_id_2, room_id_3], reverse=True) + + def test_search_term(self): + """Test that searching for a room works correctly""" + # Create two test rooms + room_id_1 = self.helper.create_room_as(self.admin_user, tok=self.admin_user_tok) + room_id_2 = self.helper.create_room_as(self.admin_user, tok=self.admin_user_tok) + + room_name_1 = "something" + room_name_2 = "else" + + # Set the name for each room + self.helper.send_state( + room_id_1, "m.room.name", {"name": room_name_1}, tok=self.admin_user_tok, + ) + self.helper.send_state( + room_id_2, "m.room.name", {"name": room_name_2}, tok=self.admin_user_tok, + ) + + def _search_test( + expected_room_id: Optional[str], + search_term: str, + expected_http_code: int = 200, + ): + """Search for a room and check that the returned room's id is a match + + Args: + expected_room_id: The room_id expected to be returned by the API. Set + to None to expect zero results for the search + search_term: The term to search for room names with + expected_http_code: The expected http code for the request + """ + url = "/_synapse/admin/v1/rooms?search_term=%s" % (search_term,) + request, channel = self.make_request( + "GET", url.encode("ascii"), access_token=self.admin_user_tok, + ) + self.render(request) + self.assertEqual(expected_http_code, channel.code, msg=channel.json_body) + + if expected_http_code != 200: + return + + # Check that rooms were returned + self.assertTrue("rooms" in channel.json_body) + rooms = channel.json_body["rooms"] + + # Check that the expected number of rooms were returned + expected_room_count = 1 if expected_room_id else 0 + self.assertEqual(len(rooms), expected_room_count) + self.assertEqual(channel.json_body["total_rooms"], expected_room_count) + + # Check that the offset is correct + # We're not paginating, so should be 0 + self.assertEqual(channel.json_body["offset"], 0) + + # Check that there is no `prev_batch` + self.assertNotIn("prev_batch", channel.json_body) + + # Check that there is no `next_batch` + self.assertNotIn("next_batch", channel.json_body) + + if expected_room_id: + # Check that the first returned room id is correct + r = rooms[0] + self.assertEqual(expected_room_id, r["room_id"]) + + # Perform search tests + _search_test(room_id_1, "something") + _search_test(room_id_1, "thing") + + _search_test(room_id_2, "else") + _search_test(room_id_2, "se") + + _search_test(None, "foo") + _search_test(None, "bar") + _search_test(None, "", expected_http_code=400) + + class JoinAliasRoomTestCase(unittest.HomeserverTestCase): servlets = [ -- cgit 1.4.1 From 13683a3a223a3f08b295973e7b8aa6e9299a2fa5 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Wed, 22 Apr 2020 13:45:40 +0100 Subject: Extend StreamChangeCache to support multiple entities per stream ID (#7303) First some background: StreamChangeCache is used to keep track of what "entities" have changed since a given stream ID. So for example, we might use it to keep track of when the last to-device message for a given user was received [1], and hence whether we need to pull any to-device messages from the database on a sync [2]. Now, it turns out that StreamChangeCache didn't support more than one thing being changed at a given stream_id (this was part of the problem with #7206). However, it's entirely valid to send to-device messages to more than one user at a time. As it turns out, this did in fact work, because *some* methods of StreamChangeCache coped ok with having multiple things changing on the same stream ID, and it seems we never actually use the methods which don't work on the stream change caches where we allow multiple changes at the same stream ID. But that feels horribly fragile, hence: let's update StreamChangeCache to properly support this, and add some typing and some more tests while we're at it. [1]: https://github.com/matrix-org/synapse/blob/release-v1.12.3/synapse/storage/data_stores/main/deviceinbox.py#L301 [2]: https://github.com/matrix-org/synapse/blob/release-v1.12.3/synapse/storage/data_stores/main/deviceinbox.py#L47-L51 --- changelog.d/7303.misc | 1 + stubs/sortedcontainers/__init__.pyi | 13 +++ stubs/sortedcontainers/sorteddict.pyi | 124 +++++++++++++++++++++++++++++ synapse/util/caches/stream_change_cache.py | 117 ++++++++++++++++----------- tests/util/test_stream_change_cache.py | 69 +++++++++++++--- tox.ini | 4 +- 6 files changed, 272 insertions(+), 56 deletions(-) create mode 100644 changelog.d/7303.misc create mode 100644 stubs/sortedcontainers/__init__.pyi create mode 100644 stubs/sortedcontainers/sorteddict.pyi (limited to 'tests') diff --git a/changelog.d/7303.misc b/changelog.d/7303.misc new file mode 100644 index 0000000000..aa89c2b254 --- /dev/null +++ b/changelog.d/7303.misc @@ -0,0 +1 @@ +Fix StreamChangeCache to work with multiple entities changing on the same stream id. diff --git a/stubs/sortedcontainers/__init__.pyi b/stubs/sortedcontainers/__init__.pyi new file mode 100644 index 0000000000..073b806d3c --- /dev/null +++ b/stubs/sortedcontainers/__init__.pyi @@ -0,0 +1,13 @@ +from .sorteddict import ( + SortedDict, + SortedKeysView, + SortedItemsView, + SortedValuesView, +) + +__all__ = [ + "SortedDict", + "SortedKeysView", + "SortedItemsView", + "SortedValuesView", +] diff --git a/stubs/sortedcontainers/sorteddict.pyi b/stubs/sortedcontainers/sorteddict.pyi new file mode 100644 index 0000000000..68779f968e --- /dev/null +++ b/stubs/sortedcontainers/sorteddict.pyi @@ -0,0 +1,124 @@ +# stub for SortedDict. This is a lightly edited copy of +# https://github.com/grantjenks/python-sortedcontainers/blob/eea42df1f7bad2792e8da77335ff888f04b9e5ae/sortedcontainers/sorteddict.pyi +# (from https://github.com/grantjenks/python-sortedcontainers/pull/107) + +from typing import ( + Any, + Callable, + Dict, + Hashable, + Iterator, + Iterable, + ItemsView, + KeysView, + List, + Mapping, + Optional, + Sequence, + Type, + TypeVar, + Tuple, + Union, + ValuesView, + overload, +) + +_T = TypeVar("_T") +_S = TypeVar("_S") +_T_h = TypeVar("_T_h", bound=Hashable) +_KT = TypeVar("_KT", bound=Hashable) # Key type. +_VT = TypeVar("_VT") # Value type. +_KT_co = TypeVar("_KT_co", covariant=True, bound=Hashable) +_VT_co = TypeVar("_VT_co", covariant=True) +_SD = TypeVar("_SD", bound=SortedDict) +_Key = Callable[[_T], Any] + +class SortedDict(Dict[_KT, _VT]): + @overload + def __init__(self, **kwargs: _VT) -> None: ... + @overload + def __init__(self, __map: Mapping[_KT, _VT], **kwargs: _VT) -> None: ... + @overload + def __init__( + self, __iterable: Iterable[Tuple[_KT, _VT]], **kwargs: _VT + ) -> None: ... + @overload + def __init__(self, __key: _Key[_KT], **kwargs: _VT) -> None: ... + @overload + def __init__( + self, __key: _Key[_KT], __map: Mapping[_KT, _VT], **kwargs: _VT + ) -> None: ... + @overload + def __init__( + self, __key: _Key[_KT], __iterable: Iterable[Tuple[_KT, _VT]], **kwargs: _VT + ) -> None: ... + @property + def key(self) -> Optional[_Key[_KT]]: ... + @property + def iloc(self) -> SortedKeysView[_KT]: ... + def clear(self) -> None: ... + def __delitem__(self, key: _KT) -> None: ... + def __iter__(self) -> Iterator[_KT]: ... + def __reversed__(self) -> Iterator[_KT]: ... + def __setitem__(self, key: _KT, value: _VT) -> None: ... + def _setitem(self, key: _KT, value: _VT) -> None: ... + def copy(self: _SD) -> _SD: ... + def __copy__(self: _SD) -> _SD: ... + @classmethod + @overload + def fromkeys(cls, seq: Iterable[_T_h]) -> SortedDict[_T_h, None]: ... + @classmethod + @overload + def fromkeys(cls, seq: Iterable[_T_h], value: _S) -> SortedDict[_T_h, _S]: ... + def keys(self) -> SortedKeysView[_KT]: ... + def items(self) -> SortedItemsView[_KT, _VT]: ... + def values(self) -> SortedValuesView[_VT]: ... + @overload + def pop(self, key: _KT) -> _VT: ... + @overload + def pop(self, key: _KT, default: _T = ...) -> Union[_VT, _T]: ... + def popitem(self, index: int = ...) -> Tuple[_KT, _VT]: ... + def peekitem(self, index: int = ...) -> Tuple[_KT, _VT]: ... + def setdefault(self, key: _KT, default: Optional[_VT] = ...) -> _VT: ... + @overload + def update(self, __map: Mapping[_KT, _VT], **kwargs: _VT) -> None: ... + @overload + def update(self, __iterable: Iterable[Tuple[_KT, _VT]], **kwargs: _VT) -> None: ... + @overload + def update(self, **kwargs: _VT) -> None: ... + def __reduce__( + self, + ) -> Tuple[ + Type[SortedDict[_KT, _VT]], Tuple[Callable[[_KT], Any], List[Tuple[_KT, _VT]]], + ]: ... + def __repr__(self) -> str: ... + def _check(self) -> None: ... + def islice( + self, start: Optional[int] = ..., stop: Optional[int] = ..., reverse=bool, + ) -> Iterator[_KT]: ... + def bisect_left(self, value: _KT) -> int: ... + def bisect_right(self, value: _KT) -> int: ... + +class SortedKeysView(KeysView[_KT_co], Sequence[_KT_co]): + @overload + def __getitem__(self, index: int) -> _KT_co: ... + @overload + def __getitem__(self, index: slice) -> List[_KT_co]: ... + def __delitem__(self, index: Union[int, slice]) -> None: ... + +class SortedItemsView( # type: ignore + ItemsView[_KT_co, _VT_co], Sequence[Tuple[_KT_co, _VT_co]] +): + def __iter__(self) -> Iterator[Tuple[_KT_co, _VT_co]]: ... + @overload + def __getitem__(self, index: int) -> Tuple[_KT_co, _VT_co]: ... + @overload + def __getitem__(self, index: slice) -> List[Tuple[_KT_co, _VT_co]]: ... + def __delitem__(self, index: Union[int, slice]) -> None: ... + +class SortedValuesView(ValuesView[_VT_co], Sequence[_VT_co]): + @overload + def __getitem__(self, index: int) -> _VT_co: ... + @overload + def __getitem__(self, index: slice) -> List[_VT_co]: ... + def __delitem__(self, index: Union[int, slice]) -> None: ... diff --git a/synapse/util/caches/stream_change_cache.py b/synapse/util/caches/stream_change_cache.py index c61d36a82e..38dc3f501e 100644 --- a/synapse/util/caches/stream_change_cache.py +++ b/synapse/util/caches/stream_change_cache.py @@ -14,6 +14,7 @@ # limitations under the License. import logging +from typing import Dict, Iterable, List, Mapping, Optional, Set from six import integer_types @@ -23,8 +24,11 @@ from synapse.util import caches logger = logging.getLogger(__name__) +# for now, assume all entities in the cache are strings +EntityType = str -class StreamChangeCache(object): + +class StreamChangeCache: """Keeps track of the stream positions of the latest change in a set of entities. Typically the entity will be a room or user id. @@ -34,10 +38,23 @@ class StreamChangeCache(object): old then the cache will simply return all given entities. """ - def __init__(self, name, current_stream_pos, max_size=10000, prefilled_cache=None): + def __init__( + self, + name: str, + current_stream_pos: int, + max_size=10000, + prefilled_cache: Optional[Mapping[EntityType, int]] = None, + ): self._max_size = int(max_size * caches.CACHE_SIZE_FACTOR) - self._entity_to_key = {} - self._cache = SortedDict() + self._entity_to_key = {} # type: Dict[EntityType, int] + + # map from stream id to the a set of entities which changed at that stream id. + self._cache = SortedDict() # type: SortedDict[int, Set[EntityType]] + + # the earliest stream_pos for which we can reliably answer + # get_all_entities_changed. In other words, one less than the earliest + # stream_pos for which we know _cache is valid. + # self._earliest_known_stream_pos = current_stream_pos self.name = name self.metrics = caches.register_cache("cache", self.name, self._cache) @@ -46,7 +63,7 @@ class StreamChangeCache(object): for entity, stream_pos in prefilled_cache.items(): self.entity_has_changed(entity, stream_pos) - def has_entity_changed(self, entity, stream_pos): + def has_entity_changed(self, entity: EntityType, stream_pos: int) -> bool: """Returns True if the entity may have been updated since stream_pos """ assert type(stream_pos) in integer_types @@ -67,22 +84,17 @@ class StreamChangeCache(object): self.metrics.inc_hits() return False - def get_entities_changed(self, entities, stream_pos): + def get_entities_changed( + self, entities: Iterable[EntityType], stream_pos: int + ) -> Set[EntityType]: """ Returns subset of entities that have had new things since the given position. Entities unknown to the cache will be returned. If the position is too old it will just return the given list. """ - assert type(stream_pos) is int - - if stream_pos >= self._earliest_known_stream_pos: - changed_entities = { - self._cache[k] - for k in self._cache.islice(start=self._cache.bisect_right(stream_pos)) - } - - result = changed_entities.intersection(entities) - + changed_entities = self.get_all_entities_changed(stream_pos) + if changed_entities is not None: + result = set(changed_entities).intersection(entities) self.metrics.inc_hits() else: result = set(entities) @@ -90,13 +102,13 @@ class StreamChangeCache(object): return result - def has_any_entity_changed(self, stream_pos): + def has_any_entity_changed(self, stream_pos: int) -> bool: """Returns if any entity has changed """ assert type(stream_pos) is int if not self._cache: - # If we have no cache, nothing can have changed. + # If the cache is empty, nothing can have changed. return False if stream_pos >= self._earliest_known_stream_pos: @@ -106,45 +118,58 @@ class StreamChangeCache(object): self.metrics.inc_misses() return True - def get_all_entities_changed(self, stream_pos): - """Returns all entites that have had new things since the given + def get_all_entities_changed(self, stream_pos: int) -> Optional[List[EntityType]]: + """Returns all entities that have had new things since the given position. If the position is too old it will return None. + + Returns the entities in the order that they were changed. """ assert type(stream_pos) is int - if stream_pos >= self._earliest_known_stream_pos: - return [ - self._cache[k] - for k in self._cache.islice(start=self._cache.bisect_right(stream_pos)) - ] - else: + if stream_pos < self._earliest_known_stream_pos: return None - def entity_has_changed(self, entity, stream_pos): + changed_entities = [] # type: List[EntityType] + + for k in self._cache.islice(start=self._cache.bisect_right(stream_pos)): + changed_entities.extend(self._cache[k]) + return changed_entities + + def entity_has_changed(self, entity: EntityType, stream_pos: int) -> None: """Informs the cache that the entity has been changed at the given position. """ assert type(stream_pos) is int - # FIXME: add a sanity check here that we are not overwriting existing - # data in self._cache - - if stream_pos > self._earliest_known_stream_pos: - old_pos = self._entity_to_key.get(entity, None) - if old_pos is not None: - stream_pos = max(stream_pos, old_pos) - self._cache.pop(old_pos, None) - self._cache[stream_pos] = entity - self._entity_to_key[entity] = stream_pos - - while len(self._cache) > self._max_size: - k, r = self._cache.popitem(0) - self._earliest_known_stream_pos = max( - k, self._earliest_known_stream_pos - ) - self._entity_to_key.pop(r, None) - - def get_max_pos_of_last_change(self, entity): + if stream_pos <= self._earliest_known_stream_pos: + return + + old_pos = self._entity_to_key.get(entity, None) + if old_pos is not None: + if old_pos >= stream_pos: + # nothing to do + return + e = self._cache[old_pos] + e.remove(entity) + if not e: + # cache at this point is now empty + del self._cache[old_pos] + + e1 = self._cache.get(stream_pos) + if e1 is None: + e1 = self._cache[stream_pos] = set() + e1.add(entity) + self._entity_to_key[entity] = stream_pos + + # if the cache is too big, remove entries + while len(self._cache) > self._max_size: + k, r = self._cache.popitem(0) + self._earliest_known_stream_pos = max(k, self._earliest_known_stream_pos) + for entity in r: + del self._entity_to_key[entity] + + def get_max_pos_of_last_change(self, entity: EntityType) -> int: + """Returns an upper bound of the stream id of the last change to an entity. """ diff --git a/tests/util/test_stream_change_cache.py b/tests/util/test_stream_change_cache.py index 72a9de5370..6857933540 100644 --- a/tests/util/test_stream_change_cache.py +++ b/tests/util/test_stream_change_cache.py @@ -28,18 +28,26 @@ class StreamChangeCacheTests(unittest.TestCase): cache.entity_has_changed("user@foo.com", 6) cache.entity_has_changed("bar@baz.net", 7) + # also test multiple things changing on the same stream ID + cache.entity_has_changed("user2@foo.com", 8) + cache.entity_has_changed("bar2@baz.net", 8) + # If it's been changed after that stream position, return True self.assertTrue(cache.has_entity_changed("user@foo.com", 4)) self.assertTrue(cache.has_entity_changed("bar@baz.net", 4)) + self.assertTrue(cache.has_entity_changed("bar2@baz.net", 4)) + self.assertTrue(cache.has_entity_changed("user2@foo.com", 4)) # If it's been changed at that stream position, return False self.assertFalse(cache.has_entity_changed("user@foo.com", 6)) + self.assertFalse(cache.has_entity_changed("user2@foo.com", 8)) # If there's no changes after that stream position, return False self.assertFalse(cache.has_entity_changed("user@foo.com", 7)) + self.assertFalse(cache.has_entity_changed("user2@foo.com", 9)) # If the entity does not exist, return False. - self.assertFalse(cache.has_entity_changed("not@here.website", 7)) + self.assertFalse(cache.has_entity_changed("not@here.website", 9)) # If we request before the stream cache's earliest known position, # return True, whether it's a known entity or not. @@ -47,7 +55,7 @@ class StreamChangeCacheTests(unittest.TestCase): self.assertTrue(cache.has_entity_changed("not@here.website", 0)) @patch("synapse.util.caches.CACHE_SIZE_FACTOR", 1.0) - def test_has_entity_changed_pops_off_start(self): + def test_entity_has_changed_pops_off_start(self): """ StreamChangeCache.entity_has_changed will respect the max size and purge the oldest items upon reaching that max size. @@ -64,11 +72,20 @@ class StreamChangeCacheTests(unittest.TestCase): # The oldest item has been popped off self.assertTrue("user@foo.com" not in cache._entity_to_key) + self.assertEqual( + cache.get_all_entities_changed(2), ["bar@baz.net", "user@elsewhere.org"], + ) + self.assertIsNone(cache.get_all_entities_changed(1)) + # If we update an existing entity, it keeps the two existing entities cache.entity_has_changed("bar@baz.net", 5) self.assertEqual( {"bar@baz.net", "user@elsewhere.org"}, set(cache._entity_to_key) ) + self.assertEqual( + cache.get_all_entities_changed(2), ["user@elsewhere.org", "bar@baz.net"], + ) + self.assertIsNone(cache.get_all_entities_changed(1)) def test_get_all_entities_changed(self): """ @@ -80,18 +97,52 @@ class StreamChangeCacheTests(unittest.TestCase): cache.entity_has_changed("user@foo.com", 2) cache.entity_has_changed("bar@baz.net", 3) + cache.entity_has_changed("anotheruser@foo.com", 3) cache.entity_has_changed("user@elsewhere.org", 4) - self.assertEqual( - cache.get_all_entities_changed(1), - ["user@foo.com", "bar@baz.net", "user@elsewhere.org"], - ) - self.assertEqual( - cache.get_all_entities_changed(2), ["bar@baz.net", "user@elsewhere.org"] - ) + r = cache.get_all_entities_changed(1) + + # either of these are valid + ok1 = [ + "user@foo.com", + "bar@baz.net", + "anotheruser@foo.com", + "user@elsewhere.org", + ] + ok2 = [ + "user@foo.com", + "anotheruser@foo.com", + "bar@baz.net", + "user@elsewhere.org", + ] + self.assertTrue(r == ok1 or r == ok2) + + r = cache.get_all_entities_changed(2) + self.assertTrue(r == ok1[1:] or r == ok2[1:]) + self.assertEqual(cache.get_all_entities_changed(3), ["user@elsewhere.org"]) self.assertEqual(cache.get_all_entities_changed(0), None) + # ... later, things gest more updates + cache.entity_has_changed("user@foo.com", 5) + cache.entity_has_changed("bar@baz.net", 5) + cache.entity_has_changed("anotheruser@foo.com", 6) + + ok1 = [ + "user@elsewhere.org", + "user@foo.com", + "bar@baz.net", + "anotheruser@foo.com", + ] + ok2 = [ + "user@elsewhere.org", + "bar@baz.net", + "user@foo.com", + "anotheruser@foo.com", + ] + r = cache.get_all_entities_changed(3) + self.assertTrue(r == ok1 or r == ok2) + def test_has_any_entity_changed(self): """ StreamChangeCache.has_any_entity_changed will return True if any diff --git a/tox.ini b/tox.ini index 42b2d74891..31011d7436 100644 --- a/tox.ini +++ b/tox.ini @@ -202,7 +202,9 @@ commands = mypy \ synapse/spam_checker_api \ synapse/storage/engines \ synapse/storage/database.py \ - synapse/streams + synapse/streams \ + synapse/util/caches/stream_change_cache.py \ + tests/util/test_stream_change_cache.py # To find all folders that pass mypy you run: # -- cgit 1.4.1 From 82d8b1dd1f9712e6fe29da671b079c646a663a48 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Wed, 22 Apr 2020 14:34:31 +0100 Subject: Another go at fixing one-word commands (#7326) I messed this up last time I tried (#7239 / e13c6c7). --- changelog.d/7326.misc | 1 + synapse/replication/tcp/commands.py | 2 +- tests/replication/tcp/test_commands.py | 42 ++++++++++++++++++++++++++++++++++ 3 files changed, 44 insertions(+), 1 deletion(-) create mode 100644 changelog.d/7326.misc create mode 100644 tests/replication/tcp/test_commands.py (limited to 'tests') diff --git a/changelog.d/7326.misc b/changelog.d/7326.misc new file mode 100644 index 0000000000..676f285377 --- /dev/null +++ b/changelog.d/7326.misc @@ -0,0 +1 @@ +Move catchup of replication streams logic to worker. diff --git a/synapse/replication/tcp/commands.py b/synapse/replication/tcp/commands.py index 5002efe6a0..f26aee83cb 100644 --- a/synapse/replication/tcp/commands.py +++ b/synapse/replication/tcp/commands.py @@ -462,7 +462,7 @@ def parse_command_from_line(line: str) -> Command: Line should already be stripped of whitespace and be checked if blank. """ - idx = line.index(" ") + idx = line.find(" ") if idx >= 0: cmd_name = line[:idx] rest_of_line = line[idx + 1 :] diff --git a/tests/replication/tcp/test_commands.py b/tests/replication/tcp/test_commands.py new file mode 100644 index 0000000000..3cbcb513cc --- /dev/null +++ b/tests/replication/tcp/test_commands.py @@ -0,0 +1,42 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +from synapse.replication.tcp.commands import ( + RdataCommand, + ReplicateCommand, + parse_command_from_line, +) + +from tests.unittest import TestCase + + +class ParseCommandTestCase(TestCase): + def test_parse_one_word_command(self): + line = "REPLICATE" + cmd = parse_command_from_line(line) + self.assertIsInstance(cmd, ReplicateCommand) + + def test_parse_rdata(self): + line = 'RDATA events 6287863 ["ev", ["$eventid", "!roomid", "type", null, null, null]]' + cmd = parse_command_from_line(line) + self.assertIsInstance(cmd, RdataCommand) + self.assertEqual(cmd.stream_name, "events") + self.assertEqual(cmd.token, 6287863) + + def test_parse_rdata_batch(self): + line = 'RDATA presence batch ["@foo:example.com", "online"]' + cmd = parse_command_from_line(line) + self.assertIsInstance(cmd, RdataCommand) + self.assertEqual(cmd.stream_name, "presence") + self.assertIsNone(cmd.token) -- cgit 1.4.1 From 2e3b9a0fcb81b539e155004ded8017ee9923eecc Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Thu, 23 Apr 2020 11:23:53 +0200 Subject: Revert "Revert "Merge pull request #7315 from matrix-org/babolivier/request_token"" This reverts commit 1adf6a55870aa08de272591ff49db9dc49738076. --- changelog.d/7315.feature | 1 + docs/sample_config.yaml | 10 ++++++ synapse/config/server.py | 21 +++++++++++++ synapse/rest/client/v2_alpha/account.py | 17 ++++++++++- synapse/rest/client/v2_alpha/register.py | 12 +++++++- tests/rest/client/v2_alpha/test_account.py | 16 ++++++++++ tests/rest/client/v2_alpha/test_register.py | 47 ++++++++++++++++++++++++++++- 7 files changed, 121 insertions(+), 3 deletions(-) create mode 100644 changelog.d/7315.feature (limited to 'tests') diff --git a/changelog.d/7315.feature b/changelog.d/7315.feature new file mode 100644 index 0000000000..ebcb4741b7 --- /dev/null +++ b/changelog.d/7315.feature @@ -0,0 +1 @@ +Allow `/requestToken` endpoints to hide the existence (or lack thereof) of 3PID associations on the homeserver. diff --git a/docs/sample_config.yaml b/docs/sample_config.yaml index ca8accbc6e..6d5f4f316d 100644 --- a/docs/sample_config.yaml +++ b/docs/sample_config.yaml @@ -414,6 +414,16 @@ retention: # longest_max_lifetime: 1y # interval: 1d +# Inhibits the /requestToken endpoints from returning an error that might leak +# information about whether an e-mail address is in use or not on this +# homeserver. +# Note that for some endpoints the error situation is the e-mail already being +# used, and for others the error is entering the e-mail being unused. +# If this option is enabled, instead of returning an error, these endpoints will +# act as if no error happened and return a fake session ID ('sid') to clients. +# +#request_token_inhibit_3pid_errors: true + ## TLS ## diff --git a/synapse/config/server.py b/synapse/config/server.py index 28e2a031fb..c6d58effd4 100644 --- a/synapse/config/server.py +++ b/synapse/config/server.py @@ -507,6 +507,17 @@ class ServerConfig(Config): self.enable_ephemeral_messages = config.get("enable_ephemeral_messages", False) + # Inhibits the /requestToken endpoints from returning an error that might leak + # information about whether an e-mail address is in use or not on this + # homeserver, and instead return a 200 with a fake sid if this kind of error is + # met, without sending anything. + # This is a compromise between sending an email, which could be a spam vector, + # and letting the client know which email address is bound to an account and + # which one isn't. + self.request_token_inhibit_3pid_errors = config.get( + "request_token_inhibit_3pid_errors", False, + ) + def has_tls_listener(self) -> bool: return any(l["tls"] for l in self.listeners) @@ -972,6 +983,16 @@ class ServerConfig(Config): # - shortest_max_lifetime: 3d # longest_max_lifetime: 1y # interval: 1d + + # Inhibits the /requestToken endpoints from returning an error that might leak + # information about whether an e-mail address is in use or not on this + # homeserver. + # Note that for some endpoints the error situation is the e-mail already being + # used, and for others the error is entering the e-mail being unused. + # If this option is enabled, instead of returning an error, these endpoints will + # act as if no error happened and return a fake session ID ('sid') to clients. + # + #request_token_inhibit_3pid_errors: true """ % locals() ) diff --git a/synapse/rest/client/v2_alpha/account.py b/synapse/rest/client/v2_alpha/account.py index 31435b1e1c..1bd0234779 100644 --- a/synapse/rest/client/v2_alpha/account.py +++ b/synapse/rest/client/v2_alpha/account.py @@ -30,7 +30,7 @@ from synapse.http.servlet import ( ) from synapse.push.mailer import Mailer, load_jinja2_templates from synapse.util.msisdn import phone_number_to_msisdn -from synapse.util.stringutils import assert_valid_client_secret +from synapse.util.stringutils import assert_valid_client_secret, random_string from synapse.util.threepids import check_3pid_allowed from ._base import client_patterns, interactive_auth_handler @@ -100,6 +100,11 @@ class EmailPasswordRequestTokenRestServlet(RestServlet): ) if existing_user_id is None: + if self.config.request_token_inhibit_3pid_errors: + # Make the client think the operation succeeded. See the rationale in the + # comments for request_token_inhibit_3pid_errors. + return 200, {"sid": random_string(16)} + raise SynapseError(400, "Email not found", Codes.THREEPID_NOT_FOUND) if self.config.threepid_behaviour_email == ThreepidBehaviour.REMOTE: @@ -390,6 +395,11 @@ class EmailThreepidRequestTokenRestServlet(RestServlet): ) if existing_user_id is not None: + if self.config.request_token_inhibit_3pid_errors: + # Make the client think the operation succeeded. See the rationale in the + # comments for request_token_inhibit_3pid_errors. + return 200, {"sid": random_string(16)} + raise SynapseError(400, "Email is already in use", Codes.THREEPID_IN_USE) if self.config.threepid_behaviour_email == ThreepidBehaviour.REMOTE: @@ -453,6 +463,11 @@ class MsisdnThreepidRequestTokenRestServlet(RestServlet): existing_user_id = await self.store.get_user_id_by_threepid("msisdn", msisdn) if existing_user_id is not None: + if self.hs.config.request_token_inhibit_3pid_errors: + # Make the client think the operation succeeded. See the rationale in the + # comments for request_token_inhibit_3pid_errors. + return 200, {"sid": random_string(16)} + raise SynapseError(400, "MSISDN is already in use", Codes.THREEPID_IN_USE) if not self.hs.config.account_threepid_delegate_msisdn: diff --git a/synapse/rest/client/v2_alpha/register.py b/synapse/rest/client/v2_alpha/register.py index 431ecf4f84..d1b5c49989 100644 --- a/synapse/rest/client/v2_alpha/register.py +++ b/synapse/rest/client/v2_alpha/register.py @@ -49,7 +49,7 @@ from synapse.http.servlet import ( from synapse.push.mailer import load_jinja2_templates from synapse.util.msisdn import phone_number_to_msisdn from synapse.util.ratelimitutils import FederationRateLimiter -from synapse.util.stringutils import assert_valid_client_secret +from synapse.util.stringutils import assert_valid_client_secret, random_string from synapse.util.threepids import check_3pid_allowed from ._base import client_patterns, interactive_auth_handler @@ -135,6 +135,11 @@ class EmailRegisterRequestTokenRestServlet(RestServlet): ) if existing_user_id is not None: + if self.hs.config.request_token_inhibit_3pid_errors: + # Make the client think the operation succeeded. See the rationale in the + # comments for request_token_inhibit_3pid_errors. + return 200, {"sid": random_string(16)} + raise SynapseError(400, "Email is already in use", Codes.THREEPID_IN_USE) if self.config.threepid_behaviour_email == ThreepidBehaviour.REMOTE: @@ -202,6 +207,11 @@ class MsisdnRegisterRequestTokenRestServlet(RestServlet): ) if existing_user_id is not None: + if self.hs.config.request_token_inhibit_3pid_errors: + # Make the client think the operation succeeded. See the rationale in the + # comments for request_token_inhibit_3pid_errors. + return 200, {"sid": random_string(16)} + raise SynapseError( 400, "Phone number is already in use", Codes.THREEPID_IN_USE ) diff --git a/tests/rest/client/v2_alpha/test_account.py b/tests/rest/client/v2_alpha/test_account.py index 45a9d445f8..0d6936fd36 100644 --- a/tests/rest/client/v2_alpha/test_account.py +++ b/tests/rest/client/v2_alpha/test_account.py @@ -179,6 +179,22 @@ class PasswordResetTestCase(unittest.HomeserverTestCase): # Assert we can't log in with the new password self.attempt_wrong_password_login("kermit", new_password) + @unittest.override_config({"request_token_inhibit_3pid_errors": True}) + def test_password_reset_bad_email_inhibit_error(self): + """Test that triggering a password reset with an email address that isn't bound + to an account doesn't leak the lack of binding for that address if configured + that way. + """ + self.register_user("kermit", "monkey") + self.login("kermit", "monkey") + + email = "test@example.com" + + client_secret = "foobar" + session_id = self._request_token(email, client_secret) + + self.assertIsNotNone(session_id) + def _request_token(self, email, client_secret): request, channel = self.make_request( "POST", diff --git a/tests/rest/client/v2_alpha/test_register.py b/tests/rest/client/v2_alpha/test_register.py index b6ed06e02d..a68a96f618 100644 --- a/tests/rest/client/v2_alpha/test_register.py +++ b/tests/rest/client/v2_alpha/test_register.py @@ -33,7 +33,11 @@ from tests import unittest class RegisterRestServletTestCase(unittest.HomeserverTestCase): - servlets = [register.register_servlets] + servlets = [ + login.register_servlets, + register.register_servlets, + synapse.rest.admin.register_servlets, + ] url = b"/_matrix/client/r0/register" def default_config(self): @@ -260,6 +264,47 @@ class RegisterRestServletTestCase(unittest.HomeserverTestCase): [["m.login.email.identity"]], (f["stages"] for f in flows) ) + @unittest.override_config( + { + "request_token_inhibit_3pid_errors": True, + "public_baseurl": "https://test_server", + "email": { + "smtp_host": "mail_server", + "smtp_port": 2525, + "notif_from": "sender@host", + }, + } + ) + def test_request_token_existing_email_inhibit_error(self): + """Test that requesting a token via this endpoint doesn't leak existing + associations if configured that way. + """ + user_id = self.register_user("kermit", "monkey") + self.login("kermit", "monkey") + + email = "test@example.com" + + # Add a threepid + self.get_success( + self.hs.get_datastore().user_add_threepid( + user_id=user_id, + medium="email", + address=email, + validated_at=0, + added_at=0, + ) + ) + + request, channel = self.make_request( + "POST", + b"register/email/requestToken", + {"client_secret": "foobar", "email": email, "send_attempt": 1}, + ) + self.render(request) + self.assertEquals(200, channel.code, channel.result) + + self.assertIsNotNone(channel.json_body.get("sid")) + class AccountValidityTestCase(unittest.HomeserverTestCase): -- cgit 1.4.1 From fb8ff79efd0897b0b7bf52b0c4bb4061a4ef4018 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Tue, 28 Apr 2020 14:21:48 +0100 Subject: Fix collation for postgres for unit tests (#7359) When running the UTs against a postgres deatbase, we need to set the collation correctly. --- changelog.d/7359.misc | 1 + tests/utils.py | 5 ++++- 2 files changed, 5 insertions(+), 1 deletion(-) create mode 100644 changelog.d/7359.misc (limited to 'tests') diff --git a/changelog.d/7359.misc b/changelog.d/7359.misc new file mode 100644 index 0000000000..b99f257d9a --- /dev/null +++ b/changelog.d/7359.misc @@ -0,0 +1 @@ +Fix collation for postgres for unit tests. diff --git a/tests/utils.py b/tests/utils.py index 2079e0143d..037cb134f0 100644 --- a/tests/utils.py +++ b/tests/utils.py @@ -74,7 +74,10 @@ def setupdb(): db_conn.autocommit = True cur = db_conn.cursor() cur.execute("DROP DATABASE IF EXISTS %s;" % (POSTGRES_BASE_DB,)) - cur.execute("CREATE DATABASE %s;" % (POSTGRES_BASE_DB,)) + cur.execute( + "CREATE DATABASE %s ENCODING 'UTF8' LC_COLLATE='C' LC_CTYPE='C' " + "template=template0;" % (POSTGRES_BASE_DB,) + ) cur.close() db_conn.close() -- cgit 1.4.1 From fce663889b150e55b19097dd9b7fed7aca8abccc Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Tue, 28 Apr 2020 17:42:03 +0100 Subject: Add some replication tests (#7278) Specifically some tests for the typing stream, which means we test streams that fetch missing updates via HTTP (rather than via the DB). We also shuffle things around a bit so that we create two separate `HomeServer` objects, rather than trying to insert a slaved store into places. Note: `test_typing.py` is heavily inspired by `test_receipts.py` --- changelog.d/7278.misc | 1 + tests/replication/tcp/streams/_base.py | 229 +++++++++++++++++++++++++-- tests/replication/tcp/streams/test_typing.py | 80 ++++++++++ 3 files changed, 299 insertions(+), 11 deletions(-) create mode 100644 changelog.d/7278.misc create mode 100644 tests/replication/tcp/streams/test_typing.py (limited to 'tests') diff --git a/changelog.d/7278.misc b/changelog.d/7278.misc new file mode 100644 index 0000000000..8a4c4328f4 --- /dev/null +++ b/changelog.d/7278.misc @@ -0,0 +1 @@ +Add some unit tests for replication. diff --git a/tests/replication/tcp/streams/_base.py b/tests/replication/tcp/streams/_base.py index 32238fe79a..82f15c64e0 100644 --- a/tests/replication/tcp/streams/_base.py +++ b/tests/replication/tcp/streams/_base.py @@ -12,34 +12,67 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. +import logging +from typing import Optional from mock import Mock +import attr + +from twisted.internet.interfaces import IConsumer, IPullProducer, IReactorTime +from twisted.internet.task import LoopingCall +from twisted.web.http import HTTPChannel + +from synapse.app.generic_worker import GenericWorkerServer +from synapse.http.site import SynapseRequest +from synapse.replication.tcp.client import ReplicationDataHandler from synapse.replication.tcp.handler import ReplicationCommandHandler from synapse.replication.tcp.protocol import ClientReplicationStreamProtocol from synapse.replication.tcp.resource import ReplicationStreamProtocolFactory +from synapse.util import Clock from tests import unittest from tests.server import FakeTransport +logger = logging.getLogger(__name__) + class BaseStreamTestCase(unittest.HomeserverTestCase): """Base class for tests of the replication streams""" - def make_homeserver(self, reactor, clock): - self.test_handler = Mock(wraps=TestReplicationDataHandler()) - return self.setup_test_homeserver(replication_data_handler=self.test_handler) - def prepare(self, reactor, clock, hs): # build a replication server server_factory = ReplicationStreamProtocolFactory(hs) self.streamer = hs.get_replication_streamer() self.server = server_factory.buildProtocol(None) - repl_handler = ReplicationCommandHandler(hs) - repl_handler.handler = self.test_handler + # Make a new HomeServer object for the worker + config = self.default_config() + config["worker_app"] = "synapse.app.generic_worker" + config["worker_replication_host"] = "testserv" + config["worker_replication_http_port"] = "8765" + + self.reactor.lookups["testserv"] = "1.2.3.4" + + self.worker_hs = self.setup_test_homeserver( + http_client=None, + homeserverToUse=GenericWorkerServer, + config=config, + reactor=self.reactor, + ) + + # Since we use sqlite in memory databases we need to make sure the + # databases objects are the same. + self.worker_hs.get_datastore().db = hs.get_datastore().db + + self.test_handler = Mock( + wraps=TestReplicationDataHandler(self.worker_hs.get_datastore()) + ) + self.worker_hs.replication_data_handler = self.test_handler + + repl_handler = ReplicationCommandHandler(self.worker_hs) self.client = ClientReplicationStreamProtocol( - hs, "client", "test", clock, repl_handler, + self.worker_hs, "client", "test", clock, repl_handler, ) self._client_transport = None @@ -74,11 +107,75 @@ class BaseStreamTestCase(unittest.HomeserverTestCase): self.streamer.on_notifier_poke() self.pump(0.1) + def handle_http_replication_attempt(self) -> SynapseRequest: + """Asserts that a connection attempt was made to the master HS on the + HTTP replication port, then proxies it to the master HS object to be + handled. + + Returns: + The request object received by master HS. + """ + + # We should have an outbound connection attempt. + clients = self.reactor.tcpClients + self.assertEqual(len(clients), 1) + (host, port, client_factory, _timeout, _bindAddress) = clients.pop(0) + self.assertEqual(host, "1.2.3.4") + self.assertEqual(port, 8765) + + # Set up client side protocol + client_protocol = client_factory.buildProtocol(None) + + request_factory = OneShotRequestFactory() -class TestReplicationDataHandler: + # Set up the server side protocol + channel = _PushHTTPChannel(self.reactor) + channel.requestFactory = request_factory + channel.site = self.site + + # Connect client to server and vice versa. + client_to_server_transport = FakeTransport( + channel, self.reactor, client_protocol + ) + client_protocol.makeConnection(client_to_server_transport) + + server_to_client_transport = FakeTransport( + client_protocol, self.reactor, channel + ) + channel.makeConnection(server_to_client_transport) + + # The request will now be processed by `self.site` and the response + # streamed back. + self.reactor.advance(0) + + # We tear down the connection so it doesn't get reused without our + # knowledge. + server_to_client_transport.loseConnection() + client_to_server_transport.loseConnection() + + return request_factory.request + + def assert_request_is_get_repl_stream_updates( + self, request: SynapseRequest, stream_name: str + ): + """Asserts that the given request is a HTTP replication request for + fetching updates for given stream. + """ + + self.assertRegex( + request.path, + br"^/_synapse/replication/get_repl_stream_updates/%s/[^/]+$" + % (stream_name.encode("ascii"),), + ) + + self.assertEqual(request.method, b"GET") + + +class TestReplicationDataHandler(ReplicationDataHandler): """Drop-in for ReplicationDataHandler which just collects RDATA rows""" - def __init__(self): + def __init__(self, hs): + super().__init__(hs) self.streams = set() self._received_rdata_rows = [] @@ -90,8 +187,118 @@ class TestReplicationDataHandler: return positions async def on_rdata(self, stream_name, token, rows): + await super().on_rdata(stream_name, token, rows) for r in rows: self._received_rdata_rows.append((stream_name, token, r)) - async def on_position(self, stream_name, token): - pass + +@attr.s() +class OneShotRequestFactory: + """A simple request factory that generates a single `SynapseRequest` and + stores it for future use. Can only be used once. + """ + + request = attr.ib(default=None) + + def __call__(self, *args, **kwargs): + assert self.request is None + + self.request = SynapseRequest(*args, **kwargs) + return self.request + + +class _PushHTTPChannel(HTTPChannel): + """A HTTPChannel that wraps pull producers to push producers. + + This is a hack to get around the fact that HTTPChannel transparently wraps a + pull producer (which is what Synapse uses to reply to requests) with + `_PullToPush` to convert it to a push producer. Unfortunately `_PullToPush` + uses the standard reactor rather than letting us use our test reactor, which + makes it very hard to test. + """ + + def __init__(self, reactor: IReactorTime): + super().__init__() + self.reactor = reactor + + self._pull_to_push_producer = None + + def registerProducer(self, producer, streaming): + # Convert pull producers to push producer. + if not streaming: + self._pull_to_push_producer = _PullToPushProducer( + self.reactor, producer, self + ) + producer = self._pull_to_push_producer + + super().registerProducer(producer, True) + + def unregisterProducer(self): + if self._pull_to_push_producer: + # We need to manually stop the _PullToPushProducer. + self._pull_to_push_producer.stop() + + +class _PullToPushProducer: + """A push producer that wraps a pull producer. + """ + + def __init__( + self, reactor: IReactorTime, producer: IPullProducer, consumer: IConsumer + ): + self._clock = Clock(reactor) + self._producer = producer + self._consumer = consumer + + # While running we use a looping call with a zero delay to call + # resumeProducing on given producer. + self._looping_call = None # type: Optional[LoopingCall] + + # We start writing next reactor tick. + self._start_loop() + + def _start_loop(self): + """Start the looping call to + """ + + if not self._looping_call: + # Start a looping call which runs every tick. + self._looping_call = self._clock.looping_call(self._run_once, 0) + + def stop(self): + """Stops calling resumeProducing. + """ + if self._looping_call: + self._looping_call.stop() + self._looping_call = None + + def pauseProducing(self): + """Implements IPushProducer + """ + self.stop() + + def resumeProducing(self): + """Implements IPushProducer + """ + self._start_loop() + + def stopProducing(self): + """Implements IPushProducer + """ + self.stop() + self._producer.stopProducing() + + def _run_once(self): + """Calls resumeProducing on producer once. + """ + + try: + self._producer.resumeProducing() + except Exception: + logger.exception("Failed to call resumeProducing") + try: + self._consumer.unregisterProducer() + except Exception: + pass + + self.stopProducing() diff --git a/tests/replication/tcp/streams/test_typing.py b/tests/replication/tcp/streams/test_typing.py new file mode 100644 index 0000000000..f0ad6402ae --- /dev/null +++ b/tests/replication/tcp/streams/test_typing.py @@ -0,0 +1,80 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +from synapse.handlers.typing import RoomMember +from synapse.replication.http import streams +from synapse.replication.tcp.streams import TypingStream + +from tests.replication.tcp.streams._base import BaseStreamTestCase + +USER_ID = "@feeling:blue" + + +class TypingStreamTestCase(BaseStreamTestCase): + servlets = [ + streams.register_servlets, + ] + + def test_typing(self): + typing = self.hs.get_typing_handler() + + room_id = "!bar:blue" + + self.reconnect() + + # make the client subscribe to the receipts stream + self.test_handler.streams.add("typing") + + typing._push_update(member=RoomMember(room_id, USER_ID), typing=True) + + self.reactor.advance(0) + + # We should now see an attempt to connect to the master + request = self.handle_http_replication_attempt() + self.assert_request_is_get_repl_stream_updates(request, "typing") + + self.test_handler.on_rdata.assert_called_once() + stream_name, token, rdata_rows = self.test_handler.on_rdata.call_args[0] + self.assertEqual(stream_name, "typing") + self.assertEqual(1, len(rdata_rows)) + row = rdata_rows[0] # type: TypingStream.TypingStreamRow + self.assertEqual(room_id, row.room_id) + self.assertEqual([USER_ID], row.user_ids) + + # Now let's disconnect and insert some data. + self.disconnect() + + self.test_handler.on_rdata.reset_mock() + + typing._push_update(member=RoomMember(room_id, USER_ID), typing=False) + + self.test_handler.on_rdata.assert_not_called() + + self.reconnect() + self.pump(0.1) + + # We should now see an attempt to connect to the master + request = self.handle_http_replication_attempt() + self.assert_request_is_get_repl_stream_updates(request, "typing") + + # The from token should be the token from the last RDATA we got. + self.assertEqual(int(request.args[b"from_token"][0]), token) + + self.test_handler.on_rdata.assert_called_once() + stream_name, token, rdata_rows = self.test_handler.on_rdata.call_args[0] + self.assertEqual(stream_name, "typing") + self.assertEqual(1, len(rdata_rows)) + row = rdata_rows[0] # type: TypingStream.TypingStreamRow + self.assertEqual(room_id, row.room_id) + self.assertEqual([], row.user_ids) -- cgit 1.4.1 From 04dd7d182d0601289e0e047243b50803f526ef69 Mon Sep 17 00:00:00 2001 From: Manuel Stahl <37705355+awesome-manuel@users.noreply.github.com> Date: Tue, 28 Apr 2020 19:19:36 +0200 Subject: Return total number of users and profile attributes in admin users endpoint (#6881) Signed-off-by: Manuel Stahl --- changelog.d/6881.misc | 1 + docs/admin_api/user_admin_api.rst | 11 +++-- synapse/rest/admin/users.py | 8 ++-- synapse/storage/data_stores/main/__init__.py | 68 ++++++++++++++++------------ tests/rest/admin/test_user.py | 2 + tests/storage/test_main.py | 46 +++++++++++++++++++ 6 files changed, 100 insertions(+), 36 deletions(-) create mode 100644 changelog.d/6881.misc create mode 100644 tests/storage/test_main.py (limited to 'tests') diff --git a/changelog.d/6881.misc b/changelog.d/6881.misc new file mode 100644 index 0000000000..03b89ccd3d --- /dev/null +++ b/changelog.d/6881.misc @@ -0,0 +1 @@ +Return total number of users and profile attributes in admin users endpoint. Contributed by Awesome Technologies Innovationslabor GmbH. diff --git a/docs/admin_api/user_admin_api.rst b/docs/admin_api/user_admin_api.rst index 9ce10119ff..927ed65f77 100644 --- a/docs/admin_api/user_admin_api.rst +++ b/docs/admin_api/user_admin_api.rst @@ -72,17 +72,22 @@ It returns a JSON body like the following: "is_guest": 0, "admin": 0, "user_type": null, - "deactivated": 0 + "deactivated": 0, + "displayname": , + "avatar_url": null }, { "name": "", "password_hash": "", "is_guest": 0, "admin": 1, "user_type": null, - "deactivated": 0 + "deactivated": 0, + "displayname": , + "avatar_url": "" } ], - "next_token": "100" + "next_token": "100", + "total": 200 } diff --git a/synapse/rest/admin/users.py b/synapse/rest/admin/users.py index 8551ac19b8..593ce011e8 100644 --- a/synapse/rest/admin/users.py +++ b/synapse/rest/admin/users.py @@ -94,10 +94,10 @@ class UsersRestServletV2(RestServlet): guests = parse_boolean(request, "guests", default=True) deactivated = parse_boolean(request, "deactivated", default=False) - users = await self.store.get_users_paginate( + users, total = await self.store.get_users_paginate( start, limit, user_id, guests, deactivated ) - ret = {"users": users} + ret = {"users": users, "total": total} if len(users) >= limit: ret["next_token"] = str(start + len(users)) @@ -199,7 +199,7 @@ class UserRestServletV2(RestServlet): user_id, threepid["medium"], threepid["address"], current_time ) - if "avatar_url" in body: + if "avatar_url" in body and type(body["avatar_url"]) == str: await self.profile_handler.set_avatar_url( target_user, requester, body["avatar_url"], True ) @@ -276,7 +276,7 @@ class UserRestServletV2(RestServlet): user_id, threepid["medium"], threepid["address"], current_time ) - if "avatar_url" in body: + if "avatar_url" in body and type(body["avatar_url"]) == str: await self.profile_handler.set_avatar_url( user_id, requester, body["avatar_url"], True ) diff --git a/synapse/storage/data_stores/main/__init__.py b/synapse/storage/data_stores/main/__init__.py index 649e835303..bd7c3a00ea 100644 --- a/synapse/storage/data_stores/main/__init__.py +++ b/synapse/storage/data_stores/main/__init__.py @@ -503,7 +503,8 @@ class DataStore( self, start, limit, name=None, guests=True, deactivated=False ): """Function to retrieve a paginated list of users from - users list. This will return a json list of users. + users list. This will return a json list of users and the + total number of users matching the filter criteria. Args: start (int): start number to begin the query from @@ -512,35 +513,44 @@ class DataStore( guests (bool): whether to in include guest users deactivated (bool): whether to include deactivated users Returns: - defer.Deferred: resolves to list[dict[str, Any]] + defer.Deferred: resolves to list[dict[str, Any]], int """ - name_filter = {} - if name: - name_filter["name"] = "%" + name + "%" - - attr_filter = {} - if not guests: - attr_filter["is_guest"] = 0 - if not deactivated: - attr_filter["deactivated"] = 0 - - return self.db.simple_select_list_paginate( - desc="get_users_paginate", - table="users", - orderby="name", - start=start, - limit=limit, - filters=name_filter, - keyvalues=attr_filter, - retcols=[ - "name", - "password_hash", - "is_guest", - "admin", - "user_type", - "deactivated", - ], - ) + + def get_users_paginate_txn(txn): + filters = [] + args = [] + + if name: + filters.append("name LIKE ?") + args.append("%" + name + "%") + + if not guests: + filters.append("is_guest = 0") + + if not deactivated: + filters.append("deactivated = 0") + + where_clause = "WHERE " + " AND ".join(filters) if len(filters) > 0 else "" + + sql = "SELECT COUNT(*) as total_users FROM users %s" % (where_clause) + txn.execute(sql, args) + count = txn.fetchone()[0] + + args = [self.hs.config.server_name] + args + [limit, start] + sql = """ + SELECT name, user_type, is_guest, admin, deactivated, displayname, avatar_url + FROM users as u + LEFT JOIN profiles AS p ON u.name = '@' || p.user_id || ':' || ? + {} + ORDER BY u.name LIMIT ? OFFSET ? + """.format( + where_clause + ) + txn.execute(sql, args) + users = self.db.cursor_to_dict(txn) + return users, count + + return self.db.runInteraction("get_users_paginate_txn", get_users_paginate_txn) def search_users(self, term): """Function to search users list for one or more users with diff --git a/tests/rest/admin/test_user.py b/tests/rest/admin/test_user.py index 6416fb5d2a..6c88ab06e2 100644 --- a/tests/rest/admin/test_user.py +++ b/tests/rest/admin/test_user.py @@ -360,6 +360,7 @@ class UsersListTestCase(unittest.HomeserverTestCase): self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) self.assertEqual(3, len(channel.json_body["users"])) + self.assertEqual(3, channel.json_body["total"]) class UserRestTestCase(unittest.HomeserverTestCase): @@ -434,6 +435,7 @@ class UserRestTestCase(unittest.HomeserverTestCase): "admin": True, "displayname": "Bob's name", "threepids": [{"medium": "email", "address": "bob@bob.bob"}], + "avatar_url": None, } ) diff --git a/tests/storage/test_main.py b/tests/storage/test_main.py new file mode 100644 index 0000000000..ab0df5ea93 --- /dev/null +++ b/tests/storage/test_main.py @@ -0,0 +1,46 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 Awesome Technologies Innovationslabor GmbH +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + + +from twisted.internet import defer + +from synapse.types import UserID + +from tests import unittest +from tests.utils import setup_test_homeserver + + +class DataStoreTestCase(unittest.TestCase): + @defer.inlineCallbacks + def setUp(self): + hs = yield setup_test_homeserver(self.addCleanup) + + self.store = hs.get_datastore() + + self.user = UserID.from_string("@abcde:test") + self.displayname = "Frank" + + @defer.inlineCallbacks + def test_get_users_paginate(self): + yield self.store.register_user(self.user.to_string(), "pass") + yield self.store.create_profile(self.user.localpart) + yield self.store.set_profile_displayname(self.user.localpart, self.displayname) + + users, total = yield self.store.get_users_paginate( + 0, 10, name="bc", guests=False + ) + + self.assertEquals(1, total) + self.assertEquals(self.displayname, users.pop()["displayname"]) -- cgit 1.4.1 From c2e1a2110fbe9ead26b4ecbb1afd504ed035a04d Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Wed, 29 Apr 2020 12:30:36 +0100 Subject: Fix limit logic for EventsStream (#7358) * Factor out functions for injecting events into database I want to add some more flexibility to the tools for injecting events into the database, and I don't want to clutter up HomeserverTestCase with them, so let's factor them out to a new file. * Rework TestReplicationDataHandler This wasn't very easy to work with: the mock wrapping was largely superfluous, and it's useful to be able to inspect the received rows, and clear out the received list. * Fix AssertionErrors being thrown by EventsStream Part of the problem was that there was an off-by-one error in the assertion, but also the limit logic was too simple. Fix it all up and add some tests. --- changelog.d/7358.bugfix | 1 + synapse/replication/tcp/handler.py | 4 +- synapse/replication/tcp/streams/events.py | 22 +- synapse/server.pyi | 5 + synapse/storage/data_stores/main/events_worker.py | 64 +++- tests/replication/tcp/streams/_base.py | 41 ++- tests/replication/tcp/streams/test_events.py | 417 ++++++++++++++++++++++ tests/replication/tcp/streams/test_receipts.py | 10 +- tests/replication/tcp/streams/test_typing.py | 11 +- tests/rest/client/v1/utils.py | 2 +- tests/test_utils/__init__.py | 20 ++ tests/test_utils/event_injection.py | 96 +++++ tests/unittest.py | 30 +- tox.ini | 2 + 14 files changed, 658 insertions(+), 67 deletions(-) create mode 100644 changelog.d/7358.bugfix create mode 100644 tests/replication/tcp/streams/test_events.py create mode 100644 tests/test_utils/event_injection.py (limited to 'tests') diff --git a/changelog.d/7358.bugfix b/changelog.d/7358.bugfix new file mode 100644 index 0000000000..f49c600173 --- /dev/null +++ b/changelog.d/7358.bugfix @@ -0,0 +1 @@ +Fix a bug where event updates might not be sent over replication to worker processes after the stream falls behind. diff --git a/synapse/replication/tcp/handler.py b/synapse/replication/tcp/handler.py index 0db5a3a24d..3a8c7c7e2d 100644 --- a/synapse/replication/tcp/handler.py +++ b/synapse/replication/tcp/handler.py @@ -87,7 +87,9 @@ class ReplicationCommandHandler: stream.NAME: stream(hs) for stream in STREAMS_MAP.values() } # type: Dict[str, Stream] - self._position_linearizer = Linearizer("replication_position") + self._position_linearizer = Linearizer( + "replication_position", clock=self._clock + ) # Map of stream to batched updates. See RdataCommand for info on how # batching works. diff --git a/synapse/replication/tcp/streams/events.py b/synapse/replication/tcp/streams/events.py index aa50492569..52df81b1bd 100644 --- a/synapse/replication/tcp/streams/events.py +++ b/synapse/replication/tcp/streams/events.py @@ -170,22 +170,16 @@ class EventsStream(Stream): limited = False upper_limit = current_token - # next up is the state delta table - - state_rows = await self._store.get_all_updated_current_state_deltas( + # next up is the state delta table. + ( + state_rows, + upper_limit, + state_rows_limited, + ) = await self._store.get_all_updated_current_state_deltas( from_token, upper_limit, target_row_count - ) # type: List[Tuple] - - # again, if we've hit the limit there, we'll need to limit the other sources - assert len(state_rows) < target_row_count - if len(state_rows) == target_row_count: - assert state_rows[-1][0] <= upper_limit - upper_limit = state_rows[-1][0] - limited = True + ) - # FIXME: is it a given that there is only one row per stream_id in the - # state_deltas table (so that we can be sure that we have got all of the - # rows for upper_limit)? + limited = limited or state_rows_limited # finally, fetch the ex-outliers rows. We assume there are few enough of these # not to bother with the limit. diff --git a/synapse/server.pyi b/synapse/server.pyi index f1a5717028..fc5886f762 100644 --- a/synapse/server.pyi +++ b/synapse/server.pyi @@ -25,6 +25,7 @@ import synapse.server_notices.server_notices_manager import synapse.server_notices.server_notices_sender import synapse.state import synapse.storage +from synapse.events.builder import EventBuilderFactory class HomeServer(object): @property @@ -121,3 +122,7 @@ class HomeServer(object): pass def get_instance_id(self) -> str: pass + def get_event_builder_factory(self) -> EventBuilderFactory: + pass + def get_storage(self) -> synapse.storage.Storage: + pass diff --git a/synapse/storage/data_stores/main/events_worker.py b/synapse/storage/data_stores/main/events_worker.py index ce8be72bfe..73df6b33ba 100644 --- a/synapse/storage/data_stores/main/events_worker.py +++ b/synapse/storage/data_stores/main/events_worker.py @@ -19,7 +19,7 @@ import itertools import logging import threading from collections import namedtuple -from typing import List, Optional +from typing import List, Optional, Tuple from canonicaljson import json from constantly import NamedConstant, Names @@ -1084,7 +1084,28 @@ class EventsWorkerStore(SQLBaseStore): "get_all_new_backfill_event_rows", get_all_new_backfill_event_rows ) - def get_all_updated_current_state_deltas(self, from_token, to_token, limit): + async def get_all_updated_current_state_deltas( + self, from_token: int, to_token: int, target_row_count: int + ) -> Tuple[List[Tuple], int, bool]: + """Fetch updates from current_state_delta_stream + + Args: + from_token: The previous stream token. Updates from this stream id will + be excluded. + + to_token: The current stream token (ie the upper limit). Updates up to this + stream id will be included (modulo the 'limit' param) + + target_row_count: The number of rows to try to return. If more rows are + available, we will set 'limited' in the result. In the event of a large + batch, we may return more rows than this. + Returns: + A triplet `(updates, new_last_token, limited)`, where: + * `updates` is a list of database tuples. + * `new_last_token` is the new position in stream. + * `limited` is whether there are more updates to fetch. + """ + def get_all_updated_current_state_deltas_txn(txn): sql = """ SELECT stream_id, room_id, type, state_key, event_id @@ -1092,10 +1113,45 @@ class EventsWorkerStore(SQLBaseStore): WHERE ? < stream_id AND stream_id <= ? ORDER BY stream_id ASC LIMIT ? """ - txn.execute(sql, (from_token, to_token, limit)) + txn.execute(sql, (from_token, to_token, target_row_count)) return txn.fetchall() - return self.db.runInteraction( + def get_deltas_for_stream_id_txn(txn, stream_id): + sql = """ + SELECT stream_id, room_id, type, state_key, event_id + FROM current_state_delta_stream + WHERE stream_id = ? + """ + txn.execute(sql, [stream_id]) + return txn.fetchall() + + # we need to make sure that, for every stream id in the results, we get *all* + # the rows with that stream id. + + rows = await self.db.runInteraction( "get_all_updated_current_state_deltas", get_all_updated_current_state_deltas_txn, + ) # type: List[Tuple] + + # if we've got fewer rows than the limit, we're good + if len(rows) < target_row_count: + return rows, to_token, False + + # we hit the limit, so reduce the upper limit so that we exclude the stream id + # of the last row in the result. + assert rows[-1][0] <= to_token + to_token = rows[-1][0] - 1 + + # search backwards through the list for the point to truncate + for idx in range(len(rows) - 1, 0, -1): + if rows[idx - 1][0] <= to_token: + return rows[:idx], to_token, True + + # bother. We didn't get a full set of changes for even a single + # stream id. let's run the query again, without a row limit, but for + # just one stream id. + to_token += 1 + rows = await self.db.runInteraction( + "get_deltas_for_stream_id", get_deltas_for_stream_id_txn, to_token ) + return rows, to_token, True diff --git a/tests/replication/tcp/streams/_base.py b/tests/replication/tcp/streams/_base.py index 82f15c64e0..83e16cfe3d 100644 --- a/tests/replication/tcp/streams/_base.py +++ b/tests/replication/tcp/streams/_base.py @@ -12,10 +12,9 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. -import logging -from typing import Optional -from mock import Mock +import logging +from typing import Any, Dict, List, Optional, Tuple import attr @@ -25,6 +24,7 @@ from twisted.web.http import HTTPChannel from synapse.app.generic_worker import GenericWorkerServer from synapse.http.site import SynapseRequest +from synapse.replication.slave.storage._base import BaseSlavedStore from synapse.replication.tcp.client import ReplicationDataHandler from synapse.replication.tcp.handler import ReplicationCommandHandler from synapse.replication.tcp.protocol import ClientReplicationStreamProtocol @@ -65,9 +65,7 @@ class BaseStreamTestCase(unittest.HomeserverTestCase): # databases objects are the same. self.worker_hs.get_datastore().db = hs.get_datastore().db - self.test_handler = Mock( - wraps=TestReplicationDataHandler(self.worker_hs.get_datastore()) - ) + self.test_handler = self._build_replication_data_handler() self.worker_hs.replication_data_handler = self.test_handler repl_handler = ReplicationCommandHandler(self.worker_hs) @@ -78,6 +76,9 @@ class BaseStreamTestCase(unittest.HomeserverTestCase): self._client_transport = None self._server_transport = None + def _build_replication_data_handler(self): + return TestReplicationDataHandler(self.worker_hs.get_datastore()) + def reconnect(self): if self._client_transport: self.client.close() @@ -174,22 +175,28 @@ class BaseStreamTestCase(unittest.HomeserverTestCase): class TestReplicationDataHandler(ReplicationDataHandler): """Drop-in for ReplicationDataHandler which just collects RDATA rows""" - def __init__(self, hs): - super().__init__(hs) - self.streams = set() - self._received_rdata_rows = [] + def __init__(self, store: BaseSlavedStore): + super().__init__(store) + + # streams to subscribe to: map from stream id to position + self.stream_positions = {} # type: Dict[str, int] + + # list of received (stream_name, token, row) tuples + self.received_rdata_rows = [] # type: List[Tuple[str, int, Any]] def get_streams_to_replicate(self): - positions = {s: 0 for s in self.streams} - for stream, token, _ in self._received_rdata_rows: - if stream in self.streams: - positions[stream] = max(token, positions.get(stream, 0)) - return positions + return self.stream_positions async def on_rdata(self, stream_name, token, rows): await super().on_rdata(stream_name, token, rows) for r in rows: - self._received_rdata_rows.append((stream_name, token, r)) + self.received_rdata_rows.append((stream_name, token, r)) + + if ( + stream_name in self.stream_positions + and token > self.stream_positions[stream_name] + ): + self.stream_positions[stream_name] = token @attr.s() @@ -221,7 +228,7 @@ class _PushHTTPChannel(HTTPChannel): super().__init__() self.reactor = reactor - self._pull_to_push_producer = None + self._pull_to_push_producer = None # type: Optional[_PullToPushProducer] def registerProducer(self, producer, streaming): # Convert pull producers to push producer. diff --git a/tests/replication/tcp/streams/test_events.py b/tests/replication/tcp/streams/test_events.py new file mode 100644 index 0000000000..1fa28084f9 --- /dev/null +++ b/tests/replication/tcp/streams/test_events.py @@ -0,0 +1,417 @@ +# -*- coding: utf-8 -*- +# Copyright 2019 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from typing import List, Optional + +from synapse.api.constants import EventTypes, Membership +from synapse.events import EventBase +from synapse.replication.tcp.streams._base import _STREAM_UPDATE_TARGET_ROW_COUNT +from synapse.replication.tcp.streams.events import ( + EventsStreamCurrentStateRow, + EventsStreamEventRow, + EventsStreamRow, +) +from synapse.rest import admin +from synapse.rest.client.v1 import login, room + +from tests.replication.tcp.streams._base import BaseStreamTestCase +from tests.test_utils.event_injection import inject_event, inject_member_event + + +class EventsStreamTestCase(BaseStreamTestCase): + servlets = [ + admin.register_servlets, + login.register_servlets, + room.register_servlets, + ] + + def prepare(self, reactor, clock, hs): + super().prepare(reactor, clock, hs) + self.user_id = self.register_user("u1", "pass") + self.user_tok = self.login("u1", "pass") + + self.reconnect() + self.test_handler.stream_positions["events"] = 0 + + self.room_id = self.helper.create_room_as(tok=self.user_tok) + self.test_handler.received_rdata_rows.clear() + + def test_update_function_event_row_limit(self): + """Test replication with many non-state events + + Checks that all events are correctly replicated when there are lots of + event rows to be replicated. + """ + # disconnect, so that we can stack up some changes + self.disconnect() + + # generate lots of non-state events. We inject them using inject_event + # so that they are not send out over replication until we call self.replicate(). + events = [ + self._inject_test_event() + for _ in range(_STREAM_UPDATE_TARGET_ROW_COUNT + 1) + ] + + # also one state event + state_event = self._inject_state_event() + + # tell the notifier to catch up to avoid duplicate rows. + # workaround for https://github.com/matrix-org/synapse/issues/7360 + # FIXME remove this when the above is fixed + self.replicate() + + # check we're testing what we think we are: no rows should yet have been + # received + self.assertEqual([], self.test_handler.received_rdata_rows) + + # now reconnect to pull the updates + self.reconnect() + self.replicate() + + # we should have received all the expected rows in the right order + received_rows = self.test_handler.received_rdata_rows + for event in events: + stream_name, token, row = received_rows.pop(0) + self.assertEqual("events", stream_name) + self.assertIsInstance(row, EventsStreamRow) + self.assertEqual(row.type, "ev") + self.assertIsInstance(row.data, EventsStreamEventRow) + self.assertEqual(row.data.event_id, event.event_id) + + stream_name, token, row = received_rows.pop(0) + self.assertIsInstance(row, EventsStreamRow) + self.assertIsInstance(row.data, EventsStreamEventRow) + self.assertEqual(row.data.event_id, state_event.event_id) + + stream_name, token, row = received_rows.pop(0) + self.assertEqual("events", stream_name) + self.assertIsInstance(row, EventsStreamRow) + self.assertEqual(row.type, "state") + self.assertIsInstance(row.data, EventsStreamCurrentStateRow) + self.assertEqual(row.data.event_id, state_event.event_id) + + self.assertEqual([], received_rows) + + def test_update_function_huge_state_change(self): + """Test replication with many state events + + Ensures that all events are correctly replicated when there are lots of + state change rows to be replicated. + """ + + # we want to generate lots of state changes at a single stream ID. + # + # We do this by having two branches in the DAG. On one, we have a moderator + # which that generates lots of state; on the other, we de-op the moderator, + # thus invalidating all the state. + + OTHER_USER = "@other_user:localhost" + + # have the user join + inject_member_event(self.hs, self.room_id, OTHER_USER, Membership.JOIN) + + # Update existing power levels with mod at PL50 + pls = self.helper.get_state( + self.room_id, EventTypes.PowerLevels, tok=self.user_tok + ) + pls["users"][OTHER_USER] = 50 + self.helper.send_state( + self.room_id, EventTypes.PowerLevels, pls, tok=self.user_tok, + ) + + # this is the point in the DAG where we make a fork + fork_point = self.get_success( + self.hs.get_datastore().get_latest_event_ids_in_room(self.room_id) + ) # type: List[str] + + events = [ + self._inject_state_event(sender=OTHER_USER) + for _ in range(_STREAM_UPDATE_TARGET_ROW_COUNT) + ] + + self.replicate() + # all those events and state changes should have landed + self.assertGreaterEqual( + len(self.test_handler.received_rdata_rows), 2 * len(events) + ) + + # disconnect, so that we can stack up the changes + self.disconnect() + self.test_handler.received_rdata_rows.clear() + + # a state event which doesn't get rolled back, to check that the state + # before the huge update comes through ok + state1 = self._inject_state_event() + + # roll back all the state by de-modding the user + prev_events = fork_point + pls["users"][OTHER_USER] = 0 + pl_event = inject_event( + self.hs, + prev_event_ids=prev_events, + type=EventTypes.PowerLevels, + state_key="", + sender=self.user_id, + room_id=self.room_id, + content=pls, + ) + + # one more bit of state that doesn't get rolled back + state2 = self._inject_state_event() + + # tell the notifier to catch up to avoid duplicate rows. + # workaround for https://github.com/matrix-org/synapse/issues/7360 + # FIXME remove this when the above is fixed + self.replicate() + + # check we're testing what we think we are: no rows should yet have been + # received + self.assertEqual([], self.test_handler.received_rdata_rows) + + # now reconnect to pull the updates + self.reconnect() + self.replicate() + + # now we should have received all the expected rows in the right order. + # + # we expect: + # + # - two rows for state1 + # - the PL event row, plus state rows for the PL event and each + # of the states that got reverted. + # - two rows for state2 + + received_rows = self.test_handler.received_rdata_rows + + # first check the first two rows, which should be state1 + + stream_name, token, row = received_rows.pop(0) + self.assertEqual("events", stream_name) + self.assertIsInstance(row, EventsStreamRow) + self.assertEqual(row.type, "ev") + self.assertIsInstance(row.data, EventsStreamEventRow) + self.assertEqual(row.data.event_id, state1.event_id) + + stream_name, token, row = received_rows.pop(0) + self.assertIsInstance(row, EventsStreamRow) + self.assertEqual(row.type, "state") + self.assertIsInstance(row.data, EventsStreamCurrentStateRow) + self.assertEqual(row.data.event_id, state1.event_id) + + # now the last two rows, which should be state2 + stream_name, token, row = received_rows.pop(-2) + self.assertEqual("events", stream_name) + self.assertIsInstance(row, EventsStreamRow) + self.assertEqual(row.type, "ev") + self.assertIsInstance(row.data, EventsStreamEventRow) + self.assertEqual(row.data.event_id, state2.event_id) + + stream_name, token, row = received_rows.pop(-1) + self.assertIsInstance(row, EventsStreamRow) + self.assertEqual(row.type, "state") + self.assertIsInstance(row.data, EventsStreamCurrentStateRow) + self.assertEqual(row.data.event_id, state2.event_id) + + # that should leave us with the rows for the PL event + self.assertEqual(len(received_rows), len(events) + 2) + + stream_name, token, row = received_rows.pop(0) + self.assertEqual("events", stream_name) + self.assertIsInstance(row, EventsStreamRow) + self.assertEqual(row.type, "ev") + self.assertIsInstance(row.data, EventsStreamEventRow) + self.assertEqual(row.data.event_id, pl_event.event_id) + + # the state rows are unsorted + state_rows = [] # type: List[EventsStreamCurrentStateRow] + for stream_name, token, row in received_rows: + self.assertEqual("events", stream_name) + self.assertIsInstance(row, EventsStreamRow) + self.assertEqual(row.type, "state") + self.assertIsInstance(row.data, EventsStreamCurrentStateRow) + state_rows.append(row.data) + + state_rows.sort(key=lambda r: r.state_key) + + sr = state_rows.pop(0) + self.assertEqual(sr.type, EventTypes.PowerLevels) + self.assertEqual(sr.event_id, pl_event.event_id) + for sr in state_rows: + self.assertEqual(sr.type, "test_state_event") + # "None" indicates the state has been deleted + self.assertIsNone(sr.event_id) + + def test_update_function_state_row_limit(self): + """Test replication with many state events over several stream ids. + """ + + # we want to generate lots of state changes, but for this test, we want to + # spread out the state changes over a few stream IDs. + # + # We do this by having two branches in the DAG. On one, we have four moderators, + # each of which that generates lots of state; on the other, we de-op the users, + # thus invalidating all the state. + + NUM_USERS = 4 + STATES_PER_USER = _STREAM_UPDATE_TARGET_ROW_COUNT // 4 + 1 + + user_ids = ["@user%i:localhost" % (i,) for i in range(NUM_USERS)] + + # have the users join + for u in user_ids: + inject_member_event(self.hs, self.room_id, u, Membership.JOIN) + + # Update existing power levels with mod at PL50 + pls = self.helper.get_state( + self.room_id, EventTypes.PowerLevels, tok=self.user_tok + ) + pls["users"].update({u: 50 for u in user_ids}) + self.helper.send_state( + self.room_id, EventTypes.PowerLevels, pls, tok=self.user_tok, + ) + + # this is the point in the DAG where we make a fork + fork_point = self.get_success( + self.hs.get_datastore().get_latest_event_ids_in_room(self.room_id) + ) # type: List[str] + + events = [] # type: List[EventBase] + for user in user_ids: + events.extend( + self._inject_state_event(sender=user) for _ in range(STATES_PER_USER) + ) + + self.replicate() + + # all those events and state changes should have landed + self.assertGreaterEqual( + len(self.test_handler.received_rdata_rows), 2 * len(events) + ) + + # disconnect, so that we can stack up the changes + self.disconnect() + self.test_handler.received_rdata_rows.clear() + + # now roll back all that state by de-modding the users + prev_events = fork_point + pl_events = [] + for u in user_ids: + pls["users"][u] = 0 + e = inject_event( + self.hs, + prev_event_ids=prev_events, + type=EventTypes.PowerLevels, + state_key="", + sender=self.user_id, + room_id=self.room_id, + content=pls, + ) + prev_events = [e.event_id] + pl_events.append(e) + + # tell the notifier to catch up to avoid duplicate rows. + # workaround for https://github.com/matrix-org/synapse/issues/7360 + # FIXME remove this when the above is fixed + self.replicate() + + # check we're testing what we think we are: no rows should yet have been + # received + self.assertEqual([], self.test_handler.received_rdata_rows) + + # now reconnect to pull the updates + self.reconnect() + self.replicate() + + # we should have received all the expected rows in the right order + + received_rows = self.test_handler.received_rdata_rows + self.assertGreaterEqual(len(received_rows), len(events)) + for i in range(NUM_USERS): + # for each user, we expect the PL event row, followed by state rows for + # the PL event and each of the states that got reverted. + stream_name, token, row = received_rows.pop(0) + self.assertEqual("events", stream_name) + self.assertIsInstance(row, EventsStreamRow) + self.assertEqual(row.type, "ev") + self.assertIsInstance(row.data, EventsStreamEventRow) + self.assertEqual(row.data.event_id, pl_events[i].event_id) + + # the state rows are unsorted + state_rows = [] # type: List[EventsStreamCurrentStateRow] + for j in range(STATES_PER_USER + 1): + stream_name, token, row = received_rows.pop(0) + self.assertEqual("events", stream_name) + self.assertIsInstance(row, EventsStreamRow) + self.assertEqual(row.type, "state") + self.assertIsInstance(row.data, EventsStreamCurrentStateRow) + state_rows.append(row.data) + + state_rows.sort(key=lambda r: r.state_key) + + sr = state_rows.pop(0) + self.assertEqual(sr.type, EventTypes.PowerLevels) + self.assertEqual(sr.event_id, pl_events[i].event_id) + for sr in state_rows: + self.assertEqual(sr.type, "test_state_event") + # "None" indicates the state has been deleted + self.assertIsNone(sr.event_id) + + self.assertEqual([], received_rows) + + event_count = 0 + + def _inject_test_event( + self, body: Optional[str] = None, sender: Optional[str] = None, **kwargs + ) -> EventBase: + if sender is None: + sender = self.user_id + + if body is None: + body = "event %i" % (self.event_count,) + self.event_count += 1 + + return inject_event( + self.hs, + room_id=self.room_id, + sender=sender, + type="test_event", + content={"body": body}, + **kwargs + ) + + def _inject_state_event( + self, + body: Optional[str] = None, + state_key: Optional[str] = None, + sender: Optional[str] = None, + ) -> EventBase: + if sender is None: + sender = self.user_id + + if state_key is None: + state_key = "state_%i" % (self.event_count,) + self.event_count += 1 + + if body is None: + body = "state event %s" % (state_key,) + + return inject_event( + self.hs, + room_id=self.room_id, + sender=sender, + type="test_state_event", + state_key=state_key, + content={"body": body}, + ) diff --git a/tests/replication/tcp/streams/test_receipts.py b/tests/replication/tcp/streams/test_receipts.py index a0206f7363..c122b8589c 100644 --- a/tests/replication/tcp/streams/test_receipts.py +++ b/tests/replication/tcp/streams/test_receipts.py @@ -12,6 +12,11 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. + +# type: ignore + +from mock import Mock + from synapse.replication.tcp.streams._base import ReceiptsStream from tests.replication.tcp.streams._base import BaseStreamTestCase @@ -20,11 +25,14 @@ USER_ID = "@feeling:blue" class ReceiptsStreamTestCase(BaseStreamTestCase): + def _build_replication_data_handler(self): + return Mock(wraps=super()._build_replication_data_handler()) + def test_receipt(self): self.reconnect() # make the client subscribe to the receipts stream - self.test_handler.streams.add("receipts") + self.test_handler.stream_positions.update({"receipts": 0}) # tell the master to send a new receipt self.get_success( diff --git a/tests/replication/tcp/streams/test_typing.py b/tests/replication/tcp/streams/test_typing.py index f0ad6402ae..4d354a9db8 100644 --- a/tests/replication/tcp/streams/test_typing.py +++ b/tests/replication/tcp/streams/test_typing.py @@ -12,6 +12,8 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. +from mock import Mock + from synapse.handlers.typing import RoomMember from synapse.replication.http import streams from synapse.replication.tcp.streams import TypingStream @@ -26,6 +28,9 @@ class TypingStreamTestCase(BaseStreamTestCase): streams.register_servlets, ] + def _build_replication_data_handler(self): + return Mock(wraps=super()._build_replication_data_handler()) + def test_typing(self): typing = self.hs.get_typing_handler() @@ -33,8 +38,8 @@ class TypingStreamTestCase(BaseStreamTestCase): self.reconnect() - # make the client subscribe to the receipts stream - self.test_handler.streams.add("typing") + # make the client subscribe to the typing stream + self.test_handler.stream_positions.update({"typing": 0}) typing._push_update(member=RoomMember(room_id, USER_ID), typing=True) @@ -75,6 +80,6 @@ class TypingStreamTestCase(BaseStreamTestCase): stream_name, token, rdata_rows = self.test_handler.on_rdata.call_args[0] self.assertEqual(stream_name, "typing") self.assertEqual(1, len(rdata_rows)) - row = rdata_rows[0] # type: TypingStream.TypingStreamRow + row = rdata_rows[0] self.assertEqual(room_id, row.room_id) self.assertEqual([], row.user_ids) diff --git a/tests/rest/client/v1/utils.py b/tests/rest/client/v1/utils.py index 371637618d..22d734e763 100644 --- a/tests/rest/client/v1/utils.py +++ b/tests/rest/client/v1/utils.py @@ -39,7 +39,7 @@ class RestHelper(object): resource = attr.ib() auth_user_id = attr.ib() - def create_room_as(self, room_creator, is_public=True, tok=None): + def create_room_as(self, room_creator=None, is_public=True, tok=None): temp_id = self.auth_user_id self.auth_user_id = room_creator path = "/_matrix/client/r0/createRoom" diff --git a/tests/test_utils/__init__.py b/tests/test_utils/__init__.py index a7310cf12a..7b345b03bb 100644 --- a/tests/test_utils/__init__.py +++ b/tests/test_utils/__init__.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2019 New Vector Ltd +# Copyright 2020 The Matrix.org Foundation C.I.C # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -16,3 +17,22 @@ """ Utilities for running the unit tests """ +from typing import Awaitable, TypeVar + +TV = TypeVar("TV") + + +def get_awaitable_result(awaitable: Awaitable[TV]) -> TV: + """Get the result from an Awaitable which should have completed + + Asserts that the given awaitable has a result ready, and returns its value + """ + i = awaitable.__await__() + try: + next(i) + except StopIteration as e: + # awaitable returned a result + return e.value + + # if next didn't raise, the awaitable hasn't completed. + raise Exception("awaitable has not yet completed") diff --git a/tests/test_utils/event_injection.py b/tests/test_utils/event_injection.py new file mode 100644 index 0000000000..8f6872761a --- /dev/null +++ b/tests/test_utils/event_injection.py @@ -0,0 +1,96 @@ +# -*- coding: utf-8 -*- +# Copyright 2018 New Vector Ltd +# Copyright 2020 The Matrix.org Foundation C.I.C +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from typing import Optional + +import synapse.server +from synapse.api.constants import EventTypes +from synapse.api.room_versions import KNOWN_ROOM_VERSIONS +from synapse.events import EventBase +from synapse.types import Collection + +from tests.test_utils import get_awaitable_result + + +""" +Utility functions for poking events into the storage of the server under test. +""" + + +def inject_member_event( + hs: synapse.server.HomeServer, + room_id: str, + sender: str, + membership: str, + target: Optional[str] = None, + extra_content: Optional[dict] = None, + **kwargs +) -> EventBase: + """Inject a membership event into a room.""" + if target is None: + target = sender + + content = {"membership": membership} + if extra_content: + content.update(extra_content) + + return inject_event( + hs, + room_id=room_id, + type=EventTypes.Member, + sender=sender, + state_key=target, + content=content, + **kwargs + ) + + +def inject_event( + hs: synapse.server.HomeServer, + room_version: Optional[str] = None, + prev_event_ids: Optional[Collection[str]] = None, + **kwargs +) -> EventBase: + """Inject a generic event into a room + + Args: + hs: the homeserver under test + room_version: the version of the room we're inserting into. + if not specified, will be looked up + prev_event_ids: prev_events for the event. If not specified, will be looked up + kwargs: fields for the event to be created + """ + test_reactor = hs.get_reactor() + + if room_version is None: + d = hs.get_datastore().get_room_version_id(kwargs["room_id"]) + test_reactor.advance(0) + room_version = get_awaitable_result(d) + + builder = hs.get_event_builder_factory().for_room_version( + KNOWN_ROOM_VERSIONS[room_version], kwargs + ) + d = hs.get_event_creation_handler().create_new_client_event( + builder, prev_event_ids=prev_event_ids + ) + test_reactor.advance(0) + event, context = get_awaitable_result(d) + + d = hs.get_storage().persistence.persist_event(event, context) + test_reactor.advance(0) + get_awaitable_result(d) + + return event diff --git a/tests/unittest.py b/tests/unittest.py index 27af5228fe..6b6f224e9c 100644 --- a/tests/unittest.py +++ b/tests/unittest.py @@ -32,7 +32,6 @@ from twisted.python.threadpool import ThreadPool from twisted.trial import unittest from synapse.api.constants import EventTypes, Membership -from synapse.api.room_versions import KNOWN_ROOM_VERSIONS from synapse.config.homeserver import HomeServerConfig from synapse.config.ratelimiting import FederationRateLimitConfig from synapse.federation.transport import server as federation_server @@ -55,6 +54,7 @@ from tests.server import ( render, setup_test_homeserver, ) +from tests.test_utils import event_injection from tests.test_utils.logging_setup import setup_logging from tests.utils import default_config, setupdb @@ -596,36 +596,14 @@ class HomeserverTestCase(TestCase): """ Inject a membership event into a room. + Deprecated: use event_injection.inject_room_member directly + Args: room: Room ID to inject the event into. user: MXID of the user to inject the membership for. membership: The membership type. """ - event_builder_factory = self.hs.get_event_builder_factory() - event_creation_handler = self.hs.get_event_creation_handler() - - room_version = self.get_success( - self.hs.get_datastore().get_room_version_id(room) - ) - - builder = event_builder_factory.for_room_version( - KNOWN_ROOM_VERSIONS[room_version], - { - "type": EventTypes.Member, - "sender": user, - "state_key": user, - "room_id": room, - "content": {"membership": membership}, - }, - ) - - event, context = self.get_success( - event_creation_handler.create_new_client_event(builder) - ) - - self.get_success( - self.hs.get_storage().persistence.persist_event(event, context) - ) + event_injection.inject_member_event(self.hs, room, user, membership) class FederatingHomeserverTestCase(HomeserverTestCase): diff --git a/tox.ini b/tox.ini index 31011d7436..2630857436 100644 --- a/tox.ini +++ b/tox.ini @@ -204,6 +204,8 @@ commands = mypy \ synapse/storage/database.py \ synapse/streams \ synapse/util/caches/stream_change_cache.py \ + tests/replication/tcp/streams \ + tests/test_utils \ tests/util/test_stream_change_cache.py # To find all folders that pass mypy you run: -- cgit 1.4.1 From 3eab76ad43e1de4074550c468d11318d497ff632 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 29 Apr 2020 14:10:59 +0100 Subject: Don't relay REMOTE_SERVER_UP cmds to same conn. (#7352) For direct TCP connections we need the master to relay REMOTE_SERVER_UP commands to the other connections so that all instances get notified about it. The old implementation just relayed to all connections, assuming that sending back to the original sender of the command was safe. This is not true for redis, where commands sent get echoed back to the sender, which was causing master to effectively infinite loop sending and then re-receiving REMOTE_SERVER_UP commands that it sent. The fix is to ensure that we only relay to *other* connections and not to the connection we received the notification from. Fixes #7334. --- changelog.d/7352.feature | 1 + synapse/notifier.py | 9 ---- synapse/replication/tcp/handler.py | 63 ++++++++++++++++++++------ synapse/replication/tcp/protocol.py | 2 +- synapse/replication/tcp/redis.py | 2 +- tests/replication/tcp/test_remote_server_up.py | 62 +++++++++++++++++++++++++ 6 files changed, 114 insertions(+), 25 deletions(-) create mode 100644 changelog.d/7352.feature create mode 100644 tests/replication/tcp/test_remote_server_up.py (limited to 'tests') diff --git a/changelog.d/7352.feature b/changelog.d/7352.feature new file mode 100644 index 0000000000..ce6140fdd1 --- /dev/null +++ b/changelog.d/7352.feature @@ -0,0 +1 @@ +Add support for running replication over Redis when using workers. diff --git a/synapse/notifier.py b/synapse/notifier.py index 6132727cbd..88a5a97caf 100644 --- a/synapse/notifier.py +++ b/synapse/notifier.py @@ -220,12 +220,6 @@ class Notifier(object): """ self.replication_callbacks.append(cb) - def add_remote_server_up_callback(self, cb: Callable[[str], None]): - """Add a callback that will be called when synapse detects a server - has been - """ - self.remote_server_up_callbacks.append(cb) - def on_new_room_event( self, event, room_stream_id, max_room_stream_id, extra_users=[] ): @@ -544,6 +538,3 @@ class Notifier(object): # circular dependencies. if self.federation_sender: self.federation_sender.wake_destination(server) - - for cb in self.remote_server_up_callbacks: - cb(server) diff --git a/synapse/replication/tcp/handler.py b/synapse/replication/tcp/handler.py index 3a8c7c7e2d..b8f49a8d0f 100644 --- a/synapse/replication/tcp/handler.py +++ b/synapse/replication/tcp/handler.py @@ -117,7 +117,6 @@ class ReplicationCommandHandler: self._server_notices_sender = None if self._is_master: self._server_notices_sender = hs.get_server_notices_sender() - self._notifier.add_remote_server_up_callback(self.send_remote_server_up) def start_replication(self, hs): """Helper method to start a replication connection to the remote server @@ -163,7 +162,7 @@ class ReplicationCommandHandler: port = hs.config.worker_replication_port hs.get_reactor().connectTCP(host, port, self._factory) - async def on_REPLICATE(self, cmd: ReplicateCommand): + async def on_REPLICATE(self, conn: AbstractConnection, cmd: ReplicateCommand): # We only want to announce positions by the writer of the streams. # Currently this is just the master process. if not self._is_master: @@ -173,7 +172,7 @@ class ReplicationCommandHandler: current_token = stream.current_token() self.send_command(PositionCommand(stream_name, current_token)) - async def on_USER_SYNC(self, cmd: UserSyncCommand): + async def on_USER_SYNC(self, conn: AbstractConnection, cmd: UserSyncCommand): user_sync_counter.inc() if self._is_master: @@ -181,17 +180,23 @@ class ReplicationCommandHandler: cmd.instance_id, cmd.user_id, cmd.is_syncing, cmd.last_sync_ms ) - async def on_CLEAR_USER_SYNC(self, cmd: ClearUserSyncsCommand): + async def on_CLEAR_USER_SYNC( + self, conn: AbstractConnection, cmd: ClearUserSyncsCommand + ): if self._is_master: await self._presence_handler.update_external_syncs_clear(cmd.instance_id) - async def on_FEDERATION_ACK(self, cmd: FederationAckCommand): + async def on_FEDERATION_ACK( + self, conn: AbstractConnection, cmd: FederationAckCommand + ): federation_ack_counter.inc() if self._federation_sender: self._federation_sender.federation_ack(cmd.token) - async def on_REMOVE_PUSHER(self, cmd: RemovePusherCommand): + async def on_REMOVE_PUSHER( + self, conn: AbstractConnection, cmd: RemovePusherCommand + ): remove_pusher_counter.inc() if self._is_master: @@ -201,7 +206,9 @@ class ReplicationCommandHandler: self._notifier.on_new_replication_data() - async def on_INVALIDATE_CACHE(self, cmd: InvalidateCacheCommand): + async def on_INVALIDATE_CACHE( + self, conn: AbstractConnection, cmd: InvalidateCacheCommand + ): invalidate_cache_counter.inc() if self._is_master: @@ -211,7 +218,7 @@ class ReplicationCommandHandler: cmd.cache_func, tuple(cmd.keys) ) - async def on_USER_IP(self, cmd: UserIpCommand): + async def on_USER_IP(self, conn: AbstractConnection, cmd: UserIpCommand): user_ip_cache_counter.inc() if self._is_master: @@ -227,7 +234,7 @@ class ReplicationCommandHandler: if self._server_notices_sender: await self._server_notices_sender.on_user_ip(cmd.user_id) - async def on_RDATA(self, cmd: RdataCommand): + async def on_RDATA(self, conn: AbstractConnection, cmd: RdataCommand): stream_name = cmd.stream_name inbound_rdata_count.labels(stream_name).inc() @@ -278,7 +285,7 @@ class ReplicationCommandHandler: logger.debug("Received rdata %s -> %s", stream_name, token) await self._replication_data_handler.on_rdata(stream_name, token, rows) - async def on_POSITION(self, cmd: PositionCommand): + async def on_POSITION(self, conn: AbstractConnection, cmd: PositionCommand): stream = self._streams.get(cmd.stream_name) if not stream: logger.error("Got POSITION for unknown stream: %s", cmd.stream_name) @@ -332,12 +339,30 @@ class ReplicationCommandHandler: self._streams_connected.add(cmd.stream_name) - async def on_REMOTE_SERVER_UP(self, cmd: RemoteServerUpCommand): + async def on_REMOTE_SERVER_UP( + self, conn: AbstractConnection, cmd: RemoteServerUpCommand + ): """"Called when get a new REMOTE_SERVER_UP command.""" self._replication_data_handler.on_remote_server_up(cmd.data) - if self._is_master: - self._notifier.notify_remote_server_up(cmd.data) + self._notifier.notify_remote_server_up(cmd.data) + + # We relay to all other connections to ensure every instance gets the + # notification. + # + # When configured to use redis we'll always only have one connection and + # so this is a no-op (all instances will have already received the same + # REMOTE_SERVER_UP command). + # + # For direct TCP connections this will relay to all other connections + # connected to us. When on master this will correctly fan out to all + # other direct TCP clients and on workers there'll only be the one + # connection to master. + # + # (The logic here should also be sound if we have a mix of Redis and + # direct TCP connections so long as there is only one traffic route + # between two instances, but that is not currently supported). + self.send_command(cmd, ignore_conn=conn) def new_connection(self, connection: AbstractConnection): """Called when we have a new connection. @@ -382,11 +407,21 @@ class ReplicationCommandHandler: """ return bool(self._connections) - def send_command(self, cmd: Command): + def send_command( + self, cmd: Command, ignore_conn: Optional[AbstractConnection] = None + ): """Send a command to all connected connections. + + Args: + cmd + ignore_conn: If set don't send command to the given connection. + Used when relaying commands from one connection to all others. """ if self._connections: for connection in self._connections: + if connection == ignore_conn: + continue + try: connection.send_command(cmd) except Exception: diff --git a/synapse/replication/tcp/protocol.py b/synapse/replication/tcp/protocol.py index e3f64eba8f..4198eece71 100644 --- a/synapse/replication/tcp/protocol.py +++ b/synapse/replication/tcp/protocol.py @@ -260,7 +260,7 @@ class BaseReplicationStreamProtocol(LineOnlyReceiver): # Then call out to the handler. cmd_func = getattr(self.command_handler, "on_%s" % (cmd.NAME,), None) if cmd_func: - await cmd_func(cmd) + await cmd_func(self, cmd) handled = True if not handled: diff --git a/synapse/replication/tcp/redis.py b/synapse/replication/tcp/redis.py index 49b3ed0c5e..617e860f95 100644 --- a/synapse/replication/tcp/redis.py +++ b/synapse/replication/tcp/redis.py @@ -112,7 +112,7 @@ class RedisSubscriber(txredisapi.SubscriberProtocol, AbstractConnection): # Then call out to the handler. cmd_func = getattr(self.handler, "on_%s" % (cmd.NAME,), None) if cmd_func: - await cmd_func(cmd) + await cmd_func(self, cmd) handled = True if not handled: diff --git a/tests/replication/tcp/test_remote_server_up.py b/tests/replication/tcp/test_remote_server_up.py new file mode 100644 index 0000000000..d1c15caeb0 --- /dev/null +++ b/tests/replication/tcp/test_remote_server_up.py @@ -0,0 +1,62 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from typing import Tuple + +from twisted.internet.interfaces import IProtocol +from twisted.test.proto_helpers import StringTransport + +from synapse.replication.tcp.resource import ReplicationStreamProtocolFactory + +from tests.unittest import HomeserverTestCase + + +class RemoteServerUpTestCase(HomeserverTestCase): + def prepare(self, reactor, clock, hs): + self.factory = ReplicationStreamProtocolFactory(hs) + + def _make_client(self) -> Tuple[IProtocol, StringTransport]: + """Create a new direct TCP replication connection + """ + + proto = self.factory.buildProtocol(("127.0.0.1", 0)) + transport = StringTransport() + proto.makeConnection(transport) + + # We can safely ignore the commands received during connection. + self.pump() + transport.clear() + + return proto, transport + + def test_relay(self): + """Test that Synapse will relay REMOTE_SERVER_UP commands to all + other connections, but not the one that sent it. + """ + + proto1, transport1 = self._make_client() + + # We shouldn't receive an echo. + proto1.dataReceived(b"REMOTE_SERVER_UP example.com\n") + self.pump() + self.assertEqual(transport1.value(), b"") + + # But we should see an echo if we connect another client + proto2, transport2 = self._make_client() + proto1.dataReceived(b"REMOTE_SERVER_UP example.com\n") + + self.pump() + self.assertEqual(transport1.value(), b"") + self.assertEqual(transport2.value(), b"REMOTE_SERVER_UP example.com\n") -- cgit 1.4.1 From 37f6823f5b91f27b9dd8de8fc0e52d5ea889647c Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 29 Apr 2020 16:23:08 +0100 Subject: Add instance name to RDATA/POSITION commands (#7364) This is primarily for allowing us to send those commands from workers, but for now simply allows us to ignore echoed RDATA/POSITION commands that we sent (we get echoes of sent commands when using redis). Currently we log a WARNING on the master process every time we receive an echoed RDATA. --- changelog.d/7364.misc | 1 + docs/tcp_replication.md | 41 +++++++++++++++++++------------- synapse/app/_base.py | 4 ++-- synapse/logging/opentracing.py | 23 ++++++++---------- synapse/replication/tcp/commands.py | 37 +++++++++++++++++++--------- synapse/replication/tcp/handler.py | 17 ++++++++++--- synapse/server.py | 13 ++++++++-- synapse/server.pyi | 2 ++ tests/replication/slave/storage/_base.py | 1 + tests/replication/tcp/test_commands.py | 6 +++-- 10 files changed, 95 insertions(+), 50 deletions(-) create mode 100644 changelog.d/7364.misc (limited to 'tests') diff --git a/changelog.d/7364.misc b/changelog.d/7364.misc new file mode 100644 index 0000000000..bb5d727cf4 --- /dev/null +++ b/changelog.d/7364.misc @@ -0,0 +1 @@ +Add an `instance_name` to `RDATA` and `POSITION` replication commands. diff --git a/docs/tcp_replication.md b/docs/tcp_replication.md index b922d9cf7e..ab2fffbfe4 100644 --- a/docs/tcp_replication.md +++ b/docs/tcp_replication.md @@ -15,15 +15,17 @@ example flow would be (where '>' indicates master to worker and > SERVER example.com < REPLICATE - > POSITION events 53 - > RDATA events 54 ["$foo1:bar.com", ...] - > RDATA events 55 ["$foo4:bar.com", ...] + > POSITION events master 53 + > RDATA events master 54 ["$foo1:bar.com", ...] + > RDATA events master 55 ["$foo4:bar.com", ...] The example shows the server accepting a new connection and sending its identity with the `SERVER` command, followed by the client server to respond with the position of all streams. The server then periodically sends `RDATA` commands -which have the format `RDATA `, where the format of -`` is defined by the individual streams. +which have the format `RDATA `, where +the format of `` is defined by the individual streams. The +`` is the name of the Synapse process that generated the data +(usually "master"). Error reporting happens by either the client or server sending an ERROR command, and usually the connection will be closed. @@ -52,7 +54,7 @@ The basic structure of the protocol is line based, where the initial word of each line specifies the command. The rest of the line is parsed based on the command. For example, the RDATA command is defined as: - RDATA + RDATA (Note that may contains spaces, but cannot contain newlines.) @@ -136,11 +138,11 @@ the wire: < NAME synapse.app.appservice < PING 1490197665618 < REPLICATE - > POSITION events 1 - > POSITION backfill 1 - > POSITION caches 1 - > RDATA caches 2 ["get_user_by_id",["@01register-user:localhost:8823"],1490197670513] - > RDATA events 14 ["$149019767112vOHxz:localhost:8823", + > POSITION events master 1 + > POSITION backfill master 1 + > POSITION caches master 1 + > RDATA caches master 2 ["get_user_by_id",["@01register-user:localhost:8823"],1490197670513] + > RDATA events master 14 ["$149019767112vOHxz:localhost:8823", "!AFDCvgApUmpdfVjIXm:localhost:8823","m.room.guest_access","",null] < PING 1490197675618 > ERROR server stopping @@ -151,10 +153,10 @@ position without needing to send data with the `RDATA` command. An example of a batched set of `RDATA` is: - > RDATA caches batch ["get_user_by_id",["@test:localhost:8823"],1490197670513] - > RDATA caches batch ["get_user_by_id",["@test2:localhost:8823"],1490197670513] - > RDATA caches batch ["get_user_by_id",["@test3:localhost:8823"],1490197670513] - > RDATA caches 54 ["get_user_by_id",["@test4:localhost:8823"],1490197670513] + > RDATA caches master batch ["get_user_by_id",["@test:localhost:8823"],1490197670513] + > RDATA caches master batch ["get_user_by_id",["@test2:localhost:8823"],1490197670513] + > RDATA caches master batch ["get_user_by_id",["@test3:localhost:8823"],1490197670513] + > RDATA caches master 54 ["get_user_by_id",["@test4:localhost:8823"],1490197670513] In this case the client shouldn't advance their caches token until it sees the the last `RDATA`. @@ -178,6 +180,11 @@ client (C): updates, and if so then fetch them out of band. Sent in response to a REPLICATE command (but can happen at any time). + The POSITION command includes the source of the stream. Currently all streams + are written by a single process (usually "master"). If fetching missing + updates via HTTP API, rather than via the DB, then processes should make the + request to the appropriate process. + #### ERROR (S, C) There was an error @@ -234,12 +241,12 @@ Each individual cache invalidation results in a row being sent down replication, which includes the cache name (the name of the function) and they key to invalidate. For example: - > RDATA caches 550953771 ["get_user_by_id", ["@bob:example.com"], 1550574873251] + > RDATA caches master 550953771 ["get_user_by_id", ["@bob:example.com"], 1550574873251] Alternatively, an entire cache can be invalidated by sending down a `null` instead of the key. For example: - > RDATA caches 550953772 ["get_user_by_id", null, 1550574873252] + > RDATA caches master 550953772 ["get_user_by_id", null, 1550574873252] However, there are times when a number of caches need to be invalidated at the same time with the same key. To reduce traffic we batch those diff --git a/synapse/app/_base.py b/synapse/app/_base.py index 4d84f4595a..628292b890 100644 --- a/synapse/app/_base.py +++ b/synapse/app/_base.py @@ -270,7 +270,7 @@ def start(hs, listeners=None): # Start the tracer synapse.logging.opentracing.init_tracer( # type: ignore[attr-defined] # noqa - hs.config + hs ) # It is now safe to start your Synapse. @@ -316,7 +316,7 @@ def setup_sentry(hs): scope.set_tag("matrix_server_name", hs.config.server_name) app = hs.config.worker_app if hs.config.worker_app else "synapse.app.homeserver" - name = hs.config.worker_name if hs.config.worker_name else "master" + name = hs.get_instance_name() scope.set_tag("worker_app", app) scope.set_tag("worker_name", name) diff --git a/synapse/logging/opentracing.py b/synapse/logging/opentracing.py index 0638cec429..5dddf57008 100644 --- a/synapse/logging/opentracing.py +++ b/synapse/logging/opentracing.py @@ -171,7 +171,7 @@ import logging import re import types from functools import wraps -from typing import Dict +from typing import TYPE_CHECKING, Dict from canonicaljson import json @@ -179,6 +179,9 @@ from twisted.internet import defer from synapse.config import ConfigError +if TYPE_CHECKING: + from synapse.server import HomeServer + # Helper class @@ -297,14 +300,11 @@ def _noop_context_manager(*args, **kwargs): # Setup -def init_tracer(config): +def init_tracer(hs: "HomeServer"): """Set the whitelists and initialise the JaegerClient tracer - - Args: - config (HomeserverConfig): The config used by the homeserver """ global opentracing - if not config.opentracer_enabled: + if not hs.config.opentracer_enabled: # We don't have a tracer opentracing = None return @@ -315,18 +315,15 @@ def init_tracer(config): "installed." ) - # Include the worker name - name = config.worker_name if config.worker_name else "master" - # Pull out the jaeger config if it was given. Otherwise set it to something sensible. # See https://github.com/jaegertracing/jaeger-client-python/blob/master/jaeger_client/config.py - set_homeserver_whitelist(config.opentracer_whitelist) + set_homeserver_whitelist(hs.config.opentracer_whitelist) JaegerConfig( - config=config.jaeger_config, - service_name="{} {}".format(config.server_name, name), - scope_manager=LogContextScopeManager(config), + config=hs.config.jaeger_config, + service_name="{} {}".format(hs.config.server_name, hs.get_instance_name()), + scope_manager=LogContextScopeManager(hs.config), ).initialize_tracer() diff --git a/synapse/replication/tcp/commands.py b/synapse/replication/tcp/commands.py index c7880d4b63..f58e384d17 100644 --- a/synapse/replication/tcp/commands.py +++ b/synapse/replication/tcp/commands.py @@ -95,7 +95,7 @@ class RdataCommand(Command): Format:: - RDATA + RDATA The `` may either be a numeric stream id OR "batch". The latter case is used to support sending multiple updates with the same stream ID. This @@ -105,33 +105,40 @@ class RdataCommand(Command): The client should batch all incoming RDATA with a token of "batch" (per stream_name) until it sees an RDATA with a numeric stream ID. + The `` is the source of the new data (usually "master"). + `` of "batch" maps to the instance variable `token` being None. An example of a batched series of RDATA:: - RDATA presence batch ["@foo:example.com", "online", ...] - RDATA presence batch ["@bar:example.com", "online", ...] - RDATA presence 59 ["@baz:example.com", "online", ...] + RDATA presence master batch ["@foo:example.com", "online", ...] + RDATA presence master batch ["@bar:example.com", "online", ...] + RDATA presence master 59 ["@baz:example.com", "online", ...] """ NAME = "RDATA" - def __init__(self, stream_name, token, row): + def __init__(self, stream_name, instance_name, token, row): self.stream_name = stream_name + self.instance_name = instance_name self.token = token self.row = row @classmethod def from_line(cls, line): - stream_name, token, row_json = line.split(" ", 2) + stream_name, instance_name, token, row_json = line.split(" ", 3) return cls( - stream_name, None if token == "batch" else int(token), json.loads(row_json) + stream_name, + instance_name, + None if token == "batch" else int(token), + json.loads(row_json), ) def to_line(self): return " ".join( ( self.stream_name, + self.instance_name, str(self.token) if self.token is not None else "batch", _json_encoder.encode(self.row), ) @@ -145,23 +152,31 @@ class PositionCommand(Command): """Sent by the server to tell the client the stream postition without needing to send an RDATA. + Format:: + + POSITION + On receipt of a POSITION command clients should check if they have missed any updates, and if so then fetch them out of band. + + The `` is the process that sent the command and is the source + of the stream. """ NAME = "POSITION" - def __init__(self, stream_name, token): + def __init__(self, stream_name, instance_name, token): self.stream_name = stream_name + self.instance_name = instance_name self.token = token @classmethod def from_line(cls, line): - stream_name, token = line.split(" ", 1) - return cls(stream_name, int(token)) + stream_name, instance_name, token = line.split(" ", 2) + return cls(stream_name, instance_name, int(token)) def to_line(self): - return " ".join((self.stream_name, str(self.token))) + return " ".join((self.stream_name, self.instance_name, str(self.token))) class ErrorCommand(_SimpleCommand): diff --git a/synapse/replication/tcp/handler.py b/synapse/replication/tcp/handler.py index b8f49a8d0f..6f7054d5af 100644 --- a/synapse/replication/tcp/handler.py +++ b/synapse/replication/tcp/handler.py @@ -79,6 +79,7 @@ class ReplicationCommandHandler: self._notifier = hs.get_notifier() self._clock = hs.get_clock() self._instance_id = hs.get_instance_id() + self._instance_name = hs.get_instance_name() # Set of streams that we've caught up with. self._streams_connected = set() # type: Set[str] @@ -156,7 +157,7 @@ class ReplicationCommandHandler: hs.config.redis.redis_host, hs.config.redis.redis_port, self._factory, ) else: - client_name = hs.config.worker_name + client_name = hs.get_instance_name() self._factory = DirectTcpReplicationClientFactory(hs, client_name, self) host = hs.config.worker_replication_host port = hs.config.worker_replication_port @@ -170,7 +171,9 @@ class ReplicationCommandHandler: for stream_name, stream in self._streams.items(): current_token = stream.current_token() - self.send_command(PositionCommand(stream_name, current_token)) + self.send_command( + PositionCommand(stream_name, self._instance_name, current_token) + ) async def on_USER_SYNC(self, conn: AbstractConnection, cmd: UserSyncCommand): user_sync_counter.inc() @@ -235,6 +238,10 @@ class ReplicationCommandHandler: await self._server_notices_sender.on_user_ip(cmd.user_id) async def on_RDATA(self, conn: AbstractConnection, cmd: RdataCommand): + if cmd.instance_name == self._instance_name: + # Ignore RDATA that are just our own echoes + return + stream_name = cmd.stream_name inbound_rdata_count.labels(stream_name).inc() @@ -286,6 +293,10 @@ class ReplicationCommandHandler: await self._replication_data_handler.on_rdata(stream_name, token, rows) async def on_POSITION(self, conn: AbstractConnection, cmd: PositionCommand): + if cmd.instance_name == self._instance_name: + # Ignore POSITION that are just our own echoes + return + stream = self._streams.get(cmd.stream_name) if not stream: logger.error("Got POSITION for unknown stream: %s", cmd.stream_name) @@ -485,7 +496,7 @@ class ReplicationCommandHandler: We need to check if the client is interested in the stream or not """ - self.send_command(RdataCommand(stream_name, token, data)) + self.send_command(RdataCommand(stream_name, self._instance_name, token, data)) UpdateToken = TypeVar("UpdateToken") diff --git a/synapse/server.py b/synapse/server.py index 9d273c980c..bf97a16c09 100644 --- a/synapse/server.py +++ b/synapse/server.py @@ -234,7 +234,8 @@ class HomeServer(object): self._listening_services = [] self.start_time = None - self.instance_id = random_string(5) + self._instance_id = random_string(5) + self._instance_name = config.worker_name or "master" self.clock = Clock(reactor) self.distributor = Distributor() @@ -254,7 +255,15 @@ class HomeServer(object): This is used to distinguish running instances in worker-based deployments. """ - return self.instance_id + return self._instance_id + + def get_instance_name(self) -> str: + """A unique name for this synapse process. + + Used to identify the process over replication and in config. Does not + change over restarts. + """ + return self._instance_name def setup(self): logger.info("Setting up.") diff --git a/synapse/server.pyi b/synapse/server.pyi index fc5886f762..18043a2593 100644 --- a/synapse/server.pyi +++ b/synapse/server.pyi @@ -122,6 +122,8 @@ class HomeServer(object): pass def get_instance_id(self) -> str: pass + def get_instance_name(self) -> str: + pass def get_event_builder_factory(self) -> EventBuilderFactory: pass def get_storage(self) -> synapse.storage.Storage: diff --git a/tests/replication/slave/storage/_base.py b/tests/replication/slave/storage/_base.py index 395c7d0306..1615dfab5e 100644 --- a/tests/replication/slave/storage/_base.py +++ b/tests/replication/slave/storage/_base.py @@ -57,6 +57,7 @@ class BaseSlavedStoreTestCase(unittest.HomeserverTestCase): # We now do some gut wrenching so that we have a client that is based # off of the slave store rather than the main store. self.replication_handler = ReplicationCommandHandler(self.hs) + self.replication_handler._instance_name = "worker" self.replication_handler._replication_data_handler = ReplicationDataHandler( self.slaved_store ) diff --git a/tests/replication/tcp/test_commands.py b/tests/replication/tcp/test_commands.py index 3cbcb513cc..7ddfd0a733 100644 --- a/tests/replication/tcp/test_commands.py +++ b/tests/replication/tcp/test_commands.py @@ -28,15 +28,17 @@ class ParseCommandTestCase(TestCase): self.assertIsInstance(cmd, ReplicateCommand) def test_parse_rdata(self): - line = 'RDATA events 6287863 ["ev", ["$eventid", "!roomid", "type", null, null, null]]' + line = 'RDATA events master 6287863 ["ev", ["$eventid", "!roomid", "type", null, null, null]]' cmd = parse_command_from_line(line) self.assertIsInstance(cmd, RdataCommand) self.assertEqual(cmd.stream_name, "events") + self.assertEqual(cmd.instance_name, "master") self.assertEqual(cmd.token, 6287863) def test_parse_rdata_batch(self): - line = 'RDATA presence batch ["@foo:example.com", "online"]' + line = 'RDATA presence master batch ["@foo:example.com", "online"]' cmd = parse_command_from_line(line) self.assertIsInstance(cmd, RdataCommand) self.assertEqual(cmd.stream_name, "presence") + self.assertEqual(cmd.instance_name, "master") self.assertIsNone(cmd.token) -- cgit 1.4.1 From 627b0f5f2753e6910adb7a877541d50f5936b8a5 Mon Sep 17 00:00:00 2001 From: Patrick Cloke Date: Thu, 30 Apr 2020 13:47:49 -0400 Subject: Persist user interactive authentication sessions (#7302) By persisting the user interactive authentication sessions to the database, this fixes situations where a user hits different works throughout their auth session and also allows sessions to persist through restarts of Synapse. --- changelog.d/7302.bugfix | 1 + synapse/app/generic_worker.py | 2 + synapse/handlers/auth.py | 175 +++++-------- synapse/handlers/cas_handler.py | 2 +- synapse/handlers/saml_handler.py | 2 +- synapse/rest/client/v2_alpha/auth.py | 4 +- synapse/rest/client/v2_alpha/register.py | 4 +- synapse/storage/data_stores/main/__init__.py | 2 + .../main/schema/delta/58/03persist_ui_auth.sql | 36 +++ synapse/storage/data_stores/main/ui_auth.py | 279 +++++++++++++++++++++ synapse/storage/engines/sqlite.py | 1 + tests/rest/client/v2_alpha/test_auth.py | 40 +++ tests/utils.py | 8 +- tox.ini | 3 +- 14 files changed, 434 insertions(+), 125 deletions(-) create mode 100644 changelog.d/7302.bugfix create mode 100644 synapse/storage/data_stores/main/schema/delta/58/03persist_ui_auth.sql create mode 100644 synapse/storage/data_stores/main/ui_auth.py (limited to 'tests') diff --git a/changelog.d/7302.bugfix b/changelog.d/7302.bugfix new file mode 100644 index 0000000000..820646d1f9 --- /dev/null +++ b/changelog.d/7302.bugfix @@ -0,0 +1 @@ +Persist user interactive authentication sessions across workers and Synapse restarts. diff --git a/synapse/app/generic_worker.py b/synapse/app/generic_worker.py index d125327f08..0ace7b787d 100644 --- a/synapse/app/generic_worker.py +++ b/synapse/app/generic_worker.py @@ -127,6 +127,7 @@ from synapse.storage.data_stores.main.monthly_active_users import ( MonthlyActiveUsersWorkerStore, ) from synapse.storage.data_stores.main.presence import UserPresenceState +from synapse.storage.data_stores.main.ui_auth import UIAuthWorkerStore from synapse.storage.data_stores.main.user_directory import UserDirectoryStore from synapse.types import ReadReceipt from synapse.util.async_helpers import Linearizer @@ -439,6 +440,7 @@ class GenericWorkerSlavedStore( # FIXME(#3714): We need to add UserDirectoryStore as we write directly # rather than going via the correct worker. UserDirectoryStore, + UIAuthWorkerStore, SlavedDeviceInboxStore, SlavedDeviceStore, SlavedReceiptsStore, diff --git a/synapse/handlers/auth.py b/synapse/handlers/auth.py index dbe165ce1e..7613e5b6ab 100644 --- a/synapse/handlers/auth.py +++ b/synapse/handlers/auth.py @@ -41,10 +41,10 @@ from synapse.handlers.ui_auth.checkers import UserInteractiveAuthChecker from synapse.http.server import finish_request from synapse.http.site import SynapseRequest from synapse.logging.context import defer_to_thread +from synapse.metrics.background_process_metrics import run_as_background_process from synapse.module_api import ModuleApi from synapse.push.mailer import load_jinja2_templates from synapse.types import Requester, UserID -from synapse.util.caches.expiringcache import ExpiringCache from ._base import BaseHandler @@ -69,15 +69,6 @@ class AuthHandler(BaseHandler): self.bcrypt_rounds = hs.config.bcrypt_rounds - # This is not a cache per se, but a store of all current sessions that - # expire after N hours - self.sessions = ExpiringCache( - cache_name="register_sessions", - clock=hs.get_clock(), - expiry_ms=self.SESSION_EXPIRE_MS, - reset_expiry_on_get=True, - ) - account_handler = ModuleApi(hs, self) self.password_providers = [ module(config=config, account_handler=account_handler) @@ -119,6 +110,15 @@ class AuthHandler(BaseHandler): self._clock = self.hs.get_clock() + # Expire old UI auth sessions after a period of time. + if hs.config.worker_app is None: + self._clock.looping_call( + run_as_background_process, + 5 * 60 * 1000, + "expire_old_sessions", + self._expire_old_sessions, + ) + # Load the SSO HTML templates. # The following template is shown to the user during a client login via SSO, @@ -301,16 +301,21 @@ class AuthHandler(BaseHandler): if "session" in authdict: sid = authdict["session"] + # Convert the URI and method to strings. + uri = request.uri.decode("utf-8") + method = request.uri.decode("utf-8") + # If there's no session ID, create a new session. if not sid: - session = self._create_session( - clientdict, (request.uri, request.method, clientdict), description + session = await self.store.create_ui_auth_session( + clientdict, uri, method, description ) - session_id = session["id"] else: - session = self._get_session_info(sid) - session_id = sid + try: + session = await self.store.get_ui_auth_session(sid) + except StoreError: + raise SynapseError(400, "Unknown session ID: %s" % (sid,)) if not clientdict: # This was designed to allow the client to omit the parameters @@ -322,15 +327,15 @@ class AuthHandler(BaseHandler): # on a homeserver. # Revisit: Assuming the REST APIs do sensible validation, the data # isn't arbitrary. - clientdict = session["clientdict"] + clientdict = session.clientdict # Ensure that the queried operation does not vary between stages of # the UI authentication session. This is done by generating a stable # comparator based on the URI, method, and body (minus the auth dict) # and storing it during the initial query. Subsequent queries ensure # that this comparator has not changed. - comparator = (request.uri, request.method, clientdict) - if session["ui_auth"] != comparator: + comparator = (uri, method, clientdict) + if (session.uri, session.method, session.clientdict) != comparator: raise SynapseError( 403, "Requested operation has changed during the UI authentication session.", @@ -338,11 +343,9 @@ class AuthHandler(BaseHandler): if not authdict: raise InteractiveAuthIncompleteError( - self._auth_dict_for_flows(flows, session_id) + self._auth_dict_for_flows(flows, session.session_id) ) - creds = session["creds"] - # check auth type currently being presented errordict = {} # type: Dict[str, Any] if "type" in authdict: @@ -350,8 +353,9 @@ class AuthHandler(BaseHandler): try: result = await self._check_auth_dict(authdict, clientip) if result: - creds[login_type] = result - self._save_session(session) + await self.store.mark_ui_auth_stage_complete( + session.session_id, login_type, result + ) except LoginError as e: if login_type == LoginType.EMAIL_IDENTITY: # riot used to have a bug where it would request a new @@ -367,6 +371,7 @@ class AuthHandler(BaseHandler): # so that the client can have another go. errordict = e.error_dict() + creds = await self.store.get_completed_ui_auth_stages(session.session_id) for f in flows: if len(set(f) - set(creds)) == 0: # it's very useful to know what args are stored, but this can @@ -380,9 +385,9 @@ class AuthHandler(BaseHandler): list(clientdict), ) - return creds, clientdict, session_id + return creds, clientdict, session.session_id - ret = self._auth_dict_for_flows(flows, session_id) + ret = self._auth_dict_for_flows(flows, session.session_id) ret["completed"] = list(creds) ret.update(errordict) raise InteractiveAuthIncompleteError(ret) @@ -399,13 +404,11 @@ class AuthHandler(BaseHandler): if "session" not in authdict: raise LoginError(400, "", Codes.MISSING_PARAM) - sess = self._get_session_info(authdict["session"]) - creds = sess["creds"] - result = await self.checkers[stagetype].check_auth(authdict, clientip) if result: - creds[stagetype] = result - self._save_session(sess) + await self.store.mark_ui_auth_stage_complete( + authdict["session"], stagetype, result + ) return True return False @@ -427,7 +430,7 @@ class AuthHandler(BaseHandler): sid = authdict["session"] return sid - def set_session_data(self, session_id: str, key: str, value: Any) -> None: + async def set_session_data(self, session_id: str, key: str, value: Any) -> None: """ Store a key-value pair into the sessions data associated with this request. This data is stored server-side and cannot be modified by @@ -438,11 +441,12 @@ class AuthHandler(BaseHandler): key: The key to store the data under value: The data to store """ - sess = self._get_session_info(session_id) - sess["serverdict"][key] = value - self._save_session(sess) + try: + await self.store.set_ui_auth_session_data(session_id, key, value) + except StoreError: + raise SynapseError(400, "Unknown session ID: %s" % (session_id,)) - def get_session_data( + async def get_session_data( self, session_id: str, key: str, default: Optional[Any] = None ) -> Any: """ @@ -453,8 +457,18 @@ class AuthHandler(BaseHandler): key: The key to store the data under default: Value to return if the key has not been set """ - sess = self._get_session_info(session_id) - return sess["serverdict"].get(key, default) + try: + return await self.store.get_ui_auth_session_data(session_id, key, default) + except StoreError: + raise SynapseError(400, "Unknown session ID: %s" % (session_id,)) + + async def _expire_old_sessions(self): + """ + Invalidate any user interactive authentication sessions that have expired. + """ + now = self._clock.time_msec() + expiration_time = now - self.SESSION_EXPIRE_MS + await self.store.delete_old_ui_auth_sessions(expiration_time) async def _check_auth_dict( self, authdict: Dict[str, Any], clientip: str @@ -534,67 +548,6 @@ class AuthHandler(BaseHandler): "params": params, } - def _create_session( - self, - clientdict: Dict[str, Any], - ui_auth: Tuple[bytes, bytes, Dict[str, Any]], - description: str, - ) -> dict: - """ - Creates a new user interactive authentication session. - - The session can be used to track data across multiple requests, e.g. for - interactive authentication. - - Each session has the following keys: - - id: - A unique identifier for this session. Passed back to the client - and returned for each stage. - clientdict: - The dictionary from the client root level, not the 'auth' key. - ui_auth: - A tuple which is checked at each stage of the authentication to - ensure that the asked for operation has not changed. - creds: - A map, which maps each auth-type (str) to the relevant identity - authenticated by that auth-type (mostly str, but for captcha, bool). - serverdict: - A map of data that is stored server-side and cannot be modified - by the client. - description: - A string description of the operation that the current - authentication is authorising. - Returns: - The newly created session. - """ - session_id = None - while session_id is None or session_id in self.sessions: - session_id = stringutils.random_string(24) - - self.sessions[session_id] = { - "id": session_id, - "clientdict": clientdict, - "ui_auth": ui_auth, - "creds": {}, - "serverdict": {}, - "description": description, - } - - return self.sessions[session_id] - - def _get_session_info(self, session_id: str) -> dict: - """ - Gets a session given a session ID. - - The session can be used to track data across multiple requests, e.g. for - interactive authentication. - """ - try: - return self.sessions[session_id] - except KeyError: - raise SynapseError(400, "Unknown session ID: %s" % (session_id,)) - async def get_access_token_for_user_id( self, user_id: str, device_id: Optional[str], valid_until_ms: Optional[int] ): @@ -994,13 +947,6 @@ class AuthHandler(BaseHandler): await self.store.user_delete_threepid(user_id, medium, address) return result - def _save_session(self, session: Dict[str, Any]) -> None: - """Update the last used time on the session to now and add it back to the session store.""" - # TODO: Persistent storage - logger.debug("Saving session %s", session) - session["last_used"] = self.hs.get_clock().time_msec() - self.sessions[session["id"]] = session - async def hash(self, password: str) -> str: """Computes a secure hash of password. @@ -1052,7 +998,7 @@ class AuthHandler(BaseHandler): else: return False - def start_sso_ui_auth(self, redirect_url: str, session_id: str) -> str: + async def start_sso_ui_auth(self, redirect_url: str, session_id: str) -> str: """ Get the HTML for the SSO redirect confirmation page. @@ -1063,12 +1009,15 @@ class AuthHandler(BaseHandler): Returns: The HTML to render. """ - session = self._get_session_info(session_id) + try: + session = await self.store.get_ui_auth_session(session_id) + except StoreError: + raise SynapseError(400, "Unknown session ID: %s" % (session_id,)) return self._sso_auth_confirm_template.render( - description=session["description"], redirect_url=redirect_url, + description=session.description, redirect_url=redirect_url, ) - def complete_sso_ui_auth( + async def complete_sso_ui_auth( self, registered_user_id: str, session_id: str, request: SynapseRequest, ): """Having figured out a mxid for this user, complete the HTTP request @@ -1080,13 +1029,11 @@ class AuthHandler(BaseHandler): process. """ # Mark the stage of the authentication as successful. - sess = self._get_session_info(session_id) - creds = sess["creds"] - # Save the user who authenticated with SSO, this will be used to ensure # that the account be modified is also the person who logged in. - creds[LoginType.SSO] = registered_user_id - self._save_session(sess) + await self.store.mark_ui_auth_stage_complete( + session_id, LoginType.SSO, registered_user_id + ) # Render the HTML and return. html_bytes = self._sso_auth_success_template.encode("utf-8") diff --git a/synapse/handlers/cas_handler.py b/synapse/handlers/cas_handler.py index 5cb3f9d133..64aaa1335c 100644 --- a/synapse/handlers/cas_handler.py +++ b/synapse/handlers/cas_handler.py @@ -206,7 +206,7 @@ class CasHandler: registered_user_id = await self._auth_handler.check_user_exists(user_id) if session: - self._auth_handler.complete_sso_ui_auth( + await self._auth_handler.complete_sso_ui_auth( registered_user_id, session, request, ) diff --git a/synapse/handlers/saml_handler.py b/synapse/handlers/saml_handler.py index 7c9454b504..96f2dd36ad 100644 --- a/synapse/handlers/saml_handler.py +++ b/synapse/handlers/saml_handler.py @@ -149,7 +149,7 @@ class SamlHandler: # Complete the interactive auth session or the login. if current_session and current_session.ui_auth_session_id: - self._auth_handler.complete_sso_ui_auth( + await self._auth_handler.complete_sso_ui_auth( user_id, current_session.ui_auth_session_id, request ) diff --git a/synapse/rest/client/v2_alpha/auth.py b/synapse/rest/client/v2_alpha/auth.py index 11599f5005..24dd3d3e96 100644 --- a/synapse/rest/client/v2_alpha/auth.py +++ b/synapse/rest/client/v2_alpha/auth.py @@ -140,7 +140,7 @@ class AuthRestServlet(RestServlet): self._cas_server_url = hs.config.cas_server_url self._cas_service_url = hs.config.cas_service_url - def on_GET(self, request, stagetype): + async def on_GET(self, request, stagetype): session = parse_string(request, "session") if not session: raise SynapseError(400, "No session supplied") @@ -180,7 +180,7 @@ class AuthRestServlet(RestServlet): else: raise SynapseError(400, "Homeserver not configured for SSO.") - html = self.auth_handler.start_sso_ui_auth(sso_redirect_url, session) + html = await self.auth_handler.start_sso_ui_auth(sso_redirect_url, session) else: raise SynapseError(404, "Unknown auth stage type") diff --git a/synapse/rest/client/v2_alpha/register.py b/synapse/rest/client/v2_alpha/register.py index d1b5c49989..af08cc6cce 100644 --- a/synapse/rest/client/v2_alpha/register.py +++ b/synapse/rest/client/v2_alpha/register.py @@ -499,7 +499,7 @@ class RegisterRestServlet(RestServlet): # registered a user for this session, so we could just return the # user here. We carry on and go through the auth checks though, # for paranoia. - registered_user_id = self.auth_handler.get_session_data( + registered_user_id = await self.auth_handler.get_session_data( session_id, "registered_user_id", None ) @@ -598,7 +598,7 @@ class RegisterRestServlet(RestServlet): # remember that we've now registered that user account, and with # what user ID (since the user may not have specified) - self.auth_handler.set_session_data( + await self.auth_handler.set_session_data( session_id, "registered_user_id", registered_user_id ) diff --git a/synapse/storage/data_stores/main/__init__.py b/synapse/storage/data_stores/main/__init__.py index bd7c3a00ea..ceba10882c 100644 --- a/synapse/storage/data_stores/main/__init__.py +++ b/synapse/storage/data_stores/main/__init__.py @@ -66,6 +66,7 @@ from .stats import StatsStore from .stream import StreamStore from .tags import TagsStore from .transactions import TransactionStore +from .ui_auth import UIAuthStore from .user_directory import UserDirectoryStore from .user_erasure_store import UserErasureStore @@ -112,6 +113,7 @@ class DataStore( StatsStore, RelationsStore, CacheInvalidationStore, + UIAuthStore, ): def __init__(self, database: Database, db_conn, hs): self.hs = hs diff --git a/synapse/storage/data_stores/main/schema/delta/58/03persist_ui_auth.sql b/synapse/storage/data_stores/main/schema/delta/58/03persist_ui_auth.sql new file mode 100644 index 0000000000..dcb593fc2d --- /dev/null +++ b/synapse/storage/data_stores/main/schema/delta/58/03persist_ui_auth.sql @@ -0,0 +1,36 @@ +/* Copyright 2020 The Matrix.org Foundation C.I.C + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +CREATE TABLE IF NOT EXISTS ui_auth_sessions( + session_id TEXT NOT NULL, -- The session ID passed to the client. + creation_time BIGINT NOT NULL, -- The time this session was created (epoch time in milliseconds). + serverdict TEXT NOT NULL, -- A JSON dictionary of arbitrary data added by Synapse. + clientdict TEXT NOT NULL, -- A JSON dictionary of arbitrary data from the client. + uri TEXT NOT NULL, -- The URI the UI authentication session is using. + method TEXT NOT NULL, -- The HTTP method the UI authentication session is using. + -- The clientdict, uri, and method make up an tuple that must be immutable + -- throughout the lifetime of the UI Auth session. + description TEXT NOT NULL, -- A human readable description of the operation which caused the UI Auth flow to occur. + UNIQUE (session_id) +); + +CREATE TABLE IF NOT EXISTS ui_auth_sessions_credentials( + session_id TEXT NOT NULL, -- The corresponding UI Auth session. + stage_type TEXT NOT NULL, -- The stage type. + result TEXT NOT NULL, -- The result of the stage verification, stored as JSON. + UNIQUE (session_id, stage_type), + FOREIGN KEY (session_id) + REFERENCES ui_auth_sessions (session_id) +); diff --git a/synapse/storage/data_stores/main/ui_auth.py b/synapse/storage/data_stores/main/ui_auth.py new file mode 100644 index 0000000000..c8eebc9378 --- /dev/null +++ b/synapse/storage/data_stores/main/ui_auth.py @@ -0,0 +1,279 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import json +from typing import Any, Dict, Optional, Union + +import attr + +import synapse.util.stringutils as stringutils +from synapse.api.errors import StoreError +from synapse.storage._base import SQLBaseStore +from synapse.types import JsonDict + + +@attr.s +class UIAuthSessionData: + session_id = attr.ib(type=str) + # The dictionary from the client root level, not the 'auth' key. + clientdict = attr.ib(type=JsonDict) + # The URI and method the session was intiatied with. These are checked at + # each stage of the authentication to ensure that the asked for operation + # has not changed. + uri = attr.ib(type=str) + method = attr.ib(type=str) + # A string description of the operation that the current authentication is + # authorising. + description = attr.ib(type=str) + + +class UIAuthWorkerStore(SQLBaseStore): + """ + Manage user interactive authentication sessions. + """ + + async def create_ui_auth_session( + self, clientdict: JsonDict, uri: str, method: str, description: str, + ) -> UIAuthSessionData: + """ + Creates a new user interactive authentication session. + + The session can be used to track the stages necessary to authenticate a + user across multiple HTTP requests. + + Args: + clientdict: + The dictionary from the client root level, not the 'auth' key. + uri: + The URI this session was initiated with, this is checked at each + stage of the authentication to ensure that the asked for + operation has not changed. + method: + The method this session was initiated with, this is checked at each + stage of the authentication to ensure that the asked for + operation has not changed. + description: + A string description of the operation that the current + authentication is authorising. + Returns: + The newly created session. + Raises: + StoreError if a unique session ID cannot be generated. + """ + # The clientdict gets stored as JSON. + clientdict_json = json.dumps(clientdict) + + # autogen a session ID and try to create it. We may clash, so just + # try a few times till one goes through, giving up eventually. + attempts = 0 + while attempts < 5: + session_id = stringutils.random_string(24) + + try: + await self.db.simple_insert( + table="ui_auth_sessions", + values={ + "session_id": session_id, + "clientdict": clientdict_json, + "uri": uri, + "method": method, + "description": description, + "serverdict": "{}", + "creation_time": self.hs.get_clock().time_msec(), + }, + desc="create_ui_auth_session", + ) + return UIAuthSessionData( + session_id, clientdict, uri, method, description + ) + except self.db.engine.module.IntegrityError: + attempts += 1 + raise StoreError(500, "Couldn't generate a session ID.") + + async def get_ui_auth_session(self, session_id: str) -> UIAuthSessionData: + """Retrieve a UI auth session. + + Args: + session_id: The ID of the session. + Returns: + A dict containing the device information. + Raises: + StoreError if the session is not found. + """ + result = await self.db.simple_select_one( + table="ui_auth_sessions", + keyvalues={"session_id": session_id}, + retcols=("clientdict", "uri", "method", "description"), + desc="get_ui_auth_session", + ) + + result["clientdict"] = json.loads(result["clientdict"]) + + return UIAuthSessionData(session_id, **result) + + async def mark_ui_auth_stage_complete( + self, session_id: str, stage_type: str, result: Union[str, bool, JsonDict], + ): + """ + Mark a session stage as completed. + + Args: + session_id: The ID of the corresponding session. + stage_type: The completed stage type. + result: The result of the stage verification. + Raises: + StoreError if the session cannot be found. + """ + # Add (or update) the results of the current stage to the database. + # + # Note that we need to allow for the same stage to complete multiple + # times here so that registration is idempotent. + try: + await self.db.simple_upsert( + table="ui_auth_sessions_credentials", + keyvalues={"session_id": session_id, "stage_type": stage_type}, + values={"result": json.dumps(result)}, + desc="mark_ui_auth_stage_complete", + ) + except self.db.engine.module.IntegrityError: + raise StoreError(400, "Unknown session ID: %s" % (session_id,)) + + async def get_completed_ui_auth_stages( + self, session_id: str + ) -> Dict[str, Union[str, bool, JsonDict]]: + """ + Retrieve the completed stages of a UI authentication session. + + Args: + session_id: The ID of the session. + Returns: + The completed stages mapped to the result of the verification of + that auth-type. + """ + results = {} + for row in await self.db.simple_select_list( + table="ui_auth_sessions_credentials", + keyvalues={"session_id": session_id}, + retcols=("stage_type", "result"), + desc="get_completed_ui_auth_stages", + ): + results[row["stage_type"]] = json.loads(row["result"]) + + return results + + async def set_ui_auth_session_data(self, session_id: str, key: str, value: Any): + """ + Store a key-value pair into the sessions data associated with this + request. This data is stored server-side and cannot be modified by + the client. + + Args: + session_id: The ID of this session as returned from check_auth + key: The key to store the data under + value: The data to store + Raises: + StoreError if the session cannot be found. + """ + await self.db.runInteraction( + "set_ui_auth_session_data", + self._set_ui_auth_session_data_txn, + session_id, + key, + value, + ) + + def _set_ui_auth_session_data_txn(self, txn, session_id: str, key: str, value: Any): + # Get the current value. + result = self.db.simple_select_one_txn( + txn, + table="ui_auth_sessions", + keyvalues={"session_id": session_id}, + retcols=("serverdict",), + ) + + # Update it and add it back to the database. + serverdict = json.loads(result["serverdict"]) + serverdict[key] = value + + self.db.simple_update_one_txn( + txn, + table="ui_auth_sessions", + keyvalues={"session_id": session_id}, + updatevalues={"serverdict": json.dumps(serverdict)}, + ) + + async def get_ui_auth_session_data( + self, session_id: str, key: str, default: Optional[Any] = None + ) -> Any: + """ + Retrieve data stored with set_session_data + + Args: + session_id: The ID of this session as returned from check_auth + key: The key to store the data under + default: Value to return if the key has not been set + Raises: + StoreError if the session cannot be found. + """ + result = await self.db.simple_select_one( + table="ui_auth_sessions", + keyvalues={"session_id": session_id}, + retcols=("serverdict",), + desc="get_ui_auth_session_data", + ) + + serverdict = json.loads(result["serverdict"]) + + return serverdict.get(key, default) + + +class UIAuthStore(UIAuthWorkerStore): + def delete_old_ui_auth_sessions(self, expiration_time: int): + """ + Remove sessions which were last used earlier than the expiration time. + + Args: + expiration_time: The latest time that is still considered valid. + This is an epoch time in milliseconds. + + """ + return self.db.runInteraction( + "delete_old_ui_auth_sessions", + self._delete_old_ui_auth_sessions_txn, + expiration_time, + ) + + def _delete_old_ui_auth_sessions_txn(self, txn, expiration_time: int): + # Get the expired sessions. + sql = "SELECT session_id FROM ui_auth_sessions WHERE creation_time <= ?" + txn.execute(sql, [expiration_time]) + session_ids = [r[0] for r in txn.fetchall()] + + # Delete the corresponding completed credentials. + self.db.simple_delete_many_txn( + txn, + table="ui_auth_sessions_credentials", + column="session_id", + iterable=session_ids, + keyvalues={}, + ) + + # Finally, delete the sessions. + self.db.simple_delete_many_txn( + txn, + table="ui_auth_sessions", + column="session_id", + iterable=session_ids, + keyvalues={}, + ) diff --git a/synapse/storage/engines/sqlite.py b/synapse/storage/engines/sqlite.py index 3bc2e8b986..215a949442 100644 --- a/synapse/storage/engines/sqlite.py +++ b/synapse/storage/engines/sqlite.py @@ -85,6 +85,7 @@ class Sqlite3Engine(BaseDatabaseEngine["sqlite3.Connection"]): prepare_database(db_conn, self, config=None) db_conn.create_function("rank", 1, _rank) + db_conn.execute("PRAGMA foreign_keys = ON;") def is_deadlock(self, error): return False diff --git a/tests/rest/client/v2_alpha/test_auth.py b/tests/rest/client/v2_alpha/test_auth.py index 624bf5ada2..587be7b2e7 100644 --- a/tests/rest/client/v2_alpha/test_auth.py +++ b/tests/rest/client/v2_alpha/test_auth.py @@ -181,3 +181,43 @@ class FallbackAuthTests(unittest.HomeserverTestCase): ) self.render(request) self.assertEqual(channel.code, 403) + + def test_complete_operation_unknown_session(self): + """ + Attempting to mark an invalid session as complete should error. + """ + + # Make the initial request to register. (Later on a different password + # will be used.) + request, channel = self.make_request( + "POST", + "register", + {"username": "user", "type": "m.login.password", "password": "bar"}, + ) + self.render(request) + + # Returns a 401 as per the spec + self.assertEqual(request.code, 401) + # Grab the session + session = channel.json_body["session"] + # Assert our configured public key is being given + self.assertEqual( + channel.json_body["params"]["m.login.recaptcha"]["public_key"], "brokencake" + ) + + request, channel = self.make_request( + "GET", "auth/m.login.recaptcha/fallback/web?session=" + session + ) + self.render(request) + self.assertEqual(request.code, 200) + + # Attempt to complete an unknown session, which should return an error. + unknown_session = session + "unknown" + request, channel = self.make_request( + "POST", + "auth/m.login.recaptcha/fallback/web?session=" + + unknown_session + + "&g-recaptcha-response=a", + ) + self.render(request) + self.assertEqual(request.code, 400) diff --git a/tests/utils.py b/tests/utils.py index 037cb134f0..f9be62b499 100644 --- a/tests/utils.py +++ b/tests/utils.py @@ -512,8 +512,8 @@ class MockClock(object): return t - def looping_call(self, function, interval): - self.loopers.append([function, interval / 1000.0, self.now]) + def looping_call(self, function, interval, *args, **kwargs): + self.loopers.append([function, interval / 1000.0, self.now, args, kwargs]) def cancel_call_later(self, timer, ignore_errs=False): if timer[2]: @@ -543,9 +543,9 @@ class MockClock(object): self.timers.append(t) for looped in self.loopers: - func, interval, last = looped + func, interval, last, args, kwargs = looped if last + interval < self.now: - func() + func(*args, **kwargs) looped[2] = self.now def advance_time_msec(self, ms): diff --git a/tox.ini b/tox.ini index 2630857436..eccc44e436 100644 --- a/tox.ini +++ b/tox.ini @@ -200,8 +200,9 @@ commands = mypy \ synapse/replication \ synapse/rest \ synapse/spam_checker_api \ - synapse/storage/engines \ + synapse/storage/data_stores/main/ui_auth.py \ synapse/storage/database.py \ + synapse/storage/engines \ synapse/streams \ synapse/util/caches/stream_change_cache.py \ tests/replication/tcp/streams \ -- cgit 1.4.1 From 6b22921b195c24762cd7c02a8b8fad75791fce70 Mon Sep 17 00:00:00 2001 From: Andrew Morgan <1342360+anoadragon453@users.noreply.github.com> Date: Fri, 1 May 2020 15:15:36 +0100 Subject: async/await is_server_admin (#7363) --- changelog.d/7363.misc | 1 + synapse/api/auth.py | 9 +- synapse/federation/federation_client.py | 5 +- synapse/groups/groups_server.py | 64 +++++------ synapse/handlers/_base.py | 16 ++- synapse/handlers/directory.py | 51 ++++----- synapse/handlers/federation.py | 21 ++-- synapse/handlers/groups_local.py | 24 ++-- synapse/handlers/message.py | 83 +++++++------- synapse/handlers/profile.py | 39 ++++--- synapse/handlers/register.py | 49 ++++---- synapse/handlers/room.py | 121 ++++++++++---------- synapse/handlers/room_member.py | 127 ++++++++++----------- synapse/server_notices/consent_server_notices.py | 11 +- .../resource_limits_server_notices.py | 35 +++--- synapse/server_notices/server_notices_manager.py | 32 +++--- synapse/server_notices/server_notices_sender.py | 12 +- synapse/storage/data_stores/main/registration.py | 5 +- tests/handlers/test_profile.py | 60 ++++++---- tests/handlers/test_register.py | 29 +++-- .../test_resource_limits_server_notices.py | 48 ++++---- tests/test_federation.py | 6 +- 22 files changed, 410 insertions(+), 438 deletions(-) create mode 100644 changelog.d/7363.misc (limited to 'tests') diff --git a/changelog.d/7363.misc b/changelog.d/7363.misc new file mode 100644 index 0000000000..1e3cddde79 --- /dev/null +++ b/changelog.d/7363.misc @@ -0,0 +1 @@ +Convert RegistrationWorkerStore.is_server_admin and dependent code to async/await. \ No newline at end of file diff --git a/synapse/api/auth.py b/synapse/api/auth.py index c1ade1333b..c5d1eb952b 100644 --- a/synapse/api/auth.py +++ b/synapse/api/auth.py @@ -537,8 +537,7 @@ class Auth(object): return defer.succeed(auth_ids) - @defer.inlineCallbacks - def check_can_change_room_list(self, room_id: str, user: UserID): + async def check_can_change_room_list(self, room_id: str, user: UserID): """Determine whether the user is allowed to edit the room's entry in the published room list. @@ -547,17 +546,17 @@ class Auth(object): user """ - is_admin = yield self.is_server_admin(user) + is_admin = await self.is_server_admin(user) if is_admin: return True user_id = user.to_string() - yield self.check_user_in_room(room_id, user_id) + await self.check_user_in_room(room_id, user_id) # We currently require the user is a "moderator" in the room. We do this # by checking if they would (theoretically) be able to change the # m.room.canonical_alias events - power_level_event = yield self.state.get_current_state( + power_level_event = await self.state.get_current_state( room_id, EventTypes.PowerLevels, "" ) diff --git a/synapse/federation/federation_client.py b/synapse/federation/federation_client.py index 58b13da616..687cd841ac 100644 --- a/synapse/federation/federation_client.py +++ b/synapse/federation/federation_client.py @@ -976,14 +976,13 @@ class FederationClient(FederationBase): return signed_events - @defer.inlineCallbacks - def forward_third_party_invite(self, destinations, room_id, event_dict): + async def forward_third_party_invite(self, destinations, room_id, event_dict): for destination in destinations: if destination == self.server_name: continue try: - yield self.transport_layer.exchange_third_party_invite( + await self.transport_layer.exchange_third_party_invite( destination=destination, room_id=room_id, event_dict=event_dict ) return None diff --git a/synapse/groups/groups_server.py b/synapse/groups/groups_server.py index 4f0dc0a209..4acb4fa489 100644 --- a/synapse/groups/groups_server.py +++ b/synapse/groups/groups_server.py @@ -748,17 +748,18 @@ class GroupsServerHandler(GroupsServerWorkerHandler): raise NotImplementedError() - @defer.inlineCallbacks - def remove_user_from_group(self, group_id, user_id, requester_user_id, content): + async def remove_user_from_group( + self, group_id, user_id, requester_user_id, content + ): """Remove a user from the group; either a user is leaving or an admin kicked them. """ - yield self.check_group_is_ours(group_id, requester_user_id, and_exists=True) + await self.check_group_is_ours(group_id, requester_user_id, and_exists=True) is_kick = False if requester_user_id != user_id: - is_admin = yield self.store.is_user_admin_in_group( + is_admin = await self.store.is_user_admin_in_group( group_id, requester_user_id ) if not is_admin: @@ -766,30 +767,29 @@ class GroupsServerHandler(GroupsServerWorkerHandler): is_kick = True - yield self.store.remove_user_from_group(group_id, user_id) + await self.store.remove_user_from_group(group_id, user_id) if is_kick: if self.hs.is_mine_id(user_id): groups_local = self.hs.get_groups_local_handler() - yield groups_local.user_removed_from_group(group_id, user_id, {}) + await groups_local.user_removed_from_group(group_id, user_id, {}) else: - yield self.transport_client.remove_user_from_group_notification( + await self.transport_client.remove_user_from_group_notification( get_domain_from_id(user_id), group_id, user_id, {} ) if not self.hs.is_mine_id(user_id): - yield self.store.maybe_delete_remote_profile_cache(user_id) + await self.store.maybe_delete_remote_profile_cache(user_id) # Delete group if the last user has left - users = yield self.store.get_users_in_group(group_id, include_private=True) + users = await self.store.get_users_in_group(group_id, include_private=True) if not users: - yield self.store.delete_group(group_id) + await self.store.delete_group(group_id) return {} - @defer.inlineCallbacks - def create_group(self, group_id, requester_user_id, content): - group = yield self.check_group_is_ours(group_id, requester_user_id) + async def create_group(self, group_id, requester_user_id, content): + group = await self.check_group_is_ours(group_id, requester_user_id) logger.info("Attempting to create group with ID: %r", group_id) @@ -799,7 +799,7 @@ class GroupsServerHandler(GroupsServerWorkerHandler): if group: raise SynapseError(400, "Group already exists") - is_admin = yield self.auth.is_server_admin( + is_admin = await self.auth.is_server_admin( UserID.from_string(requester_user_id) ) if not is_admin: @@ -822,7 +822,7 @@ class GroupsServerHandler(GroupsServerWorkerHandler): long_description = profile.get("long_description") user_profile = content.get("user_profile", {}) - yield self.store.create_group( + await self.store.create_group( group_id, requester_user_id, name=name, @@ -834,7 +834,7 @@ class GroupsServerHandler(GroupsServerWorkerHandler): if not self.hs.is_mine_id(requester_user_id): remote_attestation = content["attestation"] - yield self.attestations.verify_attestation( + await self.attestations.verify_attestation( remote_attestation, user_id=requester_user_id, group_id=group_id ) @@ -845,7 +845,7 @@ class GroupsServerHandler(GroupsServerWorkerHandler): local_attestation = None remote_attestation = None - yield self.store.add_user_to_group( + await self.store.add_user_to_group( group_id, requester_user_id, is_admin=True, @@ -855,7 +855,7 @@ class GroupsServerHandler(GroupsServerWorkerHandler): ) if not self.hs.is_mine_id(requester_user_id): - yield self.store.add_remote_profile_cache( + await self.store.add_remote_profile_cache( requester_user_id, displayname=user_profile.get("displayname"), avatar_url=user_profile.get("avatar_url"), @@ -863,8 +863,7 @@ class GroupsServerHandler(GroupsServerWorkerHandler): return {"group_id": group_id} - @defer.inlineCallbacks - def delete_group(self, group_id, requester_user_id): + async def delete_group(self, group_id, requester_user_id): """Deletes a group, kicking out all current members. Only group admins or server admins can call this request @@ -877,14 +876,14 @@ class GroupsServerHandler(GroupsServerWorkerHandler): Deferred """ - yield self.check_group_is_ours(group_id, requester_user_id, and_exists=True) + await self.check_group_is_ours(group_id, requester_user_id, and_exists=True) # Only server admins or group admins can delete groups. - is_admin = yield self.store.is_user_admin_in_group(group_id, requester_user_id) + is_admin = await self.store.is_user_admin_in_group(group_id, requester_user_id) if not is_admin: - is_admin = yield self.auth.is_server_admin( + is_admin = await self.auth.is_server_admin( UserID.from_string(requester_user_id) ) @@ -892,18 +891,17 @@ class GroupsServerHandler(GroupsServerWorkerHandler): raise SynapseError(403, "User is not an admin") # Before deleting the group lets kick everyone out of it - users = yield self.store.get_users_in_group(group_id, include_private=True) + users = await self.store.get_users_in_group(group_id, include_private=True) - @defer.inlineCallbacks - def _kick_user_from_group(user_id): + async def _kick_user_from_group(user_id): if self.hs.is_mine_id(user_id): groups_local = self.hs.get_groups_local_handler() - yield groups_local.user_removed_from_group(group_id, user_id, {}) + await groups_local.user_removed_from_group(group_id, user_id, {}) else: - yield self.transport_client.remove_user_from_group_notification( + await self.transport_client.remove_user_from_group_notification( get_domain_from_id(user_id), group_id, user_id, {} ) - yield self.store.maybe_delete_remote_profile_cache(user_id) + await self.store.maybe_delete_remote_profile_cache(user_id) # We kick users out in the order of: # 1. Non-admins @@ -922,11 +920,11 @@ class GroupsServerHandler(GroupsServerWorkerHandler): else: non_admins.append(u["user_id"]) - yield concurrently_execute(_kick_user_from_group, non_admins, 10) - yield concurrently_execute(_kick_user_from_group, admins, 10) - yield _kick_user_from_group(requester_user_id) + await concurrently_execute(_kick_user_from_group, non_admins, 10) + await concurrently_execute(_kick_user_from_group, admins, 10) + await _kick_user_from_group(requester_user_id) - yield self.store.delete_group(group_id) + await self.store.delete_group(group_id) def _parse_join_policy_from_contents(content): diff --git a/synapse/handlers/_base.py b/synapse/handlers/_base.py index 51413d910e..3b781d9836 100644 --- a/synapse/handlers/_base.py +++ b/synapse/handlers/_base.py @@ -126,30 +126,28 @@ class BaseHandler(object): retry_after_ms=int(1000 * (time_allowed - time_now)) ) - @defer.inlineCallbacks - def maybe_kick_guest_users(self, event, context=None): + async def maybe_kick_guest_users(self, event, context=None): # Technically this function invalidates current_state by changing it. # Hopefully this isn't that important to the caller. if event.type == EventTypes.GuestAccess: guest_access = event.content.get("guest_access", "forbidden") if guest_access != "can_join": if context: - current_state_ids = yield context.get_current_state_ids() - current_state = yield self.store.get_events( + current_state_ids = await context.get_current_state_ids() + current_state = await self.store.get_events( list(current_state_ids.values()) ) else: - current_state = yield self.state_handler.get_current_state( + current_state = await self.state_handler.get_current_state( event.room_id ) current_state = list(current_state.values()) logger.info("maybe_kick_guest_users %r", current_state) - yield self.kick_guest_users(current_state) + await self.kick_guest_users(current_state) - @defer.inlineCallbacks - def kick_guest_users(self, current_state): + async def kick_guest_users(self, current_state): for member_event in current_state: try: if member_event.type != EventTypes.Member: @@ -180,7 +178,7 @@ class BaseHandler(object): # homeserver. requester = synapse.types.create_requester(target_user, is_guest=True) handler = self.hs.get_room_member_handler() - yield handler.update_membership( + await handler.update_membership( requester, target_user, member_event.room_id, diff --git a/synapse/handlers/directory.py b/synapse/handlers/directory.py index 53e5f585d9..f2f16b1e43 100644 --- a/synapse/handlers/directory.py +++ b/synapse/handlers/directory.py @@ -86,8 +86,7 @@ class DirectoryHandler(BaseHandler): room_alias, room_id, servers, creator=creator ) - @defer.inlineCallbacks - def create_association( + async def create_association( self, requester: Requester, room_alias: RoomAlias, @@ -129,10 +128,10 @@ class DirectoryHandler(BaseHandler): else: # Server admins are not subject to the same constraints as normal # users when creating an alias (e.g. being in the room). - is_admin = yield self.auth.is_server_admin(requester.user) + is_admin = await self.auth.is_server_admin(requester.user) if (self.require_membership and check_membership) and not is_admin: - rooms_for_user = yield self.store.get_rooms_for_user(user_id) + rooms_for_user = await self.store.get_rooms_for_user(user_id) if room_id not in rooms_for_user: raise AuthError( 403, "You must be in the room to create an alias for it" @@ -149,7 +148,7 @@ class DirectoryHandler(BaseHandler): # per alias creation rule? raise SynapseError(403, "Not allowed to create alias") - can_create = yield self.can_modify_alias(room_alias, user_id=user_id) + can_create = await self.can_modify_alias(room_alias, user_id=user_id) if not can_create: raise AuthError( 400, @@ -157,10 +156,9 @@ class DirectoryHandler(BaseHandler): errcode=Codes.EXCLUSIVE, ) - yield self._create_association(room_alias, room_id, servers, creator=user_id) + await self._create_association(room_alias, room_id, servers, creator=user_id) - @defer.inlineCallbacks - def delete_association(self, requester: Requester, room_alias: RoomAlias): + async def delete_association(self, requester: Requester, room_alias: RoomAlias): """Remove an alias from the directory (this is only meant for human users; AS users should call @@ -184,7 +182,7 @@ class DirectoryHandler(BaseHandler): user_id = requester.user.to_string() try: - can_delete = yield self._user_can_delete_alias(room_alias, user_id) + can_delete = await self._user_can_delete_alias(room_alias, user_id) except StoreError as e: if e.code == 404: raise NotFoundError("Unknown room alias") @@ -193,7 +191,7 @@ class DirectoryHandler(BaseHandler): if not can_delete: raise AuthError(403, "You don't have permission to delete the alias.") - can_delete = yield self.can_modify_alias(room_alias, user_id=user_id) + can_delete = await self.can_modify_alias(room_alias, user_id=user_id) if not can_delete: raise SynapseError( 400, @@ -201,10 +199,10 @@ class DirectoryHandler(BaseHandler): errcode=Codes.EXCLUSIVE, ) - room_id = yield self._delete_association(room_alias) + room_id = await self._delete_association(room_alias) try: - yield self._update_canonical_alias(requester, user_id, room_id, room_alias) + await self._update_canonical_alias(requester, user_id, room_id, room_alias) except AuthError as e: logger.info("Failed to update alias events: %s", e) @@ -296,15 +294,14 @@ class DirectoryHandler(BaseHandler): Codes.NOT_FOUND, ) - @defer.inlineCallbacks - def _update_canonical_alias( + async def _update_canonical_alias( self, requester: Requester, user_id: str, room_id: str, room_alias: RoomAlias ): """ Send an updated canonical alias event if the removed alias was set as the canonical alias or listed in the alt_aliases field. """ - alias_event = yield self.state.get_current_state( + alias_event = await self.state.get_current_state( room_id, EventTypes.CanonicalAlias, "" ) @@ -335,7 +332,7 @@ class DirectoryHandler(BaseHandler): del content["alt_aliases"] if send_update: - yield self.event_creation_handler.create_and_send_nonmember_event( + await self.event_creation_handler.create_and_send_nonmember_event( requester, { "type": EventTypes.CanonicalAlias, @@ -376,8 +373,7 @@ class DirectoryHandler(BaseHandler): # either no interested services, or no service with an exclusive lock return defer.succeed(True) - @defer.inlineCallbacks - def _user_can_delete_alias(self, alias: RoomAlias, user_id: str): + async def _user_can_delete_alias(self, alias: RoomAlias, user_id: str): """Determine whether a user can delete an alias. One of the following must be true: @@ -388,24 +384,23 @@ class DirectoryHandler(BaseHandler): for the current room. """ - creator = yield self.store.get_room_alias_creator(alias.to_string()) + creator = await self.store.get_room_alias_creator(alias.to_string()) if creator is not None and creator == user_id: return True # Resolve the alias to the corresponding room. - room_mapping = yield self.get_association(alias) + room_mapping = await self.get_association(alias) room_id = room_mapping["room_id"] if not room_id: return False - res = yield self.auth.check_can_change_room_list( + res = await self.auth.check_can_change_room_list( room_id, UserID.from_string(user_id) ) return res - @defer.inlineCallbacks - def edit_published_room_list( + async def edit_published_room_list( self, requester: Requester, room_id: str, visibility: str ): """Edit the entry of the room in the published room list. @@ -433,11 +428,11 @@ class DirectoryHandler(BaseHandler): 403, "This user is not permitted to publish rooms to the room list" ) - room = yield self.store.get_room(room_id) + room = await self.store.get_room(room_id) if room is None: raise SynapseError(400, "Unknown room") - can_change_room_list = yield self.auth.check_can_change_room_list( + can_change_room_list = await self.auth.check_can_change_room_list( room_id, requester.user ) if not can_change_room_list: @@ -449,8 +444,8 @@ class DirectoryHandler(BaseHandler): making_public = visibility == "public" if making_public: - room_aliases = yield self.store.get_aliases_for_room(room_id) - canonical_alias = yield self.store.get_canonical_alias_for_room(room_id) + room_aliases = await self.store.get_aliases_for_room(room_id) + canonical_alias = await self.store.get_canonical_alias_for_room(room_id) if canonical_alias: room_aliases.append(canonical_alias) @@ -462,7 +457,7 @@ class DirectoryHandler(BaseHandler): # per alias creation rule? raise SynapseError(403, "Not allowed to publish room") - yield self.store.set_room_is_public(room_id, making_public) + await self.store.set_room_is_public(room_id, making_public) @defer.inlineCallbacks def edit_published_appservice_room_list( diff --git a/synapse/handlers/federation.py b/synapse/handlers/federation.py index 41b96c0a73..4e5c645525 100644 --- a/synapse/handlers/federation.py +++ b/synapse/handlers/federation.py @@ -2562,9 +2562,8 @@ class FederationHandler(BaseHandler): "missing": [e.event_id for e in missing_locals], } - @defer.inlineCallbacks @log_function - def exchange_third_party_invite( + async def exchange_third_party_invite( self, sender_user_id, target_user_id, room_id, signed ): third_party_invite = {"signed": signed} @@ -2580,16 +2579,16 @@ class FederationHandler(BaseHandler): "state_key": target_user_id, } - if (yield self.auth.check_host_in_room(room_id, self.hs.hostname)): - room_version = yield self.store.get_room_version_id(room_id) + if await self.auth.check_host_in_room(room_id, self.hs.hostname): + room_version = await self.store.get_room_version_id(room_id) builder = self.event_builder_factory.new(room_version, event_dict) EventValidator().validate_builder(builder) - event, context = yield self.event_creation_handler.create_new_client_event( + event, context = await self.event_creation_handler.create_new_client_event( builder=builder ) - event_allowed = yield self.third_party_event_rules.check_event_allowed( + event_allowed = await self.third_party_event_rules.check_event_allowed( event, context ) if not event_allowed: @@ -2601,7 +2600,7 @@ class FederationHandler(BaseHandler): 403, "This event is not allowed in this context", Codes.FORBIDDEN ) - event, context = yield self.add_display_name_to_third_party_invite( + event, context = await self.add_display_name_to_third_party_invite( room_version, event_dict, event, context ) @@ -2612,19 +2611,19 @@ class FederationHandler(BaseHandler): event.internal_metadata.send_on_behalf_of = self.hs.hostname try: - yield self.auth.check_from_context(room_version, event, context) + await self.auth.check_from_context(room_version, event, context) except AuthError as e: logger.warning("Denying new third party invite %r because %s", event, e) raise e - yield self._check_signature(event, context) + await self._check_signature(event, context) # We retrieve the room member handler here as to not cause a cyclic dependency member_handler = self.hs.get_room_member_handler() - yield member_handler.send_membership_event(None, event, context) + await member_handler.send_membership_event(None, event, context) else: destinations = {x.split(":", 1)[-1] for x in (sender_user_id, room_id)} - yield self.federation_client.forward_third_party_invite( + await self.federation_client.forward_third_party_invite( destinations, room_id, event_dict ) diff --git a/synapse/handlers/groups_local.py b/synapse/handlers/groups_local.py index ad22415782..ca5c83811a 100644 --- a/synapse/handlers/groups_local.py +++ b/synapse/handlers/groups_local.py @@ -284,15 +284,14 @@ class GroupsLocalHandler(GroupsLocalWorkerHandler): set_group_join_policy = _create_rerouter("set_group_join_policy") - @defer.inlineCallbacks - def create_group(self, group_id, user_id, content): + async def create_group(self, group_id, user_id, content): """Create a group """ logger.info("Asking to create group with ID: %r", group_id) if self.is_mine_id(group_id): - res = yield self.groups_server_handler.create_group( + res = await self.groups_server_handler.create_group( group_id, user_id, content ) local_attestation = None @@ -301,10 +300,10 @@ class GroupsLocalHandler(GroupsLocalWorkerHandler): local_attestation = self.attestations.create_attestation(group_id, user_id) content["attestation"] = local_attestation - content["user_profile"] = yield self.profile_handler.get_profile(user_id) + content["user_profile"] = await self.profile_handler.get_profile(user_id) try: - res = yield self.transport_client.create_group( + res = await self.transport_client.create_group( get_domain_from_id(group_id), group_id, user_id, content ) except HttpResponseException as e: @@ -313,7 +312,7 @@ class GroupsLocalHandler(GroupsLocalWorkerHandler): raise SynapseError(502, "Failed to contact group server") remote_attestation = res["attestation"] - yield self.attestations.verify_attestation( + await self.attestations.verify_attestation( remote_attestation, group_id=group_id, user_id=user_id, @@ -321,7 +320,7 @@ class GroupsLocalHandler(GroupsLocalWorkerHandler): ) is_publicised = content.get("publicise", False) - token = yield self.store.register_user_group_membership( + token = await self.store.register_user_group_membership( group_id, user_id, membership="join", @@ -482,12 +481,13 @@ class GroupsLocalHandler(GroupsLocalWorkerHandler): return {"state": "invite", "user_profile": user_profile} - @defer.inlineCallbacks - def remove_user_from_group(self, group_id, user_id, requester_user_id, content): + async def remove_user_from_group( + self, group_id, user_id, requester_user_id, content + ): """Remove a user from a group """ if user_id == requester_user_id: - token = yield self.store.register_user_group_membership( + token = await self.store.register_user_group_membership( group_id, user_id, membership="leave" ) self.notifier.on_new_event("groups_key", token, users=[user_id]) @@ -496,13 +496,13 @@ class GroupsLocalHandler(GroupsLocalWorkerHandler): # retry if the group server is currently down. if self.is_mine_id(group_id): - res = yield self.groups_server_handler.remove_user_from_group( + res = await self.groups_server_handler.remove_user_from_group( group_id, user_id, requester_user_id, content ) else: content["requester_user_id"] = requester_user_id try: - res = yield self.transport_client.remove_user_from_group( + res = await self.transport_client.remove_user_from_group( get_domain_from_id(group_id), group_id, requester_user_id, diff --git a/synapse/handlers/message.py b/synapse/handlers/message.py index 522271eed1..a324f09340 100644 --- a/synapse/handlers/message.py +++ b/synapse/handlers/message.py @@ -626,8 +626,7 @@ class EventCreationHandler(object): msg = self._block_events_without_consent_error % {"consent_uri": consent_uri} raise ConsentNotGivenError(msg=msg, consent_uri=consent_uri) - @defer.inlineCallbacks - def send_nonmember_event(self, requester, event, context, ratelimit=True): + async def send_nonmember_event(self, requester, event, context, ratelimit=True): """ Persists and notifies local clients and federation of an event. @@ -647,7 +646,7 @@ class EventCreationHandler(object): assert self.hs.is_mine(user), "User must be our own: %s" % (user,) if event.is_state(): - prev_state = yield self.deduplicate_state_event(event, context) + prev_state = await self.deduplicate_state_event(event, context) if prev_state is not None: logger.info( "Not bothering to persist state event %s duplicated by %s", @@ -656,7 +655,7 @@ class EventCreationHandler(object): ) return prev_state - yield self.handle_new_client_event( + await self.handle_new_client_event( requester=requester, event=event, context=context, ratelimit=ratelimit ) @@ -683,8 +682,7 @@ class EventCreationHandler(object): return prev_event return - @defer.inlineCallbacks - def create_and_send_nonmember_event( + async def create_and_send_nonmember_event( self, requester, event_dict, ratelimit=True, txn_id=None ): """ @@ -698,8 +696,8 @@ class EventCreationHandler(object): # a situation where event persistence can't keep up, causing # extremities to pile up, which in turn leads to state resolution # taking longer. - with (yield self.limiter.queue(event_dict["room_id"])): - event, context = yield self.create_event( + with (await self.limiter.queue(event_dict["room_id"])): + event, context = await self.create_event( requester, event_dict, token_id=requester.access_token_id, txn_id=txn_id ) @@ -709,7 +707,7 @@ class EventCreationHandler(object): spam_error = "Spam is not permitted here" raise SynapseError(403, spam_error, Codes.FORBIDDEN) - yield self.send_nonmember_event( + await self.send_nonmember_event( requester, event, context, ratelimit=ratelimit ) return event @@ -770,8 +768,7 @@ class EventCreationHandler(object): return (event, context) @measure_func("handle_new_client_event") - @defer.inlineCallbacks - def handle_new_client_event( + async def handle_new_client_event( self, requester, event, context, ratelimit=True, extra_users=[] ): """Processes a new event. This includes checking auth, persisting it, @@ -794,9 +791,9 @@ class EventCreationHandler(object): ): room_version = event.content.get("room_version", RoomVersions.V1.identifier) else: - room_version = yield self.store.get_room_version_id(event.room_id) + room_version = await self.store.get_room_version_id(event.room_id) - event_allowed = yield self.third_party_event_rules.check_event_allowed( + event_allowed = await self.third_party_event_rules.check_event_allowed( event, context ) if not event_allowed: @@ -805,7 +802,7 @@ class EventCreationHandler(object): ) try: - yield self.auth.check_from_context(room_version, event, context) + await self.auth.check_from_context(room_version, event, context) except AuthError as err: logger.warning("Denying new event %r because %s", event, err) raise err @@ -818,7 +815,7 @@ class EventCreationHandler(object): logger.exception("Failed to encode content: %r", event.content) raise - yield self.action_generator.handle_push_actions_for_event(event, context) + await self.action_generator.handle_push_actions_for_event(event, context) # reraise does not allow inlineCallbacks to preserve the stacktrace, so we # hack around with a try/finally instead. @@ -826,7 +823,7 @@ class EventCreationHandler(object): try: # If we're a worker we need to hit out to the master. if self.config.worker_app: - yield self.send_event_to_master( + await self.send_event_to_master( event_id=event.event_id, store=self.store, requester=requester, @@ -838,7 +835,7 @@ class EventCreationHandler(object): success = True return - yield self.persist_and_notify_client_event( + await self.persist_and_notify_client_event( requester, event, context, ratelimit=ratelimit, extra_users=extra_users ) @@ -883,8 +880,7 @@ class EventCreationHandler(object): Codes.BAD_ALIAS, ) - @defer.inlineCallbacks - def persist_and_notify_client_event( + async def persist_and_notify_client_event( self, requester, event, context, ratelimit=True, extra_users=[] ): """Called when we have fully built the event, have already @@ -901,7 +897,7 @@ class EventCreationHandler(object): # user is actually admin or not). is_admin_redaction = False if event.type == EventTypes.Redaction: - original_event = yield self.store.get_event( + original_event = await self.store.get_event( event.redacts, redact_behaviour=EventRedactBehaviour.AS_IS, get_prev_content=False, @@ -913,11 +909,11 @@ class EventCreationHandler(object): original_event and event.sender != original_event.sender ) - yield self.base_handler.ratelimit( + await self.base_handler.ratelimit( requester, is_admin_redaction=is_admin_redaction ) - yield self.base_handler.maybe_kick_guest_users(event, context) + await self.base_handler.maybe_kick_guest_users(event, context) if event.type == EventTypes.CanonicalAlias: # Validate a newly added alias or newly added alt_aliases. @@ -927,7 +923,7 @@ class EventCreationHandler(object): original_event_id = event.unsigned.get("replaces_state") if original_event_id: - original_event = yield self.store.get_event(original_event_id) + original_event = await self.store.get_event(original_event_id) if original_event: original_alias = original_event.content.get("alias", None) @@ -937,7 +933,7 @@ class EventCreationHandler(object): room_alias_str = event.content.get("alias", None) directory_handler = self.hs.get_handlers().directory_handler if room_alias_str and room_alias_str != original_alias: - yield self._validate_canonical_alias( + await self._validate_canonical_alias( directory_handler, room_alias_str, event.room_id ) @@ -957,7 +953,7 @@ class EventCreationHandler(object): new_alt_aliases = set(alt_aliases) - set(original_alt_aliases) if new_alt_aliases: for alias_str in new_alt_aliases: - yield self._validate_canonical_alias( + await self._validate_canonical_alias( directory_handler, alias_str, event.room_id ) @@ -969,7 +965,7 @@ class EventCreationHandler(object): def is_inviter_member_event(e): return e.type == EventTypes.Member and e.sender == event.sender - current_state_ids = yield context.get_current_state_ids() + current_state_ids = await context.get_current_state_ids() state_to_include_ids = [ e_id @@ -978,7 +974,7 @@ class EventCreationHandler(object): or k == (EventTypes.Member, event.sender) ] - state_to_include = yield self.store.get_events(state_to_include_ids) + state_to_include = await self.store.get_events(state_to_include_ids) event.unsigned["invite_room_state"] = [ { @@ -996,8 +992,8 @@ class EventCreationHandler(object): # way? If we have been invited by a remote server, we need # to get them to sign the event. - returned_invite = yield defer.ensureDeferred( - federation_handler.send_invite(invitee.domain, event) + returned_invite = await federation_handler.send_invite( + invitee.domain, event ) event.unsigned.pop("room_state", None) @@ -1005,7 +1001,7 @@ class EventCreationHandler(object): event.signatures.update(returned_invite.signatures) if event.type == EventTypes.Redaction: - original_event = yield self.store.get_event( + original_event = await self.store.get_event( event.redacts, redact_behaviour=EventRedactBehaviour.AS_IS, get_prev_content=False, @@ -1021,14 +1017,14 @@ class EventCreationHandler(object): if original_event.room_id != event.room_id: raise SynapseError(400, "Cannot redact event from a different room") - prev_state_ids = yield context.get_prev_state_ids() - auth_events_ids = yield self.auth.compute_auth_events( + prev_state_ids = await context.get_prev_state_ids() + auth_events_ids = await self.auth.compute_auth_events( event, prev_state_ids, for_verification=True ) - auth_events = yield self.store.get_events(auth_events_ids) + auth_events = await self.store.get_events(auth_events_ids) auth_events = {(e.type, e.state_key): e for e in auth_events.values()} - room_version = yield self.store.get_room_version_id(event.room_id) + room_version = await self.store.get_room_version_id(event.room_id) room_version_obj = KNOWN_ROOM_VERSIONS[room_version] if event_auth.check_redaction( @@ -1047,11 +1043,11 @@ class EventCreationHandler(object): event.internal_metadata.recheck_redaction = False if event.type == EventTypes.Create: - prev_state_ids = yield context.get_prev_state_ids() + prev_state_ids = await context.get_prev_state_ids() if prev_state_ids: raise AuthError(403, "Changing the room create event is forbidden") - event_stream_id, max_stream_id = yield self.storage.persistence.persist_event( + event_stream_id, max_stream_id = await self.storage.persistence.persist_event( event, context=context ) @@ -1059,7 +1055,7 @@ class EventCreationHandler(object): # If there's an expiry timestamp on the event, schedule its expiry. self._message_handler.maybe_schedule_expiry(event) - yield self.pusher_pool.on_new_notifications(event_stream_id, max_stream_id) + await self.pusher_pool.on_new_notifications(event_stream_id, max_stream_id) def _notify(): try: @@ -1083,13 +1079,12 @@ class EventCreationHandler(object): except Exception: logger.exception("Error bumping presence active time") - @defer.inlineCallbacks - def _send_dummy_events_to_fill_extremities(self): + async def _send_dummy_events_to_fill_extremities(self): """Background task to send dummy events into rooms that have a large number of extremities """ self._expire_rooms_to_exclude_from_dummy_event_insertion() - room_ids = yield self.store.get_rooms_with_many_extremities( + room_ids = await self.store.get_rooms_with_many_extremities( min_count=10, limit=5, room_id_filter=self._rooms_to_exclude_from_dummy_event_insertion.keys(), @@ -1099,9 +1094,9 @@ class EventCreationHandler(object): # For each room we need to find a joined member we can use to send # the dummy event with. - latest_event_ids = yield self.store.get_prev_events_for_room(room_id) + latest_event_ids = await self.store.get_prev_events_for_room(room_id) - members = yield self.state.get_current_users_in_room( + members = await self.state.get_current_users_in_room( room_id, latest_event_ids=latest_event_ids ) dummy_event_sent = False @@ -1110,7 +1105,7 @@ class EventCreationHandler(object): continue requester = create_requester(user_id) try: - event, context = yield self.create_event( + event, context = await self.create_event( requester, { "type": "org.matrix.dummy_event", @@ -1123,7 +1118,7 @@ class EventCreationHandler(object): event.internal_metadata.proactively_send = False - yield self.send_nonmember_event( + await self.send_nonmember_event( requester, event, context, ratelimit=False ) dummy_event_sent = True diff --git a/synapse/handlers/profile.py b/synapse/handlers/profile.py index 6aa1c0f5e0..302efc1b9a 100644 --- a/synapse/handlers/profile.py +++ b/synapse/handlers/profile.py @@ -141,8 +141,9 @@ class BaseProfileHandler(BaseHandler): return result["displayname"] - @defer.inlineCallbacks - def set_displayname(self, target_user, requester, new_displayname, by_admin=False): + async def set_displayname( + self, target_user, requester, new_displayname, by_admin=False + ): """Set the displayname of a user Args: @@ -158,7 +159,7 @@ class BaseProfileHandler(BaseHandler): raise AuthError(400, "Cannot set another user's displayname") if not by_admin and not self.hs.config.enable_set_displayname: - profile = yield self.store.get_profileinfo(target_user.localpart) + profile = await self.store.get_profileinfo(target_user.localpart) if profile.display_name: raise SynapseError( 400, @@ -180,15 +181,15 @@ class BaseProfileHandler(BaseHandler): if by_admin: requester = create_requester(target_user) - yield self.store.set_profile_displayname(target_user.localpart, new_displayname) + await self.store.set_profile_displayname(target_user.localpart, new_displayname) if self.hs.config.user_directory_search_all_users: - profile = yield self.store.get_profileinfo(target_user.localpart) - yield self.user_directory_handler.handle_local_profile_change( + profile = await self.store.get_profileinfo(target_user.localpart) + await self.user_directory_handler.handle_local_profile_change( target_user.to_string(), profile ) - yield self._update_join_states(requester, target_user) + await self._update_join_states(requester, target_user) @defer.inlineCallbacks def get_avatar_url(self, target_user): @@ -217,8 +218,9 @@ class BaseProfileHandler(BaseHandler): return result["avatar_url"] - @defer.inlineCallbacks - def set_avatar_url(self, target_user, requester, new_avatar_url, by_admin=False): + async def set_avatar_url( + self, target_user, requester, new_avatar_url, by_admin=False + ): """target_user is the user whose avatar_url is to be changed; auth_user is the user attempting to make this change.""" if not self.hs.is_mine(target_user): @@ -228,7 +230,7 @@ class BaseProfileHandler(BaseHandler): raise AuthError(400, "Cannot set another user's avatar_url") if not by_admin and not self.hs.config.enable_set_avatar_url: - profile = yield self.store.get_profileinfo(target_user.localpart) + profile = await self.store.get_profileinfo(target_user.localpart) if profile.avatar_url: raise SynapseError( 400, "Changing avatar is disabled on this server", Codes.FORBIDDEN @@ -243,15 +245,15 @@ class BaseProfileHandler(BaseHandler): if by_admin: requester = create_requester(target_user) - yield self.store.set_profile_avatar_url(target_user.localpart, new_avatar_url) + await self.store.set_profile_avatar_url(target_user.localpart, new_avatar_url) if self.hs.config.user_directory_search_all_users: - profile = yield self.store.get_profileinfo(target_user.localpart) - yield self.user_directory_handler.handle_local_profile_change( + profile = await self.store.get_profileinfo(target_user.localpart) + await self.user_directory_handler.handle_local_profile_change( target_user.to_string(), profile ) - yield self._update_join_states(requester, target_user) + await self._update_join_states(requester, target_user) @defer.inlineCallbacks def on_profile_query(self, args): @@ -279,21 +281,20 @@ class BaseProfileHandler(BaseHandler): return response - @defer.inlineCallbacks - def _update_join_states(self, requester, target_user): + async def _update_join_states(self, requester, target_user): if not self.hs.is_mine(target_user): return - yield self.ratelimit(requester) + await self.ratelimit(requester) - room_ids = yield self.store.get_rooms_for_user(target_user.to_string()) + room_ids = await self.store.get_rooms_for_user(target_user.to_string()) for room_id in room_ids: handler = self.hs.get_room_member_handler() try: # Assume the target_user isn't a guest, # because we don't let guests set profile or avatar data. - yield handler.update_membership( + await handler.update_membership( requester, target_user, room_id, diff --git a/synapse/handlers/register.py b/synapse/handlers/register.py index 3a65b46ecd..1e6bdac0ad 100644 --- a/synapse/handlers/register.py +++ b/synapse/handlers/register.py @@ -145,9 +145,9 @@ class RegistrationHandler(BaseHandler): """Registers a new client on the server. Args: - localpart : The local part of the user ID to register. If None, + localpart: The local part of the user ID to register. If None, one will be generated. - password (unicode) : The password to assign to this user so they can + password (unicode): The password to assign to this user so they can login again. This can be None which means they cannot login again via a password (e.g. the user is an application service user). user_type (str|None): type of user. One of the values from @@ -244,7 +244,7 @@ class RegistrationHandler(BaseHandler): fail_count += 1 if not self.hs.config.user_consent_at_registration: - yield self._auto_join_rooms(user_id) + yield defer.ensureDeferred(self._auto_join_rooms(user_id)) else: logger.info( "Skipping auto-join for %s because consent is required at registration", @@ -266,8 +266,7 @@ class RegistrationHandler(BaseHandler): return user_id - @defer.inlineCallbacks - def _auto_join_rooms(self, user_id): + async def _auto_join_rooms(self, user_id): """Automatically joins users to auto join rooms - creating the room in the first place if the user is the first to be created. @@ -281,9 +280,9 @@ class RegistrationHandler(BaseHandler): # that an auto-generated support or bot user is not a real user and will never be # the user to create the room should_auto_create_rooms = False - is_real_user = yield self.store.is_real_user(user_id) + is_real_user = await self.store.is_real_user(user_id) if self.hs.config.autocreate_auto_join_rooms and is_real_user: - count = yield self.store.count_real_users() + count = await self.store.count_real_users() should_auto_create_rooms = count == 1 for r in self.hs.config.auto_join_rooms: logger.info("Auto-joining %s to %s", user_id, r) @@ -302,7 +301,7 @@ class RegistrationHandler(BaseHandler): # getting the RoomCreationHandler during init gives a dependency # loop - yield self.hs.get_room_creation_handler().create_room( + await self.hs.get_room_creation_handler().create_room( fake_requester, config={ "preset": "public_chat", @@ -311,7 +310,7 @@ class RegistrationHandler(BaseHandler): ratelimit=False, ) else: - yield self._join_user_to_room(fake_requester, r) + await self._join_user_to_room(fake_requester, r) except ConsentNotGivenError as e: # Technically not necessary to pull out this error though # moving away from bare excepts is a good thing to do. @@ -319,15 +318,14 @@ class RegistrationHandler(BaseHandler): except Exception as e: logger.error("Failed to join new user to %r: %r", r, e) - @defer.inlineCallbacks - def post_consent_actions(self, user_id): + async def post_consent_actions(self, user_id): """A series of registration actions that can only be carried out once consent has been granted Args: user_id (str): The user to join """ - yield self._auto_join_rooms(user_id) + await self._auto_join_rooms(user_id) @defer.inlineCallbacks def appservice_register(self, user_localpart, as_token): @@ -394,14 +392,13 @@ class RegistrationHandler(BaseHandler): self._next_generated_user_id += 1 return str(id) - @defer.inlineCallbacks - def _join_user_to_room(self, requester, room_identifier): + async def _join_user_to_room(self, requester, room_identifier): room_member_handler = self.hs.get_room_member_handler() if RoomID.is_valid(room_identifier): room_id = room_identifier elif RoomAlias.is_valid(room_identifier): room_alias = RoomAlias.from_string(room_identifier) - room_id, remote_room_hosts = yield room_member_handler.lookup_room_alias( + room_id, remote_room_hosts = await room_member_handler.lookup_room_alias( room_alias ) room_id = room_id.to_string() @@ -410,7 +407,7 @@ class RegistrationHandler(BaseHandler): 400, "%s was not legal room ID or room alias" % (room_identifier,) ) - yield room_member_handler.update_membership( + await room_member_handler.update_membership( requester=requester, target=requester.user, room_id=room_id, @@ -550,8 +547,7 @@ class RegistrationHandler(BaseHandler): return (device_id, access_token) - @defer.inlineCallbacks - def post_registration_actions(self, user_id, auth_result, access_token): + async def post_registration_actions(self, user_id, auth_result, access_token): """A user has completed registration Args: @@ -562,7 +558,7 @@ class RegistrationHandler(BaseHandler): device, or None if `inhibit_login` enabled. """ if self.hs.config.worker_app: - yield self._post_registration_client( + await self._post_registration_client( user_id=user_id, auth_result=auth_result, access_token=access_token ) return @@ -574,19 +570,18 @@ class RegistrationHandler(BaseHandler): if is_threepid_reserved( self.hs.config.mau_limits_reserved_threepids, threepid ): - yield self.store.upsert_monthly_active_user(user_id) + await self.store.upsert_monthly_active_user(user_id) - yield self._register_email_threepid(user_id, threepid, access_token) + await self._register_email_threepid(user_id, threepid, access_token) if auth_result and LoginType.MSISDN in auth_result: threepid = auth_result[LoginType.MSISDN] - yield self._register_msisdn_threepid(user_id, threepid) + await self._register_msisdn_threepid(user_id, threepid) if auth_result and LoginType.TERMS in auth_result: - yield self._on_user_consented(user_id, self.hs.config.user_consent_version) + await self._on_user_consented(user_id, self.hs.config.user_consent_version) - @defer.inlineCallbacks - def _on_user_consented(self, user_id, consent_version): + async def _on_user_consented(self, user_id, consent_version): """A user consented to the terms on registration Args: @@ -595,8 +590,8 @@ class RegistrationHandler(BaseHandler): consented to. """ logger.info("%s has consented to the privacy policy", user_id) - yield self.store.user_set_consent_version(user_id, consent_version) - yield self.post_consent_actions(user_id) + await self.store.user_set_consent_version(user_id, consent_version) + await self.post_consent_actions(user_id) @defer.inlineCallbacks def _register_email_threepid(self, user_id, threepid, token): diff --git a/synapse/handlers/room.py b/synapse/handlers/room.py index 3d10e4b2d9..da12df7f53 100644 --- a/synapse/handlers/room.py +++ b/synapse/handlers/room.py @@ -148,17 +148,16 @@ class RoomCreationHandler(BaseHandler): return ret - @defer.inlineCallbacks - def _upgrade_room( + async def _upgrade_room( self, requester: Requester, old_room_id: str, new_version: RoomVersion ): user_id = requester.user.to_string() # start by allocating a new room id - r = yield self.store.get_room(old_room_id) + r = await self.store.get_room(old_room_id) if r is None: raise NotFoundError("Unknown room id %s" % (old_room_id,)) - new_room_id = yield self._generate_room_id( + new_room_id = await self._generate_room_id( creator_id=user_id, is_public=r["is_public"], room_version=new_version, ) @@ -169,7 +168,7 @@ class RoomCreationHandler(BaseHandler): ( tombstone_event, tombstone_context, - ) = yield self.event_creation_handler.create_event( + ) = await self.event_creation_handler.create_event( requester, { "type": EventTypes.Tombstone, @@ -183,12 +182,12 @@ class RoomCreationHandler(BaseHandler): }, token_id=requester.access_token_id, ) - old_room_version = yield self.store.get_room_version_id(old_room_id) - yield self.auth.check_from_context( + old_room_version = await self.store.get_room_version_id(old_room_id) + await self.auth.check_from_context( old_room_version, tombstone_event, tombstone_context ) - yield self.clone_existing_room( + await self.clone_existing_room( requester, old_room_id=old_room_id, new_room_id=new_room_id, @@ -197,32 +196,31 @@ class RoomCreationHandler(BaseHandler): ) # now send the tombstone - yield self.event_creation_handler.send_nonmember_event( + await self.event_creation_handler.send_nonmember_event( requester, tombstone_event, tombstone_context ) - old_room_state = yield tombstone_context.get_current_state_ids() + old_room_state = await tombstone_context.get_current_state_ids() # update any aliases - yield self._move_aliases_to_new_room( + await self._move_aliases_to_new_room( requester, old_room_id, new_room_id, old_room_state ) # Copy over user push rules, tags and migrate room directory state - yield self.room_member_handler.transfer_room_state_on_room_upgrade( + await self.room_member_handler.transfer_room_state_on_room_upgrade( old_room_id, new_room_id ) # finally, shut down the PLs in the old room, and update them in the new # room. - yield self._update_upgraded_room_pls( + await self._update_upgraded_room_pls( requester, old_room_id, new_room_id, old_room_state, ) return new_room_id - @defer.inlineCallbacks - def _update_upgraded_room_pls( + async def _update_upgraded_room_pls( self, requester: Requester, old_room_id: str, @@ -249,7 +247,7 @@ class RoomCreationHandler(BaseHandler): ) return - old_room_pl_state = yield self.store.get_event(old_room_pl_event_id) + old_room_pl_state = await self.store.get_event(old_room_pl_event_id) # we try to stop regular users from speaking by setting the PL required # to send regular events and invites to 'Moderator' level. That's normally @@ -278,7 +276,7 @@ class RoomCreationHandler(BaseHandler): if updated: try: - yield self.event_creation_handler.create_and_send_nonmember_event( + await self.event_creation_handler.create_and_send_nonmember_event( requester, { "type": EventTypes.PowerLevels, @@ -292,7 +290,7 @@ class RoomCreationHandler(BaseHandler): except AuthError as e: logger.warning("Unable to update PLs in old room: %s", e) - yield self.event_creation_handler.create_and_send_nonmember_event( + await self.event_creation_handler.create_and_send_nonmember_event( requester, { "type": EventTypes.PowerLevels, @@ -304,8 +302,7 @@ class RoomCreationHandler(BaseHandler): ratelimit=False, ) - @defer.inlineCallbacks - def clone_existing_room( + async def clone_existing_room( self, requester: Requester, old_room_id: str, @@ -338,7 +335,7 @@ class RoomCreationHandler(BaseHandler): # Check if old room was non-federatable # Get old room's create event - old_room_create_event = yield self.store.get_create_event_for_room(old_room_id) + old_room_create_event = await self.store.get_create_event_for_room(old_room_id) # Check if the create event specified a non-federatable room if not old_room_create_event.content.get("m.federate", True): @@ -361,11 +358,11 @@ class RoomCreationHandler(BaseHandler): (EventTypes.PowerLevels, ""), ) - old_room_state_ids = yield self.store.get_filtered_current_state_ids( + old_room_state_ids = await self.store.get_filtered_current_state_ids( old_room_id, StateFilter.from_types(types_to_copy) ) # map from event_id to BaseEvent - old_room_state_events = yield self.store.get_events(old_room_state_ids.values()) + old_room_state_events = await self.store.get_events(old_room_state_ids.values()) for k, old_event_id in iteritems(old_room_state_ids): old_event = old_room_state_events.get(old_event_id) @@ -400,7 +397,7 @@ class RoomCreationHandler(BaseHandler): if current_power_level < needed_power_level: power_levels["users"][user_id] = needed_power_level - yield self._send_events_for_new_room( + await self._send_events_for_new_room( requester, new_room_id, # we expect to override all the presets with initial_state, so this is @@ -412,12 +409,12 @@ class RoomCreationHandler(BaseHandler): ) # Transfer membership events - old_room_member_state_ids = yield self.store.get_filtered_current_state_ids( + old_room_member_state_ids = await self.store.get_filtered_current_state_ids( old_room_id, StateFilter.from_types([(EventTypes.Member, None)]) ) # map from event_id to BaseEvent - old_room_member_state_events = yield self.store.get_events( + old_room_member_state_events = await self.store.get_events( old_room_member_state_ids.values() ) for k, old_event in iteritems(old_room_member_state_events): @@ -426,7 +423,7 @@ class RoomCreationHandler(BaseHandler): "membership" in old_event.content and old_event.content["membership"] == "ban" ): - yield self.room_member_handler.update_membership( + await self.room_member_handler.update_membership( requester, UserID.from_string(old_event["state_key"]), new_room_id, @@ -438,8 +435,7 @@ class RoomCreationHandler(BaseHandler): # XXX invites/joins # XXX 3pid invites - @defer.inlineCallbacks - def _move_aliases_to_new_room( + async def _move_aliases_to_new_room( self, requester: Requester, old_room_id: str, @@ -448,13 +444,13 @@ class RoomCreationHandler(BaseHandler): ): directory_handler = self.hs.get_handlers().directory_handler - aliases = yield self.store.get_aliases_for_room(old_room_id) + aliases = await self.store.get_aliases_for_room(old_room_id) # check to see if we have a canonical alias. canonical_alias_event = None canonical_alias_event_id = old_room_state.get((EventTypes.CanonicalAlias, "")) if canonical_alias_event_id: - canonical_alias_event = yield self.store.get_event(canonical_alias_event_id) + canonical_alias_event = await self.store.get_event(canonical_alias_event_id) # first we try to remove the aliases from the old room (we suppress sending # the room_aliases event until the end). @@ -472,7 +468,7 @@ class RoomCreationHandler(BaseHandler): for alias_str in aliases: alias = RoomAlias.from_string(alias_str) try: - yield directory_handler.delete_association(requester, alias) + await directory_handler.delete_association(requester, alias) removed_aliases.append(alias_str) except SynapseError as e: logger.warning("Unable to remove alias %s from old room: %s", alias, e) @@ -485,7 +481,7 @@ class RoomCreationHandler(BaseHandler): # we can now add any aliases we successfully removed to the new room. for alias in removed_aliases: try: - yield directory_handler.create_association( + await directory_handler.create_association( requester, RoomAlias.from_string(alias), new_room_id, @@ -502,7 +498,7 @@ class RoomCreationHandler(BaseHandler): # alias event for the new room with a copy of the information. try: if canonical_alias_event: - yield self.event_creation_handler.create_and_send_nonmember_event( + await self.event_creation_handler.create_and_send_nonmember_event( requester, { "type": EventTypes.CanonicalAlias, @@ -518,8 +514,9 @@ class RoomCreationHandler(BaseHandler): # we returned the new room to the client at this point. logger.error("Unable to send updated alias events in new room: %s", e) - @defer.inlineCallbacks - def create_room(self, requester, config, ratelimit=True, creator_join_profile=None): + async def create_room( + self, requester, config, ratelimit=True, creator_join_profile=None + ): """ Creates a new room. Args: @@ -547,7 +544,7 @@ class RoomCreationHandler(BaseHandler): """ user_id = requester.user.to_string() - yield self.auth.check_auth_blocking(user_id) + await self.auth.check_auth_blocking(user_id) if ( self._server_notices_mxid is not None @@ -556,11 +553,11 @@ class RoomCreationHandler(BaseHandler): # allow the server notices mxid to create rooms is_requester_admin = True else: - is_requester_admin = yield self.auth.is_server_admin(requester.user) + is_requester_admin = await self.auth.is_server_admin(requester.user) # Check whether the third party rules allows/changes the room create # request. - event_allowed = yield self.third_party_event_rules.on_create_room( + event_allowed = await self.third_party_event_rules.on_create_room( requester, config, is_requester_admin=is_requester_admin ) if not event_allowed: @@ -574,7 +571,7 @@ class RoomCreationHandler(BaseHandler): raise SynapseError(403, "You are not permitted to create rooms") if ratelimit: - yield self.ratelimit(requester) + await self.ratelimit(requester) room_version_id = config.get( "room_version", self.config.default_room_version.identifier @@ -597,7 +594,7 @@ class RoomCreationHandler(BaseHandler): raise SynapseError(400, "Invalid characters in room alias") room_alias = RoomAlias(config["room_alias_name"], self.hs.hostname) - mapping = yield self.store.get_association_from_room_alias(room_alias) + mapping = await self.store.get_association_from_room_alias(room_alias) if mapping: raise SynapseError(400, "Room alias already taken", Codes.ROOM_IN_USE) @@ -612,7 +609,7 @@ class RoomCreationHandler(BaseHandler): except Exception: raise SynapseError(400, "Invalid user_id: %s" % (i,)) - yield self.event_creation_handler.assert_accepted_privacy_policy(requester) + await self.event_creation_handler.assert_accepted_privacy_policy(requester) power_level_content_override = config.get("power_level_content_override") if ( @@ -631,13 +628,13 @@ class RoomCreationHandler(BaseHandler): visibility = config.get("visibility", None) is_public = visibility == "public" - room_id = yield self._generate_room_id( + room_id = await self._generate_room_id( creator_id=user_id, is_public=is_public, room_version=room_version, ) directory_handler = self.hs.get_handlers().directory_handler if room_alias: - yield directory_handler.create_association( + await directory_handler.create_association( requester=requester, room_id=room_id, room_alias=room_alias, @@ -670,7 +667,7 @@ class RoomCreationHandler(BaseHandler): # override any attempt to set room versions via the creation_content creation_content["room_version"] = room_version.identifier - yield self._send_events_for_new_room( + await self._send_events_for_new_room( requester, room_id, preset_config=preset_config, @@ -684,7 +681,7 @@ class RoomCreationHandler(BaseHandler): if "name" in config: name = config["name"] - yield self.event_creation_handler.create_and_send_nonmember_event( + await self.event_creation_handler.create_and_send_nonmember_event( requester, { "type": EventTypes.Name, @@ -698,7 +695,7 @@ class RoomCreationHandler(BaseHandler): if "topic" in config: topic = config["topic"] - yield self.event_creation_handler.create_and_send_nonmember_event( + await self.event_creation_handler.create_and_send_nonmember_event( requester, { "type": EventTypes.Topic, @@ -716,7 +713,7 @@ class RoomCreationHandler(BaseHandler): if is_direct: content["is_direct"] = is_direct - yield self.room_member_handler.update_membership( + await self.room_member_handler.update_membership( requester, UserID.from_string(invitee), room_id, @@ -730,7 +727,7 @@ class RoomCreationHandler(BaseHandler): id_access_token = invite_3pid.get("id_access_token") # optional address = invite_3pid["address"] medium = invite_3pid["medium"] - yield self.hs.get_room_member_handler().do_3pid_invite( + await self.hs.get_room_member_handler().do_3pid_invite( room_id, requester.user, medium, @@ -748,8 +745,7 @@ class RoomCreationHandler(BaseHandler): return result - @defer.inlineCallbacks - def _send_events_for_new_room( + async def _send_events_for_new_room( self, creator, # A Requester object. room_id, @@ -769,11 +765,10 @@ class RoomCreationHandler(BaseHandler): return e - @defer.inlineCallbacks - def send(etype, content, **kwargs): + async def send(etype, content, **kwargs): event = create(etype, content, **kwargs) logger.debug("Sending %s in new room", etype) - yield self.event_creation_handler.create_and_send_nonmember_event( + await self.event_creation_handler.create_and_send_nonmember_event( creator, event, ratelimit=False ) @@ -784,10 +779,10 @@ class RoomCreationHandler(BaseHandler): event_keys = {"room_id": room_id, "sender": creator_id, "state_key": ""} creation_content.update({"creator": creator_id}) - yield send(etype=EventTypes.Create, content=creation_content) + await send(etype=EventTypes.Create, content=creation_content) logger.debug("Sending %s in new room", EventTypes.Member) - yield self.room_member_handler.update_membership( + await self.room_member_handler.update_membership( creator, creator.user, room_id, @@ -800,7 +795,7 @@ class RoomCreationHandler(BaseHandler): # of the first events that get sent into a room. pl_content = initial_state.pop((EventTypes.PowerLevels, ""), None) if pl_content is not None: - yield send(etype=EventTypes.PowerLevels, content=pl_content) + await send(etype=EventTypes.PowerLevels, content=pl_content) else: power_level_content = { "users": {creator_id: 100}, @@ -833,33 +828,33 @@ class RoomCreationHandler(BaseHandler): if power_level_content_override: power_level_content.update(power_level_content_override) - yield send(etype=EventTypes.PowerLevels, content=power_level_content) + await send(etype=EventTypes.PowerLevels, content=power_level_content) if room_alias and (EventTypes.CanonicalAlias, "") not in initial_state: - yield send( + await send( etype=EventTypes.CanonicalAlias, content={"alias": room_alias.to_string()}, ) if (EventTypes.JoinRules, "") not in initial_state: - yield send( + await send( etype=EventTypes.JoinRules, content={"join_rule": config["join_rules"]} ) if (EventTypes.RoomHistoryVisibility, "") not in initial_state: - yield send( + await send( etype=EventTypes.RoomHistoryVisibility, content={"history_visibility": config["history_visibility"]}, ) if config["guest_can_join"]: if (EventTypes.GuestAccess, "") not in initial_state: - yield send( + await send( etype=EventTypes.GuestAccess, content={"guest_access": "can_join"} ) for (etype, state_key), content in initial_state.items(): - yield send(etype=etype, state_key=state_key, content=content) + await send(etype=etype, state_key=state_key, content=content) @defer.inlineCallbacks def _generate_room_id( diff --git a/synapse/handlers/room_member.py b/synapse/handlers/room_member.py index c3ee8db4f0..53b49bc15f 100644 --- a/synapse/handlers/room_member.py +++ b/synapse/handlers/room_member.py @@ -142,8 +142,7 @@ class RoomMemberHandler(object): """ raise NotImplementedError() - @defer.inlineCallbacks - def _local_membership_update( + async def _local_membership_update( self, requester, target, @@ -164,7 +163,7 @@ class RoomMemberHandler(object): if requester.is_guest: content["kind"] = "guest" - event, context = yield self.event_creation_handler.create_event( + event, context = await self.event_creation_handler.create_event( requester, { "type": EventTypes.Member, @@ -182,18 +181,18 @@ class RoomMemberHandler(object): ) # Check if this event matches the previous membership event for the user. - duplicate = yield self.event_creation_handler.deduplicate_state_event( + duplicate = await self.event_creation_handler.deduplicate_state_event( event, context ) if duplicate is not None: # Discard the new event since this membership change is a no-op. return duplicate - yield self.event_creation_handler.handle_new_client_event( + await self.event_creation_handler.handle_new_client_event( requester, event, context, extra_users=[target], ratelimit=ratelimit ) - prev_state_ids = yield context.get_prev_state_ids() + prev_state_ids = await context.get_prev_state_ids() prev_member_event_id = prev_state_ids.get((EventTypes.Member, user_id), None) @@ -203,15 +202,15 @@ class RoomMemberHandler(object): # info. newly_joined = True if prev_member_event_id: - prev_member_event = yield self.store.get_event(prev_member_event_id) + prev_member_event = await self.store.get_event(prev_member_event_id) newly_joined = prev_member_event.membership != Membership.JOIN if newly_joined: - yield self._user_joined_room(target, room_id) + await self._user_joined_room(target, room_id) elif event.membership == Membership.LEAVE: if prev_member_event_id: - prev_member_event = yield self.store.get_event(prev_member_event_id) + prev_member_event = await self.store.get_event(prev_member_event_id) if prev_member_event.membership == Membership.JOIN: - yield self._user_left_room(target, room_id) + await self._user_left_room(target, room_id) return event @@ -253,8 +252,7 @@ class RoomMemberHandler(object): for tag, tag_content in room_tags.items(): yield self.store.add_tag_to_room(user_id, new_room_id, tag, tag_content) - @defer.inlineCallbacks - def update_membership( + async def update_membership( self, requester, target, @@ -269,8 +267,8 @@ class RoomMemberHandler(object): ): key = (room_id,) - with (yield self.member_linearizer.queue(key)): - result = yield self._update_membership( + with (await self.member_linearizer.queue(key)): + result = await self._update_membership( requester, target, room_id, @@ -285,8 +283,7 @@ class RoomMemberHandler(object): return result - @defer.inlineCallbacks - def _update_membership( + async def _update_membership( self, requester, target, @@ -321,7 +318,7 @@ class RoomMemberHandler(object): # if this is a join with a 3pid signature, we may need to turn a 3pid # invite into a normal invite before we can handle the join. if third_party_signed is not None: - yield self.federation_handler.exchange_third_party_invite( + await self.federation_handler.exchange_third_party_invite( third_party_signed["sender"], target.to_string(), room_id, @@ -332,7 +329,7 @@ class RoomMemberHandler(object): remote_room_hosts = [] if effective_membership_state not in ("leave", "ban"): - is_blocked = yield self.store.is_room_blocked(room_id) + is_blocked = await self.store.is_room_blocked(room_id) if is_blocked: raise SynapseError(403, "This room has been blocked on this server") @@ -351,7 +348,7 @@ class RoomMemberHandler(object): is_requester_admin = True else: - is_requester_admin = yield self.auth.is_server_admin(requester.user) + is_requester_admin = await self.auth.is_server_admin(requester.user) if not is_requester_admin: if self.config.block_non_admin_invites: @@ -370,9 +367,9 @@ class RoomMemberHandler(object): if block_invite: raise SynapseError(403, "Invites have been disabled on this server") - latest_event_ids = yield self.store.get_prev_events_for_room(room_id) + latest_event_ids = await self.store.get_prev_events_for_room(room_id) - current_state_ids = yield self.state_handler.get_current_state_ids( + current_state_ids = await self.state_handler.get_current_state_ids( room_id, latest_event_ids=latest_event_ids ) @@ -381,7 +378,7 @@ class RoomMemberHandler(object): # transitions and generic otherwise old_state_id = current_state_ids.get((EventTypes.Member, target.to_string())) if old_state_id: - old_state = yield self.store.get_event(old_state_id, allow_none=True) + old_state = await self.store.get_event(old_state_id, allow_none=True) old_membership = old_state.content.get("membership") if old_state else None if action == "unban" and old_membership != "ban": raise SynapseError( @@ -413,7 +410,7 @@ class RoomMemberHandler(object): old_membership == Membership.INVITE and effective_membership_state == Membership.LEAVE ): - is_blocked = yield self._is_server_notice_room(room_id) + is_blocked = await self._is_server_notice_room(room_id) if is_blocked: raise SynapseError( http_client.FORBIDDEN, @@ -424,18 +421,18 @@ class RoomMemberHandler(object): if action == "kick": raise AuthError(403, "The target user is not in the room") - is_host_in_room = yield self._is_host_in_room(current_state_ids) + is_host_in_room = await self._is_host_in_room(current_state_ids) if effective_membership_state == Membership.JOIN: if requester.is_guest: - guest_can_join = yield self._can_guest_join(current_state_ids) + guest_can_join = await self._can_guest_join(current_state_ids) if not guest_can_join: # This should be an auth check, but guests are a local concept, # so don't really fit into the general auth process. raise AuthError(403, "Guest access not allowed") if not is_host_in_room: - inviter = yield self._get_inviter(target.to_string(), room_id) + inviter = await self._get_inviter(target.to_string(), room_id) if inviter and not self.hs.is_mine(inviter): remote_room_hosts.append(inviter.domain) @@ -443,13 +440,13 @@ class RoomMemberHandler(object): profile = self.profile_handler if not content_specified: - content["displayname"] = yield profile.get_displayname(target) - content["avatar_url"] = yield profile.get_avatar_url(target) + content["displayname"] = await profile.get_displayname(target) + content["avatar_url"] = await profile.get_avatar_url(target) if requester.is_guest: content["kind"] = "guest" - remote_join_response = yield self._remote_join( + remote_join_response = await self._remote_join( requester, remote_room_hosts, room_id, target, content ) @@ -458,7 +455,7 @@ class RoomMemberHandler(object): elif effective_membership_state == Membership.LEAVE: if not is_host_in_room: # perhaps we've been invited - inviter = yield self._get_inviter(target.to_string(), room_id) + inviter = await self._get_inviter(target.to_string(), room_id) if not inviter: raise SynapseError(404, "Not a known room") @@ -472,12 +469,12 @@ class RoomMemberHandler(object): else: # send the rejection to the inviter's HS. remote_room_hosts = remote_room_hosts + [inviter.domain] - res = yield self._remote_reject_invite( + res = await self._remote_reject_invite( requester, remote_room_hosts, room_id, target, content, ) return res - res = yield self._local_membership_update( + res = await self._local_membership_update( requester=requester, target=target, room_id=room_id, @@ -572,8 +569,7 @@ class RoomMemberHandler(object): ) continue - @defer.inlineCallbacks - def send_membership_event(self, requester, event, context, ratelimit=True): + async def send_membership_event(self, requester, event, context, ratelimit=True): """ Change the membership status of a user in a room. @@ -599,27 +595,27 @@ class RoomMemberHandler(object): else: requester = types.create_requester(target_user) - prev_event = yield self.event_creation_handler.deduplicate_state_event( + prev_event = await self.event_creation_handler.deduplicate_state_event( event, context ) if prev_event is not None: return - prev_state_ids = yield context.get_prev_state_ids() + prev_state_ids = await context.get_prev_state_ids() if event.membership == Membership.JOIN: if requester.is_guest: - guest_can_join = yield self._can_guest_join(prev_state_ids) + guest_can_join = await self._can_guest_join(prev_state_ids) if not guest_can_join: # This should be an auth check, but guests are a local concept, # so don't really fit into the general auth process. raise AuthError(403, "Guest access not allowed") if event.membership not in (Membership.LEAVE, Membership.BAN): - is_blocked = yield self.store.is_room_blocked(room_id) + is_blocked = await self.store.is_room_blocked(room_id) if is_blocked: raise SynapseError(403, "This room has been blocked on this server") - yield self.event_creation_handler.handle_new_client_event( + await self.event_creation_handler.handle_new_client_event( requester, event, context, extra_users=[target_user], ratelimit=ratelimit ) @@ -633,15 +629,15 @@ class RoomMemberHandler(object): # info. newly_joined = True if prev_member_event_id: - prev_member_event = yield self.store.get_event(prev_member_event_id) + prev_member_event = await self.store.get_event(prev_member_event_id) newly_joined = prev_member_event.membership != Membership.JOIN if newly_joined: - yield self._user_joined_room(target_user, room_id) + await self._user_joined_room(target_user, room_id) elif event.membership == Membership.LEAVE: if prev_member_event_id: - prev_member_event = yield self.store.get_event(prev_member_event_id) + prev_member_event = await self.store.get_event(prev_member_event_id) if prev_member_event.membership == Membership.JOIN: - yield self._user_left_room(target_user, room_id) + await self._user_left_room(target_user, room_id) @defer.inlineCallbacks def _can_guest_join(self, current_state_ids): @@ -699,8 +695,7 @@ class RoomMemberHandler(object): if invite: return UserID.from_string(invite.sender) - @defer.inlineCallbacks - def do_3pid_invite( + async def do_3pid_invite( self, room_id, inviter, @@ -712,7 +707,7 @@ class RoomMemberHandler(object): id_access_token=None, ): if self.config.block_non_admin_invites: - is_requester_admin = yield self.auth.is_server_admin(requester.user) + is_requester_admin = await self.auth.is_server_admin(requester.user) if not is_requester_admin: raise SynapseError( 403, "Invites have been disabled on this server", Codes.FORBIDDEN @@ -720,9 +715,9 @@ class RoomMemberHandler(object): # We need to rate limit *before* we send out any 3PID invites, so we # can't just rely on the standard ratelimiting of events. - yield self.base_handler.ratelimit(requester) + await self.base_handler.ratelimit(requester) - can_invite = yield self.third_party_event_rules.check_threepid_can_be_invited( + can_invite = await self.third_party_event_rules.check_threepid_can_be_invited( medium, address, room_id ) if not can_invite: @@ -737,16 +732,16 @@ class RoomMemberHandler(object): 403, "Looking up third-party identifiers is denied from this server" ) - invitee = yield self.identity_handler.lookup_3pid( + invitee = await self.identity_handler.lookup_3pid( id_server, medium, address, id_access_token ) if invitee: - yield self.update_membership( + await self.update_membership( requester, UserID.from_string(invitee), room_id, "invite", txn_id=txn_id ) else: - yield self._make_and_store_3pid_invite( + await self._make_and_store_3pid_invite( requester, id_server, medium, @@ -757,8 +752,7 @@ class RoomMemberHandler(object): id_access_token=id_access_token, ) - @defer.inlineCallbacks - def _make_and_store_3pid_invite( + async def _make_and_store_3pid_invite( self, requester, id_server, @@ -769,7 +763,7 @@ class RoomMemberHandler(object): txn_id, id_access_token=None, ): - room_state = yield self.state_handler.get_current_state(room_id) + room_state = await self.state_handler.get_current_state(room_id) inviter_display_name = "" inviter_avatar_url = "" @@ -807,7 +801,7 @@ class RoomMemberHandler(object): public_keys, fallback_public_key, display_name, - ) = yield self.identity_handler.ask_id_server_for_third_party_invite( + ) = await self.identity_handler.ask_id_server_for_third_party_invite( requester=requester, id_server=id_server, medium=medium, @@ -823,7 +817,7 @@ class RoomMemberHandler(object): id_access_token=id_access_token, ) - yield self.event_creation_handler.create_and_send_nonmember_event( + await self.event_creation_handler.create_and_send_nonmember_event( requester, { "type": EventTypes.ThirdPartyInvite, @@ -917,8 +911,7 @@ class RoomMemberMasterHandler(RoomMemberHandler): return complexity["v1"] > max_complexity - @defer.inlineCallbacks - def _remote_join(self, requester, remote_room_hosts, room_id, user, content): + async def _remote_join(self, requester, remote_room_hosts, room_id, user, content): """Implements RoomMemberHandler._remote_join """ # filter ourselves out of remote_room_hosts: do_invite_join ignores it @@ -933,7 +926,7 @@ class RoomMemberMasterHandler(RoomMemberHandler): if self.hs.config.limit_remote_rooms.enabled: # Fetch the room complexity - too_complex = yield self._is_remote_room_too_complex( + too_complex = await self._is_remote_room_too_complex( room_id, remote_room_hosts ) if too_complex is True: @@ -947,12 +940,10 @@ class RoomMemberMasterHandler(RoomMemberHandler): # join dance for now, since we're kinda implicitly checking # that we are allowed to join when we decide whether or not we # need to do the invite/join dance. - yield defer.ensureDeferred( - self.federation_handler.do_invite_join( - remote_room_hosts, room_id, user.to_string(), content - ) + await self.federation_handler.do_invite_join( + remote_room_hosts, room_id, user.to_string(), content ) - yield self._user_joined_room(user, room_id) + await self._user_joined_room(user, room_id) # Check the room we just joined wasn't too large, if we didn't fetch the # complexity of it before. @@ -962,7 +953,7 @@ class RoomMemberMasterHandler(RoomMemberHandler): return # Check again, but with the local state events - too_complex = yield self._is_local_room_too_complex(room_id) + too_complex = await self._is_local_room_too_complex(room_id) if too_complex is False: # We're under the limit. @@ -970,7 +961,7 @@ class RoomMemberMasterHandler(RoomMemberHandler): # The room is too large. Leave. requester = types.create_requester(user, None, False, None) - yield self.update_membership( + await self.update_membership( requester=requester, target=user, room_id=room_id, action="leave" ) raise SynapseError( @@ -1008,12 +999,12 @@ class RoomMemberMasterHandler(RoomMemberHandler): def _user_joined_room(self, target, room_id): """Implements RoomMemberHandler._user_joined_room """ - return user_joined_room(self.distributor, target, room_id) + return defer.succeed(user_joined_room(self.distributor, target, room_id)) def _user_left_room(self, target, room_id): """Implements RoomMemberHandler._user_left_room """ - return user_left_room(self.distributor, target, room_id) + return defer.succeed(user_left_room(self.distributor, target, room_id)) @defer.inlineCallbacks def forget(self, user, room_id): diff --git a/synapse/server_notices/consent_server_notices.py b/synapse/server_notices/consent_server_notices.py index 5736c56032..3bf330da49 100644 --- a/synapse/server_notices/consent_server_notices.py +++ b/synapse/server_notices/consent_server_notices.py @@ -16,8 +16,6 @@ import logging from six import iteritems, string_types -from twisted.internet import defer - from synapse.api.errors import SynapseError from synapse.api.urls import ConsentURIBuilder from synapse.config import ConfigError @@ -59,8 +57,7 @@ class ConsentServerNotices(object): self._consent_uri_builder = ConsentURIBuilder(hs.config) - @defer.inlineCallbacks - def maybe_send_server_notice_to_user(self, user_id): + async def maybe_send_server_notice_to_user(self, user_id): """Check if we need to send a notice to this user, and does so if so Args: @@ -78,7 +75,7 @@ class ConsentServerNotices(object): return self._users_in_progress.add(user_id) try: - u = yield self._store.get_user_by_id(user_id) + u = await self._store.get_user_by_id(user_id) if u["is_guest"] and not self._send_to_guests: # don't send to guests @@ -100,8 +97,8 @@ class ConsentServerNotices(object): content = copy_with_str_subst( self._server_notice_content, {"consent_uri": consent_uri} ) - yield self._server_notices_manager.send_notice(user_id, content) - yield self._store.user_set_consent_server_notice_sent( + await self._server_notices_manager.send_notice(user_id, content) + await self._store.user_set_consent_server_notice_sent( user_id, self._current_consent_version ) except SynapseError as e: diff --git a/synapse/server_notices/resource_limits_server_notices.py b/synapse/server_notices/resource_limits_server_notices.py index ce4a828894..971771b8b2 100644 --- a/synapse/server_notices/resource_limits_server_notices.py +++ b/synapse/server_notices/resource_limits_server_notices.py @@ -50,8 +50,7 @@ class ResourceLimitsServerNotices(object): self._notifier = hs.get_notifier() - @defer.inlineCallbacks - def maybe_send_server_notice_to_user(self, user_id): + async def maybe_send_server_notice_to_user(self, user_id): """Check if we need to send a notice to this user, this will be true in two cases. 1. The server has reached its limit does not reflect this @@ -74,13 +73,13 @@ class ResourceLimitsServerNotices(object): # Don't try and send server notices unless they've been enabled return - timestamp = yield self._store.user_last_seen_monthly_active(user_id) + timestamp = await self._store.user_last_seen_monthly_active(user_id) if timestamp is None: # This user will be blocked from receiving the notice anyway. # In practice, not sure we can ever get here return - room_id = yield self._server_notices_manager.get_or_create_notice_room_for_user( + room_id = await self._server_notices_manager.get_or_create_notice_room_for_user( user_id ) @@ -88,10 +87,10 @@ class ResourceLimitsServerNotices(object): logger.warning("Failed to get server notices room") return - yield self._check_and_set_tags(user_id, room_id) + await self._check_and_set_tags(user_id, room_id) # Determine current state of room - currently_blocked, ref_events = yield self._is_room_currently_blocked(room_id) + currently_blocked, ref_events = await self._is_room_currently_blocked(room_id) limit_msg = None limit_type = None @@ -99,7 +98,7 @@ class ResourceLimitsServerNotices(object): # Normally should always pass in user_id to check_auth_blocking # if you have it, but in this case are checking what would happen # to other users if they were to arrive. - yield self._auth.check_auth_blocking() + await self._auth.check_auth_blocking() except ResourceLimitError as e: limit_msg = e.msg limit_type = e.limit_type @@ -112,22 +111,21 @@ class ResourceLimitsServerNotices(object): # We have hit the MAU limit, but MAU alerting is disabled: # reset room if necessary and return if currently_blocked: - self._remove_limit_block_notification(user_id, ref_events) + await self._remove_limit_block_notification(user_id, ref_events) return if currently_blocked and not limit_msg: # Room is notifying of a block, when it ought not to be. - yield self._remove_limit_block_notification(user_id, ref_events) + await self._remove_limit_block_notification(user_id, ref_events) elif not currently_blocked and limit_msg: # Room is not notifying of a block, when it ought to be. - yield self._apply_limit_block_notification( + await self._apply_limit_block_notification( user_id, limit_msg, limit_type ) except SynapseError as e: logger.error("Error sending resource limits server notice: %s", e) - @defer.inlineCallbacks - def _remove_limit_block_notification(self, user_id, ref_events): + async def _remove_limit_block_notification(self, user_id, ref_events): """Utility method to remove limit block notifications from the server notices room. @@ -137,12 +135,13 @@ class ResourceLimitsServerNotices(object): limit blocking and need to be preserved. """ content = {"pinned": ref_events} - yield self._server_notices_manager.send_notice( + await self._server_notices_manager.send_notice( user_id, content, EventTypes.Pinned, "" ) - @defer.inlineCallbacks - def _apply_limit_block_notification(self, user_id, event_body, event_limit_type): + async def _apply_limit_block_notification( + self, user_id, event_body, event_limit_type + ): """Utility method to apply limit block notifications in the server notices room. @@ -159,12 +158,12 @@ class ResourceLimitsServerNotices(object): "admin_contact": self._config.admin_contact, "limit_type": event_limit_type, } - event = yield self._server_notices_manager.send_notice( + event = await self._server_notices_manager.send_notice( user_id, content, EventTypes.Message ) content = {"pinned": [event.event_id]} - yield self._server_notices_manager.send_notice( + await self._server_notices_manager.send_notice( user_id, content, EventTypes.Pinned, "" ) @@ -198,7 +197,7 @@ class ResourceLimitsServerNotices(object): room_id(str): The room id of the server notices room Returns: - + Deferred[Tuple[bool, List]]: bool: Is the room currently blocked list: The list of pinned events that are unrelated to limit blocking This list can be used as a convenience in the case where the block diff --git a/synapse/server_notices/server_notices_manager.py b/synapse/server_notices/server_notices_manager.py index bf0943f265..999c621b92 100644 --- a/synapse/server_notices/server_notices_manager.py +++ b/synapse/server_notices/server_notices_manager.py @@ -14,11 +14,9 @@ # limitations under the License. import logging -from twisted.internet import defer - from synapse.api.constants import EventTypes, Membership, RoomCreationPreset from synapse.types import UserID, create_requester -from synapse.util.caches.descriptors import cachedInlineCallbacks +from synapse.util.caches.descriptors import cached logger = logging.getLogger(__name__) @@ -51,8 +49,7 @@ class ServerNoticesManager(object): """ return self._config.server_notices_mxid is not None - @defer.inlineCallbacks - def send_notice( + async def send_notice( self, user_id, event_content, type=EventTypes.Message, state_key=None ): """Send a notice to the given user @@ -68,8 +65,8 @@ class ServerNoticesManager(object): Returns: Deferred[FrozenEvent] """ - room_id = yield self.get_or_create_notice_room_for_user(user_id) - yield self.maybe_invite_user_to_room(user_id, room_id) + room_id = await self.get_or_create_notice_room_for_user(user_id) + await self.maybe_invite_user_to_room(user_id, room_id) system_mxid = self._config.server_notices_mxid requester = create_requester(system_mxid) @@ -86,13 +83,13 @@ class ServerNoticesManager(object): if state_key is not None: event_dict["state_key"] = state_key - res = yield self._event_creation_handler.create_and_send_nonmember_event( + res = await self._event_creation_handler.create_and_send_nonmember_event( requester, event_dict, ratelimit=False ) return res - @cachedInlineCallbacks() - def get_or_create_notice_room_for_user(self, user_id): + @cached() + async def get_or_create_notice_room_for_user(self, user_id): """Get the room for notices for a given user If we have not yet created a notice room for this user, create it, but don't @@ -109,7 +106,7 @@ class ServerNoticesManager(object): assert self._is_mine_id(user_id), "Cannot send server notices to remote users" - rooms = yield self._store.get_rooms_for_local_user_where_membership_is( + rooms = await self._store.get_rooms_for_local_user_where_membership_is( user_id, [Membership.INVITE, Membership.JOIN] ) for room in rooms: @@ -118,7 +115,7 @@ class ServerNoticesManager(object): # be joined. This is kinda deliberate, in that if somebody somehow # manages to invite the system user to a room, that doesn't make it # the server notices room. - user_ids = yield self._store.get_users_in_room(room.room_id) + user_ids = await self._store.get_users_in_room(room.room_id) if self.server_notices_mxid in user_ids: # we found a room which our user shares with the system notice # user @@ -146,7 +143,7 @@ class ServerNoticesManager(object): } requester = create_requester(self.server_notices_mxid) - info = yield self._room_creation_handler.create_room( + info = await self._room_creation_handler.create_room( requester, config={ "preset": RoomCreationPreset.PRIVATE_CHAT, @@ -158,7 +155,7 @@ class ServerNoticesManager(object): ) room_id = info["room_id"] - max_id = yield self._store.add_tag_to_room( + max_id = await self._store.add_tag_to_room( user_id, room_id, SERVER_NOTICE_ROOM_TAG, {} ) self._notifier.on_new_event("account_data_key", max_id, users=[user_id]) @@ -166,8 +163,7 @@ class ServerNoticesManager(object): logger.info("Created server notices room %s for %s", room_id, user_id) return room_id - @defer.inlineCallbacks - def maybe_invite_user_to_room(self, user_id: str, room_id: str): + async def maybe_invite_user_to_room(self, user_id: str, room_id: str): """Invite the given user to the given server room, unless the user has already joined or been invited to it. @@ -179,14 +175,14 @@ class ServerNoticesManager(object): # Check whether the user has already joined or been invited to this room. If # that's the case, there is no need to re-invite them. - joined_rooms = yield self._store.get_rooms_for_local_user_where_membership_is( + joined_rooms = await self._store.get_rooms_for_local_user_where_membership_is( user_id, [Membership.INVITE, Membership.JOIN] ) for room in joined_rooms: if room.room_id == room_id: return - yield self._room_member_handler.update_membership( + await self._room_member_handler.update_membership( requester=requester, target=UserID.from_string(user_id), room_id=room_id, diff --git a/synapse/server_notices/server_notices_sender.py b/synapse/server_notices/server_notices_sender.py index 652bab58e3..be74e86641 100644 --- a/synapse/server_notices/server_notices_sender.py +++ b/synapse/server_notices/server_notices_sender.py @@ -12,8 +12,6 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. -from twisted.internet import defer - from synapse.server_notices.consent_server_notices import ConsentServerNotices from synapse.server_notices.resource_limits_server_notices import ( ResourceLimitsServerNotices, @@ -36,18 +34,16 @@ class ServerNoticesSender(object): ResourceLimitsServerNotices(hs), ) - @defer.inlineCallbacks - def on_user_syncing(self, user_id): + async def on_user_syncing(self, user_id): """Called when the user performs a sync operation. Args: user_id (str): mxid of user who synced """ for sn in self._server_notices: - yield sn.maybe_send_server_notice_to_user(user_id) + await sn.maybe_send_server_notice_to_user(user_id) - @defer.inlineCallbacks - def on_user_ip(self, user_id): + async def on_user_ip(self, user_id): """Called on the master when a worker process saw a client request. Args: @@ -57,4 +53,4 @@ class ServerNoticesSender(object): # we check for notices to send to the user in on_user_ip as well as # in on_user_syncing for sn in self._server_notices: - yield sn.maybe_send_server_notice_to_user(user_id) + await sn.maybe_send_server_notice_to_user(user_id) diff --git a/synapse/storage/data_stores/main/registration.py b/synapse/storage/data_stores/main/registration.py index 3e53c8568a..efcdd2100b 100644 --- a/synapse/storage/data_stores/main/registration.py +++ b/synapse/storage/data_stores/main/registration.py @@ -273,8 +273,7 @@ class RegistrationWorkerStore(SQLBaseStore): desc="delete_account_validity_for_user", ) - @defer.inlineCallbacks - def is_server_admin(self, user): + async def is_server_admin(self, user): """Determines if a user is an admin of this homeserver. Args: @@ -283,7 +282,7 @@ class RegistrationWorkerStore(SQLBaseStore): Returns (bool): true iff the user is a server admin, false otherwise. """ - res = yield self.db.simple_select_one_onecol( + res = await self.db.simple_select_one_onecol( table="users", keyvalues={"name": user.to_string()}, retcol="admin", diff --git a/tests/handlers/test_profile.py b/tests/handlers/test_profile.py index be665262c6..8aa56f1496 100644 --- a/tests/handlers/test_profile.py +++ b/tests/handlers/test_profile.py @@ -82,18 +82,26 @@ class ProfileTestCase(unittest.TestCase): @defer.inlineCallbacks def test_set_my_name(self): - yield self.handler.set_displayname( - self.frank, synapse.types.create_requester(self.frank), "Frank Jr." + yield defer.ensureDeferred( + self.handler.set_displayname( + self.frank, synapse.types.create_requester(self.frank), "Frank Jr." + ) ) self.assertEquals( - (yield self.store.get_profile_displayname(self.frank.localpart)), + ( + yield defer.ensureDeferred( + self.store.get_profile_displayname(self.frank.localpart) + ) + ), "Frank Jr.", ) # Set displayname again - yield self.handler.set_displayname( - self.frank, synapse.types.create_requester(self.frank), "Frank" + yield defer.ensureDeferred( + self.handler.set_displayname( + self.frank, synapse.types.create_requester(self.frank), "Frank" + ) ) self.assertEquals( @@ -112,16 +120,20 @@ class ProfileTestCase(unittest.TestCase): ) # Setting displayname a second time is forbidden - d = self.handler.set_displayname( - self.frank, synapse.types.create_requester(self.frank), "Frank Jr." + d = defer.ensureDeferred( + self.handler.set_displayname( + self.frank, synapse.types.create_requester(self.frank), "Frank Jr." + ) ) yield self.assertFailure(d, SynapseError) @defer.inlineCallbacks def test_set_my_name_noauth(self): - d = self.handler.set_displayname( - self.frank, synapse.types.create_requester(self.bob), "Frank Jr." + d = defer.ensureDeferred( + self.handler.set_displayname( + self.frank, synapse.types.create_requester(self.bob), "Frank Jr." + ) ) yield self.assertFailure(d, AuthError) @@ -165,10 +177,12 @@ class ProfileTestCase(unittest.TestCase): @defer.inlineCallbacks def test_set_my_avatar(self): - yield self.handler.set_avatar_url( - self.frank, - synapse.types.create_requester(self.frank), - "http://my.server/pic.gif", + yield defer.ensureDeferred( + self.handler.set_avatar_url( + self.frank, + synapse.types.create_requester(self.frank), + "http://my.server/pic.gif", + ) ) self.assertEquals( @@ -177,10 +191,12 @@ class ProfileTestCase(unittest.TestCase): ) # Set avatar again - yield self.handler.set_avatar_url( - self.frank, - synapse.types.create_requester(self.frank), - "http://my.server/me.png", + yield defer.ensureDeferred( + self.handler.set_avatar_url( + self.frank, + synapse.types.create_requester(self.frank), + "http://my.server/me.png", + ) ) self.assertEquals( @@ -203,10 +219,12 @@ class ProfileTestCase(unittest.TestCase): ) # Set avatar a second time is forbidden - d = self.handler.set_avatar_url( - self.frank, - synapse.types.create_requester(self.frank), - "http://my.server/pic.gif", + d = defer.ensureDeferred( + self.handler.set_avatar_url( + self.frank, + synapse.types.create_requester(self.frank), + "http://my.server/pic.gif", + ) ) yield self.assertFailure(d, SynapseError) diff --git a/tests/handlers/test_register.py b/tests/handlers/test_register.py index f1dc51d6c9..1b7935cef2 100644 --- a/tests/handlers/test_register.py +++ b/tests/handlers/test_register.py @@ -175,7 +175,7 @@ class RegistrationTestCase(unittest.HomeserverTestCase): room_alias_str = "#room:test" self.hs.config.auto_join_rooms = [room_alias_str] - self.store.is_real_user = Mock(return_value=False) + self.store.is_real_user = Mock(return_value=defer.succeed(False)) user_id = self.get_success(self.handler.register_user(localpart="support")) rooms = self.get_success(self.store.get_rooms_for_user(user_id)) self.assertEqual(len(rooms), 0) @@ -187,8 +187,8 @@ class RegistrationTestCase(unittest.HomeserverTestCase): room_alias_str = "#room:test" self.hs.config.auto_join_rooms = [room_alias_str] - self.store.count_real_users = Mock(return_value=1) - self.store.is_real_user = Mock(return_value=True) + self.store.count_real_users = Mock(return_value=defer.succeed(1)) + self.store.is_real_user = Mock(return_value=defer.succeed(True)) user_id = self.get_success(self.handler.register_user(localpart="real")) rooms = self.get_success(self.store.get_rooms_for_user(user_id)) directory_handler = self.hs.get_handlers().directory_handler @@ -202,8 +202,8 @@ class RegistrationTestCase(unittest.HomeserverTestCase): room_alias_str = "#room:test" self.hs.config.auto_join_rooms = [room_alias_str] - self.store.count_real_users = Mock(return_value=2) - self.store.is_real_user = Mock(return_value=True) + self.store.count_real_users = Mock(return_value=defer.succeed(2)) + self.store.is_real_user = Mock(return_value=defer.succeed(True)) user_id = self.get_success(self.handler.register_user(localpart="real")) rooms = self.get_success(self.store.get_rooms_for_user(user_id)) self.assertEqual(len(rooms), 0) @@ -256,8 +256,9 @@ class RegistrationTestCase(unittest.HomeserverTestCase): self.handler.register_user(localpart=invalid_user_id), SynapseError ) - @defer.inlineCallbacks - def get_or_create_user(self, requester, localpart, displayname, password_hash=None): + async def get_or_create_user( + self, requester, localpart, displayname, password_hash=None + ): """Creates a new user if the user does not exist, else revokes all previous access tokens and generates a new one. @@ -272,11 +273,11 @@ class RegistrationTestCase(unittest.HomeserverTestCase): """ if localpart is None: raise SynapseError(400, "Request must include user id") - yield self.hs.get_auth().check_auth_blocking() + await self.hs.get_auth().check_auth_blocking() need_register = True try: - yield self.handler.check_username(localpart) + await self.handler.check_username(localpart) except SynapseError as e: if e.errcode == Codes.USER_IN_USE: need_register = False @@ -288,23 +289,21 @@ class RegistrationTestCase(unittest.HomeserverTestCase): token = self.macaroon_generator.generate_access_token(user_id) if need_register: - yield self.handler.register_with_store( + await self.handler.register_with_store( user_id=user_id, password_hash=password_hash, create_profile_with_displayname=user.localpart, ) else: - yield defer.ensureDeferred( - self.hs.get_auth_handler().delete_access_tokens_for_user(user_id) - ) + await self.hs.get_auth_handler().delete_access_tokens_for_user(user_id) - yield self.store.add_access_token_to_user( + await self.store.add_access_token_to_user( user_id=user_id, token=token, device_id=None, valid_until_ms=None ) if displayname is not None: # logger.info("setting user display name: %s -> %s", user_id, displayname) - yield self.hs.get_profile_handler().set_displayname( + await self.hs.get_profile_handler().set_displayname( user, requester, displayname, by_admin=True ) diff --git a/tests/server_notices/test_resource_limits_server_notices.py b/tests/server_notices/test_resource_limits_server_notices.py index 93eb053b8c..987addad9b 100644 --- a/tests/server_notices/test_resource_limits_server_notices.py +++ b/tests/server_notices/test_resource_limits_server_notices.py @@ -55,25 +55,18 @@ class TestResourceLimitsServerNotices(unittest.HomeserverTestCase): self._rlsn._store.user_last_seen_monthly_active = Mock( return_value=defer.succeed(1000) ) - self._send_notice = self._rlsn._server_notices_manager.send_notice - self._rlsn._server_notices_manager.send_notice = Mock() - self._rlsn._state.get_current_state = Mock(return_value=defer.succeed(None)) - self._rlsn._store.get_events = Mock(return_value=defer.succeed({})) - + self._rlsn._server_notices_manager.send_notice = Mock( + return_value=defer.succeed(Mock()) + ) self._send_notice = self._rlsn._server_notices_manager.send_notice self.hs.config.limit_usage_by_mau = True self.user_id = "@user_id:test" - # self.server_notices_mxid = "@server:test" - # self.server_notices_mxid_display_name = None - # self.server_notices_mxid_avatar_url = None - # self.server_notices_room_name = "Server Notices" - self._rlsn._server_notices_manager.get_or_create_notice_room_for_user = Mock( - returnValue="" + return_value=defer.succeed("!something:localhost") ) - self._rlsn._store.add_tag_to_room = Mock() + self._rlsn._store.add_tag_to_room = Mock(return_value=defer.succeed(None)) self._rlsn._store.get_tags_for_room = Mock(return_value={}) self.hs.config.admin_contact = "mailto:user@test.com" @@ -95,14 +88,13 @@ class TestResourceLimitsServerNotices(unittest.HomeserverTestCase): def test_maybe_send_server_notice_to_user_remove_blocked_notice(self): """Test when user has blocked notice, but should have it removed""" - self._rlsn._auth.check_auth_blocking = Mock() + self._rlsn._auth.check_auth_blocking = Mock(return_value=defer.succeed(None)) mock_event = Mock( type=EventTypes.Message, content={"msgtype": ServerNoticeMsgType} ) self._rlsn._store.get_events = Mock( return_value=defer.succeed({"123": mock_event}) ) - self.get_success(self._rlsn.maybe_send_server_notice_to_user(self.user_id)) # Would be better to check the content, but once == remove blocking event self._send_notice.assert_called_once() @@ -112,7 +104,7 @@ class TestResourceLimitsServerNotices(unittest.HomeserverTestCase): Test when user has blocked notice, but notice ought to be there (NOOP) """ self._rlsn._auth.check_auth_blocking = Mock( - side_effect=ResourceLimitError(403, "foo") + return_value=defer.succeed(None), side_effect=ResourceLimitError(403, "foo") ) mock_event = Mock( @@ -121,6 +113,7 @@ class TestResourceLimitsServerNotices(unittest.HomeserverTestCase): self._rlsn._store.get_events = Mock( return_value=defer.succeed({"123": mock_event}) ) + self.get_success(self._rlsn.maybe_send_server_notice_to_user(self.user_id)) self._send_notice.assert_not_called() @@ -129,9 +122,8 @@ class TestResourceLimitsServerNotices(unittest.HomeserverTestCase): """ Test when user does not have blocked notice, but should have one """ - self._rlsn._auth.check_auth_blocking = Mock( - side_effect=ResourceLimitError(403, "foo") + return_value=defer.succeed(None), side_effect=ResourceLimitError(403, "foo") ) self.get_success(self._rlsn.maybe_send_server_notice_to_user(self.user_id)) @@ -142,7 +134,7 @@ class TestResourceLimitsServerNotices(unittest.HomeserverTestCase): """ Test when user does not have blocked notice, nor should they (NOOP) """ - self._rlsn._auth.check_auth_blocking = Mock() + self._rlsn._auth.check_auth_blocking = Mock(return_value=defer.succeed(None)) self.get_success(self._rlsn.maybe_send_server_notice_to_user(self.user_id)) @@ -153,7 +145,7 @@ class TestResourceLimitsServerNotices(unittest.HomeserverTestCase): Test when user is not part of the MAU cohort - this should not ever happen - but ... """ - self._rlsn._auth.check_auth_blocking = Mock() + self._rlsn._auth.check_auth_blocking = Mock(return_value=defer.succeed(None)) self._rlsn._store.user_last_seen_monthly_active = Mock( return_value=defer.succeed(None) ) @@ -167,24 +159,28 @@ class TestResourceLimitsServerNotices(unittest.HomeserverTestCase): an alert message is not sent into the room """ self.hs.config.mau_limit_alerting = False + self._rlsn._auth.check_auth_blocking = Mock( + return_value=defer.succeed(None), side_effect=ResourceLimitError( 403, "foo", limit_type=LimitBlockingTypes.MONTHLY_ACTIVE_USER - ) + ), ) self.get_success(self._rlsn.maybe_send_server_notice_to_user(self.user_id)) - self.assertTrue(self._send_notice.call_count == 0) + self.assertEqual(self._send_notice.call_count, 0) def test_check_hs_disabled_unaffected_by_mau_alert_suppression(self): """ Test that when a server is disabled, that MAU limit alerting is ignored. """ self.hs.config.mau_limit_alerting = False + self._rlsn._auth.check_auth_blocking = Mock( + return_value=defer.succeed(None), side_effect=ResourceLimitError( 403, "foo", limit_type=LimitBlockingTypes.HS_DISABLED - ) + ), ) self.get_success(self._rlsn.maybe_send_server_notice_to_user(self.user_id)) @@ -198,10 +194,12 @@ class TestResourceLimitsServerNotices(unittest.HomeserverTestCase): """ self.hs.config.mau_limit_alerting = False self._rlsn._auth.check_auth_blocking = Mock( + return_value=defer.succeed(None), side_effect=ResourceLimitError( 403, "foo", limit_type=LimitBlockingTypes.MONTHLY_ACTIVE_USER - ) + ), ) + self._rlsn._server_notices_manager.__is_room_currently_blocked = Mock( return_value=defer.succeed((True, [])) ) @@ -256,7 +254,9 @@ class TestResourceLimitsServerNoticesWithRealRooms(unittest.HomeserverTestCase): def test_server_notice_only_sent_once(self): self.store.get_monthly_active_count = Mock(return_value=1000) - self.store.user_last_seen_monthly_active = Mock(return_value=1000) + self.store.user_last_seen_monthly_active = Mock( + return_value=defer.succeed(1000) + ) # Call the function multiple times to ensure we only send the notice once self.get_success(self._rlsn.maybe_send_server_notice_to_user(self.user_id)) diff --git a/tests/test_federation.py b/tests/test_federation.py index 9b5cf562f3..f297de95f1 100644 --- a/tests/test_federation.py +++ b/tests/test_federation.py @@ -27,8 +27,10 @@ class MessageAcceptTests(unittest.TestCase): user_id = UserID("us", "test") our_user = Requester(user_id, None, False, None, None) room_creator = self.homeserver.get_room_creation_handler() - room = room_creator.create_room( - our_user, room_creator.PRESETS_DICT["public_chat"], ratelimit=False + room = ensureDeferred( + room_creator.create_room( + our_user, room_creator.PRESETS_DICT["public_chat"], ratelimit=False + ) ) self.reactor.advance(0.1) self.room_id = self.successResultOf(room)["room_id"] -- cgit 1.4.1 From 3085cde577216519d789c8160262831cb2029972 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Fri, 1 May 2020 15:21:35 +0100 Subject: Use `stream.current_token()` and remove `stream_positions()` (#7172) We move the processing of typing and federation replication traffic into their handlers so that `Stream.current_token()` points to a valid token. This allows us to remove `get_streams_to_replicate()` and `stream_positions()`. --- changelog.d/7172.misc | 1 + synapse/app/generic_worker.py | 16 ------------ synapse/replication/slave/storage/_base.py | 15 +----------- synapse/replication/slave/storage/account_data.py | 8 ------ synapse/replication/slave/storage/deviceinbox.py | 5 ---- synapse/replication/slave/storage/devices.py | 10 -------- synapse/replication/slave/storage/events.py | 6 ----- synapse/replication/slave/storage/groups.py | 5 ---- synapse/replication/slave/storage/presence.py | 9 ------- synapse/replication/slave/storage/push_rule.py | 5 ---- synapse/replication/slave/storage/pushers.py | 5 ---- synapse/replication/slave/storage/receipts.py | 5 ---- synapse/replication/slave/storage/room.py | 5 ---- synapse/replication/tcp/client.py | 19 +------------- synapse/replication/tcp/handler.py | 10 +------- tests/replication/tcp/streams/_base.py | 30 ++++++++--------------- tests/replication/tcp/streams/test_events.py | 24 ++++++++++++------ tests/replication/tcp/streams/test_receipts.py | 3 --- tests/replication/tcp/streams/test_typing.py | 3 --- 19 files changed, 30 insertions(+), 154 deletions(-) create mode 100644 changelog.d/7172.misc (limited to 'tests') diff --git a/changelog.d/7172.misc b/changelog.d/7172.misc new file mode 100644 index 0000000000..ffecdf97fe --- /dev/null +++ b/changelog.d/7172.misc @@ -0,0 +1 @@ +Use `stream.current_token()` and remove `stream_positions()`. diff --git a/synapse/app/generic_worker.py b/synapse/app/generic_worker.py index 0ace7b787d..97b9b81237 100644 --- a/synapse/app/generic_worker.py +++ b/synapse/app/generic_worker.py @@ -413,12 +413,6 @@ class GenericWorkerTyping(object): # map room IDs to sets of users currently typing self._room_typing = {} - def stream_positions(self): - # We must update this typing token from the response of the previous - # sync. In particular, the stream id may "reset" back to zero/a low - # value which we *must* use for the next replication request. - return {"typing": self._latest_room_serial} - def process_replication_rows(self, token, rows): if self._latest_room_serial > token: # The master has gone backwards. To prevent inconsistent data, just @@ -658,13 +652,6 @@ class GenericWorkerReplicationHandler(ReplicationDataHandler): ) await self.process_and_notify(stream_name, token, rows) - def get_streams_to_replicate(self): - args = super(GenericWorkerReplicationHandler, self).get_streams_to_replicate() - args.update(self.typing_handler.stream_positions()) - if self.send_handler: - args.update(self.send_handler.stream_positions()) - return args - async def process_and_notify(self, stream_name, token, rows): try: if self.send_handler: @@ -799,9 +786,6 @@ class FederationSenderHandler(object): def wake_destination(self, server: str): self.federation_sender.wake_destination(server) - def stream_positions(self): - return {"federation": self.federation_position} - async def process_replication_rows(self, stream_name, token, rows): # The federation stream contains things that we want to send out, e.g. # presence, typing, etc. diff --git a/synapse/replication/slave/storage/_base.py b/synapse/replication/slave/storage/_base.py index 751c799d94..5d7c8871a4 100644 --- a/synapse/replication/slave/storage/_base.py +++ b/synapse/replication/slave/storage/_base.py @@ -14,7 +14,7 @@ # limitations under the License. import logging -from typing import Dict, Optional +from typing import Optional import six @@ -49,19 +49,6 @@ class BaseSlavedStore(CacheInvalidationWorkerStore): self.hs = hs - def stream_positions(self) -> Dict[str, int]: - """ - Get the current positions of all the streams this store wants to subscribe to - - Returns: - map from stream name to the most recent update we have for - that stream (ie, the point we want to start replicating from) - """ - pos = {} - if self._cache_id_gen: - pos["caches"] = self._cache_id_gen.get_current_token() - return pos - def get_cache_stream_token(self): if self._cache_id_gen: return self._cache_id_gen.get_current_token() diff --git a/synapse/replication/slave/storage/account_data.py b/synapse/replication/slave/storage/account_data.py index ebe94909cb..65e54b1c71 100644 --- a/synapse/replication/slave/storage/account_data.py +++ b/synapse/replication/slave/storage/account_data.py @@ -32,14 +32,6 @@ class SlavedAccountDataStore(TagsWorkerStore, AccountDataWorkerStore, BaseSlaved def get_max_account_data_stream_id(self): return self._account_data_id_gen.get_current_token() - def stream_positions(self): - result = super(SlavedAccountDataStore, self).stream_positions() - position = self._account_data_id_gen.get_current_token() - result["user_account_data"] = position - result["room_account_data"] = position - result["tag_account_data"] = position - return result - def process_replication_rows(self, stream_name, token, rows): if stream_name == "tag_account_data": self._account_data_id_gen.advance(token) diff --git a/synapse/replication/slave/storage/deviceinbox.py b/synapse/replication/slave/storage/deviceinbox.py index 0c237c6e0f..c923751e50 100644 --- a/synapse/replication/slave/storage/deviceinbox.py +++ b/synapse/replication/slave/storage/deviceinbox.py @@ -43,11 +43,6 @@ class SlavedDeviceInboxStore(DeviceInboxWorkerStore, BaseSlavedStore): expiry_ms=30 * 60 * 1000, ) - def stream_positions(self): - result = super(SlavedDeviceInboxStore, self).stream_positions() - result["to_device"] = self._device_inbox_id_gen.get_current_token() - return result - def process_replication_rows(self, stream_name, token, rows): if stream_name == "to_device": self._device_inbox_id_gen.advance(token) diff --git a/synapse/replication/slave/storage/devices.py b/synapse/replication/slave/storage/devices.py index 23b1650e41..58fb0eaae3 100644 --- a/synapse/replication/slave/storage/devices.py +++ b/synapse/replication/slave/storage/devices.py @@ -48,16 +48,6 @@ class SlavedDeviceStore(EndToEndKeyWorkerStore, DeviceWorkerStore, BaseSlavedSto "DeviceListFederationStreamChangeCache", device_list_max ) - def stream_positions(self): - result = super(SlavedDeviceStore, self).stream_positions() - # The user signature stream uses the same stream ID generator as the - # device list stream, so set them both to the device list ID - # generator's current token. - current_token = self._device_list_id_gen.get_current_token() - result[DeviceListsStream.NAME] = current_token - result[UserSignatureStream.NAME] = current_token - return result - def process_replication_rows(self, stream_name, token, rows): if stream_name == DeviceListsStream.NAME: self._device_list_id_gen.advance(token) diff --git a/synapse/replication/slave/storage/events.py b/synapse/replication/slave/storage/events.py index e73342c657..15011259df 100644 --- a/synapse/replication/slave/storage/events.py +++ b/synapse/replication/slave/storage/events.py @@ -93,12 +93,6 @@ class SlavedEventStore( def get_room_min_stream_ordering(self): return self._backfill_id_gen.get_current_token() - def stream_positions(self): - result = super(SlavedEventStore, self).stream_positions() - result["events"] = self._stream_id_gen.get_current_token() - result["backfill"] = -self._backfill_id_gen.get_current_token() - return result - def process_replication_rows(self, stream_name, token, rows): if stream_name == "events": self._stream_id_gen.advance(token) diff --git a/synapse/replication/slave/storage/groups.py b/synapse/replication/slave/storage/groups.py index 2d4fd08cf5..01bcf0e882 100644 --- a/synapse/replication/slave/storage/groups.py +++ b/synapse/replication/slave/storage/groups.py @@ -37,11 +37,6 @@ class SlavedGroupServerStore(GroupServerWorkerStore, BaseSlavedStore): def get_group_stream_token(self): return self._group_updates_id_gen.get_current_token() - def stream_positions(self): - result = super(SlavedGroupServerStore, self).stream_positions() - result["groups"] = self._group_updates_id_gen.get_current_token() - return result - def process_replication_rows(self, stream_name, token, rows): if stream_name == "groups": self._group_updates_id_gen.advance(token) diff --git a/synapse/replication/slave/storage/presence.py b/synapse/replication/slave/storage/presence.py index ad8f0c15a9..fae3125072 100644 --- a/synapse/replication/slave/storage/presence.py +++ b/synapse/replication/slave/storage/presence.py @@ -41,15 +41,6 @@ class SlavedPresenceStore(BaseSlavedStore): def get_current_presence_token(self): return self._presence_id_gen.get_current_token() - def stream_positions(self): - result = super(SlavedPresenceStore, self).stream_positions() - - if self.hs.config.use_presence: - position = self._presence_id_gen.get_current_token() - result["presence"] = position - - return result - def process_replication_rows(self, stream_name, token, rows): if stream_name == "presence": self._presence_id_gen.advance(token) diff --git a/synapse/replication/slave/storage/push_rule.py b/synapse/replication/slave/storage/push_rule.py index eebd5a1fb6..6138796da4 100644 --- a/synapse/replication/slave/storage/push_rule.py +++ b/synapse/replication/slave/storage/push_rule.py @@ -37,11 +37,6 @@ class SlavedPushRuleStore(SlavedEventStore, PushRulesWorkerStore): def get_max_push_rules_stream_id(self): return self._push_rules_stream_id_gen.get_current_token() - def stream_positions(self): - result = super(SlavedPushRuleStore, self).stream_positions() - result["push_rules"] = self._push_rules_stream_id_gen.get_current_token() - return result - def process_replication_rows(self, stream_name, token, rows): if stream_name == "push_rules": self._push_rules_stream_id_gen.advance(token) diff --git a/synapse/replication/slave/storage/pushers.py b/synapse/replication/slave/storage/pushers.py index bce8a3d115..67be337945 100644 --- a/synapse/replication/slave/storage/pushers.py +++ b/synapse/replication/slave/storage/pushers.py @@ -28,11 +28,6 @@ class SlavedPusherStore(PusherWorkerStore, BaseSlavedStore): db_conn, "pushers", "id", extra_tables=[("deleted_pushers", "stream_id")] ) - def stream_positions(self): - result = super(SlavedPusherStore, self).stream_positions() - result["pushers"] = self._pushers_id_gen.get_current_token() - return result - def get_pushers_stream_token(self): return self._pushers_id_gen.get_current_token() diff --git a/synapse/replication/slave/storage/receipts.py b/synapse/replication/slave/storage/receipts.py index d40dc6e1f5..993432edcb 100644 --- a/synapse/replication/slave/storage/receipts.py +++ b/synapse/replication/slave/storage/receipts.py @@ -42,11 +42,6 @@ class SlavedReceiptsStore(ReceiptsWorkerStore, BaseSlavedStore): def get_max_receipt_stream_id(self): return self._receipts_id_gen.get_current_token() - def stream_positions(self): - result = super(SlavedReceiptsStore, self).stream_positions() - result["receipts"] = self._receipts_id_gen.get_current_token() - return result - def invalidate_caches_for_receipt(self, room_id, receipt_type, user_id): self.get_receipts_for_user.invalidate((user_id, receipt_type)) self._get_linearized_receipts_for_room.invalidate_many((room_id,)) diff --git a/synapse/replication/slave/storage/room.py b/synapse/replication/slave/storage/room.py index 3a20f45316..10dda8708f 100644 --- a/synapse/replication/slave/storage/room.py +++ b/synapse/replication/slave/storage/room.py @@ -30,11 +30,6 @@ class RoomStore(RoomWorkerStore, BaseSlavedStore): def get_current_public_room_stream_id(self): return self._public_room_id_gen.get_current_token() - def stream_positions(self): - result = super(RoomStore, self).stream_positions() - result["public_rooms"] = self._public_room_id_gen.get_current_token() - return result - def process_replication_rows(self, stream_name, token, rows): if stream_name == "public_rooms": self._public_room_id_gen.advance(token) diff --git a/synapse/replication/tcp/client.py b/synapse/replication/tcp/client.py index 2d07b8b2d0..5c28fd4ac3 100644 --- a/synapse/replication/tcp/client.py +++ b/synapse/replication/tcp/client.py @@ -16,7 +16,7 @@ """ import logging -from typing import TYPE_CHECKING, Dict +from typing import TYPE_CHECKING from twisted.internet.protocol import ReconnectingClientFactory @@ -100,23 +100,6 @@ class ReplicationDataHandler: """ self.store.process_replication_rows(stream_name, token, rows) - def get_streams_to_replicate(self) -> Dict[str, int]: - """Called when a new connection has been established and we need to - subscribe to streams. - - Returns: - map from stream name to the most recent update we have for - that stream (ie, the point we want to start replicating from) - """ - args = self.store.stream_positions() - user_account_data = args.pop("user_account_data", None) - room_account_data = args.pop("room_account_data", None) - if user_account_data: - args["account_data"] = user_account_data - elif room_account_data: - args["account_data"] = room_account_data - return args - async def on_position(self, stream_name: str, token: int): self.store.process_replication_rows(stream_name, token, []) diff --git a/synapse/replication/tcp/handler.py b/synapse/replication/tcp/handler.py index 6f7054d5af..d72f3d0cf9 100644 --- a/synapse/replication/tcp/handler.py +++ b/synapse/replication/tcp/handler.py @@ -314,15 +314,7 @@ class ReplicationCommandHandler: self._pending_batches.pop(cmd.stream_name, []) # Find where we previously streamed up to. - current_token = self._replication_data_handler.get_streams_to_replicate().get( - cmd.stream_name - ) - if current_token is None: - logger.warning( - "Got POSITION for stream we're not subscribed to: %s", - cmd.stream_name, - ) - return + current_token = stream.current_token() # If the position token matches our current token then we're up to # date and there's nothing to do. Otherwise, fetch all updates diff --git a/tests/replication/tcp/streams/_base.py b/tests/replication/tcp/streams/_base.py index 83e16cfe3d..8c104f8d1d 100644 --- a/tests/replication/tcp/streams/_base.py +++ b/tests/replication/tcp/streams/_base.py @@ -14,7 +14,7 @@ # limitations under the License. import logging -from typing import Any, Dict, List, Optional, Tuple +from typing import Any, List, Optional, Tuple import attr @@ -22,13 +22,15 @@ from twisted.internet.interfaces import IConsumer, IPullProducer, IReactorTime from twisted.internet.task import LoopingCall from twisted.web.http import HTTPChannel -from synapse.app.generic_worker import GenericWorkerServer +from synapse.app.generic_worker import ( + GenericWorkerReplicationHandler, + GenericWorkerServer, +) from synapse.http.site import SynapseRequest -from synapse.replication.slave.storage._base import BaseSlavedStore -from synapse.replication.tcp.client import ReplicationDataHandler from synapse.replication.tcp.handler import ReplicationCommandHandler from synapse.replication.tcp.protocol import ClientReplicationStreamProtocol from synapse.replication.tcp.resource import ReplicationStreamProtocolFactory +from synapse.server import HomeServer from synapse.util import Clock from tests import unittest @@ -77,7 +79,7 @@ class BaseStreamTestCase(unittest.HomeserverTestCase): self._server_transport = None def _build_replication_data_handler(self): - return TestReplicationDataHandler(self.worker_hs.get_datastore()) + return TestReplicationDataHandler(self.worker_hs) def reconnect(self): if self._client_transport: @@ -172,32 +174,20 @@ class BaseStreamTestCase(unittest.HomeserverTestCase): self.assertEqual(request.method, b"GET") -class TestReplicationDataHandler(ReplicationDataHandler): +class TestReplicationDataHandler(GenericWorkerReplicationHandler): """Drop-in for ReplicationDataHandler which just collects RDATA rows""" - def __init__(self, store: BaseSlavedStore): - super().__init__(store) - - # streams to subscribe to: map from stream id to position - self.stream_positions = {} # type: Dict[str, int] + def __init__(self, hs: HomeServer): + super().__init__(hs) # list of received (stream_name, token, row) tuples self.received_rdata_rows = [] # type: List[Tuple[str, int, Any]] - def get_streams_to_replicate(self): - return self.stream_positions - async def on_rdata(self, stream_name, token, rows): await super().on_rdata(stream_name, token, rows) for r in rows: self.received_rdata_rows.append((stream_name, token, r)) - if ( - stream_name in self.stream_positions - and token > self.stream_positions[stream_name] - ): - self.stream_positions[stream_name] = token - @attr.s() class OneShotRequestFactory: diff --git a/tests/replication/tcp/streams/test_events.py b/tests/replication/tcp/streams/test_events.py index 1fa28084f9..8bd67bb9f1 100644 --- a/tests/replication/tcp/streams/test_events.py +++ b/tests/replication/tcp/streams/test_events.py @@ -43,7 +43,6 @@ class EventsStreamTestCase(BaseStreamTestCase): self.user_tok = self.login("u1", "pass") self.reconnect() - self.test_handler.stream_positions["events"] = 0 self.room_id = self.helper.create_room_as(tok=self.user_tok) self.test_handler.received_rdata_rows.clear() @@ -80,8 +79,12 @@ class EventsStreamTestCase(BaseStreamTestCase): self.reconnect() self.replicate() - # we should have received all the expected rows in the right order - received_rows = self.test_handler.received_rdata_rows + # we should have received all the expected rows in the right order (as + # well as various cache invalidation updates which we ignore) + received_rows = [ + row for row in self.test_handler.received_rdata_rows if row[0] == "events" + ] + for event in events: stream_name, token, row = received_rows.pop(0) self.assertEqual("events", stream_name) @@ -184,7 +187,8 @@ class EventsStreamTestCase(BaseStreamTestCase): self.reconnect() self.replicate() - # now we should have received all the expected rows in the right order. + # we should have received all the expected rows in the right order (as + # well as various cache invalidation updates which we ignore) # # we expect: # @@ -193,7 +197,9 @@ class EventsStreamTestCase(BaseStreamTestCase): # of the states that got reverted. # - two rows for state2 - received_rows = self.test_handler.received_rdata_rows + received_rows = [ + row for row in self.test_handler.received_rdata_rows if row[0] == "events" + ] # first check the first two rows, which should be state1 @@ -334,9 +340,11 @@ class EventsStreamTestCase(BaseStreamTestCase): self.reconnect() self.replicate() - # we should have received all the expected rows in the right order - - received_rows = self.test_handler.received_rdata_rows + # we should have received all the expected rows in the right order (as + # well as various cache invalidation updates which we ignore) + received_rows = [ + row for row in self.test_handler.received_rdata_rows if row[0] == "events" + ] self.assertGreaterEqual(len(received_rows), len(events)) for i in range(NUM_USERS): # for each user, we expect the PL event row, followed by state rows for diff --git a/tests/replication/tcp/streams/test_receipts.py b/tests/replication/tcp/streams/test_receipts.py index c122b8589c..df332ee679 100644 --- a/tests/replication/tcp/streams/test_receipts.py +++ b/tests/replication/tcp/streams/test_receipts.py @@ -31,9 +31,6 @@ class ReceiptsStreamTestCase(BaseStreamTestCase): def test_receipt(self): self.reconnect() - # make the client subscribe to the receipts stream - self.test_handler.stream_positions.update({"receipts": 0}) - # tell the master to send a new receipt self.get_success( self.hs.get_datastore().insert_receipt( diff --git a/tests/replication/tcp/streams/test_typing.py b/tests/replication/tcp/streams/test_typing.py index 4d354a9db8..e8d17ca68a 100644 --- a/tests/replication/tcp/streams/test_typing.py +++ b/tests/replication/tcp/streams/test_typing.py @@ -38,9 +38,6 @@ class TypingStreamTestCase(BaseStreamTestCase): self.reconnect() - # make the client subscribe to the typing stream - self.test_handler.stream_positions.update({"typing": 0}) - typing._push_update(member=RoomMember(room_id, USER_ID), typing=True) self.reactor.advance(0) -- cgit 1.4.1 From 0e719f23981b8294df66ba7f38b8c7cc99fad228 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Fri, 1 May 2020 17:19:56 +0100 Subject: Thread through instance name to replication client. (#7369) For in memory streams when fetching updates on workers we need to query the source of the stream, which currently is hard coded to be master. This PR threads through the source instance we received via `POSITION` through to the update function in each stream, which can then be passed to the replication client for in memory streams. --- changelog.d/7369.misc | 1 + synapse/app/generic_worker.py | 10 +++--- synapse/replication/http/_base.py | 19 +++++++++- synapse/replication/http/streams.py | 4 ++- synapse/replication/tcp/client.py | 12 ++++--- synapse/replication/tcp/handler.py | 20 ++++++++--- synapse/replication/tcp/streams/_base.py | 50 +++++++++++++++++++------- synapse/replication/tcp/streams/events.py | 10 ++++-- synapse/replication/tcp/streams/federation.py | 4 +-- tests/replication/tcp/streams/_base.py | 4 +-- tests/replication/tcp/streams/test_receipts.py | 4 +-- tests/replication/tcp/streams/test_typing.py | 4 +-- 12 files changed, 101 insertions(+), 41 deletions(-) create mode 100644 changelog.d/7369.misc (limited to 'tests') diff --git a/changelog.d/7369.misc b/changelog.d/7369.misc new file mode 100644 index 0000000000..060b09c888 --- /dev/null +++ b/changelog.d/7369.misc @@ -0,0 +1 @@ +Thread through instance name to replication client. diff --git a/synapse/app/generic_worker.py b/synapse/app/generic_worker.py index 97b9b81237..667ad20428 100644 --- a/synapse/app/generic_worker.py +++ b/synapse/app/generic_worker.py @@ -646,13 +646,11 @@ class GenericWorkerReplicationHandler(ReplicationDataHandler): else: self.send_handler = None - async def on_rdata(self, stream_name, token, rows): - await super(GenericWorkerReplicationHandler, self).on_rdata( - stream_name, token, rows - ) - await self.process_and_notify(stream_name, token, rows) + async def on_rdata(self, stream_name, instance_name, token, rows): + await super().on_rdata(stream_name, instance_name, token, rows) + await self._process_and_notify(stream_name, instance_name, token, rows) - async def process_and_notify(self, stream_name, token, rows): + async def _process_and_notify(self, stream_name, instance_name, token, rows): try: if self.send_handler: await self.send_handler.process_replication_rows( diff --git a/synapse/replication/http/_base.py b/synapse/replication/http/_base.py index 1be1ccbdf3..f88c80ae84 100644 --- a/synapse/replication/http/_base.py +++ b/synapse/replication/http/_base.py @@ -16,6 +16,7 @@ import abc import logging import re +from inspect import signature from typing import Dict, List, Tuple from six import raise_from @@ -60,6 +61,8 @@ class ReplicationEndpoint(object): must call `register` to register the path with the HTTP server. Requests can be sent by calling the client returned by `make_client`. + Requests are sent to master process by default, but can be sent to other + named processes by specifying an `instance_name` keyword argument. Attributes: NAME (str): A name for the endpoint, added to the path as well as used @@ -91,6 +94,16 @@ class ReplicationEndpoint(object): hs, "repl." + self.NAME, timeout_ms=30 * 60 * 1000 ) + # We reserve `instance_name` as a parameter to sending requests, so we + # assert here that sub classes don't try and use the name. + assert ( + "instance_name" not in self.PATH_ARGS + ), "`instance_name` is a reserved paramater name" + assert ( + "instance_name" + not in signature(self.__class__._serialize_payload).parameters + ), "`instance_name` is a reserved paramater name" + assert self.METHOD in ("PUT", "POST", "GET") @abc.abstractmethod @@ -135,7 +148,11 @@ class ReplicationEndpoint(object): @trace(opname="outgoing_replication_request") @defer.inlineCallbacks - def send_request(**kwargs): + def send_request(instance_name="master", **kwargs): + # Currently we only support sending requests to master process. + if instance_name != "master": + raise Exception("Unknown instance") + data = yield cls._serialize_payload(**kwargs) url_args = [ diff --git a/synapse/replication/http/streams.py b/synapse/replication/http/streams.py index f35cebc710..0459f582bf 100644 --- a/synapse/replication/http/streams.py +++ b/synapse/replication/http/streams.py @@ -50,6 +50,8 @@ class ReplicationGetStreamUpdates(ReplicationEndpoint): def __init__(self, hs): super().__init__(hs) + self._instance_name = hs.get_instance_name() + # We pull the streams from the replication steamer (if we try and make # them ourselves we end up in an import loop). self.streams = hs.get_replication_streamer().get_streams() @@ -67,7 +69,7 @@ class ReplicationGetStreamUpdates(ReplicationEndpoint): upto_token = parse_integer(request, "upto_token", required=True) updates, upto_token, limited = await stream.get_updates_since( - from_token, upto_token + self._instance_name, from_token, upto_token ) return ( diff --git a/synapse/replication/tcp/client.py b/synapse/replication/tcp/client.py index 5c28fd4ac3..3bbf3c3569 100644 --- a/synapse/replication/tcp/client.py +++ b/synapse/replication/tcp/client.py @@ -86,17 +86,19 @@ class ReplicationDataHandler: def __init__(self, store: BaseSlavedStore): self.store = store - async def on_rdata(self, stream_name: str, token: int, rows: list): + async def on_rdata( + self, stream_name: str, instance_name: str, token: int, rows: list + ): """Called to handle a batch of replication data with a given stream token. By default this just pokes the slave store. Can be overridden in subclasses to handle more. Args: - stream_name (str): name of the replication stream for this batch of rows - token (int): stream token for this batch of rows - rows (list): a list of Stream.ROW_TYPE objects as returned by - Stream.parse_row. + stream_name: name of the replication stream for this batch of rows + instance_name: the instance that wrote the rows. + token: stream token for this batch of rows + rows: a list of Stream.ROW_TYPE objects as returned by Stream.parse_row. """ self.store.process_replication_rows(stream_name, token, rows) diff --git a/synapse/replication/tcp/handler.py b/synapse/replication/tcp/handler.py index d72f3d0cf9..2d1d119c7c 100644 --- a/synapse/replication/tcp/handler.py +++ b/synapse/replication/tcp/handler.py @@ -278,19 +278,24 @@ class ReplicationCommandHandler: # Check if this is the last of a batch of updates rows = self._pending_batches.pop(stream_name, []) rows.append(row) - await self.on_rdata(stream_name, cmd.token, rows) + await self.on_rdata(stream_name, cmd.instance_name, cmd.token, rows) - async def on_rdata(self, stream_name: str, token: int, rows: list): + async def on_rdata( + self, stream_name: str, instance_name: str, token: int, rows: list + ): """Called to handle a batch of replication data with a given stream token. Args: stream_name: name of the replication stream for this batch of rows + instance_name: the instance that wrote the rows. token: stream token for this batch of rows rows: a list of Stream.ROW_TYPE objects as returned by Stream.parse_row. """ logger.debug("Received rdata %s -> %s", stream_name, token) - await self._replication_data_handler.on_rdata(stream_name, token, rows) + await self._replication_data_handler.on_rdata( + stream_name, instance_name, token, rows + ) async def on_POSITION(self, conn: AbstractConnection, cmd: PositionCommand): if cmd.instance_name == self._instance_name: @@ -325,7 +330,9 @@ class ReplicationCommandHandler: updates, current_token, missing_updates, - ) = await stream.get_updates_since(current_token, cmd.token) + ) = await stream.get_updates_since( + cmd.instance_name, current_token, cmd.token + ) # TODO: add some tests for this @@ -334,7 +341,10 @@ class ReplicationCommandHandler: for token, rows in _batch_updates(updates): await self.on_rdata( - cmd.stream_name, token, [stream.parse_row(row) for row in rows], + cmd.stream_name, + cmd.instance_name, + token, + [stream.parse_row(row) for row in rows], ) # We've now caught up to position sent to us, notify handler. diff --git a/synapse/replication/tcp/streams/_base.py b/synapse/replication/tcp/streams/_base.py index 4af1afd119..b0f87c365b 100644 --- a/synapse/replication/tcp/streams/_base.py +++ b/synapse/replication/tcp/streams/_base.py @@ -16,7 +16,7 @@ import logging from collections import namedtuple -from typing import Any, Awaitable, Callable, Iterable, List, Optional, Tuple +from typing import Any, Awaitable, Callable, List, Optional, Tuple import attr @@ -53,6 +53,7 @@ StreamUpdateResult = Tuple[List[Tuple[Token, StreamRow]], Token, bool] # # The arguments are: # +# * instance_name: the writer of the stream # * from_token: the previous stream token: the starting point for fetching the # updates # * to_token: the new stream token: the point to get updates up to @@ -62,7 +63,7 @@ StreamUpdateResult = Tuple[List[Tuple[Token, StreamRow]], Token, bool] # If there are more updates available, it should set `limited` in the result, and # it will be called again to get the next batch. # -UpdateFunction = Callable[[Token, Token, int], Awaitable[StreamUpdateResult]] +UpdateFunction = Callable[[str, Token, Token, int], Awaitable[StreamUpdateResult]] class Stream(object): @@ -93,6 +94,7 @@ class Stream(object): def __init__( self, + local_instance_name: str, current_token_function: Callable[[], Token], update_function: UpdateFunction, ): @@ -108,9 +110,11 @@ class Stream(object): stream tokens. See the UpdateFunction type definition for more info. Args: + local_instance_name: The instance name of the current process current_token_function: callback to get the current token, as above update_function: callback go get stream updates, as above """ + self.local_instance_name = local_instance_name self.current_token = current_token_function self.update_function = update_function @@ -135,14 +139,14 @@ class Stream(object): """ current_token = self.current_token() updates, current_token, limited = await self.get_updates_since( - self.last_token, current_token + self.local_instance_name, self.last_token, current_token ) self.last_token = current_token return updates, current_token, limited async def get_updates_since( - self, from_token: Token, upto_token: Token + self, instance_name: str, from_token: Token, upto_token: Token ) -> StreamUpdateResult: """Like get_updates except allows specifying from when we should stream updates @@ -160,19 +164,19 @@ class Stream(object): return [], upto_token, False updates, upto_token, limited = await self.update_function( - from_token, upto_token, _STREAM_UPDATE_TARGET_ROW_COUNT, + instance_name, from_token, upto_token, _STREAM_UPDATE_TARGET_ROW_COUNT, ) return updates, upto_token, limited def db_query_to_update_function( - query_function: Callable[[Token, Token, int], Awaitable[Iterable[tuple]]] + query_function: Callable[[Token, Token, int], Awaitable[List[tuple]]] ) -> UpdateFunction: """Wraps a db query function which returns a list of rows to make it suitable for use as an `update_function` for the Stream class """ - async def update_function(from_token, upto_token, limit): + async def update_function(instance_name, from_token, upto_token, limit): rows = await query_function(from_token, upto_token, limit) updates = [(row[0], row[1:]) for row in rows] limited = False @@ -193,10 +197,13 @@ def make_http_update_function(hs, stream_name: str) -> UpdateFunction: client = ReplicationGetStreamUpdates.make_client(hs) async def update_function( - from_token: int, upto_token: int, limit: int + instance_name: str, from_token: int, upto_token: int, limit: int ) -> StreamUpdateResult: result = await client( - stream_name=stream_name, from_token=from_token, upto_token=upto_token, + instance_name=instance_name, + stream_name=stream_name, + from_token=from_token, + upto_token=upto_token, ) return result["updates"], result["upto_token"], result["limited"] @@ -226,6 +233,7 @@ class BackfillStream(Stream): def __init__(self, hs): store = hs.get_datastore() super().__init__( + hs.get_instance_name(), store.get_current_backfill_token, db_query_to_update_function(store.get_all_new_backfill_event_rows), ) @@ -261,7 +269,9 @@ class PresenceStream(Stream): # Query master process update_function = make_http_update_function(hs, self.NAME) - super().__init__(store.get_current_presence_token, update_function) + super().__init__( + hs.get_instance_name(), store.get_current_presence_token, update_function + ) class TypingStream(Stream): @@ -284,7 +294,9 @@ class TypingStream(Stream): # Query master process update_function = make_http_update_function(hs, self.NAME) - super().__init__(typing_handler.get_current_token, update_function) + super().__init__( + hs.get_instance_name(), typing_handler.get_current_token, update_function + ) class ReceiptsStream(Stream): @@ -305,6 +317,7 @@ class ReceiptsStream(Stream): def __init__(self, hs): store = hs.get_datastore() super().__init__( + hs.get_instance_name(), store.get_max_receipt_stream_id, db_query_to_update_function(store.get_all_updated_receipts), ) @@ -322,14 +335,16 @@ class PushRulesStream(Stream): def __init__(self, hs): self.store = hs.get_datastore() super(PushRulesStream, self).__init__( - self._current_token, self._update_function + hs.get_instance_name(), self._current_token, self._update_function ) def _current_token(self) -> int: push_rules_token, _ = self.store.get_push_rules_stream_token() return push_rules_token - async def _update_function(self, from_token: Token, to_token: Token, limit: int): + async def _update_function( + self, instance_name: str, from_token: Token, to_token: Token, limit: int + ): rows = await self.store.get_all_push_rule_updates(from_token, to_token, limit) limited = False @@ -356,6 +371,7 @@ class PushersStream(Stream): store = hs.get_datastore() super().__init__( + hs.get_instance_name(), store.get_pushers_stream_token, db_query_to_update_function(store.get_all_updated_pushers_rows), ) @@ -387,6 +403,7 @@ class CachesStream(Stream): def __init__(self, hs): store = hs.get_datastore() super().__init__( + hs.get_instance_name(), store.get_cache_stream_token, db_query_to_update_function(store.get_all_updated_caches), ) @@ -412,6 +429,7 @@ class PublicRoomsStream(Stream): def __init__(self, hs): store = hs.get_datastore() super().__init__( + hs.get_instance_name(), store.get_current_public_room_stream_id, db_query_to_update_function(store.get_all_new_public_rooms), ) @@ -432,6 +450,7 @@ class DeviceListsStream(Stream): def __init__(self, hs): store = hs.get_datastore() super().__init__( + hs.get_instance_name(), store.get_device_stream_token, db_query_to_update_function(store.get_all_device_list_changes_for_remotes), ) @@ -449,6 +468,7 @@ class ToDeviceStream(Stream): def __init__(self, hs): store = hs.get_datastore() super().__init__( + hs.get_instance_name(), store.get_to_device_stream_token, db_query_to_update_function(store.get_all_new_device_messages), ) @@ -468,6 +488,7 @@ class TagAccountDataStream(Stream): def __init__(self, hs): store = hs.get_datastore() super().__init__( + hs.get_instance_name(), store.get_max_account_data_stream_id, db_query_to_update_function(store.get_all_updated_tags), ) @@ -487,6 +508,7 @@ class AccountDataStream(Stream): def __init__(self, hs): self.store = hs.get_datastore() super().__init__( + hs.get_instance_name(), self.store.get_max_account_data_stream_id, db_query_to_update_function(self._update_function), ) @@ -517,6 +539,7 @@ class GroupServerStream(Stream): def __init__(self, hs): store = hs.get_datastore() super().__init__( + hs.get_instance_name(), store.get_group_stream_token, db_query_to_update_function(store.get_all_groups_changes), ) @@ -534,6 +557,7 @@ class UserSignatureStream(Stream): def __init__(self, hs): store = hs.get_datastore() super().__init__( + hs.get_instance_name(), store.get_device_stream_token, db_query_to_update_function( store.get_all_user_signature_changes_for_remotes diff --git a/synapse/replication/tcp/streams/events.py b/synapse/replication/tcp/streams/events.py index 52df81b1bd..890e75d827 100644 --- a/synapse/replication/tcp/streams/events.py +++ b/synapse/replication/tcp/streams/events.py @@ -118,11 +118,17 @@ class EventsStream(Stream): def __init__(self, hs): self._store = hs.get_datastore() super().__init__( - self._store.get_current_events_token, self._update_function, + hs.get_instance_name(), + self._store.get_current_events_token, + self._update_function, ) async def _update_function( - self, from_token: Token, current_token: Token, target_row_count: int + self, + instance_name: str, + from_token: Token, + current_token: Token, + target_row_count: int, ) -> StreamUpdateResult: # the events stream merges together three separate sources: diff --git a/synapse/replication/tcp/streams/federation.py b/synapse/replication/tcp/streams/federation.py index 75133d7e40..e8bd52e389 100644 --- a/synapse/replication/tcp/streams/federation.py +++ b/synapse/replication/tcp/streams/federation.py @@ -48,8 +48,8 @@ class FederationStream(Stream): current_token = lambda: 0 update_function = self._stub_update_function - super().__init__(current_token, update_function) + super().__init__(hs.get_instance_name(), current_token, update_function) @staticmethod - async def _stub_update_function(from_token, upto_token, limit): + async def _stub_update_function(instance_name, from_token, upto_token, limit): return [], upto_token, False diff --git a/tests/replication/tcp/streams/_base.py b/tests/replication/tcp/streams/_base.py index 8c104f8d1d..7b56d2028d 100644 --- a/tests/replication/tcp/streams/_base.py +++ b/tests/replication/tcp/streams/_base.py @@ -183,8 +183,8 @@ class TestReplicationDataHandler(GenericWorkerReplicationHandler): # list of received (stream_name, token, row) tuples self.received_rdata_rows = [] # type: List[Tuple[str, int, Any]] - async def on_rdata(self, stream_name, token, rows): - await super().on_rdata(stream_name, token, rows) + async def on_rdata(self, stream_name, instance_name, token, rows): + await super().on_rdata(stream_name, instance_name, token, rows) for r in rows: self.received_rdata_rows.append((stream_name, token, r)) diff --git a/tests/replication/tcp/streams/test_receipts.py b/tests/replication/tcp/streams/test_receipts.py index df332ee679..5853314fd4 100644 --- a/tests/replication/tcp/streams/test_receipts.py +++ b/tests/replication/tcp/streams/test_receipts.py @@ -41,7 +41,7 @@ class ReceiptsStreamTestCase(BaseStreamTestCase): # there should be one RDATA command self.test_handler.on_rdata.assert_called_once() - stream_name, token, rdata_rows = self.test_handler.on_rdata.call_args[0] + stream_name, _, token, rdata_rows = self.test_handler.on_rdata.call_args[0] self.assertEqual(stream_name, "receipts") self.assertEqual(1, len(rdata_rows)) row = rdata_rows[0] # type: ReceiptsStream.ReceiptsStreamRow @@ -71,7 +71,7 @@ class ReceiptsStreamTestCase(BaseStreamTestCase): # We should now have caught up and get the missing data self.test_handler.on_rdata.assert_called_once() - stream_name, token, rdata_rows = self.test_handler.on_rdata.call_args[0] + stream_name, _, token, rdata_rows = self.test_handler.on_rdata.call_args[0] self.assertEqual(stream_name, "receipts") self.assertEqual(token, 3) self.assertEqual(1, len(rdata_rows)) diff --git a/tests/replication/tcp/streams/test_typing.py b/tests/replication/tcp/streams/test_typing.py index e8d17ca68a..d25a7b194e 100644 --- a/tests/replication/tcp/streams/test_typing.py +++ b/tests/replication/tcp/streams/test_typing.py @@ -47,7 +47,7 @@ class TypingStreamTestCase(BaseStreamTestCase): self.assert_request_is_get_repl_stream_updates(request, "typing") self.test_handler.on_rdata.assert_called_once() - stream_name, token, rdata_rows = self.test_handler.on_rdata.call_args[0] + stream_name, _, token, rdata_rows = self.test_handler.on_rdata.call_args[0] self.assertEqual(stream_name, "typing") self.assertEqual(1, len(rdata_rows)) row = rdata_rows[0] # type: TypingStream.TypingStreamRow @@ -74,7 +74,7 @@ class TypingStreamTestCase(BaseStreamTestCase): self.assertEqual(int(request.args[b"from_token"][0]), token) self.test_handler.on_rdata.assert_called_once() - stream_name, token, rdata_rows = self.test_handler.on_rdata.call_args[0] + stream_name, _, token, rdata_rows = self.test_handler.on_rdata.call_args[0] self.assertEqual(stream_name, "typing") self.assertEqual(1, len(rdata_rows)) row = rdata_rows[0] -- cgit 1.4.1 From 032e5a2acaa26511a00801eedd71e35d21c791fa Mon Sep 17 00:00:00 2001 From: Patrick Cloke Date: Fri, 1 May 2020 15:28:59 -0400 Subject: Convert synapse.server_notices to async/await. (#7394) --- changelog.d/7394.misc | 1 + synapse/server_notices/resource_limits_server_notices.py | 16 ++++++---------- .../test_resource_limits_server_notices.py | 2 +- 3 files changed, 8 insertions(+), 11 deletions(-) create mode 100644 changelog.d/7394.misc (limited to 'tests') diff --git a/changelog.d/7394.misc b/changelog.d/7394.misc new file mode 100644 index 0000000000..f1390308b3 --- /dev/null +++ b/changelog.d/7394.misc @@ -0,0 +1 @@ +Convert synapse.server_notices to async/await. diff --git a/synapse/server_notices/resource_limits_server_notices.py b/synapse/server_notices/resource_limits_server_notices.py index 971771b8b2..d97166351e 100644 --- a/synapse/server_notices/resource_limits_server_notices.py +++ b/synapse/server_notices/resource_limits_server_notices.py @@ -16,8 +16,6 @@ import logging from six import iteritems -from twisted.internet import defer - from synapse.api.constants import ( EventTypes, LimitBlockingTypes, @@ -167,8 +165,7 @@ class ResourceLimitsServerNotices(object): user_id, content, EventTypes.Pinned, "" ) - @defer.inlineCallbacks - def _check_and_set_tags(self, user_id, room_id): + async def _check_and_set_tags(self, user_id, room_id): """ Since server notices rooms were originally not with tags, important to check that tags have been set correctly @@ -176,20 +173,19 @@ class ResourceLimitsServerNotices(object): user_id(str): the user in question room_id(str): the server notices room for that user """ - tags = yield self._store.get_tags_for_room(user_id, room_id) + tags = await self._store.get_tags_for_room(user_id, room_id) need_to_set_tag = True if tags: if SERVER_NOTICE_ROOM_TAG in tags: # tag already present, nothing to do here need_to_set_tag = False if need_to_set_tag: - max_id = yield self._store.add_tag_to_room( + max_id = await self._store.add_tag_to_room( user_id, room_id, SERVER_NOTICE_ROOM_TAG, {} ) self._notifier.on_new_event("account_data_key", max_id, users=[user_id]) - @defer.inlineCallbacks - def _is_room_currently_blocked(self, room_id): + async def _is_room_currently_blocked(self, room_id): """ Determines if the room is currently blocked @@ -207,7 +203,7 @@ class ResourceLimitsServerNotices(object): currently_blocked = False pinned_state_event = None try: - pinned_state_event = yield self._state.get_current_state( + pinned_state_event = await self._state.get_current_state( room_id, event_type=EventTypes.Pinned ) except AuthError: @@ -218,7 +214,7 @@ class ResourceLimitsServerNotices(object): if pinned_state_event is not None: referenced_events = list(pinned_state_event.content.get("pinned", [])) - events = yield self._store.get_events(referenced_events) + events = await self._store.get_events(referenced_events) for event_id, event in iteritems(events): if event.type != EventTypes.Message: continue diff --git a/tests/server_notices/test_resource_limits_server_notices.py b/tests/server_notices/test_resource_limits_server_notices.py index 987addad9b..406f29a7c0 100644 --- a/tests/server_notices/test_resource_limits_server_notices.py +++ b/tests/server_notices/test_resource_limits_server_notices.py @@ -67,7 +67,7 @@ class TestResourceLimitsServerNotices(unittest.HomeserverTestCase): return_value=defer.succeed("!something:localhost") ) self._rlsn._store.add_tag_to_room = Mock(return_value=defer.succeed(None)) - self._rlsn._store.get_tags_for_room = Mock(return_value={}) + self._rlsn._store.get_tags_for_room = Mock(return_value=defer.succeed({})) self.hs.config.admin_contact = "mailto:user@test.com" def test_maybe_send_server_notice_to_user_flag_off(self): -- cgit 1.4.1 From 8123b2f90945e6bb3d65cf48730c2133b46d2b9a Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Mon, 4 May 2020 17:17:45 +0100 Subject: Add MultiWriterIdGenerator. (#7281) This will be used to coordinate stream IDs across multiple writers. Functions as the equivalent of both `StreamIdGenerator` and `SlavedIdTracker`. --- changelog.d/7281.misc | 1 + synapse/storage/util/id_generators.py | 169 ++++++++++++++++++++++++++++++- tests/storage/test_id_generators.py | 184 ++++++++++++++++++++++++++++++++++ tox.ini | 1 + 4 files changed, 353 insertions(+), 2 deletions(-) create mode 100644 changelog.d/7281.misc create mode 100644 tests/storage/test_id_generators.py (limited to 'tests') diff --git a/changelog.d/7281.misc b/changelog.d/7281.misc new file mode 100644 index 0000000000..86ad511e19 --- /dev/null +++ b/changelog.d/7281.misc @@ -0,0 +1 @@ +Add MultiWriterIdGenerator to support multiple concurrent writers of streams. diff --git a/synapse/storage/util/id_generators.py b/synapse/storage/util/id_generators.py index 9d851beaa5..86d04ea9ac 100644 --- a/synapse/storage/util/id_generators.py +++ b/synapse/storage/util/id_generators.py @@ -16,6 +16,11 @@ import contextlib import threading from collections import deque +from typing import Dict, Set, Tuple + +from typing_extensions import Deque + +from synapse.storage.database import Database, LoggingTransaction class IdGenerator(object): @@ -87,7 +92,7 @@ class StreamIdGenerator(object): self._current = (max if step > 0 else min)( self._current, _load_current_id(db_conn, table, column, step) ) - self._unfinished_ids = deque() + self._unfinished_ids = deque() # type: Deque[int] def get_next(self): """ @@ -163,7 +168,7 @@ class ChainedIdGenerator(object): self.chained_generator = chained_generator self._lock = threading.Lock() self._current_max = _load_current_id(db_conn, table, column) - self._unfinished_ids = deque() + self._unfinished_ids = deque() # type: Deque[Tuple[int, int]] def get_next(self): """ @@ -198,3 +203,163 @@ class ChainedIdGenerator(object): return stream_id - 1, chained_id return self._current_max, self.chained_generator.get_current_token() + + +class MultiWriterIdGenerator: + """An ID generator that tracks a stream that can have multiple writers. + + Uses a Postgres sequence to coordinate ID assignment, but positions of other + writers will only get updated when `advance` is called (by replication). + + Note: Only works with Postgres. + + Args: + db_conn + db + instance_name: The name of this instance. + table: Database table associated with stream. + instance_column: Column that stores the row's writer's instance name + id_column: Column that stores the stream ID. + sequence_name: The name of the postgres sequence used to generate new + IDs. + """ + + def __init__( + self, + db_conn, + db: Database, + instance_name: str, + table: str, + instance_column: str, + id_column: str, + sequence_name: str, + ): + self._db = db + self._instance_name = instance_name + self._sequence_name = sequence_name + + # We lock as some functions may be called from DB threads. + self._lock = threading.Lock() + + self._current_positions = self._load_current_ids( + db_conn, table, instance_column, id_column + ) + + # Set of local IDs that we're still processing. The current position + # should be less than the minimum of this set (if not empty). + self._unfinished_ids = set() # type: Set[int] + + def _load_current_ids( + self, db_conn, table: str, instance_column: str, id_column: str + ) -> Dict[str, int]: + sql = """ + SELECT %(instance)s, MAX(%(id)s) FROM %(table)s + GROUP BY %(instance)s + """ % { + "instance": instance_column, + "id": id_column, + "table": table, + } + + cur = db_conn.cursor() + cur.execute(sql) + + # `cur` is an iterable over returned rows, which are 2-tuples. + current_positions = dict(cur) + + cur.close() + + return current_positions + + def _load_next_id_txn(self, txn): + txn.execute("SELECT nextval(?)", (self._sequence_name,)) + (next_id,) = txn.fetchone() + return next_id + + async def get_next(self): + """ + Usage: + with await stream_id_gen.get_next() as stream_id: + # ... persist event ... + """ + next_id = await self._db.runInteraction("_load_next_id", self._load_next_id_txn) + + # Assert the fetched ID is actually greater than what we currently + # believe the ID to be. If not, then the sequence and table have got + # out of sync somehow. + assert self.get_current_token() < next_id + + with self._lock: + self._unfinished_ids.add(next_id) + + @contextlib.contextmanager + def manager(): + try: + yield next_id + finally: + self._mark_id_as_finished(next_id) + + return manager() + + def get_next_txn(self, txn: LoggingTransaction): + """ + Usage: + + stream_id = stream_id_gen.get_next(txn) + # ... persist event ... + """ + + next_id = self._load_next_id_txn(txn) + + with self._lock: + self._unfinished_ids.add(next_id) + + txn.call_after(self._mark_id_as_finished, next_id) + txn.call_on_exception(self._mark_id_as_finished, next_id) + + return next_id + + def _mark_id_as_finished(self, next_id: int): + """The ID has finished being processed so we should advance the + current poistion if possible. + """ + + with self._lock: + self._unfinished_ids.discard(next_id) + + # Figure out if its safe to advance the position by checking there + # aren't any lower allocated IDs that are yet to finish. + if all(c > next_id for c in self._unfinished_ids): + curr = self._current_positions.get(self._instance_name, 0) + self._current_positions[self._instance_name] = max(curr, next_id) + + def get_current_token(self, instance_name: str = None) -> int: + """Gets the current position of a named writer (defaults to current + instance). + + Returns 0 if we don't have a position for the named writer (likely due + to it being a new writer). + """ + + if instance_name is None: + instance_name = self._instance_name + + with self._lock: + return self._current_positions.get(instance_name, 0) + + def get_positions(self) -> Dict[str, int]: + """Get a copy of the current positon map. + """ + + with self._lock: + return dict(self._current_positions) + + def advance(self, instance_name: str, new_id: int): + """Advance the postion of the named writer to the given ID, if greater + than existing entry. + """ + + with self._lock: + self._current_positions[instance_name] = max( + new_id, self._current_positions.get(instance_name, 0) + ) diff --git a/tests/storage/test_id_generators.py b/tests/storage/test_id_generators.py new file mode 100644 index 0000000000..55e9ecf264 --- /dev/null +++ b/tests/storage/test_id_generators.py @@ -0,0 +1,184 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + + +from synapse.storage.database import Database +from synapse.storage.util.id_generators import MultiWriterIdGenerator + +from tests.unittest import HomeserverTestCase +from tests.utils import USE_POSTGRES_FOR_TESTS + + +class MultiWriterIdGeneratorTestCase(HomeserverTestCase): + if not USE_POSTGRES_FOR_TESTS: + skip = "Requires Postgres" + + def prepare(self, reactor, clock, hs): + self.store = hs.get_datastore() + self.db = self.store.db # type: Database + + self.get_success(self.db.runInteraction("_setup_db", self._setup_db)) + + def _setup_db(self, txn): + txn.execute("CREATE SEQUENCE foobar_seq") + txn.execute( + """ + CREATE TABLE foobar ( + stream_id BIGINT NOT NULL, + instance_name TEXT NOT NULL, + data TEXT + ); + """ + ) + + def _create_id_generator(self, instance_name="master") -> MultiWriterIdGenerator: + def _create(conn): + return MultiWriterIdGenerator( + conn, + self.db, + instance_name=instance_name, + table="foobar", + instance_column="instance_name", + id_column="stream_id", + sequence_name="foobar_seq", + ) + + return self.get_success(self.db.runWithConnection(_create)) + + def _insert_rows(self, instance_name: str, number: int): + def _insert(txn): + for _ in range(number): + txn.execute( + "INSERT INTO foobar VALUES (nextval('foobar_seq'), ?)", + (instance_name,), + ) + + self.get_success(self.db.runInteraction("test_single_instance", _insert)) + + def test_empty(self): + """Test an ID generator against an empty database gives sensible + current positions. + """ + + id_gen = self._create_id_generator() + + # The table is empty so we expect an empty map for positions + self.assertEqual(id_gen.get_positions(), {}) + + def test_single_instance(self): + """Test that reads and writes from a single process are handled + correctly. + """ + + # Prefill table with 7 rows written by 'master' + self._insert_rows("master", 7) + + id_gen = self._create_id_generator() + + self.assertEqual(id_gen.get_positions(), {"master": 7}) + self.assertEqual(id_gen.get_current_token("master"), 7) + + # Try allocating a new ID gen and check that we only see position + # advanced after we leave the context manager. + + async def _get_next_async(): + with await id_gen.get_next() as stream_id: + self.assertEqual(stream_id, 8) + + self.assertEqual(id_gen.get_positions(), {"master": 7}) + self.assertEqual(id_gen.get_current_token("master"), 7) + + self.get_success(_get_next_async()) + + self.assertEqual(id_gen.get_positions(), {"master": 8}) + self.assertEqual(id_gen.get_current_token("master"), 8) + + def test_multi_instance(self): + """Test that reads and writes from multiple processes are handled + correctly. + """ + self._insert_rows("first", 3) + self._insert_rows("second", 4) + + first_id_gen = self._create_id_generator("first") + second_id_gen = self._create_id_generator("second") + + self.assertEqual(first_id_gen.get_positions(), {"first": 3, "second": 7}) + self.assertEqual(first_id_gen.get_current_token("first"), 3) + self.assertEqual(first_id_gen.get_current_token("second"), 7) + + # Try allocating a new ID gen and check that we only see position + # advanced after we leave the context manager. + + async def _get_next_async(): + with await first_id_gen.get_next() as stream_id: + self.assertEqual(stream_id, 8) + + self.assertEqual( + first_id_gen.get_positions(), {"first": 3, "second": 7} + ) + + self.get_success(_get_next_async()) + + self.assertEqual(first_id_gen.get_positions(), {"first": 8, "second": 7}) + + # However the ID gen on the second instance won't have seen the update + self.assertEqual(second_id_gen.get_positions(), {"first": 3, "second": 7}) + + # ... but calling `get_next` on the second instance should give a unique + # stream ID + + async def _get_next_async(): + with await second_id_gen.get_next() as stream_id: + self.assertEqual(stream_id, 9) + + self.assertEqual( + second_id_gen.get_positions(), {"first": 3, "second": 7} + ) + + self.get_success(_get_next_async()) + + self.assertEqual(second_id_gen.get_positions(), {"first": 3, "second": 9}) + + # If the second ID gen gets told about the first, it correctly updates + second_id_gen.advance("first", 8) + self.assertEqual(second_id_gen.get_positions(), {"first": 8, "second": 9}) + + def test_get_next_txn(self): + """Test that the `get_next_txn` function works correctly. + """ + + # Prefill table with 7 rows written by 'master' + self._insert_rows("master", 7) + + id_gen = self._create_id_generator() + + self.assertEqual(id_gen.get_positions(), {"master": 7}) + self.assertEqual(id_gen.get_current_token("master"), 7) + + # Try allocating a new ID gen and check that we only see position + # advanced after we leave the context manager. + + def _get_next_txn(txn): + stream_id = id_gen.get_next_txn(txn) + self.assertEqual(stream_id, 8) + + self.assertEqual(id_gen.get_positions(), {"master": 7}) + self.assertEqual(id_gen.get_current_token("master"), 7) + + self.get_success(self.db.runInteraction("test", _get_next_txn)) + + self.assertEqual(id_gen.get_positions(), {"master": 8}) + self.assertEqual(id_gen.get_current_token("master"), 8) diff --git a/tox.ini b/tox.ini index eccc44e436..4a1509c51b 100644 --- a/tox.ini +++ b/tox.ini @@ -203,6 +203,7 @@ commands = mypy \ synapse/storage/data_stores/main/ui_auth.py \ synapse/storage/database.py \ synapse/storage/engines \ + synapse/storage/util \ synapse/streams \ synapse/util/caches/stream_change_cache.py \ tests/replication/tcp/streams \ -- cgit 1.4.1 From d5aa7d93ed1f7963524125d16ab640ebf6cb91c2 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Tue, 5 May 2020 14:15:57 +0100 Subject: Fix catchup-on-reconnect for the Federation Stream (#7374) looks like we managed to break this during the refactorathon. --- changelog.d/7374.misc | 1 + synapse/federation/send_queue.py | 40 ++++++----- synapse/federation/sender/__init__.py | 12 ++-- synapse/federation/sender/per_destination_queue.py | 6 +- synapse/federation/sender/transaction_manager.py | 6 +- synapse/replication/tcp/resource.py | 2 +- synapse/replication/tcp/streams/_base.py | 3 +- synapse/replication/tcp/streams/federation.py | 30 +++++--- tests/replication/tcp/streams/_base.py | 20 ++++-- tests/replication/tcp/streams/test_federation.py | 81 ++++++++++++++++++++++ tests/replication/tcp/streams/test_typing.py | 5 -- 11 files changed, 158 insertions(+), 48 deletions(-) create mode 100644 changelog.d/7374.misc create mode 100644 tests/replication/tcp/streams/test_federation.py (limited to 'tests') diff --git a/changelog.d/7374.misc b/changelog.d/7374.misc new file mode 100644 index 0000000000..676f285377 --- /dev/null +++ b/changelog.d/7374.misc @@ -0,0 +1 @@ +Move catchup of replication streams logic to worker. diff --git a/synapse/federation/send_queue.py b/synapse/federation/send_queue.py index e1700ca8aa..6fbacf6a3e 100644 --- a/synapse/federation/send_queue.py +++ b/synapse/federation/send_queue.py @@ -31,6 +31,7 @@ Events are replicated via a separate events stream. import logging from collections import namedtuple +from typing import List, Tuple from six import iteritems @@ -69,7 +70,11 @@ class FederationRemoteSendQueue(object): self.edus = SortedDict() # stream position -> Edu + # stream ID for the next entry into presence_changed/keyed_edu_changed/edus. self.pos = 1 + + # map from stream ID to the time that stream entry was generated, so that we + # can clear out entries after a while self.pos_time = SortedDict() # EVERYTHING IS SAD. In particular, python only makes new scopes when @@ -250,19 +255,23 @@ class FederationRemoteSendQueue(object): self._clear_queue_before_pos(token) async def get_replication_rows( - self, from_token, to_token, limit, federation_ack=None - ): + self, instance_name: str, from_token: int, to_token: int, target_row_count: int + ) -> Tuple[List[Tuple[int, Tuple]], int, bool]: """Get rows to be sent over federation between the two tokens Args: - from_token (int) - to_token(int) - limit (int) - federation_ack (int): Optional. The position where the worker is - explicitly acknowledged it has handled. Allows us to drop - data from before that point + instance_name: the name of the current process + from_token: the previous stream token: the starting point for fetching the + updates + to_token: the new stream token: the point to get updates up to + target_row_count: a target for the number of rows to be returned. + + Returns: a triplet `(updates, new_last_token, limited)`, where: + * `updates` is a list of `(token, row)` entries. + * `new_last_token` is the new position in stream. + * `limited` is whether there are more updates to fetch. """ - # TODO: Handle limit. + # TODO: Handle target_row_count. # To handle restarts where we wrap around if from_token > self.pos: @@ -270,12 +279,7 @@ class FederationRemoteSendQueue(object): # list of tuple(int, BaseFederationRow), where the first is the position # of the federation stream. - rows = [] - - # There should be only one reader, so lets delete everything its - # acknowledged its seen. - if federation_ack: - self._clear_queue_before_pos(federation_ack) + rows = [] # type: List[Tuple[int, BaseFederationRow]] # Fetch changed presence i = self.presence_changed.bisect_right(from_token) @@ -332,7 +336,11 @@ class FederationRemoteSendQueue(object): # Sort rows based on pos rows.sort() - return [(pos, row.TypeId, row.to_data()) for pos, row in rows] + return ( + [(pos, (row.TypeId, row.to_data())) for pos, row in rows], + to_token, + False, + ) class BaseFederationRow(object): diff --git a/synapse/federation/sender/__init__.py b/synapse/federation/sender/__init__.py index a477578e44..d473576902 100644 --- a/synapse/federation/sender/__init__.py +++ b/synapse/federation/sender/__init__.py @@ -14,7 +14,7 @@ # limitations under the License. import logging -from typing import Dict, Hashable, Iterable, List, Optional, Set +from typing import Dict, Hashable, Iterable, List, Optional, Set, Tuple from six import itervalues @@ -498,14 +498,16 @@ class FederationSender(object): self._get_per_destination_queue(destination).attempt_new_transaction() - def get_current_token(self) -> int: + @staticmethod + def get_current_token() -> int: # Dummy implementation for case where federation sender isn't offloaded # to a worker. return 0 + @staticmethod async def get_replication_rows( - self, from_token, to_token, limit, federation_ack=None - ): + instance_name: str, from_token: int, to_token: int, target_row_count: int + ) -> Tuple[List[Tuple[int, Tuple]], int, bool]: # Dummy implementation for case where federation sender isn't offloaded # to a worker. - return [] + return [], 0, False diff --git a/synapse/federation/sender/per_destination_queue.py b/synapse/federation/sender/per_destination_queue.py index e13cd20ffa..276a2b596f 100644 --- a/synapse/federation/sender/per_destination_queue.py +++ b/synapse/federation/sender/per_destination_queue.py @@ -15,11 +15,10 @@ # limitations under the License. import datetime import logging -from typing import Dict, Hashable, Iterable, List, Tuple +from typing import TYPE_CHECKING, Dict, Hashable, Iterable, List, Tuple from prometheus_client import Counter -import synapse.server from synapse.api.errors import ( FederationDeniedError, HttpResponseException, @@ -34,6 +33,9 @@ from synapse.storage.presence import UserPresenceState from synapse.types import ReadReceipt from synapse.util.retryutils import NotRetryingDestination, get_retry_limiter +if TYPE_CHECKING: + import synapse.server + # This is defined in the Matrix spec and enforced by the receiver. MAX_EDUS_PER_TRANSACTION = 100 diff --git a/synapse/federation/sender/transaction_manager.py b/synapse/federation/sender/transaction_manager.py index 3c2a02a3b3..a2752a54a5 100644 --- a/synapse/federation/sender/transaction_manager.py +++ b/synapse/federation/sender/transaction_manager.py @@ -13,11 +13,10 @@ # See the License for the specific language governing permissions and # limitations under the License. import logging -from typing import List +from typing import TYPE_CHECKING, List from canonicaljson import json -import synapse.server from synapse.api.errors import HttpResponseException from synapse.events import EventBase from synapse.federation.persistence import TransactionActions @@ -31,6 +30,9 @@ from synapse.logging.opentracing import ( ) from synapse.util.metrics import measure_func +if TYPE_CHECKING: + import synapse.server + logger = logging.getLogger(__name__) diff --git a/synapse/replication/tcp/resource.py b/synapse/replication/tcp/resource.py index 33d2f589ac..b690abedad 100644 --- a/synapse/replication/tcp/resource.py +++ b/synapse/replication/tcp/resource.py @@ -80,7 +80,7 @@ class ReplicationStreamer(object): for stream in STREAMS_MAP.values(): if stream == FederationStream and hs.config.send_federation: # We only support federation stream if federation sending - # hase been disabled on the master. + # has been disabled on the master. continue self.streams.append(stream(hs)) diff --git a/synapse/replication/tcp/streams/_base.py b/synapse/replication/tcp/streams/_base.py index b0f87c365b..084604e8b0 100644 --- a/synapse/replication/tcp/streams/_base.py +++ b/synapse/replication/tcp/streams/_base.py @@ -104,7 +104,8 @@ class Stream(object): implemented by subclasses. current_token_function is called to get the current token of the underlying - stream. + stream. It is only meaningful on the process that is the source of the + replication stream (ie, usually the master). update_function is called to get updates for this stream between a pair of stream tokens. See the UpdateFunction type definition for more info. diff --git a/synapse/replication/tcp/streams/federation.py b/synapse/replication/tcp/streams/federation.py index e8bd52e389..b0505b8a2c 100644 --- a/synapse/replication/tcp/streams/federation.py +++ b/synapse/replication/tcp/streams/federation.py @@ -15,7 +15,7 @@ # limitations under the License. from collections import namedtuple -from synapse.replication.tcp.streams._base import Stream, db_query_to_update_function +from synapse.replication.tcp.streams._base import Stream, make_http_update_function class FederationStream(Stream): @@ -35,21 +35,33 @@ class FederationStream(Stream): ROW_TYPE = FederationStreamRow def __init__(self, hs): - # Not all synapse instances will have a federation sender instance, - # whether that's a `FederationSender` or a `FederationRemoteSendQueue`, - # so we stub the stream out when that is the case. - if hs.config.worker_app is None or hs.should_send_federation(): + if hs.config.worker_app is None: + # master process: get updates from the FederationRemoteSendQueue. + # (if the master is configured to send federation itself, federation_sender + # will be a real FederationSender, which has stubs for current_token and + # get_replication_rows.) federation_sender = hs.get_federation_sender() current_token = federation_sender.get_current_token - update_function = db_query_to_update_function( - federation_sender.get_replication_rows - ) + update_function = federation_sender.get_replication_rows + + elif hs.should_send_federation(): + # federation sender: Query master process + update_function = make_http_update_function(hs, self.NAME) + current_token = self._stub_current_token + else: - current_token = lambda: 0 + # other worker: stub out the update function (we're not interested in + # any updates so when we get a POSITION we do nothing) update_function = self._stub_update_function + current_token = self._stub_current_token super().__init__(hs.get_instance_name(), current_token, update_function) + @staticmethod + def _stub_current_token(): + # dummy current-token method for use on workers + return 0 + @staticmethod async def _stub_update_function(instance_name, from_token, upto_token, limit): return [], upto_token, False diff --git a/tests/replication/tcp/streams/_base.py b/tests/replication/tcp/streams/_base.py index 7b56d2028d..9d4f0bbe44 100644 --- a/tests/replication/tcp/streams/_base.py +++ b/tests/replication/tcp/streams/_base.py @@ -27,6 +27,7 @@ from synapse.app.generic_worker import ( GenericWorkerServer, ) from synapse.http.site import SynapseRequest +from synapse.replication.http import streams from synapse.replication.tcp.handler import ReplicationCommandHandler from synapse.replication.tcp.protocol import ClientReplicationStreamProtocol from synapse.replication.tcp.resource import ReplicationStreamProtocolFactory @@ -42,6 +43,10 @@ logger = logging.getLogger(__name__) class BaseStreamTestCase(unittest.HomeserverTestCase): """Base class for tests of the replication streams""" + servlets = [ + streams.register_servlets, + ] + def prepare(self, reactor, clock, hs): # build a replication server server_factory = ReplicationStreamProtocolFactory(hs) @@ -49,17 +54,11 @@ class BaseStreamTestCase(unittest.HomeserverTestCase): self.server = server_factory.buildProtocol(None) # Make a new HomeServer object for the worker - config = self.default_config() - config["worker_app"] = "synapse.app.generic_worker" - config["worker_replication_host"] = "testserv" - config["worker_replication_http_port"] = "8765" - self.reactor.lookups["testserv"] = "1.2.3.4" - self.worker_hs = self.setup_test_homeserver( http_client=None, homeserverToUse=GenericWorkerServer, - config=config, + config=self._get_worker_hs_config(), reactor=self.reactor, ) @@ -78,6 +77,13 @@ class BaseStreamTestCase(unittest.HomeserverTestCase): self._client_transport = None self._server_transport = None + def _get_worker_hs_config(self) -> dict: + config = self.default_config() + config["worker_app"] = "synapse.app.generic_worker" + config["worker_replication_host"] = "testserv" + config["worker_replication_http_port"] = "8765" + return config + def _build_replication_data_handler(self): return TestReplicationDataHandler(self.worker_hs) diff --git a/tests/replication/tcp/streams/test_federation.py b/tests/replication/tcp/streams/test_federation.py new file mode 100644 index 0000000000..eea4565da3 --- /dev/null +++ b/tests/replication/tcp/streams/test_federation.py @@ -0,0 +1,81 @@ +# -*- coding: utf-8 -*- +# Copyright 2019 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from synapse.federation.send_queue import EduRow +from synapse.replication.tcp.streams.federation import FederationStream + +from tests.replication.tcp.streams._base import BaseStreamTestCase + + +class FederationStreamTestCase(BaseStreamTestCase): + def _get_worker_hs_config(self) -> dict: + # enable federation sending on the worker + config = super()._get_worker_hs_config() + # TODO: make it so we don't need both of these + config["send_federation"] = True + config["worker_app"] = "synapse.app.federation_sender" + return config + + def test_catchup(self): + """Basic test of catchup on reconnect + + Makes sure that updates sent while we are offline are received later. + """ + fed_sender = self.hs.get_federation_sender() + received_rows = self.test_handler.received_rdata_rows + + fed_sender.build_and_send_edu("testdest", "m.test_edu", {"a": "b"}) + + self.reconnect() + self.reactor.advance(0) + + # check we're testing what we think we are: no rows should yet have been + # received + self.assertEqual(received_rows, []) + + # We should now see an attempt to connect to the master + request = self.handle_http_replication_attempt() + self.assert_request_is_get_repl_stream_updates(request, "federation") + + # we should have received an update row + stream_name, token, row = received_rows.pop() + self.assertEqual(stream_name, "federation") + self.assertIsInstance(row, FederationStream.FederationStreamRow) + self.assertEqual(row.type, EduRow.TypeId) + edurow = EduRow.from_data(row.data) + self.assertEqual(edurow.edu.edu_type, "m.test_edu") + self.assertEqual(edurow.edu.origin, self.hs.hostname) + self.assertEqual(edurow.edu.destination, "testdest") + self.assertEqual(edurow.edu.content, {"a": "b"}) + + self.assertEqual(received_rows, []) + + # additional updates should be transferred without an HTTP hit + fed_sender.build_and_send_edu("testdest", "m.test1", {"c": "d"}) + self.reactor.advance(0) + # there should be no http hit + self.assertEqual(len(self.reactor.tcpClients), 0) + # ... but we should have a row + self.assertEqual(len(received_rows), 1) + + stream_name, token, row = received_rows.pop() + self.assertEqual(stream_name, "federation") + self.assertIsInstance(row, FederationStream.FederationStreamRow) + self.assertEqual(row.type, EduRow.TypeId) + edurow = EduRow.from_data(row.data) + self.assertEqual(edurow.edu.edu_type, "m.test1") + self.assertEqual(edurow.edu.origin, self.hs.hostname) + self.assertEqual(edurow.edu.destination, "testdest") + self.assertEqual(edurow.edu.content, {"c": "d"}) diff --git a/tests/replication/tcp/streams/test_typing.py b/tests/replication/tcp/streams/test_typing.py index d25a7b194e..125c63dab5 100644 --- a/tests/replication/tcp/streams/test_typing.py +++ b/tests/replication/tcp/streams/test_typing.py @@ -15,7 +15,6 @@ from mock import Mock from synapse.handlers.typing import RoomMember -from synapse.replication.http import streams from synapse.replication.tcp.streams import TypingStream from tests.replication.tcp.streams._base import BaseStreamTestCase @@ -24,10 +23,6 @@ USER_ID = "@feeling:blue" class TypingStreamTestCase(BaseStreamTestCase): - servlets = [ - streams.register_servlets, - ] - def _build_replication_data_handler(self): return Mock(wraps=super()._build_replication_data_handler()) -- cgit 1.4.1 From 7941a70fa8b297b0dec320a9b7dda01df3efe1e4 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Tue, 5 May 2020 14:17:27 +0100 Subject: Fix bug in EventContext.deserialize. (#7393) This caused `prev_state_ids` to be incorrect if the state event was not replacing an existing state entry. --- changelog.d/7393.bugfix | 1 + synapse/events/snapshot.py | 7 ++- tests/events/test_snapshot.py | 100 ++++++++++++++++++++++++++++++++++++ tests/test_utils/event_injection.py | 26 +++++++--- 4 files changed, 126 insertions(+), 8 deletions(-) create mode 100644 changelog.d/7393.bugfix create mode 100644 tests/events/test_snapshot.py (limited to 'tests') diff --git a/changelog.d/7393.bugfix b/changelog.d/7393.bugfix new file mode 100644 index 0000000000..74419af858 --- /dev/null +++ b/changelog.d/7393.bugfix @@ -0,0 +1 @@ +Fix bug in `EventContext.deserialize`. diff --git a/synapse/events/snapshot.py b/synapse/events/snapshot.py index 9ea85e93e6..7c5f620d09 100644 --- a/synapse/events/snapshot.py +++ b/synapse/events/snapshot.py @@ -322,11 +322,14 @@ class _AsyncEventContextImpl(EventContext): self._current_state_ids = yield self._storage.state.get_state_ids_for_group( self.state_group ) - if self._prev_state_id and self._event_state_key is not None: + if self._event_state_key is not None: self._prev_state_ids = dict(self._current_state_ids) key = (self._event_type, self._event_state_key) - self._prev_state_ids[key] = self._prev_state_id + if self._prev_state_id: + self._prev_state_ids[key] = self._prev_state_id + else: + self._prev_state_ids.pop(key, None) else: self._prev_state_ids = self._current_state_ids diff --git a/tests/events/test_snapshot.py b/tests/events/test_snapshot.py new file mode 100644 index 0000000000..640f5f3bce --- /dev/null +++ b/tests/events/test_snapshot.py @@ -0,0 +1,100 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from synapse.events.snapshot import EventContext +from synapse.rest import admin +from synapse.rest.client.v1 import login, room + +from tests import unittest +from tests.test_utils.event_injection import create_event + + +class TestEventContext(unittest.HomeserverTestCase): + servlets = [ + admin.register_servlets, + login.register_servlets, + room.register_servlets, + ] + + def prepare(self, reactor, clock, hs): + self.store = hs.get_datastore() + self.storage = hs.get_storage() + + self.user_id = self.register_user("u1", "pass") + self.user_tok = self.login("u1", "pass") + self.room_id = self.helper.create_room_as(tok=self.user_tok) + + def test_serialize_deserialize_msg(self): + """Test that an EventContext for a message event is the same after + serialize/deserialize. + """ + + event, context = create_event( + self.hs, room_id=self.room_id, type="m.test", sender=self.user_id, + ) + + self._check_serialize_deserialize(event, context) + + def test_serialize_deserialize_state_no_prev(self): + """Test that an EventContext for a state event (with not previous entry) + is the same after serialize/deserialize. + """ + event, context = create_event( + self.hs, + room_id=self.room_id, + type="m.test", + sender=self.user_id, + state_key="", + ) + + self._check_serialize_deserialize(event, context) + + def test_serialize_deserialize_state_prev(self): + """Test that an EventContext for a state event (which replaces a + previous entry) is the same after serialize/deserialize. + """ + event, context = create_event( + self.hs, + room_id=self.room_id, + type="m.room.member", + sender=self.user_id, + state_key=self.user_id, + content={"membership": "leave"}, + ) + + self._check_serialize_deserialize(event, context) + + def _check_serialize_deserialize(self, event, context): + serialized = self.get_success(context.serialize(event, self.store)) + + d_context = EventContext.deserialize(self.storage, serialized) + + self.assertEqual(context.state_group, d_context.state_group) + self.assertEqual(context.rejected, d_context.rejected) + self.assertEqual( + context.state_group_before_event, d_context.state_group_before_event + ) + self.assertEqual(context.prev_group, d_context.prev_group) + self.assertEqual(context.delta_ids, d_context.delta_ids) + self.assertEqual(context.app_service, d_context.app_service) + + self.assertEqual( + self.get_success(context.get_current_state_ids()), + self.get_success(d_context.get_current_state_ids()), + ) + self.assertEqual( + self.get_success(context.get_prev_state_ids()), + self.get_success(d_context.get_prev_state_ids()), + ) diff --git a/tests/test_utils/event_injection.py b/tests/test_utils/event_injection.py index 8f6872761a..431e9f8e5e 100644 --- a/tests/test_utils/event_injection.py +++ b/tests/test_utils/event_injection.py @@ -14,12 +14,13 @@ # See the License for the specific language governing permissions and # limitations under the License. -from typing import Optional +from typing import Optional, Tuple import synapse.server from synapse.api.constants import EventTypes from synapse.api.room_versions import KNOWN_ROOM_VERSIONS from synapse.events import EventBase +from synapse.events.snapshot import EventContext from synapse.types import Collection from tests.test_utils import get_awaitable_result @@ -75,6 +76,23 @@ def inject_event( """ test_reactor = hs.get_reactor() + event, context = create_event(hs, room_version, prev_event_ids, **kwargs) + + d = hs.get_storage().persistence.persist_event(event, context) + test_reactor.advance(0) + get_awaitable_result(d) + + return event + + +def create_event( + hs: synapse.server.HomeServer, + room_version: Optional[str] = None, + prev_event_ids: Optional[Collection[str]] = None, + **kwargs +) -> Tuple[EventBase, EventContext]: + test_reactor = hs.get_reactor() + if room_version is None: d = hs.get_datastore().get_room_version_id(kwargs["room_id"]) test_reactor.advance(0) @@ -89,8 +107,4 @@ def inject_event( test_reactor.advance(0) event, context = get_awaitable_result(d) - d = hs.get_storage().persistence.persist_event(event, context) - test_reactor.advance(0) - get_awaitable_result(d) - - return event + return event, context -- cgit 1.4.1 From aee9130a83eec7a295071a7e3ea87ce380c4521c Mon Sep 17 00:00:00 2001 From: Andrew Morgan <1342360+anoadragon453@users.noreply.github.com> Date: Wed, 6 May 2020 15:54:58 +0100 Subject: Stop Auth methods from polling the config on every req. (#7420) --- changelog.d/7420.misc | 1 + synapse/api/auth.py | 83 +++++----------------------------- synapse/api/auth_blocking.py | 104 +++++++++++++++++++++++++++++++++++++++++++ tests/api/test_auth.py | 36 ++++++++------- tests/handlers/test_auth.py | 23 ++++++---- tests/handlers/test_sync.py | 13 +++--- tests/test_mau.py | 14 ++++-- 7 files changed, 168 insertions(+), 106 deletions(-) create mode 100644 changelog.d/7420.misc create mode 100644 synapse/api/auth_blocking.py (limited to 'tests') diff --git a/changelog.d/7420.misc b/changelog.d/7420.misc new file mode 100644 index 0000000000..e834a9163e --- /dev/null +++ b/changelog.d/7420.misc @@ -0,0 +1 @@ +Prevent methods in `synapse.handlers.auth` from polling the homeserver config every request. \ No newline at end of file diff --git a/synapse/api/auth.py b/synapse/api/auth.py index c5d1eb952b..1ad5ff9410 100644 --- a/synapse/api/auth.py +++ b/synapse/api/auth.py @@ -26,16 +26,15 @@ from twisted.internet import defer import synapse.logging.opentracing as opentracing import synapse.types from synapse import event_auth -from synapse.api.constants import EventTypes, LimitBlockingTypes, Membership, UserTypes +from synapse.api.auth_blocking import AuthBlocking +from synapse.api.constants import EventTypes, Membership from synapse.api.errors import ( AuthError, Codes, InvalidClientTokenError, MissingClientTokenError, - ResourceLimitError, ) from synapse.api.room_versions import KNOWN_ROOM_VERSIONS -from synapse.config.server import is_threepid_reserved from synapse.events import EventBase from synapse.types import StateMap, UserID from synapse.util.caches import CACHE_SIZE_FACTOR, register_cache @@ -77,7 +76,11 @@ class Auth(object): self.token_cache = LruCache(CACHE_SIZE_FACTOR * 10000) register_cache("cache", "token_cache", self.token_cache) + self._auth_blocking = AuthBlocking(self.hs) + self._account_validity = hs.config.account_validity + self._track_appservice_user_ips = hs.config.track_appservice_user_ips + self._macaroon_secret_key = hs.config.macaroon_secret_key @defer.inlineCallbacks def check_from_context(self, room_version: str, event, context, do_sig_check=True): @@ -191,7 +194,7 @@ class Auth(object): opentracing.set_tag("authenticated_entity", user_id) opentracing.set_tag("appservice_id", app_service.id) - if ip_addr and self.hs.config.track_appservice_user_ips: + if ip_addr and self._track_appservice_user_ips: yield self.store.insert_client_ip( user_id=user_id, access_token=access_token, @@ -454,7 +457,7 @@ class Auth(object): # access_tokens include a nonce for uniqueness: any value is acceptable v.satisfy_general(lambda c: c.startswith("nonce = ")) - v.verify(macaroon, self.hs.config.macaroon_secret_key) + v.verify(macaroon, self._macaroon_secret_key) def _verify_expiry(self, caveat): prefix = "time < " @@ -663,71 +666,5 @@ class Auth(object): % (user_id, room_id), ) - @defer.inlineCallbacks - def check_auth_blocking(self, user_id=None, threepid=None, user_type=None): - """Checks if the user should be rejected for some external reason, - such as monthly active user limiting or global disable flag - - Args: - user_id(str|None): If present, checks for presence against existing - MAU cohort - - threepid(dict|None): If present, checks for presence against configured - reserved threepid. Used in cases where the user is trying register - with a MAU blocked server, normally they would be rejected but their - threepid is on the reserved list. user_id and - threepid should never be set at the same time. - - user_type(str|None): If present, is used to decide whether to check against - certain blocking reasons like MAU. - """ - - # Never fail an auth check for the server notices users or support user - # This can be a problem where event creation is prohibited due to blocking - if user_id is not None: - if user_id == self.hs.config.server_notices_mxid: - return - if (yield self.store.is_support_user(user_id)): - return - - if self.hs.config.hs_disabled: - raise ResourceLimitError( - 403, - self.hs.config.hs_disabled_message, - errcode=Codes.RESOURCE_LIMIT_EXCEEDED, - admin_contact=self.hs.config.admin_contact, - limit_type=LimitBlockingTypes.HS_DISABLED, - ) - if self.hs.config.limit_usage_by_mau is True: - assert not (user_id and threepid) - - # If the user is already part of the MAU cohort or a trial user - if user_id: - timestamp = yield self.store.user_last_seen_monthly_active(user_id) - if timestamp: - return - - is_trial = yield self.store.is_trial_user(user_id) - if is_trial: - return - elif threepid: - # If the user does not exist yet, but is signing up with a - # reserved threepid then pass auth check - if is_threepid_reserved( - self.hs.config.mau_limits_reserved_threepids, threepid - ): - return - elif user_type == UserTypes.SUPPORT: - # If the user does not exist yet and is of type "support", - # allow registration. Support users are excluded from MAU checks. - return - # Else if there is no room in the MAU bucket, bail - current_mau = yield self.store.get_monthly_active_count() - if current_mau >= self.hs.config.max_mau_value: - raise ResourceLimitError( - 403, - "Monthly Active User Limit Exceeded", - admin_contact=self.hs.config.admin_contact, - errcode=Codes.RESOURCE_LIMIT_EXCEEDED, - limit_type=LimitBlockingTypes.MONTHLY_ACTIVE_USER, - ) + def check_auth_blocking(self, *args, **kwargs): + return self._auth_blocking.check_auth_blocking(*args, **kwargs) diff --git a/synapse/api/auth_blocking.py b/synapse/api/auth_blocking.py new file mode 100644 index 0000000000..5c499b6b4e --- /dev/null +++ b/synapse/api/auth_blocking.py @@ -0,0 +1,104 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging + +from twisted.internet import defer + +from synapse.api.constants import LimitBlockingTypes, UserTypes +from synapse.api.errors import Codes, ResourceLimitError +from synapse.config.server import is_threepid_reserved + +logger = logging.getLogger(__name__) + + +class AuthBlocking(object): + def __init__(self, hs): + self.store = hs.get_datastore() + + self._server_notices_mxid = hs.config.server_notices_mxid + self._hs_disabled = hs.config.hs_disabled + self._hs_disabled_message = hs.config.hs_disabled_message + self._admin_contact = hs.config.admin_contact + self._max_mau_value = hs.config.max_mau_value + self._limit_usage_by_mau = hs.config.limit_usage_by_mau + self._mau_limits_reserved_threepids = hs.config.mau_limits_reserved_threepids + + @defer.inlineCallbacks + def check_auth_blocking(self, user_id=None, threepid=None, user_type=None): + """Checks if the user should be rejected for some external reason, + such as monthly active user limiting or global disable flag + + Args: + user_id(str|None): If present, checks for presence against existing + MAU cohort + + threepid(dict|None): If present, checks for presence against configured + reserved threepid. Used in cases where the user is trying register + with a MAU blocked server, normally they would be rejected but their + threepid is on the reserved list. user_id and + threepid should never be set at the same time. + + user_type(str|None): If present, is used to decide whether to check against + certain blocking reasons like MAU. + """ + + # Never fail an auth check for the server notices users or support user + # This can be a problem where event creation is prohibited due to blocking + if user_id is not None: + if user_id == self._server_notices_mxid: + return + if (yield self.store.is_support_user(user_id)): + return + + if self._hs_disabled: + raise ResourceLimitError( + 403, + self._hs_disabled_message, + errcode=Codes.RESOURCE_LIMIT_EXCEEDED, + admin_contact=self._admin_contact, + limit_type=LimitBlockingTypes.HS_DISABLED, + ) + if self._limit_usage_by_mau is True: + assert not (user_id and threepid) + + # If the user is already part of the MAU cohort or a trial user + if user_id: + timestamp = yield self.store.user_last_seen_monthly_active(user_id) + if timestamp: + return + + is_trial = yield self.store.is_trial_user(user_id) + if is_trial: + return + elif threepid: + # If the user does not exist yet, but is signing up with a + # reserved threepid then pass auth check + if is_threepid_reserved(self._mau_limits_reserved_threepids, threepid): + return + elif user_type == UserTypes.SUPPORT: + # If the user does not exist yet and is of type "support", + # allow registration. Support users are excluded from MAU checks. + return + # Else if there is no room in the MAU bucket, bail + current_mau = yield self.store.get_monthly_active_count() + if current_mau >= self._max_mau_value: + raise ResourceLimitError( + 403, + "Monthly Active User Limit Exceeded", + admin_contact=self._admin_contact, + errcode=Codes.RESOURCE_LIMIT_EXCEEDED, + limit_type=LimitBlockingTypes.MONTHLY_ACTIVE_USER, + ) diff --git a/tests/api/test_auth.py b/tests/api/test_auth.py index cc0b10e7f6..0bfb86bf1f 100644 --- a/tests/api/test_auth.py +++ b/tests/api/test_auth.py @@ -52,6 +52,10 @@ class AuthTestCase(unittest.TestCase): self.hs.handlers = TestHandlers(self.hs) self.auth = Auth(self.hs) + # AuthBlocking reads from the hs' config on initialization. We need to + # modify its config instead of the hs' + self.auth_blocking = self.auth._auth_blocking + self.test_user = "@foo:bar" self.test_token = b"_test_token_" @@ -321,15 +325,15 @@ class AuthTestCase(unittest.TestCase): @defer.inlineCallbacks def test_blocking_mau(self): - self.hs.config.limit_usage_by_mau = False - self.hs.config.max_mau_value = 50 + self.auth_blocking._limit_usage_by_mau = False + self.auth_blocking._max_mau_value = 50 lots_of_users = 100 small_number_of_users = 1 # Ensure no error thrown yield defer.ensureDeferred(self.auth.check_auth_blocking()) - self.hs.config.limit_usage_by_mau = True + self.auth_blocking._limit_usage_by_mau = True self.store.get_monthly_active_count = Mock( return_value=defer.succeed(lots_of_users) @@ -349,8 +353,8 @@ class AuthTestCase(unittest.TestCase): @defer.inlineCallbacks def test_blocking_mau__depending_on_user_type(self): - self.hs.config.max_mau_value = 50 - self.hs.config.limit_usage_by_mau = True + self.auth_blocking._max_mau_value = 50 + self.auth_blocking._limit_usage_by_mau = True self.store.get_monthly_active_count = Mock(return_value=defer.succeed(100)) # Support users allowed @@ -370,12 +374,12 @@ class AuthTestCase(unittest.TestCase): @defer.inlineCallbacks def test_reserved_threepid(self): - self.hs.config.limit_usage_by_mau = True - self.hs.config.max_mau_value = 1 + self.auth_blocking._limit_usage_by_mau = True + self.auth_blocking._max_mau_value = 1 self.store.get_monthly_active_count = lambda: defer.succeed(2) threepid = {"medium": "email", "address": "reserved@server.com"} unknown_threepid = {"medium": "email", "address": "unreserved@server.com"} - self.hs.config.mau_limits_reserved_threepids = [threepid] + self.auth_blocking._mau_limits_reserved_threepids = [threepid] with self.assertRaises(ResourceLimitError): yield defer.ensureDeferred(self.auth.check_auth_blocking()) @@ -389,8 +393,8 @@ class AuthTestCase(unittest.TestCase): @defer.inlineCallbacks def test_hs_disabled(self): - self.hs.config.hs_disabled = True - self.hs.config.hs_disabled_message = "Reason for being disabled" + self.auth_blocking._hs_disabled = True + self.auth_blocking._hs_disabled_message = "Reason for being disabled" with self.assertRaises(ResourceLimitError) as e: yield defer.ensureDeferred(self.auth.check_auth_blocking()) self.assertEquals(e.exception.admin_contact, self.hs.config.admin_contact) @@ -404,10 +408,10 @@ class AuthTestCase(unittest.TestCase): """ # this should be the default, but we had a bug where the test was doing the wrong # thing, so let's make it explicit - self.hs.config.server_notices_mxid = None + self.auth_blocking._server_notices_mxid = None - self.hs.config.hs_disabled = True - self.hs.config.hs_disabled_message = "Reason for being disabled" + self.auth_blocking._hs_disabled = True + self.auth_blocking._hs_disabled_message = "Reason for being disabled" with self.assertRaises(ResourceLimitError) as e: yield defer.ensureDeferred(self.auth.check_auth_blocking()) self.assertEquals(e.exception.admin_contact, self.hs.config.admin_contact) @@ -416,8 +420,8 @@ class AuthTestCase(unittest.TestCase): @defer.inlineCallbacks def test_server_notices_mxid_special_cased(self): - self.hs.config.hs_disabled = True + self.auth_blocking._hs_disabled = True user = "@user:server" - self.hs.config.server_notices_mxid = user - self.hs.config.hs_disabled_message = "Reason for being disabled" + self.auth_blocking._server_notices_mxid = user + self.auth_blocking._hs_disabled_message = "Reason for being disabled" yield defer.ensureDeferred(self.auth.check_auth_blocking(user)) diff --git a/tests/handlers/test_auth.py b/tests/handlers/test_auth.py index 52c4ac8b11..c01b04e1dc 100644 --- a/tests/handlers/test_auth.py +++ b/tests/handlers/test_auth.py @@ -39,8 +39,13 @@ class AuthTestCase(unittest.TestCase): self.hs.handlers = AuthHandlers(self.hs) self.auth_handler = self.hs.handlers.auth_handler self.macaroon_generator = self.hs.get_macaroon_generator() + # MAU tests - self.hs.config.max_mau_value = 50 + # AuthBlocking reads from the hs' config on initialization. We need to + # modify its config instead of the hs' + self.auth_blocking = self.hs.get_auth()._auth_blocking + self.auth_blocking._max_mau_value = 50 + self.small_number_of_users = 1 self.large_number_of_users = 100 @@ -119,7 +124,7 @@ class AuthTestCase(unittest.TestCase): @defer.inlineCallbacks def test_mau_limits_disabled(self): - self.hs.config.limit_usage_by_mau = False + self.auth_blocking._limit_usage_by_mau = False # Ensure does not throw exception yield defer.ensureDeferred( self.auth_handler.get_access_token_for_user_id( @@ -135,7 +140,7 @@ class AuthTestCase(unittest.TestCase): @defer.inlineCallbacks def test_mau_limits_exceeded_large(self): - self.hs.config.limit_usage_by_mau = True + self.auth_blocking._limit_usage_by_mau = True self.hs.get_datastore().get_monthly_active_count = Mock( return_value=defer.succeed(self.large_number_of_users) ) @@ -159,11 +164,11 @@ class AuthTestCase(unittest.TestCase): @defer.inlineCallbacks def test_mau_limits_parity(self): - self.hs.config.limit_usage_by_mau = True + self.auth_blocking._limit_usage_by_mau = True # If not in monthly active cohort self.hs.get_datastore().get_monthly_active_count = Mock( - return_value=defer.succeed(self.hs.config.max_mau_value) + return_value=defer.succeed(self.auth_blocking._max_mau_value) ) with self.assertRaises(ResourceLimitError): yield defer.ensureDeferred( @@ -173,7 +178,7 @@ class AuthTestCase(unittest.TestCase): ) self.hs.get_datastore().get_monthly_active_count = Mock( - return_value=defer.succeed(self.hs.config.max_mau_value) + return_value=defer.succeed(self.auth_blocking._max_mau_value) ) with self.assertRaises(ResourceLimitError): yield defer.ensureDeferred( @@ -186,7 +191,7 @@ class AuthTestCase(unittest.TestCase): return_value=defer.succeed(self.hs.get_clock().time_msec()) ) self.hs.get_datastore().get_monthly_active_count = Mock( - return_value=defer.succeed(self.hs.config.max_mau_value) + return_value=defer.succeed(self.auth_blocking._max_mau_value) ) yield defer.ensureDeferred( self.auth_handler.get_access_token_for_user_id( @@ -197,7 +202,7 @@ class AuthTestCase(unittest.TestCase): return_value=defer.succeed(self.hs.get_clock().time_msec()) ) self.hs.get_datastore().get_monthly_active_count = Mock( - return_value=defer.succeed(self.hs.config.max_mau_value) + return_value=defer.succeed(self.auth_blocking._max_mau_value) ) yield defer.ensureDeferred( self.auth_handler.validate_short_term_login_token_and_get_user_id( @@ -207,7 +212,7 @@ class AuthTestCase(unittest.TestCase): @defer.inlineCallbacks def test_mau_limits_not_exceeded(self): - self.hs.config.limit_usage_by_mau = True + self.auth_blocking._limit_usage_by_mau = True self.hs.get_datastore().get_monthly_active_count = Mock( return_value=defer.succeed(self.small_number_of_users) diff --git a/tests/handlers/test_sync.py b/tests/handlers/test_sync.py index 4cbe9784ed..e178d7765b 100644 --- a/tests/handlers/test_sync.py +++ b/tests/handlers/test_sync.py @@ -30,28 +30,31 @@ class SyncTestCase(tests.unittest.HomeserverTestCase): self.sync_handler = self.hs.get_sync_handler() self.store = self.hs.get_datastore() - def test_wait_for_sync_for_user_auth_blocking(self): + # AuthBlocking reads from the hs' config on initialization. We need to + # modify its config instead of the hs' + self.auth_blocking = self.hs.get_auth()._auth_blocking + def test_wait_for_sync_for_user_auth_blocking(self): user_id1 = "@user1:test" user_id2 = "@user2:test" sync_config = self._generate_sync_config(user_id1) self.reactor.advance(100) # So we get not 0 time - self.hs.config.limit_usage_by_mau = True - self.hs.config.max_mau_value = 1 + self.auth_blocking._limit_usage_by_mau = True + self.auth_blocking._max_mau_value = 1 # Check that the happy case does not throw errors self.get_success(self.store.upsert_monthly_active_user(user_id1)) self.get_success(self.sync_handler.wait_for_sync_for_user(sync_config)) # Test that global lock works - self.hs.config.hs_disabled = True + self.auth_blocking._hs_disabled = True e = self.get_failure( self.sync_handler.wait_for_sync_for_user(sync_config), ResourceLimitError ) self.assertEquals(e.value.errcode, Codes.RESOURCE_LIMIT_EXCEEDED) - self.hs.config.hs_disabled = False + self.auth_blocking._hs_disabled = False sync_config = self._generate_sync_config(user_id2) diff --git a/tests/test_mau.py b/tests/test_mau.py index 1fbe0d51ff..eb159e3ba5 100644 --- a/tests/test_mau.py +++ b/tests/test_mau.py @@ -19,6 +19,7 @@ import json from mock import Mock +from synapse.api.auth_blocking import AuthBlocking from synapse.api.constants import LoginType from synapse.api.errors import Codes, HttpResponseException, SynapseError from synapse.rest.client.v2_alpha import register, sync @@ -45,11 +46,17 @@ class TestMauLimit(unittest.HomeserverTestCase): self.hs.config.limit_usage_by_mau = True self.hs.config.hs_disabled = False self.hs.config.max_mau_value = 2 - self.hs.config.mau_trial_days = 0 self.hs.config.server_notices_mxid = "@server:red" self.hs.config.server_notices_mxid_display_name = None self.hs.config.server_notices_mxid_avatar_url = None self.hs.config.server_notices_room_name = "Test Server Notice Room" + self.hs.config.mau_trial_days = 0 + + # AuthBlocking reads config options during hs creation. Recreate the + # hs' copy of AuthBlocking after we've updated config values above + self.auth_blocking = AuthBlocking(self.hs) + self.hs.get_auth()._auth_blocking = self.auth_blocking + return self.hs def test_simple_deny_mau(self): @@ -121,6 +128,7 @@ class TestMauLimit(unittest.HomeserverTestCase): self.assertEqual(e.errcode, Codes.RESOURCE_LIMIT_EXCEEDED) def test_trial_users_cant_come_back(self): + self.auth_blocking._mau_trial_days = 1 self.hs.config.mau_trial_days = 1 # We should be able to register more than the limit initially @@ -169,8 +177,8 @@ class TestMauLimit(unittest.HomeserverTestCase): self.assertEqual(e.errcode, Codes.RESOURCE_LIMIT_EXCEEDED) def test_tracked_but_not_limited(self): - self.hs.config.max_mau_value = 1 # should not matter - self.hs.config.limit_usage_by_mau = False + self.auth_blocking._max_mau_value = 1 # should not matter + self.auth_blocking._limit_usage_by_mau = False self.hs.config.mau_stats_only = True # Simply being able to create 2 users indicates that the -- cgit 1.4.1 From a4a5ec4096f8de938f4a6e4264aeaaa0e0b26463 Mon Sep 17 00:00:00 2001 From: Manuel Stahl <37705355+awesome-manuel@users.noreply.github.com> Date: Thu, 7 May 2020 21:33:07 +0200 Subject: Add room details admin endpoint (#7317) --- changelog.d/7317.feature | 1 + docs/admin_api/rooms.md | 54 ++++++++++++++++++++++++++++++++ synapse/rest/admin/__init__.py | 2 ++ synapse/rest/admin/rooms.py | 26 ++++++++++++++- synapse/storage/data_stores/main/room.py | 31 ++++++++++++++++++ tests/rest/admin/test_room.py | 41 ++++++++++++++++++++++++ tests/storage/test_room.py | 11 +++++++ 7 files changed, 165 insertions(+), 1 deletion(-) create mode 100644 changelog.d/7317.feature (limited to 'tests') diff --git a/changelog.d/7317.feature b/changelog.d/7317.feature new file mode 100644 index 0000000000..23c063f280 --- /dev/null +++ b/changelog.d/7317.feature @@ -0,0 +1 @@ +Add room details admin endpoint. Contributed by Awesome Technologies Innovationslabor GmbH. diff --git a/docs/admin_api/rooms.md b/docs/admin_api/rooms.md index 26fe8b8679..624e7745ba 100644 --- a/docs/admin_api/rooms.md +++ b/docs/admin_api/rooms.md @@ -264,3 +264,57 @@ Response: Once the `next_token` parameter is no longer present, we know we've reached the end of the list. + +# DRAFT: Room Details API + +The Room Details admin API allows server admins to get all details of a room. + +This API is still a draft and details might change! + +The following fields are possible in the JSON response body: + +* `room_id` - The ID of the room. +* `name` - The name of the room. +* `canonical_alias` - The canonical (main) alias address of the room. +* `joined_members` - How many users are currently in the room. +* `joined_local_members` - How many local users are currently in the room. +* `version` - The version of the room as a string. +* `creator` - The `user_id` of the room creator. +* `encryption` - Algorithm of end-to-end encryption of messages. Is `null` if encryption is not active. +* `federatable` - Whether users on other servers can join this room. +* `public` - Whether the room is visible in room directory. +* `join_rules` - The type of rules used for users wishing to join this room. One of: ["public", "knock", "invite", "private"]. +* `guest_access` - Whether guests can join the room. One of: ["can_join", "forbidden"]. +* `history_visibility` - Who can see the room history. One of: ["invited", "joined", "shared", "world_readable"]. +* `state_events` - Total number of state_events of a room. Complexity of the room. + +## Usage + +A standard request: + +``` +GET /_synapse/admin/v1/rooms/ + +{} +``` + +Response: + +``` +{ + "room_id": "!mscvqgqpHYjBGDxNym:matrix.org", + "name": "Music Theory", + "canonical_alias": "#musictheory:matrix.org", + "joined_members": 127 + "joined_local_members": 2, + "version": "1", + "creator": "@foo:matrix.org", + "encryption": null, + "federatable": true, + "public": true, + "join_rules": "invite", + "guest_access": null, + "history_visibility": "shared", + "state_events": 93534 +} +``` diff --git a/synapse/rest/admin/__init__.py b/synapse/rest/admin/__init__.py index ed70d448a1..6b85148a32 100644 --- a/synapse/rest/admin/__init__.py +++ b/synapse/rest/admin/__init__.py @@ -32,6 +32,7 @@ from synapse.rest.admin.purge_room_servlet import PurgeRoomServlet from synapse.rest.admin.rooms import ( JoinRoomAliasServlet, ListRoomRestServlet, + RoomRestServlet, ShutdownRoomRestServlet, ) from synapse.rest.admin.server_notice_servlet import SendServerNoticeServlet @@ -193,6 +194,7 @@ def register_servlets(hs, http_server): """ register_servlets_for_client_rest_resource(hs, http_server) ListRoomRestServlet(hs).register(http_server) + RoomRestServlet(hs).register(http_server) JoinRoomAliasServlet(hs).register(http_server) PurgeRoomServlet(hs).register(http_server) SendServerNoticeServlet(hs).register(http_server) diff --git a/synapse/rest/admin/rooms.py b/synapse/rest/admin/rooms.py index d1bdb64111..7d40001988 100644 --- a/synapse/rest/admin/rooms.py +++ b/synapse/rest/admin/rooms.py @@ -26,6 +26,7 @@ from synapse.http.servlet import ( ) from synapse.rest.admin._base import ( admin_patterns, + assert_requester_is_admin, assert_user_is_admin, historical_admin_path_patterns, ) @@ -169,7 +170,7 @@ class ListRoomRestServlet(RestServlet): in a dictionary containing room information. Supports pagination. """ - PATTERNS = admin_patterns("/rooms") + PATTERNS = admin_patterns("/rooms$") def __init__(self, hs): self.store = hs.get_datastore() @@ -253,6 +254,29 @@ class ListRoomRestServlet(RestServlet): return 200, response +class RoomRestServlet(RestServlet): + """Get room details. + + TODO: Add on_POST to allow room creation without joining the room + """ + + PATTERNS = admin_patterns("/rooms/(?P[^/]+)$") + + def __init__(self, hs): + self.hs = hs + self.auth = hs.get_auth() + self.store = hs.get_datastore() + + async def on_GET(self, request, room_id): + await assert_requester_is_admin(self.auth, request) + + ret = await self.store.get_room_with_stats(room_id) + if not ret: + raise NotFoundError("Room not found") + + return 200, ret + + class JoinRoomAliasServlet(RestServlet): PATTERNS = admin_patterns("/join/(?P[^/]*)") diff --git a/synapse/storage/data_stores/main/room.py b/synapse/storage/data_stores/main/room.py index 147eba1df7..cafa664c16 100644 --- a/synapse/storage/data_stores/main/room.py +++ b/synapse/storage/data_stores/main/room.py @@ -98,6 +98,37 @@ class RoomWorkerStore(SQLBaseStore): allow_none=True, ) + def get_room_with_stats(self, room_id: str): + """Retrieve room with statistics. + + Args: + room_id: The ID of the room to retrieve. + Returns: + A dict containing the room information, or None if the room is unknown. + """ + + def get_room_with_stats_txn(txn, room_id): + sql = """ + SELECT room_id, state.name, state.canonical_alias, curr.joined_members, + curr.local_users_in_room AS joined_local_members, rooms.room_version AS version, + rooms.creator, state.encryption, state.is_federatable AS federatable, + rooms.is_public AS public, state.join_rules, state.guest_access, + state.history_visibility, curr.current_state_events AS state_events + FROM rooms + LEFT JOIN room_stats_state state USING (room_id) + LEFT JOIN room_stats_current curr USING (room_id) + WHERE room_id = ? + """ + txn.execute(sql, [room_id]) + res = self.db.cursor_to_dict(txn)[0] + res["federatable"] = bool(res["federatable"]) + res["public"] = bool(res["public"]) + return res + + return self.db.runInteraction( + "get_room_with_stats", get_room_with_stats_txn, room_id + ) + def get_public_room_ids(self): return self.db.simple_select_onecol( table="rooms", diff --git a/tests/rest/admin/test_room.py b/tests/rest/admin/test_room.py index 249c93722f..54cd24bf64 100644 --- a/tests/rest/admin/test_room.py +++ b/tests/rest/admin/test_room.py @@ -701,6 +701,47 @@ class RoomTestCase(unittest.HomeserverTestCase): _search_test(None, "bar") _search_test(None, "", expected_http_code=400) + def test_single_room(self): + """Test that a single room can be requested correctly""" + # Create two test rooms + room_id_1 = self.helper.create_room_as(self.admin_user, tok=self.admin_user_tok) + room_id_2 = self.helper.create_room_as(self.admin_user, tok=self.admin_user_tok) + + room_name_1 = "something" + room_name_2 = "else" + + # Set the name for each room + self.helper.send_state( + room_id_1, "m.room.name", {"name": room_name_1}, tok=self.admin_user_tok, + ) + self.helper.send_state( + room_id_2, "m.room.name", {"name": room_name_2}, tok=self.admin_user_tok, + ) + + url = "/_synapse/admin/v1/rooms/%s" % (room_id_1,) + request, channel = self.make_request( + "GET", url.encode("ascii"), access_token=self.admin_user_tok, + ) + self.render(request) + self.assertEqual(200, channel.code, msg=channel.json_body) + + self.assertIn("room_id", channel.json_body) + self.assertIn("name", channel.json_body) + self.assertIn("canonical_alias", channel.json_body) + self.assertIn("joined_members", channel.json_body) + self.assertIn("joined_local_members", channel.json_body) + self.assertIn("version", channel.json_body) + self.assertIn("creator", channel.json_body) + self.assertIn("encryption", channel.json_body) + self.assertIn("federatable", channel.json_body) + self.assertIn("public", channel.json_body) + self.assertIn("join_rules", channel.json_body) + self.assertIn("guest_access", channel.json_body) + self.assertIn("history_visibility", channel.json_body) + self.assertIn("state_events", channel.json_body) + + self.assertEqual(room_id_1, channel.json_body["room_id"]) + class JoinAliasRoomTestCase(unittest.HomeserverTestCase): diff --git a/tests/storage/test_room.py b/tests/storage/test_room.py index 086adeb8fd..3b78d48896 100644 --- a/tests/storage/test_room.py +++ b/tests/storage/test_room.py @@ -55,6 +55,17 @@ class RoomStoreTestCase(unittest.TestCase): (yield self.store.get_room(self.room.to_string())), ) + @defer.inlineCallbacks + def test_get_room_with_stats(self): + self.assertDictContainsSubset( + { + "room_id": self.room.to_string(), + "creator": self.u_creator.to_string(), + "public": True, + }, + (yield self.store.get_room_with_stats(self.room.to_string())), + ) + class RoomEventsStoreTestCase(unittest.TestCase): @defer.inlineCallbacks -- cgit 1.4.1 From 616af44137c78d481024da83bb51ed0d50a49522 Mon Sep 17 00:00:00 2001 From: Quentin Gliech Date: Fri, 8 May 2020 14:30:40 +0200 Subject: Implement OpenID Connect-based login (#7256) --- changelog.d/7256.feature | 1 + docs/dev/oidc.md | 175 ++++++ docs/sample_config.yaml | 95 ++++ mypy.ini | 3 + synapse/app/homeserver.py | 12 + synapse/config/_base.pyi | 2 + synapse/config/homeserver.py | 2 + synapse/config/oidc_config.py | 177 ++++++ synapse/config/sso.py | 17 +- synapse/handlers/auth.py | 4 +- synapse/handlers/oidc_handler.py | 998 +++++++++++++++++++++++++++++++++ synapse/http/client.py | 7 + synapse/python_dependencies.py | 1 + synapse/res/templates/sso_error.html | 18 + synapse/rest/client/v1/login.py | 28 +- synapse/rest/oidc/__init__.py | 27 + synapse/rest/oidc/callback_resource.py | 31 + synapse/server.py | 6 + synapse/server.pyi | 5 + tests/handlers/test_oidc.py | 565 +++++++++++++++++++ tox.ini | 1 + 21 files changed, 2163 insertions(+), 12 deletions(-) create mode 100644 changelog.d/7256.feature create mode 100644 docs/dev/oidc.md create mode 100644 synapse/config/oidc_config.py create mode 100644 synapse/handlers/oidc_handler.py create mode 100644 synapse/res/templates/sso_error.html create mode 100644 synapse/rest/oidc/__init__.py create mode 100644 synapse/rest/oidc/callback_resource.py create mode 100644 tests/handlers/test_oidc.py (limited to 'tests') diff --git a/changelog.d/7256.feature b/changelog.d/7256.feature new file mode 100644 index 0000000000..7ad767bf71 --- /dev/null +++ b/changelog.d/7256.feature @@ -0,0 +1 @@ +Add OpenID Connect login/registration support. Contributed by Quentin Gliech, on behalf of [les Connecteurs](https://connecteu.rs). diff --git a/docs/dev/oidc.md b/docs/dev/oidc.md new file mode 100644 index 0000000000..a90c5d2441 --- /dev/null +++ b/docs/dev/oidc.md @@ -0,0 +1,175 @@ +# How to test OpenID Connect + +Any OpenID Connect Provider (OP) should work with Synapse, as long as it supports the authorization code flow. +There are a few options for that: + + - start a local OP. Synapse has been tested with [Hydra][hydra] and [Dex][dex-idp]. + Note that for an OP to work, it should be served under a secure (HTTPS) origin. + A certificate signed with a self-signed, locally trusted CA should work. In that case, start Synapse with a `SSL_CERT_FILE` environment variable set to the path of the CA. + - use a publicly available OP. Synapse has been tested with [Google][google-idp]. + - setup a SaaS OP, like [Auth0][auth0] and [Okta][okta]. Auth0 has a free tier which has been tested with Synapse. + +[google-idp]: https://developers.google.com/identity/protocols/OpenIDConnect#authenticatingtheuser +[auth0]: https://auth0.com/ +[okta]: https://www.okta.com/ +[dex-idp]: https://github.com/dexidp/dex +[hydra]: https://www.ory.sh/docs/hydra/ + + +## Sample configs + +Here are a few configs for providers that should work with Synapse. + +### [Dex][dex-idp] + +[Dex][dex-idp] is a simple, open-source, certified OpenID Connect Provider. +Although it is designed to help building a full-blown provider, with some external database, it can be configured with static passwords in a config file. + +Follow the [Getting Started guide](https://github.com/dexidp/dex/blob/master/Documentation/getting-started.md) to install Dex. + +Edit `examples/config-dev.yaml` config file from the Dex repo to add a client: + +```yaml +staticClients: +- id: synapse + secret: secret + redirectURIs: + - '[synapse base url]/_synapse/oidc/callback' + name: 'Synapse' +``` + +Run with `dex serve examples/config-dex.yaml` + +Synapse config: + +```yaml +oidc_config: + enabled: true + skip_verification: true # This is needed as Dex is served on an insecure endpoint + issuer: "http://127.0.0.1:5556/dex" + discover: true + client_id: "synapse" + client_secret: "secret" + scopes: + - openid + - profile + user_mapping_provider: + config: + localpart_template: '{{ user.name }}' + display_name_template: '{{ user.name|capitalize }}' +``` + +### [Auth0][auth0] + +1. Create a regular web application for Synapse +2. Set the Allowed Callback URLs to `[synapse base url]/_synapse/oidc/callback` +3. Add a rule to add the `preferred_username` claim. +
+ Code sample + + ```js + function addPersistenceAttribute(user, context, callback) { + user.user_metadata = user.user_metadata || {}; + user.user_metadata.preferred_username = user.user_metadata.preferred_username || user.user_id; + context.idToken.preferred_username = user.user_metadata.preferred_username; + + auth0.users.updateUserMetadata(user.user_id, user.user_metadata) + .then(function(){ + callback(null, user, context); + }) + .catch(function(err){ + callback(err); + }); + } + ``` + +
+ + +```yaml +oidc_config: + enabled: true + issuer: "https://your-tier.eu.auth0.com/" # TO BE FILLED + discover: true + client_id: "your-client-id" # TO BE FILLED + client_secret: "your-client-secret" # TO BE FILLED + scopes: + - openid + - profile + user_mapping_provider: + config: + localpart_template: '{{ user.preferred_username }}' + display_name_template: '{{ user.name }}' +``` + +### GitHub + +GitHub is a bit special as it is not an OpenID Connect compliant provider, but just a regular OAuth2 provider. +The `/user` API endpoint can be used to retrieve informations from the user. +As the OIDC login mechanism needs an attribute to uniquely identify users and that endpoint does not return a `sub` property, an alternative `subject_claim` has to be set. + +1. Create a new OAuth application: https://github.com/settings/applications/new +2. Set the callback URL to `[synapse base url]/_synapse/oidc/callback` + +```yaml +oidc_config: + enabled: true + issuer: "https://github.com/" + discover: false + client_id: "your-client-id" # TO BE FILLED + client_secret: "your-client-secret" # TO BE FILLED + authorization_endpoint: "https://github.com/login/oauth/authorize" + token_endpoint: "https://github.com/login/oauth/access_token" + userinfo_endpoint: "https://api.github.com/user" + scopes: + - read:user + user_mapping_provider: + config: + subject_claim: 'id' + localpart_template: '{{ user.login }}' + display_name_template: '{{ user.name }}' +``` + +### Google + +1. Setup a project in the Google API Console +2. Obtain the OAuth 2.0 credentials (see ) +3. Add this Authorized redirect URI: `[synapse base url]/_synapse/oidc/callback` + +```yaml +oidc_config: + enabled: true + issuer: "https://accounts.google.com/" + discover: true + client_id: "your-client-id" # TO BE FILLED + client_secret: "your-client-secret" # TO BE FILLED + scopes: + - openid + - profile + user_mapping_provider: + config: + localpart_template: '{{ user.given_name|lower }}' + display_name_template: '{{ user.name }}' +``` + +### Twitch + +1. Setup a developer account on [Twitch](https://dev.twitch.tv/) +2. Obtain the OAuth 2.0 credentials by [creating an app](https://dev.twitch.tv/console/apps/) +3. Add this OAuth Redirect URL: `[synapse base url]/_synapse/oidc/callback` + +```yaml +oidc_config: + enabled: true + issuer: "https://id.twitch.tv/oauth2/" + discover: true + client_id: "your-client-id" # TO BE FILLED + client_secret: "your-client-secret" # TO BE FILLED + client_auth_method: "client_secret_post" + scopes: + - openid + user_mapping_provider: + config: + localpart_template: '{{ user.preferred_username }}' + display_name_template: '{{ user.name }}' +``` diff --git a/docs/sample_config.yaml b/docs/sample_config.yaml index 98ead7dc0e..1e397f7734 100644 --- a/docs/sample_config.yaml +++ b/docs/sample_config.yaml @@ -1470,6 +1470,94 @@ saml2_config: #template_dir: "res/templates" +# Enable OpenID Connect for registration and login. Uses authlib. +# +oidc_config: + # enable OpenID Connect. Defaults to false. + # + #enabled: true + + # use the OIDC discovery mechanism to discover endpoints. Defaults to true. + # + #discover: true + + # the OIDC issuer. Used to validate tokens and discover the providers endpoints. Required. + # + #issuer: "https://accounts.example.com/" + + # oauth2 client id to use. Required. + # + #client_id: "provided-by-your-issuer" + + # oauth2 client secret to use. Required. + # + #client_secret: "provided-by-your-issuer" + + # auth method to use when exchanging the token. + # Valid values are "client_secret_basic" (default), "client_secret_post" and "none". + # + #client_auth_method: "client_auth_basic" + + # list of scopes to ask. This should include the "openid" scope. Defaults to ["openid"]. + # + #scopes: ["openid"] + + # the oauth2 authorization endpoint. Required if provider discovery is disabled. + # + #authorization_endpoint: "https://accounts.example.com/oauth2/auth" + + # the oauth2 token endpoint. Required if provider discovery is disabled. + # + #token_endpoint: "https://accounts.example.com/oauth2/token" + + # the OIDC userinfo endpoint. Required if discovery is disabled and the "openid" scope is not asked. + # + #userinfo_endpoint: "https://accounts.example.com/userinfo" + + # URI where to fetch the JWKS. Required if discovery is disabled and the "openid" scope is used. + # + #jwks_uri: "https://accounts.example.com/.well-known/jwks.json" + + # skip metadata verification. Defaults to false. + # Use this if you are connecting to a provider that is not OpenID Connect compliant. + # Avoid this in production. + # + #skip_verification: false + + + # An external module can be provided here as a custom solution to mapping + # attributes returned from a OIDC provider onto a matrix user. + # + user_mapping_provider: + # The custom module's class. Uncomment to use a custom module. + # Default is 'synapse.handlers.oidc_handler.JinjaOidcMappingProvider'. + # + #module: mapping_provider.OidcMappingProvider + + # Custom configuration values for the module. Below options are intended + # for the built-in provider, they should be changed if using a custom + # module. This section will be passed as a Python dictionary to the + # module's `parse_config` method. + # + # Below is the config of the default mapping provider, based on Jinja2 + # templates. Those templates are used to render user attributes, where the + # userinfo object is available through the `user` variable. + # + config: + # name of the claim containing a unique identifier for the user. + # Defaults to `sub`, which OpenID Connect compliant providers should provide. + # + #subject_claim: "sub" + + # Jinja2 template for the localpart of the MXID + # + localpart_template: "{{ user.preferred_username }}" + + # Jinja2 template for the display name to set on first login. Optional. + # + #display_name_template: "{{ user.given_name }} {{ user.last_name }}" + + # Enable CAS for registration and login. # @@ -1554,6 +1642,13 @@ sso: # # This template has no additional variables. # + # * HTML page to display to users if something goes wrong during the + # OpenID Connect authentication process: 'sso_error.html'. + # + # When rendering, this template is given two variables: + # * error: the technical name of the error + # * error_description: a human-readable message for the error + # # You can see the default templates at: # https://github.com/matrix-org/synapse/tree/master/synapse/res/templates # diff --git a/mypy.ini b/mypy.ini index 69be2f67ad..3533797d68 100644 --- a/mypy.ini +++ b/mypy.ini @@ -75,3 +75,6 @@ ignore_missing_imports = True [mypy-jwt.*] ignore_missing_imports = True + +[mypy-authlib.*] +ignore_missing_imports = True diff --git a/synapse/app/homeserver.py b/synapse/app/homeserver.py index cbd1ea475a..bc8695d8dd 100644 --- a/synapse/app/homeserver.py +++ b/synapse/app/homeserver.py @@ -192,6 +192,11 @@ class SynapseHomeServer(HomeServer): } ) + if self.get_config().oidc_enabled: + from synapse.rest.oidc import OIDCResource + + resources["/_synapse/oidc"] = OIDCResource(self) + if self.get_config().saml2_enabled: from synapse.rest.saml2 import SAML2Resource @@ -422,6 +427,13 @@ def setup(config_options): # Check if it needs to be reprovisioned every day. hs.get_clock().looping_call(reprovision_acme, 24 * 60 * 60 * 1000) + # Load the OIDC provider metadatas, if OIDC is enabled. + if hs.config.oidc_enabled: + oidc = hs.get_oidc_handler() + # Loading the provider metadata also ensures the provider config is valid. + yield defer.ensureDeferred(oidc.load_metadata()) + yield defer.ensureDeferred(oidc.load_jwks()) + _base.start(hs, config.listeners) hs.get_datastore().db.updates.start_doing_background_updates() diff --git a/synapse/config/_base.pyi b/synapse/config/_base.pyi index 3053fc9d27..9e576060d4 100644 --- a/synapse/config/_base.pyi +++ b/synapse/config/_base.pyi @@ -13,6 +13,7 @@ from synapse.config import ( key, logger, metrics, + oidc_config, password, password_auth_providers, push, @@ -59,6 +60,7 @@ class RootConfig: saml2: saml2_config.SAML2Config cas: cas.CasConfig sso: sso.SSOConfig + oidc: oidc_config.OIDCConfig jwt: jwt_config.JWTConfig password: password.PasswordConfig email: emailconfig.EmailConfig diff --git a/synapse/config/homeserver.py b/synapse/config/homeserver.py index be6c6afa74..996d3e6bf7 100644 --- a/synapse/config/homeserver.py +++ b/synapse/config/homeserver.py @@ -27,6 +27,7 @@ from .jwt_config import JWTConfig from .key import KeyConfig from .logger import LoggingConfig from .metrics import MetricsConfig +from .oidc_config import OIDCConfig from .password import PasswordConfig from .password_auth_providers import PasswordAuthProviderConfig from .push import PushConfig @@ -66,6 +67,7 @@ class HomeServerConfig(RootConfig): AppServiceConfig, KeyConfig, SAML2Config, + OIDCConfig, CasConfig, SSOConfig, JWTConfig, diff --git a/synapse/config/oidc_config.py b/synapse/config/oidc_config.py new file mode 100644 index 0000000000..5af110745e --- /dev/null +++ b/synapse/config/oidc_config.py @@ -0,0 +1,177 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 Quentin Gliech +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from synapse.python_dependencies import DependencyException, check_requirements +from synapse.util.module_loader import load_module + +from ._base import Config, ConfigError + +DEFAULT_USER_MAPPING_PROVIDER = "synapse.handlers.oidc_handler.JinjaOidcMappingProvider" + + +class OIDCConfig(Config): + section = "oidc" + + def read_config(self, config, **kwargs): + self.oidc_enabled = False + + oidc_config = config.get("oidc_config") + + if not oidc_config or not oidc_config.get("enabled", False): + return + + try: + check_requirements("oidc") + except DependencyException as e: + raise ConfigError(e.message) + + public_baseurl = self.public_baseurl + if public_baseurl is None: + raise ConfigError("oidc_config requires a public_baseurl to be set") + self.oidc_callback_url = public_baseurl + "_synapse/oidc/callback" + + self.oidc_enabled = True + self.oidc_discover = oidc_config.get("discover", True) + self.oidc_issuer = oidc_config["issuer"] + self.oidc_client_id = oidc_config["client_id"] + self.oidc_client_secret = oidc_config["client_secret"] + self.oidc_client_auth_method = oidc_config.get( + "client_auth_method", "client_secret_basic" + ) + self.oidc_scopes = oidc_config.get("scopes", ["openid"]) + self.oidc_authorization_endpoint = oidc_config.get("authorization_endpoint") + self.oidc_token_endpoint = oidc_config.get("token_endpoint") + self.oidc_userinfo_endpoint = oidc_config.get("userinfo_endpoint") + self.oidc_jwks_uri = oidc_config.get("jwks_uri") + self.oidc_subject_claim = oidc_config.get("subject_claim", "sub") + self.oidc_skip_verification = oidc_config.get("skip_verification", False) + + ump_config = oidc_config.get("user_mapping_provider", {}) + ump_config.setdefault("module", DEFAULT_USER_MAPPING_PROVIDER) + ump_config.setdefault("config", {}) + + ( + self.oidc_user_mapping_provider_class, + self.oidc_user_mapping_provider_config, + ) = load_module(ump_config) + + # Ensure loaded user mapping module has defined all necessary methods + required_methods = [ + "get_remote_user_id", + "map_user_attributes", + ] + missing_methods = [ + method + for method in required_methods + if not hasattr(self.oidc_user_mapping_provider_class, method) + ] + if missing_methods: + raise ConfigError( + "Class specified by oidc_config." + "user_mapping_provider.module is missing required " + "methods: %s" % (", ".join(missing_methods),) + ) + + def generate_config_section(self, config_dir_path, server_name, **kwargs): + return """\ + # Enable OpenID Connect for registration and login. Uses authlib. + # + oidc_config: + # enable OpenID Connect. Defaults to false. + # + #enabled: true + + # use the OIDC discovery mechanism to discover endpoints. Defaults to true. + # + #discover: true + + # the OIDC issuer. Used to validate tokens and discover the providers endpoints. Required. + # + #issuer: "https://accounts.example.com/" + + # oauth2 client id to use. Required. + # + #client_id: "provided-by-your-issuer" + + # oauth2 client secret to use. Required. + # + #client_secret: "provided-by-your-issuer" + + # auth method to use when exchanging the token. + # Valid values are "client_secret_basic" (default), "client_secret_post" and "none". + # + #client_auth_method: "client_auth_basic" + + # list of scopes to ask. This should include the "openid" scope. Defaults to ["openid"]. + # + #scopes: ["openid"] + + # the oauth2 authorization endpoint. Required if provider discovery is disabled. + # + #authorization_endpoint: "https://accounts.example.com/oauth2/auth" + + # the oauth2 token endpoint. Required if provider discovery is disabled. + # + #token_endpoint: "https://accounts.example.com/oauth2/token" + + # the OIDC userinfo endpoint. Required if discovery is disabled and the "openid" scope is not asked. + # + #userinfo_endpoint: "https://accounts.example.com/userinfo" + + # URI where to fetch the JWKS. Required if discovery is disabled and the "openid" scope is used. + # + #jwks_uri: "https://accounts.example.com/.well-known/jwks.json" + + # skip metadata verification. Defaults to false. + # Use this if you are connecting to a provider that is not OpenID Connect compliant. + # Avoid this in production. + # + #skip_verification: false + + + # An external module can be provided here as a custom solution to mapping + # attributes returned from a OIDC provider onto a matrix user. + # + user_mapping_provider: + # The custom module's class. Uncomment to use a custom module. + # Default is {mapping_provider!r}. + # + #module: mapping_provider.OidcMappingProvider + + # Custom configuration values for the module. Below options are intended + # for the built-in provider, they should be changed if using a custom + # module. This section will be passed as a Python dictionary to the + # module's `parse_config` method. + # + # Below is the config of the default mapping provider, based on Jinja2 + # templates. Those templates are used to render user attributes, where the + # userinfo object is available through the `user` variable. + # + config: + # name of the claim containing a unique identifier for the user. + # Defaults to `sub`, which OpenID Connect compliant providers should provide. + # + #subject_claim: "sub" + + # Jinja2 template for the localpart of the MXID + # + localpart_template: "{{{{ user.preferred_username }}}}" + + # Jinja2 template for the display name to set on first login. Optional. + # + #display_name_template: "{{{{ user.given_name }}}} {{{{ user.last_name }}}}" + """.format( + mapping_provider=DEFAULT_USER_MAPPING_PROVIDER + ) diff --git a/synapse/config/sso.py b/synapse/config/sso.py index cac6bc0139..aff642f015 100644 --- a/synapse/config/sso.py +++ b/synapse/config/sso.py @@ -36,17 +36,13 @@ class SSOConfig(Config): if not template_dir: template_dir = pkg_resources.resource_filename("synapse", "res/templates",) - self.sso_redirect_confirm_template_dir = template_dir + self.sso_template_dir = template_dir self.sso_account_deactivated_template = self.read_file( - os.path.join( - self.sso_redirect_confirm_template_dir, "sso_account_deactivated.html" - ), + os.path.join(self.sso_template_dir, "sso_account_deactivated.html"), "sso_account_deactivated_template", ) self.sso_auth_success_template = self.read_file( - os.path.join( - self.sso_redirect_confirm_template_dir, "sso_auth_success.html" - ), + os.path.join(self.sso_template_dir, "sso_auth_success.html"), "sso_auth_success_template", ) @@ -137,6 +133,13 @@ class SSOConfig(Config): # # This template has no additional variables. # + # * HTML page to display to users if something goes wrong during the + # OpenID Connect authentication process: 'sso_error.html'. + # + # When rendering, this template is given two variables: + # * error: the technical name of the error + # * error_description: a human-readable message for the error + # # You can see the default templates at: # https://github.com/matrix-org/synapse/tree/master/synapse/res/templates # diff --git a/synapse/handlers/auth.py b/synapse/handlers/auth.py index 7613e5b6ab..f8d2331bf1 100644 --- a/synapse/handlers/auth.py +++ b/synapse/handlers/auth.py @@ -126,13 +126,13 @@ class AuthHandler(BaseHandler): # It notifies the user they are about to give access to their matrix account # to the client. self._sso_redirect_confirm_template = load_jinja2_templates( - hs.config.sso_redirect_confirm_template_dir, ["sso_redirect_confirm.html"], + hs.config.sso_template_dir, ["sso_redirect_confirm.html"], )[0] # The following template is shown during user interactive authentication # in the fallback auth scenario. It notifies the user that they are # authenticating for an operation to occur on their account. self._sso_auth_confirm_template = load_jinja2_templates( - hs.config.sso_redirect_confirm_template_dir, ["sso_auth_confirm.html"], + hs.config.sso_template_dir, ["sso_auth_confirm.html"], )[0] # The following template is shown after a successful user interactive # authentication session. It tells the user they can close the window. diff --git a/synapse/handlers/oidc_handler.py b/synapse/handlers/oidc_handler.py new file mode 100644 index 0000000000..178f263439 --- /dev/null +++ b/synapse/handlers/oidc_handler.py @@ -0,0 +1,998 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 Quentin Gliech +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import json +import logging +from typing import Dict, Generic, List, Optional, Tuple, TypeVar +from urllib.parse import urlencode + +import attr +import pymacaroons +from authlib.common.security import generate_token +from authlib.jose import JsonWebToken +from authlib.oauth2.auth import ClientAuth +from authlib.oauth2.rfc6749.parameters import prepare_grant_uri +from authlib.oidc.core import CodeIDToken, ImplicitIDToken, UserInfo +from authlib.oidc.discovery import OpenIDProviderMetadata, get_well_known_url +from jinja2 import Environment, Template +from pymacaroons.exceptions import ( + MacaroonDeserializationException, + MacaroonInvalidSignatureException, +) +from typing_extensions import TypedDict + +from twisted.web.client import readBody + +from synapse.config import ConfigError +from synapse.http.server import finish_request +from synapse.http.site import SynapseRequest +from synapse.push.mailer import load_jinja2_templates +from synapse.server import HomeServer +from synapse.types import UserID, map_username_to_mxid_localpart + +logger = logging.getLogger(__name__) + +SESSION_COOKIE_NAME = b"oidc_session" + +#: A token exchanged from the token endpoint, as per RFC6749 sec 5.1. and +#: OpenID.Core sec 3.1.3.3. +Token = TypedDict( + "Token", + { + "access_token": str, + "token_type": str, + "id_token": Optional[str], + "refresh_token": Optional[str], + "expires_in": int, + "scope": Optional[str], + }, +) + +#: A JWK, as per RFC7517 sec 4. The type could be more precise than that, but +#: there is no real point of doing this in our case. +JWK = Dict[str, str] + +#: A JWK Set, as per RFC7517 sec 5. +JWKS = TypedDict("JWKS", {"keys": List[JWK]}) + + +class OidcError(Exception): + """Used to catch errors when calling the token_endpoint + """ + + def __init__(self, error, error_description=None): + self.error = error + self.error_description = error_description + + def __str__(self): + if self.error_description: + return "{}: {}".format(self.error, self.error_description) + return self.error + + +class MappingException(Exception): + """Used to catch errors when mapping the UserInfo object + """ + + +class OidcHandler: + """Handles requests related to the OpenID Connect login flow. + """ + + def __init__(self, hs: HomeServer): + self._callback_url = hs.config.oidc_callback_url # type: str + self._scopes = hs.config.oidc_scopes # type: List[str] + self._client_auth = ClientAuth( + hs.config.oidc_client_id, + hs.config.oidc_client_secret, + hs.config.oidc_client_auth_method, + ) # type: ClientAuth + self._client_auth_method = hs.config.oidc_client_auth_method # type: str + self._subject_claim = hs.config.oidc_subject_claim + self._provider_metadata = OpenIDProviderMetadata( + issuer=hs.config.oidc_issuer, + authorization_endpoint=hs.config.oidc_authorization_endpoint, + token_endpoint=hs.config.oidc_token_endpoint, + userinfo_endpoint=hs.config.oidc_userinfo_endpoint, + jwks_uri=hs.config.oidc_jwks_uri, + ) # type: OpenIDProviderMetadata + self._provider_needs_discovery = hs.config.oidc_discover # type: bool + self._user_mapping_provider = hs.config.oidc_user_mapping_provider_class( + hs.config.oidc_user_mapping_provider_config + ) # type: OidcMappingProvider + self._skip_verification = hs.config.oidc_skip_verification # type: bool + + self._http_client = hs.get_proxied_http_client() + self._auth_handler = hs.get_auth_handler() + self._registration_handler = hs.get_registration_handler() + self._datastore = hs.get_datastore() + self._clock = hs.get_clock() + self._hostname = hs.hostname # type: str + self._server_name = hs.config.server_name # type: str + self._macaroon_secret_key = hs.config.macaroon_secret_key + self._error_template = load_jinja2_templates( + hs.config.sso_template_dir, ["sso_error.html"] + )[0] + + # identifier for the external_ids table + self._auth_provider_id = "oidc" + + def _render_error( + self, request, error: str, error_description: Optional[str] = None + ) -> None: + """Renders the error template and respond with it. + + This is used to show errors to the user. The template of this page can + be found under ``synapse/res/templates/sso_error.html``. + + Args: + request: The incoming request from the browser. + We'll respond with an HTML page describing the error. + error: A technical identifier for this error. Those include + well-known OAuth2/OIDC error types like invalid_request or + access_denied. + error_description: A human-readable description of the error. + """ + html_bytes = self._error_template.render( + error=error, error_description=error_description + ).encode("utf-8") + + request.setResponseCode(400) + request.setHeader(b"Content-Type", b"text/html; charset=utf-8") + request.setHeader(b"Content-Length", b"%i" % len(html_bytes)) + request.write(html_bytes) + finish_request(request) + + def _validate_metadata(self): + """Verifies the provider metadata. + + This checks the validity of the currently loaded provider. Not + everything is checked, only: + + - ``issuer`` + - ``authorization_endpoint`` + - ``token_endpoint`` + - ``response_types_supported`` (checks if "code" is in it) + - ``jwks_uri`` + + Raises: + ValueError: if something in the provider is not valid + """ + # Skip verification to allow non-compliant providers (e.g. issuers not running on a secure origin) + if self._skip_verification is True: + return + + m = self._provider_metadata + m.validate_issuer() + m.validate_authorization_endpoint() + m.validate_token_endpoint() + + if m.get("token_endpoint_auth_methods_supported") is not None: + m.validate_token_endpoint_auth_methods_supported() + if ( + self._client_auth_method + not in m["token_endpoint_auth_methods_supported"] + ): + raise ValueError( + '"{auth_method}" not in "token_endpoint_auth_methods_supported" ({supported!r})'.format( + auth_method=self._client_auth_method, + supported=m["token_endpoint_auth_methods_supported"], + ) + ) + + if m.get("response_types_supported") is not None: + m.validate_response_types_supported() + + if "code" not in m["response_types_supported"]: + raise ValueError( + '"code" not in "response_types_supported" (%r)' + % (m["response_types_supported"],) + ) + + # If the openid scope was not requested, we need a userinfo endpoint to fetch user infos + if self._uses_userinfo: + if m.get("userinfo_endpoint") is None: + raise ValueError( + 'provider has no "userinfo_endpoint", even though it is required because the "openid" scope is not requested' + ) + else: + # If we're not using userinfo, we need a valid jwks to validate the ID token + if m.get("jwks") is None: + if m.get("jwks_uri") is not None: + m.validate_jwks_uri() + else: + raise ValueError('"jwks_uri" must be set') + + @property + def _uses_userinfo(self) -> bool: + """Returns True if the ``userinfo_endpoint`` should be used. + + This is based on the requested scopes: if the scopes include + ``openid``, the provider should give use an ID token containing the + user informations. If not, we should fetch them using the + ``access_token`` with the ``userinfo_endpoint``. + """ + + # Maybe that should be user-configurable and not inferred? + return "openid" not in self._scopes + + async def load_metadata(self) -> OpenIDProviderMetadata: + """Load and validate the provider metadata. + + The values metadatas are discovered if ``oidc_config.discovery`` is + ``True`` and then cached. + + Raises: + ValueError: if something in the provider is not valid + + Returns: + The provider's metadata. + """ + # If we are using the OpenID Discovery documents, it needs to be loaded once + # FIXME: should there be a lock here? + if self._provider_needs_discovery: + url = get_well_known_url(self._provider_metadata["issuer"], external=True) + metadata_response = await self._http_client.get_json(url) + # TODO: maybe update the other way around to let user override some values? + self._provider_metadata.update(metadata_response) + self._provider_needs_discovery = False + + self._validate_metadata() + + return self._provider_metadata + + async def load_jwks(self, force: bool = False) -> JWKS: + """Load the JSON Web Key Set used to sign ID tokens. + + If we're not using the ``userinfo_endpoint``, user infos are extracted + from the ID token, which is a JWT signed by keys given by the provider. + The keys are then cached. + + Args: + force: Force reloading the keys. + + Returns: + The key set + + Looks like this:: + + { + 'keys': [ + { + 'kid': 'abcdef', + 'kty': 'RSA', + 'alg': 'RS256', + 'use': 'sig', + 'e': 'XXXX', + 'n': 'XXXX', + } + ] + } + """ + if self._uses_userinfo: + # We're not using jwt signing, return an empty jwk set + return {"keys": []} + + # First check if the JWKS are loaded in the provider metadata. + # It can happen either if the provider gives its JWKS in the discovery + # document directly or if it was already loaded once. + metadata = await self.load_metadata() + jwk_set = metadata.get("jwks") + if jwk_set is not None and not force: + return jwk_set + + # Loading the JWKS using the `jwks_uri` metadata + uri = metadata.get("jwks_uri") + if not uri: + raise RuntimeError('Missing "jwks_uri" in metadata') + + jwk_set = await self._http_client.get_json(uri) + + # Caching the JWKS in the provider's metadata + self._provider_metadata["jwks"] = jwk_set + return jwk_set + + async def _exchange_code(self, code: str) -> Token: + """Exchange an authorization code for a token. + + This calls the ``token_endpoint`` with the authorization code we + received in the callback to exchange it for a token. The call uses the + ``ClientAuth`` to authenticate with the client with its ID and secret. + + Args: + code: The autorization code we got from the callback. + + Returns: + A dict containing various tokens. + + May look like this:: + + { + 'token_type': 'bearer', + 'access_token': 'abcdef', + 'expires_in': 3599, + 'id_token': 'ghijkl', + 'refresh_token': 'mnopqr', + } + + Raises: + OidcError: when the ``token_endpoint`` returned an error. + """ + metadata = await self.load_metadata() + token_endpoint = metadata.get("token_endpoint") + headers = { + "Content-Type": "application/x-www-form-urlencoded", + "User-Agent": self._http_client.user_agent, + "Accept": "application/json", + } + + args = { + "grant_type": "authorization_code", + "code": code, + "redirect_uri": self._callback_url, + } + body = urlencode(args, True) + + # Fill the body/headers with credentials + uri, headers, body = self._client_auth.prepare( + method="POST", uri=token_endpoint, headers=headers, body=body + ) + headers = {k: [v] for (k, v) in headers.items()} + + # Do the actual request + # We're not using the SimpleHttpClient util methods as we don't want to + # check the HTTP status code and we do the body encoding ourself. + response = await self._http_client.request( + method="POST", uri=uri, data=body.encode("utf-8"), headers=headers, + ) + + # This is used in multiple error messages below + status = "{code} {phrase}".format( + code=response.code, phrase=response.phrase.decode("utf-8") + ) + + resp_body = await readBody(response) + + if response.code >= 500: + # In case of a server error, we should first try to decode the body + # and check for an error field. If not, we respond with a generic + # error message. + try: + resp = json.loads(resp_body.decode("utf-8")) + error = resp["error"] + description = resp.get("error_description", error) + except (ValueError, KeyError): + # Catch ValueError for the JSON decoding and KeyError for the "error" field + error = "server_error" + description = ( + ( + 'Authorization server responded with a "{status}" error ' + "while exchanging the authorization code." + ).format(status=status), + ) + + raise OidcError(error, description) + + # Since it is a not a 5xx code, body should be a valid JSON. It will + # raise if not. + resp = json.loads(resp_body.decode("utf-8")) + + if "error" in resp: + error = resp["error"] + # In case the authorization server responded with an error field, + # it should be a 4xx code. If not, warn about it but don't do + # anything special and report the original error message. + if response.code < 400: + logger.debug( + "Invalid response from the authorization server: " + 'responded with a "{status}" ' + "but body has an error field: {error!r}".format( + status=status, error=resp["error"] + ) + ) + + description = resp.get("error_description", error) + raise OidcError(error, description) + + # Now, this should not be an error. According to RFC6749 sec 5.1, it + # should be a 200 code. We're a bit more flexible than that, and will + # only throw on a 4xx code. + if response.code >= 400: + description = ( + 'Authorization server responded with a "{status}" error ' + 'but did not include an "error" field in its response.'.format( + status=status + ) + ) + logger.warning(description) + # Body was still valid JSON. Might be useful to log it for debugging. + logger.warning("Code exchange response: {resp!r}".format(resp=resp)) + raise OidcError("server_error", description) + + return resp + + async def _fetch_userinfo(self, token: Token) -> UserInfo: + """Fetch user informations from the ``userinfo_endpoint``. + + Args: + token: the token given by the ``token_endpoint``. + Must include an ``access_token`` field. + + Returns: + UserInfo: an object representing the user. + """ + metadata = await self.load_metadata() + + resp = await self._http_client.get_json( + metadata["userinfo_endpoint"], + headers={"Authorization": ["Bearer {}".format(token["access_token"])]}, + ) + + return UserInfo(resp) + + async def _parse_id_token(self, token: Token, nonce: str) -> UserInfo: + """Return an instance of UserInfo from token's ``id_token``. + + Args: + token: the token given by the ``token_endpoint``. + Must include an ``id_token`` field. + nonce: the nonce value originally sent in the initial authorization + request. This value should match the one inside the token. + + Returns: + An object representing the user. + """ + metadata = await self.load_metadata() + claims_params = { + "nonce": nonce, + "client_id": self._client_auth.client_id, + } + if "access_token" in token: + # If we got an `access_token`, there should be an `at_hash` claim + # in the `id_token` that we can check against. + claims_params["access_token"] = token["access_token"] + claims_cls = CodeIDToken + else: + claims_cls = ImplicitIDToken + + alg_values = metadata.get("id_token_signing_alg_values_supported", ["RS256"]) + + jwt = JsonWebToken(alg_values) + + claim_options = {"iss": {"values": [metadata["issuer"]]}} + + # Try to decode the keys in cache first, then retry by forcing the keys + # to be reloaded + jwk_set = await self.load_jwks() + try: + claims = jwt.decode( + token["id_token"], + key=jwk_set, + claims_cls=claims_cls, + claims_options=claim_options, + claims_params=claims_params, + ) + except ValueError: + jwk_set = await self.load_jwks(force=True) # try reloading the jwks + claims = jwt.decode( + token["id_token"], + key=jwk_set, + claims_cls=claims_cls, + claims_options=claim_options, + claims_params=claims_params, + ) + + claims.validate(leeway=120) # allows 2 min of clock skew + return UserInfo(claims) + + async def handle_redirect_request( + self, request: SynapseRequest, client_redirect_url: bytes + ) -> None: + """Handle an incoming request to /login/sso/redirect + + It redirects the browser to the authorization endpoint with a few + parameters: + + - ``client_id``: the client ID set in ``oidc_config.client_id`` + - ``response_type``: ``code`` + - ``redirect_uri``: the callback URL ; ``{base url}/_synapse/oidc/callback`` + - ``scope``: the list of scopes set in ``oidc_config.scopes`` + - ``state``: a random string + - ``nonce``: a random string + + In addition to redirecting the client, we are setting a cookie with + a signed macaroon token containing the state, the nonce and the + client_redirect_url params. Those are then checked when the client + comes back from the provider. + + + Args: + request: the incoming request from the browser. + We'll respond to it with a redirect and a cookie. + client_redirect_url: the URL that we should redirect the client to + when everything is done + """ + + state = generate_token() + nonce = generate_token() + + cookie = self._generate_oidc_session_token( + state=state, nonce=nonce, client_redirect_url=client_redirect_url.decode(), + ) + request.addCookie( + SESSION_COOKIE_NAME, + cookie, + path="/_synapse/oidc", + max_age="3600", + httpOnly=True, + sameSite="lax", + ) + + metadata = await self.load_metadata() + authorization_endpoint = metadata.get("authorization_endpoint") + uri = prepare_grant_uri( + authorization_endpoint, + client_id=self._client_auth.client_id, + response_type="code", + redirect_uri=self._callback_url, + scope=self._scopes, + state=state, + nonce=nonce, + ) + request.redirect(uri) + finish_request(request) + + async def handle_oidc_callback(self, request: SynapseRequest) -> None: + """Handle an incoming request to /_synapse/oidc/callback + + Since we might want to display OIDC-related errors in a user-friendly + way, we don't raise SynapseError from here. Instead, we call + ``self._render_error`` which displays an HTML page for the error. + + Most of the OpenID Connect logic happens here: + + - first, we check if there was any error returned by the provider and + display it + - then we fetch the session cookie, decode and verify it + - the ``state`` query parameter should match with the one stored in the + session cookie + - once we known this session is legit, exchange the code with the + provider using the ``token_endpoint`` (see ``_exchange_code``) + - once we have the token, use it to either extract the UserInfo from + the ``id_token`` (``_parse_id_token``), or use the ``access_token`` + to fetch UserInfo from the ``userinfo_endpoint`` + (``_fetch_userinfo``) + - map those UserInfo to a Matrix user (``_map_userinfo_to_user``) and + finish the login + + Args: + request: the incoming request from the browser. + """ + + # The provider might redirect with an error. + # In that case, just display it as-is. + if b"error" in request.args: + error = request.args[b"error"][0].decode() + description = request.args.get(b"error_description", [b""])[0].decode() + + # Most of the errors returned by the provider could be due by + # either the provider misbehaving or Synapse being misconfigured. + # The only exception of that is "access_denied", where the user + # probably cancelled the login flow. In other cases, log those errors. + if error != "access_denied": + logger.error("Error from the OIDC provider: %s %s", error, description) + + self._render_error(request, error, description) + return + + # Fetch the session cookie + session = request.getCookie(SESSION_COOKIE_NAME) + if session is None: + logger.info("No session cookie found") + self._render_error(request, "missing_session", "No session cookie found") + return + + # Remove the cookie. There is a good chance that if the callback failed + # once, it will fail next time and the code will already be exchanged. + # Removing it early avoids spamming the provider with token requests. + request.addCookie( + SESSION_COOKIE_NAME, + b"", + path="/_synapse/oidc", + expires="Thu, Jan 01 1970 00:00:00 UTC", + httpOnly=True, + sameSite="lax", + ) + + # Check for the state query parameter + if b"state" not in request.args: + logger.info("State parameter is missing") + self._render_error(request, "invalid_request", "State parameter is missing") + return + + state = request.args[b"state"][0].decode() + + # Deserialize the session token and verify it. + try: + nonce, client_redirect_url = self._verify_oidc_session_token(session, state) + except MacaroonDeserializationException as e: + logger.exception("Invalid session") + self._render_error(request, "invalid_session", str(e)) + return + except MacaroonInvalidSignatureException as e: + logger.exception("Could not verify session") + self._render_error(request, "mismatching_session", str(e)) + return + + # Exchange the code with the provider + if b"code" not in request.args: + logger.info("Code parameter is missing") + self._render_error(request, "invalid_request", "Code parameter is missing") + return + + logger.info("Exchanging code") + code = request.args[b"code"][0].decode() + try: + token = await self._exchange_code(code) + except OidcError as e: + logger.exception("Could not exchange code") + self._render_error(request, e.error, e.error_description) + return + + # Now that we have a token, get the userinfo, either by decoding the + # `id_token` or by fetching the `userinfo_endpoint`. + if self._uses_userinfo: + logger.info("Fetching userinfo") + try: + userinfo = await self._fetch_userinfo(token) + except Exception as e: + logger.exception("Could not fetch userinfo") + self._render_error(request, "fetch_error", str(e)) + return + else: + logger.info("Extracting userinfo from id_token") + try: + userinfo = await self._parse_id_token(token, nonce=nonce) + except Exception as e: + logger.exception("Invalid id_token") + self._render_error(request, "invalid_token", str(e)) + return + + # Call the mapper to register/login the user + try: + user_id = await self._map_userinfo_to_user(userinfo, token) + except MappingException as e: + logger.exception("Could not map user") + self._render_error(request, "mapping_error", str(e)) + return + + # and finally complete the login + await self._auth_handler.complete_sso_login( + user_id, request, client_redirect_url + ) + + def _generate_oidc_session_token( + self, + state: str, + nonce: str, + client_redirect_url: str, + duration_in_ms: int = (60 * 60 * 1000), + ) -> str: + """Generates a signed token storing data about an OIDC session. + + When Synapse initiates an authorization flow, it creates a random state + and a random nonce. Those parameters are given to the provider and + should be verified when the client comes back from the provider. + It is also used to store the client_redirect_url, which is used to + complete the SSO login flow. + + Args: + state: The ``state`` parameter passed to the OIDC provider. + nonce: The ``nonce`` parameter passed to the OIDC provider. + client_redirect_url: The URL the client gave when it initiated the + flow. + duration_in_ms: An optional duration for the token in milliseconds. + Defaults to an hour. + + Returns: + A signed macaroon token with the session informations. + """ + macaroon = pymacaroons.Macaroon( + location=self._server_name, identifier="key", key=self._macaroon_secret_key, + ) + macaroon.add_first_party_caveat("gen = 1") + macaroon.add_first_party_caveat("type = session") + macaroon.add_first_party_caveat("state = %s" % (state,)) + macaroon.add_first_party_caveat("nonce = %s" % (nonce,)) + macaroon.add_first_party_caveat( + "client_redirect_url = %s" % (client_redirect_url,) + ) + now = self._clock.time_msec() + expiry = now + duration_in_ms + macaroon.add_first_party_caveat("time < %d" % (expiry,)) + return macaroon.serialize() + + def _verify_oidc_session_token(self, session: str, state: str) -> Tuple[str, str]: + """Verifies and extract an OIDC session token. + + This verifies that a given session token was issued by this homeserver + and extract the nonce and client_redirect_url caveats. + + Args: + session: The session token to verify + state: The state the OIDC provider gave back + + Returns: + The nonce and the client_redirect_url for this session + """ + macaroon = pymacaroons.Macaroon.deserialize(session) + + v = pymacaroons.Verifier() + v.satisfy_exact("gen = 1") + v.satisfy_exact("type = session") + v.satisfy_exact("state = %s" % (state,)) + v.satisfy_general(lambda c: c.startswith("nonce = ")) + v.satisfy_general(lambda c: c.startswith("client_redirect_url = ")) + v.satisfy_general(self._verify_expiry) + + v.verify(macaroon, self._macaroon_secret_key) + + # Extract the `nonce` and `client_redirect_url` from the token + nonce = self._get_value_from_macaroon(macaroon, "nonce") + client_redirect_url = self._get_value_from_macaroon( + macaroon, "client_redirect_url" + ) + + return nonce, client_redirect_url + + def _get_value_from_macaroon(self, macaroon: pymacaroons.Macaroon, key: str) -> str: + """Extracts a caveat value from a macaroon token. + + Args: + macaroon: the token + key: the key of the caveat to extract + + Returns: + The extracted value + + Raises: + Exception: if the caveat was not in the macaroon + """ + prefix = key + " = " + for caveat in macaroon.caveats: + if caveat.caveat_id.startswith(prefix): + return caveat.caveat_id[len(prefix) :] + raise Exception("No %s caveat in macaroon" % (key,)) + + def _verify_expiry(self, caveat: str) -> bool: + prefix = "time < " + if not caveat.startswith(prefix): + return False + expiry = int(caveat[len(prefix) :]) + now = self._clock.time_msec() + return now < expiry + + async def _map_userinfo_to_user(self, userinfo: UserInfo, token: Token) -> str: + """Maps a UserInfo object to a mxid. + + UserInfo should have a claim that uniquely identifies users. This claim + is usually `sub`, but can be configured with `oidc_config.subject_claim`. + It is then used as an `external_id`. + + If we don't find the user that way, we should register the user, + mapping the localpart and the display name from the UserInfo. + + If a user already exists with the mxid we've mapped, raise an exception. + + Args: + userinfo: an object representing the user + token: a dict with the tokens obtained from the provider + + Raises: + MappingException: if there was an error while mapping some properties + + Returns: + The mxid of the user + """ + try: + remote_user_id = self._user_mapping_provider.get_remote_user_id(userinfo) + except Exception as e: + raise MappingException( + "Failed to extract subject from OIDC response: %s" % (e,) + ) + + logger.info( + "Looking for existing mapping for user %s:%s", + self._auth_provider_id, + remote_user_id, + ) + + registered_user_id = await self._datastore.get_user_by_external_id( + self._auth_provider_id, remote_user_id, + ) + + if registered_user_id is not None: + logger.info("Found existing mapping %s", registered_user_id) + return registered_user_id + + try: + attributes = await self._user_mapping_provider.map_user_attributes( + userinfo, token + ) + except Exception as e: + raise MappingException( + "Could not extract user attributes from OIDC response: " + str(e) + ) + + logger.debug( + "Retrieved user attributes from user mapping provider: %r", attributes + ) + + if not attributes["localpart"]: + raise MappingException("localpart is empty") + + localpart = map_username_to_mxid_localpart(attributes["localpart"]) + + user_id = UserID(localpart, self._hostname) + if await self._datastore.get_users_by_id_case_insensitive(user_id.to_string()): + # This mxid is taken + raise MappingException( + "mxid '{}' is already taken".format(user_id.to_string()) + ) + + # It's the first time this user is logging in and the mapped mxid was + # not taken, register the user + registered_user_id = await self._registration_handler.register_user( + localpart=localpart, default_display_name=attributes["display_name"], + ) + + await self._datastore.record_user_external_id( + self._auth_provider_id, remote_user_id, registered_user_id, + ) + return registered_user_id + + +UserAttribute = TypedDict( + "UserAttribute", {"localpart": str, "display_name": Optional[str]} +) +C = TypeVar("C") + + +class OidcMappingProvider(Generic[C]): + """A mapping provider maps a UserInfo object to user attributes. + + It should provide the API described by this class. + """ + + def __init__(self, config: C): + """ + Args: + config: A custom config object from this module, parsed by ``parse_config()`` + """ + + @staticmethod + def parse_config(config: dict) -> C: + """Parse the dict provided by the homeserver's config + + Args: + config: A dictionary containing configuration options for this provider + + Returns: + A custom config object for this module + """ + raise NotImplementedError() + + def get_remote_user_id(self, userinfo: UserInfo) -> str: + """Get a unique user ID for this user. + + Usually, in an OIDC-compliant scenario, it should be the ``sub`` claim from the UserInfo object. + + Args: + userinfo: An object representing the user given by the OIDC provider + + Returns: + A unique user ID + """ + raise NotImplementedError() + + async def map_user_attributes( + self, userinfo: UserInfo, token: Token + ) -> UserAttribute: + """Map a ``UserInfo`` objects into user attributes. + + Args: + userinfo: An object representing the user given by the OIDC provider + token: A dict with the tokens returned by the provider + + Returns: + A dict containing the ``localpart`` and (optionally) the ``display_name`` + """ + raise NotImplementedError() + + +# Used to clear out "None" values in templates +def jinja_finalize(thing): + return thing if thing is not None else "" + + +env = Environment(finalize=jinja_finalize) + + +@attr.s +class JinjaOidcMappingConfig: + subject_claim = attr.ib() # type: str + localpart_template = attr.ib() # type: Template + display_name_template = attr.ib() # type: Optional[Template] + + +class JinjaOidcMappingProvider(OidcMappingProvider[JinjaOidcMappingConfig]): + """An implementation of a mapping provider based on Jinja templates. + + This is the default mapping provider. + """ + + def __init__(self, config: JinjaOidcMappingConfig): + self._config = config + + @staticmethod + def parse_config(config: dict) -> JinjaOidcMappingConfig: + subject_claim = config.get("subject_claim", "sub") + + if "localpart_template" not in config: + raise ConfigError( + "missing key: oidc_config.user_mapping_provider.config.localpart_template" + ) + + try: + localpart_template = env.from_string(config["localpart_template"]) + except Exception as e: + raise ConfigError( + "invalid jinja template for oidc_config.user_mapping_provider.config.localpart_template: %r" + % (e,) + ) + + display_name_template = None # type: Optional[Template] + if "display_name_template" in config: + try: + display_name_template = env.from_string(config["display_name_template"]) + except Exception as e: + raise ConfigError( + "invalid jinja template for oidc_config.user_mapping_provider.config.display_name_template: %r" + % (e,) + ) + + return JinjaOidcMappingConfig( + subject_claim=subject_claim, + localpart_template=localpart_template, + display_name_template=display_name_template, + ) + + def get_remote_user_id(self, userinfo: UserInfo) -> str: + return userinfo[self._config.subject_claim] + + async def map_user_attributes( + self, userinfo: UserInfo, token: Token + ) -> UserAttribute: + localpart = self._config.localpart_template.render(user=userinfo).strip() + + display_name = None # type: Optional[str] + if self._config.display_name_template is not None: + display_name = self._config.display_name_template.render( + user=userinfo + ).strip() + + if display_name == "": + display_name = None + + return UserAttribute(localpart=localpart, display_name=display_name) diff --git a/synapse/http/client.py b/synapse/http/client.py index 3797545824..58eb47c69c 100644 --- a/synapse/http/client.py +++ b/synapse/http/client.py @@ -359,6 +359,7 @@ class SimpleHttpClient(object): actual_headers = { b"Content-Type": [b"application/x-www-form-urlencoded"], b"User-Agent": [self.user_agent], + b"Accept": [b"application/json"], } if headers: actual_headers.update(headers) @@ -399,6 +400,7 @@ class SimpleHttpClient(object): actual_headers = { b"Content-Type": [b"application/json"], b"User-Agent": [self.user_agent], + b"Accept": [b"application/json"], } if headers: actual_headers.update(headers) @@ -434,6 +436,10 @@ class SimpleHttpClient(object): ValueError: if the response was not JSON """ + actual_headers = {b"Accept": [b"application/json"]} + if headers: + actual_headers.update(headers) + body = yield self.get_raw(uri, args, headers=headers) return json.loads(body) @@ -467,6 +473,7 @@ class SimpleHttpClient(object): actual_headers = { b"Content-Type": [b"application/json"], b"User-Agent": [self.user_agent], + b"Accept": [b"application/json"], } if headers: actual_headers.update(headers) diff --git a/synapse/python_dependencies.py b/synapse/python_dependencies.py index 39c99a2802..8b4312e5a3 100644 --- a/synapse/python_dependencies.py +++ b/synapse/python_dependencies.py @@ -92,6 +92,7 @@ CONDITIONAL_REQUIREMENTS = { 'eliot<1.8.0;python_version<"3.5.3"', ], "saml2": ["pysaml2>=4.5.0"], + "oidc": ["authlib>=0.14.0"], "systemd": ["systemd-python>=231"], "url_preview": ["lxml>=3.5.0"], "test": ["mock>=2.0", "parameterized"], diff --git a/synapse/res/templates/sso_error.html b/synapse/res/templates/sso_error.html new file mode 100644 index 0000000000..43a211386b --- /dev/null +++ b/synapse/res/templates/sso_error.html @@ -0,0 +1,18 @@ + + + + + SSO error + + +

Oops! Something went wrong during authentication.

+

+ Try logging in again from your Matrix client and if the problem persists + please contact the server's administrator. +

+

Error: {{ error }}

+ {% if error_description %} + 
{{ error_description }}
+ {% endif %} + + diff --git a/synapse/rest/client/v1/login.py b/synapse/rest/client/v1/login.py index 4de2f97d06..de7eca21f8 100644 --- a/synapse/rest/client/v1/login.py +++ b/synapse/rest/client/v1/login.py @@ -83,6 +83,7 @@ class LoginRestServlet(RestServlet): self.jwt_algorithm = hs.config.jwt_algorithm self.saml2_enabled = hs.config.saml2_enabled self.cas_enabled = hs.config.cas_enabled + self.oidc_enabled = hs.config.oidc_enabled self.auth_handler = self.hs.get_auth_handler() self.registration_handler = hs.get_registration_handler() self.handlers = hs.get_handlers() @@ -96,9 +97,7 @@ class LoginRestServlet(RestServlet): flows = [] if self.jwt_enabled: flows.append({"type": LoginRestServlet.JWT_TYPE}) - if self.saml2_enabled: - flows.append({"type": LoginRestServlet.SSO_TYPE}) - flows.append({"type": LoginRestServlet.TOKEN_TYPE}) + if self.cas_enabled: flows.append({"type": LoginRestServlet.SSO_TYPE}) @@ -114,6 +113,11 @@ class LoginRestServlet(RestServlet): # fall back to the fallback API if they don't understand one of the # login flow types returned. flows.append({"type": LoginRestServlet.TOKEN_TYPE}) + elif self.saml2_enabled: + flows.append({"type": LoginRestServlet.SSO_TYPE}) + flows.append({"type": LoginRestServlet.TOKEN_TYPE}) + elif self.oidc_enabled: + flows.append({"type": LoginRestServlet.SSO_TYPE}) flows.extend( ({"type": t} for t in self.auth_handler.get_supported_login_types()) @@ -465,6 +469,22 @@ class SAMLRedirectServlet(BaseSSORedirectServlet): return self._saml_handler.handle_redirect_request(client_redirect_url) +class OIDCRedirectServlet(RestServlet): + """Implementation for /login/sso/redirect for the OIDC login flow.""" + + PATTERNS = client_patterns("/login/sso/redirect", v1=True) + + def __init__(self, hs): + self._oidc_handler = hs.get_oidc_handler() + + async def on_GET(self, request): + args = request.args + if b"redirectUrl" not in args: + return 400, "Redirect URL not specified for SSO auth" + client_redirect_url = args[b"redirectUrl"][0] + await self._oidc_handler.handle_redirect_request(request, client_redirect_url) + + def register_servlets(hs, http_server): LoginRestServlet(hs).register(http_server) if hs.config.cas_enabled: @@ -472,3 +492,5 @@ def register_servlets(hs, http_server): CasTicketServlet(hs).register(http_server) elif hs.config.saml2_enabled: SAMLRedirectServlet(hs).register(http_server) + elif hs.config.oidc_enabled: + OIDCRedirectServlet(hs).register(http_server) diff --git a/synapse/rest/oidc/__init__.py b/synapse/rest/oidc/__init__.py new file mode 100644 index 0000000000..d958dd65bb --- /dev/null +++ b/synapse/rest/oidc/__init__.py @@ -0,0 +1,27 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 Quentin Gliech +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import logging + +from twisted.web.resource import Resource + +from synapse.rest.oidc.callback_resource import OIDCCallbackResource + +logger = logging.getLogger(__name__) + + +class OIDCResource(Resource): + def __init__(self, hs): + Resource.__init__(self) + self.putChild(b"callback", OIDCCallbackResource(hs)) diff --git a/synapse/rest/oidc/callback_resource.py b/synapse/rest/oidc/callback_resource.py new file mode 100644 index 0000000000..c03194f001 --- /dev/null +++ b/synapse/rest/oidc/callback_resource.py @@ -0,0 +1,31 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 Quentin Gliech +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import logging + +from synapse.http.server import DirectServeResource, wrap_html_request_handler + +logger = logging.getLogger(__name__) + + +class OIDCCallbackResource(DirectServeResource): + isLeaf = 1 + + def __init__(self, hs): + super().__init__() + self._oidc_handler = hs.get_oidc_handler() + + @wrap_html_request_handler + async def _async_render_GET(self, request): + return await self._oidc_handler.handle_oidc_callback(request) diff --git a/synapse/server.py b/synapse/server.py index bf97a16c09..b4aea81e24 100644 --- a/synapse/server.py +++ b/synapse/server.py @@ -204,6 +204,7 @@ class HomeServer(object): "account_validity_handler", "cas_handler", "saml_handler", + "oidc_handler", "event_client_serializer", "password_policy_handler", "storage", @@ -562,6 +563,11 @@ class HomeServer(object): return SamlHandler(self) + def build_oidc_handler(self): + from synapse.handlers.oidc_handler import OidcHandler + + return OidcHandler(self) + def build_event_client_serializer(self): return EventClientSerializer(self) diff --git a/synapse/server.pyi b/synapse/server.pyi index 18043a2593..31a9cc0389 100644 --- a/synapse/server.pyi +++ b/synapse/server.pyi @@ -13,6 +13,7 @@ import synapse.handlers.device import synapse.handlers.e2e_keys import synapse.handlers.message import synapse.handlers.presence +import synapse.handlers.register import synapse.handlers.room import synapse.handlers.room_member import synapse.handlers.set_password @@ -128,3 +129,7 @@ class HomeServer(object): pass def get_storage(self) -> synapse.storage.Storage: pass + def get_registration_handler(self) -> synapse.handlers.register.RegistrationHandler: + pass + def get_macaroon_generator(self) -> synapse.handlers.auth.MacaroonGenerator: + pass diff --git a/tests/handlers/test_oidc.py b/tests/handlers/test_oidc.py new file mode 100644 index 0000000000..61963aa90d --- /dev/null +++ b/tests/handlers/test_oidc.py @@ -0,0 +1,565 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 Quentin Gliech +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import json +from urllib.parse import parse_qs, urlparse + +from mock import Mock, patch + +import attr +import pymacaroons + +from twisted.internet import defer +from twisted.python.failure import Failure +from twisted.web._newclient import ResponseDone + +from synapse.handlers.oidc_handler import ( + MappingException, + OidcError, + OidcHandler, + OidcMappingProvider, +) +from synapse.types import UserID + +from tests.unittest import HomeserverTestCase, override_config + + +@attr.s +class FakeResponse: + code = attr.ib() + body = attr.ib() + phrase = attr.ib() + + def deliverBody(self, protocol): + protocol.dataReceived(self.body) + protocol.connectionLost(Failure(ResponseDone())) + + +# These are a few constants that are used as config parameters in the tests. +ISSUER = "https://issuer/" +CLIENT_ID = "test-client-id" +CLIENT_SECRET = "test-client-secret" +BASE_URL = "https://synapse/" +CALLBACK_URL = BASE_URL + "_synapse/oidc/callback" +SCOPES = ["openid"] + +AUTHORIZATION_ENDPOINT = ISSUER + "authorize" +TOKEN_ENDPOINT = ISSUER + "token" +USERINFO_ENDPOINT = ISSUER + "userinfo" +WELL_KNOWN = ISSUER + ".well-known/openid-configuration" +JWKS_URI = ISSUER + ".well-known/jwks.json" + +# config for common cases +COMMON_CONFIG = { + "discover": False, + "authorization_endpoint": AUTHORIZATION_ENDPOINT, + "token_endpoint": TOKEN_ENDPOINT, + "jwks_uri": JWKS_URI, +} + + +# The cookie name and path don't really matter, just that it has to be coherent +# between the callback & redirect handlers. +COOKIE_NAME = b"oidc_session" +COOKIE_PATH = "/_synapse/oidc" + +MockedMappingProvider = Mock(OidcMappingProvider) + + +def simple_async_mock(return_value=None, raises=None): + # AsyncMock is not available in python3.5, this mimics part of its behaviour + async def cb(*args, **kwargs): + if raises: + raise raises + return return_value + + return Mock(side_effect=cb) + + +async def get_json(url): + # Mock get_json calls to handle jwks & oidc discovery endpoints + if url == WELL_KNOWN: + # Minimal discovery document, as defined in OpenID.Discovery + # https://openid.net/specs/openid-connect-discovery-1_0.html#ProviderMetadata + return { + "issuer": ISSUER, + "authorization_endpoint": AUTHORIZATION_ENDPOINT, + "token_endpoint": TOKEN_ENDPOINT, + "jwks_uri": JWKS_URI, + "userinfo_endpoint": USERINFO_ENDPOINT, + "response_types_supported": ["code"], + "subject_types_supported": ["public"], + "id_token_signing_alg_values_supported": ["RS256"], + } + elif url == JWKS_URI: + return {"keys": []} + + +class OidcHandlerTestCase(HomeserverTestCase): + def make_homeserver(self, reactor, clock): + + self.http_client = Mock(spec=["get_json"]) + self.http_client.get_json.side_effect = get_json + self.http_client.user_agent = "Synapse Test" + + config = self.default_config() + config["public_baseurl"] = BASE_URL + oidc_config = config.get("oidc_config", {}) + oidc_config["enabled"] = True + oidc_config["client_id"] = CLIENT_ID + oidc_config["client_secret"] = CLIENT_SECRET + oidc_config["issuer"] = ISSUER + oidc_config["scopes"] = SCOPES + oidc_config["user_mapping_provider"] = { + "module": __name__ + ".MockedMappingProvider" + } + config["oidc_config"] = oidc_config + + hs = self.setup_test_homeserver( + http_client=self.http_client, + proxied_http_client=self.http_client, + config=config, + ) + + self.handler = OidcHandler(hs) + + return hs + + def metadata_edit(self, values): + return patch.dict(self.handler._provider_metadata, values) + + def assertRenderedError(self, error, error_description=None): + args = self.handler._render_error.call_args[0] + self.assertEqual(args[1], error) + if error_description is not None: + self.assertEqual(args[2], error_description) + # Reset the render_error mock + self.handler._render_error.reset_mock() + + def test_config(self): + """Basic config correctly sets up the callback URL and client auth correctly.""" + self.assertEqual(self.handler._callback_url, CALLBACK_URL) + self.assertEqual(self.handler._client_auth.client_id, CLIENT_ID) + self.assertEqual(self.handler._client_auth.client_secret, CLIENT_SECRET) + + @override_config({"oidc_config": {"discover": True}}) + @defer.inlineCallbacks + def test_discovery(self): + """The handler should discover the endpoints from OIDC discovery document.""" + # This would throw if some metadata were invalid + metadata = yield defer.ensureDeferred(self.handler.load_metadata()) + self.http_client.get_json.assert_called_once_with(WELL_KNOWN) + + self.assertEqual(metadata.issuer, ISSUER) + self.assertEqual(metadata.authorization_endpoint, AUTHORIZATION_ENDPOINT) + self.assertEqual(metadata.token_endpoint, TOKEN_ENDPOINT) + self.assertEqual(metadata.jwks_uri, JWKS_URI) + # FIXME: it seems like authlib does not have that defined in its metadata models + # self.assertEqual(metadata.userinfo_endpoint, USERINFO_ENDPOINT) + + # subsequent calls should be cached + self.http_client.reset_mock() + yield defer.ensureDeferred(self.handler.load_metadata()) + self.http_client.get_json.assert_not_called() + + @override_config({"oidc_config": COMMON_CONFIG}) + @defer.inlineCallbacks + def test_no_discovery(self): + """When discovery is disabled, it should not try to load from discovery document.""" + yield defer.ensureDeferred(self.handler.load_metadata()) + self.http_client.get_json.assert_not_called() + + @override_config({"oidc_config": COMMON_CONFIG}) + @defer.inlineCallbacks + def test_load_jwks(self): + """JWKS loading is done once (then cached) if used.""" + jwks = yield defer.ensureDeferred(self.handler.load_jwks()) + self.http_client.get_json.assert_called_once_with(JWKS_URI) + self.assertEqual(jwks, {"keys": []}) + + # subsequent calls should be cached… + self.http_client.reset_mock() + yield defer.ensureDeferred(self.handler.load_jwks()) + self.http_client.get_json.assert_not_called() + + # …unless forced + self.http_client.reset_mock() + yield defer.ensureDeferred(self.handler.load_jwks(force=True)) + self.http_client.get_json.assert_called_once_with(JWKS_URI) + + # Throw if the JWKS uri is missing + with self.metadata_edit({"jwks_uri": None}): + with self.assertRaises(RuntimeError): + yield defer.ensureDeferred(self.handler.load_jwks(force=True)) + + # Return empty key set if JWKS are not used + self.handler._scopes = [] # not asking the openid scope + self.http_client.get_json.reset_mock() + jwks = yield defer.ensureDeferred(self.handler.load_jwks(force=True)) + self.http_client.get_json.assert_not_called() + self.assertEqual(jwks, {"keys": []}) + + @override_config({"oidc_config": COMMON_CONFIG}) + def test_validate_config(self): + """Provider metadatas are extensively validated.""" + h = self.handler + + # Default test config does not throw + h._validate_metadata() + + with self.metadata_edit({"issuer": None}): + self.assertRaisesRegex(ValueError, "issuer", h._validate_metadata) + + with self.metadata_edit({"issuer": "http://insecure/"}): + self.assertRaisesRegex(ValueError, "issuer", h._validate_metadata) + + with self.metadata_edit({"issuer": "https://invalid/?because=query"}): + self.assertRaisesRegex(ValueError, "issuer", h._validate_metadata) + + with self.metadata_edit({"authorization_endpoint": None}): + self.assertRaisesRegex( + ValueError, "authorization_endpoint", h._validate_metadata + ) + + with self.metadata_edit({"authorization_endpoint": "http://insecure/auth"}): + self.assertRaisesRegex( + ValueError, "authorization_endpoint", h._validate_metadata + ) + + with self.metadata_edit({"token_endpoint": None}): + self.assertRaisesRegex(ValueError, "token_endpoint", h._validate_metadata) + + with self.metadata_edit({"token_endpoint": "http://insecure/token"}): + self.assertRaisesRegex(ValueError, "token_endpoint", h._validate_metadata) + + with self.metadata_edit({"jwks_uri": None}): + self.assertRaisesRegex(ValueError, "jwks_uri", h._validate_metadata) + + with self.metadata_edit({"jwks_uri": "http://insecure/jwks.json"}): + self.assertRaisesRegex(ValueError, "jwks_uri", h._validate_metadata) + + with self.metadata_edit({"response_types_supported": ["id_token"]}): + self.assertRaisesRegex( + ValueError, "response_types_supported", h._validate_metadata + ) + + with self.metadata_edit( + {"token_endpoint_auth_methods_supported": ["client_secret_basic"]} + ): + # should not throw, as client_secret_basic is the default auth method + h._validate_metadata() + + with self.metadata_edit( + {"token_endpoint_auth_methods_supported": ["client_secret_post"]} + ): + self.assertRaisesRegex( + ValueError, + "token_endpoint_auth_methods_supported", + h._validate_metadata, + ) + + # Tests for configs that the userinfo endpoint + self.assertFalse(h._uses_userinfo) + h._scopes = [] # do not request the openid scope + self.assertTrue(h._uses_userinfo) + self.assertRaisesRegex(ValueError, "userinfo_endpoint", h._validate_metadata) + + with self.metadata_edit( + {"userinfo_endpoint": USERINFO_ENDPOINT, "jwks_uri": None} + ): + # Shouldn't raise with a valid userinfo, even without + h._validate_metadata() + + @override_config({"oidc_config": {"skip_verification": True}}) + def test_skip_verification(self): + """Provider metadata validation can be disabled by config.""" + with self.metadata_edit({"issuer": "http://insecure"}): + # This should not throw + self.handler._validate_metadata() + + @defer.inlineCallbacks + def test_redirect_request(self): + """The redirect request has the right arguments & generates a valid session cookie.""" + req = Mock(spec=["addCookie", "redirect", "finish"]) + yield defer.ensureDeferred( + self.handler.handle_redirect_request(req, b"http://client/redirect") + ) + url = req.redirect.call_args[0][0] + url = urlparse(url) + auth_endpoint = urlparse(AUTHORIZATION_ENDPOINT) + + self.assertEqual(url.scheme, auth_endpoint.scheme) + self.assertEqual(url.netloc, auth_endpoint.netloc) + self.assertEqual(url.path, auth_endpoint.path) + + params = parse_qs(url.query) + self.assertEqual(params["redirect_uri"], [CALLBACK_URL]) + self.assertEqual(params["response_type"], ["code"]) + self.assertEqual(params["scope"], [" ".join(SCOPES)]) + self.assertEqual(params["client_id"], [CLIENT_ID]) + self.assertEqual(len(params["state"]), 1) + self.assertEqual(len(params["nonce"]), 1) + + # Check what is in the cookie + # note: python3.5 mock does not have the .called_once() method + calls = req.addCookie.call_args_list + self.assertEqual(len(calls), 1) # called once + # For some reason, call.args does not work with python3.5 + args = calls[0][0] + kwargs = calls[0][1] + self.assertEqual(args[0], COOKIE_NAME) + self.assertEqual(kwargs["path"], COOKIE_PATH) + cookie = args[1] + + macaroon = pymacaroons.Macaroon.deserialize(cookie) + state = self.handler._get_value_from_macaroon(macaroon, "state") + nonce = self.handler._get_value_from_macaroon(macaroon, "nonce") + redirect = self.handler._get_value_from_macaroon( + macaroon, "client_redirect_url" + ) + + self.assertEqual(params["state"], [state]) + self.assertEqual(params["nonce"], [nonce]) + self.assertEqual(redirect, "http://client/redirect") + + @defer.inlineCallbacks + def test_callback_error(self): + """Errors from the provider returned in the callback are displayed.""" + self.handler._render_error = Mock() + request = Mock(args={}) + request.args[b"error"] = [b"invalid_client"] + yield defer.ensureDeferred(self.handler.handle_oidc_callback(request)) + self.assertRenderedError("invalid_client", "") + + request.args[b"error_description"] = [b"some description"] + yield defer.ensureDeferred(self.handler.handle_oidc_callback(request)) + self.assertRenderedError("invalid_client", "some description") + + @defer.inlineCallbacks + def test_callback(self): + """Code callback works and display errors if something went wrong. + + A lot of scenarios are tested here: + - when the callback works, with userinfo from ID token + - when the user mapping fails + - when ID token verification fails + - when the callback works, with userinfo fetched from the userinfo endpoint + - when the userinfo fetching fails + - when the code exchange fails + """ + token = { + "type": "bearer", + "id_token": "id_token", + "access_token": "access_token", + } + userinfo = { + "sub": "foo", + "preferred_username": "bar", + } + user_id = UserID("foo", "domain.org") + self.handler._render_error = Mock(return_value=None) + self.handler._exchange_code = simple_async_mock(return_value=token) + self.handler._parse_id_token = simple_async_mock(return_value=userinfo) + self.handler._fetch_userinfo = simple_async_mock(return_value=userinfo) + self.handler._map_userinfo_to_user = simple_async_mock(return_value=user_id) + self.handler._auth_handler.complete_sso_login = simple_async_mock() + request = Mock(spec=["args", "getCookie", "addCookie"]) + + code = "code" + state = "state" + nonce = "nonce" + client_redirect_url = "http://client/redirect" + session = self.handler._generate_oidc_session_token( + state=state, nonce=nonce, client_redirect_url=client_redirect_url, + ) + request.getCookie.return_value = session + + request.args = {} + request.args[b"code"] = [code.encode("utf-8")] + request.args[b"state"] = [state.encode("utf-8")] + + yield defer.ensureDeferred(self.handler.handle_oidc_callback(request)) + + self.handler._auth_handler.complete_sso_login.assert_called_once_with( + user_id, request, client_redirect_url, + ) + self.handler._exchange_code.assert_called_once_with(code) + self.handler._parse_id_token.assert_called_once_with(token, nonce=nonce) + self.handler._map_userinfo_to_user.assert_called_once_with(userinfo, token) + self.handler._fetch_userinfo.assert_not_called() + self.handler._render_error.assert_not_called() + + # Handle mapping errors + self.handler._map_userinfo_to_user = simple_async_mock( + raises=MappingException() + ) + yield defer.ensureDeferred(self.handler.handle_oidc_callback(request)) + self.assertRenderedError("mapping_error") + self.handler._map_userinfo_to_user = simple_async_mock(return_value=user_id) + + # Handle ID token errors + self.handler._parse_id_token = simple_async_mock(raises=Exception()) + yield defer.ensureDeferred(self.handler.handle_oidc_callback(request)) + self.assertRenderedError("invalid_token") + + self.handler._auth_handler.complete_sso_login.reset_mock() + self.handler._exchange_code.reset_mock() + self.handler._parse_id_token.reset_mock() + self.handler._map_userinfo_to_user.reset_mock() + self.handler._fetch_userinfo.reset_mock() + + # With userinfo fetching + self.handler._scopes = [] # do not ask the "openid" scope + yield defer.ensureDeferred(self.handler.handle_oidc_callback(request)) + + self.handler._auth_handler.complete_sso_login.assert_called_once_with( + user_id, request, client_redirect_url, + ) + self.handler._exchange_code.assert_called_once_with(code) + self.handler._parse_id_token.assert_not_called() + self.handler._map_userinfo_to_user.assert_called_once_with(userinfo, token) + self.handler._fetch_userinfo.assert_called_once_with(token) + self.handler._render_error.assert_not_called() + + # Handle userinfo fetching error + self.handler._fetch_userinfo = simple_async_mock(raises=Exception()) + yield defer.ensureDeferred(self.handler.handle_oidc_callback(request)) + self.assertRenderedError("fetch_error") + + # Handle code exchange failure + self.handler._exchange_code = simple_async_mock( + raises=OidcError("invalid_request") + ) + yield defer.ensureDeferred(self.handler.handle_oidc_callback(request)) + self.assertRenderedError("invalid_request") + + @defer.inlineCallbacks + def test_callback_session(self): + """The callback verifies the session presence and validity""" + self.handler._render_error = Mock(return_value=None) + request = Mock(spec=["args", "getCookie", "addCookie"]) + + # Missing cookie + request.args = {} + request.getCookie.return_value = None + yield defer.ensureDeferred(self.handler.handle_oidc_callback(request)) + self.assertRenderedError("missing_session", "No session cookie found") + + # Missing session parameter + request.args = {} + request.getCookie.return_value = "session" + yield defer.ensureDeferred(self.handler.handle_oidc_callback(request)) + self.assertRenderedError("invalid_request", "State parameter is missing") + + # Invalid cookie + request.args = {} + request.args[b"state"] = [b"state"] + request.getCookie.return_value = "session" + yield defer.ensureDeferred(self.handler.handle_oidc_callback(request)) + self.assertRenderedError("invalid_session") + + # Mismatching session + session = self.handler._generate_oidc_session_token( + state="state", nonce="nonce", client_redirect_url="http://client/redirect", + ) + request.args = {} + request.args[b"state"] = [b"mismatching state"] + request.getCookie.return_value = session + yield defer.ensureDeferred(self.handler.handle_oidc_callback(request)) + self.assertRenderedError("mismatching_session") + + # Valid session + request.args = {} + request.args[b"state"] = [b"state"] + request.getCookie.return_value = session + yield defer.ensureDeferred(self.handler.handle_oidc_callback(request)) + self.assertRenderedError("invalid_request") + + @override_config({"oidc_config": {"client_auth_method": "client_secret_post"}}) + @defer.inlineCallbacks + def test_exchange_code(self): + """Code exchange behaves correctly and handles various error scenarios.""" + token = {"type": "bearer"} + token_json = json.dumps(token).encode("utf-8") + self.http_client.request = simple_async_mock( + return_value=FakeResponse(code=200, phrase=b"OK", body=token_json) + ) + code = "code" + ret = yield defer.ensureDeferred(self.handler._exchange_code(code)) + kwargs = self.http_client.request.call_args[1] + + self.assertEqual(ret, token) + self.assertEqual(kwargs["method"], "POST") + self.assertEqual(kwargs["uri"], TOKEN_ENDPOINT) + + args = parse_qs(kwargs["data"].decode("utf-8")) + self.assertEqual(args["grant_type"], ["authorization_code"]) + self.assertEqual(args["code"], [code]) + self.assertEqual(args["client_id"], [CLIENT_ID]) + self.assertEqual(args["client_secret"], [CLIENT_SECRET]) + self.assertEqual(args["redirect_uri"], [CALLBACK_URL]) + + # Test error handling + self.http_client.request = simple_async_mock( + return_value=FakeResponse( + code=400, + phrase=b"Bad Request", + body=b'{"error": "foo", "error_description": "bar"}', + ) + ) + with self.assertRaises(OidcError) as exc: + yield defer.ensureDeferred(self.handler._exchange_code(code)) + self.assertEqual(exc.exception.error, "foo") + self.assertEqual(exc.exception.error_description, "bar") + + # Internal server error with no JSON body + self.http_client.request = simple_async_mock( + return_value=FakeResponse( + code=500, phrase=b"Internal Server Error", body=b"Not JSON", + ) + ) + with self.assertRaises(OidcError) as exc: + yield defer.ensureDeferred(self.handler._exchange_code(code)) + self.assertEqual(exc.exception.error, "server_error") + + # Internal server error with JSON body + self.http_client.request = simple_async_mock( + return_value=FakeResponse( + code=500, + phrase=b"Internal Server Error", + body=b'{"error": "internal_server_error"}', + ) + ) + with self.assertRaises(OidcError) as exc: + yield defer.ensureDeferred(self.handler._exchange_code(code)) + self.assertEqual(exc.exception.error, "internal_server_error") + + # 4xx error without "error" field + self.http_client.request = simple_async_mock( + return_value=FakeResponse(code=400, phrase=b"Bad request", body=b"{}",) + ) + with self.assertRaises(OidcError) as exc: + yield defer.ensureDeferred(self.handler._exchange_code(code)) + self.assertEqual(exc.exception.error, "server_error") + + # 2xx error with "error" field + self.http_client.request = simple_async_mock( + return_value=FakeResponse( + code=200, phrase=b"OK", body=b'{"error": "some_error"}', + ) + ) + with self.assertRaises(OidcError) as exc: + yield defer.ensureDeferred(self.handler._exchange_code(code)) + self.assertEqual(exc.exception.error, "some_error") diff --git a/tox.ini b/tox.ini index c699f3e46a..ad1902d47d 100644 --- a/tox.ini +++ b/tox.ini @@ -185,6 +185,7 @@ commands = mypy \ synapse/handlers/auth.py \ synapse/handlers/cas_handler.py \ synapse/handlers/directory.py \ + synapse/handlers/oidc_handler.py \ synapse/handlers/presence.py \ synapse/handlers/saml_handler.py \ synapse/handlers/sync.py \ -- cgit 1.4.1 From 67feea8044562764b04c4968ebf159b44eb59218 Mon Sep 17 00:00:00 2001 From: Andrew Morgan <1342360+anoadragon453@users.noreply.github.com> Date: Fri, 8 May 2020 19:25:48 +0100 Subject: Extend spam checker to allow for multiple modules (#7435) --- changelog.d/7435.feature | 1 + docs/sample_config.yaml | 15 +++++-- docs/spam_checker.md | 19 +++++---- synapse/config/spam_checker.py | 38 +++++++++++++---- synapse/events/spamcheck.py | 78 +++++++++++++++++------------------ tests/handlers/test_user_directory.py | 4 +- 6 files changed, 95 insertions(+), 60 deletions(-) create mode 100644 changelog.d/7435.feature (limited to 'tests') diff --git a/changelog.d/7435.feature b/changelog.d/7435.feature new file mode 100644 index 0000000000..399291b13b --- /dev/null +++ b/changelog.d/7435.feature @@ -0,0 +1 @@ +Allow for using more than one spam checker module at once. \ No newline at end of file diff --git a/docs/sample_config.yaml b/docs/sample_config.yaml index 1e397f7734..5abeaf519b 100644 --- a/docs/sample_config.yaml +++ b/docs/sample_config.yaml @@ -1867,10 +1867,17 @@ password_providers: # include_content: true -#spam_checker: -# module: "my_custom_project.SuperSpamChecker" -# config: -# example_option: 'things' +# Spam checkers are third-party modules that can block specific actions +# of local users, such as creating rooms and registering undesirable +# usernames, as well as remote users by redacting incoming events. +# +spam_checker: + #- module: "my_custom_project.SuperSpamChecker" + # config: + # example_option: 'things' + #- module: "some_other_project.BadEventStopper" + # config: + # example_stop_events_from: ['@bad:example.com'] # Uncomment to allow non-server-admin users to create groups on this server diff --git a/docs/spam_checker.md b/docs/spam_checker.md index 5b5f5000b7..eb10e115f9 100644 --- a/docs/spam_checker.md +++ b/docs/spam_checker.md @@ -64,10 +64,12 @@ class ExampleSpamChecker: Modify the `spam_checker` section of your `homeserver.yaml` in the following manner: -`module` should point to the fully qualified Python class that implements your -custom logic, e.g. `my_module.ExampleSpamChecker`. +Create a list entry with the keys `module` and `config`. -`config` is a dictionary that gets passed to the spam checker class. +* `module` should point to the fully qualified Python class that implements your + custom logic, e.g. `my_module.ExampleSpamChecker`. + +* `config` is a dictionary that gets passed to the spam checker class. ### Example @@ -75,12 +77,15 @@ This section might look like: ```yaml spam_checker: - module: my_module.ExampleSpamChecker - config: - # Enable or disable a specific option in ExampleSpamChecker. - my_custom_option: true + - module: my_module.ExampleSpamChecker + config: + # Enable or disable a specific option in ExampleSpamChecker. + my_custom_option: true ``` +More spam checkers can be added in tandem by appending more items to the list. An +action is blocked when at least one of the configured spam checkers flags it. + ## Examples The [Mjolnir](https://github.com/matrix-org/mjolnir) project is a full fledged diff --git a/synapse/config/spam_checker.py b/synapse/config/spam_checker.py index 36e0ddab5c..3d067d29db 100644 --- a/synapse/config/spam_checker.py +++ b/synapse/config/spam_checker.py @@ -13,6 +13,9 @@ # See the License for the specific language governing permissions and # limitations under the License. +from typing import Any, Dict, List, Tuple + +from synapse.config import ConfigError from synapse.util.module_loader import load_module from ._base import Config @@ -22,16 +25,35 @@ class SpamCheckerConfig(Config): section = "spamchecker" def read_config(self, config, **kwargs): - self.spam_checker = None + self.spam_checkers = [] # type: List[Tuple[Any, Dict]] + + spam_checkers = config.get("spam_checker") or [] + if isinstance(spam_checkers, dict): + # The spam_checker config option used to only support one + # spam checker, and thus was simply a dictionary with module + # and config keys. Support this old behaviour by checking + # to see if the option resolves to a dictionary + self.spam_checkers.append(load_module(spam_checkers)) + elif isinstance(spam_checkers, list): + for spam_checker in spam_checkers: + if not isinstance(spam_checker, dict): + raise ConfigError("spam_checker syntax is incorrect") - provider = config.get("spam_checker", None) - if provider is not None: - self.spam_checker = load_module(provider) + self.spam_checkers.append(load_module(spam_checker)) + else: + raise ConfigError("spam_checker syntax is incorrect") def generate_config_section(self, **kwargs): return """\ - #spam_checker: - # module: "my_custom_project.SuperSpamChecker" - # config: - # example_option: 'things' + # Spam checkers are third-party modules that can block specific actions + # of local users, such as creating rooms and registering undesirable + # usernames, as well as remote users by redacting incoming events. + # + spam_checker: + #- module: "my_custom_project.SuperSpamChecker" + # config: + # example_option: 'things' + #- module: "some_other_project.BadEventStopper" + # config: + # example_stop_events_from: ['@bad:example.com'] """ diff --git a/synapse/events/spamcheck.py b/synapse/events/spamcheck.py index a23b6b7b61..1ffc9525d1 100644 --- a/synapse/events/spamcheck.py +++ b/synapse/events/spamcheck.py @@ -15,7 +15,7 @@ # limitations under the License. import inspect -from typing import Dict +from typing import Any, Dict, List from synapse.spam_checker_api import SpamCheckerApi @@ -26,24 +26,17 @@ if MYPY: class SpamChecker(object): def __init__(self, hs: "synapse.server.HomeServer"): - self.spam_checker = None + self.spam_checkers = [] # type: List[Any] - module = None - config = None - try: - module, config = hs.config.spam_checker - except Exception: - pass - - if module is not None: + for module, config in hs.config.spam_checkers: # Older spam checkers don't accept the `api` argument, so we # try and detect support. spam_args = inspect.getfullargspec(module) if "api" in spam_args.args: api = SpamCheckerApi(hs) - self.spam_checker = module(config=config, api=api) + self.spam_checkers.append(module(config=config, api=api)) else: - self.spam_checker = module(config=config) + self.spam_checkers.append(module(config=config)) def check_event_for_spam(self, event: "synapse.events.EventBase") -> bool: """Checks if a given event is considered "spammy" by this server. @@ -58,10 +51,11 @@ class SpamChecker(object): Returns: True if the event is spammy. """ - if self.spam_checker is None: - return False + for spam_checker in self.spam_checkers: + if spam_checker.check_event_for_spam(event): + return True - return self.spam_checker.check_event_for_spam(event) + return False def user_may_invite( self, inviter_userid: str, invitee_userid: str, room_id: str @@ -78,12 +72,14 @@ class SpamChecker(object): Returns: True if the user may send an invite, otherwise False """ - if self.spam_checker is None: - return True + for spam_checker in self.spam_checkers: + if ( + spam_checker.user_may_invite(inviter_userid, invitee_userid, room_id) + is False + ): + return False - return self.spam_checker.user_may_invite( - inviter_userid, invitee_userid, room_id - ) + return True def user_may_create_room(self, userid: str) -> bool: """Checks if a given user may create a room @@ -96,10 +92,11 @@ class SpamChecker(object): Returns: True if the user may create a room, otherwise False """ - if self.spam_checker is None: - return True + for spam_checker in self.spam_checkers: + if spam_checker.user_may_create_room(userid) is False: + return False - return self.spam_checker.user_may_create_room(userid) + return True def user_may_create_room_alias(self, userid: str, room_alias: str) -> bool: """Checks if a given user may create a room alias @@ -113,10 +110,11 @@ class SpamChecker(object): Returns: True if the user may create a room alias, otherwise False """ - if self.spam_checker is None: - return True + for spam_checker in self.spam_checkers: + if spam_checker.user_may_create_room_alias(userid, room_alias) is False: + return False - return self.spam_checker.user_may_create_room_alias(userid, room_alias) + return True def user_may_publish_room(self, userid: str, room_id: str) -> bool: """Checks if a given user may publish a room to the directory @@ -130,10 +128,11 @@ class SpamChecker(object): Returns: True if the user may publish the room, otherwise False """ - if self.spam_checker is None: - return True + for spam_checker in self.spam_checkers: + if spam_checker.user_may_publish_room(userid, room_id) is False: + return False - return self.spam_checker.user_may_publish_room(userid, room_id) + return True def check_username_for_spam(self, user_profile: Dict[str, str]) -> bool: """Checks if a user ID or display name are considered "spammy" by this server. @@ -150,13 +149,14 @@ class SpamChecker(object): Returns: True if the user is spammy. """ - if self.spam_checker is None: - return False - - # For backwards compatibility, if the method does not exist on the spam checker, fallback to not interfering. - checker = getattr(self.spam_checker, "check_username_for_spam", None) - if not checker: - return False - # Make a copy of the user profile object to ensure the spam checker - # cannot modify it. - return checker(user_profile.copy()) + for spam_checker in self.spam_checkers: + # For backwards compatibility, only run if the method exists on the + # spam checker + checker = getattr(spam_checker, "check_username_for_spam", None) + if checker: + # Make a copy of the user profile object to ensure the spam checker + # cannot modify it. + if checker(user_profile.copy()): + return True + + return False diff --git a/tests/handlers/test_user_directory.py b/tests/handlers/test_user_directory.py index 7b92bdbc47..572df8d80b 100644 --- a/tests/handlers/test_user_directory.py +++ b/tests/handlers/test_user_directory.py @@ -185,7 +185,7 @@ class UserDirectoryTestCase(unittest.HomeserverTestCase): # Allow all users. return False - spam_checker.spam_checker = AllowAll() + spam_checker.spam_checkers = [AllowAll()] # The results do not change: # We get one search result when searching for user2 by user1. @@ -198,7 +198,7 @@ class UserDirectoryTestCase(unittest.HomeserverTestCase): # All users are spammy. return True - spam_checker.spam_checker = BlockAll() + spam_checker.spam_checkers = [BlockAll()] # User1 now gets no search results for any of the other users. s = self.get_success(self.handler.search_users(u1, "user2", 10)) -- cgit 1.4.1 From 0ad6d28b0dec06d5e7478984280b4e81ef0f0256 Mon Sep 17 00:00:00 2001 From: Patrick Cloke Date: Fri, 8 May 2020 16:08:58 -0400 Subject: Rework UI Auth session validation for registration (#7455) Be less strict about validation of UI authentication sessions during registration to match client expecations. --- changelog.d/7455.bugfix | 1 + synapse/handlers/auth.py | 54 +++-- synapse/rest/client/v2_alpha/register.py | 1 + synapse/storage/data_stores/main/ui_auth.py | 21 ++ tests/rest/client/v2_alpha/test_auth.py | 304 ++++++++++++++++++++-------- tox.ini | 1 + 6 files changed, 280 insertions(+), 102 deletions(-) create mode 100644 changelog.d/7455.bugfix (limited to 'tests') diff --git a/changelog.d/7455.bugfix b/changelog.d/7455.bugfix new file mode 100644 index 0000000000..d1693a7f22 --- /dev/null +++ b/changelog.d/7455.bugfix @@ -0,0 +1 @@ +Ensure that a user inteactive authentication session is tied to a single request. diff --git a/synapse/handlers/auth.py b/synapse/handlers/auth.py index 7613e5b6ab..9c71702371 100644 --- a/synapse/handlers/auth.py +++ b/synapse/handlers/auth.py @@ -252,6 +252,7 @@ class AuthHandler(BaseHandler): clientdict: Dict[str, Any], clientip: str, description: str, + validate_clientdict: bool = True, ) -> Tuple[dict, dict, str]: """ Takes a dictionary sent by the client in the login / registration @@ -277,6 +278,10 @@ class AuthHandler(BaseHandler): description: A human readable string to be displayed to the user that describes the operation happening on their account. + validate_clientdict: Whether to validate that the operation happening + on the account has not changed. If this is false, + the client dict is persisted instead of validated. + Returns: A tuple of (creds, params, session_id). @@ -317,30 +322,51 @@ class AuthHandler(BaseHandler): except StoreError: raise SynapseError(400, "Unknown session ID: %s" % (sid,)) + # If the client provides parameters, update what is persisted, + # otherwise use whatever was last provided. + # + # This was designed to allow the client to omit the parameters + # and just supply the session in subsequent calls so it split + # auth between devices by just sharing the session, (eg. so you + # could continue registration from your phone having clicked the + # email auth link on there). It's probably too open to abuse + # because it lets unauthenticated clients store arbitrary objects + # on a homeserver. + # + # Revisit: Assuming the REST APIs do sensible validation, the data + # isn't arbitrary. + # + # Note that the registration endpoint explicitly removes the + # "initial_device_display_name" parameter if it is provided + # without a "password" parameter. See the changes to + # synapse.rest.client.v2_alpha.register.RegisterRestServlet.on_POST + # in commit 544722bad23fc31056b9240189c3cbbbf0ffd3f9. if not clientdict: - # This was designed to allow the client to omit the parameters - # and just supply the session in subsequent calls so it split - # auth between devices by just sharing the session, (eg. so you - # could continue registration from your phone having clicked the - # email auth link on there). It's probably too open to abuse - # because it lets unauthenticated clients store arbitrary objects - # on a homeserver. - # Revisit: Assuming the REST APIs do sensible validation, the data - # isn't arbitrary. clientdict = session.clientdict # Ensure that the queried operation does not vary between stages of # the UI authentication session. This is done by generating a stable - # comparator based on the URI, method, and body (minus the auth dict) - # and storing it during the initial query. Subsequent queries ensure - # that this comparator has not changed. - comparator = (uri, method, clientdict) - if (session.uri, session.method, session.clientdict) != comparator: + # comparator based on the URI, method, and client dict (minus the + # auth dict) and storing it during the initial query. Subsequent + # queries ensure that this comparator has not changed. + if validate_clientdict: + session_comparator = (session.uri, session.method, session.clientdict) + comparator = (uri, method, clientdict) + else: + session_comparator = (session.uri, session.method) # type: ignore + comparator = (uri, method) # type: ignore + + if session_comparator != comparator: raise SynapseError( 403, "Requested operation has changed during the UI authentication session.", ) + # For backwards compatibility the registration endpoint persists + # changes to the client dict instead of validating them. + if not validate_clientdict: + await self.store.set_ui_auth_clientdict(sid, clientdict) + if not authdict: raise InteractiveAuthIncompleteError( self._auth_dict_for_flows(flows, session.session_id) diff --git a/synapse/rest/client/v2_alpha/register.py b/synapse/rest/client/v2_alpha/register.py index af08cc6cce..e77dd6bf92 100644 --- a/synapse/rest/client/v2_alpha/register.py +++ b/synapse/rest/client/v2_alpha/register.py @@ -516,6 +516,7 @@ class RegisterRestServlet(RestServlet): body, self.hs.get_ip_from_request(request), "register a new account", + validate_clientdict=False, ) # Check that we're not trying to register a denied 3pid. diff --git a/synapse/storage/data_stores/main/ui_auth.py b/synapse/storage/data_stores/main/ui_auth.py index c8eebc9378..1d8ee22fb1 100644 --- a/synapse/storage/data_stores/main/ui_auth.py +++ b/synapse/storage/data_stores/main/ui_auth.py @@ -172,6 +172,27 @@ class UIAuthWorkerStore(SQLBaseStore): return results + async def set_ui_auth_clientdict( + self, session_id: str, clientdict: JsonDict + ) -> None: + """ + Store an updated clientdict for a given session ID. + + Args: + session_id: The ID of this session as returned from check_auth + clientdict: + The dictionary from the client root level, not the 'auth' key. + """ + # The clientdict gets stored as JSON. + clientdict_json = json.dumps(clientdict) + + self.db.simple_update_one( + table="ui_auth_sessions", + keyvalues={"session_id": session_id}, + updatevalues={"clientdict": clientdict_json}, + desc="set_ui_auth_client_dict", + ) + async def set_ui_auth_session_data(self, session_id: str, key: str, value: Any): """ Store a key-value pair into the sessions data associated with this diff --git a/tests/rest/client/v2_alpha/test_auth.py b/tests/rest/client/v2_alpha/test_auth.py index 587be7b2e7..a56c50a5b7 100644 --- a/tests/rest/client/v2_alpha/test_auth.py +++ b/tests/rest/client/v2_alpha/test_auth.py @@ -12,16 +12,20 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. - +from typing import List, Union from twisted.internet.defer import succeed import synapse.rest.admin from synapse.api.constants import LoginType from synapse.handlers.ui_auth.checkers import UserInteractiveAuthChecker -from synapse.rest.client.v2_alpha import auth, register +from synapse.http.site import SynapseRequest +from synapse.rest.client.v1 import login +from synapse.rest.client.v2_alpha import auth, devices, register +from synapse.types import JsonDict from tests import unittest +from tests.server import FakeChannel class DummyRecaptchaChecker(UserInteractiveAuthChecker): @@ -34,11 +38,15 @@ class DummyRecaptchaChecker(UserInteractiveAuthChecker): return succeed(True) +class DummyPasswordChecker(UserInteractiveAuthChecker): + def check_auth(self, authdict, clientip): + return succeed(authdict["identifier"]["user"]) + + class FallbackAuthTests(unittest.HomeserverTestCase): servlets = [ auth.register_servlets, - synapse.rest.admin.register_servlets_for_client_rest_resource, register.register_servlets, ] hijack_auth = False @@ -59,79 +67,84 @@ class FallbackAuthTests(unittest.HomeserverTestCase): auth_handler = hs.get_auth_handler() auth_handler.checkers[LoginType.RECAPTCHA] = self.recaptcha_checker - @unittest.INFO - def test_fallback_captcha(self): - + def register(self, expected_response: int, body: JsonDict) -> FakeChannel: + """Make a register request.""" request, channel = self.make_request( - "POST", - "register", - {"username": "user", "type": "m.login.password", "password": "bar"}, - ) + "POST", "register", body + ) # type: SynapseRequest, FakeChannel self.render(request) - # Returns a 401 as per the spec - self.assertEqual(request.code, 401) - # Grab the session - session = channel.json_body["session"] - # Assert our configured public key is being given - self.assertEqual( - channel.json_body["params"]["m.login.recaptcha"]["public_key"], "brokencake" - ) + self.assertEqual(request.code, expected_response) + return channel + + def recaptcha( + self, session: str, expected_post_response: int, post_session: str = None + ) -> None: + """Get and respond to a fallback recaptcha. Returns the second request.""" + if post_session is None: + post_session = session request, channel = self.make_request( "GET", "auth/m.login.recaptcha/fallback/web?session=" + session - ) + ) # type: SynapseRequest, FakeChannel self.render(request) self.assertEqual(request.code, 200) request, channel = self.make_request( "POST", "auth/m.login.recaptcha/fallback/web?session=" - + session + + post_session + "&g-recaptcha-response=a", ) self.render(request) - self.assertEqual(request.code, 200) + self.assertEqual(request.code, expected_post_response) # The recaptcha handler is called with the response given attempts = self.recaptcha_checker.recaptcha_attempts self.assertEqual(len(attempts), 1) self.assertEqual(attempts[0][0]["response"], "a") - # also complete the dummy auth - request, channel = self.make_request( - "POST", "register", {"auth": {"session": session, "type": "m.login.dummy"}} + @unittest.INFO + def test_fallback_captcha(self): + """Ensure that fallback auth via a captcha works.""" + # Returns a 401 as per the spec + channel = self.register( + 401, {"username": "user", "type": "m.login.password", "password": "bar"}, ) - self.render(request) + + # Grab the session + session = channel.json_body["session"] + # Assert our configured public key is being given + self.assertEqual( + channel.json_body["params"]["m.login.recaptcha"]["public_key"], "brokencake" + ) + + # Complete the recaptcha step. + self.recaptcha(session, 200) + + # also complete the dummy auth + self.register(200, {"auth": {"session": session, "type": "m.login.dummy"}}) # Now we should have fulfilled a complete auth flow, including # the recaptcha fallback step, we can then send a # request to the register API with the session in the authdict. - request, channel = self.make_request( - "POST", "register", {"auth": {"session": session}} - ) - self.render(request) - self.assertEqual(channel.code, 200) + channel = self.register(200, {"auth": {"session": session}}) # We're given a registered user. self.assertEqual(channel.json_body["user_id"], "@user:test") - def test_cannot_change_operation(self): + def test_legacy_registration(self): """ - The initial requested operation cannot be modified during the user interactive authentication session. + Registration allows the parameters to vary through the process. """ # Make the initial request to register. (Later on a different password # will be used.) - request, channel = self.make_request( - "POST", - "register", - {"username": "user", "type": "m.login.password", "password": "bar"}, + # Returns a 401 as per the spec + channel = self.register( + 401, {"username": "user", "type": "m.login.password", "password": "bar"}, ) - self.render(request) - # Returns a 401 as per the spec - self.assertEqual(request.code, 401) # Grab the session session = channel.json_body["session"] # Assert our configured public key is being given @@ -139,65 +152,39 @@ class FallbackAuthTests(unittest.HomeserverTestCase): channel.json_body["params"]["m.login.recaptcha"]["public_key"], "brokencake" ) - request, channel = self.make_request( - "GET", "auth/m.login.recaptcha/fallback/web?session=" + session - ) - self.render(request) - self.assertEqual(request.code, 200) - - request, channel = self.make_request( - "POST", - "auth/m.login.recaptcha/fallback/web?session=" - + session - + "&g-recaptcha-response=a", - ) - self.render(request) - self.assertEqual(request.code, 200) - - # The recaptcha handler is called with the response given - attempts = self.recaptcha_checker.recaptcha_attempts - self.assertEqual(len(attempts), 1) - self.assertEqual(attempts[0][0]["response"], "a") + # Complete the recaptcha step. + self.recaptcha(session, 200) # also complete the dummy auth - request, channel = self.make_request( - "POST", "register", {"auth": {"session": session, "type": "m.login.dummy"}} - ) - self.render(request) + self.register(200, {"auth": {"session": session, "type": "m.login.dummy"}}) # Now we should have fulfilled a complete auth flow, including # the recaptcha fallback step. Make the initial request again, but - # with a different password. This causes the request to fail since the - # operaiton was modified during the ui auth session. - request, channel = self.make_request( - "POST", - "register", + # with a changed password. This still completes. + channel = self.register( + 200, { "username": "user", "type": "m.login.password", - "password": "foo", # Note this doesn't match the original request. + "password": "foo", # Note that this is different. "auth": {"session": session}, }, ) - self.render(request) - self.assertEqual(channel.code, 403) + + # We're given a registered user. + self.assertEqual(channel.json_body["user_id"], "@user:test") def test_complete_operation_unknown_session(self): """ Attempting to mark an invalid session as complete should error. """ - # Make the initial request to register. (Later on a different password # will be used.) - request, channel = self.make_request( - "POST", - "register", - {"username": "user", "type": "m.login.password", "password": "bar"}, + # Returns a 401 as per the spec + channel = self.register( + 401, {"username": "user", "type": "m.login.password", "password": "bar"} ) - self.render(request) - # Returns a 401 as per the spec - self.assertEqual(request.code, 401) # Grab the session session = channel.json_body["session"] # Assert our configured public key is being given @@ -205,19 +192,160 @@ class FallbackAuthTests(unittest.HomeserverTestCase): channel.json_body["params"]["m.login.recaptcha"]["public_key"], "brokencake" ) + # Attempt to complete the recaptcha step with an unknown session. + # This results in an error. + self.recaptcha(session, 400, session + "unknown") + + +class UIAuthTests(unittest.HomeserverTestCase): + servlets = [ + auth.register_servlets, + devices.register_servlets, + login.register_servlets, + synapse.rest.admin.register_servlets_for_client_rest_resource, + register.register_servlets, + ] + + def prepare(self, reactor, clock, hs): + auth_handler = hs.get_auth_handler() + auth_handler.checkers[LoginType.PASSWORD] = DummyPasswordChecker(hs) + + self.user_pass = "pass" + self.user = self.register_user("test", self.user_pass) + self.user_tok = self.login("test", self.user_pass) + + def get_device_ids(self) -> List[str]: + # Get the list of devices so one can be deleted. request, channel = self.make_request( - "GET", "auth/m.login.recaptcha/fallback/web?session=" + session - ) + "GET", "devices", access_token=self.user_tok, + ) # type: SynapseRequest, FakeChannel self.render(request) + + # Get the ID of the device. self.assertEqual(request.code, 200) + return [d["device_id"] for d in channel.json_body["devices"]] - # Attempt to complete an unknown session, which should return an error. - unknown_session = session + "unknown" + def delete_device( + self, device: str, expected_response: int, body: Union[bytes, JsonDict] = b"" + ) -> FakeChannel: + """Delete an individual device.""" request, channel = self.make_request( - "POST", - "auth/m.login.recaptcha/fallback/web?session=" - + unknown_session - + "&g-recaptcha-response=a", - ) + "DELETE", "devices/" + device, body, access_token=self.user_tok + ) # type: SynapseRequest, FakeChannel self.render(request) - self.assertEqual(request.code, 400) + + # Ensure the response is sane. + self.assertEqual(request.code, expected_response) + + return channel + + def delete_devices(self, expected_response: int, body: JsonDict) -> FakeChannel: + """Delete 1 or more devices.""" + # Note that this uses the delete_devices endpoint so that we can modify + # the payload half-way through some tests. + request, channel = self.make_request( + "POST", "delete_devices", body, access_token=self.user_tok, + ) # type: SynapseRequest, FakeChannel + self.render(request) + + # Ensure the response is sane. + self.assertEqual(request.code, expected_response) + + return channel + + def test_ui_auth(self): + """ + Test user interactive authentication outside of registration. + """ + device_id = self.get_device_ids()[0] + + # Attempt to delete this device. + # Returns a 401 as per the spec + channel = self.delete_device(device_id, 401) + + # Grab the session + session = channel.json_body["session"] + # Ensure that flows are what is expected. + self.assertIn({"stages": ["m.login.password"]}, channel.json_body["flows"]) + + # Make another request providing the UI auth flow. + self.delete_device( + device_id, + 200, + { + "auth": { + "type": "m.login.password", + "identifier": {"type": "m.id.user", "user": self.user}, + "password": self.user_pass, + "session": session, + }, + }, + ) + + def test_cannot_change_body(self): + """ + The initial requested client dict cannot be modified during the user interactive authentication session. + """ + # Create a second login. + self.login("test", self.user_pass) + + device_ids = self.get_device_ids() + self.assertEqual(len(device_ids), 2) + + # Attempt to delete the first device. + # Returns a 401 as per the spec + channel = self.delete_devices(401, {"devices": [device_ids[0]]}) + + # Grab the session + session = channel.json_body["session"] + # Ensure that flows are what is expected. + self.assertIn({"stages": ["m.login.password"]}, channel.json_body["flows"]) + + # Make another request providing the UI auth flow, but try to delete the + # second device. This results in an error. + self.delete_devices( + 403, + { + "devices": [device_ids[1]], + "auth": { + "type": "m.login.password", + "identifier": {"type": "m.id.user", "user": self.user}, + "password": self.user_pass, + "session": session, + }, + }, + ) + + def test_cannot_change_uri(self): + """ + The initial requested URI cannot be modified during the user interactive authentication session. + """ + # Create a second login. + self.login("test", self.user_pass) + + device_ids = self.get_device_ids() + self.assertEqual(len(device_ids), 2) + + # Attempt to delete the first device. + # Returns a 401 as per the spec + channel = self.delete_device(device_ids[0], 401) + + # Grab the session + session = channel.json_body["session"] + # Ensure that flows are what is expected. + self.assertIn({"stages": ["m.login.password"]}, channel.json_body["flows"]) + + # Make another request providing the UI auth flow, but try to delete the + # second device. This results in an error. + self.delete_device( + device_ids[1], + 403, + { + "auth": { + "type": "m.login.password", + "identifier": {"type": "m.id.user", "user": self.user}, + "password": self.user_pass, + "session": session, + }, + }, + ) diff --git a/tox.ini b/tox.ini index eccc44e436..8aef52021d 100644 --- a/tox.ini +++ b/tox.ini @@ -207,6 +207,7 @@ commands = mypy \ synapse/util/caches/stream_change_cache.py \ tests/replication/tcp/streams \ tests/test_utils \ + tests/rest/client/v2_alpha/test_auth.py \ tests/util/test_stream_change_cache.py # To find all folders that pass mypy you run: -- cgit 1.4.1 From 7cb8b4bc67042a39bd1b0e05df46089a2fce1955 Mon Sep 17 00:00:00 2001 From: Amber Brown Date: Tue, 12 May 2020 03:45:23 +1000 Subject: Allow configuration of Synapse's cache without using synctl or environment variables (#6391) --- changelog.d/6391.feature | 1 + docs/sample_config.yaml | 43 +++++- synapse/api/auth.py | 4 +- synapse/app/homeserver.py | 5 +- synapse/config/cache.py | 164 ++++++++++++++++++++++ synapse/config/database.py | 6 - synapse/config/homeserver.py | 2 + synapse/http/client.py | 6 +- synapse/metrics/_exposition.py | 12 +- synapse/push/bulk_push_rule_evaluator.py | 4 +- synapse/push/push_rule_evaluator.py | 4 +- synapse/replication/slave/storage/client_ips.py | 3 +- synapse/state/__init__.py | 4 +- synapse/storage/data_stores/main/client_ips.py | 3 +- synapse/storage/data_stores/main/events_worker.py | 5 +- synapse/storage/data_stores/state/store.py | 6 +- synapse/util/caches/__init__.py | 144 ++++++++++--------- synapse/util/caches/descriptors.py | 36 ++++- synapse/util/caches/expiringcache.py | 29 +++- synapse/util/caches/lrucache.py | 52 +++++-- synapse/util/caches/response_cache.py | 2 +- synapse/util/caches/stream_change_cache.py | 33 ++++- synapse/util/caches/ttlcache.py | 2 +- tests/config/test_cache.py | 127 +++++++++++++++++ tests/storage/test__base.py | 8 +- tests/storage/test_appservice.py | 10 +- tests/storage/test_base.py | 3 +- tests/test_metrics.py | 34 +++++ tests/util/test_expiring_cache.py | 2 +- tests/util/test_lrucache.py | 6 +- tests/util/test_stream_change_cache.py | 5 +- tests/utils.py | 1 + 32 files changed, 620 insertions(+), 146 deletions(-) create mode 100644 changelog.d/6391.feature create mode 100644 synapse/config/cache.py create mode 100644 tests/config/test_cache.py (limited to 'tests') diff --git a/changelog.d/6391.feature b/changelog.d/6391.feature new file mode 100644 index 0000000000..f123426e23 --- /dev/null +++ b/changelog.d/6391.feature @@ -0,0 +1 @@ +Synapse's cache factor can now be configured in `homeserver.yaml` by the `caches.global_factor` setting. Additionally, `caches.per_cache_factors` controls the cache factors for individual caches. diff --git a/docs/sample_config.yaml b/docs/sample_config.yaml index 5abeaf519b..8a8415b9a2 100644 --- a/docs/sample_config.yaml +++ b/docs/sample_config.yaml @@ -603,6 +603,45 @@ acme: +## Caching ## + +# Caching can be configured through the following options. +# +# A cache 'factor' is a multiplier that can be applied to each of +# Synapse's caches in order to increase or decrease the maximum +# number of entries that can be stored. + +# The number of events to cache in memory. Not affected by +# caches.global_factor. +# +#event_cache_size: 10K + +caches: + # Controls the global cache factor, which is the default cache factor + # for all caches if a specific factor for that cache is not otherwise + # set. + # + # This can also be set by the "SYNAPSE_CACHE_FACTOR" environment + # variable. Setting by environment variable takes priority over + # setting through the config file. + # + # Defaults to 0.5, which will half the size of all caches. + # + #global_factor: 1.0 + + # A dictionary of cache name to cache factor for that individual + # cache. Overrides the global cache factor for a given cache. + # + # These can also be set through environment variables comprised + # of "SYNAPSE_CACHE_FACTOR_" + the name of the cache in capital + # letters and underscores. Setting by environment variable + # takes priority over setting through the config file. + # Ex. SYNAPSE_CACHE_FACTOR_GET_USERS_WHO_SHARE_ROOM_WITH_USER=2.0 + # + per_cache_factors: + #get_users_who_share_room_with_user: 2.0 + + ## Database ## # The 'database' setting defines the database that synapse uses to store all of @@ -646,10 +685,6 @@ database: args: database: DATADIR/homeserver.db -# Number of events to cache in memory. -# -#event_cache_size: 10K - ## Logging ## diff --git a/synapse/api/auth.py b/synapse/api/auth.py index 1ad5ff9410..e009b1a760 100644 --- a/synapse/api/auth.py +++ b/synapse/api/auth.py @@ -37,7 +37,7 @@ from synapse.api.errors import ( from synapse.api.room_versions import KNOWN_ROOM_VERSIONS from synapse.events import EventBase from synapse.types import StateMap, UserID -from synapse.util.caches import CACHE_SIZE_FACTOR, register_cache +from synapse.util.caches import register_cache from synapse.util.caches.lrucache import LruCache from synapse.util.metrics import Measure @@ -73,7 +73,7 @@ class Auth(object): self.store = hs.get_datastore() self.state = hs.get_state_handler() - self.token_cache = LruCache(CACHE_SIZE_FACTOR * 10000) + self.token_cache = LruCache(10000) register_cache("cache", "token_cache", self.token_cache) self._auth_blocking = AuthBlocking(self.hs) diff --git a/synapse/app/homeserver.py b/synapse/app/homeserver.py index bc8695d8dd..d7f337e586 100644 --- a/synapse/app/homeserver.py +++ b/synapse/app/homeserver.py @@ -69,7 +69,6 @@ from synapse.server import HomeServer from synapse.storage import DataStore from synapse.storage.engines import IncorrectDatabaseSetup from synapse.storage.prepare_database import UpgradeDatabaseException -from synapse.util.caches import CACHE_SIZE_FACTOR from synapse.util.httpresourcetree import create_resource_tree from synapse.util.manhole import manhole from synapse.util.module_loader import load_module @@ -516,8 +515,8 @@ def phone_stats_home(hs, stats, stats_process=_stats_process): daily_sent_messages = yield hs.get_datastore().count_daily_sent_messages() stats["daily_sent_messages"] = daily_sent_messages - stats["cache_factor"] = CACHE_SIZE_FACTOR - stats["event_cache_size"] = hs.config.event_cache_size + stats["cache_factor"] = hs.config.caches.global_factor + stats["event_cache_size"] = hs.config.caches.event_cache_size # # Performance statistics diff --git a/synapse/config/cache.py b/synapse/config/cache.py new file mode 100644 index 0000000000..91036a012e --- /dev/null +++ b/synapse/config/cache.py @@ -0,0 +1,164 @@ +# -*- coding: utf-8 -*- +# Copyright 2019 Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import os +from typing import Callable, Dict + +from ._base import Config, ConfigError + +# The prefix for all cache factor-related environment variables +_CACHES = {} +_CACHE_PREFIX = "SYNAPSE_CACHE_FACTOR" +_DEFAULT_FACTOR_SIZE = 0.5 +_DEFAULT_EVENT_CACHE_SIZE = "10K" + + +class CacheProperties(object): + def __init__(self): + # The default factor size for all caches + self.default_factor_size = float( + os.environ.get(_CACHE_PREFIX, _DEFAULT_FACTOR_SIZE) + ) + self.resize_all_caches_func = None + + +properties = CacheProperties() + + +def add_resizable_cache(cache_name: str, cache_resize_callback: Callable): + """Register a cache that's size can dynamically change + + Args: + cache_name: A reference to the cache + cache_resize_callback: A callback function that will be ran whenever + the cache needs to be resized + """ + _CACHES[cache_name.lower()] = cache_resize_callback + + # Ensure all loaded caches are sized appropriately + # + # This method should only run once the config has been read, + # as it uses values read from it + if properties.resize_all_caches_func: + properties.resize_all_caches_func() + + +class CacheConfig(Config): + section = "caches" + _environ = os.environ + + @staticmethod + def reset(): + """Resets the caches to their defaults. Used for tests.""" + properties.default_factor_size = float( + os.environ.get(_CACHE_PREFIX, _DEFAULT_FACTOR_SIZE) + ) + properties.resize_all_caches_func = None + _CACHES.clear() + + def generate_config_section(self, **kwargs): + return """\ + ## Caching ## + + # Caching can be configured through the following options. + # + # A cache 'factor' is a multiplier that can be applied to each of + # Synapse's caches in order to increase or decrease the maximum + # number of entries that can be stored. + + # The number of events to cache in memory. Not affected by + # caches.global_factor. + # + #event_cache_size: 10K + + caches: + # Controls the global cache factor, which is the default cache factor + # for all caches if a specific factor for that cache is not otherwise + # set. + # + # This can also be set by the "SYNAPSE_CACHE_FACTOR" environment + # variable. Setting by environment variable takes priority over + # setting through the config file. + # + # Defaults to 0.5, which will half the size of all caches. + # + #global_factor: 1.0 + + # A dictionary of cache name to cache factor for that individual + # cache. Overrides the global cache factor for a given cache. + # + # These can also be set through environment variables comprised + # of "SYNAPSE_CACHE_FACTOR_" + the name of the cache in capital + # letters and underscores. Setting by environment variable + # takes priority over setting through the config file. + # Ex. SYNAPSE_CACHE_FACTOR_GET_USERS_WHO_SHARE_ROOM_WITH_USER=2.0 + # + per_cache_factors: + #get_users_who_share_room_with_user: 2.0 + """ + + def read_config(self, config, **kwargs): + self.event_cache_size = self.parse_size( + config.get("event_cache_size", _DEFAULT_EVENT_CACHE_SIZE) + ) + self.cache_factors = {} # type: Dict[str, float] + + cache_config = config.get("caches") or {} + self.global_factor = cache_config.get( + "global_factor", properties.default_factor_size + ) + if not isinstance(self.global_factor, (int, float)): + raise ConfigError("caches.global_factor must be a number.") + + # Set the global one so that it's reflected in new caches + properties.default_factor_size = self.global_factor + + # Load cache factors from the config + individual_factors = cache_config.get("per_cache_factors") or {} + if not isinstance(individual_factors, dict): + raise ConfigError("caches.per_cache_factors must be a dictionary") + + # Override factors from environment if necessary + individual_factors.update( + { + key[len(_CACHE_PREFIX) + 1 :].lower(): float(val) + for key, val in self._environ.items() + if key.startswith(_CACHE_PREFIX + "_") + } + ) + + for cache, factor in individual_factors.items(): + if not isinstance(factor, (int, float)): + raise ConfigError( + "caches.per_cache_factors.%s must be a number" % (cache.lower(),) + ) + self.cache_factors[cache.lower()] = factor + + # Resize all caches (if necessary) with the new factors we've loaded + self.resize_all_caches() + + # Store this function so that it can be called from other classes without + # needing an instance of Config + properties.resize_all_caches_func = self.resize_all_caches + + def resize_all_caches(self): + """Ensure all cache sizes are up to date + + For each cache, run the mapped callback function with either + a specific cache factor or the default, global one. + """ + for cache_name, callback in _CACHES.items(): + new_factor = self.cache_factors.get(cache_name, self.global_factor) + callback(new_factor) diff --git a/synapse/config/database.py b/synapse/config/database.py index 5b662d1b01..1064c2697b 100644 --- a/synapse/config/database.py +++ b/synapse/config/database.py @@ -68,10 +68,6 @@ database: name: sqlite3 args: database: %(database_path)s - -# Number of events to cache in memory. -# -#event_cache_size: 10K """ @@ -116,8 +112,6 @@ class DatabaseConfig(Config): self.databases = [] def read_config(self, config, **kwargs): - self.event_cache_size = self.parse_size(config.get("event_cache_size", "10K")) - # We *experimentally* support specifying multiple databases via the # `databases` key. This is a map from a label to database config in the # same format as the `database` config option, plus an extra diff --git a/synapse/config/homeserver.py b/synapse/config/homeserver.py index 996d3e6bf7..2c7b3a699f 100644 --- a/synapse/config/homeserver.py +++ b/synapse/config/homeserver.py @@ -17,6 +17,7 @@ from ._base import RootConfig from .api import ApiConfig from .appservice import AppServiceConfig +from .cache import CacheConfig from .captcha import CaptchaConfig from .cas import CasConfig from .consent_config import ConsentConfig @@ -55,6 +56,7 @@ class HomeServerConfig(RootConfig): config_classes = [ ServerConfig, TlsConfig, + CacheConfig, DatabaseConfig, LoggingConfig, RatelimitConfig, diff --git a/synapse/http/client.py b/synapse/http/client.py index 58eb47c69c..3cef747a4d 100644 --- a/synapse/http/client.py +++ b/synapse/http/client.py @@ -49,7 +49,6 @@ from synapse.http.proxyagent import ProxyAgent from synapse.logging.context import make_deferred_yieldable from synapse.logging.opentracing import set_tag, start_active_span, tags from synapse.util.async_helpers import timeout_deferred -from synapse.util.caches import CACHE_SIZE_FACTOR logger = logging.getLogger(__name__) @@ -241,7 +240,10 @@ class SimpleHttpClient(object): # tends to do so in batches, so we need to allow the pool to keep # lots of idle connections around. pool = HTTPConnectionPool(self.reactor) - pool.maxPersistentPerHost = max((100 * CACHE_SIZE_FACTOR, 5)) + # XXX: The justification for using the cache factor here is that larger instances + # will need both more cache and more connections. + # Still, this should probably be a separate dial + pool.maxPersistentPerHost = max((100 * hs.config.caches.global_factor, 5)) pool.cachedConnectionTimeout = 2 * 60 self.agent = ProxyAgent( diff --git a/synapse/metrics/_exposition.py b/synapse/metrics/_exposition.py index a248103191..ab7f948ed4 100644 --- a/synapse/metrics/_exposition.py +++ b/synapse/metrics/_exposition.py @@ -33,6 +33,8 @@ from prometheus_client import REGISTRY from twisted.web.resource import Resource +from synapse.util import caches + try: from prometheus_client.samples import Sample except ImportError: @@ -103,13 +105,15 @@ def nameify_sample(sample): def generate_latest(registry, emit_help=False): - output = [] - for metric in registry.collect(): + # Trigger the cache metrics to be rescraped, which updates the common + # metrics but do not produce metrics themselves + for collector in caches.collectors_by_name.values(): + collector.collect() - if metric.name.startswith("__unused"): - continue + output = [] + for metric in registry.collect(): if not metric.samples: # No samples, don't bother. continue diff --git a/synapse/push/bulk_push_rule_evaluator.py b/synapse/push/bulk_push_rule_evaluator.py index 433ca2f416..e75d964ac8 100644 --- a/synapse/push/bulk_push_rule_evaluator.py +++ b/synapse/push/bulk_push_rule_evaluator.py @@ -51,6 +51,7 @@ push_rules_delta_state_cache_metric = register_cache( "cache", "push_rules_delta_state_cache_metric", cache=[], # Meaningless size, as this isn't a cache that stores values + resizable=False, ) @@ -67,7 +68,8 @@ class BulkPushRuleEvaluator(object): self.room_push_rule_cache_metrics = register_cache( "cache", "room_push_rule_cache", - cache=[], # Meaningless size, as this isn't a cache that stores values + cache=[], # Meaningless size, as this isn't a cache that stores values, + resizable=False, ) @defer.inlineCallbacks diff --git a/synapse/push/push_rule_evaluator.py b/synapse/push/push_rule_evaluator.py index 4cd702b5fa..11032491af 100644 --- a/synapse/push/push_rule_evaluator.py +++ b/synapse/push/push_rule_evaluator.py @@ -22,7 +22,7 @@ from six import string_types from synapse.events import EventBase from synapse.types import UserID -from synapse.util.caches import CACHE_SIZE_FACTOR, register_cache +from synapse.util.caches import register_cache from synapse.util.caches.lrucache import LruCache logger = logging.getLogger(__name__) @@ -165,7 +165,7 @@ class PushRuleEvaluatorForEvent(object): # Caches (string, is_glob, word_boundary) -> regex for push. See _glob_matches -regex_cache = LruCache(50000 * CACHE_SIZE_FACTOR) +regex_cache = LruCache(50000) register_cache("cache", "regex_push_cache", regex_cache) diff --git a/synapse/replication/slave/storage/client_ips.py b/synapse/replication/slave/storage/client_ips.py index fbf996e33a..1a38f53dfb 100644 --- a/synapse/replication/slave/storage/client_ips.py +++ b/synapse/replication/slave/storage/client_ips.py @@ -15,7 +15,6 @@ from synapse.storage.data_stores.main.client_ips import LAST_SEEN_GRANULARITY from synapse.storage.database import Database -from synapse.util.caches import CACHE_SIZE_FACTOR from synapse.util.caches.descriptors import Cache from ._base import BaseSlavedStore @@ -26,7 +25,7 @@ class SlavedClientIpStore(BaseSlavedStore): super(SlavedClientIpStore, self).__init__(database, db_conn, hs) self.client_ip_last_seen = Cache( - name="client_ip_last_seen", keylen=4, max_entries=50000 * CACHE_SIZE_FACTOR + name="client_ip_last_seen", keylen=4, max_entries=50000 ) def insert_client_ip(self, user_id, access_token, ip, user_agent, device_id): diff --git a/synapse/state/__init__.py b/synapse/state/__init__.py index 4afefc6b1d..2fa529fcd0 100644 --- a/synapse/state/__init__.py +++ b/synapse/state/__init__.py @@ -35,7 +35,6 @@ from synapse.state import v1, v2 from synapse.storage.data_stores.main.events_worker import EventRedactBehaviour from synapse.types import StateMap from synapse.util.async_helpers import Linearizer -from synapse.util.caches import get_cache_factor_for from synapse.util.caches.expiringcache import ExpiringCache from synapse.util.metrics import Measure, measure_func @@ -53,7 +52,6 @@ state_groups_histogram = Histogram( KeyStateTuple = namedtuple("KeyStateTuple", ("context", "type", "state_key")) -SIZE_OF_CACHE = 100000 * get_cache_factor_for("state_cache") EVICTION_TIMEOUT_SECONDS = 60 * 60 @@ -447,7 +445,7 @@ class StateResolutionHandler(object): self._state_cache = ExpiringCache( cache_name="state_cache", clock=self.clock, - max_len=SIZE_OF_CACHE, + max_len=100000, expiry_ms=EVICTION_TIMEOUT_SECONDS * 1000, iterable=True, reset_expiry_on_get=True, diff --git a/synapse/storage/data_stores/main/client_ips.py b/synapse/storage/data_stores/main/client_ips.py index 92bc06919b..71f8d43a76 100644 --- a/synapse/storage/data_stores/main/client_ips.py +++ b/synapse/storage/data_stores/main/client_ips.py @@ -22,7 +22,6 @@ from twisted.internet import defer from synapse.metrics.background_process_metrics import wrap_as_background_process from synapse.storage._base import SQLBaseStore from synapse.storage.database import Database, make_tuple_comparison_clause -from synapse.util.caches import CACHE_SIZE_FACTOR from synapse.util.caches.descriptors import Cache logger = logging.getLogger(__name__) @@ -361,7 +360,7 @@ class ClientIpStore(ClientIpBackgroundUpdateStore): def __init__(self, database: Database, db_conn, hs): self.client_ip_last_seen = Cache( - name="client_ip_last_seen", keylen=4, max_entries=50000 * CACHE_SIZE_FACTOR + name="client_ip_last_seen", keylen=4, max_entries=50000 ) super(ClientIpStore, self).__init__(database, db_conn, hs) diff --git a/synapse/storage/data_stores/main/events_worker.py b/synapse/storage/data_stores/main/events_worker.py index 73df6b33ba..b8c1bbdf99 100644 --- a/synapse/storage/data_stores/main/events_worker.py +++ b/synapse/storage/data_stores/main/events_worker.py @@ -75,7 +75,10 @@ class EventsWorkerStore(SQLBaseStore): super(EventsWorkerStore, self).__init__(database, db_conn, hs) self._get_event_cache = Cache( - "*getEvent*", keylen=3, max_entries=hs.config.event_cache_size + "*getEvent*", + keylen=3, + max_entries=hs.config.caches.event_cache_size, + apply_cache_factor_from_config=False, ) self._event_fetch_lock = threading.Condition() diff --git a/synapse/storage/data_stores/state/store.py b/synapse/storage/data_stores/state/store.py index 57a5267663..f3ad1e4369 100644 --- a/synapse/storage/data_stores/state/store.py +++ b/synapse/storage/data_stores/state/store.py @@ -28,7 +28,6 @@ from synapse.storage.data_stores.state.bg_updates import StateBackgroundUpdateSt from synapse.storage.database import Database from synapse.storage.state import StateFilter from synapse.types import StateMap -from synapse.util.caches import get_cache_factor_for from synapse.util.caches.descriptors import cached from synapse.util.caches.dictionary_cache import DictionaryCache @@ -90,11 +89,10 @@ class StateGroupDataStore(StateBackgroundUpdateStore, SQLBaseStore): self._state_group_cache = DictionaryCache( "*stateGroupCache*", # TODO: this hasn't been tuned yet - 50000 * get_cache_factor_for("stateGroupCache"), + 50000, ) self._state_group_members_cache = DictionaryCache( - "*stateGroupMembersCache*", - 500000 * get_cache_factor_for("stateGroupMembersCache"), + "*stateGroupMembersCache*", 500000, ) @cached(max_entries=10000, iterable=True) diff --git a/synapse/util/caches/__init__.py b/synapse/util/caches/__init__.py index da5077b471..4b8a0c7a8f 100644 --- a/synapse/util/caches/__init__.py +++ b/synapse/util/caches/__init__.py @@ -1,6 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2015, 2016 OpenMarket Ltd -# Copyright 2019 The Matrix.org Foundation C.I.C. +# Copyright 2019, 2020 The Matrix.org Foundation C.I.C. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -15,27 +15,17 @@ # limitations under the License. import logging -import os -from typing import Dict +from typing import Callable, Dict, Optional import six from six.moves import intern -from prometheus_client.core import REGISTRY, Gauge, GaugeMetricFamily - -logger = logging.getLogger(__name__) - -CACHE_SIZE_FACTOR = float(os.environ.get("SYNAPSE_CACHE_FACTOR", 0.5)) +import attr +from prometheus_client.core import Gauge +from synapse.config.cache import add_resizable_cache -def get_cache_factor_for(cache_name): - env_var = "SYNAPSE_CACHE_FACTOR_" + cache_name.upper() - factor = os.environ.get(env_var) - if factor: - return float(factor) - - return CACHE_SIZE_FACTOR - +logger = logging.getLogger(__name__) caches_by_name = {} collectors_by_name = {} # type: Dict @@ -44,6 +34,7 @@ cache_size = Gauge("synapse_util_caches_cache:size", "", ["name"]) cache_hits = Gauge("synapse_util_caches_cache:hits", "", ["name"]) cache_evicted = Gauge("synapse_util_caches_cache:evicted_size", "", ["name"]) cache_total = Gauge("synapse_util_caches_cache:total", "", ["name"]) +cache_max_size = Gauge("synapse_util_caches_cache_max_size", "", ["name"]) response_cache_size = Gauge("synapse_util_caches_response_cache:size", "", ["name"]) response_cache_hits = Gauge("synapse_util_caches_response_cache:hits", "", ["name"]) @@ -53,67 +44,82 @@ response_cache_evicted = Gauge( response_cache_total = Gauge("synapse_util_caches_response_cache:total", "", ["name"]) -def register_cache(cache_type, cache_name, cache, collect_callback=None): - """Register a cache object for metric collection. +@attr.s +class CacheMetric(object): + + _cache = attr.ib() + _cache_type = attr.ib(type=str) + _cache_name = attr.ib(type=str) + _collect_callback = attr.ib(type=Optional[Callable]) + + hits = attr.ib(default=0) + misses = attr.ib(default=0) + evicted_size = attr.ib(default=0) + + def inc_hits(self): + self.hits += 1 + + def inc_misses(self): + self.misses += 1 + + def inc_evictions(self, size=1): + self.evicted_size += size + + def describe(self): + return [] + + def collect(self): + try: + if self._cache_type == "response_cache": + response_cache_size.labels(self._cache_name).set(len(self._cache)) + response_cache_hits.labels(self._cache_name).set(self.hits) + response_cache_evicted.labels(self._cache_name).set(self.evicted_size) + response_cache_total.labels(self._cache_name).set( + self.hits + self.misses + ) + else: + cache_size.labels(self._cache_name).set(len(self._cache)) + cache_hits.labels(self._cache_name).set(self.hits) + cache_evicted.labels(self._cache_name).set(self.evicted_size) + cache_total.labels(self._cache_name).set(self.hits + self.misses) + if getattr(self._cache, "max_size", None): + cache_max_size.labels(self._cache_name).set(self._cache.max_size) + if self._collect_callback: + self._collect_callback() + except Exception as e: + logger.warning("Error calculating metrics for %s: %s", self._cache_name, e) + raise + + +def register_cache( + cache_type: str, + cache_name: str, + cache, + collect_callback: Optional[Callable] = None, + resizable: bool = True, + resize_callback: Optional[Callable] = None, +) -> CacheMetric: + """Register a cache object for metric collection and resizing. Args: - cache_type (str): - cache_name (str): name of the cache - cache (object): cache itself - collect_callback (callable|None): if not None, a function which is called during - metric collection to update additional metrics. + cache_type + cache_name: name of the cache + cache: cache itself + collect_callback: If given, a function which is called during metric + collection to update additional metrics. + resizable: Whether this cache supports being resized. + resize_callback: A function which can be called to resize the cache. Returns: CacheMetric: an object which provides inc_{hits,misses,evictions} methods """ + if resizable: + if not resize_callback: + resize_callback = getattr(cache, "set_cache_factor") + add_resizable_cache(cache_name, resize_callback) - # Check if the metric is already registered. Unregister it, if so. - # This usually happens during tests, as at runtime these caches are - # effectively singletons. + metric = CacheMetric(cache, cache_type, cache_name, collect_callback) metric_name = "cache_%s_%s" % (cache_type, cache_name) - if metric_name in collectors_by_name.keys(): - REGISTRY.unregister(collectors_by_name[metric_name]) - - class CacheMetric(object): - - hits = 0 - misses = 0 - evicted_size = 0 - - def inc_hits(self): - self.hits += 1 - - def inc_misses(self): - self.misses += 1 - - def inc_evictions(self, size=1): - self.evicted_size += size - - def describe(self): - return [] - - def collect(self): - try: - if cache_type == "response_cache": - response_cache_size.labels(cache_name).set(len(cache)) - response_cache_hits.labels(cache_name).set(self.hits) - response_cache_evicted.labels(cache_name).set(self.evicted_size) - response_cache_total.labels(cache_name).set(self.hits + self.misses) - else: - cache_size.labels(cache_name).set(len(cache)) - cache_hits.labels(cache_name).set(self.hits) - cache_evicted.labels(cache_name).set(self.evicted_size) - cache_total.labels(cache_name).set(self.hits + self.misses) - if collect_callback: - collect_callback() - except Exception as e: - logger.warning("Error calculating metrics for %s: %s", cache_name, e) - raise - - yield GaugeMetricFamily("__unused", "") - - metric = CacheMetric() - REGISTRY.register(metric) caches_by_name[cache_name] = cache collectors_by_name[metric_name] = metric return metric diff --git a/synapse/util/caches/descriptors.py b/synapse/util/caches/descriptors.py index 2e8f6543e5..cd48262420 100644 --- a/synapse/util/caches/descriptors.py +++ b/synapse/util/caches/descriptors.py @@ -13,6 +13,7 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. + import functools import inspect import logging @@ -30,7 +31,6 @@ from twisted.internet import defer from synapse.logging.context import make_deferred_yieldable, preserve_fn from synapse.util import unwrapFirstError from synapse.util.async_helpers import ObservableDeferred -from synapse.util.caches import get_cache_factor_for from synapse.util.caches.lrucache import LruCache from synapse.util.caches.treecache import TreeCache, iterate_tree_cache_entry @@ -81,7 +81,6 @@ class CacheEntry(object): class Cache(object): __slots__ = ( "cache", - "max_entries", "name", "keylen", "thread", @@ -89,7 +88,29 @@ class Cache(object): "_pending_deferred_cache", ) - def __init__(self, name, max_entries=1000, keylen=1, tree=False, iterable=False): + def __init__( + self, + name: str, + max_entries: int = 1000, + keylen: int = 1, + tree: bool = False, + iterable: bool = False, + apply_cache_factor_from_config: bool = True, + ): + """ + Args: + name: The name of the cache + max_entries: Maximum amount of entries that the cache will hold + keylen: The length of the tuple used as the cache key + tree: Use a TreeCache instead of a dict as the underlying cache type + iterable: If True, count each item in the cached object as an entry, + rather than each cached object + apply_cache_factor_from_config: Whether cache factors specified in the + config file affect `max_entries` + + Returns: + Cache + """ cache_type = TreeCache if tree else dict self._pending_deferred_cache = cache_type() @@ -99,6 +120,7 @@ class Cache(object): cache_type=cache_type, size_callback=(lambda d: len(d)) if iterable else None, evicted_callback=self._on_evicted, + apply_cache_factor_from_config=apply_cache_factor_from_config, ) self.name = name @@ -111,6 +133,10 @@ class Cache(object): collect_callback=self._metrics_collection_callback, ) + @property + def max_entries(self): + return self.cache.max_size + def _on_evicted(self, evicted_count): self.metrics.inc_evictions(evicted_count) @@ -370,13 +396,11 @@ class CacheDescriptor(_CacheDescriptorBase): cache_context=cache_context, ) - max_entries = int(max_entries * get_cache_factor_for(orig.__name__)) - self.max_entries = max_entries self.tree = tree self.iterable = iterable - def __get__(self, obj, objtype=None): + def __get__(self, obj, owner): cache = Cache( name=self.orig.__name__, max_entries=self.max_entries, diff --git a/synapse/util/caches/expiringcache.py b/synapse/util/caches/expiringcache.py index cddf1ed515..2726b67b6d 100644 --- a/synapse/util/caches/expiringcache.py +++ b/synapse/util/caches/expiringcache.py @@ -18,6 +18,7 @@ from collections import OrderedDict from six import iteritems, itervalues +from synapse.config import cache as cache_config from synapse.metrics.background_process_metrics import run_as_background_process from synapse.util.caches import register_cache @@ -51,15 +52,16 @@ class ExpiringCache(object): an item on access. Defaults to False. iterable (bool): If true, the size is calculated by summing the sizes of all entries, rather than the number of entries. - """ self._cache_name = cache_name + self._original_max_size = max_len + + self._max_size = int(max_len * cache_config.properties.default_factor_size) + self._clock = clock - self._max_len = max_len self._expiry_ms = expiry_ms - self._reset_expiry_on_get = reset_expiry_on_get self._cache = OrderedDict() @@ -82,9 +84,11 @@ class ExpiringCache(object): def __setitem__(self, key, value): now = self._clock.time_msec() self._cache[key] = _CacheEntry(now, value) + self.evict() + def evict(self): # Evict if there are now too many items - while self._max_len and len(self) > self._max_len: + while self._max_size and len(self) > self._max_size: _key, value = self._cache.popitem(last=False) if self.iterable: self.metrics.inc_evictions(len(value.value)) @@ -170,6 +174,23 @@ class ExpiringCache(object): else: return len(self._cache) + def set_cache_factor(self, factor: float) -> bool: + """ + Set the cache factor for this individual cache. + + This will trigger a resize if it changes, which may require evicting + items from the cache. + + Returns: + bool: Whether the cache changed size or not. + """ + new_size = int(self._original_max_size * factor) + if new_size != self._max_size: + self._max_size = new_size + self.evict() + return True + return False + class _CacheEntry(object): __slots__ = ["time", "value"] diff --git a/synapse/util/caches/lrucache.py b/synapse/util/caches/lrucache.py index 1536cb64f3..29fabac3cd 100644 --- a/synapse/util/caches/lrucache.py +++ b/synapse/util/caches/lrucache.py @@ -13,10 +13,11 @@ # See the License for the specific language governing permissions and # limitations under the License. - import threading from functools import wraps +from typing import Callable, Optional, Type, Union +from synapse.config import cache as cache_config from synapse.util.caches.treecache import TreeCache @@ -52,17 +53,18 @@ class LruCache(object): def __init__( self, - max_size, - keylen=1, - cache_type=dict, - size_callback=None, - evicted_callback=None, + max_size: int, + keylen: int = 1, + cache_type: Type[Union[dict, TreeCache]] = dict, + size_callback: Optional[Callable] = None, + evicted_callback: Optional[Callable] = None, + apply_cache_factor_from_config: bool = True, ): """ Args: - max_size (int): + max_size: The maximum amount of entries the cache can hold - keylen (int): + keylen: The length of the tuple used as the cache key cache_type (type): type of underlying cache to be used. Typically one of dict @@ -73,9 +75,23 @@ class LruCache(object): evicted_callback (func(int)|None): if not None, called on eviction with the size of the evicted entry + + apply_cache_factor_from_config (bool): If true, `max_size` will be + multiplied by a cache factor derived from the homeserver config """ cache = cache_type() self.cache = cache # Used for introspection. + + # Save the original max size, and apply the default size factor. + self._original_max_size = max_size + # We previously didn't apply the cache factor here, and as such some caches were + # not affected by the global cache factor. Add an option here to disable applying + # the cache factor when a cache is created + if apply_cache_factor_from_config: + self.max_size = int(max_size * cache_config.properties.default_factor_size) + else: + self.max_size = int(max_size) + list_root = _Node(None, None, None, None) list_root.next_node = list_root list_root.prev_node = list_root @@ -83,7 +99,7 @@ class LruCache(object): lock = threading.Lock() def evict(): - while cache_len() > max_size: + while cache_len() > self.max_size: todelete = list_root.prev_node evicted_len = delete_node(todelete) cache.pop(todelete.key, None) @@ -236,6 +252,7 @@ class LruCache(object): return key in cache self.sentinel = object() + self._on_resize = evict self.get = cache_get self.set = cache_set self.setdefault = cache_set_default @@ -266,3 +283,20 @@ class LruCache(object): def __contains__(self, key): return self.contains(key) + + def set_cache_factor(self, factor: float) -> bool: + """ + Set the cache factor for this individual cache. + + This will trigger a resize if it changes, which may require evicting + items from the cache. + + Returns: + bool: Whether the cache changed size or not. + """ + new_size = int(self._original_max_size * factor) + if new_size != self.max_size: + self.max_size = new_size + self._on_resize() + return True + return False diff --git a/synapse/util/caches/response_cache.py b/synapse/util/caches/response_cache.py index b68f9fe0d4..a6c60888e5 100644 --- a/synapse/util/caches/response_cache.py +++ b/synapse/util/caches/response_cache.py @@ -38,7 +38,7 @@ class ResponseCache(object): self.timeout_sec = timeout_ms / 1000.0 self._name = name - self._metrics = register_cache("response_cache", name, self) + self._metrics = register_cache("response_cache", name, self, resizable=False) def size(self): return len(self.pending_result_cache) diff --git a/synapse/util/caches/stream_change_cache.py b/synapse/util/caches/stream_change_cache.py index e54f80d76e..2a161bf244 100644 --- a/synapse/util/caches/stream_change_cache.py +++ b/synapse/util/caches/stream_change_cache.py @@ -14,6 +14,7 @@ # limitations under the License. import logging +import math from typing import Dict, FrozenSet, List, Mapping, Optional, Set, Union from six import integer_types @@ -46,7 +47,8 @@ class StreamChangeCache: max_size=10000, prefilled_cache: Optional[Mapping[EntityType, int]] = None, ): - self._max_size = int(max_size * caches.CACHE_SIZE_FACTOR) + self._original_max_size = max_size + self._max_size = math.floor(max_size) self._entity_to_key = {} # type: Dict[EntityType, int] # map from stream id to the a set of entities which changed at that stream id. @@ -58,12 +60,31 @@ class StreamChangeCache: # self._earliest_known_stream_pos = current_stream_pos self.name = name - self.metrics = caches.register_cache("cache", self.name, self._cache) + self.metrics = caches.register_cache( + "cache", self.name, self._cache, resize_callback=self.set_cache_factor + ) if prefilled_cache: for entity, stream_pos in prefilled_cache.items(): self.entity_has_changed(entity, stream_pos) + def set_cache_factor(self, factor: float) -> bool: + """ + Set the cache factor for this individual cache. + + This will trigger a resize if it changes, which may require evicting + items from the cache. + + Returns: + bool: Whether the cache changed size or not. + """ + new_size = math.floor(self._original_max_size * factor) + if new_size != self._max_size: + self.max_size = new_size + self._evict() + return True + return False + def has_entity_changed(self, entity: EntityType, stream_pos: int) -> bool: """Returns True if the entity may have been updated since stream_pos """ @@ -171,6 +192,7 @@ class StreamChangeCache: e1 = self._cache[stream_pos] = set() e1.add(entity) self._entity_to_key[entity] = stream_pos + self._evict() # if the cache is too big, remove entries while len(self._cache) > self._max_size: @@ -179,6 +201,13 @@ class StreamChangeCache: for entity in r: del self._entity_to_key[entity] + def _evict(self): + while len(self._cache) > self._max_size: + k, r = self._cache.popitem(0) + self._earliest_known_stream_pos = max(k, self._earliest_known_stream_pos) + for entity in r: + self._entity_to_key.pop(entity, None) + def get_max_pos_of_last_change(self, entity: EntityType) -> int: """Returns an upper bound of the stream id of the last change to an diff --git a/synapse/util/caches/ttlcache.py b/synapse/util/caches/ttlcache.py index 99646c7cf0..6437aa907e 100644 --- a/synapse/util/caches/ttlcache.py +++ b/synapse/util/caches/ttlcache.py @@ -38,7 +38,7 @@ class TTLCache(object): self._timer = timer - self._metrics = register_cache("ttl", cache_name, self) + self._metrics = register_cache("ttl", cache_name, self, resizable=False) def set(self, key, value, ttl): """Add/update an entry in the cache diff --git a/tests/config/test_cache.py b/tests/config/test_cache.py new file mode 100644 index 0000000000..2920279125 --- /dev/null +++ b/tests/config/test_cache.py @@ -0,0 +1,127 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from synapse.config._base import Config, RootConfig +from synapse.config.cache import CacheConfig, add_resizable_cache +from synapse.util.caches.lrucache import LruCache + +from tests.unittest import TestCase + + +class FakeServer(Config): + section = "server" + + +class TestConfig(RootConfig): + config_classes = [FakeServer, CacheConfig] + + +class CacheConfigTests(TestCase): + def setUp(self): + # Reset caches before each test + TestConfig().caches.reset() + + def test_individual_caches_from_environ(self): + """ + Individual cache factors will be loaded from the environment. + """ + config = {} + t = TestConfig() + t.caches._environ = { + "SYNAPSE_CACHE_FACTOR_SOMETHING_OR_OTHER": "2", + "SYNAPSE_NOT_CACHE": "BLAH", + } + t.read_config(config, config_dir_path="", data_dir_path="") + + self.assertEqual(dict(t.caches.cache_factors), {"something_or_other": 2.0}) + + def test_config_overrides_environ(self): + """ + Individual cache factors defined in the environment will take precedence + over those in the config. + """ + config = {"caches": {"per_cache_factors": {"foo": 2, "bar": 3}}} + t = TestConfig() + t.caches._environ = { + "SYNAPSE_CACHE_FACTOR_SOMETHING_OR_OTHER": "2", + "SYNAPSE_CACHE_FACTOR_FOO": 1, + } + t.read_config(config, config_dir_path="", data_dir_path="") + + self.assertEqual( + dict(t.caches.cache_factors), + {"foo": 1.0, "bar": 3.0, "something_or_other": 2.0}, + ) + + def test_individual_instantiated_before_config_load(self): + """ + If a cache is instantiated before the config is read, it will be given + the default cache size in the interim, and then resized once the config + is loaded. + """ + cache = LruCache(100) + + add_resizable_cache("foo", cache_resize_callback=cache.set_cache_factor) + self.assertEqual(cache.max_size, 50) + + config = {"caches": {"per_cache_factors": {"foo": 3}}} + t = TestConfig() + t.read_config(config, config_dir_path="", data_dir_path="") + + self.assertEqual(cache.max_size, 300) + + def test_individual_instantiated_after_config_load(self): + """ + If a cache is instantiated after the config is read, it will be + immediately resized to the correct size given the per_cache_factor if + there is one. + """ + config = {"caches": {"per_cache_factors": {"foo": 2}}} + t = TestConfig() + t.read_config(config, config_dir_path="", data_dir_path="") + + cache = LruCache(100) + add_resizable_cache("foo", cache_resize_callback=cache.set_cache_factor) + self.assertEqual(cache.max_size, 200) + + def test_global_instantiated_before_config_load(self): + """ + If a cache is instantiated before the config is read, it will be given + the default cache size in the interim, and then resized to the new + default cache size once the config is loaded. + """ + cache = LruCache(100) + add_resizable_cache("foo", cache_resize_callback=cache.set_cache_factor) + self.assertEqual(cache.max_size, 50) + + config = {"caches": {"global_factor": 4}} + t = TestConfig() + t.read_config(config, config_dir_path="", data_dir_path="") + + self.assertEqual(cache.max_size, 400) + + def test_global_instantiated_after_config_load(self): + """ + If a cache is instantiated after the config is read, it will be + immediately resized to the correct size given the global factor if there + is no per-cache factor. + """ + config = {"caches": {"global_factor": 1.5}} + t = TestConfig() + t.read_config(config, config_dir_path="", data_dir_path="") + + cache = LruCache(100) + add_resizable_cache("foo", cache_resize_callback=cache.set_cache_factor) + self.assertEqual(cache.max_size, 150) diff --git a/tests/storage/test__base.py b/tests/storage/test__base.py index e37260a820..5a50e4fdd4 100644 --- a/tests/storage/test__base.py +++ b/tests/storage/test__base.py @@ -25,8 +25,8 @@ from synapse.util.caches.descriptors import Cache, cached from tests import unittest -class CacheTestCase(unittest.TestCase): - def setUp(self): +class CacheTestCase(unittest.HomeserverTestCase): + def prepare(self, reactor, clock, homeserver): self.cache = Cache("test") def test_empty(self): @@ -96,7 +96,7 @@ class CacheTestCase(unittest.TestCase): cache.get(3) -class CacheDecoratorTestCase(unittest.TestCase): +class CacheDecoratorTestCase(unittest.HomeserverTestCase): @defer.inlineCallbacks def test_passthrough(self): class A(object): @@ -239,7 +239,7 @@ class CacheDecoratorTestCase(unittest.TestCase): callcount2 = [0] class A(object): - @cached(max_entries=4) # HACK: This makes it 2 due to cache factor + @cached(max_entries=2) def func(self, key): callcount[0] += 1 return key diff --git a/tests/storage/test_appservice.py b/tests/storage/test_appservice.py index 31710949a8..ef296e7dab 100644 --- a/tests/storage/test_appservice.py +++ b/tests/storage/test_appservice.py @@ -43,7 +43,7 @@ class ApplicationServiceStoreTestCase(unittest.TestCase): ) hs.config.app_service_config_files = self.as_yaml_files - hs.config.event_cache_size = 1 + hs.config.caches.event_cache_size = 1 hs.config.password_providers = [] self.as_token = "token1" @@ -110,7 +110,7 @@ class ApplicationServiceTransactionStoreTestCase(unittest.TestCase): ) hs.config.app_service_config_files = self.as_yaml_files - hs.config.event_cache_size = 1 + hs.config.caches.event_cache_size = 1 hs.config.password_providers = [] self.as_list = [ @@ -422,7 +422,7 @@ class ApplicationServiceStoreConfigTestCase(unittest.TestCase): ) hs.config.app_service_config_files = [f1, f2] - hs.config.event_cache_size = 1 + hs.config.caches.event_cache_size = 1 hs.config.password_providers = [] database = hs.get_datastores().databases[0] @@ -440,7 +440,7 @@ class ApplicationServiceStoreConfigTestCase(unittest.TestCase): ) hs.config.app_service_config_files = [f1, f2] - hs.config.event_cache_size = 1 + hs.config.caches.event_cache_size = 1 hs.config.password_providers = [] with self.assertRaises(ConfigError) as cm: @@ -464,7 +464,7 @@ class ApplicationServiceStoreConfigTestCase(unittest.TestCase): ) hs.config.app_service_config_files = [f1, f2] - hs.config.event_cache_size = 1 + hs.config.caches.event_cache_size = 1 hs.config.password_providers = [] with self.assertRaises(ConfigError) as cm: diff --git a/tests/storage/test_base.py b/tests/storage/test_base.py index cdee0a9e60..278961c331 100644 --- a/tests/storage/test_base.py +++ b/tests/storage/test_base.py @@ -51,7 +51,8 @@ class SQLBaseStoreTestCase(unittest.TestCase): config = Mock() config._disable_native_upserts = True - config.event_cache_size = 1 + config.caches = Mock() + config.caches.event_cache_size = 1 hs = TestHomeServer("test", config=config) sqlite_config = {"name": "sqlite3"} diff --git a/tests/test_metrics.py b/tests/test_metrics.py index 270f853d60..f5f63d8ed6 100644 --- a/tests/test_metrics.py +++ b/tests/test_metrics.py @@ -15,6 +15,7 @@ # limitations under the License. from synapse.metrics import REGISTRY, InFlightGauge, generate_latest +from synapse.util.caches.descriptors import Cache from tests import unittest @@ -129,3 +130,36 @@ class BuildInfoTests(unittest.TestCase): self.assertTrue(b"osversion=" in items[0]) self.assertTrue(b"pythonversion=" in items[0]) self.assertTrue(b"version=" in items[0]) + + +class CacheMetricsTests(unittest.HomeserverTestCase): + def test_cache_metric(self): + """ + Caches produce metrics reflecting their state when scraped. + """ + CACHE_NAME = "cache_metrics_test_fgjkbdfg" + cache = Cache(CACHE_NAME, max_entries=777) + + items = { + x.split(b"{")[0].decode("ascii"): x.split(b" ")[1].decode("ascii") + for x in filter( + lambda x: b"cache_metrics_test_fgjkbdfg" in x, + generate_latest(REGISTRY).split(b"\n"), + ) + } + + self.assertEqual(items["synapse_util_caches_cache_size"], "0.0") + self.assertEqual(items["synapse_util_caches_cache_max_size"], "777.0") + + cache.prefill("1", "hi") + + items = { + x.split(b"{")[0].decode("ascii"): x.split(b" ")[1].decode("ascii") + for x in filter( + lambda x: b"cache_metrics_test_fgjkbdfg" in x, + generate_latest(REGISTRY).split(b"\n"), + ) + } + + self.assertEqual(items["synapse_util_caches_cache_size"], "1.0") + self.assertEqual(items["synapse_util_caches_cache_max_size"], "777.0") diff --git a/tests/util/test_expiring_cache.py b/tests/util/test_expiring_cache.py index 50bc7702d2..49ffeebd0e 100644 --- a/tests/util/test_expiring_cache.py +++ b/tests/util/test_expiring_cache.py @@ -21,7 +21,7 @@ from tests.utils import MockClock from .. import unittest -class ExpiringCacheTestCase(unittest.TestCase): +class ExpiringCacheTestCase(unittest.HomeserverTestCase): def test_get_set(self): clock = MockClock() cache = ExpiringCache("test", clock, max_len=1) diff --git a/tests/util/test_lrucache.py b/tests/util/test_lrucache.py index 786947375d..0adb2174af 100644 --- a/tests/util/test_lrucache.py +++ b/tests/util/test_lrucache.py @@ -22,7 +22,7 @@ from synapse.util.caches.treecache import TreeCache from .. import unittest -class LruCacheTestCase(unittest.TestCase): +class LruCacheTestCase(unittest.HomeserverTestCase): def test_get_set(self): cache = LruCache(1) cache["key"] = "value" @@ -84,7 +84,7 @@ class LruCacheTestCase(unittest.TestCase): self.assertEquals(len(cache), 0) -class LruCacheCallbacksTestCase(unittest.TestCase): +class LruCacheCallbacksTestCase(unittest.HomeserverTestCase): def test_get(self): m = Mock() cache = LruCache(1) @@ -233,7 +233,7 @@ class LruCacheCallbacksTestCase(unittest.TestCase): self.assertEquals(m3.call_count, 1) -class LruCacheSizedTestCase(unittest.TestCase): +class LruCacheSizedTestCase(unittest.HomeserverTestCase): def test_evict(self): cache = LruCache(5, size_callback=len) cache["key1"] = [0] diff --git a/tests/util/test_stream_change_cache.py b/tests/util/test_stream_change_cache.py index 6857933540..13b753e367 100644 --- a/tests/util/test_stream_change_cache.py +++ b/tests/util/test_stream_change_cache.py @@ -1,11 +1,9 @@ -from mock import patch - from synapse.util.caches.stream_change_cache import StreamChangeCache from tests import unittest -class StreamChangeCacheTests(unittest.TestCase): +class StreamChangeCacheTests(unittest.HomeserverTestCase): """ Tests for StreamChangeCache. """ @@ -54,7 +52,6 @@ class StreamChangeCacheTests(unittest.TestCase): self.assertTrue(cache.has_entity_changed("user@foo.com", 0)) self.assertTrue(cache.has_entity_changed("not@here.website", 0)) - @patch("synapse.util.caches.CACHE_SIZE_FACTOR", 1.0) def test_entity_has_changed_pops_off_start(self): """ StreamChangeCache.entity_has_changed will respect the max size and diff --git a/tests/utils.py b/tests/utils.py index f9be62b499..59c020a051 100644 --- a/tests/utils.py +++ b/tests/utils.py @@ -167,6 +167,7 @@ def default_config(name, parse=False): # disable user directory updates, because they get done in the # background, which upsets the test runner. "update_user_directory": False, + "caches": {"global_factor": 1}, } if parse: -- cgit 1.4.1 From 1a1da60ad2c9172fe487cd38a164b39df60f4cb5 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Tue, 12 May 2020 11:20:48 +0100 Subject: Fix new flake8 errors (#7470) --- changelog.d/7470.misc | 1 + synapse/app/_base.py | 5 +++-- synapse/config/server.py | 2 +- synapse/notifier.py | 10 ++++++---- synapse/push/mailer.py | 7 +++++-- synapse/storage/database.py | 4 ++-- tests/config/test_load.py | 2 +- 7 files changed, 19 insertions(+), 12 deletions(-) create mode 100644 changelog.d/7470.misc (limited to 'tests') diff --git a/changelog.d/7470.misc b/changelog.d/7470.misc new file mode 100644 index 0000000000..45e66ecf48 --- /dev/null +++ b/changelog.d/7470.misc @@ -0,0 +1 @@ +Fix linting errors in new version of Flake8. diff --git a/synapse/app/_base.py b/synapse/app/_base.py index 628292b890..dedff81af3 100644 --- a/synapse/app/_base.py +++ b/synapse/app/_base.py @@ -22,6 +22,7 @@ import sys import traceback from daemonize import Daemonize +from typing_extensions import NoReturn from twisted.internet import defer, error, reactor from twisted.protocols.tls import TLSMemoryBIOFactory @@ -139,9 +140,9 @@ def start_reactor( run() -def quit_with_error(error_string): +def quit_with_error(error_string: str) -> NoReturn: message_lines = error_string.split("\n") - line_length = max(len(l) for l in message_lines if len(l) < 80) + 2 + line_length = max(len(line) for line in message_lines if len(line) < 80) + 2 sys.stderr.write("*" * line_length + "\n") for line in message_lines: sys.stderr.write(" %s\n" % (line.rstrip(),)) diff --git a/synapse/config/server.py b/synapse/config/server.py index 6d88231843..ed28da3deb 100644 --- a/synapse/config/server.py +++ b/synapse/config/server.py @@ -522,7 +522,7 @@ class ServerConfig(Config): ) def has_tls_listener(self) -> bool: - return any(l["tls"] for l in self.listeners) + return any(listener["tls"] for listener in self.listeners) def generate_config_section( self, server_name, data_dir_path, open_private_ports, listeners, **kwargs diff --git a/synapse/notifier.py b/synapse/notifier.py index 71d9ed62b0..87c120a59c 100644 --- a/synapse/notifier.py +++ b/synapse/notifier.py @@ -15,7 +15,7 @@ import logging from collections import namedtuple -from typing import Callable, List +from typing import Callable, Iterable, List, TypeVar from prometheus_client import Counter @@ -42,12 +42,14 @@ users_woken_by_stream_counter = Counter( "synapse_notifier_users_woken_by_stream", "", ["stream"] ) +T = TypeVar("T") + # TODO(paul): Should be shared somewhere -def count(func, l): - """Return the number of items in l for which func returns true.""" +def count(func: Callable[[T], bool], it: Iterable[T]) -> int: + """Return the number of items in it for which func returns true.""" n = 0 - for x in l: + for x in it: if func(x): n += 1 return n diff --git a/synapse/push/mailer.py b/synapse/push/mailer.py index 73580c1c6c..ab33abbeed 100644 --- a/synapse/push/mailer.py +++ b/synapse/push/mailer.py @@ -19,6 +19,7 @@ import logging import time from email.mime.multipart import MIMEMultipart from email.mime.text import MIMEText +from typing import Iterable, List, TypeVar from six.moves import urllib @@ -41,6 +42,8 @@ from synapse.visibility import filter_events_for_client logger = logging.getLogger(__name__) +T = TypeVar("T") + MESSAGE_FROM_PERSON_IN_ROOM = ( "You have a message on %(app)s from %(person)s in the %(room)s room..." @@ -638,10 +641,10 @@ def safe_text(raw_text): ) -def deduped_ordered_list(l): +def deduped_ordered_list(it: Iterable[T]) -> List[T]: seen = set() ret = [] - for item in l: + for item in it: if item not in seen: seen.add(item) ret.append(item) diff --git a/synapse/storage/database.py b/synapse/storage/database.py index 2b635d6ca0..c3d0863429 100644 --- a/synapse/storage/database.py +++ b/synapse/storage/database.py @@ -214,9 +214,9 @@ class LoggingTransaction: def executemany(self, sql: str, *args: Any): self._do_execute(self.txn.executemany, sql, *args) - def _make_sql_one_line(self, sql): + def _make_sql_one_line(self, sql: str) -> str: "Strip newlines out of SQL so that the loggers in the DB are on one line" - return " ".join(l.strip() for l in sql.splitlines() if l.strip()) + return " ".join(line.strip() for line in sql.splitlines() if line.strip()) def _do_execute(self, func, sql, *args): sql = self._make_sql_one_line(sql) diff --git a/tests/config/test_load.py b/tests/config/test_load.py index b3e557bd6a..734a9983e8 100644 --- a/tests/config/test_load.py +++ b/tests/config/test_load.py @@ -122,7 +122,7 @@ class ConfigLoadingTestCase(unittest.TestCase): with open(self.file, "r") as f: contents = f.readlines() - contents = [l for l in contents if needle not in l] + contents = [line for line in contents if needle not in line] with open(self.file, "w") as f: f.write("".join(contents)) -- cgit 1.4.1 From edd3b0747cc651d224fc1bf81ae7fbd6a25a2ea5 Mon Sep 17 00:00:00 2001 From: Patrick Cloke Date: Wed, 13 May 2020 08:24:50 -0400 Subject: Fix new flake8 errors (#7489) This is a cherry-pick of 1a1da60ad2c9172fe487cd38a164b39df60f4cb5 (#7470) to the release-v1.13.0 branch. --- changelog.d/7470.misc | 1 + synapse/app/_base.py | 5 +++-- synapse/config/server.py | 2 +- synapse/notifier.py | 10 ++++++---- synapse/push/mailer.py | 7 +++++-- synapse/storage/database.py | 4 ++-- tests/config/test_load.py | 2 +- 7 files changed, 19 insertions(+), 12 deletions(-) create mode 100644 changelog.d/7470.misc (limited to 'tests') diff --git a/changelog.d/7470.misc b/changelog.d/7470.misc new file mode 100644 index 0000000000..45e66ecf48 --- /dev/null +++ b/changelog.d/7470.misc @@ -0,0 +1 @@ +Fix linting errors in new version of Flake8. diff --git a/synapse/app/_base.py b/synapse/app/_base.py index 628292b890..dedff81af3 100644 --- a/synapse/app/_base.py +++ b/synapse/app/_base.py @@ -22,6 +22,7 @@ import sys import traceback from daemonize import Daemonize +from typing_extensions import NoReturn from twisted.internet import defer, error, reactor from twisted.protocols.tls import TLSMemoryBIOFactory @@ -139,9 +140,9 @@ def start_reactor( run() -def quit_with_error(error_string): +def quit_with_error(error_string: str) -> NoReturn: message_lines = error_string.split("\n") - line_length = max(len(l) for l in message_lines if len(l) < 80) + 2 + line_length = max(len(line) for line in message_lines if len(line) < 80) + 2 sys.stderr.write("*" * line_length + "\n") for line in message_lines: sys.stderr.write(" %s\n" % (line.rstrip(),)) diff --git a/synapse/config/server.py b/synapse/config/server.py index 6d88231843..ed28da3deb 100644 --- a/synapse/config/server.py +++ b/synapse/config/server.py @@ -522,7 +522,7 @@ class ServerConfig(Config): ) def has_tls_listener(self) -> bool: - return any(l["tls"] for l in self.listeners) + return any(listener["tls"] for listener in self.listeners) def generate_config_section( self, server_name, data_dir_path, open_private_ports, listeners, **kwargs diff --git a/synapse/notifier.py b/synapse/notifier.py index 71d9ed62b0..87c120a59c 100644 --- a/synapse/notifier.py +++ b/synapse/notifier.py @@ -15,7 +15,7 @@ import logging from collections import namedtuple -from typing import Callable, List +from typing import Callable, Iterable, List, TypeVar from prometheus_client import Counter @@ -42,12 +42,14 @@ users_woken_by_stream_counter = Counter( "synapse_notifier_users_woken_by_stream", "", ["stream"] ) +T = TypeVar("T") + # TODO(paul): Should be shared somewhere -def count(func, l): - """Return the number of items in l for which func returns true.""" +def count(func: Callable[[T], bool], it: Iterable[T]) -> int: + """Return the number of items in it for which func returns true.""" n = 0 - for x in l: + for x in it: if func(x): n += 1 return n diff --git a/synapse/push/mailer.py b/synapse/push/mailer.py index 73580c1c6c..ab33abbeed 100644 --- a/synapse/push/mailer.py +++ b/synapse/push/mailer.py @@ -19,6 +19,7 @@ import logging import time from email.mime.multipart import MIMEMultipart from email.mime.text import MIMEText +from typing import Iterable, List, TypeVar from six.moves import urllib @@ -41,6 +42,8 @@ from synapse.visibility import filter_events_for_client logger = logging.getLogger(__name__) +T = TypeVar("T") + MESSAGE_FROM_PERSON_IN_ROOM = ( "You have a message on %(app)s from %(person)s in the %(room)s room..." @@ -638,10 +641,10 @@ def safe_text(raw_text): ) -def deduped_ordered_list(l): +def deduped_ordered_list(it: Iterable[T]) -> List[T]: seen = set() ret = [] - for item in l: + for item in it: if item not in seen: seen.add(item) ret.append(item) diff --git a/synapse/storage/database.py b/synapse/storage/database.py index a7cd97b0b0..50f475bfd3 100644 --- a/synapse/storage/database.py +++ b/synapse/storage/database.py @@ -212,9 +212,9 @@ class LoggingTransaction: def executemany(self, sql: str, *args: Any): self._do_execute(self.txn.executemany, sql, *args) - def _make_sql_one_line(self, sql): + def _make_sql_one_line(self, sql: str) -> str: "Strip newlines out of SQL so that the loggers in the DB are on one line" - return " ".join(l.strip() for l in sql.splitlines() if l.strip()) + return " ".join(line.strip() for line in sql.splitlines() if line.strip()) def _do_execute(self, func, sql, *args): sql = self._make_sql_one_line(sql) diff --git a/tests/config/test_load.py b/tests/config/test_load.py index b3e557bd6a..734a9983e8 100644 --- a/tests/config/test_load.py +++ b/tests/config/test_load.py @@ -122,7 +122,7 @@ class ConfigLoadingTestCase(unittest.TestCase): with open(self.file, "r") as f: contents = f.readlines() - contents = [l for l in contents if needle not in l] + contents = [line for line in contents if needle not in line] with open(self.file, "w") as f: f.write("".join(contents)) -- cgit 1.4.1 From 782e4e64df58dd9e114e7e1e467ae55219336b79 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 13 May 2020 13:38:22 +0100 Subject: Shuffle persist event data store functions. (#7440) The aim here is to get to a stage where we have a `PersistEventStore` that holds all the write methods used during event persistence, so that we can take that class out of the `DataStore` mixin and instansiate it separately. This will allow us to instansiate it on processes other than master, while also ensuring it is only available on processes that are configured to write to events stream. This is a bit of an architectural change, where we end up with multiple classes per data store (rather than one per data store we have now). We end up having: 1. Storage classes that provide high level APIs that can talk to multiple data stores. 2. Data store modules that consist of classes that must point at the same database instance. 3. Classes in a data store that can be instantiated on processes depending on config. --- changelog.d/7440.misc | 1 + synapse/storage/data_stores/__init__.py | 9 + synapse/storage/data_stores/main/__init__.py | 8 +- synapse/storage/data_stores/main/censor_events.py | 213 ++++ .../storage/data_stores/main/event_federation.py | 83 -- .../storage/data_stores/main/event_push_actions.py | 74 -- synapse/storage/data_stores/main/events.py | 1216 ++++++++------------ synapse/storage/data_stores/main/events_worker.py | 152 ++- synapse/storage/data_stores/main/metrics.py | 128 +++ synapse/storage/data_stores/main/purge_events.py | 399 +++++++ synapse/storage/data_stores/main/rejections.py | 11 - synapse/storage/data_stores/main/relations.py | 60 +- synapse/storage/data_stores/main/room.py | 49 - synapse/storage/data_stores/main/roommember.py | 90 -- synapse/storage/data_stores/main/search.py | 23 - synapse/storage/data_stores/main/signatures.py | 30 - synapse/storage/data_stores/main/state.py | 35 - synapse/storage/persist_events.py | 27 +- tests/storage/test_event_push_actions.py | 3 +- 19 files changed, 1376 insertions(+), 1235 deletions(-) create mode 100644 changelog.d/7440.misc create mode 100644 synapse/storage/data_stores/main/censor_events.py create mode 100644 synapse/storage/data_stores/main/metrics.py create mode 100644 synapse/storage/data_stores/main/purge_events.py (limited to 'tests') diff --git a/changelog.d/7440.misc b/changelog.d/7440.misc new file mode 100644 index 0000000000..2505cdc66f --- /dev/null +++ b/changelog.d/7440.misc @@ -0,0 +1 @@ +Refactor event persistence database functions in preparation for allowing them to be run on non-master processes. diff --git a/synapse/storage/data_stores/__init__.py b/synapse/storage/data_stores/__init__.py index e1d03429ca..7c14ba91c2 100644 --- a/synapse/storage/data_stores/__init__.py +++ b/synapse/storage/data_stores/__init__.py @@ -15,6 +15,7 @@ import logging +from synapse.storage.data_stores.main.events import PersistEventsStore from synapse.storage.data_stores.state import StateGroupDataStore from synapse.storage.database import Database, make_conn from synapse.storage.engines import create_engine @@ -39,6 +40,7 @@ class DataStores(object): self.databases = [] self.main = None self.state = None + self.persist_events = None for database_config in hs.config.database.databases: db_name = database_config.name @@ -64,6 +66,13 @@ class DataStores(object): self.main = main_store_class(database, db_conn, hs) + # If we're on a process that can persist events (currently + # master), also instansiate a `PersistEventsStore` + if hs.config.worker.worker_app is None: + self.persist_events = PersistEventsStore( + hs, database, self.main + ) + if "state" in database_config.data_stores: logger.info("Starting 'state' data store") diff --git a/synapse/storage/data_stores/main/__init__.py b/synapse/storage/data_stores/main/__init__.py index cd2a1f0461..5df9dce79d 100644 --- a/synapse/storage/data_stores/main/__init__.py +++ b/synapse/storage/data_stores/main/__init__.py @@ -34,6 +34,7 @@ from synapse.util.caches.stream_change_cache import StreamChangeCache from .account_data import AccountDataStore from .appservice import ApplicationServiceStore, ApplicationServiceTransactionStore from .cache import CacheInvalidationWorkerStore +from .censor_events import CensorEventsStore from .client_ips import ClientIpStore from .deviceinbox import DeviceInboxStore from .devices import DeviceStore @@ -42,16 +43,17 @@ from .e2e_room_keys import EndToEndRoomKeyStore from .end_to_end_keys import EndToEndKeyStore from .event_federation import EventFederationStore from .event_push_actions import EventPushActionsStore -from .events import EventsStore from .events_bg_updates import EventsBackgroundUpdatesStore from .filtering import FilteringStore from .group_server import GroupServerStore from .keys import KeyStore from .media_repository import MediaRepositoryStore +from .metrics import ServerMetricsStore from .monthly_active_users import MonthlyActiveUsersStore from .openid import OpenIdStore from .presence import PresenceStore, UserPresenceState from .profile import ProfileStore +from .purge_events import PurgeEventsStore from .push_rule import PushRuleStore from .pusher import PusherStore from .receipts import ReceiptsStore @@ -88,7 +90,7 @@ class DataStore( StateStore, SignatureStore, ApplicationServiceStore, - EventsStore, + PurgeEventsStore, EventFederationStore, MediaRepositoryStore, RejectionsStore, @@ -113,8 +115,10 @@ class DataStore( MonthlyActiveUsersStore, StatsStore, RelationsStore, + CensorEventsStore, UIAuthStore, CacheInvalidationWorkerStore, + ServerMetricsStore, ): def __init__(self, database: Database, db_conn, hs): self.hs = hs diff --git a/synapse/storage/data_stores/main/censor_events.py b/synapse/storage/data_stores/main/censor_events.py new file mode 100644 index 0000000000..49683b4d5a --- /dev/null +++ b/synapse/storage/data_stores/main/censor_events.py @@ -0,0 +1,213 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging +from typing import TYPE_CHECKING + +from twisted.internet import defer + +from synapse.events.utils import prune_event_dict +from synapse.metrics.background_process_metrics import run_as_background_process +from synapse.storage._base import SQLBaseStore +from synapse.storage.data_stores.main.cache import CacheInvalidationWorkerStore +from synapse.storage.data_stores.main.events import encode_json +from synapse.storage.data_stores.main.events_worker import EventsWorkerStore +from synapse.storage.database import Database + +if TYPE_CHECKING: + from synapse.server import HomeServer + + +logger = logging.getLogger(__name__) + + +class CensorEventsStore(CacheInvalidationWorkerStore, EventsWorkerStore, SQLBaseStore): + def __init__(self, database: Database, db_conn, hs: "HomeServer"): + super().__init__(database, db_conn, hs) + + # This should only exist on master for now + assert ( + hs.config.worker.worker_app is None + ), "Can only instantiate CensorEventsStore on master" + + def _censor_redactions(): + return run_as_background_process( + "_censor_redactions", self._censor_redactions + ) + + if self.hs.config.redaction_retention_period is not None: + hs.get_clock().looping_call(_censor_redactions, 5 * 60 * 1000) + + async def _censor_redactions(self): + """Censors all redactions older than the configured period that haven't + been censored yet. + + By censor we mean update the event_json table with the redacted event. + """ + + if self.hs.config.redaction_retention_period is None: + return + + if not ( + await self.db.updates.has_completed_background_update( + "redactions_have_censored_ts_idx" + ) + ): + # We don't want to run this until the appropriate index has been + # created. + return + + before_ts = self._clock.time_msec() - self.hs.config.redaction_retention_period + + # We fetch all redactions that: + # 1. point to an event we have, + # 2. has a received_ts from before the cut off, and + # 3. we haven't yet censored. + # + # This is limited to 100 events to ensure that we don't try and do too + # much at once. We'll get called again so this should eventually catch + # up. + sql = """ + SELECT redactions.event_id, redacts FROM redactions + LEFT JOIN events AS original_event ON ( + redacts = original_event.event_id + ) + WHERE NOT have_censored + AND redactions.received_ts <= ? + ORDER BY redactions.received_ts ASC + LIMIT ? + """ + + rows = await self.db.execute( + "_censor_redactions_fetch", None, sql, before_ts, 100 + ) + + updates = [] + + for redaction_id, event_id in rows: + redaction_event = await self.get_event(redaction_id, allow_none=True) + original_event = await self.get_event( + event_id, allow_rejected=True, allow_none=True + ) + + # The SQL above ensures that we have both the redaction and + # original event, so if the `get_event` calls return None it + # means that the redaction wasn't allowed. Either way we know that + # the result won't change so we mark the fact that we've checked. + if ( + redaction_event + and original_event + and original_event.internal_metadata.is_redacted() + ): + # Redaction was allowed + pruned_json = encode_json( + prune_event_dict( + original_event.room_version, original_event.get_dict() + ) + ) + else: + # Redaction wasn't allowed + pruned_json = None + + updates.append((redaction_id, event_id, pruned_json)) + + def _update_censor_txn(txn): + for redaction_id, event_id, pruned_json in updates: + if pruned_json: + self._censor_event_txn(txn, event_id, pruned_json) + + self.db.simple_update_one_txn( + txn, + table="redactions", + keyvalues={"event_id": redaction_id}, + updatevalues={"have_censored": True}, + ) + + await self.db.runInteraction("_update_censor_txn", _update_censor_txn) + + def _censor_event_txn(self, txn, event_id, pruned_json): + """Censor an event by replacing its JSON in the event_json table with the + provided pruned JSON. + + Args: + txn (LoggingTransaction): The database transaction. + event_id (str): The ID of the event to censor. + pruned_json (str): The pruned JSON + """ + self.db.simple_update_one_txn( + txn, + table="event_json", + keyvalues={"event_id": event_id}, + updatevalues={"json": pruned_json}, + ) + + @defer.inlineCallbacks + def expire_event(self, event_id): + """Retrieve and expire an event that has expired, and delete its associated + expiry timestamp. If the event can't be retrieved, delete its associated + timestamp so we don't try to expire it again in the future. + + Args: + event_id (str): The ID of the event to delete. + """ + # Try to retrieve the event's content from the database or the event cache. + event = yield self.get_event(event_id) + + def delete_expired_event_txn(txn): + # Delete the expiry timestamp associated with this event from the database. + self._delete_event_expiry_txn(txn, event_id) + + if not event: + # If we can't find the event, log a warning and delete the expiry date + # from the database so that we don't try to expire it again in the + # future. + logger.warning( + "Can't expire event %s because we don't have it.", event_id + ) + return + + # Prune the event's dict then convert it to JSON. + pruned_json = encode_json( + prune_event_dict(event.room_version, event.get_dict()) + ) + + # Update the event_json table to replace the event's JSON with the pruned + # JSON. + self._censor_event_txn(txn, event.event_id, pruned_json) + + # We need to invalidate the event cache entry for this event because we + # changed its content in the database. We can't call + # self._invalidate_cache_and_stream because self.get_event_cache isn't of the + # right type. + txn.call_after(self._get_event_cache.invalidate, (event.event_id,)) + # Send that invalidation to replication so that other workers also invalidate + # the event cache. + self._send_invalidation_to_replication( + txn, "_get_event_cache", (event.event_id,) + ) + + yield self.db.runInteraction("delete_expired_event", delete_expired_event_txn) + + def _delete_event_expiry_txn(self, txn, event_id): + """Delete the expiry timestamp associated with an event ID without deleting the + actual event. + + Args: + txn (LoggingTransaction): The transaction to use to perform the deletion. + event_id (str): The event ID to delete the associated expiry timestamp of. + """ + return self.db.simple_delete_txn( + txn=txn, table="event_expiry", keyvalues={"event_id": event_id} + ) diff --git a/synapse/storage/data_stores/main/event_federation.py b/synapse/storage/data_stores/main/event_federation.py index b99439cc37..24ce8c4330 100644 --- a/synapse/storage/data_stores/main/event_federation.py +++ b/synapse/storage/data_stores/main/event_federation.py @@ -640,89 +640,6 @@ class EventFederationStore(EventFederationWorkerStore): self._delete_old_forward_extrem_cache, 60 * 60 * 1000 ) - def _update_min_depth_for_room_txn(self, txn, room_id, depth): - min_depth = self._get_min_depth_interaction(txn, room_id) - - if min_depth is not None and depth >= min_depth: - return - - self.db.simple_upsert_txn( - txn, - table="room_depth", - keyvalues={"room_id": room_id}, - values={"min_depth": depth}, - ) - - def _handle_mult_prev_events(self, txn, events): - """ - For the given event, update the event edges table and forward and - backward extremities tables. - """ - self.db.simple_insert_many_txn( - txn, - table="event_edges", - values=[ - { - "event_id": ev.event_id, - "prev_event_id": e_id, - "room_id": ev.room_id, - "is_state": False, - } - for ev in events - for e_id in ev.prev_event_ids() - ], - ) - - self._update_backward_extremeties(txn, events) - - def _update_backward_extremeties(self, txn, events): - """Updates the event_backward_extremities tables based on the new/updated - events being persisted. - - This is called for new events *and* for events that were outliers, but - are now being persisted as non-outliers. - - Forward extremities are handled when we first start persisting the events. - """ - events_by_room = {} - for ev in events: - events_by_room.setdefault(ev.room_id, []).append(ev) - - query = ( - "INSERT INTO event_backward_extremities (event_id, room_id)" - " SELECT ?, ? WHERE NOT EXISTS (" - " SELECT 1 FROM event_backward_extremities" - " WHERE event_id = ? AND room_id = ?" - " )" - " AND NOT EXISTS (" - " SELECT 1 FROM events WHERE event_id = ? AND room_id = ? " - " AND outlier = ?" - " )" - ) - - txn.executemany( - query, - [ - (e_id, ev.room_id, e_id, ev.room_id, e_id, ev.room_id, False) - for ev in events - for e_id in ev.prev_event_ids() - if not ev.internal_metadata.is_outlier() - ], - ) - - query = ( - "DELETE FROM event_backward_extremities" - " WHERE event_id = ? AND room_id = ?" - ) - txn.executemany( - query, - [ - (ev.event_id, ev.room_id) - for ev in events - if not ev.internal_metadata.is_outlier() - ], - ) - def _delete_old_forward_extrem_cache(self): def _delete_old_forward_extrem_cache_txn(txn): # Delete entries older than a month, while making sure we don't delete diff --git a/synapse/storage/data_stores/main/event_push_actions.py b/synapse/storage/data_stores/main/event_push_actions.py index 8eed590929..0321274de2 100644 --- a/synapse/storage/data_stores/main/event_push_actions.py +++ b/synapse/storage/data_stores/main/event_push_actions.py @@ -652,69 +652,6 @@ class EventPushActionsStore(EventPushActionsWorkerStore): self._start_rotate_notifs, 30 * 60 * 1000 ) - def _set_push_actions_for_event_and_users_txn( - self, txn, events_and_contexts, all_events_and_contexts - ): - """Handles moving push actions from staging table to main - event_push_actions table for all events in `events_and_contexts`. - - Also ensures that all events in `all_events_and_contexts` are removed - from the push action staging area. - - Args: - events_and_contexts (list[(EventBase, EventContext)]): events - we are persisting - all_events_and_contexts (list[(EventBase, EventContext)]): all - events that we were going to persist. This includes events - we've already persisted, etc, that wouldn't appear in - events_and_context. - """ - - sql = """ - INSERT INTO event_push_actions ( - room_id, event_id, user_id, actions, stream_ordering, - topological_ordering, notif, highlight - ) - SELECT ?, event_id, user_id, actions, ?, ?, notif, highlight - FROM event_push_actions_staging - WHERE event_id = ? - """ - - if events_and_contexts: - txn.executemany( - sql, - ( - ( - event.room_id, - event.internal_metadata.stream_ordering, - event.depth, - event.event_id, - ) - for event, _ in events_and_contexts - ), - ) - - for event, _ in events_and_contexts: - user_ids = self.db.simple_select_onecol_txn( - txn, - table="event_push_actions_staging", - keyvalues={"event_id": event.event_id}, - retcol="user_id", - ) - - for uid in user_ids: - txn.call_after( - self.get_unread_event_push_actions_by_room_for_user.invalidate_many, - (event.room_id, uid), - ) - - # Now we delete the staging area for *all* events that were being - # persisted. - txn.executemany( - "DELETE FROM event_push_actions_staging WHERE event_id = ?", - ((event.event_id,) for event, _ in all_events_and_contexts), - ) - @defer.inlineCallbacks def get_push_actions_for_user( self, user_id, before=None, limit=50, only_highlight=False @@ -763,17 +700,6 @@ class EventPushActionsStore(EventPushActionsWorkerStore): ) return result[0] or 0 - def _remove_push_actions_for_event_id_txn(self, txn, room_id, event_id): - # Sad that we have to blow away the cache for the whole room here - txn.call_after( - self.get_unread_event_push_actions_by_room_for_user.invalidate_many, - (room_id,), - ) - txn.execute( - "DELETE FROM event_push_actions WHERE room_id = ? AND event_id = ?", - (room_id, event_id), - ) - def _remove_old_push_actions_before_txn( self, txn, room_id, user_id, stream_ordering ): diff --git a/synapse/storage/data_stores/main/events.py b/synapse/storage/data_stores/main/events.py index e71c23541d..092739962b 100644 --- a/synapse/storage/data_stores/main/events.py +++ b/synapse/storage/data_stores/main/events.py @@ -17,39 +17,44 @@ import itertools import logging -from collections import Counter as c_counter, OrderedDict, namedtuple +from collections import OrderedDict, namedtuple from functools import wraps -from typing import Dict, List, Tuple +from typing import TYPE_CHECKING, Dict, Iterable, List, Tuple -from six import iteritems, text_type +from six import integer_types, iteritems, text_type from six.moves import range +import attr from canonicaljson import json from prometheus_client import Counter from twisted.internet import defer import synapse.metrics -from synapse.api.constants import EventContentFields, EventTypes -from synapse.api.errors import SynapseError +from synapse.api.constants import ( + EventContentFields, + EventTypes, + Membership, + RelationTypes, +) from synapse.api.room_versions import RoomVersions +from synapse.crypto.event_signing import compute_event_reference_hash from synapse.events import EventBase # noqa: F401 from synapse.events.snapshot import EventContext # noqa: F401 -from synapse.events.utils import prune_event_dict from synapse.logging.utils import log_function -from synapse.metrics import BucketCollector -from synapse.metrics.background_process_metrics import run_as_background_process from synapse.storage._base import make_in_list_sql_clause -from synapse.storage.data_stores.main.event_federation import EventFederationStore -from synapse.storage.data_stores.main.events_worker import EventsWorkerStore -from synapse.storage.data_stores.main.state import StateGroupWorkerStore +from synapse.storage.data_stores.main.search import SearchEntry from synapse.storage.database import Database, LoggingTransaction -from synapse.storage.persist_events import DeltaState -from synapse.types import RoomStreamToken, StateMap, get_domain_from_id -from synapse.util.caches.descriptors import cached, cachedInlineCallbacks +from synapse.storage.util.id_generators import StreamIdGenerator +from synapse.types import StateMap, get_domain_from_id from synapse.util.frozenutils import frozendict_json_encoder from synapse.util.iterutils import batch_iter +if TYPE_CHECKING: + from synapse.storage.data_stores.main import DataStore + from synapse.server import HomeServer + + logger = logging.getLogger(__name__) persist_event_counter = Counter("synapse_storage_events_persisted_events", "") @@ -94,58 +99,49 @@ def _retry_on_integrity_error(func): return f -# inherits from EventFederationStore so that we can call _update_backward_extremities -# and _handle_mult_prev_events (though arguably those could both be moved in here) -class EventsStore( - StateGroupWorkerStore, EventFederationStore, EventsWorkerStore, -): - def __init__(self, database: Database, db_conn, hs): - super(EventsStore, self).__init__(database, db_conn, hs) +@attr.s(slots=True) +class DeltaState: + """Deltas to use to update the `current_state_events` table. - # Collect metrics on the number of forward extremities that exist. - # Counter of number of extremities to count - self._current_forward_extremities_amount = c_counter() + Attributes: + to_delete: List of type/state_keys to delete from current state + to_insert: Map of state to upsert into current state + no_longer_in_room: The server is not longer in the room, so the room + should e.g. be removed from `current_state_events` table. + """ - BucketCollector( - "synapse_forward_extremities", - lambda: self._current_forward_extremities_amount, - buckets=[1, 2, 3, 5, 7, 10, 15, 20, 50, 100, 200, 500, "+Inf"], - ) + to_delete = attr.ib(type=List[Tuple[str, str]]) + to_insert = attr.ib(type=StateMap[str]) + no_longer_in_room = attr.ib(type=bool, default=False) - # Read the extrems every 60 minutes - def read_forward_extremities(): - # run as a background process to make sure that the database transactions - # have a logcontext to report to - return run_as_background_process( - "read_forward_extremities", self._read_forward_extremities - ) - hs.get_clock().looping_call(read_forward_extremities, 60 * 60 * 1000) +class PersistEventsStore: + """Contains all the functions for writing events to the database. - def _censor_redactions(): - return run_as_background_process( - "_censor_redactions", self._censor_redactions - ) + Should only be instansiated on one process (when using a worker mode setup). + + Note: This is not part of the `DataStore` mixin. + """ - if self.hs.config.redaction_retention_period is not None: - hs.get_clock().looping_call(_censor_redactions, 5 * 60 * 1000) + def __init__(self, hs: "HomeServer", db: Database, main_data_store: "DataStore"): + self.hs = hs + self.db = db + self.store = main_data_store + self.database_engine = db.engine + self._clock = hs.get_clock() self._ephemeral_messages_enabled = hs.config.enable_ephemeral_messages self.is_mine_id = hs.is_mine_id - @defer.inlineCallbacks - def _read_forward_extremities(self): - def fetch(txn): - txn.execute( - """ - select count(*) c from event_forward_extremities - group by room_id - """ - ) - return txn.fetchall() + # Ideally we'd move these ID gens here, unfortunately some other ID + # generators are chained off them so doing so is a bit of a PITA. + self._backfill_id_gen = self.store._backfill_id_gen # type: StreamIdGenerator + self._stream_id_gen = self.store._stream_id_gen # type: StreamIdGenerator - res = yield self.db.runInteraction("read_forward_extremities", fetch) - self._current_forward_extremities_amount = c_counter([x[0] for x in res]) + # This should only exist on master for now + assert ( + hs.config.worker.worker_app is None + ), "Can only instantiate PersistEventsStore on master" @_retry_on_integrity_error @defer.inlineCallbacks @@ -237,10 +233,10 @@ class EventsStore( event_counter.labels(event.type, origin_type, origin_entity).inc() for room_id, new_state in iteritems(current_state_for_room): - self.get_current_state_ids.prefill((room_id,), new_state) + self.store.get_current_state_ids.prefill((room_id,), new_state) for room_id, latest_event_ids in iteritems(new_forward_extremeties): - self.get_latest_event_ids_in_room.prefill( + self.store.get_latest_event_ids_in_room.prefill( (room_id,), list(latest_event_ids) ) @@ -586,7 +582,7 @@ class EventsStore( ) txn.call_after( - self._curr_state_delta_stream_cache.entity_has_changed, + self.store._curr_state_delta_stream_cache.entity_has_changed, room_id, stream_id, ) @@ -606,10 +602,13 @@ class EventsStore( for member in members_changed: txn.call_after( - self.get_rooms_for_user_with_stream_ordering.invalidate, (member,) + self.store.get_rooms_for_user_with_stream_ordering.invalidate, + (member,), ) - self._invalidate_state_caches_and_stream(txn, room_id, members_changed) + self.store._invalidate_state_caches_and_stream( + txn, room_id, members_changed + ) def _upsert_room_version_txn(self, txn: LoggingTransaction, room_id: str): """Update the room version in the database based off current state @@ -647,7 +646,9 @@ class EventsStore( self.db.simple_delete_txn( txn, table="event_forward_extremities", keyvalues={"room_id": room_id} ) - txn.call_after(self.get_latest_event_ids_in_room.invalidate, (room_id,)) + txn.call_after( + self.store.get_latest_event_ids_in_room.invalidate, (room_id,) + ) self.db.simple_insert_many_txn( txn, @@ -713,10 +714,10 @@ class EventsStore( depth_updates = {} for event, context in events_and_contexts: # Remove the any existing cache entries for the event_ids - txn.call_after(self._invalidate_get_event_cache, event.event_id) + txn.call_after(self.store._invalidate_get_event_cache, event.event_id) if not backfilled: txn.call_after( - self._events_stream_cache.entity_has_changed, + self.store._events_stream_cache.entity_has_changed, event.room_id, event.internal_metadata.stream_ordering, ) @@ -1088,13 +1089,15 @@ class EventsStore( def prefill(): for cache_entry in to_prefill: - self._get_event_cache.prefill((cache_entry[0].event_id,), cache_entry) + self.store._get_event_cache.prefill( + (cache_entry[0].event_id,), cache_entry + ) txn.call_after(prefill) def _store_redaction(self, txn, event): # invalidate the cache for the redacted event - txn.call_after(self._invalidate_get_event_cache, event.redacts) + txn.call_after(self.store._invalidate_get_event_cache, event.redacts) self.db.simple_insert_txn( txn, @@ -1106,783 +1109,484 @@ class EventsStore( }, ) - async def _censor_redactions(self): - """Censors all redactions older than the configured period that haven't - been censored yet. + def insert_labels_for_event_txn( + self, txn, event_id, labels, room_id, topological_ordering + ): + """Store the mapping between an event's ID and its labels, with one row per + (event_id, label) tuple. - By censor we mean update the event_json table with the redacted event. + Args: + txn (LoggingTransaction): The transaction to execute. + event_id (str): The event's ID. + labels (list[str]): A list of text labels. + room_id (str): The ID of the room the event was sent to. + topological_ordering (int): The position of the event in the room's topology. """ + return self.db.simple_insert_many_txn( + txn=txn, + table="event_labels", + values=[ + { + "event_id": event_id, + "label": label, + "room_id": room_id, + "topological_ordering": topological_ordering, + } + for label in labels + ], + ) - if self.hs.config.redaction_retention_period is None: - return - - if not ( - await self.db.updates.has_completed_background_update( - "redactions_have_censored_ts_idx" - ) - ): - # We don't want to run this until the appropriate index has been - # created. - return - - before_ts = self._clock.time_msec() - self.hs.config.redaction_retention_period + def _insert_event_expiry_txn(self, txn, event_id, expiry_ts): + """Save the expiry timestamp associated with a given event ID. - # We fetch all redactions that: - # 1. point to an event we have, - # 2. has a received_ts from before the cut off, and - # 3. we haven't yet censored. - # - # This is limited to 100 events to ensure that we don't try and do too - # much at once. We'll get called again so this should eventually catch - # up. - sql = """ - SELECT redactions.event_id, redacts FROM redactions - LEFT JOIN events AS original_event ON ( - redacts = original_event.event_id - ) - WHERE NOT have_censored - AND redactions.received_ts <= ? - ORDER BY redactions.received_ts ASC - LIMIT ? + Args: + txn (LoggingTransaction): The database transaction to use. + event_id (str): The event ID the expiry timestamp is associated with. + expiry_ts (int): The timestamp at which to expire (delete) the event. """ - - rows = await self.db.execute( - "_censor_redactions_fetch", None, sql, before_ts, 100 + return self.db.simple_insert_txn( + txn=txn, + table="event_expiry", + values={"event_id": event_id, "expiry_ts": expiry_ts}, ) - updates = [] + def _store_event_reference_hashes_txn(self, txn, events): + """Store a hash for a PDU + Args: + txn (cursor): + events (list): list of Events. + """ - for redaction_id, event_id in rows: - redaction_event = await self.get_event(redaction_id, allow_none=True) - original_event = await self.get_event( - event_id, allow_rejected=True, allow_none=True + vals = [] + for event in events: + ref_alg, ref_hash_bytes = compute_event_reference_hash(event) + vals.append( + { + "event_id": event.event_id, + "algorithm": ref_alg, + "hash": memoryview(ref_hash_bytes), + } ) - # The SQL above ensures that we have both the redaction and - # original event, so if the `get_event` calls return None it - # means that the redaction wasn't allowed. Either way we know that - # the result won't change so we mark the fact that we've checked. - if ( - redaction_event - and original_event - and original_event.internal_metadata.is_redacted() - ): - # Redaction was allowed - pruned_json = encode_json( - prune_event_dict( - original_event.room_version, original_event.get_dict() - ) - ) - else: - # Redaction wasn't allowed - pruned_json = None - - updates.append((redaction_id, event_id, pruned_json)) - - def _update_censor_txn(txn): - for redaction_id, event_id, pruned_json in updates: - if pruned_json: - self._censor_event_txn(txn, event_id, pruned_json) - - self.db.simple_update_one_txn( - txn, - table="redactions", - keyvalues={"event_id": redaction_id}, - updatevalues={"have_censored": True}, - ) - - await self.db.runInteraction("_update_censor_txn", _update_censor_txn) + self.db.simple_insert_many_txn(txn, table="event_reference_hashes", values=vals) - def _censor_event_txn(self, txn, event_id, pruned_json): - """Censor an event by replacing its JSON in the event_json table with the - provided pruned JSON. - - Args: - txn (LoggingTransaction): The database transaction. - event_id (str): The ID of the event to censor. - pruned_json (str): The pruned JSON + def _store_room_members_txn(self, txn, events, backfilled): + """Store a room member in the database. """ - self.db.simple_update_one_txn( + self.db.simple_insert_many_txn( txn, - table="event_json", - keyvalues={"event_id": event_id}, - updatevalues={"json": pruned_json}, + table="room_memberships", + values=[ + { + "event_id": event.event_id, + "user_id": event.state_key, + "sender": event.user_id, + "room_id": event.room_id, + "membership": event.membership, + "display_name": event.content.get("displayname", None), + "avatar_url": event.content.get("avatar_url", None), + } + for event in events + ], ) - @defer.inlineCallbacks - def count_daily_messages(self): - """ - Returns an estimate of the number of messages sent in the last day. - - If it has been significantly less or more than one day since the last - call to this function, it will return None. - """ - - def _count_messages(txn): - sql = """ - SELECT COALESCE(COUNT(*), 0) FROM events - WHERE type = 'm.room.message' - AND stream_ordering > ? - """ - txn.execute(sql, (self.stream_ordering_day_ago,)) - (count,) = txn.fetchone() - return count - - ret = yield self.db.runInteraction("count_messages", _count_messages) - return ret - - @defer.inlineCallbacks - def count_daily_sent_messages(self): - def _count_messages(txn): - # This is good enough as if you have silly characters in your own - # hostname then thats your own fault. - like_clause = "%:" + self.hs.hostname - - sql = """ - SELECT COALESCE(COUNT(*), 0) FROM events - WHERE type = 'm.room.message' - AND sender LIKE ? - AND stream_ordering > ? - """ - - txn.execute(sql, (like_clause, self.stream_ordering_day_ago)) - (count,) = txn.fetchone() - return count - - ret = yield self.db.runInteraction("count_daily_sent_messages", _count_messages) - return ret - - @defer.inlineCallbacks - def count_daily_active_rooms(self): - def _count(txn): - sql = """ - SELECT COALESCE(COUNT(DISTINCT room_id), 0) FROM events - WHERE type = 'm.room.message' - AND stream_ordering > ? - """ - txn.execute(sql, (self.stream_ordering_day_ago,)) - (count,) = txn.fetchone() - return count - - ret = yield self.db.runInteraction("count_daily_active_rooms", _count) - return ret - - @cached(num_args=5, max_entries=10) - def get_all_new_events( - self, - last_backfill_id, - last_forward_id, - current_backfill_id, - current_forward_id, - limit, - ): - """Get all the new events that have arrived at the server either as - new events or as backfilled events""" - have_backfill_events = last_backfill_id != current_backfill_id - have_forward_events = last_forward_id != current_forward_id - - if not have_backfill_events and not have_forward_events: - return defer.succeed(AllNewEventsResult([], [], [], [], [])) - - def get_all_new_events_txn(txn): - sql = ( - "SELECT e.stream_ordering, e.event_id, e.room_id, e.type," - " state_key, redacts" - " FROM events AS e" - " LEFT JOIN redactions USING (event_id)" - " LEFT JOIN state_events USING (event_id)" - " WHERE ? < stream_ordering AND stream_ordering <= ?" - " ORDER BY stream_ordering ASC" - " LIMIT ?" + for event in events: + txn.call_after( + self.store._membership_stream_cache.entity_has_changed, + event.state_key, + event.internal_metadata.stream_ordering, ) - if have_forward_events: - txn.execute(sql, (last_forward_id, current_forward_id, limit)) - new_forward_events = txn.fetchall() - - if len(new_forward_events) == limit: - upper_bound = new_forward_events[-1][0] - else: - upper_bound = current_forward_id - - sql = ( - "SELECT event_stream_ordering, event_id, state_group" - " FROM ex_outlier_stream" - " WHERE ? > event_stream_ordering" - " AND event_stream_ordering >= ?" - " ORDER BY event_stream_ordering DESC" - ) - txn.execute(sql, (last_forward_id, upper_bound)) - forward_ex_outliers = txn.fetchall() - else: - new_forward_events = [] - forward_ex_outliers = [] - - sql = ( - "SELECT -e.stream_ordering, e.event_id, e.room_id, e.type," - " state_key, redacts" - " FROM events AS e" - " LEFT JOIN redactions USING (event_id)" - " LEFT JOIN state_events USING (event_id)" - " WHERE ? > stream_ordering AND stream_ordering >= ?" - " ORDER BY stream_ordering DESC" - " LIMIT ?" + txn.call_after( + self.store.get_invited_rooms_for_local_user.invalidate, + (event.state_key,), ) - if have_backfill_events: - txn.execute(sql, (-last_backfill_id, -current_backfill_id, limit)) - new_backfill_events = txn.fetchall() - if len(new_backfill_events) == limit: - upper_bound = new_backfill_events[-1][0] - else: - upper_bound = current_backfill_id - - sql = ( - "SELECT -event_stream_ordering, event_id, state_group" - " FROM ex_outlier_stream" - " WHERE ? > event_stream_ordering" - " AND event_stream_ordering >= ?" - " ORDER BY event_stream_ordering DESC" - ) - txn.execute(sql, (-last_backfill_id, -upper_bound)) - backward_ex_outliers = txn.fetchall() - else: - new_backfill_events = [] - backward_ex_outliers = [] - - return AllNewEventsResult( - new_forward_events, - new_backfill_events, - forward_ex_outliers, - backward_ex_outliers, + # We update the local_invites table only if the event is "current", + # i.e., its something that has just happened. If the event is an + # outlier it is only current if its an "out of band membership", + # like a remote invite or a rejection of a remote invite. + is_new_state = not backfilled and ( + not event.internal_metadata.is_outlier() + or event.internal_metadata.is_out_of_band_membership() ) + is_mine = self.is_mine_id(event.state_key) + if is_new_state and is_mine: + if event.membership == Membership.INVITE: + self.db.simple_insert_txn( + txn, + table="local_invites", + values={ + "event_id": event.event_id, + "invitee": event.state_key, + "inviter": event.sender, + "room_id": event.room_id, + "stream_id": event.internal_metadata.stream_ordering, + }, + ) + else: + sql = ( + "UPDATE local_invites SET stream_id = ?, replaced_by = ? WHERE" + " room_id = ? AND invitee = ? AND locally_rejected is NULL" + " AND replaced_by is NULL" + ) - return self.db.runInteraction("get_all_new_events", get_all_new_events_txn) - - def purge_history(self, room_id, token, delete_local_events): - """Deletes room history before a certain point - - Args: - room_id (str): + txn.execute( + sql, + ( + event.internal_metadata.stream_ordering, + event.event_id, + event.room_id, + event.state_key, + ), + ) - token (str): A topological token to delete events before + # We also update the `local_current_membership` table with + # latest invite info. This will usually get updated by the + # `current_state_events` handling, unless its an outlier. + if event.internal_metadata.is_outlier(): + # This should only happen for out of band memberships, so + # we add a paranoia check. + assert event.internal_metadata.is_out_of_band_membership() + + self.db.simple_upsert_txn( + txn, + table="local_current_membership", + keyvalues={ + "room_id": event.room_id, + "user_id": event.state_key, + }, + values={ + "event_id": event.event_id, + "membership": event.membership, + }, + ) - delete_local_events (bool): - if True, we will delete local events as well as remote ones - (instead of just marking them as outliers and deleting their - state groups). + def _handle_event_relations(self, txn, event): + """Handles inserting relation data during peristence of events - Returns: - Deferred[set[int]]: The set of state groups that are referenced by - deleted events. + Args: + txn + event (EventBase) """ + relation = event.content.get("m.relates_to") + if not relation: + # No relations + return - return self.db.runInteraction( - "purge_history", - self._purge_history_txn, - room_id, - token, - delete_local_events, - ) + rel_type = relation.get("rel_type") + if rel_type not in ( + RelationTypes.ANNOTATION, + RelationTypes.REFERENCE, + RelationTypes.REPLACE, + ): + # Unknown relation type + return - def _purge_history_txn(self, txn, room_id, token_str, delete_local_events): - token = RoomStreamToken.parse(token_str) - - # Tables that should be pruned: - # event_auth - # event_backward_extremities - # event_edges - # event_forward_extremities - # event_json - # event_push_actions - # event_reference_hashes - # event_search - # event_to_state_groups - # events - # rejections - # room_depth - # state_groups - # state_groups_state - - # we will build a temporary table listing the events so that we don't - # have to keep shovelling the list back and forth across the - # connection. Annoyingly the python sqlite driver commits the - # transaction on CREATE, so let's do this first. - # - # furthermore, we might already have the table from a previous (failed) - # purge attempt, so let's drop the table first. + parent_id = relation.get("event_id") + if not parent_id: + # Invalid relation + return - txn.execute("DROP TABLE IF EXISTS events_to_purge") + aggregation_key = relation.get("key") - txn.execute( - "CREATE TEMPORARY TABLE events_to_purge (" - " event_id TEXT NOT NULL," - " should_delete BOOLEAN NOT NULL" - ")" + self.db.simple_insert_txn( + txn, + table="event_relations", + values={ + "event_id": event.event_id, + "relates_to_id": parent_id, + "relation_type": rel_type, + "aggregation_key": aggregation_key, + }, ) - # First ensure that we're not about to delete all the forward extremeties - txn.execute( - "SELECT e.event_id, e.depth FROM events as e " - "INNER JOIN event_forward_extremities as f " - "ON e.event_id = f.event_id " - "AND e.room_id = f.room_id " - "WHERE f.room_id = ?", - (room_id,), + txn.call_after(self.store.get_relations_for_event.invalidate_many, (parent_id,)) + txn.call_after( + self.store.get_aggregation_groups_for_event.invalidate_many, (parent_id,) ) - rows = txn.fetchall() - max_depth = max(row[1] for row in rows) - - if max_depth < token.topological: - # We need to ensure we don't delete all the events from the database - # otherwise we wouldn't be able to send any events (due to not - # having any backwards extremeties) - raise SynapseError( - 400, "topological_ordering is greater than forward extremeties" - ) - - logger.info("[purge] looking for events to delete") - - should_delete_expr = "state_key IS NULL" - should_delete_params = () - if not delete_local_events: - should_delete_expr += " AND event_id NOT LIKE ?" - - # We include the parameter twice since we use the expression twice - should_delete_params += ("%:" + self.hs.hostname, "%:" + self.hs.hostname) - should_delete_params += (room_id, token.topological) + if rel_type == RelationTypes.REPLACE: + txn.call_after(self.store.get_applicable_edit.invalidate, (parent_id,)) - # Note that we insert events that are outliers and aren't going to be - # deleted, as nothing will happen to them. - txn.execute( - "INSERT INTO events_to_purge" - " SELECT event_id, %s" - " FROM events AS e LEFT JOIN state_events USING (event_id)" - " WHERE (NOT outlier OR (%s)) AND e.room_id = ? AND topological_ordering < ?" - % (should_delete_expr, should_delete_expr), - should_delete_params, - ) + def _handle_redaction(self, txn, redacted_event_id): + """Handles receiving a redaction and checking whether we need to remove + any redacted relations from the database. - # We create the indices *after* insertion as that's a lot faster. + Args: + txn + redacted_event_id (str): The event that was redacted. + """ - # create an index on should_delete because later we'll be looking for - # the should_delete / shouldn't_delete subsets - txn.execute( - "CREATE INDEX events_to_purge_should_delete" - " ON events_to_purge(should_delete)" + self.db.simple_delete_txn( + txn, table="event_relations", keyvalues={"event_id": redacted_event_id} ) - # We do joins against events_to_purge for e.g. calculating state - # groups to purge, etc., so lets make an index. - txn.execute("CREATE INDEX events_to_purge_id ON events_to_purge(event_id)") - - txn.execute("SELECT event_id, should_delete FROM events_to_purge") - event_rows = txn.fetchall() - logger.info( - "[purge] found %i events before cutoff, of which %i can be deleted", - len(event_rows), - sum(1 for e in event_rows if e[1]), - ) + def _store_room_topic_txn(self, txn, event): + if hasattr(event, "content") and "topic" in event.content: + self.store_event_search_txn( + txn, event, "content.topic", event.content["topic"] + ) - logger.info("[purge] Finding new backward extremities") + def _store_room_name_txn(self, txn, event): + if hasattr(event, "content") and "name" in event.content: + self.store_event_search_txn( + txn, event, "content.name", event.content["name"] + ) - # We calculate the new entries for the backward extremeties by finding - # events to be purged that are pointed to by events we're not going to - # purge. - txn.execute( - "SELECT DISTINCT e.event_id FROM events_to_purge AS e" - " INNER JOIN event_edges AS ed ON e.event_id = ed.prev_event_id" - " LEFT JOIN events_to_purge AS ep2 ON ed.event_id = ep2.event_id" - " WHERE ep2.event_id IS NULL" - ) - new_backwards_extrems = txn.fetchall() + def _store_room_message_txn(self, txn, event): + if hasattr(event, "content") and "body" in event.content: + self.store_event_search_txn( + txn, event, "content.body", event.content["body"] + ) - logger.info("[purge] replacing backward extremities: %r", new_backwards_extrems) + def _store_retention_policy_for_room_txn(self, txn, event): + if hasattr(event, "content") and ( + "min_lifetime" in event.content or "max_lifetime" in event.content + ): + if ( + "min_lifetime" in event.content + and not isinstance(event.content.get("min_lifetime"), integer_types) + ) or ( + "max_lifetime" in event.content + and not isinstance(event.content.get("max_lifetime"), integer_types) + ): + # Ignore the event if one of the value isn't an integer. + return - txn.execute( - "DELETE FROM event_backward_extremities WHERE room_id = ?", (room_id,) - ) + self.db.simple_insert_txn( + txn=txn, + table="room_retention", + values={ + "room_id": event.room_id, + "event_id": event.event_id, + "min_lifetime": event.content.get("min_lifetime"), + "max_lifetime": event.content.get("max_lifetime"), + }, + ) - # Update backward extremeties - txn.executemany( - "INSERT INTO event_backward_extremities (room_id, event_id)" - " VALUES (?, ?)", - [(room_id, event_id) for event_id, in new_backwards_extrems], - ) + self.store._invalidate_cache_and_stream( + txn, self.store.get_retention_policy_for_room, (event.room_id,) + ) - logger.info("[purge] finding state groups referenced by deleted events") + def store_event_search_txn(self, txn, event, key, value): + """Add event to the search table - # Get all state groups that are referenced by events that are to be - # deleted. - txn.execute( - """ - SELECT DISTINCT state_group FROM events_to_purge - INNER JOIN event_to_state_groups USING (event_id) + Args: + txn (cursor): + event (EventBase): + key (str): + value (str): """ + self.store.store_search_entries_txn( + txn, + ( + SearchEntry( + key=key, + value=value, + event_id=event.event_id, + room_id=event.room_id, + stream_ordering=event.internal_metadata.stream_ordering, + origin_server_ts=event.origin_server_ts, + ), + ), ) - referenced_state_groups = {sg for sg, in txn} - logger.info( - "[purge] found %i referenced state groups", len(referenced_state_groups) - ) + def _set_push_actions_for_event_and_users_txn( + self, txn, events_and_contexts, all_events_and_contexts + ): + """Handles moving push actions from staging table to main + event_push_actions table for all events in `events_and_contexts`. - logger.info("[purge] removing events from event_to_state_groups") - txn.execute( - "DELETE FROM event_to_state_groups " - "WHERE event_id IN (SELECT event_id from events_to_purge)" - ) - for event_id, _ in event_rows: - txn.call_after(self._get_state_group_for_event.invalidate, (event_id,)) + Also ensures that all events in `all_events_and_contexts` are removed + from the push action staging area. - # Delete all remote non-state events - for table in ( - "events", - "event_json", - "event_auth", - "event_edges", - "event_forward_extremities", - "event_reference_hashes", - "event_search", - "rejections", - ): - logger.info("[purge] removing events from %s", table) + Args: + events_and_contexts (list[(EventBase, EventContext)]): events + we are persisting + all_events_and_contexts (list[(EventBase, EventContext)]): all + events that we were going to persist. This includes events + we've already persisted, etc, that wouldn't appear in + events_and_context. + """ - txn.execute( - "DELETE FROM %s WHERE event_id IN (" - " SELECT event_id FROM events_to_purge WHERE should_delete" - ")" % (table,) + sql = """ + INSERT INTO event_push_actions ( + room_id, event_id, user_id, actions, stream_ordering, + topological_ordering, notif, highlight ) + SELECT ?, event_id, user_id, actions, ?, ?, notif, highlight + FROM event_push_actions_staging + WHERE event_id = ? + """ - # event_push_actions lacks an index on event_id, and has one on - # (room_id, event_id) instead. - for table in ("event_push_actions",): - logger.info("[purge] removing events from %s", table) + if events_and_contexts: + txn.executemany( + sql, + ( + ( + event.room_id, + event.internal_metadata.stream_ordering, + event.depth, + event.event_id, + ) + for event, _ in events_and_contexts + ), + ) - txn.execute( - "DELETE FROM %s WHERE room_id = ? AND event_id IN (" - " SELECT event_id FROM events_to_purge WHERE should_delete" - ")" % (table,), - (room_id,), + for event, _ in events_and_contexts: + user_ids = self.db.simple_select_onecol_txn( + txn, + table="event_push_actions_staging", + keyvalues={"event_id": event.event_id}, + retcol="user_id", ) - # Mark all state and own events as outliers - logger.info("[purge] marking remaining events as outliers") - txn.execute( - "UPDATE events SET outlier = ?" - " WHERE event_id IN (" - " SELECT event_id FROM events_to_purge " - " WHERE NOT should_delete" - ")", - (True,), + for uid in user_ids: + txn.call_after( + self.store.get_unread_event_push_actions_by_room_for_user.invalidate_many, + (event.room_id, uid), + ) + + # Now we delete the staging area for *all* events that were being + # persisted. + txn.executemany( + "DELETE FROM event_push_actions_staging WHERE event_id = ?", + ((event.event_id,) for event, _ in all_events_and_contexts), ) - # synapse tries to take out an exclusive lock on room_depth whenever it - # persists events (because upsert), and once we run this update, we - # will block that for the rest of our transaction. - # - # So, let's stick it at the end so that we don't block event - # persistence. - # - # We do this by calculating the minimum depth of the backwards - # extremities. However, the events in event_backward_extremities - # are ones we don't have yet so we need to look at the events that - # point to it via event_edges table. - txn.execute( - """ - SELECT COALESCE(MIN(depth), 0) - FROM event_backward_extremities AS eb - INNER JOIN event_edges AS eg ON eg.prev_event_id = eb.event_id - INNER JOIN events AS e ON e.event_id = eg.event_id - WHERE eb.room_id = ? - """, + def _remove_push_actions_for_event_id_txn(self, txn, room_id, event_id): + # Sad that we have to blow away the cache for the whole room here + txn.call_after( + self.store.get_unread_event_push_actions_by_room_for_user.invalidate_many, (room_id,), ) - (min_depth,) = txn.fetchone() - - logger.info("[purge] updating room_depth to %d", min_depth) - txn.execute( - "UPDATE room_depth SET min_depth = ? WHERE room_id = ?", - (min_depth, room_id), + "DELETE FROM event_push_actions WHERE room_id = ? AND event_id = ?", + (room_id, event_id), ) - # finally, drop the temp table. this will commit the txn in sqlite, - # so make sure to keep this actually last. - txn.execute("DROP TABLE events_to_purge") - - logger.info("[purge] done") - - return referenced_state_groups - - def purge_room(self, room_id): - """Deletes all record of a room + def _store_rejections_txn(self, txn, event_id, reason): + self.db.simple_insert_txn( + txn, + table="rejections", + values={ + "event_id": event_id, + "reason": reason, + "last_check": self._clock.time_msec(), + }, + ) - Args: - room_id (str) + def _store_event_state_mappings_txn( + self, txn, events_and_contexts: Iterable[Tuple[EventBase, EventContext]] + ): + state_groups = {} + for event, context in events_and_contexts: + if event.internal_metadata.is_outlier(): + continue - Returns: - Deferred[List[int]]: The list of state groups to delete. - """ + # if the event was rejected, just give it the same state as its + # predecessor. + if context.rejected: + state_groups[event.event_id] = context.state_group_before_event + continue - return self.db.runInteraction("purge_room", self._purge_room_txn, room_id) + state_groups[event.event_id] = context.state_group - def _purge_room_txn(self, txn, room_id): - # First we fetch all the state groups that should be deleted, before - # we delete that information. - txn.execute( - """ - SELECT DISTINCT state_group FROM events - INNER JOIN event_to_state_groups USING(event_id) - WHERE events.room_id = ? - """, - (room_id,), + self.db.simple_insert_many_txn( + txn, + table="event_to_state_groups", + values=[ + {"state_group": state_group_id, "event_id": event_id} + for event_id, state_group_id in iteritems(state_groups) + ], ) - state_groups = [row[0] for row in txn] - - # Now we delete tables which lack an index on room_id but have one on event_id - for table in ( - "event_auth", - "event_edges", - "event_push_actions_staging", - "event_reference_hashes", - "event_relations", - "event_to_state_groups", - "redactions", - "rejections", - "state_events", - ): - logger.info("[purge] removing %s from %s", room_id, table) - - txn.execute( - """ - DELETE FROM %s WHERE event_id IN ( - SELECT event_id FROM events WHERE room_id=? - ) - """ - % (table,), - (room_id,), + for event_id, state_group_id in iteritems(state_groups): + txn.call_after( + self.store._get_state_group_for_event.prefill, + (event_id,), + state_group_id, ) - # and finally, the tables with an index on room_id (or no useful index) - for table in ( - "current_state_events", - "event_backward_extremities", - "event_forward_extremities", - "event_json", - "event_push_actions", - "event_search", - "events", - "group_rooms", - "public_room_list_stream", - "receipts_graph", - "receipts_linearized", - "room_aliases", - "room_depth", - "room_memberships", - "room_stats_state", - "room_stats_current", - "room_stats_historical", - "room_stats_earliest_token", - "rooms", - "stream_ordering_to_exterm", - "users_in_public_rooms", - "users_who_share_private_rooms", - # no useful index, but let's clear them anyway - "appservice_room_list", - "e2e_room_keys", - "event_push_summary", - "pusher_throttle", - "group_summary_rooms", - "local_invites", - "room_account_data", - "room_tags", - "local_current_membership", - ): - logger.info("[purge] removing %s from %s", room_id, table) - txn.execute("DELETE FROM %s WHERE room_id=?" % (table,), (room_id,)) - - # Other tables we do NOT need to clear out: - # - # - blocked_rooms - # This is important, to make sure that we don't accidentally rejoin a blocked - # room after it was purged - # - # - user_directory - # This has a room_id column, but it is unused - # - - # Other tables that we might want to consider clearing out include: - # - # - event_reports - # Given that these are intended for abuse management my initial - # inclination is to leave them in place. - # - # - current_state_delta_stream - # - ex_outlier_stream - # - room_tags_revisions - # The problem with these is that they are largeish and there is no room_id - # index on them. In any case we should be clearing out 'stream' tables - # periodically anyway (#5888) - - # TODO: we could probably usefully do a bunch of cache invalidation here + def _update_min_depth_for_room_txn(self, txn, room_id, depth): + min_depth = self.store._get_min_depth_interaction(txn, room_id) - logger.info("[purge] done") - - return state_groups - - async def is_event_after(self, event_id1, event_id2): - """Returns True if event_id1 is after event_id2 in the stream - """ - to_1, so_1 = await self._get_event_ordering(event_id1) - to_2, so_2 = await self._get_event_ordering(event_id2) - return (to_1, so_1) > (to_2, so_2) + if min_depth is not None and depth >= min_depth: + return - @cachedInlineCallbacks(max_entries=5000) - def _get_event_ordering(self, event_id): - res = yield self.db.simple_select_one( - table="events", - retcols=["topological_ordering", "stream_ordering"], - keyvalues={"event_id": event_id}, - allow_none=True, + self.db.simple_upsert_txn( + txn, + table="room_depth", + keyvalues={"room_id": room_id}, + values={"min_depth": depth}, ) - if not res: - raise SynapseError(404, "Could not find event %s" % (event_id,)) - - return (int(res["topological_ordering"]), int(res["stream_ordering"])) - - def insert_labels_for_event_txn( - self, txn, event_id, labels, room_id, topological_ordering - ): - """Store the mapping between an event's ID and its labels, with one row per - (event_id, label) tuple. - - Args: - txn (LoggingTransaction): The transaction to execute. - event_id (str): The event's ID. - labels (list[str]): A list of text labels. - room_id (str): The ID of the room the event was sent to. - topological_ordering (int): The position of the event in the room's topology. + def _handle_mult_prev_events(self, txn, events): """ - return self.db.simple_insert_many_txn( - txn=txn, - table="event_labels", + For the given event, update the event edges table and forward and + backward extremities tables. + """ + self.db.simple_insert_many_txn( + txn, + table="event_edges", values=[ { - "event_id": event_id, - "label": label, - "room_id": room_id, - "topological_ordering": topological_ordering, + "event_id": ev.event_id, + "prev_event_id": e_id, + "room_id": ev.room_id, + "is_state": False, } - for label in labels + for ev in events + for e_id in ev.prev_event_ids() ], ) - def _insert_event_expiry_txn(self, txn, event_id, expiry_ts): - """Save the expiry timestamp associated with a given event ID. - - Args: - txn (LoggingTransaction): The database transaction to use. - event_id (str): The event ID the expiry timestamp is associated with. - expiry_ts (int): The timestamp at which to expire (delete) the event. - """ - return self.db.simple_insert_txn( - txn=txn, - table="event_expiry", - values={"event_id": event_id, "expiry_ts": expiry_ts}, - ) - - @defer.inlineCallbacks - def expire_event(self, event_id): - """Retrieve and expire an event that has expired, and delete its associated - expiry timestamp. If the event can't be retrieved, delete its associated - timestamp so we don't try to expire it again in the future. - - Args: - event_id (str): The ID of the event to delete. - """ - # Try to retrieve the event's content from the database or the event cache. - event = yield self.get_event(event_id) - - def delete_expired_event_txn(txn): - # Delete the expiry timestamp associated with this event from the database. - self._delete_event_expiry_txn(txn, event_id) - - if not event: - # If we can't find the event, log a warning and delete the expiry date - # from the database so that we don't try to expire it again in the - # future. - logger.warning( - "Can't expire event %s because we don't have it.", event_id - ) - return - - # Prune the event's dict then convert it to JSON. - pruned_json = encode_json( - prune_event_dict(event.room_version, event.get_dict()) - ) - - # Update the event_json table to replace the event's JSON with the pruned - # JSON. - self._censor_event_txn(txn, event.event_id, pruned_json) - - # We need to invalidate the event cache entry for this event because we - # changed its content in the database. We can't call - # self._invalidate_cache_and_stream because self.get_event_cache isn't of the - # right type. - txn.call_after(self._get_event_cache.invalidate, (event.event_id,)) - # Send that invalidation to replication so that other workers also invalidate - # the event cache. - self._send_invalidation_to_replication( - txn, "_get_event_cache", (event.event_id,) - ) + self._update_backward_extremeties(txn, events) - yield self.db.runInteraction("delete_expired_event", delete_expired_event_txn) + def _update_backward_extremeties(self, txn, events): + """Updates the event_backward_extremities tables based on the new/updated + events being persisted. - def _delete_event_expiry_txn(self, txn, event_id): - """Delete the expiry timestamp associated with an event ID without deleting the - actual event. + This is called for new events *and* for events that were outliers, but + are now being persisted as non-outliers. - Args: - txn (LoggingTransaction): The transaction to use to perform the deletion. - event_id (str): The event ID to delete the associated expiry timestamp of. + Forward extremities are handled when we first start persisting the events. """ - return self.db.simple_delete_txn( - txn=txn, table="event_expiry", keyvalues={"event_id": event_id} + events_by_room = {} + for ev in events: + events_by_room.setdefault(ev.room_id, []).append(ev) + + query = ( + "INSERT INTO event_backward_extremities (event_id, room_id)" + " SELECT ?, ? WHERE NOT EXISTS (" + " SELECT 1 FROM event_backward_extremities" + " WHERE event_id = ? AND room_id = ?" + " )" + " AND NOT EXISTS (" + " SELECT 1 FROM events WHERE event_id = ? AND room_id = ? " + " AND outlier = ?" + " )" ) - def get_next_event_to_expire(self): - """Retrieve the entry with the lowest expiry timestamp in the event_expiry - table, or None if there's no more event to expire. - - Returns: Deferred[Optional[Tuple[str, int]]] - A tuple containing the event ID as its first element and an expiry timestamp - as its second one, if there's at least one row in the event_expiry table. - None otherwise. - """ - - def get_next_event_to_expire_txn(txn): - txn.execute( - """ - SELECT event_id, expiry_ts FROM event_expiry - ORDER BY expiry_ts ASC LIMIT 1 - """ - ) - - return txn.fetchone() - - return self.db.runInteraction( - desc="get_next_event_to_expire", func=get_next_event_to_expire_txn + txn.executemany( + query, + [ + (e_id, ev.room_id, e_id, ev.room_id, e_id, ev.room_id, False) + for ev in events + for e_id in ev.prev_event_ids() + if not ev.internal_metadata.is_outlier() + ], ) - -AllNewEventsResult = namedtuple( - "AllNewEventsResult", - [ - "new_forward_events", - "new_backfill_events", - "forward_ex_outliers", - "backward_ex_outliers", - ], -) + query = ( + "DELETE FROM event_backward_extremities" + " WHERE event_id = ? AND room_id = ?" + ) + txn.executemany( + query, + [ + (ev.event_id, ev.room_id) + for ev in events + if not ev.internal_metadata.is_outlier() + ], + ) diff --git a/synapse/storage/data_stores/main/events_worker.py b/synapse/storage/data_stores/main/events_worker.py index b8c1bbdf99..970c31bd05 100644 --- a/synapse/storage/data_stores/main/events_worker.py +++ b/synapse/storage/data_stores/main/events_worker.py @@ -27,7 +27,7 @@ from constantly import NamedConstant, Names from twisted.internet import defer from synapse.api.constants import EventTypes -from synapse.api.errors import NotFoundError +from synapse.api.errors import NotFoundError, SynapseError from synapse.api.room_versions import ( KNOWN_ROOM_VERSIONS, EventFormatVersions, @@ -40,7 +40,7 @@ from synapse.metrics.background_process_metrics import run_as_background_process from synapse.storage._base import SQLBaseStore, make_in_list_sql_clause from synapse.storage.database import Database from synapse.types import get_domain_from_id -from synapse.util.caches.descriptors import Cache +from synapse.util.caches.descriptors import Cache, cached, cachedInlineCallbacks from synapse.util.iterutils import batch_iter from synapse.util.metrics import Measure @@ -1157,4 +1157,152 @@ class EventsWorkerStore(SQLBaseStore): rows = await self.db.runInteraction( "get_deltas_for_stream_id", get_deltas_for_stream_id_txn, to_token ) + return rows, to_token, True + + @cached(num_args=5, max_entries=10) + def get_all_new_events( + self, + last_backfill_id, + last_forward_id, + current_backfill_id, + current_forward_id, + limit, + ): + """Get all the new events that have arrived at the server either as + new events or as backfilled events""" + have_backfill_events = last_backfill_id != current_backfill_id + have_forward_events = last_forward_id != current_forward_id + + if not have_backfill_events and not have_forward_events: + return defer.succeed(AllNewEventsResult([], [], [], [], [])) + + def get_all_new_events_txn(txn): + sql = ( + "SELECT e.stream_ordering, e.event_id, e.room_id, e.type," + " state_key, redacts" + " FROM events AS e" + " LEFT JOIN redactions USING (event_id)" + " LEFT JOIN state_events USING (event_id)" + " WHERE ? < stream_ordering AND stream_ordering <= ?" + " ORDER BY stream_ordering ASC" + " LIMIT ?" + ) + if have_forward_events: + txn.execute(sql, (last_forward_id, current_forward_id, limit)) + new_forward_events = txn.fetchall() + + if len(new_forward_events) == limit: + upper_bound = new_forward_events[-1][0] + else: + upper_bound = current_forward_id + + sql = ( + "SELECT event_stream_ordering, event_id, state_group" + " FROM ex_outlier_stream" + " WHERE ? > event_stream_ordering" + " AND event_stream_ordering >= ?" + " ORDER BY event_stream_ordering DESC" + ) + txn.execute(sql, (last_forward_id, upper_bound)) + forward_ex_outliers = txn.fetchall() + else: + new_forward_events = [] + forward_ex_outliers = [] + + sql = ( + "SELECT -e.stream_ordering, e.event_id, e.room_id, e.type," + " state_key, redacts" + " FROM events AS e" + " LEFT JOIN redactions USING (event_id)" + " LEFT JOIN state_events USING (event_id)" + " WHERE ? > stream_ordering AND stream_ordering >= ?" + " ORDER BY stream_ordering DESC" + " LIMIT ?" + ) + if have_backfill_events: + txn.execute(sql, (-last_backfill_id, -current_backfill_id, limit)) + new_backfill_events = txn.fetchall() + + if len(new_backfill_events) == limit: + upper_bound = new_backfill_events[-1][0] + else: + upper_bound = current_backfill_id + + sql = ( + "SELECT -event_stream_ordering, event_id, state_group" + " FROM ex_outlier_stream" + " WHERE ? > event_stream_ordering" + " AND event_stream_ordering >= ?" + " ORDER BY event_stream_ordering DESC" + ) + txn.execute(sql, (-last_backfill_id, -upper_bound)) + backward_ex_outliers = txn.fetchall() + else: + new_backfill_events = [] + backward_ex_outliers = [] + + return AllNewEventsResult( + new_forward_events, + new_backfill_events, + forward_ex_outliers, + backward_ex_outliers, + ) + + return self.db.runInteraction("get_all_new_events", get_all_new_events_txn) + + async def is_event_after(self, event_id1, event_id2): + """Returns True if event_id1 is after event_id2 in the stream + """ + to_1, so_1 = await self._get_event_ordering(event_id1) + to_2, so_2 = await self._get_event_ordering(event_id2) + return (to_1, so_1) > (to_2, so_2) + + @cachedInlineCallbacks(max_entries=5000) + def _get_event_ordering(self, event_id): + res = yield self.db.simple_select_one( + table="events", + retcols=["topological_ordering", "stream_ordering"], + keyvalues={"event_id": event_id}, + allow_none=True, + ) + + if not res: + raise SynapseError(404, "Could not find event %s" % (event_id,)) + + return (int(res["topological_ordering"]), int(res["stream_ordering"])) + + def get_next_event_to_expire(self): + """Retrieve the entry with the lowest expiry timestamp in the event_expiry + table, or None if there's no more event to expire. + + Returns: Deferred[Optional[Tuple[str, int]]] + A tuple containing the event ID as its first element and an expiry timestamp + as its second one, if there's at least one row in the event_expiry table. + None otherwise. + """ + + def get_next_event_to_expire_txn(txn): + txn.execute( + """ + SELECT event_id, expiry_ts FROM event_expiry + ORDER BY expiry_ts ASC LIMIT 1 + """ + ) + + return txn.fetchone() + + return self.db.runInteraction( + desc="get_next_event_to_expire", func=get_next_event_to_expire_txn + ) + + +AllNewEventsResult = namedtuple( + "AllNewEventsResult", + [ + "new_forward_events", + "new_backfill_events", + "forward_ex_outliers", + "backward_ex_outliers", + ], +) diff --git a/synapse/storage/data_stores/main/metrics.py b/synapse/storage/data_stores/main/metrics.py new file mode 100644 index 0000000000..dad5bbc602 --- /dev/null +++ b/synapse/storage/data_stores/main/metrics.py @@ -0,0 +1,128 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import typing +from collections import Counter + +from twisted.internet import defer + +from synapse.metrics import BucketCollector +from synapse.metrics.background_process_metrics import run_as_background_process +from synapse.storage._base import SQLBaseStore +from synapse.storage.data_stores.main.event_push_actions import ( + EventPushActionsWorkerStore, +) +from synapse.storage.database import Database + + +class ServerMetricsStore(EventPushActionsWorkerStore, SQLBaseStore): + """Functions to pull various metrics from the DB, for e.g. phone home + stats and prometheus metrics. + """ + + def __init__(self, database: Database, db_conn, hs): + super().__init__(database, db_conn, hs) + + # Collect metrics on the number of forward extremities that exist. + # Counter of number of extremities to count + self._current_forward_extremities_amount = ( + Counter() + ) # type: typing.Counter[int] + + BucketCollector( + "synapse_forward_extremities", + lambda: self._current_forward_extremities_amount, + buckets=[1, 2, 3, 5, 7, 10, 15, 20, 50, 100, 200, 500, "+Inf"], + ) + + # Read the extrems every 60 minutes + def read_forward_extremities(): + # run as a background process to make sure that the database transactions + # have a logcontext to report to + return run_as_background_process( + "read_forward_extremities", self._read_forward_extremities + ) + + hs.get_clock().looping_call(read_forward_extremities, 60 * 60 * 1000) + + async def _read_forward_extremities(self): + def fetch(txn): + txn.execute( + """ + select count(*) c from event_forward_extremities + group by room_id + """ + ) + return txn.fetchall() + + res = await self.db.runInteraction("read_forward_extremities", fetch) + self._current_forward_extremities_amount = Counter([x[0] for x in res]) + + @defer.inlineCallbacks + def count_daily_messages(self): + """ + Returns an estimate of the number of messages sent in the last day. + + If it has been significantly less or more than one day since the last + call to this function, it will return None. + """ + + def _count_messages(txn): + sql = """ + SELECT COALESCE(COUNT(*), 0) FROM events + WHERE type = 'm.room.message' + AND stream_ordering > ? + """ + txn.execute(sql, (self.stream_ordering_day_ago,)) + (count,) = txn.fetchone() + return count + + ret = yield self.db.runInteraction("count_messages", _count_messages) + return ret + + @defer.inlineCallbacks + def count_daily_sent_messages(self): + def _count_messages(txn): + # This is good enough as if you have silly characters in your own + # hostname then thats your own fault. + like_clause = "%:" + self.hs.hostname + + sql = """ + SELECT COALESCE(COUNT(*), 0) FROM events + WHERE type = 'm.room.message' + AND sender LIKE ? + AND stream_ordering > ? + """ + + txn.execute(sql, (like_clause, self.stream_ordering_day_ago)) + (count,) = txn.fetchone() + return count + + ret = yield self.db.runInteraction("count_daily_sent_messages", _count_messages) + return ret + + @defer.inlineCallbacks + def count_daily_active_rooms(self): + def _count(txn): + sql = """ + SELECT COALESCE(COUNT(DISTINCT room_id), 0) FROM events + WHERE type = 'm.room.message' + AND stream_ordering > ? + """ + txn.execute(sql, (self.stream_ordering_day_ago,)) + (count,) = txn.fetchone() + return count + + ret = yield self.db.runInteraction("count_daily_active_rooms", _count) + return ret diff --git a/synapse/storage/data_stores/main/purge_events.py b/synapse/storage/data_stores/main/purge_events.py new file mode 100644 index 0000000000..a93e1ef198 --- /dev/null +++ b/synapse/storage/data_stores/main/purge_events.py @@ -0,0 +1,399 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging +from typing import Any, Tuple + +from synapse.api.errors import SynapseError +from synapse.storage._base import SQLBaseStore +from synapse.storage.data_stores.main.state import StateGroupWorkerStore +from synapse.types import RoomStreamToken + +logger = logging.getLogger(__name__) + + +class PurgeEventsStore(StateGroupWorkerStore, SQLBaseStore): + def purge_history(self, room_id, token, delete_local_events): + """Deletes room history before a certain point + + Args: + room_id (str): + + token (str): A topological token to delete events before + + delete_local_events (bool): + if True, we will delete local events as well as remote ones + (instead of just marking them as outliers and deleting their + state groups). + + Returns: + Deferred[set[int]]: The set of state groups that are referenced by + deleted events. + """ + + return self.db.runInteraction( + "purge_history", + self._purge_history_txn, + room_id, + token, + delete_local_events, + ) + + def _purge_history_txn(self, txn, room_id, token_str, delete_local_events): + token = RoomStreamToken.parse(token_str) + + # Tables that should be pruned: + # event_auth + # event_backward_extremities + # event_edges + # event_forward_extremities + # event_json + # event_push_actions + # event_reference_hashes + # event_search + # event_to_state_groups + # events + # rejections + # room_depth + # state_groups + # state_groups_state + + # we will build a temporary table listing the events so that we don't + # have to keep shovelling the list back and forth across the + # connection. Annoyingly the python sqlite driver commits the + # transaction on CREATE, so let's do this first. + # + # furthermore, we might already have the table from a previous (failed) + # purge attempt, so let's drop the table first. + + txn.execute("DROP TABLE IF EXISTS events_to_purge") + + txn.execute( + "CREATE TEMPORARY TABLE events_to_purge (" + " event_id TEXT NOT NULL," + " should_delete BOOLEAN NOT NULL" + ")" + ) + + # First ensure that we're not about to delete all the forward extremeties + txn.execute( + "SELECT e.event_id, e.depth FROM events as e " + "INNER JOIN event_forward_extremities as f " + "ON e.event_id = f.event_id " + "AND e.room_id = f.room_id " + "WHERE f.room_id = ?", + (room_id,), + ) + rows = txn.fetchall() + max_depth = max(row[1] for row in rows) + + if max_depth < token.topological: + # We need to ensure we don't delete all the events from the database + # otherwise we wouldn't be able to send any events (due to not + # having any backwards extremeties) + raise SynapseError( + 400, "topological_ordering is greater than forward extremeties" + ) + + logger.info("[purge] looking for events to delete") + + should_delete_expr = "state_key IS NULL" + should_delete_params = () # type: Tuple[Any, ...] + if not delete_local_events: + should_delete_expr += " AND event_id NOT LIKE ?" + + # We include the parameter twice since we use the expression twice + should_delete_params += ("%:" + self.hs.hostname, "%:" + self.hs.hostname) + + should_delete_params += (room_id, token.topological) + + # Note that we insert events that are outliers and aren't going to be + # deleted, as nothing will happen to them. + txn.execute( + "INSERT INTO events_to_purge" + " SELECT event_id, %s" + " FROM events AS e LEFT JOIN state_events USING (event_id)" + " WHERE (NOT outlier OR (%s)) AND e.room_id = ? AND topological_ordering < ?" + % (should_delete_expr, should_delete_expr), + should_delete_params, + ) + + # We create the indices *after* insertion as that's a lot faster. + + # create an index on should_delete because later we'll be looking for + # the should_delete / shouldn't_delete subsets + txn.execute( + "CREATE INDEX events_to_purge_should_delete" + " ON events_to_purge(should_delete)" + ) + + # We do joins against events_to_purge for e.g. calculating state + # groups to purge, etc., so lets make an index. + txn.execute("CREATE INDEX events_to_purge_id ON events_to_purge(event_id)") + + txn.execute("SELECT event_id, should_delete FROM events_to_purge") + event_rows = txn.fetchall() + logger.info( + "[purge] found %i events before cutoff, of which %i can be deleted", + len(event_rows), + sum(1 for e in event_rows if e[1]), + ) + + logger.info("[purge] Finding new backward extremities") + + # We calculate the new entries for the backward extremeties by finding + # events to be purged that are pointed to by events we're not going to + # purge. + txn.execute( + "SELECT DISTINCT e.event_id FROM events_to_purge AS e" + " INNER JOIN event_edges AS ed ON e.event_id = ed.prev_event_id" + " LEFT JOIN events_to_purge AS ep2 ON ed.event_id = ep2.event_id" + " WHERE ep2.event_id IS NULL" + ) + new_backwards_extrems = txn.fetchall() + + logger.info("[purge] replacing backward extremities: %r", new_backwards_extrems) + + txn.execute( + "DELETE FROM event_backward_extremities WHERE room_id = ?", (room_id,) + ) + + # Update backward extremeties + txn.executemany( + "INSERT INTO event_backward_extremities (room_id, event_id)" + " VALUES (?, ?)", + [(room_id, event_id) for event_id, in new_backwards_extrems], + ) + + logger.info("[purge] finding state groups referenced by deleted events") + + # Get all state groups that are referenced by events that are to be + # deleted. + txn.execute( + """ + SELECT DISTINCT state_group FROM events_to_purge + INNER JOIN event_to_state_groups USING (event_id) + """ + ) + + referenced_state_groups = {sg for sg, in txn} + logger.info( + "[purge] found %i referenced state groups", len(referenced_state_groups) + ) + + logger.info("[purge] removing events from event_to_state_groups") + txn.execute( + "DELETE FROM event_to_state_groups " + "WHERE event_id IN (SELECT event_id from events_to_purge)" + ) + for event_id, _ in event_rows: + txn.call_after(self._get_state_group_for_event.invalidate, (event_id,)) + + # Delete all remote non-state events + for table in ( + "events", + "event_json", + "event_auth", + "event_edges", + "event_forward_extremities", + "event_reference_hashes", + "event_search", + "rejections", + ): + logger.info("[purge] removing events from %s", table) + + txn.execute( + "DELETE FROM %s WHERE event_id IN (" + " SELECT event_id FROM events_to_purge WHERE should_delete" + ")" % (table,) + ) + + # event_push_actions lacks an index on event_id, and has one on + # (room_id, event_id) instead. + for table in ("event_push_actions",): + logger.info("[purge] removing events from %s", table) + + txn.execute( + "DELETE FROM %s WHERE room_id = ? AND event_id IN (" + " SELECT event_id FROM events_to_purge WHERE should_delete" + ")" % (table,), + (room_id,), + ) + + # Mark all state and own events as outliers + logger.info("[purge] marking remaining events as outliers") + txn.execute( + "UPDATE events SET outlier = ?" + " WHERE event_id IN (" + " SELECT event_id FROM events_to_purge " + " WHERE NOT should_delete" + ")", + (True,), + ) + + # synapse tries to take out an exclusive lock on room_depth whenever it + # persists events (because upsert), and once we run this update, we + # will block that for the rest of our transaction. + # + # So, let's stick it at the end so that we don't block event + # persistence. + # + # We do this by calculating the minimum depth of the backwards + # extremities. However, the events in event_backward_extremities + # are ones we don't have yet so we need to look at the events that + # point to it via event_edges table. + txn.execute( + """ + SELECT COALESCE(MIN(depth), 0) + FROM event_backward_extremities AS eb + INNER JOIN event_edges AS eg ON eg.prev_event_id = eb.event_id + INNER JOIN events AS e ON e.event_id = eg.event_id + WHERE eb.room_id = ? + """, + (room_id,), + ) + (min_depth,) = txn.fetchone() + + logger.info("[purge] updating room_depth to %d", min_depth) + + txn.execute( + "UPDATE room_depth SET min_depth = ? WHERE room_id = ?", + (min_depth, room_id), + ) + + # finally, drop the temp table. this will commit the txn in sqlite, + # so make sure to keep this actually last. + txn.execute("DROP TABLE events_to_purge") + + logger.info("[purge] done") + + return referenced_state_groups + + def purge_room(self, room_id): + """Deletes all record of a room + + Args: + room_id (str) + + Returns: + Deferred[List[int]]: The list of state groups to delete. + """ + + return self.db.runInteraction("purge_room", self._purge_room_txn, room_id) + + def _purge_room_txn(self, txn, room_id): + # First we fetch all the state groups that should be deleted, before + # we delete that information. + txn.execute( + """ + SELECT DISTINCT state_group FROM events + INNER JOIN event_to_state_groups USING(event_id) + WHERE events.room_id = ? + """, + (room_id,), + ) + + state_groups = [row[0] for row in txn] + + # Now we delete tables which lack an index on room_id but have one on event_id + for table in ( + "event_auth", + "event_edges", + "event_push_actions_staging", + "event_reference_hashes", + "event_relations", + "event_to_state_groups", + "redactions", + "rejections", + "state_events", + ): + logger.info("[purge] removing %s from %s", room_id, table) + + txn.execute( + """ + DELETE FROM %s WHERE event_id IN ( + SELECT event_id FROM events WHERE room_id=? + ) + """ + % (table,), + (room_id,), + ) + + # and finally, the tables with an index on room_id (or no useful index) + for table in ( + "current_state_events", + "event_backward_extremities", + "event_forward_extremities", + "event_json", + "event_push_actions", + "event_search", + "events", + "group_rooms", + "public_room_list_stream", + "receipts_graph", + "receipts_linearized", + "room_aliases", + "room_depth", + "room_memberships", + "room_stats_state", + "room_stats_current", + "room_stats_historical", + "room_stats_earliest_token", + "rooms", + "stream_ordering_to_exterm", + "users_in_public_rooms", + "users_who_share_private_rooms", + # no useful index, but let's clear them anyway + "appservice_room_list", + "e2e_room_keys", + "event_push_summary", + "pusher_throttle", + "group_summary_rooms", + "local_invites", + "room_account_data", + "room_tags", + "local_current_membership", + ): + logger.info("[purge] removing %s from %s", room_id, table) + txn.execute("DELETE FROM %s WHERE room_id=?" % (table,), (room_id,)) + + # Other tables we do NOT need to clear out: + # + # - blocked_rooms + # This is important, to make sure that we don't accidentally rejoin a blocked + # room after it was purged + # + # - user_directory + # This has a room_id column, but it is unused + # + + # Other tables that we might want to consider clearing out include: + # + # - event_reports + # Given that these are intended for abuse management my initial + # inclination is to leave them in place. + # + # - current_state_delta_stream + # - ex_outlier_stream + # - room_tags_revisions + # The problem with these is that they are largeish and there is no room_id + # index on them. In any case we should be clearing out 'stream' tables + # periodically anyway (#5888) + + # TODO: we could probably usefully do a bunch of cache invalidation here + + logger.info("[purge] done") + + return state_groups diff --git a/synapse/storage/data_stores/main/rejections.py b/synapse/storage/data_stores/main/rejections.py index 1c07c7a425..27e5a2084a 100644 --- a/synapse/storage/data_stores/main/rejections.py +++ b/synapse/storage/data_stores/main/rejections.py @@ -21,17 +21,6 @@ logger = logging.getLogger(__name__) class RejectionsStore(SQLBaseStore): - def _store_rejections_txn(self, txn, event_id, reason): - self.db.simple_insert_txn( - txn, - table="rejections", - values={ - "event_id": event_id, - "reason": reason, - "last_check": self._clock.time_msec(), - }, - ) - def get_rejection_reason(self, event_id): return self.db.simple_select_one_onecol( table="rejections", diff --git a/synapse/storage/data_stores/main/relations.py b/synapse/storage/data_stores/main/relations.py index 046c2b4845..7d477f8d01 100644 --- a/synapse/storage/data_stores/main/relations.py +++ b/synapse/storage/data_stores/main/relations.py @@ -324,62 +324,4 @@ class RelationsWorkerStore(SQLBaseStore): class RelationsStore(RelationsWorkerStore): - def _handle_event_relations(self, txn, event): - """Handles inserting relation data during peristence of events - - Args: - txn - event (EventBase) - """ - relation = event.content.get("m.relates_to") - if not relation: - # No relations - return - - rel_type = relation.get("rel_type") - if rel_type not in ( - RelationTypes.ANNOTATION, - RelationTypes.REFERENCE, - RelationTypes.REPLACE, - ): - # Unknown relation type - return - - parent_id = relation.get("event_id") - if not parent_id: - # Invalid relation - return - - aggregation_key = relation.get("key") - - self.db.simple_insert_txn( - txn, - table="event_relations", - values={ - "event_id": event.event_id, - "relates_to_id": parent_id, - "relation_type": rel_type, - "aggregation_key": aggregation_key, - }, - ) - - txn.call_after(self.get_relations_for_event.invalidate_many, (parent_id,)) - txn.call_after( - self.get_aggregation_groups_for_event.invalidate_many, (parent_id,) - ) - - if rel_type == RelationTypes.REPLACE: - txn.call_after(self.get_applicable_edit.invalidate, (parent_id,)) - - def _handle_redaction(self, txn, redacted_event_id): - """Handles receiving a redaction and checking whether we need to remove - any redacted relations from the database. - - Args: - txn - redacted_event_id (str): The event that was redacted. - """ - - self.db.simple_delete_txn( - txn, table="event_relations", keyvalues={"event_id": redacted_event_id} - ) + pass diff --git a/synapse/storage/data_stores/main/room.py b/synapse/storage/data_stores/main/room.py index cafa664c16..46f643c6b9 100644 --- a/synapse/storage/data_stores/main/room.py +++ b/synapse/storage/data_stores/main/room.py @@ -21,8 +21,6 @@ from abc import abstractmethod from enum import Enum from typing import Any, Dict, List, Optional, Tuple -from six import integer_types - from canonicaljson import json from twisted.internet import defer @@ -1302,53 +1300,6 @@ class RoomStore(RoomBackgroundUpdateStore, RoomWorkerStore, SearchStore): return self.db.runInteraction("get_rooms", f) - def _store_room_topic_txn(self, txn, event): - if hasattr(event, "content") and "topic" in event.content: - self.store_event_search_txn( - txn, event, "content.topic", event.content["topic"] - ) - - def _store_room_name_txn(self, txn, event): - if hasattr(event, "content") and "name" in event.content: - self.store_event_search_txn( - txn, event, "content.name", event.content["name"] - ) - - def _store_room_message_txn(self, txn, event): - if hasattr(event, "content") and "body" in event.content: - self.store_event_search_txn( - txn, event, "content.body", event.content["body"] - ) - - def _store_retention_policy_for_room_txn(self, txn, event): - if hasattr(event, "content") and ( - "min_lifetime" in event.content or "max_lifetime" in event.content - ): - if ( - "min_lifetime" in event.content - and not isinstance(event.content.get("min_lifetime"), integer_types) - ) or ( - "max_lifetime" in event.content - and not isinstance(event.content.get("max_lifetime"), integer_types) - ): - # Ignore the event if one of the value isn't an integer. - return - - self.db.simple_insert_txn( - txn=txn, - table="room_retention", - values={ - "room_id": event.room_id, - "event_id": event.event_id, - "min_lifetime": event.content.get("min_lifetime"), - "max_lifetime": event.content.get("max_lifetime"), - }, - ) - - self._invalidate_cache_and_stream( - txn, self.get_retention_policy_for_room, (event.room_id,) - ) - def add_event_report( self, room_id, event_id, user_id, reason, content, received_ts ): diff --git a/synapse/storage/data_stores/main/roommember.py b/synapse/storage/data_stores/main/roommember.py index ae9a76713c..04ffdcf934 100644 --- a/synapse/storage/data_stores/main/roommember.py +++ b/synapse/storage/data_stores/main/roommember.py @@ -1050,96 +1050,6 @@ class RoomMemberStore(RoomMemberWorkerStore, RoomMemberBackgroundUpdateStore): def __init__(self, database: Database, db_conn, hs): super(RoomMemberStore, self).__init__(database, db_conn, hs) - def _store_room_members_txn(self, txn, events, backfilled): - """Store a room member in the database. - """ - self.db.simple_insert_many_txn( - txn, - table="room_memberships", - values=[ - { - "event_id": event.event_id, - "user_id": event.state_key, - "sender": event.user_id, - "room_id": event.room_id, - "membership": event.membership, - "display_name": event.content.get("displayname", None), - "avatar_url": event.content.get("avatar_url", None), - } - for event in events - ], - ) - - for event in events: - txn.call_after( - self._membership_stream_cache.entity_has_changed, - event.state_key, - event.internal_metadata.stream_ordering, - ) - txn.call_after( - self.get_invited_rooms_for_local_user.invalidate, (event.state_key,) - ) - - # We update the local_invites table only if the event is "current", - # i.e., its something that has just happened. If the event is an - # outlier it is only current if its an "out of band membership", - # like a remote invite or a rejection of a remote invite. - is_new_state = not backfilled and ( - not event.internal_metadata.is_outlier() - or event.internal_metadata.is_out_of_band_membership() - ) - is_mine = self.hs.is_mine_id(event.state_key) - if is_new_state and is_mine: - if event.membership == Membership.INVITE: - self.db.simple_insert_txn( - txn, - table="local_invites", - values={ - "event_id": event.event_id, - "invitee": event.state_key, - "inviter": event.sender, - "room_id": event.room_id, - "stream_id": event.internal_metadata.stream_ordering, - }, - ) - else: - sql = ( - "UPDATE local_invites SET stream_id = ?, replaced_by = ? WHERE" - " room_id = ? AND invitee = ? AND locally_rejected is NULL" - " AND replaced_by is NULL" - ) - - txn.execute( - sql, - ( - event.internal_metadata.stream_ordering, - event.event_id, - event.room_id, - event.state_key, - ), - ) - - # We also update the `local_current_membership` table with - # latest invite info. This will usually get updated by the - # `current_state_events` handling, unless its an outlier. - if event.internal_metadata.is_outlier(): - # This should only happen for out of band memberships, so - # we add a paranoia check. - assert event.internal_metadata.is_out_of_band_membership() - - self.db.simple_upsert_txn( - txn, - table="local_current_membership", - keyvalues={ - "room_id": event.room_id, - "user_id": event.state_key, - }, - values={ - "event_id": event.event_id, - "membership": event.membership, - }, - ) - @defer.inlineCallbacks def locally_reject_invite(self, user_id, room_id): sql = ( diff --git a/synapse/storage/data_stores/main/search.py b/synapse/storage/data_stores/main/search.py index 47ebb8a214..ee75b92344 100644 --- a/synapse/storage/data_stores/main/search.py +++ b/synapse/storage/data_stores/main/search.py @@ -347,29 +347,6 @@ class SearchStore(SearchBackgroundUpdateStore): def __init__(self, database: Database, db_conn, hs): super(SearchStore, self).__init__(database, db_conn, hs) - def store_event_search_txn(self, txn, event, key, value): - """Add event to the search table - - Args: - txn (cursor): - event (EventBase): - key (str): - value (str): - """ - self.store_search_entries_txn( - txn, - ( - SearchEntry( - key=key, - value=value, - event_id=event.event_id, - room_id=event.room_id, - stream_ordering=event.internal_metadata.stream_ordering, - origin_server_ts=event.origin_server_ts, - ), - ), - ) - @defer.inlineCallbacks def search_msgs(self, room_ids, search_term, keys): """Performs a full text search over events with given keys. diff --git a/synapse/storage/data_stores/main/signatures.py b/synapse/storage/data_stores/main/signatures.py index 563216b63c..36244d9f5d 100644 --- a/synapse/storage/data_stores/main/signatures.py +++ b/synapse/storage/data_stores/main/signatures.py @@ -13,23 +13,13 @@ # See the License for the specific language governing permissions and # limitations under the License. -import six - from unpaddedbase64 import encode_base64 from twisted.internet import defer -from synapse.crypto.event_signing import compute_event_reference_hash from synapse.storage._base import SQLBaseStore from synapse.util.caches.descriptors import cached, cachedList -# py2 sqlite has buffer hardcoded as only binary type, so we must use it, -# despite being deprecated and removed in favor of memoryview -if six.PY2: - db_binary_type = six.moves.builtins.buffer -else: - db_binary_type = memoryview - class SignatureWorkerStore(SQLBaseStore): @cached() @@ -79,23 +69,3 @@ class SignatureWorkerStore(SQLBaseStore): class SignatureStore(SignatureWorkerStore): """Persistence for event signatures and hashes""" - - def _store_event_reference_hashes_txn(self, txn, events): - """Store a hash for a PDU - Args: - txn (cursor): - events (list): list of Events. - """ - - vals = [] - for event in events: - ref_alg, ref_hash_bytes = compute_event_reference_hash(event) - vals.append( - { - "event_id": event.event_id, - "algorithm": ref_alg, - "hash": db_binary_type(ref_hash_bytes), - } - ) - - self.db.simple_insert_many_txn(txn, table="event_reference_hashes", values=vals) diff --git a/synapse/storage/data_stores/main/state.py b/synapse/storage/data_stores/main/state.py index 3a3b9a8e72..21052fcc7a 100644 --- a/synapse/storage/data_stores/main/state.py +++ b/synapse/storage/data_stores/main/state.py @@ -16,17 +16,12 @@ import collections.abc import logging from collections import namedtuple -from typing import Iterable, Tuple - -from six import iteritems from twisted.internet import defer from synapse.api.constants import EventTypes, Membership from synapse.api.errors import NotFoundError, UnsupportedRoomVersionError from synapse.api.room_versions import KNOWN_ROOM_VERSIONS, RoomVersion -from synapse.events import EventBase -from synapse.events.snapshot import EventContext from synapse.storage._base import SQLBaseStore from synapse.storage.data_stores.main.events_worker import EventsWorkerStore from synapse.storage.data_stores.main.roommember import RoomMemberWorkerStore @@ -473,33 +468,3 @@ class StateStore(StateGroupWorkerStore, MainStateBackgroundUpdateStore): def __init__(self, database: Database, db_conn, hs): super(StateStore, self).__init__(database, db_conn, hs) - - def _store_event_state_mappings_txn( - self, txn, events_and_contexts: Iterable[Tuple[EventBase, EventContext]] - ): - state_groups = {} - for event, context in events_and_contexts: - if event.internal_metadata.is_outlier(): - continue - - # if the event was rejected, just give it the same state as its - # predecessor. - if context.rejected: - state_groups[event.event_id] = context.state_group_before_event - continue - - state_groups[event.event_id] = context.state_group - - self.db.simple_insert_many_txn( - txn, - table="event_to_state_groups", - values=[ - {"state_group": state_group_id, "event_id": event_id} - for event_id, state_group_id in iteritems(state_groups) - ], - ) - - for event_id, state_group_id in iteritems(state_groups): - txn.call_after( - self._get_state_group_for_event.prefill, (event_id,), state_group_id - ) diff --git a/synapse/storage/persist_events.py b/synapse/storage/persist_events.py index 0f9ac1cf09..41881ea20b 100644 --- a/synapse/storage/persist_events.py +++ b/synapse/storage/persist_events.py @@ -23,7 +23,6 @@ from typing import Iterable, List, Optional, Set, Tuple from six import iteritems from six.moves import range -import attr from prometheus_client import Counter, Histogram from twisted.internet import defer @@ -35,6 +34,7 @@ from synapse.logging.context import PreserveLoggingContext, make_deferred_yielda from synapse.metrics.background_process_metrics import run_as_background_process from synapse.state import StateResolutionStore from synapse.storage.data_stores import DataStores +from synapse.storage.data_stores.main.events import DeltaState from synapse.types import StateMap from synapse.util.async_helpers import ObservableDeferred from synapse.util.metrics import Measure @@ -73,22 +73,6 @@ stale_forward_extremities_counter = Histogram( ) -@attr.s(slots=True) -class DeltaState: - """Deltas to use to update the `current_state_events` table. - - Attributes: - to_delete: List of type/state_keys to delete from current state - to_insert: Map of state to upsert into current state - no_longer_in_room: The server is not longer in the room, so the room - should e.g. be removed from `current_state_events` table. - """ - - to_delete = attr.ib(type=List[Tuple[str, str]]) - to_insert = attr.ib(type=StateMap[str]) - no_longer_in_room = attr.ib(type=bool, default=False) - - class _EventPeristenceQueue(object): """Queues up events so that they can be persisted in bulk with only one concurrent transaction per room. @@ -205,6 +189,7 @@ class EventsPersistenceStorage(object): # store for now. self.main_store = stores.main self.state_store = stores.state + self.persist_events_store = stores.persist_events self._clock = hs.get_clock() self.is_mine_id = hs.is_mine_id @@ -445,7 +430,7 @@ class EventsPersistenceStorage(object): if current_state is not None: current_state_for_room[room_id] = current_state - await self.main_store._persist_events_and_state_updates( + await self.persist_events_store._persist_events_and_state_updates( chunk, current_state_for_room=current_state_for_room, state_delta_for_room=state_delta_for_room, @@ -491,13 +476,15 @@ class EventsPersistenceStorage(object): ) # Remove any events which are prev_events of any existing events. - existing_prevs = await self.main_store._get_events_which_are_prevs(result) + existing_prevs = await self.persist_events_store._get_events_which_are_prevs( + result + ) result.difference_update(existing_prevs) # Finally handle the case where the new events have soft-failed prev # events. If they do we need to remove them and their prev events, # otherwise we end up with dangling extremities. - existing_prevs = await self.main_store._get_prevs_before_rejected( + existing_prevs = await self.persist_events_store._get_prevs_before_rejected( e_id for event in new_events for e_id in event.prev_event_ids() ) result.difference_update(existing_prevs) diff --git a/tests/storage/test_event_push_actions.py b/tests/storage/test_event_push_actions.py index d4bcf1821e..b45bc9c115 100644 --- a/tests/storage/test_event_push_actions.py +++ b/tests/storage/test_event_push_actions.py @@ -35,6 +35,7 @@ class EventPushActionsStoreTestCase(tests.unittest.TestCase): def setUp(self): hs = yield tests.utils.setup_test_homeserver(self.addCleanup) self.store = hs.get_datastore() + self.persist_events_store = hs.get_datastores().persist_events @defer.inlineCallbacks def test_get_unread_push_actions_for_user_in_range_for_http(self): @@ -76,7 +77,7 @@ class EventPushActionsStoreTestCase(tests.unittest.TestCase): ) yield self.store.db.runInteraction( "", - self.store._set_push_actions_for_event_and_users_txn, + self.persist_events_store._set_push_actions_for_event_and_users_txn, [(event, None)], [(event, None)], ) -- cgit 1.4.1 From 18c1e52d82a1d564906bd08e9af74cdb0f6efcd1 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 13 May 2020 16:01:47 +0100 Subject: Clean up replication unit tests. (#7490) --- changelog.d/7490.misc | 1 + tests/replication/_base.py | 307 +++++++++++++++++++++++ tests/replication/slave/storage/_base.py | 50 +--- tests/replication/slave/storage/test_events.py | 8 +- tests/replication/tcp/streams/_base.py | 307 ----------------------- tests/replication/tcp/streams/test_events.py | 2 +- tests/replication/tcp/streams/test_federation.py | 2 +- tests/replication/tcp/streams/test_receipts.py | 2 +- tests/replication/tcp/streams/test_typing.py | 2 +- 9 files changed, 323 insertions(+), 358 deletions(-) create mode 100644 changelog.d/7490.misc create mode 100644 tests/replication/_base.py delete mode 100644 tests/replication/tcp/streams/_base.py (limited to 'tests') diff --git a/changelog.d/7490.misc b/changelog.d/7490.misc new file mode 100644 index 0000000000..7debf7871b --- /dev/null +++ b/changelog.d/7490.misc @@ -0,0 +1 @@ +Clean up replication unit tests. diff --git a/tests/replication/_base.py b/tests/replication/_base.py new file mode 100644 index 0000000000..9d4f0bbe44 --- /dev/null +++ b/tests/replication/_base.py @@ -0,0 +1,307 @@ +# -*- coding: utf-8 -*- +# Copyright 2019 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging +from typing import Any, List, Optional, Tuple + +import attr + +from twisted.internet.interfaces import IConsumer, IPullProducer, IReactorTime +from twisted.internet.task import LoopingCall +from twisted.web.http import HTTPChannel + +from synapse.app.generic_worker import ( + GenericWorkerReplicationHandler, + GenericWorkerServer, +) +from synapse.http.site import SynapseRequest +from synapse.replication.http import streams +from synapse.replication.tcp.handler import ReplicationCommandHandler +from synapse.replication.tcp.protocol import ClientReplicationStreamProtocol +from synapse.replication.tcp.resource import ReplicationStreamProtocolFactory +from synapse.server import HomeServer +from synapse.util import Clock + +from tests import unittest +from tests.server import FakeTransport + +logger = logging.getLogger(__name__) + + +class BaseStreamTestCase(unittest.HomeserverTestCase): + """Base class for tests of the replication streams""" + + servlets = [ + streams.register_servlets, + ] + + def prepare(self, reactor, clock, hs): + # build a replication server + server_factory = ReplicationStreamProtocolFactory(hs) + self.streamer = hs.get_replication_streamer() + self.server = server_factory.buildProtocol(None) + + # Make a new HomeServer object for the worker + self.reactor.lookups["testserv"] = "1.2.3.4" + self.worker_hs = self.setup_test_homeserver( + http_client=None, + homeserverToUse=GenericWorkerServer, + config=self._get_worker_hs_config(), + reactor=self.reactor, + ) + + # Since we use sqlite in memory databases we need to make sure the + # databases objects are the same. + self.worker_hs.get_datastore().db = hs.get_datastore().db + + self.test_handler = self._build_replication_data_handler() + self.worker_hs.replication_data_handler = self.test_handler + + repl_handler = ReplicationCommandHandler(self.worker_hs) + self.client = ClientReplicationStreamProtocol( + self.worker_hs, "client", "test", clock, repl_handler, + ) + + self._client_transport = None + self._server_transport = None + + def _get_worker_hs_config(self) -> dict: + config = self.default_config() + config["worker_app"] = "synapse.app.generic_worker" + config["worker_replication_host"] = "testserv" + config["worker_replication_http_port"] = "8765" + return config + + def _build_replication_data_handler(self): + return TestReplicationDataHandler(self.worker_hs) + + def reconnect(self): + if self._client_transport: + self.client.close() + + if self._server_transport: + self.server.close() + + self._client_transport = FakeTransport(self.server, self.reactor) + self.client.makeConnection(self._client_transport) + + self._server_transport = FakeTransport(self.client, self.reactor) + self.server.makeConnection(self._server_transport) + + def disconnect(self): + if self._client_transport: + self._client_transport = None + self.client.close() + + if self._server_transport: + self._server_transport = None + self.server.close() + + def replicate(self): + """Tell the master side of replication that something has happened, and then + wait for the replication to occur. + """ + self.streamer.on_notifier_poke() + self.pump(0.1) + + def handle_http_replication_attempt(self) -> SynapseRequest: + """Asserts that a connection attempt was made to the master HS on the + HTTP replication port, then proxies it to the master HS object to be + handled. + + Returns: + The request object received by master HS. + """ + + # We should have an outbound connection attempt. + clients = self.reactor.tcpClients + self.assertEqual(len(clients), 1) + (host, port, client_factory, _timeout, _bindAddress) = clients.pop(0) + self.assertEqual(host, "1.2.3.4") + self.assertEqual(port, 8765) + + # Set up client side protocol + client_protocol = client_factory.buildProtocol(None) + + request_factory = OneShotRequestFactory() + + # Set up the server side protocol + channel = _PushHTTPChannel(self.reactor) + channel.requestFactory = request_factory + channel.site = self.site + + # Connect client to server and vice versa. + client_to_server_transport = FakeTransport( + channel, self.reactor, client_protocol + ) + client_protocol.makeConnection(client_to_server_transport) + + server_to_client_transport = FakeTransport( + client_protocol, self.reactor, channel + ) + channel.makeConnection(server_to_client_transport) + + # The request will now be processed by `self.site` and the response + # streamed back. + self.reactor.advance(0) + + # We tear down the connection so it doesn't get reused without our + # knowledge. + server_to_client_transport.loseConnection() + client_to_server_transport.loseConnection() + + return request_factory.request + + def assert_request_is_get_repl_stream_updates( + self, request: SynapseRequest, stream_name: str + ): + """Asserts that the given request is a HTTP replication request for + fetching updates for given stream. + """ + + self.assertRegex( + request.path, + br"^/_synapse/replication/get_repl_stream_updates/%s/[^/]+$" + % (stream_name.encode("ascii"),), + ) + + self.assertEqual(request.method, b"GET") + + +class TestReplicationDataHandler(GenericWorkerReplicationHandler): + """Drop-in for ReplicationDataHandler which just collects RDATA rows""" + + def __init__(self, hs: HomeServer): + super().__init__(hs) + + # list of received (stream_name, token, row) tuples + self.received_rdata_rows = [] # type: List[Tuple[str, int, Any]] + + async def on_rdata(self, stream_name, instance_name, token, rows): + await super().on_rdata(stream_name, instance_name, token, rows) + for r in rows: + self.received_rdata_rows.append((stream_name, token, r)) + + +@attr.s() +class OneShotRequestFactory: + """A simple request factory that generates a single `SynapseRequest` and + stores it for future use. Can only be used once. + """ + + request = attr.ib(default=None) + + def __call__(self, *args, **kwargs): + assert self.request is None + + self.request = SynapseRequest(*args, **kwargs) + return self.request + + +class _PushHTTPChannel(HTTPChannel): + """A HTTPChannel that wraps pull producers to push producers. + + This is a hack to get around the fact that HTTPChannel transparently wraps a + pull producer (which is what Synapse uses to reply to requests) with + `_PullToPush` to convert it to a push producer. Unfortunately `_PullToPush` + uses the standard reactor rather than letting us use our test reactor, which + makes it very hard to test. + """ + + def __init__(self, reactor: IReactorTime): + super().__init__() + self.reactor = reactor + + self._pull_to_push_producer = None # type: Optional[_PullToPushProducer] + + def registerProducer(self, producer, streaming): + # Convert pull producers to push producer. + if not streaming: + self._pull_to_push_producer = _PullToPushProducer( + self.reactor, producer, self + ) + producer = self._pull_to_push_producer + + super().registerProducer(producer, True) + + def unregisterProducer(self): + if self._pull_to_push_producer: + # We need to manually stop the _PullToPushProducer. + self._pull_to_push_producer.stop() + + +class _PullToPushProducer: + """A push producer that wraps a pull producer. + """ + + def __init__( + self, reactor: IReactorTime, producer: IPullProducer, consumer: IConsumer + ): + self._clock = Clock(reactor) + self._producer = producer + self._consumer = consumer + + # While running we use a looping call with a zero delay to call + # resumeProducing on given producer. + self._looping_call = None # type: Optional[LoopingCall] + + # We start writing next reactor tick. + self._start_loop() + + def _start_loop(self): + """Start the looping call to + """ + + if not self._looping_call: + # Start a looping call which runs every tick. + self._looping_call = self._clock.looping_call(self._run_once, 0) + + def stop(self): + """Stops calling resumeProducing. + """ + if self._looping_call: + self._looping_call.stop() + self._looping_call = None + + def pauseProducing(self): + """Implements IPushProducer + """ + self.stop() + + def resumeProducing(self): + """Implements IPushProducer + """ + self._start_loop() + + def stopProducing(self): + """Implements IPushProducer + """ + self.stop() + self._producer.stopProducing() + + def _run_once(self): + """Calls resumeProducing on producer once. + """ + + try: + self._producer.resumeProducing() + except Exception: + logger.exception("Failed to call resumeProducing") + try: + self._consumer.unregisterProducer() + except Exception: + pass + + self.stopProducing() diff --git a/tests/replication/slave/storage/_base.py b/tests/replication/slave/storage/_base.py index 1615dfab5e..32cb04645f 100644 --- a/tests/replication/slave/storage/_base.py +++ b/tests/replication/slave/storage/_base.py @@ -15,23 +15,13 @@ from mock import Mock, NonCallableMock -from synapse.replication.tcp.client import ( - DirectTcpReplicationClientFactory, - ReplicationDataHandler, -) -from synapse.replication.tcp.handler import ReplicationCommandHandler -from synapse.replication.tcp.resource import ReplicationStreamProtocolFactory -from synapse.storage.database import make_conn +from tests.replication._base import BaseStreamTestCase -from tests import unittest -from tests.server import FakeTransport - -class BaseSlavedStoreTestCase(unittest.HomeserverTestCase): +class BaseSlavedStoreTestCase(BaseStreamTestCase): def make_homeserver(self, reactor, clock): hs = self.setup_test_homeserver( - "blue", federation_client=Mock(), ratelimiter=NonCallableMock(spec_set=["can_do_action"]), ) @@ -41,39 +31,13 @@ class BaseSlavedStoreTestCase(unittest.HomeserverTestCase): return hs def prepare(self, reactor, clock, hs): + super().prepare(reactor, clock, hs) - db_config = hs.config.database.get_single_database() - self.master_store = self.hs.get_datastore() - self.storage = hs.get_storage() - database = hs.get_datastores().databases[0] - self.slaved_store = self.STORE_TYPE( - database, make_conn(db_config, database.engine), self.hs - ) - self.event_id = 0 - - server_factory = ReplicationStreamProtocolFactory(self.hs) - self.streamer = hs.get_replication_streamer() - - # We now do some gut wrenching so that we have a client that is based - # off of the slave store rather than the main store. - self.replication_handler = ReplicationCommandHandler(self.hs) - self.replication_handler._instance_name = "worker" - self.replication_handler._replication_data_handler = ReplicationDataHandler( - self.slaved_store - ) + self.reconnect() - client_factory = DirectTcpReplicationClientFactory( - self.hs, "client_name", self.replication_handler - ) - client_factory.handler = self.replication_handler - - server = server_factory.buildProtocol(None) - client = client_factory.buildProtocol(None) - - client.makeConnection(FakeTransport(server, reactor)) - - self.server_to_client_transport = FakeTransport(client, reactor) - server.makeConnection(self.server_to_client_transport) + self.master_store = hs.get_datastore() + self.slaved_store = self.worker_hs.get_datastore() + self.storage = hs.get_storage() def replicate(self): """Tell the master side of replication that something has happened, and then diff --git a/tests/replication/slave/storage/test_events.py b/tests/replication/slave/storage/test_events.py index f0561b30e3..0fee8a71c4 100644 --- a/tests/replication/slave/storage/test_events.py +++ b/tests/replication/slave/storage/test_events.py @@ -24,10 +24,10 @@ from synapse.storage.roommember import RoomsForUser from ._base import BaseSlavedStoreTestCase -USER_ID = "@feeling:blue" -USER_ID_2 = "@bright:blue" +USER_ID = "@feeling:test" +USER_ID_2 = "@bright:test" OUTLIER = {"outlier": True} -ROOM_ID = "!room:blue" +ROOM_ID = "!room:test" logger = logging.getLogger(__name__) @@ -239,7 +239,7 @@ class SlavedEventStoreTestCase(BaseSlavedStoreTestCase): self.check("get_rooms_for_user_with_stream_ordering", (USER_ID_2,), set()) # limit the replication rate - repl_transport = self.server_to_client_transport + repl_transport = self._server_transport repl_transport.autoflush = False # build the join and message events and persist them in the same batch. diff --git a/tests/replication/tcp/streams/_base.py b/tests/replication/tcp/streams/_base.py deleted file mode 100644 index 9d4f0bbe44..0000000000 --- a/tests/replication/tcp/streams/_base.py +++ /dev/null @@ -1,307 +0,0 @@ -# -*- coding: utf-8 -*- -# Copyright 2019 New Vector Ltd -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - -import logging -from typing import Any, List, Optional, Tuple - -import attr - -from twisted.internet.interfaces import IConsumer, IPullProducer, IReactorTime -from twisted.internet.task import LoopingCall -from twisted.web.http import HTTPChannel - -from synapse.app.generic_worker import ( - GenericWorkerReplicationHandler, - GenericWorkerServer, -) -from synapse.http.site import SynapseRequest -from synapse.replication.http import streams -from synapse.replication.tcp.handler import ReplicationCommandHandler -from synapse.replication.tcp.protocol import ClientReplicationStreamProtocol -from synapse.replication.tcp.resource import ReplicationStreamProtocolFactory -from synapse.server import HomeServer -from synapse.util import Clock - -from tests import unittest -from tests.server import FakeTransport - -logger = logging.getLogger(__name__) - - -class BaseStreamTestCase(unittest.HomeserverTestCase): - """Base class for tests of the replication streams""" - - servlets = [ - streams.register_servlets, - ] - - def prepare(self, reactor, clock, hs): - # build a replication server - server_factory = ReplicationStreamProtocolFactory(hs) - self.streamer = hs.get_replication_streamer() - self.server = server_factory.buildProtocol(None) - - # Make a new HomeServer object for the worker - self.reactor.lookups["testserv"] = "1.2.3.4" - self.worker_hs = self.setup_test_homeserver( - http_client=None, - homeserverToUse=GenericWorkerServer, - config=self._get_worker_hs_config(), - reactor=self.reactor, - ) - - # Since we use sqlite in memory databases we need to make sure the - # databases objects are the same. - self.worker_hs.get_datastore().db = hs.get_datastore().db - - self.test_handler = self._build_replication_data_handler() - self.worker_hs.replication_data_handler = self.test_handler - - repl_handler = ReplicationCommandHandler(self.worker_hs) - self.client = ClientReplicationStreamProtocol( - self.worker_hs, "client", "test", clock, repl_handler, - ) - - self._client_transport = None - self._server_transport = None - - def _get_worker_hs_config(self) -> dict: - config = self.default_config() - config["worker_app"] = "synapse.app.generic_worker" - config["worker_replication_host"] = "testserv" - config["worker_replication_http_port"] = "8765" - return config - - def _build_replication_data_handler(self): - return TestReplicationDataHandler(self.worker_hs) - - def reconnect(self): - if self._client_transport: - self.client.close() - - if self._server_transport: - self.server.close() - - self._client_transport = FakeTransport(self.server, self.reactor) - self.client.makeConnection(self._client_transport) - - self._server_transport = FakeTransport(self.client, self.reactor) - self.server.makeConnection(self._server_transport) - - def disconnect(self): - if self._client_transport: - self._client_transport = None - self.client.close() - - if self._server_transport: - self._server_transport = None - self.server.close() - - def replicate(self): - """Tell the master side of replication that something has happened, and then - wait for the replication to occur. - """ - self.streamer.on_notifier_poke() - self.pump(0.1) - - def handle_http_replication_attempt(self) -> SynapseRequest: - """Asserts that a connection attempt was made to the master HS on the - HTTP replication port, then proxies it to the master HS object to be - handled. - - Returns: - The request object received by master HS. - """ - - # We should have an outbound connection attempt. - clients = self.reactor.tcpClients - self.assertEqual(len(clients), 1) - (host, port, client_factory, _timeout, _bindAddress) = clients.pop(0) - self.assertEqual(host, "1.2.3.4") - self.assertEqual(port, 8765) - - # Set up client side protocol - client_protocol = client_factory.buildProtocol(None) - - request_factory = OneShotRequestFactory() - - # Set up the server side protocol - channel = _PushHTTPChannel(self.reactor) - channel.requestFactory = request_factory - channel.site = self.site - - # Connect client to server and vice versa. - client_to_server_transport = FakeTransport( - channel, self.reactor, client_protocol - ) - client_protocol.makeConnection(client_to_server_transport) - - server_to_client_transport = FakeTransport( - client_protocol, self.reactor, channel - ) - channel.makeConnection(server_to_client_transport) - - # The request will now be processed by `self.site` and the response - # streamed back. - self.reactor.advance(0) - - # We tear down the connection so it doesn't get reused without our - # knowledge. - server_to_client_transport.loseConnection() - client_to_server_transport.loseConnection() - - return request_factory.request - - def assert_request_is_get_repl_stream_updates( - self, request: SynapseRequest, stream_name: str - ): - """Asserts that the given request is a HTTP replication request for - fetching updates for given stream. - """ - - self.assertRegex( - request.path, - br"^/_synapse/replication/get_repl_stream_updates/%s/[^/]+$" - % (stream_name.encode("ascii"),), - ) - - self.assertEqual(request.method, b"GET") - - -class TestReplicationDataHandler(GenericWorkerReplicationHandler): - """Drop-in for ReplicationDataHandler which just collects RDATA rows""" - - def __init__(self, hs: HomeServer): - super().__init__(hs) - - # list of received (stream_name, token, row) tuples - self.received_rdata_rows = [] # type: List[Tuple[str, int, Any]] - - async def on_rdata(self, stream_name, instance_name, token, rows): - await super().on_rdata(stream_name, instance_name, token, rows) - for r in rows: - self.received_rdata_rows.append((stream_name, token, r)) - - -@attr.s() -class OneShotRequestFactory: - """A simple request factory that generates a single `SynapseRequest` and - stores it for future use. Can only be used once. - """ - - request = attr.ib(default=None) - - def __call__(self, *args, **kwargs): - assert self.request is None - - self.request = SynapseRequest(*args, **kwargs) - return self.request - - -class _PushHTTPChannel(HTTPChannel): - """A HTTPChannel that wraps pull producers to push producers. - - This is a hack to get around the fact that HTTPChannel transparently wraps a - pull producer (which is what Synapse uses to reply to requests) with - `_PullToPush` to convert it to a push producer. Unfortunately `_PullToPush` - uses the standard reactor rather than letting us use our test reactor, which - makes it very hard to test. - """ - - def __init__(self, reactor: IReactorTime): - super().__init__() - self.reactor = reactor - - self._pull_to_push_producer = None # type: Optional[_PullToPushProducer] - - def registerProducer(self, producer, streaming): - # Convert pull producers to push producer. - if not streaming: - self._pull_to_push_producer = _PullToPushProducer( - self.reactor, producer, self - ) - producer = self._pull_to_push_producer - - super().registerProducer(producer, True) - - def unregisterProducer(self): - if self._pull_to_push_producer: - # We need to manually stop the _PullToPushProducer. - self._pull_to_push_producer.stop() - - -class _PullToPushProducer: - """A push producer that wraps a pull producer. - """ - - def __init__( - self, reactor: IReactorTime, producer: IPullProducer, consumer: IConsumer - ): - self._clock = Clock(reactor) - self._producer = producer - self._consumer = consumer - - # While running we use a looping call with a zero delay to call - # resumeProducing on given producer. - self._looping_call = None # type: Optional[LoopingCall] - - # We start writing next reactor tick. - self._start_loop() - - def _start_loop(self): - """Start the looping call to - """ - - if not self._looping_call: - # Start a looping call which runs every tick. - self._looping_call = self._clock.looping_call(self._run_once, 0) - - def stop(self): - """Stops calling resumeProducing. - """ - if self._looping_call: - self._looping_call.stop() - self._looping_call = None - - def pauseProducing(self): - """Implements IPushProducer - """ - self.stop() - - def resumeProducing(self): - """Implements IPushProducer - """ - self._start_loop() - - def stopProducing(self): - """Implements IPushProducer - """ - self.stop() - self._producer.stopProducing() - - def _run_once(self): - """Calls resumeProducing on producer once. - """ - - try: - self._producer.resumeProducing() - except Exception: - logger.exception("Failed to call resumeProducing") - try: - self._consumer.unregisterProducer() - except Exception: - pass - - self.stopProducing() diff --git a/tests/replication/tcp/streams/test_events.py b/tests/replication/tcp/streams/test_events.py index 8bd67bb9f1..51bf0ef4e9 100644 --- a/tests/replication/tcp/streams/test_events.py +++ b/tests/replication/tcp/streams/test_events.py @@ -26,7 +26,7 @@ from synapse.replication.tcp.streams.events import ( from synapse.rest import admin from synapse.rest.client.v1 import login, room -from tests.replication.tcp.streams._base import BaseStreamTestCase +from tests.replication._base import BaseStreamTestCase from tests.test_utils.event_injection import inject_event, inject_member_event diff --git a/tests/replication/tcp/streams/test_federation.py b/tests/replication/tcp/streams/test_federation.py index eea4565da3..2babea4e3e 100644 --- a/tests/replication/tcp/streams/test_federation.py +++ b/tests/replication/tcp/streams/test_federation.py @@ -16,7 +16,7 @@ from synapse.federation.send_queue import EduRow from synapse.replication.tcp.streams.federation import FederationStream -from tests.replication.tcp.streams._base import BaseStreamTestCase +from tests.replication._base import BaseStreamTestCase class FederationStreamTestCase(BaseStreamTestCase): diff --git a/tests/replication/tcp/streams/test_receipts.py b/tests/replication/tcp/streams/test_receipts.py index 5853314fd4..56b062ecc1 100644 --- a/tests/replication/tcp/streams/test_receipts.py +++ b/tests/replication/tcp/streams/test_receipts.py @@ -19,7 +19,7 @@ from mock import Mock from synapse.replication.tcp.streams._base import ReceiptsStream -from tests.replication.tcp.streams._base import BaseStreamTestCase +from tests.replication._base import BaseStreamTestCase USER_ID = "@feeling:blue" diff --git a/tests/replication/tcp/streams/test_typing.py b/tests/replication/tcp/streams/test_typing.py index 125c63dab5..fd62b26356 100644 --- a/tests/replication/tcp/streams/test_typing.py +++ b/tests/replication/tcp/streams/test_typing.py @@ -17,7 +17,7 @@ from mock import Mock from synapse.handlers.typing import RoomMember from synapse.replication.tcp.streams import TypingStream -from tests.replication.tcp.streams._base import BaseStreamTestCase +from tests.replication._base import BaseStreamTestCase USER_ID = "@feeling:blue" -- cgit 1.4.1 From 5d64fefd6c7790dac0209c6c32cdb97cd6cd8820 Mon Sep 17 00:00:00 2001 From: Patrick Cloke Date: Wed, 13 May 2020 14:26:44 -0400 Subject: Do not validate that the client dict is stable during UI Auth. (#7483) This backs out some of the validation for the client dictionary and logs if this changes during a user interactive authentication session instead. --- changelog.d/7483.bugfix | 1 + synapse/handlers/auth.py | 37 +++++++++++---------- synapse/rest/client/v2_alpha/register.py | 1 - tests/rest/client/v2_alpha/test_auth.py | 55 ++++++-------------------------- 4 files changed, 29 insertions(+), 65 deletions(-) create mode 100644 changelog.d/7483.bugfix (limited to 'tests') diff --git a/changelog.d/7483.bugfix b/changelog.d/7483.bugfix new file mode 100644 index 0000000000..e1bc324617 --- /dev/null +++ b/changelog.d/7483.bugfix @@ -0,0 +1 @@ +Restore compatibility with non-compliant clients during the user interactive authentication process. diff --git a/synapse/handlers/auth.py b/synapse/handlers/auth.py index 9c71702371..5c20e29171 100644 --- a/synapse/handlers/auth.py +++ b/synapse/handlers/auth.py @@ -252,7 +252,6 @@ class AuthHandler(BaseHandler): clientdict: Dict[str, Any], clientip: str, description: str, - validate_clientdict: bool = True, ) -> Tuple[dict, dict, str]: """ Takes a dictionary sent by the client in the login / registration @@ -278,10 +277,6 @@ class AuthHandler(BaseHandler): description: A human readable string to be displayed to the user that describes the operation happening on their account. - validate_clientdict: Whether to validate that the operation happening - on the account has not changed. If this is false, - the client dict is persisted instead of validated. - Returns: A tuple of (creds, params, session_id). @@ -346,26 +341,30 @@ class AuthHandler(BaseHandler): # Ensure that the queried operation does not vary between stages of # the UI authentication session. This is done by generating a stable - # comparator based on the URI, method, and client dict (minus the - # auth dict) and storing it during the initial query. Subsequent + # comparator and storing it during the initial query. Subsequent # queries ensure that this comparator has not changed. - if validate_clientdict: - session_comparator = (session.uri, session.method, session.clientdict) - comparator = (uri, method, clientdict) - else: - session_comparator = (session.uri, session.method) # type: ignore - comparator = (uri, method) # type: ignore - - if session_comparator != comparator: + # + # The comparator is based on the requested URI and HTTP method. The + # client dict (minus the auth dict) should also be checked, but some + # clients are not spec compliant, just warn for now if the client + # dict changes. + if (session.uri, session.method) != (uri, method): raise SynapseError( 403, "Requested operation has changed during the UI authentication session.", ) - # For backwards compatibility the registration endpoint persists - # changes to the client dict instead of validating them. - if not validate_clientdict: - await self.store.set_ui_auth_clientdict(sid, clientdict) + if session.clientdict != clientdict: + logger.warning( + "Requested operation has changed during the UI " + "authentication session. A future version of Synapse " + "will remove this capability." + ) + + # For backwards compatibility, changes to the client dict are + # persisted as clients modify them throughout their user interactive + # authentication flow. + await self.store.set_ui_auth_clientdict(sid, clientdict) if not authdict: raise InteractiveAuthIncompleteError( diff --git a/synapse/rest/client/v2_alpha/register.py b/synapse/rest/client/v2_alpha/register.py index e77dd6bf92..af08cc6cce 100644 --- a/synapse/rest/client/v2_alpha/register.py +++ b/synapse/rest/client/v2_alpha/register.py @@ -516,7 +516,6 @@ class RegisterRestServlet(RestServlet): body, self.hs.get_ip_from_request(request), "register a new account", - validate_clientdict=False, ) # Check that we're not trying to register a denied 3pid. diff --git a/tests/rest/client/v2_alpha/test_auth.py b/tests/rest/client/v2_alpha/test_auth.py index a56c50a5b7..293ccfba2b 100644 --- a/tests/rest/client/v2_alpha/test_auth.py +++ b/tests/rest/client/v2_alpha/test_auth.py @@ -133,47 +133,6 @@ class FallbackAuthTests(unittest.HomeserverTestCase): # We're given a registered user. self.assertEqual(channel.json_body["user_id"], "@user:test") - def test_legacy_registration(self): - """ - Registration allows the parameters to vary through the process. - """ - - # Make the initial request to register. (Later on a different password - # will be used.) - # Returns a 401 as per the spec - channel = self.register( - 401, {"username": "user", "type": "m.login.password", "password": "bar"}, - ) - - # Grab the session - session = channel.json_body["session"] - # Assert our configured public key is being given - self.assertEqual( - channel.json_body["params"]["m.login.recaptcha"]["public_key"], "brokencake" - ) - - # Complete the recaptcha step. - self.recaptcha(session, 200) - - # also complete the dummy auth - self.register(200, {"auth": {"session": session, "type": "m.login.dummy"}}) - - # Now we should have fulfilled a complete auth flow, including - # the recaptcha fallback step. Make the initial request again, but - # with a changed password. This still completes. - channel = self.register( - 200, - { - "username": "user", - "type": "m.login.password", - "password": "foo", # Note that this is different. - "auth": {"session": session}, - }, - ) - - # We're given a registered user. - self.assertEqual(channel.json_body["user_id"], "@user:test") - def test_complete_operation_unknown_session(self): """ Attempting to mark an invalid session as complete should error. @@ -282,9 +241,15 @@ class UIAuthTests(unittest.HomeserverTestCase): }, ) - def test_cannot_change_body(self): + def test_can_change_body(self): """ - The initial requested client dict cannot be modified during the user interactive authentication session. + The client dict can be modified during the user interactive authentication session. + + Note that it is not spec compliant to modify the client dict during a + user interactive authentication session, but many clients currently do. + + When Synapse is updated to be spec compliant, the call to re-use the + session ID should be rejected. """ # Create a second login. self.login("test", self.user_pass) @@ -302,9 +267,9 @@ class UIAuthTests(unittest.HomeserverTestCase): self.assertIn({"stages": ["m.login.password"]}, channel.json_body["flows"]) # Make another request providing the UI auth flow, but try to delete the - # second device. This results in an error. + # second device. self.delete_devices( - 403, + 200, { "devices": [device_ids[1]], "auth": { -- cgit 1.4.1 From a0e063387d5c8390b19cf69359a90924ecd4fbda Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Thu, 14 May 2020 10:07:54 +0100 Subject: Stop `get_joined_users` corruption from custom statuses (#7376) Fix a bug where the `get_joined_users` cache could be corrupted by custom status events (or other state events with a state_key matching the user ID). The bug was introduced by #2229, but has largely gone unnoticed since then. Fixes #7099, #7373. --- changelog.d/7376.bugfix | 1 + synapse/storage/data_stores/main/roommember.py | 3 +- tests/storage/test_roommember.py | 50 +++++++++++++++++++++++++- 3 files changed, 52 insertions(+), 2 deletions(-) create mode 100644 changelog.d/7376.bugfix (limited to 'tests') diff --git a/changelog.d/7376.bugfix b/changelog.d/7376.bugfix new file mode 100644 index 0000000000..b7b67e328e --- /dev/null +++ b/changelog.d/7376.bugfix @@ -0,0 +1 @@ +Fix a bug which could cause messages not to be sent over federation, when state events with state keys matching user IDs (such as custom user statuses) were received. diff --git a/synapse/storage/data_stores/main/roommember.py b/synapse/storage/data_stores/main/roommember.py index d5bd0cb5cf..e626b7f6f7 100644 --- a/synapse/storage/data_stores/main/roommember.py +++ b/synapse/storage/data_stores/main/roommember.py @@ -576,7 +576,8 @@ class RoomMemberWorkerStore(EventsWorkerStore): if key[0] == EventTypes.Member ] for etype, state_key in context.delta_ids: - users_in_room.pop(state_key, None) + if etype == EventTypes.Member: + users_in_room.pop(state_key, None) # We check if we have any of the member event ids in the event cache # before we ask the DB diff --git a/tests/storage/test_roommember.py b/tests/storage/test_roommember.py index 00df0ea68e..5dd46005e6 100644 --- a/tests/storage/test_roommember.py +++ b/tests/storage/test_roommember.py @@ -22,6 +22,8 @@ from synapse.rest.client.v1 import login, room from synapse.types import Requester, UserID from tests import unittest +from tests.test_utils import event_injection +from tests.utils import TestHomeServer class RoomMemberStoreTestCase(unittest.HomeserverTestCase): @@ -38,7 +40,7 @@ class RoomMemberStoreTestCase(unittest.HomeserverTestCase): ) return hs - def prepare(self, reactor, clock, hs): + def prepare(self, reactor, clock, hs: TestHomeServer): # We can't test the RoomMemberStore on its own without the other event # storage logic @@ -114,6 +116,52 @@ class RoomMemberStoreTestCase(unittest.HomeserverTestCase): # It now knows about Charlie's server. self.assertEqual(self.store._known_servers_count, 2) + def test_get_joined_users_from_context(self): + room = self.helper.create_room_as(self.u_alice, tok=self.t_alice) + bob_event = event_injection.inject_member_event( + self.hs, room, self.u_bob, Membership.JOIN + ) + + # first, create a regular event + event, context = event_injection.create_event( + self.hs, + room_id=room, + sender=self.u_alice, + prev_event_ids=[bob_event.event_id], + type="m.test.1", + content={}, + ) + + users = self.get_success( + self.store.get_joined_users_from_context(event, context) + ) + self.assertEqual(users.keys(), {self.u_alice, self.u_bob}) + + # Regression test for #7376: create a state event whose key matches bob's + # user_id, but which is *not* a membership event, and persist that; then check + # that `get_joined_users_from_context` returns the correct users for the next event. + non_member_event = event_injection.inject_event( + self.hs, + room_id=room, + sender=self.u_bob, + prev_event_ids=[bob_event.event_id], + type="m.test.2", + state_key=self.u_bob, + content={}, + ) + event, context = event_injection.create_event( + self.hs, + room_id=room, + sender=self.u_alice, + prev_event_ids=[non_member_event.event_id], + type="m.test.3", + content={}, + ) + users = self.get_success( + self.store.get_joined_users_from_context(event, context) + ) + self.assertEqual(users.keys(), {self.u_alice, self.u_bob}) + class CurrentStateMembershipUpdateTestCase(unittest.HomeserverTestCase): def prepare(self, reactor, clock, homeserver): -- cgit 1.4.1 From 225c16508705ecfdde44e3c90060609fab020e32 Mon Sep 17 00:00:00 2001 From: Andrew Morgan <1342360+anoadragon453@users.noreply.github.com> Date: Thu, 14 May 2020 16:32:49 +0100 Subject: Allow expired accounts to logout (#7443) --- changelog.d/7443.bugfix | 1 + synapse/api/auth.py | 50 ++++++++++++++------- synapse/rest/client/v1/logout.py | 6 +-- tests/rest/client/v1/test_login.py | 69 ++++++++++++++++++++++++++++- tests/rest/client/v2_alpha/test_register.py | 36 ++++++++++++++- 5 files changed, 140 insertions(+), 22 deletions(-) create mode 100644 changelog.d/7443.bugfix (limited to 'tests') diff --git a/changelog.d/7443.bugfix b/changelog.d/7443.bugfix new file mode 100644 index 0000000000..1dddec59ed --- /dev/null +++ b/changelog.d/7443.bugfix @@ -0,0 +1 @@ +Allow expired user accounts to log out their device sessions. \ No newline at end of file diff --git a/synapse/api/auth.py b/synapse/api/auth.py index e009b1a760..3c660318fc 100644 --- a/synapse/api/auth.py +++ b/synapse/api/auth.py @@ -22,6 +22,7 @@ import pymacaroons from netaddr import IPAddress from twisted.internet import defer +from twisted.web.server import Request import synapse.logging.opentracing as opentracing import synapse.types @@ -162,19 +163,25 @@ class Auth(object): @defer.inlineCallbacks def get_user_by_req( - self, request, allow_guest=False, rights="access", allow_expired=False + self, + request: Request, + allow_guest: bool = False, + rights: str = "access", + allow_expired: bool = False, ): """ Get a registered user's ID. Args: - request - An HTTP request with an access_token query parameter. - allow_expired - Whether to allow the request through even if the account is - expired. If true, Synapse will still require an access token to be - provided but won't check if the account it belongs to has expired. This - works thanks to /login delivering access tokens regardless of accounts' - expiration. + request: An HTTP request with an access_token query parameter. + allow_guest: If False, will raise an AuthError if the user making the + request is a guest. + rights: The operation being performed; the access token must allow this + allow_expired: If True, allow the request through even if the account + is expired, or session token lifetime has ended. Note that + /login will deliver access tokens regardless of expiration. + Returns: - defer.Deferred: resolves to a ``synapse.types.Requester`` object + defer.Deferred: resolves to a `synapse.types.Requester` object Raises: InvalidClientCredentialsError if no user by that token exists or the token is invalid. @@ -205,7 +212,9 @@ class Auth(object): return synapse.types.create_requester(user_id, app_service=app_service) - user_info = yield self.get_user_by_access_token(access_token, rights) + user_info = yield self.get_user_by_access_token( + access_token, rights, allow_expired=allow_expired + ) user = user_info["user"] token_id = user_info["token_id"] is_guest = user_info["is_guest"] @@ -280,13 +289,17 @@ class Auth(object): return user_id, app_service @defer.inlineCallbacks - def get_user_by_access_token(self, token, rights="access"): + def get_user_by_access_token( + self, token: str, rights: str = "access", allow_expired: bool = False, + ): """ Validate access token and get user_id from it Args: - token (str): The access token to get the user by. - rights (str): The operation being performed; the access token must - allow this. + token: The access token to get the user by + rights: The operation being performed; the access token must + allow this + allow_expired: If False, raises an InvalidClientTokenError + if the token is expired Returns: Deferred[dict]: dict that includes: `user` (UserID) @@ -294,8 +307,10 @@ class Auth(object): `token_id` (int|None): access token id. May be None if guest `device_id` (str|None): device corresponding to access token Raises: + InvalidClientTokenError if a user by that token exists, but the token is + expired InvalidClientCredentialsError if no user by that token exists or the token - is invalid. + is invalid """ if rights == "access": @@ -304,7 +319,8 @@ class Auth(object): if r: valid_until_ms = r["valid_until_ms"] if ( - valid_until_ms is not None + not allow_expired + and valid_until_ms is not None and valid_until_ms < self.clock.time_msec() ): # there was a valid access token, but it has expired. @@ -575,7 +591,7 @@ class Auth(object): return user_level >= send_level @staticmethod - def has_access_token(request): + def has_access_token(request: Request): """Checks if the request has an access_token. Returns: @@ -586,7 +602,7 @@ class Auth(object): return bool(query_params) or bool(auth_headers) @staticmethod - def get_access_token_from_request(request): + def get_access_token_from_request(request: Request): """Extracts the access_token from the request. Args: diff --git a/synapse/rest/client/v1/logout.py b/synapse/rest/client/v1/logout.py index 1cf3caf832..b0c30b65be 100644 --- a/synapse/rest/client/v1/logout.py +++ b/synapse/rest/client/v1/logout.py @@ -34,10 +34,10 @@ class LogoutRestServlet(RestServlet): return 200, {} async def on_POST(self, request): - requester = await self.auth.get_user_by_req(request) + requester = await self.auth.get_user_by_req(request, allow_expired=True) if requester.device_id is None: - # the acccess token wasn't associated with a device. + # The access token wasn't associated with a device. # Just delete the access token access_token = self.auth.get_access_token_from_request(request) await self._auth_handler.delete_access_token(access_token) @@ -62,7 +62,7 @@ class LogoutAllRestServlet(RestServlet): return 200, {} async def on_POST(self, request): - requester = await self.auth.get_user_by_req(request) + requester = await self.auth.get_user_by_req(request, allow_expired=True) user_id = requester.user.to_string() # first delete all of the user's devices diff --git a/tests/rest/client/v1/test_login.py b/tests/rest/client/v1/test_login.py index 1856c7ffd5..eb8f6264fd 100644 --- a/tests/rest/client/v1/test_login.py +++ b/tests/rest/client/v1/test_login.py @@ -4,7 +4,7 @@ import urllib.parse from mock import Mock import synapse.rest.admin -from synapse.rest.client.v1 import login +from synapse.rest.client.v1 import login, logout from synapse.rest.client.v2_alpha import devices from synapse.rest.client.v2_alpha.account import WhoamiRestServlet @@ -20,6 +20,7 @@ class LoginRestServletTestCase(unittest.HomeserverTestCase): servlets = [ synapse.rest.admin.register_servlets_for_client_rest_resource, login.register_servlets, + logout.register_servlets, devices.register_servlets, lambda hs, http_server: WhoamiRestServlet(hs).register(http_server), ] @@ -256,6 +257,72 @@ class LoginRestServletTestCase(unittest.HomeserverTestCase): self.render(request) self.assertEquals(channel.code, 200, channel.result) + @override_config({"session_lifetime": "24h"}) + def test_session_can_hard_logout_after_being_soft_logged_out(self): + self.register_user("kermit", "monkey") + + # log in as normal + access_token = self.login("kermit", "monkey") + + # we should now be able to make requests with the access token + request, channel = self.make_request( + b"GET", TEST_URL, access_token=access_token + ) + self.render(request) + self.assertEquals(channel.code, 200, channel.result) + + # time passes + self.reactor.advance(24 * 3600) + + # ... and we should be soft-logouted + request, channel = self.make_request( + b"GET", TEST_URL, access_token=access_token + ) + self.render(request) + self.assertEquals(channel.code, 401, channel.result) + self.assertEquals(channel.json_body["errcode"], "M_UNKNOWN_TOKEN") + self.assertEquals(channel.json_body["soft_logout"], True) + + # Now try to hard logout this session + request, channel = self.make_request( + b"POST", "/logout", access_token=access_token + ) + self.render(request) + self.assertEquals(channel.result["code"], b"200", channel.result) + + @override_config({"session_lifetime": "24h"}) + def test_session_can_hard_logout_all_sessions_after_being_soft_logged_out(self): + self.register_user("kermit", "monkey") + + # log in as normal + access_token = self.login("kermit", "monkey") + + # we should now be able to make requests with the access token + request, channel = self.make_request( + b"GET", TEST_URL, access_token=access_token + ) + self.render(request) + self.assertEquals(channel.code, 200, channel.result) + + # time passes + self.reactor.advance(24 * 3600) + + # ... and we should be soft-logouted + request, channel = self.make_request( + b"GET", TEST_URL, access_token=access_token + ) + self.render(request) + self.assertEquals(channel.code, 401, channel.result) + self.assertEquals(channel.json_body["errcode"], "M_UNKNOWN_TOKEN") + self.assertEquals(channel.json_body["soft_logout"], True) + + # Now try to hard log out all of the user's sessions + request, channel = self.make_request( + b"POST", "/logout/all", access_token=access_token + ) + self.render(request) + self.assertEquals(channel.result["code"], b"200", channel.result) + class CASTestCase(unittest.HomeserverTestCase): diff --git a/tests/rest/client/v2_alpha/test_register.py b/tests/rest/client/v2_alpha/test_register.py index a68a96f618..5637ce2090 100644 --- a/tests/rest/client/v2_alpha/test_register.py +++ b/tests/rest/client/v2_alpha/test_register.py @@ -25,7 +25,7 @@ import synapse.rest.admin from synapse.api.constants import LoginType from synapse.api.errors import Codes from synapse.appservice import ApplicationService -from synapse.rest.client.v1 import login +from synapse.rest.client.v1 import login, logout from synapse.rest.client.v2_alpha import account, account_validity, register, sync from tests import unittest @@ -313,6 +313,7 @@ class AccountValidityTestCase(unittest.HomeserverTestCase): synapse.rest.admin.register_servlets_for_client_rest_resource, login.register_servlets, sync.register_servlets, + logout.register_servlets, account_validity.register_servlets, ] @@ -405,6 +406,39 @@ class AccountValidityTestCase(unittest.HomeserverTestCase): channel.json_body["errcode"], Codes.EXPIRED_ACCOUNT, channel.result ) + def test_logging_out_expired_user(self): + user_id = self.register_user("kermit", "monkey") + tok = self.login("kermit", "monkey") + + self.register_user("admin", "adminpassword", admin=True) + admin_tok = self.login("admin", "adminpassword") + + url = "/_matrix/client/unstable/admin/account_validity/validity" + params = { + "user_id": user_id, + "expiration_ts": 0, + "enable_renewal_emails": False, + } + request_data = json.dumps(params) + request, channel = self.make_request( + b"POST", url, request_data, access_token=admin_tok + ) + self.render(request) + self.assertEquals(channel.result["code"], b"200", channel.result) + + # Try to log the user out + request, channel = self.make_request(b"POST", "/logout", access_token=tok) + self.render(request) + self.assertEquals(channel.result["code"], b"200", channel.result) + + # Log the user in again (allowed for expired accounts) + tok = self.login("kermit", "monkey") + + # Try to log out all of the user's sessions + request, channel = self.make_request(b"POST", "/logout/all", access_token=tok) + self.render(request) + self.assertEquals(channel.result["code"], b"200", channel.result) + class AccountValidityRenewalByEmailTestCase(unittest.HomeserverTestCase): -- cgit 1.4.1 From fef3ff5cc42e91cc31d61094c4db2638532a53a4 Mon Sep 17 00:00:00 2001 From: Patrick Cloke Date: Thu, 14 May 2020 12:38:17 -0400 Subject: Enforce MSC2209: auth rules for notifications in power level event (#7502) In a new room version, the "notifications" key of power level events are subject to restricted auth rules. --- changelog.d/7502.feature | 1 + synapse/api/room_versions.py | 22 +++++++++++++++++++++- synapse/event_auth.py | 12 ++++++++++-- tests/test_event_auth.py | 33 +++++++++++++++++++++++++++++++++ 4 files changed, 65 insertions(+), 3 deletions(-) create mode 100644 changelog.d/7502.feature (limited to 'tests') diff --git a/changelog.d/7502.feature b/changelog.d/7502.feature new file mode 100644 index 0000000000..059a05bfca --- /dev/null +++ b/changelog.d/7502.feature @@ -0,0 +1 @@ +Add additional authentication checks for m.room.power_levels event per [MSC2209](https://github.com/matrix-org/matrix-doc/pull/2209). diff --git a/synapse/api/room_versions.py b/synapse/api/room_versions.py index 871179749a..af3612ed61 100644 --- a/synapse/api/room_versions.py +++ b/synapse/api/room_versions.py @@ -58,7 +58,11 @@ class RoomVersion(object): enforce_key_validity = attr.ib() # bool # bool: before MSC2261/MSC2432, m.room.aliases had special auth rules and redaction rules - special_case_aliases_auth = attr.ib(type=bool, default=False) + special_case_aliases_auth = attr.ib(type=bool) + + # bool: MSC2209: Check 'notifications' key while verifying + # m.room.power_levels auth rules. + limit_notifications_power_levels = attr.ib(type=bool) class RoomVersions(object): @@ -69,6 +73,7 @@ class RoomVersions(object): StateResolutionVersions.V1, enforce_key_validity=False, special_case_aliases_auth=True, + limit_notifications_power_levels=False, ) V2 = RoomVersion( "2", @@ -77,6 +82,7 @@ class RoomVersions(object): StateResolutionVersions.V2, enforce_key_validity=False, special_case_aliases_auth=True, + limit_notifications_power_levels=False, ) V3 = RoomVersion( "3", @@ -85,6 +91,7 @@ class RoomVersions(object): StateResolutionVersions.V2, enforce_key_validity=False, special_case_aliases_auth=True, + limit_notifications_power_levels=False, ) V4 = RoomVersion( "4", @@ -93,6 +100,7 @@ class RoomVersions(object): StateResolutionVersions.V2, enforce_key_validity=False, special_case_aliases_auth=True, + limit_notifications_power_levels=False, ) V5 = RoomVersion( "5", @@ -101,6 +109,7 @@ class RoomVersions(object): StateResolutionVersions.V2, enforce_key_validity=True, special_case_aliases_auth=True, + limit_notifications_power_levels=False, ) MSC2432_DEV = RoomVersion( "org.matrix.msc2432", @@ -109,6 +118,16 @@ class RoomVersions(object): StateResolutionVersions.V2, enforce_key_validity=True, special_case_aliases_auth=False, + limit_notifications_power_levels=False, + ) + MSC2209_DEV = RoomVersion( + "org.matrix.msc2209", + RoomDisposition.UNSTABLE, + EventFormatVersions.V3, + StateResolutionVersions.V2, + enforce_key_validity=True, + special_case_aliases_auth=True, + limit_notifications_power_levels=True, ) @@ -121,5 +140,6 @@ KNOWN_ROOM_VERSIONS = { RoomVersions.V4, RoomVersions.V5, RoomVersions.MSC2432_DEV, + RoomVersions.MSC2209_DEV, ) } # type: Dict[str, RoomVersion] diff --git a/synapse/event_auth.py b/synapse/event_auth.py index 46beb5334f..5a5b568a95 100644 --- a/synapse/event_auth.py +++ b/synapse/event_auth.py @@ -181,7 +181,7 @@ def check( _can_send_event(event, auth_events) if event.type == EventTypes.PowerLevels: - _check_power_levels(event, auth_events) + _check_power_levels(room_version_obj, event, auth_events) if event.type == EventTypes.Redaction: check_redaction(room_version_obj, event, auth_events) @@ -442,7 +442,7 @@ def check_redaction(room_version_obj: RoomVersion, event, auth_events): raise AuthError(403, "You don't have permission to redact events") -def _check_power_levels(event, auth_events): +def _check_power_levels(room_version_obj, event, auth_events): user_list = event.content.get("users", {}) # Validate users for k, v in user_list.items(): @@ -484,6 +484,14 @@ def _check_power_levels(event, auth_events): for ev_id in set(list(old_list) + list(new_list)): levels_to_check.append((ev_id, "events")) + # MSC2209 specifies these checks should also be done for the "notifications" + # key. + if room_version_obj.limit_notifications_power_levels: + old_list = current_state.content.get("notifications", {}) + new_list = event.content.get("notifications", {}) + for ev_id in set(list(old_list) + list(new_list)): + levels_to_check.append((ev_id, "notifications")) + old_state = current_state.content new_state = event.content diff --git a/tests/test_event_auth.py b/tests/test_event_auth.py index 6c2351cf55..f2def601fb 100644 --- a/tests/test_event_auth.py +++ b/tests/test_event_auth.py @@ -165,6 +165,39 @@ class EventAuthTestCase(unittest.TestCase): do_sig_check=False, ) + def test_msc2209(self): + """ + Notifications power levels get checked due to MSC2209. + """ + creator = "@creator:example.com" + pleb = "@joiner:example.com" + + auth_events = { + ("m.room.create", ""): _create_event(creator), + ("m.room.member", creator): _join_event(creator), + ("m.room.power_levels", ""): _power_levels_event( + creator, {"state_default": "30", "users": {pleb: "30"}} + ), + ("m.room.member", pleb): _join_event(pleb), + } + + # pleb should be able to modify the notifications power level. + event_auth.check( + RoomVersions.V1, + _power_levels_event(pleb, {"notifications": {"room": 100}}), + auth_events, + do_sig_check=False, + ) + + # But an MSC2209 room rejects this change. + with self.assertRaises(AuthError): + event_auth.check( + RoomVersions.MSC2209_DEV, + _power_levels_event(pleb, {"notifications": {"room": 100}}), + auth_events, + do_sig_check=False, + ) + # helpers for making events -- cgit 1.4.1 From 56b66db78a3a6a22f65b219f4dc12899111f742b Mon Sep 17 00:00:00 2001 From: Patrick Cloke Date: Thu, 14 May 2020 13:24:01 -0400 Subject: Strictly enforce canonicaljson requirements in a new room version (#7381) --- changelog.d/7381.bugfix | 1 + synapse/api/room_versions.py | 24 ++++++++++++- synapse/events/utils.py | 35 +++++++++++++++++- synapse/events/validator.py | 7 ++++ synapse/federation/federation_base.py | 6 +++- synapse/util/frozenutils.py | 2 +- tests/handlers/test_federation.py | 67 ++++++++++++++++++++++++++++++++++- 7 files changed, 137 insertions(+), 5 deletions(-) create mode 100644 changelog.d/7381.bugfix (limited to 'tests') diff --git a/changelog.d/7381.bugfix b/changelog.d/7381.bugfix new file mode 100644 index 0000000000..e5f93571dc --- /dev/null +++ b/changelog.d/7381.bugfix @@ -0,0 +1 @@ +Add an experimental room version which strictly adheres to the canonical JSON specification. diff --git a/synapse/api/room_versions.py b/synapse/api/room_versions.py index af3612ed61..0901afb900 100644 --- a/synapse/api/room_versions.py +++ b/synapse/api/room_versions.py @@ -59,7 +59,11 @@ class RoomVersion(object): # bool: before MSC2261/MSC2432, m.room.aliases had special auth rules and redaction rules special_case_aliases_auth = attr.ib(type=bool) - + # Strictly enforce canonicaljson, do not allow: + # * Integers outside the range of [-2 ^ 53 + 1, 2 ^ 53 - 1] + # * Floats + # * NaN, Infinity, -Infinity + strict_canonicaljson = attr.ib(type=bool) # bool: MSC2209: Check 'notifications' key while verifying # m.room.power_levels auth rules. limit_notifications_power_levels = attr.ib(type=bool) @@ -73,6 +77,7 @@ class RoomVersions(object): StateResolutionVersions.V1, enforce_key_validity=False, special_case_aliases_auth=True, + strict_canonicaljson=False, limit_notifications_power_levels=False, ) V2 = RoomVersion( @@ -82,6 +87,7 @@ class RoomVersions(object): StateResolutionVersions.V2, enforce_key_validity=False, special_case_aliases_auth=True, + strict_canonicaljson=False, limit_notifications_power_levels=False, ) V3 = RoomVersion( @@ -91,6 +97,7 @@ class RoomVersions(object): StateResolutionVersions.V2, enforce_key_validity=False, special_case_aliases_auth=True, + strict_canonicaljson=False, limit_notifications_power_levels=False, ) V4 = RoomVersion( @@ -100,6 +107,7 @@ class RoomVersions(object): StateResolutionVersions.V2, enforce_key_validity=False, special_case_aliases_auth=True, + strict_canonicaljson=False, limit_notifications_power_levels=False, ) V5 = RoomVersion( @@ -109,6 +117,7 @@ class RoomVersions(object): StateResolutionVersions.V2, enforce_key_validity=True, special_case_aliases_auth=True, + strict_canonicaljson=False, limit_notifications_power_levels=False, ) MSC2432_DEV = RoomVersion( @@ -118,6 +127,17 @@ class RoomVersions(object): StateResolutionVersions.V2, enforce_key_validity=True, special_case_aliases_auth=False, + strict_canonicaljson=False, + limit_notifications_power_levels=False, + ) + STRICT_CANONICALJSON = RoomVersion( + "org.matrix.strict_canonicaljson", + RoomDisposition.UNSTABLE, + EventFormatVersions.V3, + StateResolutionVersions.V2, + enforce_key_validity=True, + special_case_aliases_auth=True, + strict_canonicaljson=True, limit_notifications_power_levels=False, ) MSC2209_DEV = RoomVersion( @@ -127,6 +147,7 @@ class RoomVersions(object): StateResolutionVersions.V2, enforce_key_validity=True, special_case_aliases_auth=True, + strict_canonicaljson=False, limit_notifications_power_levels=True, ) @@ -140,6 +161,7 @@ KNOWN_ROOM_VERSIONS = { RoomVersions.V4, RoomVersions.V5, RoomVersions.MSC2432_DEV, + RoomVersions.STRICT_CANONICALJSON, RoomVersions.MSC2209_DEV, ) } # type: Dict[str, RoomVersion] diff --git a/synapse/events/utils.py b/synapse/events/utils.py index b75b097e5e..dd340be9a7 100644 --- a/synapse/events/utils.py +++ b/synapse/events/utils.py @@ -14,7 +14,7 @@ # limitations under the License. import collections import re -from typing import Mapping, Union +from typing import Any, Mapping, Union from six import string_types @@ -23,6 +23,7 @@ from frozendict import frozendict from twisted.internet import defer from synapse.api.constants import EventTypes, RelationTypes +from synapse.api.errors import Codes, SynapseError from synapse.api.room_versions import RoomVersion from synapse.util.async_helpers import yieldable_gather_results @@ -449,3 +450,35 @@ def copy_power_levels_contents( raise TypeError("Invalid power_levels value for %s: %r" % (k, v)) return power_levels + + +def validate_canonicaljson(value: Any): + """ + Ensure that the JSON object is valid according to the rules of canonical JSON. + + See the appendix section 3.1: Canonical JSON. + + This rejects JSON that has: + * An integer outside the range of [-2 ^ 53 + 1, 2 ^ 53 - 1] + * Floats + * NaN, Infinity, -Infinity + """ + if isinstance(value, int): + if value <= -(2 ** 53) or 2 ** 53 <= value: + raise SynapseError(400, "JSON integer out of range", Codes.BAD_JSON) + + elif isinstance(value, float): + # Note that Infinity, -Infinity, and NaN are also considered floats. + raise SynapseError(400, "Bad JSON value: float", Codes.BAD_JSON) + + elif isinstance(value, (dict, frozendict)): + for v in value.values(): + validate_canonicaljson(v) + + elif isinstance(value, (list, tuple)): + for i in value: + validate_canonicaljson(i) + + elif not isinstance(value, (bool, str)) and value is not None: + # Other potential JSON values (bool, None, str) are safe. + raise SynapseError(400, "Unknown JSON value", Codes.BAD_JSON) diff --git a/synapse/events/validator.py b/synapse/events/validator.py index 9b90c9ce04..b001c64bb4 100644 --- a/synapse/events/validator.py +++ b/synapse/events/validator.py @@ -18,6 +18,7 @@ from six import integer_types, string_types from synapse.api.constants import MAX_ALIAS_LENGTH, EventTypes, Membership from synapse.api.errors import Codes, SynapseError from synapse.api.room_versions import EventFormatVersions +from synapse.events.utils import validate_canonicaljson from synapse.types import EventID, RoomID, UserID @@ -55,6 +56,12 @@ class EventValidator(object): if not isinstance(getattr(event, s), string_types): raise SynapseError(400, "'%s' not a string type" % (s,)) + # Depending on the room version, ensure the data is spec compliant JSON. + if event.room_version.strict_canonicaljson: + # Note that only the client controlled portion of the event is + # checked, since we trust the portions of the event we created. + validate_canonicaljson(event.content) + if event.type == EventTypes.Aliases: if "aliases" in event.content: for alias in event.content["aliases"]: diff --git a/synapse/federation/federation_base.py b/synapse/federation/federation_base.py index 4b115aac04..c0012c6872 100644 --- a/synapse/federation/federation_base.py +++ b/synapse/federation/federation_base.py @@ -29,7 +29,7 @@ from synapse.api.room_versions import EventFormatVersions, RoomVersion from synapse.crypto.event_signing import check_event_content_hash from synapse.crypto.keyring import Keyring from synapse.events import EventBase, make_event_from_dict -from synapse.events.utils import prune_event +from synapse.events.utils import prune_event, validate_canonicaljson from synapse.http.servlet import assert_params_in_dict from synapse.logging.context import ( PreserveLoggingContext, @@ -302,6 +302,10 @@ def event_from_pdu_json( elif depth > MAX_DEPTH: raise SynapseError(400, "Depth too large", Codes.BAD_JSON) + # Validate that the JSON conforms to the specification. + if room_version.strict_canonicaljson: + validate_canonicaljson(pdu_json) + event = make_event_from_dict(pdu_json, room_version) event.internal_metadata.outlier = outlier diff --git a/synapse/util/frozenutils.py b/synapse/util/frozenutils.py index f2ccd5e7c6..9815bb8667 100644 --- a/synapse/util/frozenutils.py +++ b/synapse/util/frozenutils.py @@ -65,5 +65,5 @@ def _handle_frozendict(obj): ) -# A JSONEncoder which is capable of encoding frozendics without barfing +# A JSONEncoder which is capable of encoding frozendicts without barfing frozendict_json_encoder = json.JSONEncoder(default=_handle_frozendict) diff --git a/tests/handlers/test_federation.py b/tests/handlers/test_federation.py index 132e35651d..dfef58e704 100644 --- a/tests/handlers/test_federation.py +++ b/tests/handlers/test_federation.py @@ -13,9 +13,12 @@ # See the License for the specific language governing permissions and # limitations under the License. import logging +from unittest import TestCase from synapse.api.constants import EventTypes -from synapse.api.errors import AuthError, Codes +from synapse.api.errors import AuthError, Codes, SynapseError +from synapse.api.room_versions import RoomVersions +from synapse.events import EventBase from synapse.federation.federation_base import event_from_pdu_json from synapse.logging.context import LoggingContext, run_in_background from synapse.rest import admin @@ -207,3 +210,65 @@ class FederationTestCase(unittest.HomeserverTestCase): self.assertEqual(r[(EventTypes.Member, other_user)], join_event.event_id) return join_event + + +class EventFromPduTestCase(TestCase): + def test_valid_json(self): + """Valid JSON should be turned into an event.""" + ev = event_from_pdu_json( + { + "type": EventTypes.Message, + "content": {"bool": True, "null": None, "int": 1, "str": "foobar"}, + "room_id": "!room:test", + "sender": "@user:test", + "depth": 1, + "prev_events": [], + "auth_events": [], + "origin_server_ts": 1234, + }, + RoomVersions.STRICT_CANONICALJSON, + ) + + self.assertIsInstance(ev, EventBase) + + def test_invalid_numbers(self): + """Invalid values for an integer should be rejected, all floats should be rejected.""" + for value in [ + -(2 ** 53), + 2 ** 53, + 1.0, + float("inf"), + float("-inf"), + float("nan"), + ]: + with self.assertRaises(SynapseError): + event_from_pdu_json( + { + "type": EventTypes.Message, + "content": {"foo": value}, + "room_id": "!room:test", + "sender": "@user:test", + "depth": 1, + "prev_events": [], + "auth_events": [], + "origin_server_ts": 1234, + }, + RoomVersions.STRICT_CANONICALJSON, + ) + + def test_invalid_nested(self): + """List and dictionaries are recursively searched.""" + with self.assertRaises(SynapseError): + event_from_pdu_json( + { + "type": EventTypes.Message, + "content": {"foo": [{"bar": 2 ** 56}]}, + "room_id": "!room:test", + "sender": "@user:test", + "depth": 1, + "prev_events": [], + "auth_events": [], + "origin_server_ts": 1234, + }, + RoomVersions.STRICT_CANONICALJSON, + ) -- cgit 1.4.1 From 08bc80ef096497ca8197657c98a3a710623b3135 Mon Sep 17 00:00:00 2001 From: Patrick Cloke Date: Fri, 15 May 2020 09:30:10 -0400 Subject: Implement room version 6 (MSC2240). (#7506) --- changelog.d/7506.feature | 1 + synapse/api/room_versions.py | 30 ++++-------------------------- tests/events/test_utils.py | 2 +- tests/handlers/test_federation.py | 6 +++--- tests/test_event_auth.py | 16 +++++----------- 5 files changed, 14 insertions(+), 41 deletions(-) create mode 100644 changelog.d/7506.feature (limited to 'tests') diff --git a/changelog.d/7506.feature b/changelog.d/7506.feature new file mode 100644 index 0000000000..1407c91689 --- /dev/null +++ b/changelog.d/7506.feature @@ -0,0 +1 @@ +Implement room version 6 per [MSC2240](https://github.com/matrix-org/matrix-doc/pull/2240). diff --git a/synapse/api/room_versions.py b/synapse/api/room_versions.py index 0901afb900..d7baf2bc39 100644 --- a/synapse/api/room_versions.py +++ b/synapse/api/room_versions.py @@ -120,34 +120,14 @@ class RoomVersions(object): strict_canonicaljson=False, limit_notifications_power_levels=False, ) - MSC2432_DEV = RoomVersion( - "org.matrix.msc2432", - RoomDisposition.UNSTABLE, + V6 = RoomVersion( + "6", + RoomDisposition.STABLE, EventFormatVersions.V3, StateResolutionVersions.V2, enforce_key_validity=True, special_case_aliases_auth=False, - strict_canonicaljson=False, - limit_notifications_power_levels=False, - ) - STRICT_CANONICALJSON = RoomVersion( - "org.matrix.strict_canonicaljson", - RoomDisposition.UNSTABLE, - EventFormatVersions.V3, - StateResolutionVersions.V2, - enforce_key_validity=True, - special_case_aliases_auth=True, strict_canonicaljson=True, - limit_notifications_power_levels=False, - ) - MSC2209_DEV = RoomVersion( - "org.matrix.msc2209", - RoomDisposition.UNSTABLE, - EventFormatVersions.V3, - StateResolutionVersions.V2, - enforce_key_validity=True, - special_case_aliases_auth=True, - strict_canonicaljson=False, limit_notifications_power_levels=True, ) @@ -160,8 +140,6 @@ KNOWN_ROOM_VERSIONS = { RoomVersions.V3, RoomVersions.V4, RoomVersions.V5, - RoomVersions.MSC2432_DEV, - RoomVersions.STRICT_CANONICALJSON, - RoomVersions.MSC2209_DEV, + RoomVersions.V6, ) } # type: Dict[str, RoomVersion] diff --git a/tests/events/test_utils.py b/tests/events/test_utils.py index ab5f5ac549..c1274c14af 100644 --- a/tests/events/test_utils.py +++ b/tests/events/test_utils.py @@ -156,7 +156,7 @@ class PruneEventTestCase(unittest.TestCase): "signatures": {}, "unsigned": {}, }, - room_version=RoomVersions.MSC2432_DEV, + room_version=RoomVersions.V6, ) diff --git a/tests/handlers/test_federation.py b/tests/handlers/test_federation.py index dfef58e704..96fea58673 100644 --- a/tests/handlers/test_federation.py +++ b/tests/handlers/test_federation.py @@ -226,7 +226,7 @@ class EventFromPduTestCase(TestCase): "auth_events": [], "origin_server_ts": 1234, }, - RoomVersions.STRICT_CANONICALJSON, + RoomVersions.V6, ) self.assertIsInstance(ev, EventBase) @@ -253,7 +253,7 @@ class EventFromPduTestCase(TestCase): "auth_events": [], "origin_server_ts": 1234, }, - RoomVersions.STRICT_CANONICALJSON, + RoomVersions.V6, ) def test_invalid_nested(self): @@ -270,5 +270,5 @@ class EventFromPduTestCase(TestCase): "auth_events": [], "origin_server_ts": 1234, }, - RoomVersions.STRICT_CANONICALJSON, + RoomVersions.V6, ) diff --git a/tests/test_event_auth.py b/tests/test_event_auth.py index f2def601fb..69b4c5d6c2 100644 --- a/tests/test_event_auth.py +++ b/tests/test_event_auth.py @@ -136,21 +136,18 @@ class EventAuthTestCase(unittest.TestCase): # creator should be able to send aliases event_auth.check( - RoomVersions.MSC2432_DEV, - _alias_event(creator), - auth_events, - do_sig_check=False, + RoomVersions.V6, _alias_event(creator), auth_events, do_sig_check=False, ) # No particular checks are done on the state key. event_auth.check( - RoomVersions.MSC2432_DEV, + RoomVersions.V6, _alias_event(creator, state_key=""), auth_events, do_sig_check=False, ) event_auth.check( - RoomVersions.MSC2432_DEV, + RoomVersions.V6, _alias_event(creator, state_key="test.com"), auth_events, do_sig_check=False, @@ -159,10 +156,7 @@ class EventAuthTestCase(unittest.TestCase): # Per standard auth rules, the member must be in the room. with self.assertRaises(AuthError): event_auth.check( - RoomVersions.MSC2432_DEV, - _alias_event(other), - auth_events, - do_sig_check=False, + RoomVersions.V6, _alias_event(other), auth_events, do_sig_check=False, ) def test_msc2209(self): @@ -192,7 +186,7 @@ class EventAuthTestCase(unittest.TestCase): # But an MSC2209 room rejects this change. with self.assertRaises(AuthError): event_auth.check( - RoomVersions.MSC2209_DEV, + RoomVersions.V6, _power_levels_event(pleb, {"notifications": {"room": 100}}), auth_events, do_sig_check=False, -- cgit 1.4.1 From a3cf36f76ed41222241393adf608d0e640bb51b8 Mon Sep 17 00:00:00 2001 From: Patrick Cloke Date: Fri, 15 May 2020 12:26:02 -0400 Subject: Support UI Authentication for OpenID Connect accounts (#7457) --- changelog.d/7457.feature | 1 + synapse/handlers/auth.py | 4 +- synapse/handlers/oidc_handler.py | 76 +++++++++++++++++++++++++++--------- synapse/rest/client/v1/login.py | 31 +++++++++------ synapse/rest/client/v2_alpha/auth.py | 19 +++++++-- tests/handlers/test_oidc.py | 15 ++++--- 6 files changed, 105 insertions(+), 41 deletions(-) create mode 100644 changelog.d/7457.feature (limited to 'tests') diff --git a/changelog.d/7457.feature b/changelog.d/7457.feature new file mode 100644 index 0000000000..7ad767bf71 --- /dev/null +++ b/changelog.d/7457.feature @@ -0,0 +1 @@ +Add OpenID Connect login/registration support. Contributed by Quentin Gliech, on behalf of [les Connecteurs](https://connecteu.rs). diff --git a/synapse/handlers/auth.py b/synapse/handlers/auth.py index 524281d2f1..75b39e878c 100644 --- a/synapse/handlers/auth.py +++ b/synapse/handlers/auth.py @@ -80,7 +80,9 @@ class AuthHandler(BaseHandler): self.hs = hs # FIXME better possibility to access registrationHandler later? self.macaroon_gen = hs.get_macaroon_generator() self._password_enabled = hs.config.password_enabled - self._sso_enabled = hs.config.saml2_enabled or hs.config.cas_enabled + self._sso_enabled = ( + hs.config.cas_enabled or hs.config.saml2_enabled or hs.config.oidc_enabled + ) # we keep this as a list despite the O(N^2) implication so that we can # keep PASSWORD first and avoid confusing clients which pick the first diff --git a/synapse/handlers/oidc_handler.py b/synapse/handlers/oidc_handler.py index 178f263439..4ba8c7fda5 100644 --- a/synapse/handlers/oidc_handler.py +++ b/synapse/handlers/oidc_handler.py @@ -311,7 +311,7 @@ class OidcHandler: ``ClientAuth`` to authenticate with the client with its ID and secret. Args: - code: The autorization code we got from the callback. + code: The authorization code we got from the callback. Returns: A dict containing various tokens. @@ -497,11 +497,14 @@ class OidcHandler: return UserInfo(claims) async def handle_redirect_request( - self, request: SynapseRequest, client_redirect_url: bytes - ) -> None: + self, + request: SynapseRequest, + client_redirect_url: bytes, + ui_auth_session_id: Optional[str] = None, + ) -> str: """Handle an incoming request to /login/sso/redirect - It redirects the browser to the authorization endpoint with a few + It returns a redirect to the authorization endpoint with a few parameters: - ``client_id``: the client ID set in ``oidc_config.client_id`` @@ -511,24 +514,32 @@ class OidcHandler: - ``state``: a random string - ``nonce``: a random string - In addition to redirecting the client, we are setting a cookie with + In addition generating a redirect URL, we are setting a cookie with a signed macaroon token containing the state, the nonce and the client_redirect_url params. Those are then checked when the client comes back from the provider. - Args: request: the incoming request from the browser. We'll respond to it with a redirect and a cookie. client_redirect_url: the URL that we should redirect the client to when everything is done + ui_auth_session_id: The session ID of the ongoing UI Auth (or + None if this is a login). + + Returns: + The redirect URL to the authorization endpoint. + """ state = generate_token() nonce = generate_token() cookie = self._generate_oidc_session_token( - state=state, nonce=nonce, client_redirect_url=client_redirect_url.decode(), + state=state, + nonce=nonce, + client_redirect_url=client_redirect_url.decode(), + ui_auth_session_id=ui_auth_session_id, ) request.addCookie( SESSION_COOKIE_NAME, @@ -541,7 +552,7 @@ class OidcHandler: metadata = await self.load_metadata() authorization_endpoint = metadata.get("authorization_endpoint") - uri = prepare_grant_uri( + return prepare_grant_uri( authorization_endpoint, client_id=self._client_auth.client_id, response_type="code", @@ -550,8 +561,6 @@ class OidcHandler: state=state, nonce=nonce, ) - request.redirect(uri) - finish_request(request) async def handle_oidc_callback(self, request: SynapseRequest) -> None: """Handle an incoming request to /_synapse/oidc/callback @@ -625,7 +634,11 @@ class OidcHandler: # Deserialize the session token and verify it. try: - nonce, client_redirect_url = self._verify_oidc_session_token(session, state) + ( + nonce, + client_redirect_url, + ui_auth_session_id, + ) = self._verify_oidc_session_token(session, state) except MacaroonDeserializationException as e: logger.exception("Invalid session") self._render_error(request, "invalid_session", str(e)) @@ -678,15 +691,21 @@ class OidcHandler: return # and finally complete the login - await self._auth_handler.complete_sso_login( - user_id, request, client_redirect_url - ) + if ui_auth_session_id: + await self._auth_handler.complete_sso_ui_auth( + user_id, ui_auth_session_id, request + ) + else: + await self._auth_handler.complete_sso_login( + user_id, request, client_redirect_url + ) def _generate_oidc_session_token( self, state: str, nonce: str, client_redirect_url: str, + ui_auth_session_id: Optional[str], duration_in_ms: int = (60 * 60 * 1000), ) -> str: """Generates a signed token storing data about an OIDC session. @@ -702,6 +721,8 @@ class OidcHandler: nonce: The ``nonce`` parameter passed to the OIDC provider. client_redirect_url: The URL the client gave when it initiated the flow. + ui_auth_session_id: The session ID of the ongoing UI Auth (or + None if this is a login). duration_in_ms: An optional duration for the token in milliseconds. Defaults to an hour. @@ -718,12 +739,19 @@ class OidcHandler: macaroon.add_first_party_caveat( "client_redirect_url = %s" % (client_redirect_url,) ) + if ui_auth_session_id: + macaroon.add_first_party_caveat( + "ui_auth_session_id = %s" % (ui_auth_session_id,) + ) now = self._clock.time_msec() expiry = now + duration_in_ms macaroon.add_first_party_caveat("time < %d" % (expiry,)) + return macaroon.serialize() - def _verify_oidc_session_token(self, session: str, state: str) -> Tuple[str, str]: + def _verify_oidc_session_token( + self, session: str, state: str + ) -> Tuple[str, str, Optional[str]]: """Verifies and extract an OIDC session token. This verifies that a given session token was issued by this homeserver @@ -734,7 +762,7 @@ class OidcHandler: state: The state the OIDC provider gave back Returns: - The nonce and the client_redirect_url for this session + The nonce, client_redirect_url, and ui_auth_session_id for this session """ macaroon = pymacaroons.Macaroon.deserialize(session) @@ -744,17 +772,27 @@ class OidcHandler: v.satisfy_exact("state = %s" % (state,)) v.satisfy_general(lambda c: c.startswith("nonce = ")) v.satisfy_general(lambda c: c.startswith("client_redirect_url = ")) + # Sometimes there's a UI auth session ID, it seems to be OK to attempt + # to always satisfy this. + v.satisfy_general(lambda c: c.startswith("ui_auth_session_id = ")) v.satisfy_general(self._verify_expiry) v.verify(macaroon, self._macaroon_secret_key) - # Extract the `nonce` and `client_redirect_url` from the token + # Extract the `nonce`, `client_redirect_url`, and maybe the + # `ui_auth_session_id` from the token. nonce = self._get_value_from_macaroon(macaroon, "nonce") client_redirect_url = self._get_value_from_macaroon( macaroon, "client_redirect_url" ) + try: + ui_auth_session_id = self._get_value_from_macaroon( + macaroon, "ui_auth_session_id" + ) # type: Optional[str] + except ValueError: + ui_auth_session_id = None - return nonce, client_redirect_url + return nonce, client_redirect_url, ui_auth_session_id def _get_value_from_macaroon(self, macaroon: pymacaroons.Macaroon, key: str) -> str: """Extracts a caveat value from a macaroon token. @@ -773,7 +811,7 @@ class OidcHandler: for caveat in macaroon.caveats: if caveat.caveat_id.startswith(prefix): return caveat.caveat_id[len(prefix) :] - raise Exception("No %s caveat in macaroon" % (key,)) + raise ValueError("No %s caveat in macaroon" % (key,)) def _verify_expiry(self, caveat: str) -> bool: prefix = "time < " diff --git a/synapse/rest/client/v1/login.py b/synapse/rest/client/v1/login.py index de7eca21f8..d89b2e5532 100644 --- a/synapse/rest/client/v1/login.py +++ b/synapse/rest/client/v1/login.py @@ -401,19 +401,22 @@ class BaseSSORedirectServlet(RestServlet): PATTERNS = client_patterns("/login/(cas|sso)/redirect", v1=True) - def on_GET(self, request: SynapseRequest): + async def on_GET(self, request: SynapseRequest): args = request.args if b"redirectUrl" not in args: return 400, "Redirect URL not specified for SSO auth" client_redirect_url = args[b"redirectUrl"][0] - sso_url = self.get_sso_url(client_redirect_url) + sso_url = await self.get_sso_url(request, client_redirect_url) request.redirect(sso_url) finish_request(request) - def get_sso_url(self, client_redirect_url: bytes) -> bytes: + async def get_sso_url( + self, request: SynapseRequest, client_redirect_url: bytes + ) -> bytes: """Get the URL to redirect to, to perform SSO auth Args: + request: The client request to redirect. client_redirect_url: the URL that we should redirect the client to when everything is done @@ -428,7 +431,9 @@ class CasRedirectServlet(BaseSSORedirectServlet): def __init__(self, hs): self._cas_handler = hs.get_cas_handler() - def get_sso_url(self, client_redirect_url: bytes) -> bytes: + async def get_sso_url( + self, request: SynapseRequest, client_redirect_url: bytes + ) -> bytes: return self._cas_handler.get_redirect_url( {"redirectUrl": client_redirect_url} ).encode("ascii") @@ -465,11 +470,13 @@ class SAMLRedirectServlet(BaseSSORedirectServlet): def __init__(self, hs): self._saml_handler = hs.get_saml_handler() - def get_sso_url(self, client_redirect_url: bytes) -> bytes: + async def get_sso_url( + self, request: SynapseRequest, client_redirect_url: bytes + ) -> bytes: return self._saml_handler.handle_redirect_request(client_redirect_url) -class OIDCRedirectServlet(RestServlet): +class OIDCRedirectServlet(BaseSSORedirectServlet): """Implementation for /login/sso/redirect for the OIDC login flow.""" PATTERNS = client_patterns("/login/sso/redirect", v1=True) @@ -477,12 +484,12 @@ class OIDCRedirectServlet(RestServlet): def __init__(self, hs): self._oidc_handler = hs.get_oidc_handler() - async def on_GET(self, request): - args = request.args - if b"redirectUrl" not in args: - return 400, "Redirect URL not specified for SSO auth" - client_redirect_url = args[b"redirectUrl"][0] - await self._oidc_handler.handle_redirect_request(request, client_redirect_url) + async def get_sso_url( + self, request: SynapseRequest, client_redirect_url: bytes + ) -> bytes: + return await self._oidc_handler.handle_redirect_request( + request, client_redirect_url + ) def register_servlets(hs, http_server): diff --git a/synapse/rest/client/v2_alpha/auth.py b/synapse/rest/client/v2_alpha/auth.py index 24dd3d3e96..7bca1326d5 100644 --- a/synapse/rest/client/v2_alpha/auth.py +++ b/synapse/rest/client/v2_alpha/auth.py @@ -131,14 +131,19 @@ class AuthRestServlet(RestServlet): self.registration_handler = hs.get_registration_handler() # SSO configuration. - self._saml_enabled = hs.config.saml2_enabled - if self._saml_enabled: - self._saml_handler = hs.get_saml_handler() self._cas_enabled = hs.config.cas_enabled if self._cas_enabled: self._cas_handler = hs.get_cas_handler() self._cas_server_url = hs.config.cas_server_url self._cas_service_url = hs.config.cas_service_url + self._saml_enabled = hs.config.saml2_enabled + if self._saml_enabled: + self._saml_handler = hs.get_saml_handler() + self._oidc_enabled = hs.config.oidc_enabled + if self._oidc_enabled: + self._oidc_handler = hs.get_oidc_handler() + self._cas_server_url = hs.config.cas_server_url + self._cas_service_url = hs.config.cas_service_url async def on_GET(self, request, stagetype): session = parse_string(request, "session") @@ -172,11 +177,17 @@ class AuthRestServlet(RestServlet): ) elif self._saml_enabled: - client_redirect_url = "" + client_redirect_url = b"" sso_redirect_url = self._saml_handler.handle_redirect_request( client_redirect_url, session ) + elif self._oidc_enabled: + client_redirect_url = b"" + sso_redirect_url = await self._oidc_handler.handle_redirect_request( + request, client_redirect_url, session + ) + else: raise SynapseError(400, "Homeserver not configured for SSO.") diff --git a/tests/handlers/test_oidc.py b/tests/handlers/test_oidc.py index 61963aa90d..1bb25ab684 100644 --- a/tests/handlers/test_oidc.py +++ b/tests/handlers/test_oidc.py @@ -292,11 +292,10 @@ class OidcHandlerTestCase(HomeserverTestCase): @defer.inlineCallbacks def test_redirect_request(self): """The redirect request has the right arguments & generates a valid session cookie.""" - req = Mock(spec=["addCookie", "redirect", "finish"]) - yield defer.ensureDeferred( + req = Mock(spec=["addCookie"]) + url = yield defer.ensureDeferred( self.handler.handle_redirect_request(req, b"http://client/redirect") ) - url = req.redirect.call_args[0][0] url = urlparse(url) auth_endpoint = urlparse(AUTHORIZATION_ENDPOINT) @@ -382,7 +381,10 @@ class OidcHandlerTestCase(HomeserverTestCase): nonce = "nonce" client_redirect_url = "http://client/redirect" session = self.handler._generate_oidc_session_token( - state=state, nonce=nonce, client_redirect_url=client_redirect_url, + state=state, + nonce=nonce, + client_redirect_url=client_redirect_url, + ui_auth_session_id=None, ) request.getCookie.return_value = session @@ -472,7 +474,10 @@ class OidcHandlerTestCase(HomeserverTestCase): # Mismatching session session = self.handler._generate_oidc_session_token( - state="state", nonce="nonce", client_redirect_url="http://client/redirect", + state="state", + nonce="nonce", + client_redirect_url="http://client/redirect", + ui_auth_session_id=None, ) request.args = {} request.args[b"state"] = [b"mismatching state"] -- cgit 1.4.1 From 6c1f7c722f0baade9aecf41f600fcced670c4fcb Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Fri, 15 May 2020 19:03:25 +0100 Subject: Fix limit logic for AccountDataStream (#7384) Make sure that the AccountDataStream presents complete updates, in the right order. This is much the same fix as #7337 and #7358, but applied to a different stream. --- changelog.d/7384.bugfix | 1 + synapse/replication/tcp/streams/_base.py | 68 +++++++++--- synapse/storage/data_stores/main/account_data.py | 62 +++++++---- tests/replication/tcp/streams/test_account_data.py | 117 +++++++++++++++++++++ 4 files changed, 217 insertions(+), 31 deletions(-) create mode 100644 changelog.d/7384.bugfix create mode 100644 tests/replication/tcp/streams/test_account_data.py (limited to 'tests') diff --git a/changelog.d/7384.bugfix b/changelog.d/7384.bugfix new file mode 100644 index 0000000000..f49c600173 --- /dev/null +++ b/changelog.d/7384.bugfix @@ -0,0 +1 @@ +Fix a bug where event updates might not be sent over replication to worker processes after the stream falls behind. diff --git a/synapse/replication/tcp/streams/_base.py b/synapse/replication/tcp/streams/_base.py index b48a6a3e91..d42aaff055 100644 --- a/synapse/replication/tcp/streams/_base.py +++ b/synapse/replication/tcp/streams/_base.py @@ -14,14 +14,27 @@ # See the License for the specific language governing permissions and # limitations under the License. +import heapq import logging from collections import namedtuple -from typing import Any, Awaitable, Callable, List, Optional, Tuple +from typing import ( + TYPE_CHECKING, + Any, + Awaitable, + Callable, + List, + Optional, + Tuple, + TypeVar, +) import attr from synapse.replication.http.streams import ReplicationGetStreamUpdates +if TYPE_CHECKING: + import synapse.server + logger = logging.getLogger(__name__) # the number of rows to request from an update_function. @@ -37,7 +50,7 @@ Token = int # parsing with Stream.parse_row (which turns it into a `ROW_TYPE`). Normally it's # just a row from a database query, though this is dependent on the stream in question. # -StreamRow = Tuple +StreamRow = TypeVar("StreamRow", bound=Tuple) # The type returned by the update_function of a stream, as well as get_updates(), # get_updates_since, etc. @@ -533,32 +546,63 @@ class AccountDataStream(Stream): """ AccountDataStreamRow = namedtuple( - "AccountDataStream", ("user_id", "room_id", "data_type") # str # str # str + "AccountDataStream", + ("user_id", "room_id", "data_type"), # str # Optional[str] # str ) NAME = "account_data" ROW_TYPE = AccountDataStreamRow - def __init__(self, hs): + def __init__(self, hs: "synapse.server.HomeServer"): self.store = hs.get_datastore() super().__init__( hs.get_instance_name(), current_token_without_instance(self.store.get_max_account_data_stream_id), - db_query_to_update_function(self._update_function), + self._update_function, + ) + + async def _update_function( + self, instance_name: str, from_token: int, to_token: int, limit: int + ) -> StreamUpdateResult: + limited = False + global_results = await self.store.get_updated_global_account_data( + from_token, to_token, limit ) - async def _update_function(self, from_token, to_token, limit): - global_results, room_results = await self.store.get_all_updated_account_data( - from_token, from_token, to_token, limit + # if the global results hit the limit, we'll need to limit the room results to + # the same stream token. + if len(global_results) >= limit: + to_token = global_results[-1][0] + limited = True + + room_results = await self.store.get_updated_room_account_data( + from_token, to_token, limit ) - results = list(room_results) - results.extend( - (stream_id, user_id, None, account_data_type) + # likewise, if the room results hit the limit, limit the global results to + # the same stream token. + if len(room_results) >= limit: + to_token = room_results[-1][0] + limited = True + + # convert the global results to the right format, and limit them to the to_token + # at the same time + global_rows = ( + (stream_id, (user_id, None, account_data_type)) for stream_id, user_id, account_data_type in global_results + if stream_id <= to_token + ) + + # we know that the room_results are already limited to `to_token` so no need + # for a check on `stream_id` here. + room_rows = ( + (stream_id, (user_id, room_id, account_data_type)) + for stream_id, user_id, room_id, account_data_type in room_results ) - return results + # we need to return a sorted list, so merge them together. + updates = list(heapq.merge(room_rows, global_rows)) + return updates, to_token, limited class GroupServerStream(Stream): diff --git a/synapse/storage/data_stores/main/account_data.py b/synapse/storage/data_stores/main/account_data.py index 46b494b334..f9eef1b78e 100644 --- a/synapse/storage/data_stores/main/account_data.py +++ b/synapse/storage/data_stores/main/account_data.py @@ -16,6 +16,7 @@ import abc import logging +from typing import List, Tuple from canonicaljson import json @@ -175,41 +176,64 @@ class AccountDataWorkerStore(SQLBaseStore): "get_account_data_for_room_and_type", get_account_data_for_room_and_type_txn ) - def get_all_updated_account_data( - self, last_global_id, last_room_id, current_id, limit - ): - """Get all the client account_data that has changed on the server + async def get_updated_global_account_data( + self, last_id: int, current_id: int, limit: int + ) -> List[Tuple[int, str, str]]: + """Get the global account_data that has changed, for the account_data stream + Args: - last_global_id(int): The position to fetch from for top level data - last_room_id(int): The position to fetch from for per room data - current_id(int): The position to fetch up to. + last_id: the last stream_id from the previous batch. + current_id: the maximum stream_id to return up to + limit: the maximum number of rows to return + Returns: - A deferred pair of lists of tuples of stream_id int, user_id string, - room_id string, and type string. + A list of tuples of stream_id int, user_id string, + and type string. """ - if last_room_id == current_id and last_global_id == current_id: - return defer.succeed(([], [])) + if last_id == current_id: + return [] - def get_updated_account_data_txn(txn): + def get_updated_global_account_data_txn(txn): sql = ( "SELECT stream_id, user_id, account_data_type" " FROM account_data WHERE ? < stream_id AND stream_id <= ?" " ORDER BY stream_id ASC LIMIT ?" ) - txn.execute(sql, (last_global_id, current_id, limit)) - global_results = txn.fetchall() + txn.execute(sql, (last_id, current_id, limit)) + return txn.fetchall() + + return await self.db.runInteraction( + "get_updated_global_account_data", get_updated_global_account_data_txn + ) + + async def get_updated_room_account_data( + self, last_id: int, current_id: int, limit: int + ) -> List[Tuple[int, str, str, str]]: + """Get the global account_data that has changed, for the account_data stream + Args: + last_id: the last stream_id from the previous batch. + current_id: the maximum stream_id to return up to + limit: the maximum number of rows to return + + Returns: + A list of tuples of stream_id int, user_id string, + room_id string and type string. + """ + if last_id == current_id: + return [] + + def get_updated_room_account_data_txn(txn): sql = ( "SELECT stream_id, user_id, room_id, account_data_type" " FROM room_account_data WHERE ? < stream_id AND stream_id <= ?" " ORDER BY stream_id ASC LIMIT ?" ) - txn.execute(sql, (last_room_id, current_id, limit)) - room_results = txn.fetchall() - return global_results, room_results + txn.execute(sql, (last_id, current_id, limit)) + return txn.fetchall() - return self.db.runInteraction( - "get_all_updated_account_data_txn", get_updated_account_data_txn + return await self.db.runInteraction( + "get_updated_room_account_data", get_updated_room_account_data_txn ) def get_updated_account_data_for_user(self, user_id, stream_id): diff --git a/tests/replication/tcp/streams/test_account_data.py b/tests/replication/tcp/streams/test_account_data.py new file mode 100644 index 0000000000..6a5116dd2a --- /dev/null +++ b/tests/replication/tcp/streams/test_account_data.py @@ -0,0 +1,117 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from synapse.replication.tcp.streams._base import ( + _STREAM_UPDATE_TARGET_ROW_COUNT, + AccountDataStream, +) + +from tests.replication._base import BaseStreamTestCase + + +class AccountDataStreamTestCase(BaseStreamTestCase): + def test_update_function_room_account_data_limit(self): + """Test replication with many room account data updates + """ + store = self.hs.get_datastore() + + # generate lots of account data updates + updates = [] + for i in range(_STREAM_UPDATE_TARGET_ROW_COUNT + 5): + update = "m.test_type.%i" % (i,) + self.get_success( + store.add_account_data_to_room("test_user", "test_room", update, {}) + ) + updates.append(update) + + # also one global update + self.get_success(store.add_account_data_for_user("test_user", "m.global", {})) + + # tell the notifier to catch up to avoid duplicate rows. + # workaround for https://github.com/matrix-org/synapse/issues/7360 + # FIXME remove this when the above is fixed + self.replicate() + + # check we're testing what we think we are: no rows should yet have been + # received + self.assertEqual([], self.test_handler.received_rdata_rows) + + # now reconnect to pull the updates + self.reconnect() + self.replicate() + + # we should have received all the expected rows in the right order + received_rows = self.test_handler.received_rdata_rows + + for t in updates: + (stream_name, token, row) = received_rows.pop(0) + self.assertEqual(stream_name, AccountDataStream.NAME) + self.assertIsInstance(row, AccountDataStream.AccountDataStreamRow) + self.assertEqual(row.data_type, t) + self.assertEqual(row.room_id, "test_room") + + (stream_name, token, row) = received_rows.pop(0) + self.assertIsInstance(row, AccountDataStream.AccountDataStreamRow) + self.assertEqual(row.data_type, "m.global") + self.assertIsNone(row.room_id) + + self.assertEqual([], received_rows) + + def test_update_function_global_account_data_limit(self): + """Test replication with many global account data updates + """ + store = self.hs.get_datastore() + + # generate lots of account data updates + updates = [] + for i in range(_STREAM_UPDATE_TARGET_ROW_COUNT + 5): + update = "m.test_type.%i" % (i,) + self.get_success(store.add_account_data_for_user("test_user", update, {})) + updates.append(update) + + # also one per-room update + self.get_success( + store.add_account_data_to_room("test_user", "test_room", "m.per_room", {}) + ) + + # tell the notifier to catch up to avoid duplicate rows. + # workaround for https://github.com/matrix-org/synapse/issues/7360 + # FIXME remove this when the above is fixed + self.replicate() + + # check we're testing what we think we are: no rows should yet have been + # received + self.assertEqual([], self.test_handler.received_rdata_rows) + + # now reconnect to pull the updates + self.reconnect() + self.replicate() + + # we should have received all the expected rows in the right order + received_rows = self.test_handler.received_rdata_rows + + for t in updates: + (stream_name, token, row) = received_rows.pop(0) + self.assertEqual(stream_name, AccountDataStream.NAME) + self.assertIsInstance(row, AccountDataStream.AccountDataStreamRow) + self.assertEqual(row.data_type, t) + self.assertIsNone(row.room_id) + + (stream_name, token, row) = received_rows.pop(0) + self.assertIsInstance(row, AccountDataStream.AccountDataStreamRow) + self.assertEqual(row.data_type, "m.per_room") + self.assertEqual(row.room_id, "test_room") + + self.assertEqual([], received_rows) -- cgit 1.4.1 From 164f50f5f25a3204cce5fd2c8f196e9a9d4deb5d Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Mon, 18 May 2020 10:43:05 +0100 Subject: fix mypy for tests/replication (#7518) --- changelog.d/7518.misc | 1 + tests/replication/slave/storage/test_events.py | 16 +++++----------- tests/replication/tcp/test_commands.py | 4 ++-- tox.ini | 2 +- 4 files changed, 9 insertions(+), 14 deletions(-) create mode 100644 changelog.d/7518.misc (limited to 'tests') diff --git a/changelog.d/7518.misc b/changelog.d/7518.misc new file mode 100644 index 0000000000..f6e143fe1c --- /dev/null +++ b/changelog.d/7518.misc @@ -0,0 +1 @@ +Fix typing annotations in `tests.replication`. diff --git a/tests/replication/slave/storage/test_events.py b/tests/replication/slave/storage/test_events.py index 0fee8a71c4..1a88c7fb80 100644 --- a/tests/replication/slave/storage/test_events.py +++ b/tests/replication/slave/storage/test_events.py @@ -17,11 +17,12 @@ from canonicaljson import encode_canonical_json from synapse.api.room_versions import RoomVersions from synapse.events import FrozenEvent, _EventInternalMetadata, make_event_from_dict -from synapse.events.snapshot import EventContext from synapse.handlers.room import RoomEventSource from synapse.replication.slave.storage.events import SlavedEventStore from synapse.storage.roommember import RoomsForUser +from tests.server import FakeTransport + from ._base import BaseSlavedStoreTestCase USER_ID = "@feeling:test" @@ -240,6 +241,7 @@ class SlavedEventStoreTestCase(BaseSlavedStoreTestCase): # limit the replication rate repl_transport = self._server_transport + assert isinstance(repl_transport, FakeTransport) repl_transport.autoflush = False # build the join and message events and persist them in the same batch. @@ -322,7 +324,6 @@ class SlavedEventStoreTestCase(BaseSlavedStoreTestCase): type="m.room.message", key=None, internal={}, - state=None, depth=None, prev_events=[], auth_events=[], @@ -362,15 +363,8 @@ class SlavedEventStoreTestCase(BaseSlavedStoreTestCase): event = make_event_from_dict(event_dict, internal_metadata_dict=internal) self.event_id += 1 - - if state is not None: - state_ids = {key: e.event_id for key, e in state.items()} - context = EventContext.with_state( - state_group=None, current_state_ids=state_ids, prev_state_ids=state_ids - ) - else: - state_handler = self.hs.get_state_handler() - context = self.get_success(state_handler.compute_event_context(event)) + state_handler = self.hs.get_state_handler() + context = self.get_success(state_handler.compute_event_context(event)) self.master_store.add_push_actions_to_staging( event.event_id, {user_id: actions for user_id, actions in push_actions} diff --git a/tests/replication/tcp/test_commands.py b/tests/replication/tcp/test_commands.py index 7ddfd0a733..60c10a441a 100644 --- a/tests/replication/tcp/test_commands.py +++ b/tests/replication/tcp/test_commands.py @@ -30,7 +30,7 @@ class ParseCommandTestCase(TestCase): def test_parse_rdata(self): line = 'RDATA events master 6287863 ["ev", ["$eventid", "!roomid", "type", null, null, null]]' cmd = parse_command_from_line(line) - self.assertIsInstance(cmd, RdataCommand) + assert isinstance(cmd, RdataCommand) self.assertEqual(cmd.stream_name, "events") self.assertEqual(cmd.instance_name, "master") self.assertEqual(cmd.token, 6287863) @@ -38,7 +38,7 @@ class ParseCommandTestCase(TestCase): def test_parse_rdata_batch(self): line = 'RDATA presence master batch ["@foo:example.com", "online"]' cmd = parse_command_from_line(line) - self.assertIsInstance(cmd, RdataCommand) + assert isinstance(cmd, RdataCommand) self.assertEqual(cmd.stream_name, "presence") self.assertEqual(cmd.instance_name, "master") self.assertIsNone(cmd.token) diff --git a/tox.ini b/tox.ini index 5a1fa610b6..3bb4d45e2a 100644 --- a/tox.ini +++ b/tox.ini @@ -207,7 +207,7 @@ commands = mypy \ synapse/storage/util \ synapse/streams \ synapse/util/caches/stream_change_cache.py \ - tests/replication/tcp/streams \ + tests/replication \ tests/test_utils \ tests/rest/client/v2_alpha/test_auth.py \ tests/util/test_stream_change_cache.py -- cgit 1.4.1 From d1ae1015ecb7dcb36d7cb39d5d41733e1ced2a52 Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Thu, 21 May 2020 17:41:12 +0200 Subject: Retry to sync out of sync device lists (#7453) When a call to `user_device_resync` fails, we don't currently mark the remote user's device list as out of sync, nor do we retry to sync it. https://github.com/matrix-org/synapse/pull/6776 introduced some code infrastructure to mark device lists as stale/out of sync. This commit uses that code infrastructure to mark device lists as out of sync if processing an incoming device list update makes the device handler realise that the device list is out of sync, but we can't resync right now. It also adds a looping call to retry all failed resync every 30s. This shouldn't cause too much spam in the logs as this commit also removes the "Failed to handle device list update for..." warning logs when catching `NotRetryingDestination`. Fixes #7418 --- changelog.d/7453.bugfix | 1 + synapse/handlers/device.py | 80 ++++++++++++++++++++++++++--- synapse/storage/data_stores/main/devices.py | 34 +++++++----- tests/test_federation.py | 63 ++++++++++++++++++++++- 4 files changed, 158 insertions(+), 20 deletions(-) create mode 100644 changelog.d/7453.bugfix (limited to 'tests') diff --git a/changelog.d/7453.bugfix b/changelog.d/7453.bugfix new file mode 100644 index 0000000000..629a3d61e2 --- /dev/null +++ b/changelog.d/7453.bugfix @@ -0,0 +1 @@ +Fix a bug that would cause Synapse not to resync out-of-sync device lists. diff --git a/synapse/handlers/device.py b/synapse/handlers/device.py index 9bd941b5a0..29a19b4572 100644 --- a/synapse/handlers/device.py +++ b/synapse/handlers/device.py @@ -29,6 +29,7 @@ from synapse.api.errors import ( SynapseError, ) from synapse.logging.opentracing import log_kv, set_tag, trace +from synapse.metrics.background_process_metrics import run_as_background_process from synapse.types import RoomStreamToken, get_domain_from_id from synapse.util import stringutils from synapse.util.async_helpers import Linearizer @@ -535,6 +536,15 @@ class DeviceListUpdater(object): iterable=True, ) + # Attempt to resync out of sync device lists every 30s. + self._resync_retry_in_progress = False + self.clock.looping_call( + run_as_background_process, + 30 * 1000, + func=self._maybe_retry_device_resync, + desc="_maybe_retry_device_resync", + ) + @trace @defer.inlineCallbacks def incoming_device_list_update(self, origin, edu_content): @@ -679,11 +689,50 @@ class DeviceListUpdater(object): return False @defer.inlineCallbacks - def user_device_resync(self, user_id): + def _maybe_retry_device_resync(self): + """Retry to resync device lists that are out of sync, except if another retry is + in progress. + """ + if self._resync_retry_in_progress: + return + + try: + # Prevent another call of this function to retry resyncing device lists so + # we don't send too many requests. + self._resync_retry_in_progress = True + # Get all of the users that need resyncing. + need_resync = yield self.store.get_user_ids_requiring_device_list_resync() + # Iterate over the set of user IDs. + for user_id in need_resync: + # Try to resync the current user's devices list. Exception handling + # isn't necessary here, since user_device_resync catches all instances + # of "Exception" that might be raised from the federation request. This + # means that if an exception is raised by this function, it must be + # because of a database issue, which means _maybe_retry_device_resync + # probably won't be able to go much further anyway. + result = yield self.user_device_resync( + user_id=user_id, mark_failed_as_stale=False, + ) + # user_device_resync only returns a result if it managed to successfully + # resync and update the database. Updating the table of users requiring + # resync isn't necessary here as user_device_resync already does it + # (through self.store.update_remote_device_list_cache). + if result: + logger.debug( + "Successfully resynced the device list for %s" % user_id, + ) + finally: + # Allow future calls to retry resyncinc out of sync device lists. + self._resync_retry_in_progress = False + + @defer.inlineCallbacks + def user_device_resync(self, user_id, mark_failed_as_stale=True): """Fetches all devices for a user and updates the device cache with them. Args: user_id (str): The user's id whose device_list will be updated. + mark_failed_as_stale (bool): Whether to mark the user's device list as stale + if the attempt to resync failed. Returns: Deferred[dict]: a dict with device info as under the "devices" in the result of this request: @@ -694,10 +743,23 @@ class DeviceListUpdater(object): origin = get_domain_from_id(user_id) try: result = yield self.federation.query_user_devices(origin, user_id) - except (NotRetryingDestination, RequestSendFailed, HttpResponseException): - # TODO: Remember that we are now out of sync and try again - # later - logger.warning("Failed to handle device list update for %s", user_id) + except NotRetryingDestination: + if mark_failed_as_stale: + # Mark the remote user's device list as stale so we know we need to retry + # it later. + yield self.store.mark_remote_user_device_cache_as_stale(user_id) + + return + except (RequestSendFailed, HttpResponseException) as e: + logger.warning( + "Failed to handle device list update for %s: %s", user_id, e, + ) + + if mark_failed_as_stale: + # Mark the remote user's device list as stale so we know we need to retry + # it later. + yield self.store.mark_remote_user_device_cache_as_stale(user_id) + # We abort on exceptions rather than accepting the update # as otherwise synapse will 'forget' that its device list # is out of date. If we bail then we will retry the resync @@ -711,13 +773,17 @@ class DeviceListUpdater(object): logger.info(e) return except Exception as e: - # TODO: Remember that we are now out of sync and try again - # later set_tag("error", True) log_kv( {"message": "Exception raised by federation request", "exception": e} ) logger.exception("Failed to handle device list update for %s", user_id) + + if mark_failed_as_stale: + # Mark the remote user's device list as stale so we know we need to retry + # it later. + yield self.store.mark_remote_user_device_cache_as_stale(user_id) + return log_kv({"result": result}) stream_id = result["stream_id"] diff --git a/synapse/storage/data_stores/main/devices.py b/synapse/storage/data_stores/main/devices.py index fe6d6ecfe0..0e8378714a 100644 --- a/synapse/storage/data_stores/main/devices.py +++ b/synapse/storage/data_stores/main/devices.py @@ -15,7 +15,7 @@ # See the License for the specific language governing permissions and # limitations under the License. import logging -from typing import List, Tuple +from typing import List, Optional, Set, Tuple from six import iteritems @@ -649,21 +649,31 @@ class DeviceWorkerStore(SQLBaseStore): return results @defer.inlineCallbacks - def get_user_ids_requiring_device_list_resync(self, user_ids: Collection[str]): + def get_user_ids_requiring_device_list_resync( + self, user_ids: Optional[Collection[str]] = None, + ) -> Set[str]: """Given a list of remote users return the list of users that we - should resync the device lists for. + should resync the device lists for. If None is given instead of a list, + return every user that we should resync the device lists for. Returns: - Deferred[Set[str]] + The IDs of users whose device lists need resync. """ - - rows = yield self.db.simple_select_many_batch( - table="device_lists_remote_resync", - column="user_id", - iterable=user_ids, - retcols=("user_id",), - desc="get_user_ids_requiring_device_list_resync", - ) + if user_ids: + rows = yield self.db.simple_select_many_batch( + table="device_lists_remote_resync", + column="user_id", + iterable=user_ids, + retcols=("user_id",), + desc="get_user_ids_requiring_device_list_resync_with_iterable", + ) + else: + rows = yield self.db.simple_select_list( + table="device_lists_remote_resync", + keyvalues=None, + retcols=("user_id",), + desc="get_user_ids_requiring_device_list_resync", + ) return {row["user_id"] for row in rows} diff --git a/tests/test_federation.py b/tests/test_federation.py index f297de95f1..13ff14863e 100644 --- a/tests/test_federation.py +++ b/tests/test_federation.py @@ -6,12 +6,13 @@ from synapse.events import make_event_from_dict from synapse.logging.context import LoggingContext from synapse.types import Requester, UserID from synapse.util import Clock +from synapse.util.retryutils import NotRetryingDestination from tests import unittest from tests.server import ThreadedMemoryReactorClock, setup_test_homeserver -class MessageAcceptTests(unittest.TestCase): +class MessageAcceptTests(unittest.HomeserverTestCase): def setUp(self): self.http_client = Mock() @@ -145,3 +146,63 @@ class MessageAcceptTests(unittest.TestCase): # Make sure the invalid event isn't there extrem = maybeDeferred(self.store.get_latest_event_ids_in_room, self.room_id) self.assertEqual(self.successResultOf(extrem)[0], "$join:test.serv") + + def test_retry_device_list_resync(self): + """Tests that device lists are marked as stale if they couldn't be synced, and + that stale device lists are retried periodically. + """ + remote_user_id = "@john:test_remote" + remote_origin = "test_remote" + + # Track the number of attempts to resync the user's device list. + self.resync_attempts = 0 + + # When this function is called, increment the number of resync attempts (only if + # we're querying devices for the right user ID), then raise a + # NotRetryingDestination error to fail the resync gracefully. + def query_user_devices(destination, user_id): + if user_id == remote_user_id: + self.resync_attempts += 1 + + raise NotRetryingDestination(0, 0, destination) + + # Register the mock on the federation client. + federation_client = self.homeserver.get_federation_client() + federation_client.query_user_devices = Mock(side_effect=query_user_devices) + + # Register a mock on the store so that the incoming update doesn't fail because + # we don't share a room with the user. + store = self.homeserver.get_datastore() + store.get_rooms_for_user = Mock(return_value=["!someroom:test"]) + + # Manually inject a fake device list update. We need this update to include at + # least one prev_id so that the user's device list will need to be retried. + device_list_updater = self.homeserver.get_device_handler().device_list_updater + self.get_success( + device_list_updater.incoming_device_list_update( + origin=remote_origin, + edu_content={ + "deleted": False, + "device_display_name": "Mobile", + "device_id": "QBUAZIFURK", + "prev_id": [5], + "stream_id": 6, + "user_id": remote_user_id, + }, + ) + ) + + # Check that there was one resync attempt. + self.assertEqual(self.resync_attempts, 1) + + # Check that the resync attempt failed and caused the user's device list to be + # marked as stale. + need_resync = self.get_success( + store.get_user_ids_requiring_device_list_resync() + ) + self.assertIn(remote_user_id, need_resync) + + # Check that waiting for 30 seconds caused Synapse to retry resyncing the device + # list. + self.reactor.advance(30) + self.assertEqual(self.resync_attempts, 2) -- cgit 1.4.1 From 06a02bc1ce9ef23a6dff28dbfd30f910ae330b1d Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Fri, 22 May 2020 13:41:11 +0100 Subject: Convert sending mail to async/await. (#7557) Mainly because sometimes the email push code raises exceptions where the stack traces have gotten lost, which is hopefully fixed by this. --- changelog.d/7557.misc | 1 + synapse/handlers/identity.py | 9 ++-- synapse/push/emailpusher.py | 38 ++++++-------- synapse/push/mailer.py | 84 +++++++++++++----------------- tests/rest/client/v2_alpha/test_account.py | 4 +- 5 files changed, 60 insertions(+), 76 deletions(-) create mode 100644 changelog.d/7557.misc (limited to 'tests') diff --git a/changelog.d/7557.misc b/changelog.d/7557.misc new file mode 100644 index 0000000000..c850a2bc0c --- /dev/null +++ b/changelog.d/7557.misc @@ -0,0 +1 @@ +Convert sending mail to async/await. diff --git a/synapse/handlers/identity.py b/synapse/handlers/identity.py index 0f0e632b62..9ed0d23b0f 100644 --- a/synapse/handlers/identity.py +++ b/synapse/handlers/identity.py @@ -290,8 +290,7 @@ class IdentityHandler(BaseHandler): return changed - @defer.inlineCallbacks - def send_threepid_validation( + async def send_threepid_validation( self, email_address, client_secret, @@ -319,7 +318,7 @@ class IdentityHandler(BaseHandler): """ # Check that this email/client_secret/send_attempt combo is new or # greater than what we've seen previously - session = yield self.store.get_threepid_validation_session( + session = await self.store.get_threepid_validation_session( "email", client_secret, address=email_address, validated=False ) @@ -353,7 +352,7 @@ class IdentityHandler(BaseHandler): # Send the mail with the link containing the token, client_secret # and session_id try: - yield send_email_func(email_address, token, client_secret, session_id) + await send_email_func(email_address, token, client_secret, session_id) except Exception: logger.exception( "Error sending threepid validation email to %s", email_address @@ -364,7 +363,7 @@ class IdentityHandler(BaseHandler): self.hs.clock.time_msec() + self.hs.config.email_validation_token_lifetime ) - yield self.store.start_or_continue_validation_session( + await self.store.start_or_continue_validation_session( "email", email_address, session_id, diff --git a/synapse/push/emailpusher.py b/synapse/push/emailpusher.py index ba4551d619..568c13eaea 100644 --- a/synapse/push/emailpusher.py +++ b/synapse/push/emailpusher.py @@ -15,7 +15,6 @@ import logging -from twisted.internet import defer from twisted.internet.error import AlreadyCalled, AlreadyCancelled from synapse.metrics.background_process_metrics import run_as_background_process @@ -132,8 +131,7 @@ class EmailPusher(object): self._is_processing = False self._start_processing() - @defer.inlineCallbacks - def _process(self): + async def _process(self): # we should never get here if we are already processing assert not self._is_processing @@ -142,7 +140,7 @@ class EmailPusher(object): if self.throttle_params is None: # this is our first loop: load up the throttle params - self.throttle_params = yield self.store.get_throttle_params_by_room( + self.throttle_params = await self.store.get_throttle_params_by_room( self.pusher_id ) @@ -151,7 +149,7 @@ class EmailPusher(object): while True: starting_max_ordering = self.max_stream_ordering try: - yield self._unsafe_process() + await self._unsafe_process() except Exception: logger.exception("Exception processing notifs") if self.max_stream_ordering == starting_max_ordering: @@ -159,8 +157,7 @@ class EmailPusher(object): finally: self._is_processing = False - @defer.inlineCallbacks - def _unsafe_process(self): + async def _unsafe_process(self): """ Main logic of the push loop without the wrapper function that sets up logging, measures and guards against multiple instances of it @@ -168,12 +165,12 @@ class EmailPusher(object): """ start = 0 if INCLUDE_ALL_UNREAD_NOTIFS else self.last_stream_ordering fn = self.store.get_unread_push_actions_for_user_in_range_for_email - unprocessed = yield fn(self.user_id, start, self.max_stream_ordering) + unprocessed = await fn(self.user_id, start, self.max_stream_ordering) soonest_due_at = None if not unprocessed: - yield self.save_last_stream_ordering_and_success(self.max_stream_ordering) + await self.save_last_stream_ordering_and_success(self.max_stream_ordering) return for push_action in unprocessed: @@ -201,15 +198,15 @@ class EmailPusher(object): "throttle_ms": self.get_room_throttle_ms(push_action["room_id"]), } - yield self.send_notification(unprocessed, reason) + await self.send_notification(unprocessed, reason) - yield self.save_last_stream_ordering_and_success( + await self.save_last_stream_ordering_and_success( max(ea["stream_ordering"] for ea in unprocessed) ) # we update the throttle on all the possible unprocessed push actions for ea in unprocessed: - yield self.sent_notif_update_throttle(ea["room_id"], ea) + await self.sent_notif_update_throttle(ea["room_id"], ea) break else: if soonest_due_at is None or should_notify_at < soonest_due_at: @@ -227,14 +224,13 @@ class EmailPusher(object): self.seconds_until(soonest_due_at), self.on_timer ) - @defer.inlineCallbacks - def save_last_stream_ordering_and_success(self, last_stream_ordering): + async def save_last_stream_ordering_and_success(self, last_stream_ordering): if last_stream_ordering is None: # This happens if we haven't yet processed anything return self.last_stream_ordering = last_stream_ordering - pusher_still_exists = yield self.store.update_pusher_last_stream_ordering_and_success( + pusher_still_exists = await self.store.update_pusher_last_stream_ordering_and_success( self.app_id, self.email, self.user_id, @@ -275,13 +271,12 @@ class EmailPusher(object): may_send_at = last_sent_ts + throttle_ms return may_send_at - @defer.inlineCallbacks - def sent_notif_update_throttle(self, room_id, notified_push_action): + async def sent_notif_update_throttle(self, room_id, notified_push_action): # We have sent a notification, so update the throttle accordingly. # If the event that triggered the notif happened more than # THROTTLE_RESET_AFTER_MS after the previous one that triggered a # notif, we release the throttle. Otherwise, the throttle is increased. - time_of_previous_notifs = yield self.store.get_time_of_last_push_action_before( + time_of_previous_notifs = await self.store.get_time_of_last_push_action_before( notified_push_action["stream_ordering"] ) @@ -310,14 +305,13 @@ class EmailPusher(object): "last_sent_ts": self.clock.time_msec(), "throttle_ms": new_throttle_ms, } - yield self.store.set_throttle_params( + await self.store.set_throttle_params( self.pusher_id, room_id, self.throttle_params[room_id] ) - @defer.inlineCallbacks - def send_notification(self, push_actions, reason): + async def send_notification(self, push_actions, reason): logger.info("Sending notif email for user %r", self.user_id) - yield self.mailer.send_notification_mail( + await self.mailer.send_notification_mail( self.app_id, self.user_id, self.email, push_actions, reason ) diff --git a/synapse/push/mailer.py b/synapse/push/mailer.py index ab33abbeed..d57a66a697 100644 --- a/synapse/push/mailer.py +++ b/synapse/push/mailer.py @@ -26,8 +26,6 @@ from six.moves import urllib import bleach import jinja2 -from twisted.internet import defer - from synapse.api.constants import EventTypes from synapse.api.errors import StoreError from synapse.logging.context import make_deferred_yieldable @@ -127,8 +125,7 @@ class Mailer(object): logger.info("Created Mailer for app_name %s" % app_name) - @defer.inlineCallbacks - def send_password_reset_mail(self, email_address, token, client_secret, sid): + async def send_password_reset_mail(self, email_address, token, client_secret, sid): """Send an email with a password reset link to a user Args: @@ -149,14 +146,13 @@ class Mailer(object): template_vars = {"link": link} - yield self.send_email( + await self.send_email( email_address, "[%s] Password Reset" % self.hs.config.server_name, template_vars, ) - @defer.inlineCallbacks - def send_registration_mail(self, email_address, token, client_secret, sid): + async def send_registration_mail(self, email_address, token, client_secret, sid): """Send an email with a registration confirmation link to a user Args: @@ -177,14 +173,13 @@ class Mailer(object): template_vars = {"link": link} - yield self.send_email( + await self.send_email( email_address, "[%s] Register your Email Address" % self.hs.config.server_name, template_vars, ) - @defer.inlineCallbacks - def send_add_threepid_mail(self, email_address, token, client_secret, sid): + async def send_add_threepid_mail(self, email_address, token, client_secret, sid): """Send an email with a validation link to a user for adding a 3pid to their account Args: @@ -206,20 +201,19 @@ class Mailer(object): template_vars = {"link": link} - yield self.send_email( + await self.send_email( email_address, "[%s] Validate Your Email" % self.hs.config.server_name, template_vars, ) - @defer.inlineCallbacks - def send_notification_mail( + async def send_notification_mail( self, app_id, user_id, email_address, push_actions, reason ): """Send email regarding a user's room notifications""" rooms_in_order = deduped_ordered_list([pa["room_id"] for pa in push_actions]) - notif_events = yield self.store.get_events( + notif_events = await self.store.get_events( [pa["event_id"] for pa in push_actions] ) @@ -232,7 +226,7 @@ class Mailer(object): state_by_room = {} try: - user_display_name = yield self.store.get_profile_displayname( + user_display_name = await self.store.get_profile_displayname( UserID.from_string(user_id).localpart ) if user_display_name is None: @@ -240,14 +234,13 @@ class Mailer(object): except StoreError: user_display_name = user_id - @defer.inlineCallbacks - def _fetch_room_state(room_id): - room_state = yield self.store.get_current_state_ids(room_id) + async def _fetch_room_state(room_id): + room_state = await self.store.get_current_state_ids(room_id) state_by_room[room_id] = room_state # Run at most 3 of these at once: sync does 10 at a time but email # notifs are much less realtime than sync so we can afford to wait a bit. - yield concurrently_execute(_fetch_room_state, rooms_in_order, 3) + await concurrently_execute(_fetch_room_state, rooms_in_order, 3) # actually sort our so-called rooms_in_order list, most recent room first rooms_in_order.sort(key=lambda r: -(notifs_by_room[r][-1]["received_ts"] or 0)) @@ -255,19 +248,19 @@ class Mailer(object): rooms = [] for r in rooms_in_order: - roomvars = yield self.get_room_vars( + roomvars = await self.get_room_vars( r, user_id, notifs_by_room[r], notif_events, state_by_room[r] ) rooms.append(roomvars) - reason["room_name"] = yield calculate_room_name( + reason["room_name"] = await calculate_room_name( self.store, state_by_room[reason["room_id"]], user_id, fallback_to_members=True, ) - summary_text = yield self.make_summary_text( + summary_text = await self.make_summary_text( notifs_by_room, state_by_room, notif_events, user_id, reason ) @@ -282,12 +275,11 @@ class Mailer(object): "reason": reason, } - yield self.send_email( + await self.send_email( email_address, "[%s] %s" % (self.app_name, summary_text), template_vars ) - @defer.inlineCallbacks - def send_email(self, email_address, subject, template_vars): + async def send_email(self, email_address, subject, template_vars): """Send an email with the given information and template text""" try: from_string = self.hs.config.email_notif_from % {"app": self.app_name} @@ -317,7 +309,7 @@ class Mailer(object): logger.info("Sending email to %s" % email_address) - yield make_deferred_yieldable( + await make_deferred_yieldable( self.sendmail( self.hs.config.email_smtp_host, raw_from, @@ -332,13 +324,14 @@ class Mailer(object): ) ) - @defer.inlineCallbacks - def get_room_vars(self, room_id, user_id, notifs, notif_events, room_state_ids): + async def get_room_vars( + self, room_id, user_id, notifs, notif_events, room_state_ids + ): my_member_event_id = room_state_ids[("m.room.member", user_id)] - my_member_event = yield self.store.get_event(my_member_event_id) + my_member_event = await self.store.get_event(my_member_event_id) is_invite = my_member_event.content["membership"] == "invite" - room_name = yield calculate_room_name(self.store, room_state_ids, user_id) + room_name = await calculate_room_name(self.store, room_state_ids, user_id) room_vars = { "title": room_name, @@ -350,7 +343,7 @@ class Mailer(object): if not is_invite: for n in notifs: - notifvars = yield self.get_notif_vars( + notifvars = await self.get_notif_vars( n, user_id, notif_events[n["event_id"]], room_state_ids ) @@ -377,9 +370,8 @@ class Mailer(object): return room_vars - @defer.inlineCallbacks - def get_notif_vars(self, notif, user_id, notif_event, room_state_ids): - results = yield self.store.get_events_around( + async def get_notif_vars(self, notif, user_id, notif_event, room_state_ids): + results = await self.store.get_events_around( notif["room_id"], notif["event_id"], before_limit=CONTEXT_BEFORE, @@ -392,25 +384,24 @@ class Mailer(object): "messages": [], } - the_events = yield filter_events_for_client( + the_events = await filter_events_for_client( self.storage, user_id, results["events_before"] ) the_events.append(notif_event) for event in the_events: - messagevars = yield self.get_message_vars(notif, event, room_state_ids) + messagevars = await self.get_message_vars(notif, event, room_state_ids) if messagevars is not None: ret["messages"].append(messagevars) return ret - @defer.inlineCallbacks - def get_message_vars(self, notif, event, room_state_ids): + async def get_message_vars(self, notif, event, room_state_ids): if event.type != EventTypes.Message: return sender_state_event_id = room_state_ids[("m.room.member", event.sender)] - sender_state_event = yield self.store.get_event(sender_state_event_id) + sender_state_event = await self.store.get_event(sender_state_event_id) sender_name = name_from_member_event(sender_state_event) sender_avatar_url = sender_state_event.content.get("avatar_url") @@ -460,8 +451,7 @@ class Mailer(object): return messagevars - @defer.inlineCallbacks - def make_summary_text( + async def make_summary_text( self, notifs_by_room, room_state_ids, notif_events, user_id, reason ): if len(notifs_by_room) == 1: @@ -471,17 +461,17 @@ class Mailer(object): # If the room has some kind of name, use it, but we don't # want the generated-from-names one here otherwise we'll # end up with, "new message from Bob in the Bob room" - room_name = yield calculate_room_name( + room_name = await calculate_room_name( self.store, room_state_ids[room_id], user_id, fallback_to_members=False ) my_member_event_id = room_state_ids[room_id][("m.room.member", user_id)] - my_member_event = yield self.store.get_event(my_member_event_id) + my_member_event = await self.store.get_event(my_member_event_id) if my_member_event.content["membership"] == "invite": inviter_member_event_id = room_state_ids[room_id][ ("m.room.member", my_member_event.sender) ] - inviter_member_event = yield self.store.get_event( + inviter_member_event = await self.store.get_event( inviter_member_event_id ) inviter_name = name_from_member_event(inviter_member_event) @@ -506,7 +496,7 @@ class Mailer(object): state_event_id = room_state_ids[room_id][ ("m.room.member", event.sender) ] - state_event = yield self.store.get_event(state_event_id) + state_event = await self.store.get_event(state_event_id) sender_name = name_from_member_event(state_event) if sender_name is not None and room_name is not None: @@ -535,7 +525,7 @@ class Mailer(object): } ) - member_events = yield self.store.get_events( + member_events = await self.store.get_events( [ room_state_ids[room_id][("m.room.member", s)] for s in sender_ids @@ -567,7 +557,7 @@ class Mailer(object): } ) - member_events = yield self.store.get_events( + member_events = await self.store.get_events( [room_state_ids[room_id][("m.room.member", s)] for s in sender_ids] ) diff --git a/tests/rest/client/v2_alpha/test_account.py b/tests/rest/client/v2_alpha/test_account.py index 0d6936fd36..3ab611f618 100644 --- a/tests/rest/client/v2_alpha/test_account.py +++ b/tests/rest/client/v2_alpha/test_account.py @@ -46,7 +46,7 @@ class PasswordResetTestCase(unittest.HomeserverTestCase): # Email config. self.email_attempts = [] - def sendmail(smtphost, from_addr, to_addrs, msg, **kwargs): + async def sendmail(smtphost, from_addr, to_addrs, msg, **kwargs): self.email_attempts.append(msg) return @@ -358,7 +358,7 @@ class ThreepidEmailRestTestCase(unittest.HomeserverTestCase): # Email config. self.email_attempts = [] - def sendmail(smtphost, from_addr, to_addrs, msg, **kwargs): + async def sendmail(smtphost, from_addr, to_addrs, msg, **kwargs): self.email_attempts.append(msg) config["email"] = { -- cgit 1.4.1 From 1531b214fc57714c14046a8f66c7b5fe5ec5dcdd Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Fri, 22 May 2020 14:21:54 +0100 Subject: Add ability to wait for replication streams (#7542) The idea here is that if an instance persists an event via the replication HTTP API it can return before we receive that event over replication, which can lead to races where code assumes that persisting an event immediately updates various caches (e.g. current state of the room). Most of Synapse doesn't hit such races, so we don't do the waiting automagically, instead we do so where necessary to avoid unnecessary delays. We may decide to change our minds here if it turns out there are a lot of subtle races going on. People probably want to look at this commit by commit. --- changelog.d/7542.misc | 1 + synapse/handlers/federation.py | 33 ++++++--- synapse/handlers/message.py | 36 ++++++--- synapse/handlers/room.py | 65 +++++++++++----- synapse/handlers/room_member.py | 65 ++++++++++------ synapse/handlers/room_member_worker.py | 11 +-- synapse/replication/http/federation.py | 13 +++- synapse/replication/http/membership.py | 14 ++-- synapse/replication/http/send_event.py | 4 +- synapse/replication/http/streams.py | 5 +- synapse/replication/tcp/client.py | 90 ++++++++++++++++++++++- synapse/rest/admin/rooms.py | 10 ++- synapse/rest/client/v1/room.py | 20 +++-- synapse/rest/client/v2_alpha/relations.py | 2 +- synapse/server.py | 5 ++ synapse/server.pyi | 5 ++ synapse/server_notices/server_notices_manager.py | 6 +- synapse/storage/data_stores/main/events_worker.py | 6 +- synapse/storage/data_stores/main/roommember.py | 2 + tests/federation/test_complexity.py | 8 +- tests/handlers/test_typing.py | 5 +- tests/storage/test_cleanup_extrems.py | 4 +- tests/storage/test_event_metrics.py | 2 +- tests/test_federation.py | 4 +- 24 files changed, 304 insertions(+), 112 deletions(-) create mode 100644 changelog.d/7542.misc (limited to 'tests') diff --git a/changelog.d/7542.misc b/changelog.d/7542.misc new file mode 100644 index 0000000000..7dd9b4823b --- /dev/null +++ b/changelog.d/7542.misc @@ -0,0 +1 @@ +Add ability to wait for replication streams. diff --git a/synapse/handlers/federation.py b/synapse/handlers/federation.py index bb03cc9add..e354c803db 100644 --- a/synapse/handlers/federation.py +++ b/synapse/handlers/federation.py @@ -126,6 +126,7 @@ class FederationHandler(BaseHandler): self._server_notices_mxid = hs.config.server_notices_mxid self.config = hs.config self.http_client = hs.get_simple_http_client() + self._replication = hs.get_replication_data_handler() self._send_events_to_master = ReplicationFederationSendEventsRestServlet.make_client( hs @@ -1221,7 +1222,7 @@ class FederationHandler(BaseHandler): async def do_invite_join( self, target_hosts: Iterable[str], room_id: str, joinee: str, content: JsonDict - ) -> None: + ) -> Tuple[str, int]: """ Attempts to join the `joinee` to the room `room_id` via the servers contained in `target_hosts`. @@ -1304,15 +1305,23 @@ class FederationHandler(BaseHandler): room_id=room_id, room_version=room_version_obj, ) - await self._persist_auth_tree( + max_stream_id = await self._persist_auth_tree( origin, auth_chain, state, event, room_version_obj ) + # We wait here until this instance has seen the events come down + # replication (if we're using replication) as the below uses caches. + # + # TODO: Currently the events stream is written to from master + await self._replication.wait_for_stream_position( + "master", "events", max_stream_id + ) + # Check whether this room is the result of an upgrade of a room we already know # about. If so, migrate over user information predecessor = await self.store.get_room_predecessor(room_id) if not predecessor or not isinstance(predecessor.get("room_id"), str): - return + return event.event_id, max_stream_id old_room_id = predecessor["room_id"] logger.debug( "Found predecessor for %s during remote join: %s", room_id, old_room_id @@ -1325,6 +1334,7 @@ class FederationHandler(BaseHandler): ) logger.debug("Finished joining %s to %s", joinee, room_id) + return event.event_id, max_stream_id finally: room_queue = self.room_queues[room_id] del self.room_queues[room_id] @@ -1554,7 +1564,7 @@ class FederationHandler(BaseHandler): async def do_remotely_reject_invite( self, target_hosts: Iterable[str], room_id: str, user_id: str, content: JsonDict - ) -> EventBase: + ) -> Tuple[EventBase, int]: origin, event, room_version = await self._make_and_verify_event( target_hosts, room_id, user_id, "leave", content=content ) @@ -1574,9 +1584,9 @@ class FederationHandler(BaseHandler): await self.federation_client.send_leave(target_hosts, event) context = await self.state_handler.compute_event_context(event) - await self.persist_events_and_notify([(event, context)]) + stream_id = await self.persist_events_and_notify([(event, context)]) - return event + return event, stream_id async def _make_and_verify_event( self, @@ -1888,7 +1898,7 @@ class FederationHandler(BaseHandler): state: List[EventBase], event: EventBase, room_version: RoomVersion, - ) -> None: + ) -> int: """Checks the auth chain is valid (and passes auth checks) for the state and event. Then persists the auth chain and state atomically. Persists the event separately. Notifies about the persisted events @@ -1982,7 +1992,7 @@ class FederationHandler(BaseHandler): event, old_state=state ) - await self.persist_events_and_notify([(event, new_event_context)]) + return await self.persist_events_and_notify([(event, new_event_context)]) async def _prep_event( self, @@ -2835,7 +2845,7 @@ class FederationHandler(BaseHandler): self, event_and_contexts: Sequence[Tuple[EventBase, EventContext]], backfilled: bool = False, - ) -> None: + ) -> int: """Persists events and tells the notifier/pushers about them, if necessary. @@ -2845,11 +2855,12 @@ class FederationHandler(BaseHandler): backfilling or not """ if self.config.worker_app: - await self._send_events_to_master( + result = await self._send_events_to_master( store=self.store, event_and_contexts=event_and_contexts, backfilled=backfilled, ) + return result["max_stream_id"] else: max_stream_id = await self.storage.persistence.persist_events( event_and_contexts, backfilled=backfilled @@ -2864,6 +2875,8 @@ class FederationHandler(BaseHandler): for event, _ in event_and_contexts: await self._notify_persisted_event(event, max_stream_id) + return max_stream_id + async def _notify_persisted_event( self, event: EventBase, max_stream_id: int ) -> None: diff --git a/synapse/handlers/message.py b/synapse/handlers/message.py index 8f362896a2..f445e2aa2a 100644 --- a/synapse/handlers/message.py +++ b/synapse/handlers/message.py @@ -15,7 +15,7 @@ # See the License for the specific language governing permissions and # limitations under the License. import logging -from typing import Optional +from typing import Optional, Tuple from six import iteritems, itervalues, string_types @@ -42,6 +42,7 @@ from synapse.api.errors import ( ) from synapse.api.room_versions import KNOWN_ROOM_VERSIONS, RoomVersions from synapse.api.urls import ConsentURIBuilder +from synapse.events import EventBase from synapse.events.validator import EventValidator from synapse.logging.context import run_in_background from synapse.metrics.background_process_metrics import run_as_background_process @@ -630,7 +631,9 @@ class EventCreationHandler(object): msg = self._block_events_without_consent_error % {"consent_uri": consent_uri} raise ConsentNotGivenError(msg=msg, consent_uri=consent_uri) - async def send_nonmember_event(self, requester, event, context, ratelimit=True): + async def send_nonmember_event( + self, requester, event, context, ratelimit=True + ) -> int: """ Persists and notifies local clients and federation of an event. @@ -639,6 +642,9 @@ class EventCreationHandler(object): context (Context) the context of the event. ratelimit (bool): Whether to rate limit this send. is_guest (bool): Whether the sender is a guest. + + Return: + The stream_id of the persisted event. """ if event.type == EventTypes.Member: raise SynapseError( @@ -659,7 +665,7 @@ class EventCreationHandler(object): ) return prev_state - await self.handle_new_client_event( + return await self.handle_new_client_event( requester=requester, event=event, context=context, ratelimit=ratelimit ) @@ -688,7 +694,7 @@ class EventCreationHandler(object): async def create_and_send_nonmember_event( self, requester, event_dict, ratelimit=True, txn_id=None - ): + ) -> Tuple[EventBase, int]: """ Creates an event, then sends it. @@ -711,10 +717,10 @@ class EventCreationHandler(object): spam_error = "Spam is not permitted here" raise SynapseError(403, spam_error, Codes.FORBIDDEN) - await self.send_nonmember_event( + stream_id = await self.send_nonmember_event( requester, event, context, ratelimit=ratelimit ) - return event + return event, stream_id @measure_func("create_new_client_event") @defer.inlineCallbacks @@ -774,7 +780,7 @@ class EventCreationHandler(object): @measure_func("handle_new_client_event") async def handle_new_client_event( self, requester, event, context, ratelimit=True, extra_users=[] - ): + ) -> int: """Processes a new event. This includes checking auth, persisting it, notifying users, sending to remote servers, etc. @@ -787,6 +793,9 @@ class EventCreationHandler(object): context (EventContext) ratelimit (bool) extra_users (list(UserID)): Any extra users to notify about event + + Return: + The stream_id of the persisted event. """ if event.is_state() and (event.type, event.state_key) == ( @@ -827,7 +836,7 @@ class EventCreationHandler(object): try: # If we're a worker we need to hit out to the master. if self.config.worker_app: - await self.send_event_to_master( + result = await self.send_event_to_master( event_id=event.event_id, store=self.store, requester=requester, @@ -836,14 +845,17 @@ class EventCreationHandler(object): ratelimit=ratelimit, extra_users=extra_users, ) + stream_id = result["stream_id"] + event.internal_metadata.stream_ordering = stream_id success = True - return + return stream_id - await self.persist_and_notify_client_event( + stream_id = await self.persist_and_notify_client_event( requester, event, context, ratelimit=ratelimit, extra_users=extra_users ) success = True + return stream_id finally: if not success: # Ensure that we actually remove the entries in the push actions @@ -886,7 +898,7 @@ class EventCreationHandler(object): async def persist_and_notify_client_event( self, requester, event, context, ratelimit=True, extra_users=[] - ): + ) -> int: """Called when we have fully built the event, have already calculated the push actions for the event, and checked auth. @@ -1076,6 +1088,8 @@ class EventCreationHandler(object): # matters as sometimes presence code can take a while. run_in_background(self._bump_active_time, requester.user) + return event_stream_id + async def _bump_active_time(self, user): try: presence = self.hs.get_presence_handler() diff --git a/synapse/handlers/room.py b/synapse/handlers/room.py index 13850ba672..2698a129ca 100644 --- a/synapse/handlers/room.py +++ b/synapse/handlers/room.py @@ -22,6 +22,7 @@ import logging import math import string from collections import OrderedDict +from typing import Tuple from six import iteritems, string_types @@ -518,7 +519,7 @@ class RoomCreationHandler(BaseHandler): async def create_room( self, requester, config, ratelimit=True, creator_join_profile=None - ): + ) -> Tuple[dict, int]: """ Creates a new room. Args: @@ -535,9 +536,9 @@ class RoomCreationHandler(BaseHandler): `avatar_url` and/or `displayname`. Returns: - Deferred[dict]: - a dict containing the keys `room_id` and, if an alias was - requested, `room_alias`. + First, a dict containing the keys `room_id` and, if an alias + was, requested, `room_alias`. Secondly, the stream_id of the + last persisted event. Raises: SynapseError if the room ID couldn't be stored, or something went horribly wrong. @@ -669,7 +670,7 @@ class RoomCreationHandler(BaseHandler): # override any attempt to set room versions via the creation_content creation_content["room_version"] = room_version.identifier - await self._send_events_for_new_room( + last_stream_id = await self._send_events_for_new_room( requester, room_id, preset_config=preset_config, @@ -683,7 +684,10 @@ class RoomCreationHandler(BaseHandler): if "name" in config: name = config["name"] - await self.event_creation_handler.create_and_send_nonmember_event( + ( + _, + last_stream_id, + ) = await self.event_creation_handler.create_and_send_nonmember_event( requester, { "type": EventTypes.Name, @@ -697,7 +701,10 @@ class RoomCreationHandler(BaseHandler): if "topic" in config: topic = config["topic"] - await self.event_creation_handler.create_and_send_nonmember_event( + ( + _, + last_stream_id, + ) = await self.event_creation_handler.create_and_send_nonmember_event( requester, { "type": EventTypes.Topic, @@ -715,7 +722,7 @@ class RoomCreationHandler(BaseHandler): if is_direct: content["is_direct"] = is_direct - await self.room_member_handler.update_membership( + _, last_stream_id = await self.room_member_handler.update_membership( requester, UserID.from_string(invitee), room_id, @@ -729,7 +736,7 @@ class RoomCreationHandler(BaseHandler): id_access_token = invite_3pid.get("id_access_token") # optional address = invite_3pid["address"] medium = invite_3pid["medium"] - await self.hs.get_room_member_handler().do_3pid_invite( + last_stream_id = await self.hs.get_room_member_handler().do_3pid_invite( room_id, requester.user, medium, @@ -745,7 +752,7 @@ class RoomCreationHandler(BaseHandler): if room_alias: result["room_alias"] = room_alias.to_string() - return result + return result, last_stream_id async def _send_events_for_new_room( self, @@ -758,7 +765,13 @@ class RoomCreationHandler(BaseHandler): room_alias=None, power_level_content_override=None, # Doesn't apply when initial state has power level state event content creator_join_profile=None, - ): + ) -> int: + """Sends the initial events into a new room. + + Returns: + The stream_id of the last event persisted. + """ + def create(etype, content, **kwargs): e = {"type": etype, "content": content} @@ -767,12 +780,16 @@ class RoomCreationHandler(BaseHandler): return e - async def send(etype, content, **kwargs): + async def send(etype, content, **kwargs) -> int: event = create(etype, content, **kwargs) logger.debug("Sending %s in new room", etype) - await self.event_creation_handler.create_and_send_nonmember_event( + ( + _, + last_stream_id, + ) = await self.event_creation_handler.create_and_send_nonmember_event( creator, event, ratelimit=False ) + return last_stream_id config = RoomCreationHandler.PRESETS_DICT[preset_config] @@ -797,7 +814,9 @@ class RoomCreationHandler(BaseHandler): # of the first events that get sent into a room. pl_content = initial_state.pop((EventTypes.PowerLevels, ""), None) if pl_content is not None: - await send(etype=EventTypes.PowerLevels, content=pl_content) + last_sent_stream_id = await send( + etype=EventTypes.PowerLevels, content=pl_content + ) else: power_level_content = { "users": {creator_id: 100}, @@ -830,33 +849,39 @@ class RoomCreationHandler(BaseHandler): if power_level_content_override: power_level_content.update(power_level_content_override) - await send(etype=EventTypes.PowerLevels, content=power_level_content) + last_sent_stream_id = await send( + etype=EventTypes.PowerLevels, content=power_level_content + ) if room_alias and (EventTypes.CanonicalAlias, "") not in initial_state: - await send( + last_sent_stream_id = await send( etype=EventTypes.CanonicalAlias, content={"alias": room_alias.to_string()}, ) if (EventTypes.JoinRules, "") not in initial_state: - await send( + last_sent_stream_id = await send( etype=EventTypes.JoinRules, content={"join_rule": config["join_rules"]} ) if (EventTypes.RoomHistoryVisibility, "") not in initial_state: - await send( + last_sent_stream_id = await send( etype=EventTypes.RoomHistoryVisibility, content={"history_visibility": config["history_visibility"]}, ) if config["guest_can_join"]: if (EventTypes.GuestAccess, "") not in initial_state: - await send( + last_sent_stream_id = await send( etype=EventTypes.GuestAccess, content={"guest_access": "can_join"} ) for (etype, state_key), content in initial_state.items(): - await send(etype=etype, state_key=state_key, content=content) + last_sent_stream_id = await send( + etype=etype, state_key=state_key, content=content + ) + + return last_sent_stream_id async def _generate_room_id( self, creator_id: str, is_public: str, room_version: RoomVersion, diff --git a/synapse/handlers/room_member.py b/synapse/handlers/room_member.py index e51e1c32fe..691b6705b2 100644 --- a/synapse/handlers/room_member.py +++ b/synapse/handlers/room_member.py @@ -17,7 +17,7 @@ import abc import logging -from typing import Dict, Iterable, List, Optional, Tuple, Union +from typing import Dict, Iterable, List, Optional, Tuple from six.moves import http_client @@ -84,7 +84,7 @@ class RoomMemberHandler(object): room_id: str, user: UserID, content: dict, - ) -> Optional[dict]: + ) -> Tuple[str, int]: """Try and join a room that this server is not in Args: @@ -104,7 +104,7 @@ class RoomMemberHandler(object): room_id: str, target: UserID, content: dict, - ) -> dict: + ) -> Tuple[Optional[str], int]: """Attempt to reject an invite for a room this server is not in. If we fail to do so we locally mark the invite as rejected. @@ -154,7 +154,7 @@ class RoomMemberHandler(object): ratelimit: bool = True, content: Optional[dict] = None, require_consent: bool = True, - ) -> EventBase: + ) -> Tuple[str, int]: user_id = target.to_string() if content is None: @@ -187,9 +187,10 @@ class RoomMemberHandler(object): ) if duplicate is not None: # Discard the new event since this membership change is a no-op. - return duplicate + _, stream_id = await self.store.get_event_ordering(duplicate.event_id) + return duplicate.event_id, stream_id - await self.event_creation_handler.handle_new_client_event( + stream_id = await self.event_creation_handler.handle_new_client_event( requester, event, context, extra_users=[target], ratelimit=ratelimit ) @@ -213,7 +214,7 @@ class RoomMemberHandler(object): if prev_member_event.membership == Membership.JOIN: await self._user_left_room(target, room_id) - return event + return event.event_id, stream_id async def copy_room_tags_and_direct_to_room( self, old_room_id, new_room_id, user_id @@ -263,7 +264,7 @@ class RoomMemberHandler(object): ratelimit: bool = True, content: Optional[dict] = None, require_consent: bool = True, - ) -> Union[EventBase, Optional[dict]]: + ) -> Tuple[Optional[str], int]: key = (room_id,) with (await self.member_linearizer.queue(key)): @@ -294,7 +295,7 @@ class RoomMemberHandler(object): ratelimit: bool = True, content: Optional[dict] = None, require_consent: bool = True, - ) -> Union[EventBase, Optional[dict]]: + ) -> Tuple[Optional[str], int]: content_specified = bool(content) if content is None: content = {} @@ -398,7 +399,13 @@ class RoomMemberHandler(object): same_membership = old_membership == effective_membership_state same_sender = requester.user.to_string() == old_state.sender if same_sender and same_membership and same_content: - return old_state + _, stream_id = await self.store.get_event_ordering( + old_state.event_id + ) + return ( + old_state.event_id, + stream_id, + ) if old_membership in ["ban", "leave"] and action == "kick": raise AuthError(403, "The target user is not in the room") @@ -705,7 +712,7 @@ class RoomMemberHandler(object): requester: Requester, txn_id: Optional[str], id_access_token: Optional[str] = None, - ) -> None: + ) -> int: if self.config.block_non_admin_invites: is_requester_admin = await self.auth.is_server_admin(requester.user) if not is_requester_admin: @@ -737,11 +744,11 @@ class RoomMemberHandler(object): ) if invitee: - await self.update_membership( + _, stream_id = await self.update_membership( requester, UserID.from_string(invitee), room_id, "invite", txn_id=txn_id ) else: - await self._make_and_store_3pid_invite( + stream_id = await self._make_and_store_3pid_invite( requester, id_server, medium, @@ -752,6 +759,8 @@ class RoomMemberHandler(object): id_access_token=id_access_token, ) + return stream_id + async def _make_and_store_3pid_invite( self, requester: Requester, @@ -762,7 +771,7 @@ class RoomMemberHandler(object): user: UserID, txn_id: Optional[str], id_access_token: Optional[str] = None, - ) -> None: + ) -> int: room_state = await self.state_handler.get_current_state(room_id) inviter_display_name = "" @@ -817,7 +826,10 @@ class RoomMemberHandler(object): id_access_token=id_access_token, ) - await self.event_creation_handler.create_and_send_nonmember_event( + ( + event, + stream_id, + ) = await self.event_creation_handler.create_and_send_nonmember_event( requester, { "type": EventTypes.ThirdPartyInvite, @@ -835,6 +847,7 @@ class RoomMemberHandler(object): ratelimit=False, txn_id=txn_id, ) + return stream_id async def _is_host_in_room( self, current_state_ids: Dict[Tuple[str, str], str] @@ -916,7 +929,7 @@ class RoomMemberMasterHandler(RoomMemberHandler): room_id: str, user: UserID, content: dict, - ) -> None: + ) -> Tuple[str, int]: """Implements RoomMemberHandler._remote_join """ # filter ourselves out of remote_room_hosts: do_invite_join ignores it @@ -945,7 +958,7 @@ class RoomMemberMasterHandler(RoomMemberHandler): # join dance for now, since we're kinda implicitly checking # that we are allowed to join when we decide whether or not we # need to do the invite/join dance. - await self.federation_handler.do_invite_join( + event_id, stream_id = await self.federation_handler.do_invite_join( remote_room_hosts, room_id, user.to_string(), content ) await self._user_joined_room(user, room_id) @@ -955,14 +968,14 @@ class RoomMemberMasterHandler(RoomMemberHandler): if self.hs.config.limit_remote_rooms.enabled: if too_complex is False: # We checked, and we're under the limit. - return + return event_id, stream_id # Check again, but with the local state events too_complex = await self._is_local_room_too_complex(room_id) if too_complex is False: # We're under the limit. - return + return event_id, stream_id # The room is too large. Leave. requester = types.create_requester(user, None, False, None) @@ -975,6 +988,8 @@ class RoomMemberMasterHandler(RoomMemberHandler): errcode=Codes.RESOURCE_LIMIT_EXCEEDED, ) + return event_id, stream_id + async def _remote_reject_invite( self, requester: Requester, @@ -982,15 +997,15 @@ class RoomMemberMasterHandler(RoomMemberHandler): room_id: str, target: UserID, content: dict, - ) -> dict: + ) -> Tuple[Optional[str], int]: """Implements RoomMemberHandler._remote_reject_invite """ fed_handler = self.federation_handler try: - ret = await fed_handler.do_remotely_reject_invite( + event, stream_id = await fed_handler.do_remotely_reject_invite( remote_room_hosts, room_id, target.to_string(), content=content, ) - return ret + return event.event_id, stream_id except Exception as e: # if we were unable to reject the exception, just mark # it as rejected on our end and plough ahead. @@ -1000,8 +1015,10 @@ class RoomMemberMasterHandler(RoomMemberHandler): # logger.warning("Failed to reject invite: %s", e) - await self.store.locally_reject_invite(target.to_string(), room_id) - return {} + stream_id = await self.store.locally_reject_invite( + target.to_string(), room_id + ) + return None, stream_id async def _user_joined_room(self, target: UserID, room_id: str) -> None: """Implements RoomMemberHandler._user_joined_room diff --git a/synapse/handlers/room_member_worker.py b/synapse/handlers/room_member_worker.py index 5c776cc0be..02e0c4103d 100644 --- a/synapse/handlers/room_member_worker.py +++ b/synapse/handlers/room_member_worker.py @@ -14,7 +14,7 @@ # limitations under the License. import logging -from typing import List, Optional +from typing import List, Optional, Tuple from synapse.api.errors import SynapseError from synapse.handlers.room_member import RoomMemberHandler @@ -43,7 +43,7 @@ class RoomMemberWorkerHandler(RoomMemberHandler): room_id: str, user: UserID, content: dict, - ) -> Optional[dict]: + ) -> Tuple[str, int]: """Implements RoomMemberHandler._remote_join """ if len(remote_room_hosts) == 0: @@ -59,7 +59,7 @@ class RoomMemberWorkerHandler(RoomMemberHandler): await self._user_joined_room(user, room_id) - return ret + return ret["event_id"], ret["stream_id"] async def _remote_reject_invite( self, @@ -68,16 +68,17 @@ class RoomMemberWorkerHandler(RoomMemberHandler): room_id: str, target: UserID, content: dict, - ) -> dict: + ) -> Tuple[Optional[str], int]: """Implements RoomMemberHandler._remote_reject_invite """ - return await self._remote_reject_client( + ret = await self._remote_reject_client( requester=requester, remote_room_hosts=remote_room_hosts, room_id=room_id, user_id=target.to_string(), content=content, ) + return ret["event_id"], ret["stream_id"] async def _user_joined_room(self, target: UserID, room_id: str) -> None: """Implements RoomMemberHandler._user_joined_room diff --git a/synapse/replication/http/federation.py b/synapse/replication/http/federation.py index 7e23b565b9..c287c4e269 100644 --- a/synapse/replication/http/federation.py +++ b/synapse/replication/http/federation.py @@ -29,7 +29,7 @@ logger = logging.getLogger(__name__) class ReplicationFederationSendEventsRestServlet(ReplicationEndpoint): """Handles events newly received from federation, including persisting and - notifying. + notifying. Returns the maximum stream ID of the persisted events. The API looks like: @@ -46,6 +46,13 @@ class ReplicationFederationSendEventsRestServlet(ReplicationEndpoint): "context": { .. serialized event context .. }, }], "backfilled": false + } + + 200 OK + + { + "max_stream_id": 32443, + } """ NAME = "fed_send_events" @@ -115,11 +122,11 @@ class ReplicationFederationSendEventsRestServlet(ReplicationEndpoint): logger.info("Got %d events from federation", len(event_and_contexts)) - await self.federation_handler.persist_events_and_notify( + max_stream_id = await self.federation_handler.persist_events_and_notify( event_and_contexts, backfilled ) - return 200, {} + return 200, {"max_stream_id": max_stream_id} class ReplicationFederationSendEduRestServlet(ReplicationEndpoint): diff --git a/synapse/replication/http/membership.py b/synapse/replication/http/membership.py index 3577611fd7..050fd34562 100644 --- a/synapse/replication/http/membership.py +++ b/synapse/replication/http/membership.py @@ -76,11 +76,11 @@ class ReplicationRemoteJoinRestServlet(ReplicationEndpoint): logger.info("remote_join: %s into room: %s", user_id, room_id) - await self.federation_handler.do_invite_join( + event_id, stream_id = await self.federation_handler.do_invite_join( remote_room_hosts, room_id, user_id, event_content ) - return 200, {} + return 200, {"event_id": event_id, "stream_id": stream_id} class ReplicationRemoteRejectInviteRestServlet(ReplicationEndpoint): @@ -136,10 +136,10 @@ class ReplicationRemoteRejectInviteRestServlet(ReplicationEndpoint): logger.info("remote_reject_invite: %s out of room: %s", user_id, room_id) try: - event = await self.federation_handler.do_remotely_reject_invite( + event, stream_id = await self.federation_handler.do_remotely_reject_invite( remote_room_hosts, room_id, user_id, event_content, ) - ret = event.get_pdu_json() + event_id = event.event_id except Exception as e: # if we were unable to reject the exception, just mark # it as rejected on our end and plough ahead. @@ -149,10 +149,10 @@ class ReplicationRemoteRejectInviteRestServlet(ReplicationEndpoint): # logger.warning("Failed to reject invite: %s", e) - await self.store.locally_reject_invite(user_id, room_id) - ret = {} + stream_id = await self.store.locally_reject_invite(user_id, room_id) + event_id = None - return 200, ret + return 200, {"event_id": event_id, "stream_id": stream_id} class ReplicationUserJoinedLeftRoomRestServlet(ReplicationEndpoint): diff --git a/synapse/replication/http/send_event.py b/synapse/replication/http/send_event.py index b74b088ff4..c981723c1a 100644 --- a/synapse/replication/http/send_event.py +++ b/synapse/replication/http/send_event.py @@ -119,11 +119,11 @@ class ReplicationSendEventRestServlet(ReplicationEndpoint): "Got event to send with ID: %s into room: %s", event.event_id, event.room_id ) - await self.event_creation_handler.persist_and_notify_client_event( + stream_id = await self.event_creation_handler.persist_and_notify_client_event( requester, event, context, ratelimit=ratelimit, extra_users=extra_users ) - return 200, {} + return 200, {"stream_id": stream_id} def register_servlets(hs, http_server): diff --git a/synapse/replication/http/streams.py b/synapse/replication/http/streams.py index b705a8e16c..bde97eef32 100644 --- a/synapse/replication/http/streams.py +++ b/synapse/replication/http/streams.py @@ -51,10 +51,7 @@ class ReplicationGetStreamUpdates(ReplicationEndpoint): super().__init__(hs) self._instance_name = hs.get_instance_name() - - # We pull the streams from the replication handler (if we try and make - # them ourselves we end up in an import loop). - self.streams = hs.get_tcp_replication().get_streams() + self.streams = hs.get_replication_streams() @staticmethod def _serialize_payload(stream_name, from_token, upto_token): diff --git a/synapse/replication/tcp/client.py b/synapse/replication/tcp/client.py index 28826302f5..508ad1b720 100644 --- a/synapse/replication/tcp/client.py +++ b/synapse/replication/tcp/client.py @@ -14,19 +14,23 @@ # limitations under the License. """A replication client for use by synapse workers. """ - +import heapq import logging -from typing import TYPE_CHECKING, Tuple +from typing import TYPE_CHECKING, Dict, List, Tuple +from twisted.internet.defer import Deferred from twisted.internet.protocol import ReconnectingClientFactory from synapse.api.constants import EventTypes +from synapse.logging.context import PreserveLoggingContext, make_deferred_yieldable from synapse.replication.tcp.protocol import ClientReplicationStreamProtocol from synapse.replication.tcp.streams.events import ( EventsStream, EventsStreamEventRow, EventsStreamRow, ) +from synapse.util.async_helpers import timeout_deferred +from synapse.util.metrics import Measure if TYPE_CHECKING: from synapse.server import HomeServer @@ -35,6 +39,10 @@ if TYPE_CHECKING: logger = logging.getLogger(__name__) +# How long we allow callers to wait for replication updates before timing out. +_WAIT_FOR_REPLICATION_TIMEOUT_SECONDS = 30 + + class DirectTcpReplicationClientFactory(ReconnectingClientFactory): """Factory for building connections to the master. Will reconnect if the connection is lost. @@ -92,6 +100,16 @@ class ReplicationDataHandler: self.store = hs.get_datastore() self.pusher_pool = hs.get_pusherpool() self.notifier = hs.get_notifier() + self._reactor = hs.get_reactor() + self._clock = hs.get_clock() + self._streams = hs.get_replication_streams() + self._instance_name = hs.get_instance_name() + + # Map from stream to list of deferreds waiting for the stream to + # arrive at a particular position. The lists are sorted by stream position. + self._streams_to_waiters = ( + {} + ) # type: Dict[str, List[Tuple[int, Deferred[None]]]] async def on_rdata( self, stream_name: str, instance_name: str, token: int, rows: list @@ -131,8 +149,76 @@ class ReplicationDataHandler: await self.pusher_pool.on_new_notifications(token, token) + # Notify any waiting deferreds. The list is ordered by position so we + # just iterate through the list until we reach a position that is + # greater than the received row position. + waiting_list = self._streams_to_waiters.get(stream_name, []) + + # Index of first item with a position after the current token, i.e we + # have called all deferreds before this index. If not overwritten by + # loop below means either a) no items in list so no-op or b) all items + # in list were called and so the list should be cleared. Setting it to + # `len(list)` works for both cases. + index_of_first_deferred_not_called = len(waiting_list) + + for idx, (position, deferred) in enumerate(waiting_list): + if position <= token: + try: + with PreserveLoggingContext(): + deferred.callback(None) + except Exception: + # The deferred has been cancelled or timed out. + pass + else: + # The list is sorted by position so we don't need to continue + # checking any futher entries in the list. + index_of_first_deferred_not_called = idx + break + + # Drop all entries in the waiting list that were called in the above + # loop. (This maintains the order so no need to resort) + waiting_list[:] = waiting_list[index_of_first_deferred_not_called:] + async def on_position(self, stream_name: str, instance_name: str, token: int): self.store.process_replication_rows(stream_name, instance_name, token, []) def on_remote_server_up(self, server: str): """Called when get a new REMOTE_SERVER_UP command.""" + + async def wait_for_stream_position( + self, instance_name: str, stream_name: str, position: int + ): + """Wait until this instance has received updates up to and including + the given stream position. + """ + + if instance_name == self._instance_name: + # We don't get told about updates written by this process, and + # anyway in that case we don't need to wait. + return + + current_position = self._streams[stream_name].current_token(self._instance_name) + if position <= current_position: + # We're already past the position + return + + # Create a new deferred that times out after N seconds, as we don't want + # to wedge here forever. + deferred = Deferred() + deferred = timeout_deferred( + deferred, _WAIT_FOR_REPLICATION_TIMEOUT_SECONDS, self._reactor + ) + + waiting_list = self._streams_to_waiters.setdefault(stream_name, []) + + # We insert into the list using heapq as it is more efficient than + # pushing then resorting each time. + heapq.heappush(waiting_list, (position, deferred)) + + # We measure here to get in flight counts and average waiting time. + with Measure(self._clock, "repl.wait_for_stream_position"): + logger.info("Waiting for repl stream %r to reach %s", stream_name, position) + await make_deferred_yieldable(deferred) + logger.info( + "Finished waiting for repl stream %r to reach %s", stream_name, position + ) diff --git a/synapse/rest/admin/rooms.py b/synapse/rest/admin/rooms.py index 7d40001988..0a13e1ed34 100644 --- a/synapse/rest/admin/rooms.py +++ b/synapse/rest/admin/rooms.py @@ -59,6 +59,7 @@ class ShutdownRoomRestServlet(RestServlet): self.event_creation_handler = hs.get_event_creation_handler() self.room_member_handler = hs.get_room_member_handler() self.auth = hs.get_auth() + self._replication = hs.get_replication_data_handler() async def on_POST(self, request, room_id): requester = await self.auth.get_user_by_req(request) @@ -73,7 +74,7 @@ class ShutdownRoomRestServlet(RestServlet): message = content.get("message", self.DEFAULT_MESSAGE) room_name = content.get("room_name", "Content Violation Notification") - info = await self._room_creation_handler.create_room( + info, stream_id = await self._room_creation_handler.create_room( room_creator_requester, config={ "preset": "public_chat", @@ -94,6 +95,13 @@ class ShutdownRoomRestServlet(RestServlet): # desirable in case the first attempt at blocking the room failed below. await self.store.block_room(room_id, requester_user_id) + # We now wait for the create room to come back in via replication so + # that we can assume that all the joins/invites have propogated before + # we try and auto join below. + # + # TODO: Currently the events stream is written to from master + await self._replication.wait_for_stream_position("master", "events", stream_id) + users = await self.state.get_current_users_in_room(room_id) kicked_users = [] failed_to_kick_users = [] diff --git a/synapse/rest/client/v1/room.py b/synapse/rest/client/v1/room.py index 6b5830cc3f..105e0cf4d2 100644 --- a/synapse/rest/client/v1/room.py +++ b/synapse/rest/client/v1/room.py @@ -93,7 +93,7 @@ class RoomCreateRestServlet(TransactionRestServlet): async def on_POST(self, request): requester = await self.auth.get_user_by_req(request) - info = await self._room_creation_handler.create_room( + info, _ = await self._room_creation_handler.create_room( requester, self.get_room_config(request) ) @@ -202,7 +202,7 @@ class RoomStateEventRestServlet(TransactionRestServlet): if event_type == EventTypes.Member: membership = content.get("membership", None) - event = await self.room_member_handler.update_membership( + event_id, _ = await self.room_member_handler.update_membership( requester, target=UserID.from_string(state_key), room_id=room_id, @@ -210,14 +210,18 @@ class RoomStateEventRestServlet(TransactionRestServlet): content=content, ) else: - event = await self.event_creation_handler.create_and_send_nonmember_event( + ( + event, + _, + ) = await self.event_creation_handler.create_and_send_nonmember_event( requester, event_dict, txn_id=txn_id ) + event_id = event.event_id ret = {} # type: dict - if event: - set_tag("event_id", event.event_id) - ret = {"event_id": event.event_id} + if event_id: + set_tag("event_id", event_id) + ret = {"event_id": event_id} return 200, ret @@ -247,7 +251,7 @@ class RoomSendEventRestServlet(TransactionRestServlet): if b"ts" in request.args and requester.app_service: event_dict["origin_server_ts"] = parse_integer(request, "ts", 0) - event = await self.event_creation_handler.create_and_send_nonmember_event( + event, _ = await self.event_creation_handler.create_and_send_nonmember_event( requester, event_dict, txn_id=txn_id ) @@ -781,7 +785,7 @@ class RoomRedactEventRestServlet(TransactionRestServlet): requester = await self.auth.get_user_by_req(request) content = parse_json_object_from_request(request) - event = await self.event_creation_handler.create_and_send_nonmember_event( + event, _ = await self.event_creation_handler.create_and_send_nonmember_event( requester, { "type": EventTypes.Redaction, diff --git a/synapse/rest/client/v2_alpha/relations.py b/synapse/rest/client/v2_alpha/relations.py index 63f07b63da..89002ffbff 100644 --- a/synapse/rest/client/v2_alpha/relations.py +++ b/synapse/rest/client/v2_alpha/relations.py @@ -111,7 +111,7 @@ class RelationSendServlet(RestServlet): "sender": requester.user.to_string(), } - event = await self.event_creation_handler.create_and_send_nonmember_event( + event, _ = await self.event_creation_handler.create_and_send_nonmember_event( requester, event_dict=event_dict, txn_id=txn_id ) diff --git a/synapse/server.py b/synapse/server.py index c530f1aa1a..ca2deb49bb 100644 --- a/synapse/server.py +++ b/synapse/server.py @@ -90,6 +90,7 @@ from synapse.push.pusherpool import PusherPool from synapse.replication.tcp.client import ReplicationDataHandler from synapse.replication.tcp.handler import ReplicationCommandHandler from synapse.replication.tcp.resource import ReplicationStreamer +from synapse.replication.tcp.streams import STREAMS_MAP from synapse.rest.media.v1.media_repository import ( MediaRepository, MediaRepositoryResource, @@ -210,6 +211,7 @@ class HomeServer(object): "storage", "replication_streamer", "replication_data_handler", + "replication_streams", ] REQUIRED_ON_MASTER_STARTUP = ["user_directory_handler", "stats_handler"] @@ -583,6 +585,9 @@ class HomeServer(object): def build_replication_data_handler(self): return ReplicationDataHandler(self) + def build_replication_streams(self): + return {stream.NAME: stream(self) for stream in STREAMS_MAP.values()} + def remove_pusher(self, app_id, push_key, user_id): return self.get_pusherpool().remove_pusher(app_id, push_key, user_id) diff --git a/synapse/server.pyi b/synapse/server.pyi index 9e7fad7e6e..fe8024d2d4 100644 --- a/synapse/server.pyi +++ b/synapse/server.pyi @@ -1,3 +1,5 @@ +from typing import Dict + import twisted.internet import synapse.api.auth @@ -28,6 +30,7 @@ import synapse.server_notices.server_notices_sender import synapse.state import synapse.storage from synapse.events.builder import EventBuilderFactory +from synapse.replication.tcp.streams import Stream class HomeServer(object): @property @@ -136,3 +139,5 @@ class HomeServer(object): pass def get_pusherpool(self) -> synapse.push.pusherpool.PusherPool: pass + def get_replication_streams(self) -> Dict[str, Stream]: + pass diff --git a/synapse/server_notices/server_notices_manager.py b/synapse/server_notices/server_notices_manager.py index 999c621b92..bf2454c01c 100644 --- a/synapse/server_notices/server_notices_manager.py +++ b/synapse/server_notices/server_notices_manager.py @@ -83,10 +83,10 @@ class ServerNoticesManager(object): if state_key is not None: event_dict["state_key"] = state_key - res = await self._event_creation_handler.create_and_send_nonmember_event( + event, _ = await self._event_creation_handler.create_and_send_nonmember_event( requester, event_dict, ratelimit=False ) - return res + return event @cached() async def get_or_create_notice_room_for_user(self, user_id): @@ -143,7 +143,7 @@ class ServerNoticesManager(object): } requester = create_requester(self.server_notices_mxid) - info = await self._room_creation_handler.create_room( + info, _ = await self._room_creation_handler.create_room( requester, config={ "preset": RoomCreationPreset.PRIVATE_CHAT, diff --git a/synapse/storage/data_stores/main/events_worker.py b/synapse/storage/data_stores/main/events_worker.py index 9130b74eb5..b880a71782 100644 --- a/synapse/storage/data_stores/main/events_worker.py +++ b/synapse/storage/data_stores/main/events_worker.py @@ -1289,12 +1289,12 @@ class EventsWorkerStore(SQLBaseStore): async def is_event_after(self, event_id1, event_id2): """Returns True if event_id1 is after event_id2 in the stream """ - to_1, so_1 = await self._get_event_ordering(event_id1) - to_2, so_2 = await self._get_event_ordering(event_id2) + to_1, so_1 = await self.get_event_ordering(event_id1) + to_2, so_2 = await self.get_event_ordering(event_id2) return (to_1, so_1) > (to_2, so_2) @cachedInlineCallbacks(max_entries=5000) - def _get_event_ordering(self, event_id): + def get_event_ordering(self, event_id): res = yield self.db.simple_select_one( table="events", retcols=["topological_ordering", "stream_ordering"], diff --git a/synapse/storage/data_stores/main/roommember.py b/synapse/storage/data_stores/main/roommember.py index 1e9c850152..7c5ca81ae0 100644 --- a/synapse/storage/data_stores/main/roommember.py +++ b/synapse/storage/data_stores/main/roommember.py @@ -1069,6 +1069,8 @@ class RoomMemberStore(RoomMemberWorkerStore, RoomMemberBackgroundUpdateStore): with self._stream_id_gen.get_next() as stream_ordering: yield self.db.runInteraction("locally_reject_invite", f, stream_ordering) + return stream_ordering + def forget(self, user_id, room_id): """Indicate that user_id wishes to discard history for room_id.""" diff --git a/tests/federation/test_complexity.py b/tests/federation/test_complexity.py index 94980733c4..0c9987be54 100644 --- a/tests/federation/test_complexity.py +++ b/tests/federation/test_complexity.py @@ -79,7 +79,9 @@ class RoomComplexityTests(unittest.FederatingHomeserverTestCase): # Mock out some things, because we don't want to test the whole join fed_transport.client.get_json = Mock(return_value=defer.succeed({"v1": 9999})) - handler.federation_handler.do_invite_join = Mock(return_value=defer.succeed(1)) + handler.federation_handler.do_invite_join = Mock( + return_value=defer.succeed(("", 1)) + ) d = handler._remote_join( None, @@ -115,7 +117,9 @@ class RoomComplexityTests(unittest.FederatingHomeserverTestCase): # Mock out some things, because we don't want to test the whole join fed_transport.client.get_json = Mock(return_value=defer.succeed(None)) - handler.federation_handler.do_invite_join = Mock(return_value=defer.succeed(1)) + handler.federation_handler.do_invite_join = Mock( + return_value=defer.succeed(("", 1)) + ) # Artificially raise the complexity self.hs.get_datastore().get_current_state_event_counts = lambda x: defer.succeed( diff --git a/tests/handlers/test_typing.py b/tests/handlers/test_typing.py index 51e2b37218..2fa8d4739b 100644 --- a/tests/handlers/test_typing.py +++ b/tests/handlers/test_typing.py @@ -86,7 +86,10 @@ class TypingNotificationsTestCase(unittest.HomeserverTestCase): reactor.pump((1000,)) hs = self.setup_test_homeserver( - notifier=Mock(), http_client=mock_federation_client, keyring=mock_keyring + notifier=Mock(), + http_client=mock_federation_client, + keyring=mock_keyring, + replication_streams={}, ) hs.datastores = datastores diff --git a/tests/storage/test_cleanup_extrems.py b/tests/storage/test_cleanup_extrems.py index 0e04b2cf92..43425c969a 100644 --- a/tests/storage/test_cleanup_extrems.py +++ b/tests/storage/test_cleanup_extrems.py @@ -39,7 +39,7 @@ class CleanupExtremBackgroundUpdateStoreTestCase(HomeserverTestCase): # Create a test user and room self.user = UserID("alice", "test") self.requester = Requester(self.user, None, False, None, None) - info = self.get_success(self.room_creator.create_room(self.requester, {})) + info, _ = self.get_success(self.room_creator.create_room(self.requester, {})) self.room_id = info["room_id"] def run_background_update(self): @@ -261,7 +261,7 @@ class CleanupExtremDummyEventsTestCase(HomeserverTestCase): self.user = UserID.from_string(self.register_user("user1", "password")) self.token1 = self.login("user1", "password") self.requester = Requester(self.user, None, False, None, None) - info = self.get_success(self.room_creator.create_room(self.requester, {})) + info, _ = self.get_success(self.room_creator.create_room(self.requester, {})) self.room_id = info["room_id"] self.event_creator = homeserver.get_event_creation_handler() homeserver.config.user_consent_version = self.CONSENT_VERSION diff --git a/tests/storage/test_event_metrics.py b/tests/storage/test_event_metrics.py index a7b7fd36d3..a7b85004e5 100644 --- a/tests/storage/test_event_metrics.py +++ b/tests/storage/test_event_metrics.py @@ -33,7 +33,7 @@ class ExtremStatisticsTestCase(HomeserverTestCase): events = [(3, 2), (6, 2), (4, 6)] for event_count, extrems in events: - info = self.get_success(room_creator.create_room(requester, {})) + info, _ = self.get_success(room_creator.create_room(requester, {})) room_id = info["room_id"] last_event = None diff --git a/tests/test_federation.py b/tests/test_federation.py index 13ff14863e..c5099dd039 100644 --- a/tests/test_federation.py +++ b/tests/test_federation.py @@ -28,13 +28,13 @@ class MessageAcceptTests(unittest.HomeserverTestCase): user_id = UserID("us", "test") our_user = Requester(user_id, None, False, None, None) room_creator = self.homeserver.get_room_creation_handler() - room = ensureDeferred( + room_deferred = ensureDeferred( room_creator.create_room( our_user, room_creator.PRESETS_DICT["public_chat"], ratelimit=False ) ) self.reactor.advance(0.1) - self.room_id = self.successResultOf(room)["room_id"] + self.room_id = self.successResultOf(room_deferred)[0]["room_id"] self.store = self.homeserver.get_datastore() -- cgit 1.4.1 From 4429764c9f00bd1d266f08e08a20daac1b849d46 Mon Sep 17 00:00:00 2001 From: Patrick Cloke Date: Fri, 22 May 2020 09:30:07 -0400 Subject: Return 200 OK for all OPTIONS requests (#7534) --- changelog.d/7534.misc | 1 + synapse/app/generic_worker.py | 5 ++-- synapse/app/homeserver.py | 14 ++++++++---- synapse/http/server.py | 23 ++++++++++++++++--- tests/test_server.py | 53 +++++++++++++++++++++++++++++++++++++++++++ 5 files changed, 85 insertions(+), 11 deletions(-) create mode 100644 changelog.d/7534.misc (limited to 'tests') diff --git a/changelog.d/7534.misc b/changelog.d/7534.misc new file mode 100644 index 0000000000..9088fb65b8 --- /dev/null +++ b/changelog.d/7534.misc @@ -0,0 +1 @@ +All endpoints now respond with a 200 OK for `OPTIONS` requests. \ No newline at end of file diff --git a/synapse/app/generic_worker.py b/synapse/app/generic_worker.py index a45c876213..a37520000a 100644 --- a/synapse/app/generic_worker.py +++ b/synapse/app/generic_worker.py @@ -22,7 +22,6 @@ from typing import Dict, Iterable from typing_extensions import ContextManager from twisted.internet import defer, reactor -from twisted.web.resource import NoResource import synapse import synapse.events @@ -41,7 +40,7 @@ from synapse.config.logger import setup_logging from synapse.federation import send_queue from synapse.federation.transport.server import TransportLayerServer from synapse.handlers.presence import BasePresenceHandler, get_interested_parties -from synapse.http.server import JsonResource +from synapse.http.server import JsonResource, OptionsResource from synapse.http.servlet import RestServlet, parse_json_object_from_request from synapse.http.site import SynapseSite from synapse.logging.context import LoggingContext @@ -566,7 +565,7 @@ class GenericWorkerServer(HomeServer): if name == "replication": resources[REPLICATION_PREFIX] = ReplicationRestResource(self) - root_resource = create_resource_tree(resources, NoResource()) + root_resource = create_resource_tree(resources, OptionsResource()) _base.listen_tcp( bind_addresses, diff --git a/synapse/app/homeserver.py b/synapse/app/homeserver.py index d7f337e586..93a5ba2100 100644 --- a/synapse/app/homeserver.py +++ b/synapse/app/homeserver.py @@ -31,7 +31,7 @@ from prometheus_client import Gauge from twisted.application import service from twisted.internet import defer, reactor from twisted.python.failure import Failure -from twisted.web.resource import EncodingResourceWrapper, IResource, NoResource +from twisted.web.resource import EncodingResourceWrapper, IResource from twisted.web.server import GzipEncoderFactory from twisted.web.static import File @@ -52,7 +52,11 @@ from synapse.config._base import ConfigError from synapse.config.homeserver import HomeServerConfig from synapse.federation.transport.server import TransportLayerServer from synapse.http.additional_resource import AdditionalResource -from synapse.http.server import RootRedirect +from synapse.http.server import ( + OptionsResource, + RootOptionsRedirectResource, + RootRedirect, +) from synapse.http.site import SynapseSite from synapse.logging.context import LoggingContext from synapse.metrics import METRICS_PREFIX, MetricsResource, RegistryProxy @@ -121,11 +125,11 @@ class SynapseHomeServer(HomeServer): # try to find something useful to redirect '/' to if WEB_CLIENT_PREFIX in resources: - root_resource = RootRedirect(WEB_CLIENT_PREFIX) + root_resource = RootOptionsRedirectResource(WEB_CLIENT_PREFIX) elif STATIC_PREFIX in resources: - root_resource = RootRedirect(STATIC_PREFIX) + root_resource = RootOptionsRedirectResource(STATIC_PREFIX) else: - root_resource = NoResource() + root_resource = OptionsResource() root_resource = create_resource_tree(resources, root_resource) diff --git a/synapse/http/server.py b/synapse/http/server.py index 042a605198..33fcfbea6e 100644 --- a/synapse/http/server.py +++ b/synapse/http/server.py @@ -350,9 +350,6 @@ class JsonResource(HttpServer, resource.Resource): register_paths, so will return (possibly via Deferred) either None, or a tuple of (http code, response body). """ - if request.method == b"OPTIONS": - return _options_handler, "options_request_handler", {} - request_path = request.path.decode("ascii") # Loop through all the registered callbacks to check if the method @@ -448,6 +445,26 @@ class RootRedirect(resource.Resource): return resource.Resource.getChild(self, name, request) +class OptionsResource(resource.Resource): + """Responds to OPTION requests for itself and all children.""" + + def render_OPTIONS(self, request): + code, response_json_object = _options_handler(request) + + return respond_with_json( + request, code, response_json_object, send_cors=False, canonical_json=False, + ) + + def getChildWithDefault(self, path, request): + if request.method == b"OPTIONS": + return self # select ourselves as the child to render + return resource.Resource.getChildWithDefault(self, path, request) + + +class RootOptionsRedirectResource(OptionsResource, RootRedirect): + pass + + def respond_with_json( request, code, diff --git a/tests/test_server.py b/tests/test_server.py index 0d57eed268..437f925bf9 100644 --- a/tests/test_server.py +++ b/tests/test_server.py @@ -27,6 +27,7 @@ from synapse.api.errors import Codes, RedirectException, SynapseError from synapse.http.server import ( DirectServeResource, JsonResource, + OptionsResource, wrap_html_request_handler, ) from synapse.http.site import SynapseSite, logger @@ -168,6 +169,58 @@ class JsonResourceTests(unittest.TestCase): self.assertEqual(channel.json_body["errcode"], "M_UNRECOGNIZED") +class OptionsResourceTests(unittest.TestCase): + def setUp(self): + self.reactor = ThreadedMemoryReactorClock() + + class DummyResource(Resource): + isLeaf = True + + def render(self, request): + return request.path + + # Setup a resource with some children. + self.resource = OptionsResource() + self.resource.putChild(b"res", DummyResource()) + + def _make_request(self, method, path): + """Create a request from the method/path and return a channel with the response.""" + request, channel = make_request(self.reactor, method, path, shorthand=False) + request.prepath = [] # This doesn't get set properly by make_request. + + # Create a site and query for the resource. + site = SynapseSite("test", "site_tag", {}, self.resource, "1.0") + request.site = site + resource = site.getResourceFor(request) + + # Finally, render the resource and return the channel. + render(request, resource, self.reactor) + return channel + + def test_unknown_options_request(self): + """An OPTIONS requests to an unknown URL still returns 200 OK.""" + channel = self._make_request(b"OPTIONS", b"/foo/") + self.assertEqual(channel.result["code"], b"200") + self.assertEqual(channel.result["body"], b"{}") + + def test_known_options_request(self): + """An OPTIONS requests to an known URL still returns 200 OK.""" + channel = self._make_request(b"OPTIONS", b"/res/") + self.assertEqual(channel.result["code"], b"200") + self.assertEqual(channel.result["body"], b"{}") + + def test_unknown_request(self): + """A non-OPTIONS request to an unknown URL should 404.""" + channel = self._make_request(b"GET", b"/foo/") + self.assertEqual(channel.result["code"], b"404") + + def test_known_request(self): + """A non-OPTIONS request to an known URL should query the proper resource.""" + channel = self._make_request(b"GET", b"/res/") + self.assertEqual(channel.result["code"], b"200") + self.assertEqual(channel.result["body"], b"/res/") + + class WrapHtmlRequestHandlerTests(unittest.TestCase): class TestResource(DirectServeResource): callback = None -- cgit 1.4.1 From 2901f54359bba1ccbe2aac52fd9ff255aa6072b7 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Fri, 22 May 2020 17:42:39 +0100 Subject: Fix missing CORS headers on OPTION responses (#7560) Broke in #7534. --- changelog.d/7560.misc | 1 + synapse/http/server.py | 2 +- tests/test_server.py | 28 ++++++++++++++++++++++++++++ 3 files changed, 30 insertions(+), 1 deletion(-) create mode 100644 changelog.d/7560.misc (limited to 'tests') diff --git a/changelog.d/7560.misc b/changelog.d/7560.misc new file mode 100644 index 0000000000..9088fb65b8 --- /dev/null +++ b/changelog.d/7560.misc @@ -0,0 +1 @@ +All endpoints now respond with a 200 OK for `OPTIONS` requests. \ No newline at end of file diff --git a/synapse/http/server.py b/synapse/http/server.py index 33fcfbea6e..9cc2e2e154 100644 --- a/synapse/http/server.py +++ b/synapse/http/server.py @@ -452,7 +452,7 @@ class OptionsResource(resource.Resource): code, response_json_object = _options_handler(request) return respond_with_json( - request, code, response_json_object, send_cors=False, canonical_json=False, + request, code, response_json_object, send_cors=True, canonical_json=False, ) def getChildWithDefault(self, path, request): diff --git a/tests/test_server.py b/tests/test_server.py index 437f925bf9..e9a43b1e45 100644 --- a/tests/test_server.py +++ b/tests/test_server.py @@ -203,12 +203,40 @@ class OptionsResourceTests(unittest.TestCase): self.assertEqual(channel.result["code"], b"200") self.assertEqual(channel.result["body"], b"{}") + # Ensure the correct CORS headers have been added + self.assertTrue( + channel.headers.hasHeader(b"Access-Control-Allow-Origin"), + "has CORS Origin header", + ) + self.assertTrue( + channel.headers.hasHeader(b"Access-Control-Allow-Methods"), + "has CORS Methods header", + ) + self.assertTrue( + channel.headers.hasHeader(b"Access-Control-Allow-Headers"), + "has CORS Headers header", + ) + def test_known_options_request(self): """An OPTIONS requests to an known URL still returns 200 OK.""" channel = self._make_request(b"OPTIONS", b"/res/") self.assertEqual(channel.result["code"], b"200") self.assertEqual(channel.result["body"], b"{}") + # Ensure the correct CORS headers have been added + self.assertTrue( + channel.headers.hasHeader(b"Access-Control-Allow-Origin"), + "has CORS Origin header", + ) + self.assertTrue( + channel.headers.hasHeader(b"Access-Control-Allow-Methods"), + "has CORS Methods header", + ) + self.assertTrue( + channel.headers.hasHeader(b"Access-Control-Allow-Headers"), + "has CORS Headers header", + ) + def test_unknown_request(self): """A non-OPTIONS request to an unknown URL should 404.""" channel = self._make_request(b"GET", b"/foo/") -- cgit 1.4.1 From f4269694ce51a04761e43fc5897d6f8fe0ea18cc Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Fri, 22 May 2020 21:47:07 +0100 Subject: Optimise some references to hs.config (#7546) These are surprisingly expensive, and we only really need to do them at startup. --- changelog.d/7546.misc | 1 + synapse/handlers/message.py | 8 +- .../resource_limits_server_notices.py | 15 ++- .../data_stores/main/monthly_active_users.py | 28 ++++-- .../test_resource_limits_server_notices.py | 60 ++++++----- tests/storage/test_client_ips.py | 13 +-- tests/storage/test_monthly_active_users.py | 110 +++++++++++---------- tests/test_mau.py | 63 ++++++------ 8 files changed, 162 insertions(+), 136 deletions(-) create mode 100644 changelog.d/7546.misc (limited to 'tests') diff --git a/changelog.d/7546.misc b/changelog.d/7546.misc new file mode 100644 index 0000000000..0a9704789b --- /dev/null +++ b/changelog.d/7546.misc @@ -0,0 +1 @@ +Optimise some references to `hs.config`. diff --git a/synapse/handlers/message.py b/synapse/handlers/message.py index ea25f0515a..681f92cafd 100644 --- a/synapse/handlers/message.py +++ b/synapse/handlers/message.py @@ -366,7 +366,9 @@ class EventCreationHandler(object): self.notifier = hs.get_notifier() self.config = hs.config self.require_membership_for_aliases = hs.config.require_membership_for_aliases - self._instance_name = hs.get_instance_name() + self._is_event_writer = ( + self.config.worker.writers.events == hs.get_instance_name() + ) self.room_invite_state_types = self.hs.config.room_invite_state_types @@ -836,7 +838,7 @@ class EventCreationHandler(object): success = False try: # If we're a worker we need to hit out to the master. - if self.config.worker.writers.events != self._instance_name: + if not self._is_event_writer: result = await self.send_event( instance_name=self.config.worker.writers.events, event_id=event.event_id, @@ -906,7 +908,7 @@ class EventCreationHandler(object): This should only be run on the instance in charge of persisting events. """ - assert self.config.worker.writers.events == self._instance_name + assert self._is_event_writer if ratelimit: # We check if this is a room admin redacting an event so that we diff --git a/synapse/server_notices/resource_limits_server_notices.py b/synapse/server_notices/resource_limits_server_notices.py index d97166351e..73f2cedb5c 100644 --- a/synapse/server_notices/resource_limits_server_notices.py +++ b/synapse/server_notices/resource_limits_server_notices.py @@ -48,6 +48,12 @@ class ResourceLimitsServerNotices(object): self._notifier = hs.get_notifier() + self._enabled = ( + hs.config.limit_usage_by_mau + and self._server_notices_manager.is_enabled() + and not hs.config.hs_disabled + ) + async def maybe_send_server_notice_to_user(self, user_id): """Check if we need to send a notice to this user, this will be true in two cases. @@ -61,14 +67,7 @@ class ResourceLimitsServerNotices(object): Returns: Deferred """ - if self._config.hs_disabled is True: - return - - if self._config.limit_usage_by_mau is False: - return - - if not self._server_notices_manager.is_enabled(): - # Don't try and send server notices unless they've been enabled + if not self._enabled: return timestamp = await self._store.user_last_seen_monthly_active(user_id) diff --git a/synapse/storage/data_stores/main/monthly_active_users.py b/synapse/storage/data_stores/main/monthly_active_users.py index 925bc5691b..a42c52f323 100644 --- a/synapse/storage/data_stores/main/monthly_active_users.py +++ b/synapse/storage/data_stores/main/monthly_active_users.py @@ -122,6 +122,10 @@ class MonthlyActiveUsersStore(MonthlyActiveUsersWorkerStore): def __init__(self, database: Database, db_conn, hs): super(MonthlyActiveUsersStore, self).__init__(database, db_conn, hs) + self._limit_usage_by_mau = hs.config.limit_usage_by_mau + self._mau_stats_only = hs.config.mau_stats_only + self._max_mau_value = hs.config.max_mau_value + # Do not add more reserved users than the total allowable number # cur = LoggingTransaction( self.db.new_transaction( @@ -130,7 +134,7 @@ class MonthlyActiveUsersStore(MonthlyActiveUsersWorkerStore): [], [], self._initialise_reserved_users, - hs.config.mau_limits_reserved_threepids[: self.hs.config.max_mau_value], + hs.config.mau_limits_reserved_threepids[: self._max_mau_value], ) def _initialise_reserved_users(self, txn, threepids): @@ -142,6 +146,15 @@ class MonthlyActiveUsersStore(MonthlyActiveUsersWorkerStore): threepids (list[dict]): List of threepid dicts to reserve """ + # XXX what is this function trying to achieve? It upserts into + # monthly_active_users for each *registered* reserved mau user, but why? + # + # - shouldn't there already be an entry for each reserved user (at least + # if they have been active recently)? + # + # - if it's important that the timestamp is kept up to date, why do we only + # run this at startup? + for tp in threepids: user_id = self.get_user_id_by_threepid_txn(txn, tp["medium"], tp["address"]) @@ -191,8 +204,7 @@ class MonthlyActiveUsersStore(MonthlyActiveUsersWorkerStore): txn.execute(sql, query_args) - max_mau_value = self.hs.config.max_mau_value - if self.hs.config.limit_usage_by_mau: + if self._limit_usage_by_mau: # If MAU user count still exceeds the MAU threshold, then delete on # a least recently active basis. # Note it is not possible to write this query using OFFSET due to @@ -210,13 +222,13 @@ class MonthlyActiveUsersStore(MonthlyActiveUsersWorkerStore): LIMIT ? ) """ - txn.execute(sql, (max_mau_value,)) + txn.execute(sql, ((self._max_mau_value),)) # Need if/else since 'AND user_id NOT IN ({})' fails on Postgres # when len(reserved_users) == 0. Works fine on sqlite. else: # Must be >= 0 for postgres num_of_non_reserved_users_to_remove = max( - max_mau_value - len(reserved_users), 0 + self._max_mau_value - len(reserved_users), 0 ) # It is important to filter reserved users twice to guard @@ -335,7 +347,7 @@ class MonthlyActiveUsersStore(MonthlyActiveUsersWorkerStore): Args: user_id(str): the user_id to query """ - if self.hs.config.limit_usage_by_mau or self.hs.config.mau_stats_only: + if self._limit_usage_by_mau or self._mau_stats_only: # Trial users and guests should not be included as part of MAU group is_guest = yield self.is_guest(user_id) if is_guest: @@ -356,11 +368,11 @@ class MonthlyActiveUsersStore(MonthlyActiveUsersWorkerStore): # In the case where mau_stats_only is True and limit_usage_by_mau is # False, there is no point in checking get_monthly_active_count - it # adds no value and will break the logic if max_mau_value is exceeded. - if not self.hs.config.limit_usage_by_mau: + if not self._limit_usage_by_mau: yield self.upsert_monthly_active_user(user_id) else: count = yield self.get_monthly_active_count() - if count < self.hs.config.max_mau_value: + if count < self._max_mau_value: yield self.upsert_monthly_active_user(user_id) elif now - last_seen_timestamp > LAST_SEEN_GRANULARITY: yield self.upsert_monthly_active_user(user_id) diff --git a/tests/server_notices/test_resource_limits_server_notices.py b/tests/server_notices/test_resource_limits_server_notices.py index 406f29a7c0..99908edba3 100644 --- a/tests/server_notices/test_resource_limits_server_notices.py +++ b/tests/server_notices/test_resource_limits_server_notices.py @@ -27,20 +27,33 @@ from synapse.server_notices.resource_limits_server_notices import ( ) from tests import unittest +from tests.unittest import override_config +from tests.utils import default_config class TestResourceLimitsServerNotices(unittest.HomeserverTestCase): - def make_homeserver(self, reactor, clock): - hs_config = self.default_config() - hs_config["server_notices"] = { - "system_mxid_localpart": "server", - "system_mxid_display_name": "test display name", - "system_mxid_avatar_url": None, - "room_name": "Server Notices", - } + def default_config(self): + config = default_config("test") + + config.update( + { + "admin_contact": "mailto:user@test.com", + "limit_usage_by_mau": True, + "server_notices": { + "system_mxid_localpart": "server", + "system_mxid_display_name": "test display name", + "system_mxid_avatar_url": None, + "room_name": "Server Notices", + }, + } + ) + + # apply any additional config which was specified via the override_config + # decorator. + if self._extra_config is not None: + config.update(self._extra_config) - hs = self.setup_test_homeserver(config=hs_config) - return hs + return config def prepare(self, reactor, clock, hs): self.server_notices_sender = self.hs.get_server_notices_sender() @@ -60,7 +73,6 @@ class TestResourceLimitsServerNotices(unittest.HomeserverTestCase): ) self._send_notice = self._rlsn._server_notices_manager.send_notice - self.hs.config.limit_usage_by_mau = True self.user_id = "@user_id:test" self._rlsn._server_notices_manager.get_or_create_notice_room_for_user = Mock( @@ -68,21 +80,17 @@ class TestResourceLimitsServerNotices(unittest.HomeserverTestCase): ) self._rlsn._store.add_tag_to_room = Mock(return_value=defer.succeed(None)) self._rlsn._store.get_tags_for_room = Mock(return_value=defer.succeed({})) - self.hs.config.admin_contact = "mailto:user@test.com" - - def test_maybe_send_server_notice_to_user_flag_off(self): - """Tests cases where the flags indicate nothing to do""" - # test hs disabled case - self.hs.config.hs_disabled = True + @override_config({"hs_disabled": True}) + def test_maybe_send_server_notice_disabled_hs(self): + """If the HS is disabled, we should not send notices""" self.get_success(self._rlsn.maybe_send_server_notice_to_user(self.user_id)) - self._send_notice.assert_not_called() - # Test when mau limiting disabled - self.hs.config.hs_disabled = False - self.hs.config.limit_usage_by_mau = False - self.get_success(self._rlsn.maybe_send_server_notice_to_user(self.user_id)) + @override_config({"limit_usage_by_mau": False}) + def test_maybe_send_server_notice_to_user_flag_off(self): + """If mau limiting is disabled, we should not send notices""" + self.get_success(self._rlsn.maybe_send_server_notice_to_user(self.user_id)) self._send_notice.assert_not_called() def test_maybe_send_server_notice_to_user_remove_blocked_notice(self): @@ -153,13 +161,12 @@ class TestResourceLimitsServerNotices(unittest.HomeserverTestCase): self._send_notice.assert_not_called() + @override_config({"mau_limit_alerting": False}) def test_maybe_send_server_notice_when_alerting_suppressed_room_unblocked(self): """ Test that when server is over MAU limit and alerting is suppressed, then an alert message is not sent into the room """ - self.hs.config.mau_limit_alerting = False - self._rlsn._auth.check_auth_blocking = Mock( return_value=defer.succeed(None), side_effect=ResourceLimitError( @@ -170,12 +177,11 @@ class TestResourceLimitsServerNotices(unittest.HomeserverTestCase): self.assertEqual(self._send_notice.call_count, 0) + @override_config({"mau_limit_alerting": False}) def test_check_hs_disabled_unaffected_by_mau_alert_suppression(self): """ Test that when a server is disabled, that MAU limit alerting is ignored. """ - self.hs.config.mau_limit_alerting = False - self._rlsn._auth.check_auth_blocking = Mock( return_value=defer.succeed(None), side_effect=ResourceLimitError( @@ -187,12 +193,12 @@ class TestResourceLimitsServerNotices(unittest.HomeserverTestCase): # Would be better to check contents, but 2 calls == set blocking event self.assertEqual(self._send_notice.call_count, 2) + @override_config({"mau_limit_alerting": False}) def test_maybe_send_server_notice_when_alerting_suppressed_room_blocked(self): """ When the room is already in a blocked state, test that when alerting is suppressed that the room is returned to an unblocked state. """ - self.hs.config.mau_limit_alerting = False self._rlsn._auth.check_auth_blocking = Mock( return_value=defer.succeed(None), side_effect=ResourceLimitError( diff --git a/tests/storage/test_client_ips.py b/tests/storage/test_client_ips.py index bf674dd184..3b483bc7f0 100644 --- a/tests/storage/test_client_ips.py +++ b/tests/storage/test_client_ips.py @@ -23,6 +23,7 @@ from synapse.http.site import XForwardedForRequest from synapse.rest.client.v1 import login from tests import unittest +from tests.unittest import override_config class ClientIpStoreTestCase(unittest.HomeserverTestCase): @@ -137,9 +138,8 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): ], ) + @override_config({"limit_usage_by_mau": False, "max_mau_value": 50}) def test_disabled_monthly_active_user(self): - self.hs.config.limit_usage_by_mau = False - self.hs.config.max_mau_value = 50 user_id = "@user:server" self.get_success( self.store.insert_client_ip( @@ -149,9 +149,8 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): active = self.get_success(self.store.user_last_seen_monthly_active(user_id)) self.assertFalse(active) + @override_config({"limit_usage_by_mau": True, "max_mau_value": 50}) def test_adding_monthly_active_user_when_full(self): - self.hs.config.limit_usage_by_mau = True - self.hs.config.max_mau_value = 50 lots_of_users = 100 user_id = "@user:server" @@ -166,9 +165,8 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): active = self.get_success(self.store.user_last_seen_monthly_active(user_id)) self.assertFalse(active) + @override_config({"limit_usage_by_mau": True, "max_mau_value": 50}) def test_adding_monthly_active_user_when_space(self): - self.hs.config.limit_usage_by_mau = True - self.hs.config.max_mau_value = 50 user_id = "@user:server" active = self.get_success(self.store.user_last_seen_monthly_active(user_id)) self.assertFalse(active) @@ -184,9 +182,8 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase): active = self.get_success(self.store.user_last_seen_monthly_active(user_id)) self.assertTrue(active) + @override_config({"limit_usage_by_mau": True, "max_mau_value": 50}) def test_updating_monthly_active_user_when_space(self): - self.hs.config.limit_usage_by_mau = True - self.hs.config.max_mau_value = 50 user_id = "@user:server" self.get_success(self.store.register_user(user_id=user_id, password_hash=None)) diff --git a/tests/storage/test_monthly_active_users.py b/tests/storage/test_monthly_active_users.py index bc53bf0951..447fcb3a1c 100644 --- a/tests/storage/test_monthly_active_users.py +++ b/tests/storage/test_monthly_active_users.py @@ -19,94 +19,106 @@ from twisted.internet import defer from synapse.api.constants import UserTypes from tests import unittest +from tests.unittest import default_config, override_config FORTY_DAYS = 40 * 24 * 60 * 60 +def gen_3pids(count): + """Generate `count` threepids as a list.""" + return [ + {"medium": "email", "address": "user%i@matrix.org" % i} for i in range(count) + ] + + class MonthlyActiveUsersTestCase(unittest.HomeserverTestCase): - def make_homeserver(self, reactor, clock): + def default_config(self): + config = default_config("test") + + config.update({"limit_usage_by_mau": True, "max_mau_value": 50}) - hs = self.setup_test_homeserver() - self.store = hs.get_datastore() - hs.config.limit_usage_by_mau = True - hs.config.max_mau_value = 50 + # apply any additional config which was specified via the override_config + # decorator. + if self._extra_config is not None: + config.update(self._extra_config) + return config + + def prepare(self, reactor, clock, homeserver): + self.store = homeserver.get_datastore() # Advance the clock a bit reactor.advance(FORTY_DAYS) - return hs - + @override_config({"max_mau_value": 3, "mau_limit_reserved_threepids": gen_3pids(3)}) def test_initialise_reserved_users(self): - self.hs.config.max_mau_value = 5 + threepids = self.hs.config.mau_limits_reserved_threepids + + # register three users, of which two have reserved 3pids, and a third + # which is a support user. user1 = "@user1:server" - user1_email = "user1@matrix.org" + user1_email = threepids[0]["address"] user2 = "@user2:server" - user2_email = "user2@matrix.org" + user2_email = threepids[1]["address"] user3 = "@user3:server" - user3_email = "user3@matrix.org" - threepids = [ - {"medium": "email", "address": user1_email}, - {"medium": "email", "address": user2_email}, - {"medium": "email", "address": user3_email}, - ] - self.hs.config.mau_limits_reserved_threepids = threepids - # -1 because user3 is a support user and does not count - user_num = len(threepids) - 1 - - self.store.register_user(user_id=user1, password_hash=None) - self.store.register_user(user_id=user2, password_hash=None) - self.store.register_user( - user_id=user3, password_hash=None, user_type=UserTypes.SUPPORT - ) + self.store.register_user(user_id=user1) + self.store.register_user(user_id=user2) + self.store.register_user(user_id=user3, user_type=UserTypes.SUPPORT) self.pump() now = int(self.hs.get_clock().time_msec()) self.store.user_add_threepid(user1, "email", user1_email, now, now) self.store.user_add_threepid(user2, "email", user2_email, now, now) + # XXX why are we doing this here? this function is only run at startup + # so it is odd to re-run it here. self.store.db.runInteraction( "initialise", self.store._initialise_reserved_users, threepids ) self.pump() - active_count = self.store.get_monthly_active_count() + # the number of users we expect will be counted against the mau limit + # -1 because user3 is a support user and does not count + user_num = len(threepids) - 1 - # Test total counts, ensure user3 (support user) is not counted - self.assertEquals(self.get_success(active_count), user_num) + # Check the number of active users. Ensure user3 (support user) is not counted + active_count = self.get_success(self.store.get_monthly_active_count()) + self.assertEquals(active_count, user_num) - # Test user is marked as active + # Test each of the registered users is marked as active timestamp = self.store.user_last_seen_monthly_active(user1) self.assertTrue(self.get_success(timestamp)) timestamp = self.store.user_last_seen_monthly_active(user2) self.assertTrue(self.get_success(timestamp)) - # Test that users are never removed from the db. + # Test that users with reserved 3pids are not removed from the MAU table + # XXX some of this is redundant. poking things into the config shouldn't + # work, and in any case it's not obvious what we expect to happen when + # we advance the reactor. self.hs.config.max_mau_value = 0 - self.reactor.advance(FORTY_DAYS) self.hs.config.max_mau_value = 5 - self.store.reap_monthly_active_users() self.pump() active_count = self.store.get_monthly_active_count() self.assertEquals(self.get_success(active_count), user_num) - # Test that regular users are removed from the db + # Add some more users and check they are counted as active ru_count = 2 self.store.upsert_monthly_active_user("@ru1:server") self.store.upsert_monthly_active_user("@ru2:server") self.pump() - active_count = self.store.get_monthly_active_count() self.assertEqual(self.get_success(active_count), user_num + ru_count) - self.hs.config.max_mau_value = user_num + + # now run the reaper and check that the number of active users is reduced + # to max_mau_value self.store.reap_monthly_active_users() self.pump() active_count = self.store.get_monthly_active_count() - self.assertEquals(self.get_success(active_count), user_num) + self.assertEquals(self.get_success(active_count), 3) def test_can_insert_and_count_mau(self): count = self.store.get_monthly_active_count() @@ -136,8 +148,8 @@ class MonthlyActiveUsersTestCase(unittest.HomeserverTestCase): result = self.store.user_last_seen_monthly_active(user_id3) self.assertNotEqual(self.get_success(result), 0) + @override_config({"max_mau_value": 5}) def test_reap_monthly_active_users(self): - self.hs.config.max_mau_value = 5 initial_users = 10 for i in range(initial_users): self.store.upsert_monthly_active_user("@user%d:server" % i) @@ -158,19 +170,19 @@ class MonthlyActiveUsersTestCase(unittest.HomeserverTestCase): count = self.store.get_monthly_active_count() self.assertEquals(self.get_success(count), 0) + # Note that below says mau_limit (no s), this is the name of the config + # value, although it gets stored on the config object as mau_limits. + @override_config({"max_mau_value": 5, "mau_limit_reserved_threepids": gen_3pids(5)}) def test_reap_monthly_active_users_reserved_users(self): """ Tests that reaping correctly handles reaping where reserved users are present""" - - self.hs.config.max_mau_value = 5 - initial_users = 5 + threepids = self.hs.config.mau_limits_reserved_threepids + initial_users = len(threepids) reserved_user_number = initial_users - 1 - threepids = [] for i in range(initial_users): user = "@user%d:server" % i - email = "user%d@example.com" % i + email = "user%d@matrix.org" % i self.get_success(self.store.upsert_monthly_active_user(user)) - threepids.append({"medium": "email", "address": email}) # Need to ensure that the most recent entries in the # monthly_active_users table are reserved now = int(self.hs.get_clock().time_msec()) @@ -182,7 +194,6 @@ class MonthlyActiveUsersTestCase(unittest.HomeserverTestCase): self.store.user_add_threepid(user, "email", email, now, now) ) - self.hs.config.mau_limits_reserved_threepids = threepids self.store.db.runInteraction( "initialise", self.store._initialise_reserved_users, threepids ) @@ -279,11 +290,11 @@ class MonthlyActiveUsersTestCase(unittest.HomeserverTestCase): self.pump() self.assertEqual(self.get_success(count), 0) + # Note that the max_mau_value setting should not matter. + @override_config( + {"limit_usage_by_mau": False, "mau_stats_only": True, "max_mau_value": 1} + ) def test_track_monthly_users_without_cap(self): - self.hs.config.limit_usage_by_mau = False - self.hs.config.mau_stats_only = True - self.hs.config.max_mau_value = 1 # should not matter - count = self.store.get_monthly_active_count() self.assertEqual(0, self.get_success(count)) @@ -294,9 +305,8 @@ class MonthlyActiveUsersTestCase(unittest.HomeserverTestCase): count = self.store.get_monthly_active_count() self.assertEqual(2, self.get_success(count)) + @override_config({"limit_usage_by_mau": False, "mau_stats_only": False}) def test_no_users_when_not_tracking(self): - self.hs.config.limit_usage_by_mau = False - self.hs.config.mau_stats_only = False self.store.upsert_monthly_active_user = Mock() self.store.populate_monthly_active_users("@user:sever") diff --git a/tests/test_mau.py b/tests/test_mau.py index eb159e3ba5..8a97f0998d 100644 --- a/tests/test_mau.py +++ b/tests/test_mau.py @@ -17,47 +17,44 @@ import json -from mock import Mock - -from synapse.api.auth_blocking import AuthBlocking from synapse.api.constants import LoginType from synapse.api.errors import Codes, HttpResponseException, SynapseError from synapse.rest.client.v2_alpha import register, sync from tests import unittest +from tests.unittest import override_config +from tests.utils import default_config class TestMauLimit(unittest.HomeserverTestCase): servlets = [register.register_servlets, sync.register_servlets] - def make_homeserver(self, reactor, clock): + def default_config(self): + config = default_config("test") - self.hs = self.setup_test_homeserver( - "red", http_client=None, federation_client=Mock() + config.update( + { + "registrations_require_3pid": [], + "limit_usage_by_mau": True, + "max_mau_value": 2, + "mau_trial_days": 0, + "server_notices": { + "system_mxid_localpart": "server", + "room_name": "Test Server Notice Room", + }, + } ) - self.store = self.hs.get_datastore() - - self.hs.config.registrations_require_3pid = [] - self.hs.config.enable_registration_captcha = False - self.hs.config.recaptcha_public_key = [] - - self.hs.config.limit_usage_by_mau = True - self.hs.config.hs_disabled = False - self.hs.config.max_mau_value = 2 - self.hs.config.server_notices_mxid = "@server:red" - self.hs.config.server_notices_mxid_display_name = None - self.hs.config.server_notices_mxid_avatar_url = None - self.hs.config.server_notices_room_name = "Test Server Notice Room" - self.hs.config.mau_trial_days = 0 + # apply any additional config which was specified via the override_config + # decorator. + if self._extra_config is not None: + config.update(self._extra_config) - # AuthBlocking reads config options during hs creation. Recreate the - # hs' copy of AuthBlocking after we've updated config values above - self.auth_blocking = AuthBlocking(self.hs) - self.hs.get_auth()._auth_blocking = self.auth_blocking + return config - return self.hs + def prepare(self, reactor, clock, homeserver): + self.store = homeserver.get_datastore() def test_simple_deny_mau(self): # Create and sync so that the MAU counts get updated @@ -66,6 +63,9 @@ class TestMauLimit(unittest.HomeserverTestCase): token2 = self.create_user("kermit2") self.do_sync_for_user(token2) + # check we're testing what we think we are: there should be two active users + self.assertEqual(self.get_success(self.store.get_monthly_active_count()), 2) + # We've created and activated two users, we shouldn't be able to # register new users with self.assertRaises(SynapseError) as cm: @@ -93,9 +93,8 @@ class TestMauLimit(unittest.HomeserverTestCase): token3 = self.create_user("kermit3") self.do_sync_for_user(token3) + @override_config({"mau_trial_days": 1}) def test_trial_delay(self): - self.hs.config.mau_trial_days = 1 - # We should be able to register more than the limit initially token1 = self.create_user("kermit1") self.do_sync_for_user(token1) @@ -127,8 +126,8 @@ class TestMauLimit(unittest.HomeserverTestCase): self.assertEqual(e.code, 403) self.assertEqual(e.errcode, Codes.RESOURCE_LIMIT_EXCEEDED) + @override_config({"mau_trial_days": 1}) def test_trial_users_cant_come_back(self): - self.auth_blocking._mau_trial_days = 1 self.hs.config.mau_trial_days = 1 # We should be able to register more than the limit initially @@ -176,11 +175,11 @@ class TestMauLimit(unittest.HomeserverTestCase): self.assertEqual(e.code, 403) self.assertEqual(e.errcode, Codes.RESOURCE_LIMIT_EXCEEDED) + @override_config( + # max_mau_value should not matter + {"max_mau_value": 1, "limit_usage_by_mau": False, "mau_stats_only": True} + ) def test_tracked_but_not_limited(self): - self.auth_blocking._max_mau_value = 1 # should not matter - self.auth_blocking._limit_usage_by_mau = False - self.hs.config.mau_stats_only = True - # Simply being able to create 2 users indicates that the # limit was not reached. token1 = self.create_user("kermit1") -- cgit 1.4.1 From 00db90f4098e400306b3f25f14ba7a302d6f7ee8 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Tue, 26 May 2020 11:41:38 +0100 Subject: Fix recording of federation stream token (#7564) A couple of changes of significance: * remove the `_last_ack < federation_position` condition, so that updates will still be correctly processed after restart * Correctly wire up send_federation_ack to the right class. --- changelog.d/7565.bugfix | 1 + synapse/app/generic_worker.py | 68 +++++++++++++++++++----------- tests/replication/test_federation_ack.py | 71 ++++++++++++++++++++++++++++++++ 3 files changed, 116 insertions(+), 24 deletions(-) create mode 100644 changelog.d/7565.bugfix create mode 100644 tests/replication/test_federation_ack.py (limited to 'tests') diff --git a/changelog.d/7565.bugfix b/changelog.d/7565.bugfix new file mode 100644 index 0000000000..35fabb9a0b --- /dev/null +++ b/changelog.d/7565.bugfix @@ -0,0 +1 @@ +Fix exception `'GenericWorkerReplicationHandler' object has no attribute 'send_federation_ack'`, introduced in v1.13.0. diff --git a/synapse/app/generic_worker.py b/synapse/app/generic_worker.py index 2906b93f6a..440341cb3c 100644 --- a/synapse/app/generic_worker.py +++ b/synapse/app/generic_worker.py @@ -17,7 +17,7 @@ import contextlib import logging import sys -from typing import Dict, Iterable +from typing import Dict, Iterable, Optional, Set from typing_extensions import ContextManager @@ -677,10 +677,9 @@ class GenericWorkerReplicationHandler(ReplicationDataHandler): self.notify_pushers = hs.config.start_pushers self.pusher_pool = hs.get_pusherpool() + self.send_handler = None # type: Optional[FederationSenderHandler] if hs.config.send_federation: - self.send_handler = FederationSenderHandler(hs, self) - else: - self.send_handler = None + self.send_handler = FederationSenderHandler(hs) async def on_rdata(self, stream_name, instance_name, token, rows): await super().on_rdata(stream_name, instance_name, token, rows) @@ -718,7 +717,7 @@ class GenericWorkerReplicationHandler(ReplicationDataHandler): if entities: self.notifier.on_new_event("to_device_key", token, users=entities) elif stream_name == DeviceListsStream.NAME: - all_room_ids = set() + all_room_ids = set() # type: Set[str] for row in rows: if row.entity.startswith("@"): room_ids = await self.store.get_rooms_for_user(row.entity) @@ -769,24 +768,33 @@ class GenericWorkerReplicationHandler(ReplicationDataHandler): class FederationSenderHandler(object): - """Processes the replication stream and forwards the appropriate entries - to the federation sender. + """Processes the fedration replication stream + + This class is only instantiate on the worker responsible for sending outbound + federation transactions. It receives rows from the replication stream and forwards + the appropriate entries to the FederationSender class. """ - def __init__(self, hs: GenericWorkerServer, replication_client): + def __init__(self, hs: GenericWorkerServer): self.store = hs.get_datastore() self._is_mine_id = hs.is_mine_id self.federation_sender = hs.get_federation_sender() - self.replication_client = replication_client - + self._hs = hs + + # if the worker is restarted, we want to pick up where we left off in + # the replication stream, so load the position from the database. + # + # XXX is this actually worthwhile? Whenever the master is restarted, we'll + # drop some rows anyway (which is mostly fine because we're only dropping + # typing and presence notifications). If the replication stream is + # unreliable, why do we do all this hoop-jumping to store the position in the + # database? See also https://github.com/matrix-org/synapse/issues/7535. + # self.federation_position = self.store.federation_out_pos_startup - self._fed_position_linearizer = Linearizer(name="_fed_position_linearizer") + self._fed_position_linearizer = Linearizer(name="_fed_position_linearizer") self._last_ack = self.federation_position - self._room_serials = {} - self._room_typing = {} - def on_start(self): # There may be some events that are persisted but haven't been sent, # so send them now. @@ -849,22 +857,34 @@ class FederationSenderHandler(object): await self.federation_sender.send_read_receipt(receipt_info) async def update_token(self, token): + """Update the record of where we have processed to in the federation stream. + + Called after we have processed a an update received over replication. Sends + a FEDERATION_ACK back to the master, and stores the token that we have processed + in `federation_stream_position` so that we can restart where we left off. + """ try: self.federation_position = token # We linearize here to ensure we don't have races updating the token + # + # XXX this appears to be redundant, since the ReplicationCommandHandler + # has a linearizer which ensures that we only process one line of + # replication data at a time. Should we remove it, or is it doing useful + # service for robustness? Or could we replace it with an assertion that + # we're not being re-entered? + with (await self._fed_position_linearizer.queue(None)): - if self._last_ack < self.federation_position: - await self.store.update_federation_out_pos( - "federation", self.federation_position - ) + await self.store.update_federation_out_pos( + "federation", self.federation_position + ) - # We ACK this token over replication so that the master can drop - # its in memory queues - self.replication_client.send_federation_ack( - self.federation_position - ) - self._last_ack = self.federation_position + # We ACK this token over replication so that the master can drop + # its in memory queues + self._hs.get_tcp_replication().send_federation_ack( + self.federation_position + ) + self._last_ack = self.federation_position except Exception: logger.exception("Error updating federation stream position") diff --git a/tests/replication/test_federation_ack.py b/tests/replication/test_federation_ack.py new file mode 100644 index 0000000000..5448d9f0dc --- /dev/null +++ b/tests/replication/test_federation_ack.py @@ -0,0 +1,71 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import mock + +from synapse.app.generic_worker import GenericWorkerServer +from synapse.replication.tcp.commands import FederationAckCommand +from synapse.replication.tcp.protocol import AbstractConnection +from synapse.replication.tcp.streams.federation import FederationStream + +from tests.unittest import HomeserverTestCase + + +class FederationAckTestCase(HomeserverTestCase): + def default_config(self) -> dict: + config = super().default_config() + config["worker_app"] = "synapse.app.federation_sender" + config["send_federation"] = True + return config + + def make_homeserver(self, reactor, clock): + hs = self.setup_test_homeserver(homeserverToUse=GenericWorkerServer) + return hs + + def test_federation_ack_sent(self): + """A FEDERATION_ACK should be sent back after each RDATA federation + + This test checks that the federation sender is correctly sending back + FEDERATION_ACK messages. The test works by spinning up a federation_sender + worker server, and then fishing out its ReplicationCommandHandler. We wire + the RCH up to a mock connection (so that we can observe the command being sent) + and then poke in an RDATA row. + + XXX: it might be nice to do this by pretending to be a synapse master worker + (or a redis server), and having the worker connect to us via a mocked-up TCP + transport, rather than assuming that the implementation has a + ReplicationCommandHandler. + """ + rch = self.hs.get_tcp_replication() + + # wire up the ReplicationCommandHandler to a mock connection + mock_connection = mock.Mock(spec=AbstractConnection) + rch.new_connection(mock_connection) + + # tell it it received an RDATA row + self.get_success( + rch.on_rdata( + "federation", + "master", + token=10, + rows=[FederationStream.FederationStreamRow(type="x", data=[1, 2, 3])], + ) + ) + + # now check that the FEDERATION_ACK was sent + mock_connection.send_command.assert_called_once() + cmd = mock_connection.send_command.call_args[0][0] + assert isinstance(cmd, FederationAckCommand) + self.assertEqual(cmd.token, 10) -- cgit 1.4.1 From eefc6b3a0d08cd2a64be7c78c0a4a651cc965be9 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 27 May 2020 12:04:37 +0100 Subject: Don't apply cache factor to event cache. (#7578) This is already correctly done when we instansiate the cache, but wasn't when it got reloaded (which always happens at least once on startup). --- changelog.d/7578.bugfix | 1 + synapse/util/caches/lrucache.py | 4 ++++ tests/config/test_cache.py | 16 ++++++++++++++++ 3 files changed, 21 insertions(+) create mode 100644 changelog.d/7578.bugfix (limited to 'tests') diff --git a/changelog.d/7578.bugfix b/changelog.d/7578.bugfix new file mode 100644 index 0000000000..cd29307361 --- /dev/null +++ b/changelog.d/7578.bugfix @@ -0,0 +1 @@ +Fix cache config to not apply cache factor to event cache. Regression in v1.14.0rc1. diff --git a/synapse/util/caches/lrucache.py b/synapse/util/caches/lrucache.py index 29fabac3cd..df4ea5901d 100644 --- a/synapse/util/caches/lrucache.py +++ b/synapse/util/caches/lrucache.py @@ -81,6 +81,7 @@ class LruCache(object): """ cache = cache_type() self.cache = cache # Used for introspection. + self.apply_cache_factor_from_config = apply_cache_factor_from_config # Save the original max size, and apply the default size factor. self._original_max_size = max_size @@ -294,6 +295,9 @@ class LruCache(object): Returns: bool: Whether the cache changed size or not. """ + if not self.apply_cache_factor_from_config: + return False + new_size = int(self._original_max_size * factor) if new_size != self.max_size: self.max_size = new_size diff --git a/tests/config/test_cache.py b/tests/config/test_cache.py index 2920279125..b45e0cc536 100644 --- a/tests/config/test_cache.py +++ b/tests/config/test_cache.py @@ -125,3 +125,19 @@ class CacheConfigTests(TestCase): cache = LruCache(100) add_resizable_cache("foo", cache_resize_callback=cache.set_cache_factor) self.assertEqual(cache.max_size, 150) + + def test_apply_cache_factor_from_config(self): + """Caches can disable applying cache factor updates, mainly used by + event cache size. + """ + + config = {"caches": {"event_cache_size": "10k"}} + t = TestConfig() + t.read_config(config, config_dir_path="", data_dir_path="") + + cache = LruCache( + max_size=t.caches.event_cache_size, apply_cache_factor_from_config=False, + ) + add_resizable_cache("event_cache", cache_resize_callback=cache.set_cache_factor) + + self.assertEqual(cache.max_size, 10240) -- cgit 1.4.1 From 4ba55559acf56a041e47ec0d74890d4ad3e0ddb7 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 27 May 2020 13:17:01 +0100 Subject: Fix specifying cache factors via env vars with * in name. (#7580) This mostly applise to `*stateGroupCache*` and co. Broke in #6391. --- changelog.d/7580.bugfix | 1 + docs/sample_config.yaml | 6 ++++++ synapse/config/cache.py | 44 +++++++++++++++++++++++++++++++++++++++----- tests/config/test_cache.py | 28 ++++++++++++++++++++++++++++ 4 files changed, 74 insertions(+), 5 deletions(-) create mode 100644 changelog.d/7580.bugfix (limited to 'tests') diff --git a/changelog.d/7580.bugfix b/changelog.d/7580.bugfix new file mode 100644 index 0000000000..b255dc2a12 --- /dev/null +++ b/changelog.d/7580.bugfix @@ -0,0 +1 @@ +Fix specifying individual cache factors for caches with special characters in their name. Regression in v1.14.0rc1. diff --git a/docs/sample_config.yaml b/docs/sample_config.yaml index 0e1be153c7..48f273b0b2 100644 --- a/docs/sample_config.yaml +++ b/docs/sample_config.yaml @@ -643,6 +643,12 @@ caches: # takes priority over setting through the config file. # Ex. SYNAPSE_CACHE_FACTOR_GET_USERS_WHO_SHARE_ROOM_WITH_USER=2.0 # + # Some caches have '*' and other characters that are not + # alphanumeric or underscores. These caches can be named with or + # without the special characters stripped. For example, to specify + # the cache factor for `*stateGroupCache*` via an environment + # variable would be `SYNAPSE_CACHE_FACTOR_STATEGROUPCACHE=2`. + # per_cache_factors: #get_users_who_share_room_with_user: 2.0 diff --git a/synapse/config/cache.py b/synapse/config/cache.py index 91036a012e..acc31652de 100644 --- a/synapse/config/cache.py +++ b/synapse/config/cache.py @@ -14,13 +14,17 @@ # limitations under the License. import os +import re from typing import Callable, Dict from ._base import Config, ConfigError # The prefix for all cache factor-related environment variables -_CACHES = {} _CACHE_PREFIX = "SYNAPSE_CACHE_FACTOR" + +# Map from canonicalised cache name to cache. +_CACHES = {} + _DEFAULT_FACTOR_SIZE = 0.5 _DEFAULT_EVENT_CACHE_SIZE = "10K" @@ -37,6 +41,20 @@ class CacheProperties(object): properties = CacheProperties() +def _canonicalise_cache_name(cache_name: str) -> str: + """Gets the canonical form of the cache name. + + Since we specify cache names in config and environment variables we need to + ignore case and special characters. For example, some caches have asterisks + in their name to donate that they're not attached to a particular database + function, and these asterisks need to be stripped out + """ + + cache_name = re.sub(r"[^A-Za-z_1-9]", "", cache_name) + + return cache_name.lower() + + def add_resizable_cache(cache_name: str, cache_resize_callback: Callable): """Register a cache that's size can dynamically change @@ -45,7 +63,10 @@ def add_resizable_cache(cache_name: str, cache_resize_callback: Callable): cache_resize_callback: A callback function that will be ran whenever the cache needs to be resized """ - _CACHES[cache_name.lower()] = cache_resize_callback + # Some caches have '*' in them which we strip out. + cache_name = _canonicalise_cache_name(cache_name) + + _CACHES[cache_name] = cache_resize_callback # Ensure all loaded caches are sized appropriately # @@ -105,6 +126,12 @@ class CacheConfig(Config): # takes priority over setting through the config file. # Ex. SYNAPSE_CACHE_FACTOR_GET_USERS_WHO_SHARE_ROOM_WITH_USER=2.0 # + # Some caches have '*' and other characters that are not + # alphanumeric or underscores. These caches can be named with or + # without the special characters stripped. For example, to specify + # the cache factor for `*stateGroupCache*` via an environment + # variable would be `SYNAPSE_CACHE_FACTOR_STATEGROUPCACHE=2`. + # per_cache_factors: #get_users_who_share_room_with_user: 2.0 """ @@ -130,10 +157,17 @@ class CacheConfig(Config): if not isinstance(individual_factors, dict): raise ConfigError("caches.per_cache_factors must be a dictionary") + # Canonicalise the cache names *before* updating with the environment + # variables. + individual_factors = { + _canonicalise_cache_name(key): val + for key, val in individual_factors.items() + } + # Override factors from environment if necessary individual_factors.update( { - key[len(_CACHE_PREFIX) + 1 :].lower(): float(val) + _canonicalise_cache_name(key[len(_CACHE_PREFIX) + 1 :]): float(val) for key, val in self._environ.items() if key.startswith(_CACHE_PREFIX + "_") } @@ -142,9 +176,9 @@ class CacheConfig(Config): for cache, factor in individual_factors.items(): if not isinstance(factor, (int, float)): raise ConfigError( - "caches.per_cache_factors.%s must be a number" % (cache.lower(),) + "caches.per_cache_factors.%s must be a number" % (cache,) ) - self.cache_factors[cache.lower()] = factor + self.cache_factors[cache] = factor # Resize all caches (if necessary) with the new factors we've loaded self.resize_all_caches() diff --git a/tests/config/test_cache.py b/tests/config/test_cache.py index b45e0cc536..d3ec24c975 100644 --- a/tests/config/test_cache.py +++ b/tests/config/test_cache.py @@ -126,6 +126,34 @@ class CacheConfigTests(TestCase): add_resizable_cache("foo", cache_resize_callback=cache.set_cache_factor) self.assertEqual(cache.max_size, 150) + def test_cache_with_asterisk_in_name(self): + """Some caches have asterisks in their name, test that they are set correctly. + """ + + config = { + "caches": { + "per_cache_factors": {"*cache_a*": 5, "cache_b": 6, "cache_c": 2} + } + } + t = TestConfig() + t.caches._environ = { + "SYNAPSE_CACHE_FACTOR_CACHE_A": "2", + "SYNAPSE_CACHE_FACTOR_CACHE_B": 3, + } + t.read_config(config, config_dir_path="", data_dir_path="") + + cache_a = LruCache(100) + add_resizable_cache("*cache_a*", cache_resize_callback=cache_a.set_cache_factor) + self.assertEqual(cache_a.max_size, 200) + + cache_b = LruCache(100) + add_resizable_cache("*Cache_b*", cache_resize_callback=cache_b.set_cache_factor) + self.assertEqual(cache_b.max_size, 300) + + cache_c = LruCache(100) + add_resizable_cache("*cache_c*", cache_resize_callback=cache_c.set_cache_factor) + self.assertEqual(cache_c.max_size, 200) + def test_apply_cache_factor_from_config(self): """Caches can disable applying cache factor updates, mainly used by event cache size. -- cgit 1.4.1 From a72d5f39db55dfecb48291acdd6566c6556e0b0b Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 27 May 2020 19:41:06 +0100 Subject: Add test for Linearizer.is_queued(..) --- tests/util/test_linearizer.py | 32 ++++++++++++++++++++++++++++++++ 1 file changed, 32 insertions(+) (limited to 'tests') diff --git a/tests/util/test_linearizer.py b/tests/util/test_linearizer.py index 852ef23185..ca3858b184 100644 --- a/tests/util/test_linearizer.py +++ b/tests/util/test_linearizer.py @@ -45,6 +45,38 @@ class LinearizerTestCase(unittest.TestCase): with (yield d2): pass + @defer.inlineCallbacks + def test_linearizer_is_queued(self): + linearizer = Linearizer() + + key = object() + + d1 = linearizer.queue(key) + cm1 = yield d1 + + # Since d1 gets called immediately, "is_queued" should return false. + self.assertFalse(linearizer.is_queued(key)) + + d2 = linearizer.queue(key) + self.assertFalse(d2.called) + + # Now d2 is queued up behind successful completion of cm1 + self.assertTrue(linearizer.is_queued(key)) + + with cm1: + self.assertFalse(d2.called) + + # cm1 still not done, so d2 still queued. + self.assertTrue(linearizer.is_queued(key)) + + # And now d2 is called and nothing is in the queue again + self.assertFalse(linearizer.is_queued(key)) + + with (yield d2): + self.assertFalse(linearizer.is_queued(key)) + + self.assertFalse(linearizer.is_queued(key)) + def test_lots_of_queued_things(self): # we have one slow thing, and lots of fast things queued up behind it. # it should *not* explode the stack. -- cgit 1.4.1 From 901b1fa561e3cc661d78aa96d59802cf2078cb0d Mon Sep 17 00:00:00 2001 From: Dirk Klimpel <5740567+dklimpel@users.noreply.github.com> Date: Mon, 1 Jun 2020 16:34:33 +0200 Subject: Email notifications for new users when creating via the Admin API. (#7267) --- changelog.d/7267.bugfix | 1 + synapse/rest/admin/users.py | 16 +++++++++ tests/rest/admin/test_user.py | 75 +++++++++++++++++++++++++++++++++++++++++++ 3 files changed, 92 insertions(+) create mode 100644 changelog.d/7267.bugfix (limited to 'tests') diff --git a/changelog.d/7267.bugfix b/changelog.d/7267.bugfix new file mode 100644 index 0000000000..0af316c1a2 --- /dev/null +++ b/changelog.d/7267.bugfix @@ -0,0 +1 @@ +Fix email notifications not being enabled for new users when created via the Admin API. diff --git a/synapse/rest/admin/users.py b/synapse/rest/admin/users.py index e7f6928c85..82251dbe5f 100644 --- a/synapse/rest/admin/users.py +++ b/synapse/rest/admin/users.py @@ -142,6 +142,7 @@ class UserRestServletV2(RestServlet): self.set_password_handler = hs.get_set_password_handler() self.deactivate_account_handler = hs.get_deactivate_account_handler() self.registration_handler = hs.get_registration_handler() + self.pusher_pool = hs.get_pusherpool() async def on_GET(self, request, user_id): await assert_requester_is_admin(self.auth, request) @@ -281,6 +282,21 @@ class UserRestServletV2(RestServlet): await self.auth_handler.add_threepid( user_id, threepid["medium"], threepid["address"], current_time ) + if ( + self.hs.config.email_enable_notifs + and self.hs.config.email_notif_for_new_users + ): + await self.pusher_pool.add_pusher( + user_id=user_id, + access_token=None, + kind="email", + app_id="m.email", + app_display_name="Email Notifications", + device_display_name=threepid["address"], + pushkey=threepid["address"], + lang=None, # We don't know a user's language here + data={}, + ) if "avatar_url" in body and type(body["avatar_url"]) == str: await self.profile_handler.set_avatar_url( diff --git a/tests/rest/admin/test_user.py b/tests/rest/admin/test_user.py index 6c88ab06e2..e29cc24a8a 100644 --- a/tests/rest/admin/test_user.py +++ b/tests/rest/admin/test_user.py @@ -516,6 +516,81 @@ class UserRestTestCase(unittest.HomeserverTestCase): self.assertEqual(False, channel.json_body["is_guest"]) self.assertEqual(False, channel.json_body["deactivated"]) + def test_create_user_email_notif_for_new_users(self): + """ + Check that a new regular user is created successfully and + got an email pusher. + """ + self.hs.config.registration_shared_secret = None + self.hs.config.email_enable_notifs = True + self.hs.config.email_notif_for_new_users = True + url = "/_synapse/admin/v2/users/@bob:test" + + # Create user + body = json.dumps( + { + "password": "abc123", + "threepids": [{"medium": "email", "address": "bob@bob.bob"}], + } + ) + + request, channel = self.make_request( + "PUT", + url, + access_token=self.admin_user_tok, + content=body.encode(encoding="utf_8"), + ) + self.render(request) + + self.assertEqual(201, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@bob:test", channel.json_body["name"]) + self.assertEqual("email", channel.json_body["threepids"][0]["medium"]) + self.assertEqual("bob@bob.bob", channel.json_body["threepids"][0]["address"]) + + pushers = self.get_success( + self.store.get_pushers_by({"user_name": "@bob:test"}) + ) + pushers = list(pushers) + self.assertEqual(len(pushers), 1) + self.assertEqual("@bob:test", pushers[0]["user_name"]) + + def test_create_user_email_no_notif_for_new_users(self): + """ + Check that a new regular user is created successfully and + got not an email pusher. + """ + self.hs.config.registration_shared_secret = None + self.hs.config.email_enable_notifs = False + self.hs.config.email_notif_for_new_users = False + url = "/_synapse/admin/v2/users/@bob:test" + + # Create user + body = json.dumps( + { + "password": "abc123", + "threepids": [{"medium": "email", "address": "bob@bob.bob"}], + } + ) + + request, channel = self.make_request( + "PUT", + url, + access_token=self.admin_user_tok, + content=body.encode(encoding="utf_8"), + ) + self.render(request) + + self.assertEqual(201, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@bob:test", channel.json_body["name"]) + self.assertEqual("email", channel.json_body["threepids"][0]["medium"]) + self.assertEqual("bob@bob.bob", channel.json_body["threepids"][0]["address"]) + + pushers = self.get_success( + self.store.get_pushers_by({"user_name": "@bob:test"}) + ) + pushers = list(pushers) + self.assertEqual(len(pushers), 0) + def test_set_password(self): """ Test setting a new password for another user. -- cgit 1.4.1 From 33c39ab93c5cb9b82faa2717c62a5ffaf6780a86 Mon Sep 17 00:00:00 2001 From: Brendan Abolivier Date: Mon, 1 Jun 2020 17:47:30 +0200 Subject: Process cross-signing keys when resyncing device lists (#7594) It looks like `user_device_resync` was ignoring cross-signing keys from the results received from the remote server. This patch fixes this, by processing these keys using the same process `_handle_signing_key_updates` does (and effectively factor that part out of that function). --- changelog.d/7594.bugfix | 1 + synapse/handlers/device.py | 58 +++++++++++++++++++++++++++++++++++++++++++- synapse/handlers/e2e_keys.py | 22 ++++------------- tests/test_federation.py | 56 ++++++++++++++++++++++++++++++++++++++++++ 4 files changed, 119 insertions(+), 18 deletions(-) create mode 100644 changelog.d/7594.bugfix (limited to 'tests') diff --git a/changelog.d/7594.bugfix b/changelog.d/7594.bugfix new file mode 100644 index 0000000000..f0c067e184 --- /dev/null +++ b/changelog.d/7594.bugfix @@ -0,0 +1 @@ +Fix a bug causing the cross-signing keys to be ignored when resyncing a device list. diff --git a/synapse/handlers/device.py b/synapse/handlers/device.py index 2cbb695bb1..230d170258 100644 --- a/synapse/handlers/device.py +++ b/synapse/handlers/device.py @@ -15,6 +15,7 @@ # See the License for the specific language governing permissions and # limitations under the License. import logging +from typing import Any, Dict, Optional from six import iteritems, itervalues @@ -30,7 +31,11 @@ from synapse.api.errors import ( ) from synapse.logging.opentracing import log_kv, set_tag, trace from synapse.metrics.background_process_metrics import run_as_background_process -from synapse.types import RoomStreamToken, get_domain_from_id +from synapse.types import ( + RoomStreamToken, + get_domain_from_id, + get_verify_key_from_cross_signing_key, +) from synapse.util import stringutils from synapse.util.async_helpers import Linearizer from synapse.util.caches.expiringcache import ExpiringCache @@ -795,6 +800,13 @@ class DeviceListUpdater(object): stream_id = result["stream_id"] devices = result["devices"] + # Get the master key and the self-signing key for this user if provided in the + # response (None if not in the response). + # The response will not contain the user signing key, as this key is only used by + # its owner, thus it doesn't make sense to send it over federation. + master_key = result.get("master_key") + self_signing_key = result.get("self_signing_key") + # If the remote server has more than ~1000 devices for this user # we assume that something is going horribly wrong (e.g. a bot # that logs in and creates a new device every time it tries to @@ -824,6 +836,13 @@ class DeviceListUpdater(object): yield self.store.update_remote_device_list_cache(user_id, devices, stream_id) device_ids = [device["device_id"] for device in devices] + + # Handle cross-signing keys. + cross_signing_device_ids = yield self.process_cross_signing_key_update( + user_id, master_key, self_signing_key, + ) + device_ids = device_ids + cross_signing_device_ids + yield self.device_handler.notify_device_update(user_id, device_ids) # We clobber the seen updates since we've re-synced from a given @@ -831,3 +850,40 @@ class DeviceListUpdater(object): self._seen_updates[user_id] = {stream_id} defer.returnValue(result) + + @defer.inlineCallbacks + def process_cross_signing_key_update( + self, + user_id: str, + master_key: Optional[Dict[str, Any]], + self_signing_key: Optional[Dict[str, Any]], + ) -> list: + """Process the given new master and self-signing key for the given remote user. + + Args: + user_id: The ID of the user these keys are for. + master_key: The dict of the cross-signing master key as returned by the + remote server. + self_signing_key: The dict of the cross-signing self-signing key as returned + by the remote server. + + Return: + The device IDs for the given keys. + """ + device_ids = [] + + if master_key: + yield self.store.set_e2e_cross_signing_key(user_id, "master", master_key) + _, verify_key = get_verify_key_from_cross_signing_key(master_key) + # verify_key is a VerifyKey from signedjson, which uses + # .version to denote the portion of the key ID after the + # algorithm and colon, which is the device ID + device_ids.append(verify_key.version) + if self_signing_key: + yield self.store.set_e2e_cross_signing_key( + user_id, "self_signing", self_signing_key + ) + _, verify_key = get_verify_key_from_cross_signing_key(self_signing_key) + device_ids.append(verify_key.version) + + return device_ids diff --git a/synapse/handlers/e2e_keys.py b/synapse/handlers/e2e_keys.py index 8f1bc0323c..774a252619 100644 --- a/synapse/handlers/e2e_keys.py +++ b/synapse/handlers/e2e_keys.py @@ -1291,6 +1291,7 @@ class SigningKeyEduUpdater(object): """ device_handler = self.e2e_keys_handler.device_handler + device_list_updater = device_handler.device_list_updater with (yield self._remote_edu_linearizer.queue(user_id)): pending_updates = self._pending_updates.pop(user_id, []) @@ -1303,22 +1304,9 @@ class SigningKeyEduUpdater(object): logger.info("pending updates: %r", pending_updates) for master_key, self_signing_key in pending_updates: - if master_key: - yield self.store.set_e2e_cross_signing_key( - user_id, "master", master_key - ) - _, verify_key = get_verify_key_from_cross_signing_key(master_key) - # verify_key is a VerifyKey from signedjson, which uses - # .version to denote the portion of the key ID after the - # algorithm and colon, which is the device ID - device_ids.append(verify_key.version) - if self_signing_key: - yield self.store.set_e2e_cross_signing_key( - user_id, "self_signing", self_signing_key - ) - _, verify_key = get_verify_key_from_cross_signing_key( - self_signing_key - ) - device_ids.append(verify_key.version) + new_device_ids = yield device_list_updater.process_cross_signing_key_update( + user_id, master_key, self_signing_key, + ) + device_ids = device_ids + new_device_ids yield device_handler.notify_device_update(user_id, device_ids) diff --git a/tests/test_federation.py b/tests/test_federation.py index c5099dd039..c662195eec 100644 --- a/tests/test_federation.py +++ b/tests/test_federation.py @@ -206,3 +206,59 @@ class MessageAcceptTests(unittest.HomeserverTestCase): # list. self.reactor.advance(30) self.assertEqual(self.resync_attempts, 2) + + def test_cross_signing_keys_retry(self): + """Tests that resyncing a device list correctly processes cross-signing keys from + the remote server. + """ + remote_user_id = "@john:test_remote" + remote_master_key = "85T7JXPFBAySB/jwby4S3lBPTqY3+Zg53nYuGmu1ggY" + remote_self_signing_key = "QeIiFEjluPBtI7WQdG365QKZcFs9kqmHir6RBD0//nQ" + + # Register mock device list retrieval on the federation client. + federation_client = self.homeserver.get_federation_client() + federation_client.query_user_devices = Mock( + return_value={ + "user_id": remote_user_id, + "stream_id": 1, + "devices": [], + "master_key": { + "user_id": remote_user_id, + "usage": ["master"], + "keys": {"ed25519:" + remote_master_key: remote_master_key}, + }, + "self_signing_key": { + "user_id": remote_user_id, + "usage": ["self_signing"], + "keys": { + "ed25519:" + remote_self_signing_key: remote_self_signing_key + }, + }, + } + ) + + # Resync the device list. + device_handler = self.homeserver.get_device_handler() + self.get_success( + device_handler.device_list_updater.user_device_resync(remote_user_id), + ) + + # Retrieve the cross-signing keys for this user. + keys = self.get_success( + self.store.get_e2e_cross_signing_keys_bulk(user_ids=[remote_user_id]), + ) + self.assertTrue(remote_user_id in keys) + + # Check that the master key is the one returned by the mock. + master_key = keys[remote_user_id]["master"] + self.assertEqual(len(master_key["keys"]), 1) + self.assertTrue("ed25519:" + remote_master_key in master_key["keys"].keys()) + self.assertTrue(remote_master_key in master_key["keys"].values()) + + # Check that the self-signing key is the one returned by the mock. + self_signing_key = keys[remote_user_id]["self_signing"] + self.assertEqual(len(self_signing_key["keys"]), 1) + self.assertTrue( + "ed25519:" + remote_self_signing_key in self_signing_key["keys"].keys(), + ) + self.assertTrue(remote_self_signing_key in self_signing_key["keys"].values()) -- cgit 1.4.1 From fe434cd3c94dfc98954aea908e188e5d97df60db Mon Sep 17 00:00:00 2001 From: Olof Johansson Date: Mon, 1 Jun 2020 18:55:07 +0200 Subject: Fix a bug in automatic user creation with m.login.jwt. (#7585) --- changelog.d/7585.bugfix | 1 + synapse/rest/client/v1/login.py | 15 ++-- tests/rest/client/v1/test_login.py | 153 +++++++++++++++++++++++++++++++++++++ 3 files changed, 162 insertions(+), 7 deletions(-) create mode 100644 changelog.d/7585.bugfix (limited to 'tests') diff --git a/changelog.d/7585.bugfix b/changelog.d/7585.bugfix new file mode 100644 index 0000000000..263295599d --- /dev/null +++ b/changelog.d/7585.bugfix @@ -0,0 +1 @@ +Fix a bug in automatic user creation during first time login with `m.login.jwt`. Regression in v1.6.0. Contributed by @olof. diff --git a/synapse/rest/client/v1/login.py b/synapse/rest/client/v1/login.py index d89b2e5532..36aca82346 100644 --- a/synapse/rest/client/v1/login.py +++ b/synapse/rest/client/v1/login.py @@ -299,7 +299,7 @@ class LoginRestServlet(RestServlet): return result async def _complete_login( - self, user_id, login_submission, callback=None, create_non_existant_users=False + self, user_id, login_submission, callback=None, create_non_existent_users=False ): """Called when we've successfully authed the user and now need to actually login them in (e.g. create devices). This gets called on @@ -312,7 +312,7 @@ class LoginRestServlet(RestServlet): user_id (str): ID of the user to register. login_submission (dict): Dictionary of login information. callback (func|None): Callback function to run after registration. - create_non_existant_users (bool): Whether to create the user if + create_non_existent_users (bool): Whether to create the user if they don't exist. Defaults to False. Returns: @@ -331,12 +331,13 @@ class LoginRestServlet(RestServlet): update=True, ) - if create_non_existant_users: - user_id = await self.auth_handler.check_user_exists(user_id) - if not user_id: - user_id = await self.registration_handler.register_user( + if create_non_existent_users: + canonical_uid = await self.auth_handler.check_user_exists(user_id) + if not canonical_uid: + canonical_uid = await self.registration_handler.register_user( localpart=UserID.from_string(user_id).localpart ) + user_id = canonical_uid device_id = login_submission.get("device_id") initial_display_name = login_submission.get("initial_device_display_name") @@ -391,7 +392,7 @@ class LoginRestServlet(RestServlet): user_id = UserID(user, self.hs.hostname).to_string() result = await self._complete_login( - user_id, login_submission, create_non_existant_users=True + user_id, login_submission, create_non_existent_users=True ) return result diff --git a/tests/rest/client/v1/test_login.py b/tests/rest/client/v1/test_login.py index eb8f6264fd..0f0f7ca72d 100644 --- a/tests/rest/client/v1/test_login.py +++ b/tests/rest/client/v1/test_login.py @@ -1,8 +1,11 @@ import json +import time import urllib.parse from mock import Mock +import jwt + import synapse.rest.admin from synapse.rest.client.v1 import login, logout from synapse.rest.client.v2_alpha import devices @@ -473,3 +476,153 @@ class CASTestCase(unittest.HomeserverTestCase): # Because the user is deactivated they are served an error template. self.assertEqual(channel.code, 403) self.assertIn(b"SSO account deactivated", channel.result["body"]) + + +class JWTTestCase(unittest.HomeserverTestCase): + servlets = [ + synapse.rest.admin.register_servlets_for_client_rest_resource, + login.register_servlets, + ] + + jwt_secret = "secret" + + def make_homeserver(self, reactor, clock): + self.hs = self.setup_test_homeserver() + self.hs.config.jwt_enabled = True + self.hs.config.jwt_secret = self.jwt_secret + self.hs.config.jwt_algorithm = "HS256" + return self.hs + + def jwt_encode(self, token, secret=jwt_secret): + return jwt.encode(token, secret, "HS256").decode("ascii") + + def jwt_login(self, *args): + params = json.dumps({"type": "m.login.jwt", "token": self.jwt_encode(*args)}) + request, channel = self.make_request(b"POST", LOGIN_URL, params) + self.render(request) + return channel + + def test_login_jwt_valid_registered(self): + self.register_user("kermit", "monkey") + channel = self.jwt_login({"sub": "kermit"}) + self.assertEqual(channel.result["code"], b"200", channel.result) + self.assertEqual(channel.json_body["user_id"], "@kermit:test") + + def test_login_jwt_valid_unregistered(self): + channel = self.jwt_login({"sub": "frog"}) + self.assertEqual(channel.result["code"], b"200", channel.result) + self.assertEqual(channel.json_body["user_id"], "@frog:test") + + def test_login_jwt_invalid_signature(self): + channel = self.jwt_login({"sub": "frog"}, "notsecret") + self.assertEqual(channel.result["code"], b"401", channel.result) + self.assertEqual(channel.json_body["errcode"], "M_UNAUTHORIZED") + self.assertEqual(channel.json_body["error"], "Invalid JWT") + + def test_login_jwt_expired(self): + channel = self.jwt_login({"sub": "frog", "exp": 864000}) + self.assertEqual(channel.result["code"], b"401", channel.result) + self.assertEqual(channel.json_body["errcode"], "M_UNAUTHORIZED") + self.assertEqual(channel.json_body["error"], "JWT expired") + + def test_login_jwt_not_before(self): + now = int(time.time()) + channel = self.jwt_login({"sub": "frog", "nbf": now + 3600}) + self.assertEqual(channel.result["code"], b"401", channel.result) + self.assertEqual(channel.json_body["errcode"], "M_UNAUTHORIZED") + self.assertEqual(channel.json_body["error"], "Invalid JWT") + + def test_login_no_sub(self): + channel = self.jwt_login({"username": "root"}) + self.assertEqual(channel.result["code"], b"401", channel.result) + self.assertEqual(channel.json_body["errcode"], "M_UNAUTHORIZED") + self.assertEqual(channel.json_body["error"], "Invalid JWT") + + def test_login_no_token(self): + params = json.dumps({"type": "m.login.jwt"}) + request, channel = self.make_request(b"POST", LOGIN_URL, params) + self.render(request) + self.assertEqual(channel.result["code"], b"401", channel.result) + self.assertEqual(channel.json_body["errcode"], "M_UNAUTHORIZED") + self.assertEqual(channel.json_body["error"], "Token field for JWT is missing") + + +# The JWTPubKeyTestCase is a complement to JWTTestCase where we instead use +# RSS256, with a public key configured in synapse as "jwt_secret", and tokens +# signed by the private key. +class JWTPubKeyTestCase(unittest.HomeserverTestCase): + servlets = [ + login.register_servlets, + ] + + # This key's pubkey is used as the jwt_secret setting of synapse. Valid + # tokens are signed by this and validated using the pubkey. It is generated + # with `openssl genrsa 512` (not a secure way to generate real keys, but + # good enough for tests!) + jwt_privatekey = "\n".join( + [ + "-----BEGIN RSA PRIVATE KEY-----", + "MIIBPAIBAAJBAM50f1Q5gsdmzifLstzLHb5NhfajiOt7TKO1vSEWdq7u9x8SMFiB", + "492RM9W/XFoh8WUfL9uL6Now6tPRDsWv3xsCAwEAAQJAUv7OOSOtiU+wzJq82rnk", + "yR4NHqt7XX8BvkZPM7/+EjBRanmZNSp5kYZzKVaZ/gTOM9+9MwlmhidrUOweKfB/", + "kQIhAPZwHazbjo7dYlJs7wPQz1vd+aHSEH+3uQKIysebkmm3AiEA1nc6mDdmgiUq", + "TpIN8A4MBKmfZMWTLq6z05y/qjKyxb0CIQDYJxCwTEenIaEa4PdoJl+qmXFasVDN", + "ZU0+XtNV7yul0wIhAMI9IhiStIjS2EppBa6RSlk+t1oxh2gUWlIh+YVQfZGRAiEA", + "tqBR7qLZGJ5CVKxWmNhJZGt1QHoUtOch8t9C4IdOZ2g=", + "-----END RSA PRIVATE KEY-----", + ] + ) + + # Generated with `openssl rsa -in foo.key -pubout`, with the the above + # private key placed in foo.key (jwt_privatekey). + jwt_pubkey = "\n".join( + [ + "-----BEGIN PUBLIC KEY-----", + "MFwwDQYJKoZIhvcNAQEBBQADSwAwSAJBAM50f1Q5gsdmzifLstzLHb5NhfajiOt7", + "TKO1vSEWdq7u9x8SMFiB492RM9W/XFoh8WUfL9uL6Now6tPRDsWv3xsCAwEAAQ==", + "-----END PUBLIC KEY-----", + ] + ) + + # This key is used to sign tokens that shouldn't be accepted by synapse. + # Generated just like jwt_privatekey. + bad_privatekey = "\n".join( + [ + "-----BEGIN RSA PRIVATE KEY-----", + "MIIBOgIBAAJBAL//SQrKpKbjCCnv/FlasJCv+t3k/MPsZfniJe4DVFhsktF2lwQv", + "gLjmQD3jBUTz+/FndLSBvr3F4OHtGL9O/osCAwEAAQJAJqH0jZJW7Smzo9ShP02L", + "R6HRZcLExZuUrWI+5ZSP7TaZ1uwJzGFspDrunqaVoPobndw/8VsP8HFyKtceC7vY", + "uQIhAPdYInDDSJ8rFKGiy3Ajv5KWISBicjevWHF9dbotmNO9AiEAxrdRJVU+EI9I", + "eB4qRZpY6n4pnwyP0p8f/A3NBaQPG+cCIFlj08aW/PbxNdqYoBdeBA0xDrXKfmbb", + "iwYxBkwL0JCtAiBYmsi94sJn09u2Y4zpuCbJeDPKzWkbuwQh+W1fhIWQJQIhAKR0", + "KydN6cRLvphNQ9c/vBTdlzWxzcSxREpguC7F1J1m", + "-----END RSA PRIVATE KEY-----", + ] + ) + + def make_homeserver(self, reactor, clock): + self.hs = self.setup_test_homeserver() + self.hs.config.jwt_enabled = True + self.hs.config.jwt_secret = self.jwt_pubkey + self.hs.config.jwt_algorithm = "RS256" + return self.hs + + def jwt_encode(self, token, secret=jwt_privatekey): + return jwt.encode(token, secret, "RS256").decode("ascii") + + def jwt_login(self, *args): + params = json.dumps({"type": "m.login.jwt", "token": self.jwt_encode(*args)}) + request, channel = self.make_request(b"POST", LOGIN_URL, params) + self.render(request) + return channel + + def test_login_jwt_valid(self): + channel = self.jwt_login({"sub": "kermit"}) + self.assertEqual(channel.result["code"], b"200", channel.result) + self.assertEqual(channel.json_body["user_id"], "@kermit:test") + + def test_login_jwt_invalid_signature(self): + channel = self.jwt_login({"sub": "frog"}, self.bad_privatekey) + self.assertEqual(channel.result["code"], b"401", channel.result) + self.assertEqual(channel.json_body["errcode"], "M_UNAUTHORIZED") + self.assertEqual(channel.json_body["error"], "Invalid JWT") -- cgit 1.4.1 From 0188daf32c5d978c33b2bba9eb2b0b63262ca5fe Mon Sep 17 00:00:00 2001 From: Andrew Morgan <1342360+anoadragon453@users.noreply.github.com> Date: Wed, 3 Jun 2020 16:39:30 +0100 Subject: Replace instances of reactor pumping with get_success. (#7619) Calls `self.get_success` on all deferred methods instead of abusing `self.pump()`. This has the benefit of working with coroutines, as well as checking that method execution completed successfully. There are also a few small cleanups that I made in the process. --- changelog.d/7619.misc | 1 + tests/storage/test_monthly_active_users.py | 267 ++++++++++++++++------------- tests/test_mau.py | 5 +- 3 files changed, 152 insertions(+), 121 deletions(-) create mode 100644 changelog.d/7619.misc (limited to 'tests') diff --git a/changelog.d/7619.misc b/changelog.d/7619.misc new file mode 100644 index 0000000000..23a8b30b19 --- /dev/null +++ b/changelog.d/7619.misc @@ -0,0 +1 @@ +Check that all asynchronous tasks succeed and general cleanup of `MonthlyActiveUsersTestCase` and `TestMauLimit`. diff --git a/tests/storage/test_monthly_active_users.py b/tests/storage/test_monthly_active_users.py index 447fcb3a1c..9c04e92577 100644 --- a/tests/storage/test_monthly_active_users.py +++ b/tests/storage/test_monthly_active_users.py @@ -61,21 +61,27 @@ class MonthlyActiveUsersTestCase(unittest.HomeserverTestCase): user2_email = threepids[1]["address"] user3 = "@user3:server" - self.store.register_user(user_id=user1) - self.store.register_user(user_id=user2) - self.store.register_user(user_id=user3, user_type=UserTypes.SUPPORT) - self.pump() + self.get_success(self.store.register_user(user_id=user1)) + self.get_success(self.store.register_user(user_id=user2)) + self.get_success( + self.store.register_user(user_id=user3, user_type=UserTypes.SUPPORT) + ) now = int(self.hs.get_clock().time_msec()) - self.store.user_add_threepid(user1, "email", user1_email, now, now) - self.store.user_add_threepid(user2, "email", user2_email, now, now) + self.get_success( + self.store.user_add_threepid(user1, "email", user1_email, now, now) + ) + self.get_success( + self.store.user_add_threepid(user2, "email", user2_email, now, now) + ) # XXX why are we doing this here? this function is only run at startup # so it is odd to re-run it here. - self.store.db.runInteraction( - "initialise", self.store._initialise_reserved_users, threepids + self.get_success( + self.store.db.runInteraction( + "initialise", self.store._initialise_reserved_users, threepids + ) ) - self.pump() # the number of users we expect will be counted against the mau limit # -1 because user3 is a support user and does not count @@ -83,13 +89,13 @@ class MonthlyActiveUsersTestCase(unittest.HomeserverTestCase): # Check the number of active users. Ensure user3 (support user) is not counted active_count = self.get_success(self.store.get_monthly_active_count()) - self.assertEquals(active_count, user_num) + self.assertEqual(active_count, user_num) # Test each of the registered users is marked as active - timestamp = self.store.user_last_seen_monthly_active(user1) - self.assertTrue(self.get_success(timestamp)) - timestamp = self.store.user_last_seen_monthly_active(user2) - self.assertTrue(self.get_success(timestamp)) + timestamp = self.get_success(self.store.user_last_seen_monthly_active(user1)) + self.assertGreater(timestamp, 0) + timestamp = self.get_success(self.store.user_last_seen_monthly_active(user2)) + self.assertGreater(timestamp, 0) # Test that users with reserved 3pids are not removed from the MAU table # XXX some of this is redundant. poking things into the config shouldn't @@ -98,77 +104,79 @@ class MonthlyActiveUsersTestCase(unittest.HomeserverTestCase): self.hs.config.max_mau_value = 0 self.reactor.advance(FORTY_DAYS) self.hs.config.max_mau_value = 5 - self.store.reap_monthly_active_users() - self.pump() - active_count = self.store.get_monthly_active_count() - self.assertEquals(self.get_success(active_count), user_num) + self.get_success(self.store.reap_monthly_active_users()) + + active_count = self.get_success(self.store.get_monthly_active_count()) + self.assertEqual(active_count, user_num) # Add some more users and check they are counted as active ru_count = 2 - self.store.upsert_monthly_active_user("@ru1:server") - self.store.upsert_monthly_active_user("@ru2:server") - self.pump() - active_count = self.store.get_monthly_active_count() - self.assertEqual(self.get_success(active_count), user_num + ru_count) + + self.get_success(self.store.upsert_monthly_active_user("@ru1:server")) + self.get_success(self.store.upsert_monthly_active_user("@ru2:server")) + + active_count = self.get_success(self.store.get_monthly_active_count()) + self.assertEqual(active_count, user_num + ru_count) # now run the reaper and check that the number of active users is reduced # to max_mau_value - self.store.reap_monthly_active_users() - self.pump() + self.get_success(self.store.reap_monthly_active_users()) - active_count = self.store.get_monthly_active_count() - self.assertEquals(self.get_success(active_count), 3) + active_count = self.get_success(self.store.get_monthly_active_count()) + self.assertEqual(active_count, 3) def test_can_insert_and_count_mau(self): - count = self.store.get_monthly_active_count() - self.assertEqual(0, self.get_success(count)) + count = self.get_success(self.store.get_monthly_active_count()) + self.assertEqual(count, 0) - self.store.upsert_monthly_active_user("@user:server") - self.pump() + d = self.store.upsert_monthly_active_user("@user:server") + self.get_success(d) - count = self.store.get_monthly_active_count() - self.assertEqual(1, self.get_success(count)) + count = self.get_success(self.store.get_monthly_active_count()) + self.assertEqual(count, 1) def test_user_last_seen_monthly_active(self): user_id1 = "@user1:server" user_id2 = "@user2:server" user_id3 = "@user3:server" - result = self.store.user_last_seen_monthly_active(user_id1) - self.assertFalse(self.get_success(result) == 0) + result = self.get_success(self.store.user_last_seen_monthly_active(user_id1)) + self.assertNotEqual(result, 0) - self.store.upsert_monthly_active_user(user_id1) - self.store.upsert_monthly_active_user(user_id2) - self.pump() + self.get_success(self.store.upsert_monthly_active_user(user_id1)) + self.get_success(self.store.upsert_monthly_active_user(user_id2)) - result = self.store.user_last_seen_monthly_active(user_id1) - self.assertGreater(self.get_success(result), 0) + result = self.get_success(self.store.user_last_seen_monthly_active(user_id1)) + self.assertGreater(result, 0) - result = self.store.user_last_seen_monthly_active(user_id3) - self.assertNotEqual(self.get_success(result), 0) + result = self.get_success(self.store.user_last_seen_monthly_active(user_id3)) + self.assertNotEqual(result, 0) @override_config({"max_mau_value": 5}) def test_reap_monthly_active_users(self): initial_users = 10 for i in range(initial_users): - self.store.upsert_monthly_active_user("@user%d:server" % i) - self.pump() + self.get_success( + self.store.upsert_monthly_active_user("@user%d:server" % i) + ) - count = self.store.get_monthly_active_count() - self.assertTrue(self.get_success(count), initial_users) + count = self.get_success(self.store.get_monthly_active_count()) + self.assertEqual(count, initial_users) - self.store.reap_monthly_active_users() - self.pump() - count = self.store.get_monthly_active_count() - self.assertEquals(self.get_success(count), self.hs.config.max_mau_value) + d = self.store.reap_monthly_active_users() + self.get_success(d) + + count = self.get_success(self.store.get_monthly_active_count()) + self.assertEqual(count, self.hs.config.max_mau_value) self.reactor.advance(FORTY_DAYS) - self.store.reap_monthly_active_users() - self.pump() - count = self.store.get_monthly_active_count() - self.assertEquals(self.get_success(count), 0) + d = self.store.reap_monthly_active_users() + self.get_success(d) + + count = self.get_success(self.store.get_monthly_active_count()) + self.assertEqual(count, 0) # Note that below says mau_limit (no s), this is the name of the config # value, although it gets stored on the config object as mau_limits. @@ -182,7 +190,9 @@ class MonthlyActiveUsersTestCase(unittest.HomeserverTestCase): for i in range(initial_users): user = "@user%d:server" % i email = "user%d@matrix.org" % i + self.get_success(self.store.upsert_monthly_active_user(user)) + # Need to ensure that the most recent entries in the # monthly_active_users table are reserved now = int(self.hs.get_clock().time_msec()) @@ -194,26 +204,37 @@ class MonthlyActiveUsersTestCase(unittest.HomeserverTestCase): self.store.user_add_threepid(user, "email", email, now, now) ) - self.store.db.runInteraction( + d = self.store.db.runInteraction( "initialise", self.store._initialise_reserved_users, threepids ) - count = self.store.get_monthly_active_count() - self.assertTrue(self.get_success(count), initial_users) + self.get_success(d) - users = self.store.get_registered_reserved_users() - self.assertEquals(len(self.get_success(users)), reserved_user_number) + count = self.get_success(self.store.get_monthly_active_count()) + self.assertEqual(count, initial_users) - self.get_success(self.store.reap_monthly_active_users()) - count = self.store.get_monthly_active_count() - self.assertEquals(self.get_success(count), self.hs.config.max_mau_value) + users = self.get_success(self.store.get_registered_reserved_users()) + self.assertEqual(len(users), reserved_user_number) + + d = self.store.reap_monthly_active_users() + self.get_success(d) + + count = self.get_success(self.store.get_monthly_active_count()) + self.assertEqual(count, self.hs.config.max_mau_value) def test_populate_monthly_users_is_guest(self): # Test that guest users are not added to mau list user_id = "@user_id:host" - self.store.register_user(user_id=user_id, password_hash=None, make_guest=True) + + d = self.store.register_user( + user_id=user_id, password_hash=None, make_guest=True + ) + self.get_success(d) + self.store.upsert_monthly_active_user = Mock() - self.store.populate_monthly_active_users(user_id) - self.pump() + + d = self.store.populate_monthly_active_users(user_id) + self.get_success(d) + self.store.upsert_monthly_active_user.assert_not_called() def test_populate_monthly_users_should_update(self): @@ -224,8 +245,9 @@ class MonthlyActiveUsersTestCase(unittest.HomeserverTestCase): self.store.user_last_seen_monthly_active = Mock( return_value=defer.succeed(None) ) - self.store.populate_monthly_active_users("user_id") - self.pump() + d = self.store.populate_monthly_active_users("user_id") + self.get_success(d) + self.store.upsert_monthly_active_user.assert_called_once() def test_populate_monthly_users_should_not_update(self): @@ -235,16 +257,18 @@ class MonthlyActiveUsersTestCase(unittest.HomeserverTestCase): self.store.user_last_seen_monthly_active = Mock( return_value=defer.succeed(self.hs.get_clock().time_msec()) ) - self.store.populate_monthly_active_users("user_id") - self.pump() + + d = self.store.populate_monthly_active_users("user_id") + self.get_success(d) + self.store.upsert_monthly_active_user.assert_not_called() def test_get_reserved_real_user_account(self): # Test no reserved users, or reserved threepids users = self.get_success(self.store.get_registered_reserved_users()) - self.assertEquals(len(users), 0) - # Test reserved users but no registered users + self.assertEqual(len(users), 0) + # Test reserved users but no registered users user1 = "@user1:example.com" user2 = "@user2:example.com" @@ -254,63 +278,64 @@ class MonthlyActiveUsersTestCase(unittest.HomeserverTestCase): {"medium": "email", "address": user1_email}, {"medium": "email", "address": user2_email}, ] + self.hs.config.mau_limits_reserved_threepids = threepids - self.store.db.runInteraction( + d = self.store.db.runInteraction( "initialise", self.store._initialise_reserved_users, threepids ) + self.get_success(d) - self.pump() users = self.get_success(self.store.get_registered_reserved_users()) - self.assertEquals(len(users), 0) + self.assertEqual(len(users), 0) - # Test reserved registed users - self.store.register_user(user_id=user1, password_hash=None) - self.store.register_user(user_id=user2, password_hash=None) - self.pump() + # Test reserved registered users + self.get_success(self.store.register_user(user_id=user1, password_hash=None)) + self.get_success(self.store.register_user(user_id=user2, password_hash=None)) now = int(self.hs.get_clock().time_msec()) self.store.user_add_threepid(user1, "email", user1_email, now, now) self.store.user_add_threepid(user2, "email", user2_email, now, now) users = self.get_success(self.store.get_registered_reserved_users()) - self.assertEquals(len(users), len(threepids)) + self.assertEqual(len(users), len(threepids)) def test_support_user_not_add_to_mau_limits(self): support_user_id = "@support:test" - count = self.store.get_monthly_active_count() - self.pump() - self.assertEqual(self.get_success(count), 0) - self.store.register_user( + count = self.get_success(self.store.get_monthly_active_count()) + self.assertEqual(count, 0) + + d = self.store.register_user( user_id=support_user_id, password_hash=None, user_type=UserTypes.SUPPORT ) + self.get_success(d) - self.store.upsert_monthly_active_user(support_user_id) - count = self.store.get_monthly_active_count() - self.pump() - self.assertEqual(self.get_success(count), 0) + d = self.store.upsert_monthly_active_user(support_user_id) + self.get_success(d) + + d = self.store.get_monthly_active_count() + count = self.get_success(d) + self.assertEqual(count, 0) # Note that the max_mau_value setting should not matter. @override_config( {"limit_usage_by_mau": False, "mau_stats_only": True, "max_mau_value": 1} ) def test_track_monthly_users_without_cap(self): - count = self.store.get_monthly_active_count() - self.assertEqual(0, self.get_success(count)) + count = self.get_success(self.store.get_monthly_active_count()) + self.assertEqual(0, count) - self.store.upsert_monthly_active_user("@user1:server") - self.store.upsert_monthly_active_user("@user2:server") - self.pump() + self.get_success(self.store.upsert_monthly_active_user("@user1:server")) + self.get_success(self.store.upsert_monthly_active_user("@user2:server")) - count = self.store.get_monthly_active_count() - self.assertEqual(2, self.get_success(count)) + count = self.get_success(self.store.get_monthly_active_count()) + self.assertEqual(2, count) @override_config({"limit_usage_by_mau": False, "mau_stats_only": False}) def test_no_users_when_not_tracking(self): self.store.upsert_monthly_active_user = Mock() - self.store.populate_monthly_active_users("@user:sever") - self.pump() + self.get_success(self.store.populate_monthly_active_users("@user:sever")) self.store.upsert_monthly_active_user.assert_not_called() @@ -325,33 +350,39 @@ class MonthlyActiveUsersTestCase(unittest.HomeserverTestCase): service2 = "service2" native = "native" - self.store.register_user( - user_id=appservice1_user1, password_hash=None, appservice_id=service1 + self.get_success( + self.store.register_user( + user_id=appservice1_user1, password_hash=None, appservice_id=service1 + ) + ) + self.get_success( + self.store.register_user( + user_id=appservice1_user2, password_hash=None, appservice_id=service1 + ) ) - self.store.register_user( - user_id=appservice1_user2, password_hash=None, appservice_id=service1 + self.get_success( + self.store.register_user( + user_id=appservice2_user1, password_hash=None, appservice_id=service2 + ) ) - self.store.register_user( - user_id=appservice2_user1, password_hash=None, appservice_id=service2 + self.get_success( + self.store.register_user(user_id=native_user1, password_hash=None) ) - self.store.register_user(user_id=native_user1, password_hash=None) - self.pump() - count = self.store.get_monthly_active_count_by_service() - self.assertEqual({}, self.get_success(count)) + count = self.get_success(self.store.get_monthly_active_count_by_service()) + self.assertEqual(count, {}) - self.store.upsert_monthly_active_user(native_user1) - self.store.upsert_monthly_active_user(appservice1_user1) - self.store.upsert_monthly_active_user(appservice1_user2) - self.store.upsert_monthly_active_user(appservice2_user1) - self.pump() + self.get_success(self.store.upsert_monthly_active_user(native_user1)) + self.get_success(self.store.upsert_monthly_active_user(appservice1_user1)) + self.get_success(self.store.upsert_monthly_active_user(appservice1_user2)) + self.get_success(self.store.upsert_monthly_active_user(appservice2_user1)) - count = self.store.get_monthly_active_count() - self.assertEqual(4, self.get_success(count)) + count = self.get_success(self.store.get_monthly_active_count()) + self.assertEqual(count, 4) - count = self.store.get_monthly_active_count_by_service() - result = self.get_success(count) + d = self.store.get_monthly_active_count_by_service() + result = self.get_success(d) - self.assertEqual(2, result[service1]) - self.assertEqual(1, result[service2]) - self.assertEqual(1, result[native]) + self.assertEqual(result[service1], 2) + self.assertEqual(result[service2], 1) + self.assertEqual(result[native], 1) diff --git a/tests/test_mau.py b/tests/test_mau.py index 8a97f0998d..49667ed7f4 100644 --- a/tests/test_mau.py +++ b/tests/test_mau.py @@ -85,7 +85,7 @@ class TestMauLimit(unittest.HomeserverTestCase): # Advance time by 31 days self.reactor.advance(31 * 24 * 60 * 60) - self.store.reap_monthly_active_users() + self.get_success(self.store.reap_monthly_active_users()) self.reactor.advance(0) @@ -147,8 +147,7 @@ class TestMauLimit(unittest.HomeserverTestCase): # Advance by 2 months so everyone falls out of MAU self.reactor.advance(60 * 24 * 60 * 60) - self.store.reap_monthly_active_users() - self.reactor.advance(0) + self.get_success(self.store.reap_monthly_active_users()) # We can create as many new users as we want token4 = self.create_user("kermit4") -- cgit 1.4.1 From c389bfb6eac421b84d3f91b344e8f3a91e421e83 Mon Sep 17 00:00:00 2001 From: Andrew Morgan <1342360+anoadragon453@users.noreply.github.com> Date: Thu, 4 Jun 2020 20:03:40 +0100 Subject: Fix encryption algorithm typos in tests/comments (#7637) @uhoreg has confirmed these were both typos. They are only in comments and tests though, rather than anything critical. Introduced in: * https://github.com/matrix-org/synapse/pull/7157 * https://github.com/matrix-org/synapse/pull/5726 --- changelog.d/7637.misc | 1 + synapse/rest/client/v2_alpha/keys.py | 8 ++++---- tests/federation/test_federation_sender.py | 2 +- tests/handlers/test_e2e_keys.py | 10 +++++----- 4 files changed, 11 insertions(+), 10 deletions(-) create mode 100644 changelog.d/7637.misc (limited to 'tests') diff --git a/changelog.d/7637.misc b/changelog.d/7637.misc new file mode 100644 index 0000000000..90d41fa775 --- /dev/null +++ b/changelog.d/7637.misc @@ -0,0 +1 @@ +Fix typos of `m.olm.curve25519-aes-sha2` and `m.megolm.v1.aes-sha2` in comments, test files. \ No newline at end of file diff --git a/synapse/rest/client/v2_alpha/keys.py b/synapse/rest/client/v2_alpha/keys.py index 8f41a3edbf..24bb090822 100644 --- a/synapse/rest/client/v2_alpha/keys.py +++ b/synapse/rest/client/v2_alpha/keys.py @@ -42,7 +42,7 @@ class KeyUploadServlet(RestServlet): "device_id": "", "valid_until_ts": , "algorithms": [ - "m.olm.curve25519-aes-sha256", + "m.olm.curve25519-aes-sha2", ] "keys": { ":": "", @@ -124,7 +124,7 @@ class KeyQueryServlet(RestServlet): "device_id": "", // Duplicated to be signed "valid_until_ts": , "algorithms": [ // List of supported algorithms - "m.olm.curve25519-aes-sha256", + "m.olm.curve25519-aes-sha2", ], "keys": { // Must include a ed25519 signing key ":": "", @@ -285,8 +285,8 @@ class SignaturesUploadServlet(RestServlet): "user_id": "", "device_id": "", "algorithms": [ - "m.olm.curve25519-aes-sha256", - "m.megolm.v1.aes-sha" + "m.olm.curve25519-aes-sha2", + "m.megolm.v1.aes-sha2" ], "keys": { ":": "", diff --git a/tests/federation/test_federation_sender.py b/tests/federation/test_federation_sender.py index 33105576af..ff12539041 100644 --- a/tests/federation/test_federation_sender.py +++ b/tests/federation/test_federation_sender.py @@ -536,7 +536,7 @@ def build_device_dict(user_id: str, device_id: str, sk: SigningKey): return { "user_id": user_id, "device_id": device_id, - "algorithms": ["m.olm.curve25519-aes-sha256", "m.megolm.v1.aes-sha"], + "algorithms": ["m.olm.curve25519-aes-sha2", "m.megolm.v1.aes-sha2"], "keys": { "curve25519:" + device_id: "curve25519+key", key_id(sk): encode_pubkey(sk), diff --git a/tests/handlers/test_e2e_keys.py b/tests/handlers/test_e2e_keys.py index 854eb6c024..e1e144b2e7 100644 --- a/tests/handlers/test_e2e_keys.py +++ b/tests/handlers/test_e2e_keys.py @@ -222,7 +222,7 @@ class E2eKeysHandlerTestCase(unittest.TestCase): device_key_1 = { "user_id": local_user, "device_id": "abc", - "algorithms": ["m.olm.curve25519-aes-sha256", "m.megolm.v1.aes-sha"], + "algorithms": ["m.olm.curve25519-aes-sha2", "m.megolm.v1.aes-sha2"], "keys": { "ed25519:abc": "base64+ed25519+key", "curve25519:abc": "base64+curve25519+key", @@ -232,7 +232,7 @@ class E2eKeysHandlerTestCase(unittest.TestCase): device_key_2 = { "user_id": local_user, "device_id": "def", - "algorithms": ["m.olm.curve25519-aes-sha256", "m.megolm.v1.aes-sha"], + "algorithms": ["m.olm.curve25519-aes-sha2", "m.megolm.v1.aes-sha2"], "keys": { "ed25519:def": "base64+ed25519+key", "curve25519:def": "base64+curve25519+key", @@ -315,7 +315,7 @@ class E2eKeysHandlerTestCase(unittest.TestCase): device_key = { "user_id": local_user, "device_id": device_id, - "algorithms": ["m.olm.curve25519-aes-sha256", "m.megolm.v1.aes-sha"], + "algorithms": ["m.olm.curve25519-aes-sha2", "m.megolm.v1.aes-sha2"], "keys": {"curve25519:xyz": "curve25519+key", "ed25519:xyz": device_pubkey}, "signatures": {local_user: {"ed25519:xyz": "something"}}, } @@ -391,8 +391,8 @@ class E2eKeysHandlerTestCase(unittest.TestCase): "user_id": local_user, "device_id": device_id, "algorithms": [ - "m.olm.curve25519-aes-sha256", - "m.megolm.v1.aes-sha", + "m.olm.curve25519-aes-sha2", + "m.megolm.v1.aes-sha2", ], "keys": { "curve25519:xyz": "curve25519+key", -- cgit 1.4.1 From f4e6495b5d3267976f34088fa7459b388b801eb6 Mon Sep 17 00:00:00 2001 From: Andrew Morgan <1342360+anoadragon453@users.noreply.github.com> Date: Fri, 5 Jun 2020 10:47:20 +0100 Subject: Performance improvements and refactor of Ratelimiter (#7595) While working on https://github.com/matrix-org/synapse/issues/5665 I found myself digging into the `Ratelimiter` class and seeing that it was both: * Rather undocumented, and * causing a *lot* of config checks This PR attempts to refactor and comment the `Ratelimiter` class, as well as encourage config file accesses to only be done at instantiation. Best to be reviewed commit-by-commit. --- changelog.d/7595.misc | 1 + synapse/api/ratelimiting.py | 153 +++++++++++++++++++++------- synapse/config/ratelimiting.py | 8 +- synapse/handlers/_base.py | 60 ++++++----- synapse/handlers/auth.py | 24 ++--- synapse/handlers/message.py | 1 - synapse/handlers/register.py | 9 +- synapse/rest/client/v1/login.py | 65 ++++-------- synapse/rest/client/v2_alpha/register.py | 16 +-- synapse/server.py | 17 ++-- synapse/util/ratelimitutils.py | 2 +- tests/api/test_ratelimiting.py | 96 +++++++++++++---- tests/handlers/test_profile.py | 6 +- tests/replication/slave/storage/_base.py | 9 +- tests/rest/client/v1/test_events.py | 8 +- tests/rest/client/v1/test_login.py | 49 +++++++-- tests/rest/client/v1/test_rooms.py | 9 +- tests/rest/client/v1/test_typing.py | 10 +- tests/rest/client/v2_alpha/test_register.py | 9 +- 19 files changed, 322 insertions(+), 230 deletions(-) create mode 100644 changelog.d/7595.misc (limited to 'tests') diff --git a/changelog.d/7595.misc b/changelog.d/7595.misc new file mode 100644 index 0000000000..7a0646b1a3 --- /dev/null +++ b/changelog.d/7595.misc @@ -0,0 +1 @@ +Refactor `Ratelimiter` to limit the amount of expensive config value accesses. diff --git a/synapse/api/ratelimiting.py b/synapse/api/ratelimiting.py index 7a049b3af7..ec6b3a69a2 100644 --- a/synapse/api/ratelimiting.py +++ b/synapse/api/ratelimiting.py @@ -1,4 +1,5 @@ # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2020 The Matrix.org Foundation C.I.C. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -16,75 +17,157 @@ from collections import OrderedDict from typing import Any, Optional, Tuple from synapse.api.errors import LimitExceededError +from synapse.util import Clock class Ratelimiter(object): """ - Ratelimit message sending by user. + Ratelimit actions marked by arbitrary keys. + + Args: + clock: A homeserver clock, for retrieving the current time + rate_hz: The long term number of actions that can be performed in a second. + burst_count: How many actions that can be performed before being limited. """ - def __init__(self): - self.message_counts = ( - OrderedDict() - ) # type: OrderedDict[Any, Tuple[float, int, Optional[float]]] + def __init__(self, clock: Clock, rate_hz: float, burst_count: int): + self.clock = clock + self.rate_hz = rate_hz + self.burst_count = burst_count + + # A ordered dictionary keeping track of actions, when they were last + # performed and how often. Each entry is a mapping from a key of arbitrary type + # to a tuple representing: + # * How many times an action has occurred since a point in time + # * The point in time + # * The rate_hz of this particular entry. This can vary per request + self.actions = OrderedDict() # type: OrderedDict[Any, Tuple[float, int, float]] - def can_do_action(self, key, time_now_s, rate_hz, burst_count, update=True): + def can_do_action( + self, + key: Any, + rate_hz: Optional[float] = None, + burst_count: Optional[int] = None, + update: bool = True, + _time_now_s: Optional[int] = None, + ) -> Tuple[bool, float]: """Can the entity (e.g. user or IP address) perform the action? + Args: key: The key we should use when rate limiting. Can be a user ID (when sending events), an IP address, etc. - time_now_s: The time now. - rate_hz: The long term number of messages a user can send in a - second. - burst_count: How many messages the user can send before being - limited. - update (bool): Whether to update the message rates or not. This is - useful to check if a message would be allowed to be sent before - its ready to be actually sent. + rate_hz: The long term number of actions that can be performed in a second. + Overrides the value set during instantiation if set. + burst_count: How many actions that can be performed before being limited. + Overrides the value set during instantiation if set. + update: Whether to count this check as performing the action + _time_now_s: The current time. Optional, defaults to the current time according + to self.clock. Only used by tests. + Returns: - A pair of a bool indicating if they can send a message now and a - time in seconds of when they can next send a message. + A tuple containing: + * A bool indicating if they can perform the action now + * The reactor timestamp for when the action can be performed next. + -1 if rate_hz is less than or equal to zero """ - self.prune_message_counts(time_now_s) - message_count, time_start, _ignored = self.message_counts.get( - key, (0.0, time_now_s, None) - ) + # Override default values if set + time_now_s = _time_now_s if _time_now_s is not None else self.clock.time() + rate_hz = rate_hz if rate_hz is not None else self.rate_hz + burst_count = burst_count if burst_count is not None else self.burst_count + + # Remove any expired entries + self._prune_message_counts(time_now_s) + + # Check if there is an existing count entry for this key + action_count, time_start, _ = self.actions.get(key, (0.0, time_now_s, 0.0)) + + # Check whether performing another action is allowed time_delta = time_now_s - time_start - sent_count = message_count - time_delta * rate_hz - if sent_count < 0: + performed_count = action_count - time_delta * rate_hz + if performed_count < 0: + # Allow, reset back to count 1 allowed = True time_start = time_now_s - message_count = 1.0 - elif sent_count > burst_count - 1.0: + action_count = 1.0 + elif performed_count > burst_count - 1.0: + # Deny, we have exceeded our burst count allowed = False else: + # We haven't reached our limit yet allowed = True - message_count += 1 + action_count += 1.0 if update: - self.message_counts[key] = (message_count, time_start, rate_hz) + self.actions[key] = (action_count, time_start, rate_hz) if rate_hz > 0: - time_allowed = time_start + (message_count - burst_count + 1) / rate_hz + # Find out when the count of existing actions expires + time_allowed = time_start + (action_count - burst_count + 1) / rate_hz + + # Don't give back a time in the past if time_allowed < time_now_s: time_allowed = time_now_s + else: + # XXX: Why is this -1? This seems to only be used in + # self.ratelimit. I guess so that clients get a time in the past and don't + # feel afraid to try again immediately time_allowed = -1 return allowed, time_allowed - def prune_message_counts(self, time_now_s): - for key in list(self.message_counts.keys()): - message_count, time_start, rate_hz = self.message_counts[key] + def _prune_message_counts(self, time_now_s: int): + """Remove message count entries that have not exceeded their defined + rate_hz limit + + Args: + time_now_s: The current time + """ + # We create a copy of the key list here as the dictionary is modified during + # the loop + for key in list(self.actions.keys()): + action_count, time_start, rate_hz = self.actions[key] + + # Rate limit = "seconds since we started limiting this action" * rate_hz + # If this limit has not been exceeded, wipe our record of this action time_delta = time_now_s - time_start - if message_count - time_delta * rate_hz > 0: - break + if action_count - time_delta * rate_hz > 0: + continue else: - del self.message_counts[key] + del self.actions[key] + + def ratelimit( + self, + key: Any, + rate_hz: Optional[float] = None, + burst_count: Optional[int] = None, + update: bool = True, + _time_now_s: Optional[int] = None, + ): + """Checks if an action can be performed. If not, raises a LimitExceededError + + Args: + key: An arbitrary key used to classify an action + rate_hz: The long term number of actions that can be performed in a second. + Overrides the value set during instantiation if set. + burst_count: How many actions that can be performed before being limited. + Overrides the value set during instantiation if set. + update: Whether to count this check as performing the action + _time_now_s: The current time. Optional, defaults to the current time according + to self.clock. Only used by tests. + + Raises: + LimitExceededError: If an action could not be performed, along with the time in + milliseconds until the action can be performed again + """ + time_now_s = _time_now_s if _time_now_s is not None else self.clock.time() - def ratelimit(self, key, time_now_s, rate_hz, burst_count, update=True): allowed, time_allowed = self.can_do_action( - key, time_now_s, rate_hz, burst_count, update + key, + rate_hz=rate_hz, + burst_count=burst_count, + update=update, + _time_now_s=time_now_s, ) if not allowed: diff --git a/synapse/config/ratelimiting.py b/synapse/config/ratelimiting.py index 4a3bfc4354..2dd94bae2b 100644 --- a/synapse/config/ratelimiting.py +++ b/synapse/config/ratelimiting.py @@ -12,11 +12,17 @@ # See the License for the specific language governing permissions and # limitations under the License. +from typing import Dict + from ._base import Config class RateLimitConfig(object): - def __init__(self, config, defaults={"per_second": 0.17, "burst_count": 3.0}): + def __init__( + self, + config: Dict[str, float], + defaults={"per_second": 0.17, "burst_count": 3.0}, + ): self.per_second = config.get("per_second", defaults["per_second"]) self.burst_count = config.get("burst_count", defaults["burst_count"]) diff --git a/synapse/handlers/_base.py b/synapse/handlers/_base.py index 3b781d9836..61dc4beafe 100644 --- a/synapse/handlers/_base.py +++ b/synapse/handlers/_base.py @@ -19,7 +19,7 @@ from twisted.internet import defer import synapse.types from synapse.api.constants import EventTypes, Membership -from synapse.api.errors import LimitExceededError +from synapse.api.ratelimiting import Ratelimiter from synapse.types import UserID logger = logging.getLogger(__name__) @@ -44,11 +44,26 @@ class BaseHandler(object): self.notifier = hs.get_notifier() self.state_handler = hs.get_state_handler() self.distributor = hs.get_distributor() - self.ratelimiter = hs.get_ratelimiter() - self.admin_redaction_ratelimiter = hs.get_admin_redaction_ratelimiter() self.clock = hs.get_clock() self.hs = hs + # The rate_hz and burst_count are overridden on a per-user basis + self.request_ratelimiter = Ratelimiter( + clock=self.clock, rate_hz=0, burst_count=0 + ) + self._rc_message = self.hs.config.rc_message + + # Check whether ratelimiting room admin message redaction is enabled + # by the presence of rate limits in the config + if self.hs.config.rc_admin_redaction: + self.admin_redaction_ratelimiter = Ratelimiter( + clock=self.clock, + rate_hz=self.hs.config.rc_admin_redaction.per_second, + burst_count=self.hs.config.rc_admin_redaction.burst_count, + ) + else: + self.admin_redaction_ratelimiter = None + self.server_name = hs.hostname self.event_builder_factory = hs.get_event_builder_factory() @@ -70,7 +85,6 @@ class BaseHandler(object): Raises: LimitExceededError if the request should be ratelimited """ - time_now = self.clock.time() user_id = requester.user.to_string() # The AS user itself is never rate limited. @@ -83,48 +97,32 @@ class BaseHandler(object): if requester.app_service and not requester.app_service.is_rate_limited(): return + messages_per_second = self._rc_message.per_second + burst_count = self._rc_message.burst_count + # Check if there is a per user override in the DB. override = yield self.store.get_ratelimit_for_user(user_id) if override: - # If overriden with a null Hz then ratelimiting has been entirely + # If overridden with a null Hz then ratelimiting has been entirely # disabled for the user if not override.messages_per_second: return messages_per_second = override.messages_per_second burst_count = override.burst_count + + if is_admin_redaction and self.admin_redaction_ratelimiter: + # If we have separate config for admin redactions, use a separate + # ratelimiter as to not have user_ids clash + self.admin_redaction_ratelimiter.ratelimit(user_id, update=update) else: - # We default to different values if this is an admin redaction and - # the config is set - if is_admin_redaction and self.hs.config.rc_admin_redaction: - messages_per_second = self.hs.config.rc_admin_redaction.per_second - burst_count = self.hs.config.rc_admin_redaction.burst_count - else: - messages_per_second = self.hs.config.rc_message.per_second - burst_count = self.hs.config.rc_message.burst_count - - if is_admin_redaction and self.hs.config.rc_admin_redaction: - # If we have separate config for admin redactions we use a separate - # ratelimiter - allowed, time_allowed = self.admin_redaction_ratelimiter.can_do_action( - user_id, - time_now, - rate_hz=messages_per_second, - burst_count=burst_count, - update=update, - ) - else: - allowed, time_allowed = self.ratelimiter.can_do_action( + # Override rate and burst count per-user + self.request_ratelimiter.ratelimit( user_id, - time_now, rate_hz=messages_per_second, burst_count=burst_count, update=update, ) - if not allowed: - raise LimitExceededError( - retry_after_ms=int(1000 * (time_allowed - time_now)) - ) async def maybe_kick_guest_users(self, event, context=None): # Technically this function invalidates current_state by changing it. diff --git a/synapse/handlers/auth.py b/synapse/handlers/auth.py index 75b39e878c..119678e67b 100644 --- a/synapse/handlers/auth.py +++ b/synapse/handlers/auth.py @@ -108,7 +108,11 @@ class AuthHandler(BaseHandler): # Ratelimiter for failed auth during UIA. Uses same ratelimit config # as per `rc_login.failed_attempts`. - self._failed_uia_attempts_ratelimiter = Ratelimiter() + self._failed_uia_attempts_ratelimiter = Ratelimiter( + clock=self.clock, + rate_hz=self.hs.config.rc_login_failed_attempts.per_second, + burst_count=self.hs.config.rc_login_failed_attempts.burst_count, + ) self._clock = self.hs.get_clock() @@ -196,13 +200,7 @@ class AuthHandler(BaseHandler): user_id = requester.user.to_string() # Check if we should be ratelimited due to too many previous failed attempts - self._failed_uia_attempts_ratelimiter.ratelimit( - user_id, - time_now_s=self._clock.time(), - rate_hz=self.hs.config.rc_login_failed_attempts.per_second, - burst_count=self.hs.config.rc_login_failed_attempts.burst_count, - update=False, - ) + self._failed_uia_attempts_ratelimiter.ratelimit(user_id, update=False) # build a list of supported flows flows = [[login_type] for login_type in self._supported_ui_auth_types] @@ -212,14 +210,8 @@ class AuthHandler(BaseHandler): flows, request, request_body, clientip, description ) except LoginError: - # Update the ratelimite to say we failed (`can_do_action` doesn't raise). - self._failed_uia_attempts_ratelimiter.can_do_action( - user_id, - time_now_s=self._clock.time(), - rate_hz=self.hs.config.rc_login_failed_attempts.per_second, - burst_count=self.hs.config.rc_login_failed_attempts.burst_count, - update=True, - ) + # Update the ratelimiter to say we failed (`can_do_action` doesn't raise). + self._failed_uia_attempts_ratelimiter.can_do_action(user_id) raise # find the completed login type diff --git a/synapse/handlers/message.py b/synapse/handlers/message.py index 681f92cafd..649ca1f08a 100644 --- a/synapse/handlers/message.py +++ b/synapse/handlers/message.py @@ -362,7 +362,6 @@ class EventCreationHandler(object): self.profile_handler = hs.get_profile_handler() self.event_builder_factory = hs.get_event_builder_factory() self.server_name = hs.hostname - self.ratelimiter = hs.get_ratelimiter() self.notifier = hs.get_notifier() self.config = hs.config self.require_membership_for_aliases = hs.config.require_membership_for_aliases diff --git a/synapse/handlers/register.py b/synapse/handlers/register.py index 55a03e53ea..cd746be7c8 100644 --- a/synapse/handlers/register.py +++ b/synapse/handlers/register.py @@ -425,14 +425,7 @@ class RegistrationHandler(BaseHandler): if not address: return - time_now = self.clock.time() - - self.ratelimiter.ratelimit( - address, - time_now_s=time_now, - rate_hz=self.hs.config.rc_registration.per_second, - burst_count=self.hs.config.rc_registration.burst_count, - ) + self.ratelimiter.ratelimit(address) def register_with_store( self, diff --git a/synapse/rest/client/v1/login.py b/synapse/rest/client/v1/login.py index 6ac7c5142b..dceb2792fa 100644 --- a/synapse/rest/client/v1/login.py +++ b/synapse/rest/client/v1/login.py @@ -87,11 +87,22 @@ class LoginRestServlet(RestServlet): self.auth_handler = self.hs.get_auth_handler() self.registration_handler = hs.get_registration_handler() self.handlers = hs.get_handlers() - self._clock = hs.get_clock() self._well_known_builder = WellKnownBuilder(hs) - self._address_ratelimiter = Ratelimiter() - self._account_ratelimiter = Ratelimiter() - self._failed_attempts_ratelimiter = Ratelimiter() + self._address_ratelimiter = Ratelimiter( + clock=hs.get_clock(), + rate_hz=self.hs.config.rc_login_address.per_second, + burst_count=self.hs.config.rc_login_address.burst_count, + ) + self._account_ratelimiter = Ratelimiter( + clock=hs.get_clock(), + rate_hz=self.hs.config.rc_login_account.per_second, + burst_count=self.hs.config.rc_login_account.burst_count, + ) + self._failed_attempts_ratelimiter = Ratelimiter( + clock=hs.get_clock(), + rate_hz=self.hs.config.rc_login_failed_attempts.per_second, + burst_count=self.hs.config.rc_login_failed_attempts.burst_count, + ) def on_GET(self, request): flows = [] @@ -124,13 +135,7 @@ class LoginRestServlet(RestServlet): return 200, {} async def on_POST(self, request): - self._address_ratelimiter.ratelimit( - request.getClientIP(), - time_now_s=self.hs.clock.time(), - rate_hz=self.hs.config.rc_login_address.per_second, - burst_count=self.hs.config.rc_login_address.burst_count, - update=True, - ) + self._address_ratelimiter.ratelimit(request.getClientIP()) login_submission = parse_json_object_from_request(request) try: @@ -198,13 +203,7 @@ class LoginRestServlet(RestServlet): # We also apply account rate limiting using the 3PID as a key, as # otherwise using 3PID bypasses the ratelimiting based on user ID. - self._failed_attempts_ratelimiter.ratelimit( - (medium, address), - time_now_s=self._clock.time(), - rate_hz=self.hs.config.rc_login_failed_attempts.per_second, - burst_count=self.hs.config.rc_login_failed_attempts.burst_count, - update=False, - ) + self._failed_attempts_ratelimiter.ratelimit((medium, address), update=False) # Check for login providers that support 3pid login types ( @@ -238,13 +237,7 @@ class LoginRestServlet(RestServlet): # If it returned None but the 3PID was bound then we won't hit # this code path, which is fine as then the per-user ratelimit # will kick in below. - self._failed_attempts_ratelimiter.can_do_action( - (medium, address), - time_now_s=self._clock.time(), - rate_hz=self.hs.config.rc_login_failed_attempts.per_second, - burst_count=self.hs.config.rc_login_failed_attempts.burst_count, - update=True, - ) + self._failed_attempts_ratelimiter.can_do_action((medium, address)) raise LoginError(403, "", errcode=Codes.FORBIDDEN) identifier = {"type": "m.id.user", "user": user_id} @@ -263,11 +256,7 @@ class LoginRestServlet(RestServlet): # Check if we've hit the failed ratelimit (but don't update it) self._failed_attempts_ratelimiter.ratelimit( - qualified_user_id.lower(), - time_now_s=self._clock.time(), - rate_hz=self.hs.config.rc_login_failed_attempts.per_second, - burst_count=self.hs.config.rc_login_failed_attempts.burst_count, - update=False, + qualified_user_id.lower(), update=False ) try: @@ -279,13 +268,7 @@ class LoginRestServlet(RestServlet): # limiter. Using `can_do_action` avoids us raising a ratelimit # exception and masking the LoginError. The actual ratelimiting # should have happened above. - self._failed_attempts_ratelimiter.can_do_action( - qualified_user_id.lower(), - time_now_s=self._clock.time(), - rate_hz=self.hs.config.rc_login_failed_attempts.per_second, - burst_count=self.hs.config.rc_login_failed_attempts.burst_count, - update=True, - ) + self._failed_attempts_ratelimiter.can_do_action(qualified_user_id.lower()) raise result = await self._complete_login( @@ -318,13 +301,7 @@ class LoginRestServlet(RestServlet): # Before we actually log them in we check if they've already logged in # too often. This happens here rather than before as we don't # necessarily know the user before now. - self._account_ratelimiter.ratelimit( - user_id.lower(), - time_now_s=self._clock.time(), - rate_hz=self.hs.config.rc_login_account.per_second, - burst_count=self.hs.config.rc_login_account.burst_count, - update=True, - ) + self._account_ratelimiter.ratelimit(user_id.lower()) if create_non_existent_users: canonical_uid = await self.auth_handler.check_user_exists(user_id) diff --git a/synapse/rest/client/v2_alpha/register.py b/synapse/rest/client/v2_alpha/register.py index addd4cae19..b9ffe86b2a 100644 --- a/synapse/rest/client/v2_alpha/register.py +++ b/synapse/rest/client/v2_alpha/register.py @@ -26,7 +26,6 @@ import synapse.types from synapse.api.constants import LoginType from synapse.api.errors import ( Codes, - LimitExceededError, SynapseError, ThreepidValidationError, UnrecognizedRequestError, @@ -396,20 +395,7 @@ class RegisterRestServlet(RestServlet): client_addr = request.getClientIP() - time_now = self.clock.time() - - allowed, time_allowed = self.ratelimiter.can_do_action( - client_addr, - time_now_s=time_now, - rate_hz=self.hs.config.rc_registration.per_second, - burst_count=self.hs.config.rc_registration.burst_count, - update=False, - ) - - if not allowed: - raise LimitExceededError( - retry_after_ms=int(1000 * (time_allowed - time_now)) - ) + self.ratelimiter.ratelimit(client_addr, update=False) kind = b"user" if b"kind" in request.args: diff --git a/synapse/server.py b/synapse/server.py index ca2deb49bb..fe94836a2c 100644 --- a/synapse/server.py +++ b/synapse/server.py @@ -242,9 +242,12 @@ class HomeServer(object): self.clock = Clock(reactor) self.distributor = Distributor() - self.ratelimiter = Ratelimiter() - self.admin_redaction_ratelimiter = Ratelimiter() - self.registration_ratelimiter = Ratelimiter() + + self.registration_ratelimiter = Ratelimiter( + clock=self.clock, + rate_hz=config.rc_registration.per_second, + burst_count=config.rc_registration.burst_count, + ) self.datastores = None @@ -314,15 +317,9 @@ class HomeServer(object): def get_distributor(self): return self.distributor - def get_ratelimiter(self): - return self.ratelimiter - - def get_registration_ratelimiter(self): + def get_registration_ratelimiter(self) -> Ratelimiter: return self.registration_ratelimiter - def get_admin_redaction_ratelimiter(self): - return self.admin_redaction_ratelimiter - def build_federation_client(self): return FederationClient(self) diff --git a/synapse/util/ratelimitutils.py b/synapse/util/ratelimitutils.py index 5ca4521ce3..e5efdfcd02 100644 --- a/synapse/util/ratelimitutils.py +++ b/synapse/util/ratelimitutils.py @@ -43,7 +43,7 @@ class FederationRateLimiter(object): self.ratelimiters = collections.defaultdict(new_limiter) def ratelimit(self, host): - """Used to ratelimit an incoming request from given host + """Used to ratelimit an incoming request from a given host Example usage: diff --git a/tests/api/test_ratelimiting.py b/tests/api/test_ratelimiting.py index dbdd427cac..d580e729c5 100644 --- a/tests/api/test_ratelimiting.py +++ b/tests/api/test_ratelimiting.py @@ -1,39 +1,97 @@ -from synapse.api.ratelimiting import Ratelimiter +from synapse.api.ratelimiting import LimitExceededError, Ratelimiter from tests import unittest class TestRatelimiter(unittest.TestCase): - def test_allowed(self): - limiter = Ratelimiter() - allowed, time_allowed = limiter.can_do_action( - key="test_id", time_now_s=0, rate_hz=0.1, burst_count=1 - ) + def test_allowed_via_can_do_action(self): + limiter = Ratelimiter(clock=None, rate_hz=0.1, burst_count=1) + allowed, time_allowed = limiter.can_do_action(key="test_id", _time_now_s=0) self.assertTrue(allowed) self.assertEquals(10.0, time_allowed) - allowed, time_allowed = limiter.can_do_action( - key="test_id", time_now_s=5, rate_hz=0.1, burst_count=1 - ) + allowed, time_allowed = limiter.can_do_action(key="test_id", _time_now_s=5) self.assertFalse(allowed) self.assertEquals(10.0, time_allowed) - allowed, time_allowed = limiter.can_do_action( - key="test_id", time_now_s=10, rate_hz=0.1, burst_count=1 - ) + allowed, time_allowed = limiter.can_do_action(key="test_id", _time_now_s=10) self.assertTrue(allowed) self.assertEquals(20.0, time_allowed) - def test_pruning(self): - limiter = Ratelimiter() + def test_allowed_via_ratelimit(self): + limiter = Ratelimiter(clock=None, rate_hz=0.1, burst_count=1) + + # Shouldn't raise + limiter.ratelimit(key="test_id", _time_now_s=0) + + # Should raise + with self.assertRaises(LimitExceededError) as context: + limiter.ratelimit(key="test_id", _time_now_s=5) + self.assertEqual(context.exception.retry_after_ms, 5000) + + # Shouldn't raise + limiter.ratelimit(key="test_id", _time_now_s=10) + + def test_allowed_via_can_do_action_and_overriding_parameters(self): + """Test that we can override options of can_do_action that would otherwise fail + an action + """ + # Create a Ratelimiter with a very low allowed rate_hz and burst_count + limiter = Ratelimiter(clock=None, rate_hz=0.1, burst_count=1) + + # First attempt should be allowed + allowed, time_allowed = limiter.can_do_action(("test_id",), _time_now_s=0,) + self.assertTrue(allowed) + self.assertEqual(10.0, time_allowed) + + # Second attempt, 1s later, will fail + allowed, time_allowed = limiter.can_do_action(("test_id",), _time_now_s=1,) + self.assertFalse(allowed) + self.assertEqual(10.0, time_allowed) + + # But, if we allow 10 actions/sec for this request, we should be allowed + # to continue. allowed, time_allowed = limiter.can_do_action( - key="test_id_1", time_now_s=0, rate_hz=0.1, burst_count=1 + ("test_id",), _time_now_s=1, rate_hz=10.0 ) + self.assertTrue(allowed) + self.assertEqual(1.1, time_allowed) - self.assertIn("test_id_1", limiter.message_counts) - + # Similarly if we allow a burst of 10 actions allowed, time_allowed = limiter.can_do_action( - key="test_id_2", time_now_s=10, rate_hz=0.1, burst_count=1 + ("test_id",), _time_now_s=1, burst_count=10 ) + self.assertTrue(allowed) + self.assertEqual(1.0, time_allowed) + + def test_allowed_via_ratelimit_and_overriding_parameters(self): + """Test that we can override options of the ratelimit method that would otherwise + fail an action + """ + # Create a Ratelimiter with a very low allowed rate_hz and burst_count + limiter = Ratelimiter(clock=None, rate_hz=0.1, burst_count=1) + + # First attempt should be allowed + limiter.ratelimit(key=("test_id",), _time_now_s=0) + + # Second attempt, 1s later, will fail + with self.assertRaises(LimitExceededError) as context: + limiter.ratelimit(key=("test_id",), _time_now_s=1) + self.assertEqual(context.exception.retry_after_ms, 9000) + + # But, if we allow 10 actions/sec for this request, we should be allowed + # to continue. + limiter.ratelimit(key=("test_id",), _time_now_s=1, rate_hz=10.0) + + # Similarly if we allow a burst of 10 actions + limiter.ratelimit(key=("test_id",), _time_now_s=1, burst_count=10) + + def test_pruning(self): + limiter = Ratelimiter(clock=None, rate_hz=0.1, burst_count=1) + limiter.can_do_action(key="test_id_1", _time_now_s=0) + + self.assertIn("test_id_1", limiter.actions) + + limiter.can_do_action(key="test_id_2", _time_now_s=10) - self.assertNotIn("test_id_1", limiter.message_counts) + self.assertNotIn("test_id_1", limiter.actions) diff --git a/tests/handlers/test_profile.py b/tests/handlers/test_profile.py index 8aa56f1496..29dd7d9c6e 100644 --- a/tests/handlers/test_profile.py +++ b/tests/handlers/test_profile.py @@ -14,7 +14,7 @@ # limitations under the License. -from mock import Mock, NonCallableMock +from mock import Mock from twisted.internet import defer @@ -55,12 +55,8 @@ class ProfileTestCase(unittest.TestCase): federation_client=self.mock_federation, federation_server=Mock(), federation_registry=self.mock_registry, - ratelimiter=NonCallableMock(spec_set=["can_do_action"]), ) - self.ratelimiter = hs.get_ratelimiter() - self.ratelimiter.can_do_action.return_value = (True, 0) - self.store = hs.get_datastore() self.frank = UserID.from_string("@1234ABCD:test") diff --git a/tests/replication/slave/storage/_base.py b/tests/replication/slave/storage/_base.py index 32cb04645f..56497b8476 100644 --- a/tests/replication/slave/storage/_base.py +++ b/tests/replication/slave/storage/_base.py @@ -13,7 +13,7 @@ # See the License for the specific language governing permissions and # limitations under the License. -from mock import Mock, NonCallableMock +from mock import Mock from tests.replication._base import BaseStreamTestCase @@ -21,12 +21,7 @@ from tests.replication._base import BaseStreamTestCase class BaseSlavedStoreTestCase(BaseStreamTestCase): def make_homeserver(self, reactor, clock): - hs = self.setup_test_homeserver( - federation_client=Mock(), - ratelimiter=NonCallableMock(spec_set=["can_do_action"]), - ) - - hs.get_ratelimiter().can_do_action.return_value = (True, 0) + hs = self.setup_test_homeserver(federation_client=Mock()) return hs diff --git a/tests/rest/client/v1/test_events.py b/tests/rest/client/v1/test_events.py index b54b06482b..f75520877f 100644 --- a/tests/rest/client/v1/test_events.py +++ b/tests/rest/client/v1/test_events.py @@ -15,7 +15,7 @@ """ Tests REST events for /events paths.""" -from mock import Mock, NonCallableMock +from mock import Mock import synapse.rest.admin from synapse.rest.client.v1 import events, login, room @@ -40,11 +40,7 @@ class EventStreamPermissionsTestCase(unittest.HomeserverTestCase): config["enable_registration"] = True config["auto_join_rooms"] = [] - hs = self.setup_test_homeserver( - config=config, ratelimiter=NonCallableMock(spec_set=["can_do_action"]) - ) - self.ratelimiter = hs.get_ratelimiter() - self.ratelimiter.can_do_action.return_value = (True, 0) + hs = self.setup_test_homeserver(config=config) hs.get_handlers().federation_handler = Mock() diff --git a/tests/rest/client/v1/test_login.py b/tests/rest/client/v1/test_login.py index 0f0f7ca72d..9033f09fd2 100644 --- a/tests/rest/client/v1/test_login.py +++ b/tests/rest/client/v1/test_login.py @@ -29,7 +29,6 @@ class LoginRestServletTestCase(unittest.HomeserverTestCase): ] def make_homeserver(self, reactor, clock): - self.hs = self.setup_test_homeserver() self.hs.config.enable_registration = True self.hs.config.registrations_require_3pid = [] @@ -38,10 +37,20 @@ class LoginRestServletTestCase(unittest.HomeserverTestCase): return self.hs + @override_config( + { + "rc_login": { + "address": {"per_second": 0.17, "burst_count": 5}, + # Prevent the account login ratelimiter from raising first + # + # This is normally covered by the default test homeserver config + # which sets these values to 10000, but as we're overriding the entire + # rc_login dict here, we need to set this manually as well + "account": {"per_second": 10000, "burst_count": 10000}, + } + } + ) def test_POST_ratelimiting_per_address(self): - self.hs.config.rc_login_address.burst_count = 5 - self.hs.config.rc_login_address.per_second = 0.17 - # Create different users so we're sure not to be bothered by the per-user # ratelimiter. for i in range(0, 6): @@ -80,10 +89,20 @@ class LoginRestServletTestCase(unittest.HomeserverTestCase): self.assertEquals(channel.result["code"], b"200", channel.result) + @override_config( + { + "rc_login": { + "account": {"per_second": 0.17, "burst_count": 5}, + # Prevent the address login ratelimiter from raising first + # + # This is normally covered by the default test homeserver config + # which sets these values to 10000, but as we're overriding the entire + # rc_login dict here, we need to set this manually as well + "address": {"per_second": 10000, "burst_count": 10000}, + } + } + ) def test_POST_ratelimiting_per_account(self): - self.hs.config.rc_login_account.burst_count = 5 - self.hs.config.rc_login_account.per_second = 0.17 - self.register_user("kermit", "monkey") for i in range(0, 6): @@ -119,10 +138,20 @@ class LoginRestServletTestCase(unittest.HomeserverTestCase): self.assertEquals(channel.result["code"], b"200", channel.result) + @override_config( + { + "rc_login": { + # Prevent the address login ratelimiter from raising first + # + # This is normally covered by the default test homeserver config + # which sets these values to 10000, but as we're overriding the entire + # rc_login dict here, we need to set this manually as well + "address": {"per_second": 10000, "burst_count": 10000}, + "failed_attempts": {"per_second": 0.17, "burst_count": 5}, + } + } + ) def test_POST_ratelimiting_per_account_failed_attempts(self): - self.hs.config.rc_login_failed_attempts.burst_count = 5 - self.hs.config.rc_login_failed_attempts.per_second = 0.17 - self.register_user("kermit", "monkey") for i in range(0, 6): diff --git a/tests/rest/client/v1/test_rooms.py b/tests/rest/client/v1/test_rooms.py index 7dd86d0c27..4886bbb401 100644 --- a/tests/rest/client/v1/test_rooms.py +++ b/tests/rest/client/v1/test_rooms.py @@ -20,7 +20,7 @@ import json -from mock import Mock, NonCallableMock +from mock import Mock from six.moves.urllib import parse as urlparse from twisted.internet import defer @@ -46,13 +46,8 @@ class RoomBase(unittest.HomeserverTestCase): def make_homeserver(self, reactor, clock): self.hs = self.setup_test_homeserver( - "red", - http_client=None, - federation_client=Mock(), - ratelimiter=NonCallableMock(spec_set=["can_do_action"]), + "red", http_client=None, federation_client=Mock(), ) - self.ratelimiter = self.hs.get_ratelimiter() - self.ratelimiter.can_do_action.return_value = (True, 0) self.hs.get_federation_handler = Mock(return_value=Mock()) diff --git a/tests/rest/client/v1/test_typing.py b/tests/rest/client/v1/test_typing.py index 4bc3aaf02d..18260bb90e 100644 --- a/tests/rest/client/v1/test_typing.py +++ b/tests/rest/client/v1/test_typing.py @@ -16,7 +16,7 @@ """Tests REST events for /rooms paths.""" -from mock import Mock, NonCallableMock +from mock import Mock from twisted.internet import defer @@ -39,17 +39,11 @@ class RoomTypingTestCase(unittest.HomeserverTestCase): def make_homeserver(self, reactor, clock): hs = self.setup_test_homeserver( - "red", - http_client=None, - federation_client=Mock(), - ratelimiter=NonCallableMock(spec_set=["can_do_action"]), + "red", http_client=None, federation_client=Mock(), ) self.event_source = hs.get_event_sources().sources["typing"] - self.ratelimiter = hs.get_ratelimiter() - self.ratelimiter.can_do_action.return_value = (True, 0) - hs.get_handlers().federation_handler = Mock() def get_user_by_access_token(token=None, allow_guest=False): diff --git a/tests/rest/client/v2_alpha/test_register.py b/tests/rest/client/v2_alpha/test_register.py index 5637ce2090..7deaf5b24a 100644 --- a/tests/rest/client/v2_alpha/test_register.py +++ b/tests/rest/client/v2_alpha/test_register.py @@ -29,6 +29,7 @@ from synapse.rest.client.v1 import login, logout from synapse.rest.client.v2_alpha import account, account_validity, register, sync from tests import unittest +from tests.unittest import override_config class RegisterRestServletTestCase(unittest.HomeserverTestCase): @@ -146,10 +147,8 @@ class RegisterRestServletTestCase(unittest.HomeserverTestCase): self.assertEquals(channel.result["code"], b"403", channel.result) self.assertEquals(channel.json_body["error"], "Guest access is disabled") + @override_config({"rc_registration": {"per_second": 0.17, "burst_count": 5}}) def test_POST_ratelimiting_guest(self): - self.hs.config.rc_registration.burst_count = 5 - self.hs.config.rc_registration.per_second = 0.17 - for i in range(0, 6): url = self.url + b"?kind=guest" request, channel = self.make_request(b"POST", url, b"{}") @@ -168,10 +167,8 @@ class RegisterRestServletTestCase(unittest.HomeserverTestCase): self.assertEquals(channel.result["code"], b"200", channel.result) + @override_config({"rc_registration": {"per_second": 0.17, "burst_count": 5}}) def test_POST_ratelimiting(self): - self.hs.config.rc_registration.burst_count = 5 - self.hs.config.rc_registration.per_second = 0.17 - for i in range(0, 6): params = { "username": "kermit" + str(i), -- cgit 1.4.1 From e55ee7c32fb3d958fb9da4e34133f53a6c57d2ea Mon Sep 17 00:00:00 2001 From: WGH Date: Fri, 5 Jun 2020 13:54:27 +0300 Subject: Add support for webp thumbnailing (#7586) Closes #4382 Signed-off-by: Maxim Plotnikov --- changelog.d/7586.feature | 1 + synapse/config/repository.py | 1 + tests/rest/media/v1/test_media_storage.py | 135 ++++++++++++++++++++++-------- 3 files changed, 101 insertions(+), 36 deletions(-) create mode 100644 changelog.d/7586.feature (limited to 'tests') diff --git a/changelog.d/7586.feature b/changelog.d/7586.feature new file mode 100644 index 0000000000..ef0231d823 --- /dev/null +++ b/changelog.d/7586.feature @@ -0,0 +1 @@ +Add support for generating thumbnails for WebP images. Previously, users would see an empty box instead of preview image. diff --git a/synapse/config/repository.py b/synapse/config/repository.py index 9d2ce20220..b751d02d37 100644 --- a/synapse/config/repository.py +++ b/synapse/config/repository.py @@ -70,6 +70,7 @@ def parse_thumbnail_requirements(thumbnail_sizes): jpeg_thumbnail = ThumbnailRequirement(width, height, method, "image/jpeg") png_thumbnail = ThumbnailRequirement(width, height, method, "image/png") requirements.setdefault("image/jpeg", []).append(jpeg_thumbnail) + requirements.setdefault("image/webp", []).append(jpeg_thumbnail) requirements.setdefault("image/gif", []).append(png_thumbnail) requirements.setdefault("image/png", []).append(png_thumbnail) return { diff --git a/tests/rest/media/v1/test_media_storage.py b/tests/rest/media/v1/test_media_storage.py index 1809ceb839..1ca648ef2b 100644 --- a/tests/rest/media/v1/test_media_storage.py +++ b/tests/rest/media/v1/test_media_storage.py @@ -18,10 +18,16 @@ import os import shutil import tempfile from binascii import unhexlify +from io import BytesIO +from typing import Optional from mock import Mock from six.moves.urllib import parse +import attr +import PIL.Image as Image +from parameterized import parameterized_class + from twisted.internet.defer import Deferred from synapse.logging.context import make_deferred_yieldable @@ -94,6 +100,68 @@ class MediaStorageTests(unittest.HomeserverTestCase): self.assertEqual(test_body, body) +@attr.s +class _TestImage: + """An image for testing thumbnailing with the expected results + + Attributes: + data: The raw image to thumbnail + content_type: The type of the image as a content type, e.g. "image/png" + extension: The extension associated with the format, e.g. ".png" + expected_cropped: The expected bytes from cropped thumbnailing, or None if + test should just check for success. + expected_scaled: The expected bytes from scaled thumbnailing, or None if + test should just check for a valid image returned. + """ + + data = attr.ib(type=bytes) + content_type = attr.ib(type=bytes) + extension = attr.ib(type=bytes) + expected_cropped = attr.ib(type=Optional[bytes]) + expected_scaled = attr.ib(type=Optional[bytes]) + + +@parameterized_class( + ("test_image",), + [ + # smol png + ( + _TestImage( + unhexlify( + b"89504e470d0a1a0a0000000d4948445200000001000000010806" + b"0000001f15c4890000000a49444154789c63000100000500010d" + b"0a2db40000000049454e44ae426082" + ), + b"image/png", + b".png", + unhexlify( + b"89504e470d0a1a0a0000000d4948445200000020000000200806" + b"000000737a7af40000001a49444154789cedc101010000008220" + b"ffaf6e484001000000ef0610200001194334ee0000000049454e" + b"44ae426082" + ), + unhexlify( + b"89504e470d0a1a0a0000000d4948445200000001000000010806" + b"0000001f15c4890000000d49444154789c636060606000000005" + b"0001a5f645400000000049454e44ae426082" + ), + ), + ), + # small lossless webp + ( + _TestImage( + unhexlify( + b"524946461a000000574542505650384c0d0000002f0000001007" + b"1011118888fe0700" + ), + b"image/webp", + b".webp", + None, + None, + ), + ), + ], +) class MediaRepoTests(unittest.HomeserverTestCase): hijack_auth = True @@ -151,13 +219,6 @@ class MediaRepoTests(unittest.HomeserverTestCase): self.download_resource = self.media_repo.children[b"download"] self.thumbnail_resource = self.media_repo.children[b"thumbnail"] - # smol png - self.end_content = unhexlify( - b"89504e470d0a1a0a0000000d4948445200000001000000010806" - b"0000001f15c4890000000a49444154789c63000100000500010d" - b"0a2db40000000049454e44ae426082" - ) - self.media_id = "example.com/12345" def _req(self, content_disposition): @@ -176,14 +237,14 @@ class MediaRepoTests(unittest.HomeserverTestCase): self.assertEqual(self.fetches[0][3], {"allow_remote": "false"}) headers = { - b"Content-Length": [b"%d" % (len(self.end_content))], - b"Content-Type": [b"image/png"], + b"Content-Length": [b"%d" % (len(self.test_image.data))], + b"Content-Type": [self.test_image.content_type], } if content_disposition: headers[b"Content-Disposition"] = [content_disposition] self.fetches[0][0].callback( - (self.end_content, (len(self.end_content), headers)) + (self.test_image.data, (len(self.test_image.data), headers)) ) self.pump() @@ -196,12 +257,15 @@ class MediaRepoTests(unittest.HomeserverTestCase): If the filename is filename= then Synapse will decode it as an ASCII string, and use filename= in the response. """ - channel = self._req(b"inline; filename=out.png") + channel = self._req(b"inline; filename=out" + self.test_image.extension) headers = channel.headers - self.assertEqual(headers.getRawHeaders(b"Content-Type"), [b"image/png"]) self.assertEqual( - headers.getRawHeaders(b"Content-Disposition"), [b"inline; filename=out.png"] + headers.getRawHeaders(b"Content-Type"), [self.test_image.content_type] + ) + self.assertEqual( + headers.getRawHeaders(b"Content-Disposition"), + [b"inline; filename=out" + self.test_image.extension], ) def test_disposition_filenamestar_utf8escaped(self): @@ -211,13 +275,17 @@ class MediaRepoTests(unittest.HomeserverTestCase): response. """ filename = parse.quote("\u2603".encode("utf8")).encode("ascii") - channel = self._req(b"inline; filename*=utf-8''" + filename + b".png") + channel = self._req( + b"inline; filename*=utf-8''" + filename + self.test_image.extension + ) headers = channel.headers - self.assertEqual(headers.getRawHeaders(b"Content-Type"), [b"image/png"]) + self.assertEqual( + headers.getRawHeaders(b"Content-Type"), [self.test_image.content_type] + ) self.assertEqual( headers.getRawHeaders(b"Content-Disposition"), - [b"inline; filename*=utf-8''" + filename + b".png"], + [b"inline; filename*=utf-8''" + filename + self.test_image.extension], ) def test_disposition_none(self): @@ -228,27 +296,16 @@ class MediaRepoTests(unittest.HomeserverTestCase): channel = self._req(None) headers = channel.headers - self.assertEqual(headers.getRawHeaders(b"Content-Type"), [b"image/png"]) + self.assertEqual( + headers.getRawHeaders(b"Content-Type"), [self.test_image.content_type] + ) self.assertEqual(headers.getRawHeaders(b"Content-Disposition"), None) def test_thumbnail_crop(self): - expected_body = unhexlify( - b"89504e470d0a1a0a0000000d4948445200000020000000200806" - b"000000737a7af40000001a49444154789cedc101010000008220" - b"ffaf6e484001000000ef0610200001194334ee0000000049454e" - b"44ae426082" - ) - - self._test_thumbnail("crop", expected_body) + self._test_thumbnail("crop", self.test_image.expected_cropped) def test_thumbnail_scale(self): - expected_body = unhexlify( - b"89504e470d0a1a0a0000000d4948445200000001000000010806" - b"0000001f15c4890000000d49444154789c636060606000000005" - b"0001a5f645400000000049454e44ae426082" - ) - - self._test_thumbnail("scale", expected_body) + self._test_thumbnail("scale", self.test_image.expected_scaled) def _test_thumbnail(self, method, expected_body): params = "?width=32&height=32&method=" + method @@ -259,13 +316,19 @@ class MediaRepoTests(unittest.HomeserverTestCase): self.pump() headers = { - b"Content-Length": [b"%d" % (len(self.end_content))], - b"Content-Type": [b"image/png"], + b"Content-Length": [b"%d" % (len(self.test_image.data))], + b"Content-Type": [self.test_image.content_type], } self.fetches[0][0].callback( - (self.end_content, (len(self.end_content), headers)) + (self.test_image.data, (len(self.test_image.data), headers)) ) self.pump() self.assertEqual(channel.code, 200) - self.assertEqual(channel.result["body"], expected_body, channel.result["body"]) + if expected_body is not None: + self.assertEqual( + channel.result["body"], expected_body, channel.result["body"] + ) + else: + # ensure that the result is at least some valid image + Image.open(BytesIO(channel.result["body"])) -- cgit 1.4.1 From 2970ce83674a4d910ebc46b505c9dcb83a15a1b9 Mon Sep 17 00:00:00 2001 From: Dirk Klimpel <5740567+dklimpel@users.noreply.github.com> Date: Fri, 5 Jun 2020 14:07:22 +0200 Subject: Add device management to admin API (#7481) - Admin is able to - change displaynames - delete devices - list devices - get device informations Fixes #7330 --- changelog.d/7481.feature | 1 + docs/admin_api/user_admin_api.rst | 209 +++++++++++++++ synapse/rest/admin/__init__.py | 8 + synapse/rest/admin/devices.py | 161 ++++++++++++ tests/rest/admin/test_device.py | 541 ++++++++++++++++++++++++++++++++++++++ 5 files changed, 920 insertions(+) create mode 100644 changelog.d/7481.feature create mode 100644 synapse/rest/admin/devices.py create mode 100644 tests/rest/admin/test_device.py (limited to 'tests') diff --git a/changelog.d/7481.feature b/changelog.d/7481.feature new file mode 100644 index 0000000000..f167f3632c --- /dev/null +++ b/changelog.d/7481.feature @@ -0,0 +1 @@ +Add admin APIs to allow server admins to manage users' devices. Contributed by @dklimpel. diff --git a/docs/admin_api/user_admin_api.rst b/docs/admin_api/user_admin_api.rst index 776e71ec04..a3d52b282b 100644 --- a/docs/admin_api/user_admin_api.rst +++ b/docs/admin_api/user_admin_api.rst @@ -1,3 +1,5 @@ +.. contents:: + Create or modify Account ======================== @@ -245,3 +247,210 @@ with a body of: } including an ``access_token`` of a server admin. + + +User devices +============ + +List all devices +---------------- +Gets information about all devices for a specific ``user_id``. + +**Usage** + +A standard request to query the devices of an user: + +:: + + GET /_synapse/admin/v2/users//devices + + {} + +Response: + +.. code:: json + + { + "devices": [ + { + "device_id": "QBUAZIFURK", + "display_name": "android", + "last_seen_ip": "1.2.3.4", + "last_seen_ts": 1474491775024, + "user_id": "" + }, + { + "device_id": "AUIECTSRND", + "display_name": "ios", + "last_seen_ip": "1.2.3.5", + "last_seen_ts": 1474491775025, + "user_id": "" + } + ] + } + +**Parameters** + +The following query parameters are available: + +- ``user_id`` - fully qualified: for example, ``@user:server.com``. + +The following fields are possible in the JSON response body: + +- ``devices`` - An array of objects, each containing information about a device. + Device objects contain the following fields: + + - ``device_id`` - Identifier of device. + - ``display_name`` - Display name set by the user for this device. + Absent if no name has been set. + - ``last_seen_ip`` - The IP address where this device was last seen. + (May be a few minutes out of date, for efficiency reasons). + - ``last_seen_ts`` - The timestamp (in milliseconds since the unix epoch) when this + devices was last seen. (May be a few minutes out of date, for efficiency reasons). + - ``user_id`` - Owner of device. + +Delete multiple devices +------------------ +Deletes the given devices for a specific ``user_id``, and invalidates +any access token associated with them. + +**Usage** + +A standard request to delete devices: + +:: + + POST /_synapse/admin/v2/users//delete_devices + + { + "devices": [ + "QBUAZIFURK", + "AUIECTSRND" + ], + } + + +Response: + +.. code:: json + + {} + +**Parameters** + +The following query parameters are available: + +- ``user_id`` - fully qualified: for example, ``@user:server.com``. + +The following fields are required in the JSON request body: + +- ``devices`` - The list of device IDs to delete. + +Show a device +--------------- +Gets information on a single device, by ``device_id`` for a specific ``user_id``. + +**Usage** + +A standard request to get a device: + +:: + + GET /_synapse/admin/v2/users//devices/ + + {} + + +Response: + +.. code:: json + + { + "device_id": "", + "display_name": "android", + "last_seen_ip": "1.2.3.4", + "last_seen_ts": 1474491775024, + "user_id": "" + } + +**Parameters** + +The following query parameters are available: + +- ``user_id`` - fully qualified: for example, ``@user:server.com``. +- ``device_id`` - The device to retrieve. + +The following fields are possible in the JSON response body: + +- ``device_id`` - Identifier of device. +- ``display_name`` - Display name set by the user for this device. + Absent if no name has been set. +- ``last_seen_ip`` - The IP address where this device was last seen. + (May be a few minutes out of date, for efficiency reasons). +- ``last_seen_ts`` - The timestamp (in milliseconds since the unix epoch) when this + devices was last seen. (May be a few minutes out of date, for efficiency reasons). +- ``user_id`` - Owner of device. + +Update a device +--------------- +Updates the metadata on the given ``device_id`` for a specific ``user_id``. + +**Usage** + +A standard request to update a device: + +:: + + PUT /_synapse/admin/v2/users//devices/ + + { + "display_name": "My other phone" + } + + +Response: + +.. code:: json + + {} + +**Parameters** + +The following query parameters are available: + +- ``user_id`` - fully qualified: for example, ``@user:server.com``. +- ``device_id`` - The device to update. + +The following fields are required in the JSON request body: + +- ``display_name`` - The new display name for this device. If not given, + the display name is unchanged. + +Delete a device +--------------- +Deletes the given ``device_id`` for a specific ``user_id``, +and invalidates any access token associated with it. + +**Usage** + +A standard request for delete a device: + +:: + + DELETE /_synapse/admin/v2/users//devices/ + + {} + + +Response: + +.. code:: json + + {} + +**Parameters** + +The following query parameters are available: + +- ``user_id`` - fully qualified: for example, ``@user:server.com``. +- ``device_id`` - The device to delete. diff --git a/synapse/rest/admin/__init__.py b/synapse/rest/admin/__init__.py index 6b85148a32..9eda592de9 100644 --- a/synapse/rest/admin/__init__.py +++ b/synapse/rest/admin/__init__.py @@ -26,6 +26,11 @@ from synapse.rest.admin._base import ( assert_requester_is_admin, historical_admin_path_patterns, ) +from synapse.rest.admin.devices import ( + DeleteDevicesRestServlet, + DeviceRestServlet, + DevicesRestServlet, +) from synapse.rest.admin.groups import DeleteGroupAdminRestServlet from synapse.rest.admin.media import ListMediaInRoom, register_servlets_for_media_repo from synapse.rest.admin.purge_room_servlet import PurgeRoomServlet @@ -202,6 +207,9 @@ def register_servlets(hs, http_server): UserAdminServlet(hs).register(http_server) UserRestServletV2(hs).register(http_server) UsersRestServletV2(hs).register(http_server) + DeviceRestServlet(hs).register(http_server) + DevicesRestServlet(hs).register(http_server) + DeleteDevicesRestServlet(hs).register(http_server) def register_servlets_for_client_rest_resource(hs, http_server): diff --git a/synapse/rest/admin/devices.py b/synapse/rest/admin/devices.py new file mode 100644 index 0000000000..8d32677339 --- /dev/null +++ b/synapse/rest/admin/devices.py @@ -0,0 +1,161 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 Dirk Klimpel +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import logging +import re + +from synapse.api.errors import NotFoundError, SynapseError +from synapse.http.servlet import ( + RestServlet, + assert_params_in_dict, + parse_json_object_from_request, +) +from synapse.rest.admin._base import assert_requester_is_admin +from synapse.types import UserID + +logger = logging.getLogger(__name__) + + +class DeviceRestServlet(RestServlet): + """ + Get, update or delete the given user's device + """ + + PATTERNS = ( + re.compile( + "^/_synapse/admin/v2/users/(?P[^/]*)/devices/(?P[^/]*)$" + ), + ) + + def __init__(self, hs): + super(DeviceRestServlet, self).__init__() + self.hs = hs + self.auth = hs.get_auth() + self.device_handler = hs.get_device_handler() + self.store = hs.get_datastore() + + async def on_GET(self, request, user_id, device_id): + await assert_requester_is_admin(self.auth, request) + + target_user = UserID.from_string(user_id) + if not self.hs.is_mine(target_user): + raise SynapseError(400, "Can only lookup local users") + + u = await self.store.get_user_by_id(target_user.to_string()) + if u is None: + raise NotFoundError("Unknown user") + + device = await self.device_handler.get_device( + target_user.to_string(), device_id + ) + return 200, device + + async def on_DELETE(self, request, user_id, device_id): + await assert_requester_is_admin(self.auth, request) + + target_user = UserID.from_string(user_id) + if not self.hs.is_mine(target_user): + raise SynapseError(400, "Can only lookup local users") + + u = await self.store.get_user_by_id(target_user.to_string()) + if u is None: + raise NotFoundError("Unknown user") + + await self.device_handler.delete_device(target_user.to_string(), device_id) + return 200, {} + + async def on_PUT(self, request, user_id, device_id): + await assert_requester_is_admin(self.auth, request) + + target_user = UserID.from_string(user_id) + if not self.hs.is_mine(target_user): + raise SynapseError(400, "Can only lookup local users") + + u = await self.store.get_user_by_id(target_user.to_string()) + if u is None: + raise NotFoundError("Unknown user") + + body = parse_json_object_from_request(request, allow_empty_body=True) + await self.device_handler.update_device( + target_user.to_string(), device_id, body + ) + return 200, {} + + +class DevicesRestServlet(RestServlet): + """ + Retrieve the given user's devices + """ + + PATTERNS = (re.compile("^/_synapse/admin/v2/users/(?P[^/]*)/devices$"),) + + def __init__(self, hs): + """ + Args: + hs (synapse.server.HomeServer): server + """ + self.hs = hs + self.auth = hs.get_auth() + self.device_handler = hs.get_device_handler() + self.store = hs.get_datastore() + + async def on_GET(self, request, user_id): + await assert_requester_is_admin(self.auth, request) + + target_user = UserID.from_string(user_id) + if not self.hs.is_mine(target_user): + raise SynapseError(400, "Can only lookup local users") + + u = await self.store.get_user_by_id(target_user.to_string()) + if u is None: + raise NotFoundError("Unknown user") + + devices = await self.device_handler.get_devices_by_user(target_user.to_string()) + return 200, {"devices": devices} + + +class DeleteDevicesRestServlet(RestServlet): + """ + API for bulk deletion of devices. Accepts a JSON object with a devices + key which lists the device_ids to delete. + """ + + PATTERNS = ( + re.compile("^/_synapse/admin/v2/users/(?P[^/]*)/delete_devices$"), + ) + + def __init__(self, hs): + self.hs = hs + self.auth = hs.get_auth() + self.device_handler = hs.get_device_handler() + self.store = hs.get_datastore() + + async def on_POST(self, request, user_id): + await assert_requester_is_admin(self.auth, request) + + target_user = UserID.from_string(user_id) + if not self.hs.is_mine(target_user): + raise SynapseError(400, "Can only lookup local users") + + u = await self.store.get_user_by_id(target_user.to_string()) + if u is None: + raise NotFoundError("Unknown user") + + body = parse_json_object_from_request(request, allow_empty_body=False) + assert_params_in_dict(body, ["devices"]) + + await self.device_handler.delete_devices( + target_user.to_string(), body["devices"] + ) + return 200, {} diff --git a/tests/rest/admin/test_device.py b/tests/rest/admin/test_device.py new file mode 100644 index 0000000000..faa7f381a9 --- /dev/null +++ b/tests/rest/admin/test_device.py @@ -0,0 +1,541 @@ +# -*- coding: utf-8 -*- +# Copyright 2020 Dirk Klimpel +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import json +import urllib.parse + +import synapse.rest.admin +from synapse.api.errors import Codes +from synapse.rest.client.v1 import login + +from tests import unittest + + +class DeviceRestTestCase(unittest.HomeserverTestCase): + + servlets = [ + synapse.rest.admin.register_servlets, + login.register_servlets, + ] + + def prepare(self, reactor, clock, hs): + self.handler = hs.get_device_handler() + + self.admin_user = self.register_user("admin", "pass", admin=True) + self.admin_user_tok = self.login("admin", "pass") + + self.other_user = self.register_user("user", "pass") + self.other_user_token = self.login("user", "pass") + res = self.get_success(self.handler.get_devices_by_user(self.other_user)) + self.other_user_device_id = res[0]["device_id"] + + self.url = "/_synapse/admin/v2/users/%s/devices/%s" % ( + urllib.parse.quote(self.other_user), + self.other_user_device_id, + ) + + def test_no_auth(self): + """ + Try to get a device of an user without authentication. + """ + request, channel = self.make_request("GET", self.url, b"{}") + self.render(request) + + self.assertEqual(401, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual(Codes.MISSING_TOKEN, channel.json_body["errcode"]) + + request, channel = self.make_request("PUT", self.url, b"{}") + self.render(request) + + self.assertEqual(401, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual(Codes.MISSING_TOKEN, channel.json_body["errcode"]) + + request, channel = self.make_request("DELETE", self.url, b"{}") + self.render(request) + + self.assertEqual(401, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual(Codes.MISSING_TOKEN, channel.json_body["errcode"]) + + def test_requester_is_no_admin(self): + """ + If the user is not a server admin, an error is returned. + """ + request, channel = self.make_request( + "GET", self.url, access_token=self.other_user_token, + ) + self.render(request) + + self.assertEqual(403, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual(Codes.FORBIDDEN, channel.json_body["errcode"]) + + request, channel = self.make_request( + "PUT", self.url, access_token=self.other_user_token, + ) + self.render(request) + + self.assertEqual(403, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual(Codes.FORBIDDEN, channel.json_body["errcode"]) + + request, channel = self.make_request( + "DELETE", self.url, access_token=self.other_user_token, + ) + self.render(request) + + self.assertEqual(403, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual(Codes.FORBIDDEN, channel.json_body["errcode"]) + + def test_user_does_not_exist(self): + """ + Tests that a lookup for a user that does not exist returns a 404 + """ + url = ( + "/_synapse/admin/v2/users/@unknown_person:test/devices/%s" + % self.other_user_device_id + ) + + request, channel = self.make_request( + "GET", url, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(404, channel.code, msg=channel.json_body) + self.assertEqual(Codes.NOT_FOUND, channel.json_body["errcode"]) + + request, channel = self.make_request( + "PUT", url, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(404, channel.code, msg=channel.json_body) + self.assertEqual(Codes.NOT_FOUND, channel.json_body["errcode"]) + + request, channel = self.make_request( + "DELETE", url, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(404, channel.code, msg=channel.json_body) + self.assertEqual(Codes.NOT_FOUND, channel.json_body["errcode"]) + + def test_user_is_not_local(self): + """ + Tests that a lookup for a user that is not a local returns a 400 + """ + url = ( + "/_synapse/admin/v2/users/@unknown_person:unknown_domain/devices/%s" + % self.other_user_device_id + ) + + request, channel = self.make_request( + "GET", url, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(400, channel.code, msg=channel.json_body) + self.assertEqual("Can only lookup local users", channel.json_body["error"]) + + request, channel = self.make_request( + "PUT", url, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(400, channel.code, msg=channel.json_body) + self.assertEqual("Can only lookup local users", channel.json_body["error"]) + + request, channel = self.make_request( + "DELETE", url, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(400, channel.code, msg=channel.json_body) + self.assertEqual("Can only lookup local users", channel.json_body["error"]) + + def test_unknown_device(self): + """ + Tests that a lookup for a device that does not exist returns either 404 or 200. + """ + url = "/_synapse/admin/v2/users/%s/devices/unknown_device" % urllib.parse.quote( + self.other_user + ) + + request, channel = self.make_request( + "GET", url, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(404, channel.code, msg=channel.json_body) + self.assertEqual(Codes.NOT_FOUND, channel.json_body["errcode"]) + + request, channel = self.make_request( + "PUT", url, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(200, channel.code, msg=channel.json_body) + + request, channel = self.make_request( + "DELETE", url, access_token=self.admin_user_tok, + ) + self.render(request) + + # Delete unknown device returns status 200 + self.assertEqual(200, channel.code, msg=channel.json_body) + + def test_update_device_too_long_display_name(self): + """ + Update a device with a display name that is invalid (too long). + """ + # Set iniital display name. + update = {"display_name": "new display"} + self.get_success( + self.handler.update_device( + self.other_user, self.other_user_device_id, update + ) + ) + + # Request to update a device display name with a new value that is longer than allowed. + update = { + "display_name": "a" + * (synapse.handlers.device.MAX_DEVICE_DISPLAY_NAME_LEN + 1) + } + + body = json.dumps(update) + request, channel = self.make_request( + "PUT", + self.url, + access_token=self.admin_user_tok, + content=body.encode(encoding="utf_8"), + ) + self.render(request) + + self.assertEqual(400, channel.code, msg=channel.json_body) + self.assertEqual(Codes.UNKNOWN, channel.json_body["errcode"]) + + # Ensure the display name was not updated. + request, channel = self.make_request( + "GET", self.url, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(200, channel.code, msg=channel.json_body) + self.assertEqual("new display", channel.json_body["display_name"]) + + def test_update_no_display_name(self): + """ + Tests that a update for a device without JSON returns a 200 + """ + # Set iniital display name. + update = {"display_name": "new display"} + self.get_success( + self.handler.update_device( + self.other_user, self.other_user_device_id, update + ) + ) + + request, channel = self.make_request( + "PUT", self.url, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(200, channel.code, msg=channel.json_body) + + # Ensure the display name was not updated. + request, channel = self.make_request( + "GET", self.url, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(200, channel.code, msg=channel.json_body) + self.assertEqual("new display", channel.json_body["display_name"]) + + def test_update_display_name(self): + """ + Tests a normal successful update of display name + """ + # Set new display_name + body = json.dumps({"display_name": "new displayname"}) + request, channel = self.make_request( + "PUT", + self.url, + access_token=self.admin_user_tok, + content=body.encode(encoding="utf_8"), + ) + self.render(request) + + self.assertEqual(200, channel.code, msg=channel.json_body) + + # Check new display_name + request, channel = self.make_request( + "GET", self.url, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(200, channel.code, msg=channel.json_body) + self.assertEqual("new displayname", channel.json_body["display_name"]) + + def test_get_device(self): + """ + Tests that a normal lookup for a device is successfully + """ + request, channel = self.make_request( + "GET", self.url, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(200, channel.code, msg=channel.json_body) + self.assertEqual(self.other_user, channel.json_body["user_id"]) + # Check that all fields are available + self.assertIn("user_id", channel.json_body) + self.assertIn("device_id", channel.json_body) + self.assertIn("display_name", channel.json_body) + self.assertIn("last_seen_ip", channel.json_body) + self.assertIn("last_seen_ts", channel.json_body) + + def test_delete_device(self): + """ + Tests that a remove of a device is successfully + """ + # Count number of devies of an user. + res = self.get_success(self.handler.get_devices_by_user(self.other_user)) + number_devices = len(res) + self.assertEqual(1, number_devices) + + # Delete device + request, channel = self.make_request( + "DELETE", self.url, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(200, channel.code, msg=channel.json_body) + + # Ensure that the number of devices is decreased + res = self.get_success(self.handler.get_devices_by_user(self.other_user)) + self.assertEqual(number_devices - 1, len(res)) + + +class DevicesRestTestCase(unittest.HomeserverTestCase): + + servlets = [ + synapse.rest.admin.register_servlets, + login.register_servlets, + ] + + def prepare(self, reactor, clock, hs): + self.admin_user = self.register_user("admin", "pass", admin=True) + self.admin_user_tok = self.login("admin", "pass") + + self.other_user = self.register_user("user", "pass") + + self.url = "/_synapse/admin/v2/users/%s/devices" % urllib.parse.quote( + self.other_user + ) + + def test_no_auth(self): + """ + Try to list devices of an user without authentication. + """ + request, channel = self.make_request("GET", self.url, b"{}") + self.render(request) + + self.assertEqual(401, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual(Codes.MISSING_TOKEN, channel.json_body["errcode"]) + + def test_requester_is_no_admin(self): + """ + If the user is not a server admin, an error is returned. + """ + other_user_token = self.login("user", "pass") + + request, channel = self.make_request( + "GET", self.url, access_token=other_user_token, + ) + self.render(request) + + self.assertEqual(403, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual(Codes.FORBIDDEN, channel.json_body["errcode"]) + + def test_user_does_not_exist(self): + """ + Tests that a lookup for a user that does not exist returns a 404 + """ + url = "/_synapse/admin/v2/users/@unknown_person:test/devices" + request, channel = self.make_request( + "GET", url, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(404, channel.code, msg=channel.json_body) + self.assertEqual(Codes.NOT_FOUND, channel.json_body["errcode"]) + + def test_user_is_not_local(self): + """ + Tests that a lookup for a user that is not a local returns a 400 + """ + url = "/_synapse/admin/v2/users/@unknown_person:unknown_domain/devices" + + request, channel = self.make_request( + "GET", url, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(400, channel.code, msg=channel.json_body) + self.assertEqual("Can only lookup local users", channel.json_body["error"]) + + def test_get_devices(self): + """ + Tests that a normal lookup for devices is successfully + """ + # Create devices + number_devices = 5 + for n in range(number_devices): + self.login("user", "pass") + + # Get devices + request, channel = self.make_request( + "GET", self.url, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(200, channel.code, msg=channel.json_body) + self.assertEqual(number_devices, len(channel.json_body["devices"])) + self.assertEqual(self.other_user, channel.json_body["devices"][0]["user_id"]) + # Check that all fields are available + for d in channel.json_body["devices"]: + self.assertIn("user_id", d) + self.assertIn("device_id", d) + self.assertIn("display_name", d) + self.assertIn("last_seen_ip", d) + self.assertIn("last_seen_ts", d) + + +class DeleteDevicesRestTestCase(unittest.HomeserverTestCase): + + servlets = [ + synapse.rest.admin.register_servlets, + login.register_servlets, + ] + + def prepare(self, reactor, clock, hs): + self.handler = hs.get_device_handler() + + self.admin_user = self.register_user("admin", "pass", admin=True) + self.admin_user_tok = self.login("admin", "pass") + + self.other_user = self.register_user("user", "pass") + + self.url = "/_synapse/admin/v2/users/%s/delete_devices" % urllib.parse.quote( + self.other_user + ) + + def test_no_auth(self): + """ + Try to delete devices of an user without authentication. + """ + request, channel = self.make_request("POST", self.url, b"{}") + self.render(request) + + self.assertEqual(401, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual(Codes.MISSING_TOKEN, channel.json_body["errcode"]) + + def test_requester_is_no_admin(self): + """ + If the user is not a server admin, an error is returned. + """ + other_user_token = self.login("user", "pass") + + request, channel = self.make_request( + "POST", self.url, access_token=other_user_token, + ) + self.render(request) + + self.assertEqual(403, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual(Codes.FORBIDDEN, channel.json_body["errcode"]) + + def test_user_does_not_exist(self): + """ + Tests that a lookup for a user that does not exist returns a 404 + """ + url = "/_synapse/admin/v2/users/@unknown_person:test/delete_devices" + request, channel = self.make_request( + "POST", url, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(404, channel.code, msg=channel.json_body) + self.assertEqual(Codes.NOT_FOUND, channel.json_body["errcode"]) + + def test_user_is_not_local(self): + """ + Tests that a lookup for a user that is not a local returns a 400 + """ + url = "/_synapse/admin/v2/users/@unknown_person:unknown_domain/delete_devices" + + request, channel = self.make_request( + "POST", url, access_token=self.admin_user_tok, + ) + self.render(request) + + self.assertEqual(400, channel.code, msg=channel.json_body) + self.assertEqual("Can only lookup local users", channel.json_body["error"]) + + def test_unknown_devices(self): + """ + Tests that a remove of a device that does not exist returns 200. + """ + body = json.dumps({"devices": ["unknown_device1", "unknown_device2"]}) + request, channel = self.make_request( + "POST", + self.url, + access_token=self.admin_user_tok, + content=body.encode(encoding="utf_8"), + ) + self.render(request) + + # Delete unknown devices returns status 200 + self.assertEqual(200, channel.code, msg=channel.json_body) + + def test_delete_devices(self): + """ + Tests that a remove of devices is successfully + """ + + # Create devices + number_devices = 5 + for n in range(number_devices): + self.login("user", "pass") + + # Get devices + res = self.get_success(self.handler.get_devices_by_user(self.other_user)) + self.assertEqual(number_devices, len(res)) + + # Create list of device IDs + device_ids = [] + for d in res: + device_ids.append(str(d["device_id"])) + + # Delete devices + body = json.dumps({"devices": device_ids}) + request, channel = self.make_request( + "POST", + self.url, + access_token=self.admin_user_tok, + content=body.encode(encoding="utf_8"), + ) + self.render(request) + + self.assertEqual(200, channel.code, msg=channel.json_body) + + res = self.get_success(self.handler.get_devices_by_user(self.other_user)) + self.assertEqual(0, len(res)) -- cgit 1.4.1 From 908f9e2d24617a62f5e2fe52aa68941c64b0fde3 Mon Sep 17 00:00:00 2001 From: Dirk Klimpel <5740567+dklimpel@users.noreply.github.com> Date: Fri, 5 Jun 2020 14:08:49 +0200 Subject: Allow new users to be registered via the admin API even if the monthly active user limit has been reached (#7263) --- changelog.d/7263.bugfix | 1 + synapse/handlers/register.py | 7 +- synapse/rest/admin/users.py | 2 + tests/rest/admin/test_user.py | 178 ++++++++++++++++++++++++++++++++++++++---- 4 files changed, 172 insertions(+), 16 deletions(-) create mode 100644 changelog.d/7263.bugfix (limited to 'tests') diff --git a/changelog.d/7263.bugfix b/changelog.d/7263.bugfix new file mode 100644 index 0000000000..0b4739261c --- /dev/null +++ b/changelog.d/7263.bugfix @@ -0,0 +1 @@ +Allow new users to be registered via the admin API even if the monthly active user limit has been reached. Contributed by @dkimpel. diff --git a/synapse/handlers/register.py b/synapse/handlers/register.py index cd746be7c8..ffda09226c 100644 --- a/synapse/handlers/register.py +++ b/synapse/handlers/register.py @@ -150,6 +150,7 @@ class RegistrationHandler(BaseHandler): default_display_name=None, address=None, bind_emails=[], + by_admin=False, ): """Registers a new client on the server. @@ -165,6 +166,8 @@ class RegistrationHandler(BaseHandler): will be set to this. Defaults to 'localpart'. address (str|None): the IP address used to perform the registration. bind_emails (List[str]): list of emails to bind to this account. + by_admin (bool): True if this registration is being made via the + admin api, otherwise False. Returns: Deferred[str]: user_id Raises: @@ -172,7 +175,9 @@ class RegistrationHandler(BaseHandler): """ yield self.check_registration_ratelimit(address) - yield self.auth.check_auth_blocking(threepid=threepid) + # do not check_auth_blocking if the call is coming through the Admin API + if not by_admin: + yield self.auth.check_auth_blocking(threepid=threepid) if localpart is not None: yield self.check_username(localpart, guest_access_token=guest_access_token) diff --git a/synapse/rest/admin/users.py b/synapse/rest/admin/users.py index 82251dbe5f..fefc8f71fa 100644 --- a/synapse/rest/admin/users.py +++ b/synapse/rest/admin/users.py @@ -270,6 +270,7 @@ class UserRestServletV2(RestServlet): admin=bool(admin), default_display_name=displayname, user_type=user_type, + by_admin=True, ) if "threepids" in body: @@ -432,6 +433,7 @@ class UserRegisterServlet(RestServlet): password_hash=password_hash, admin=bool(admin), user_type=user_type, + by_admin=True, ) result = await register._create_registration_details(user_id, body) diff --git a/tests/rest/admin/test_user.py b/tests/rest/admin/test_user.py index e29cc24a8a..cca5f548e6 100644 --- a/tests/rest/admin/test_user.py +++ b/tests/rest/admin/test_user.py @@ -22,9 +22,12 @@ from mock import Mock import synapse.rest.admin from synapse.api.constants import UserTypes +from synapse.api.errors import HttpResponseException, ResourceLimitError from synapse.rest.client.v1 import login +from synapse.rest.client.v2_alpha import sync from tests import unittest +from tests.unittest import override_config class UserRegisterTestCase(unittest.HomeserverTestCase): @@ -320,6 +323,52 @@ class UserRegisterTestCase(unittest.HomeserverTestCase): self.assertEqual(400, int(channel.result["code"]), msg=channel.result["body"]) self.assertEqual("Invalid user type", channel.json_body["error"]) + @override_config( + {"limit_usage_by_mau": True, "max_mau_value": 2, "mau_trial_days": 0} + ) + def test_register_mau_limit_reached(self): + """ + Check we can register a user via the shared secret registration API + even if the MAU limit is reached. + """ + handler = self.hs.get_registration_handler() + store = self.hs.get_datastore() + + # Set monthly active users to the limit + store.get_monthly_active_count = Mock(return_value=self.hs.config.max_mau_value) + # Check that the blocking of monthly active users is working as expected + # The registration of a new user fails due to the limit + self.get_failure( + handler.register_user(localpart="local_part"), ResourceLimitError + ) + + # Register new user with admin API + request, channel = self.make_request("GET", self.url) + self.render(request) + nonce = channel.json_body["nonce"] + + want_mac = hmac.new(key=b"shared", digestmod=hashlib.sha1) + want_mac.update( + nonce.encode("ascii") + b"\x00bob\x00abc123\x00admin\x00support" + ) + want_mac = want_mac.hexdigest() + + body = json.dumps( + { + "nonce": nonce, + "username": "bob", + "password": "abc123", + "admin": True, + "user_type": UserTypes.SUPPORT, + "mac": want_mac, + } + ) + request, channel = self.make_request("POST", self.url, body.encode("utf8")) + self.render(request) + + self.assertEqual(200, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@bob:test", channel.json_body["user_id"]) + class UsersListTestCase(unittest.HomeserverTestCase): @@ -368,6 +417,7 @@ class UserRestTestCase(unittest.HomeserverTestCase): servlets = [ synapse.rest.admin.register_servlets, login.register_servlets, + sync.register_servlets, ] def prepare(self, reactor, clock, hs): @@ -386,7 +436,6 @@ class UserRestTestCase(unittest.HomeserverTestCase): """ If the user is not a server admin, an error is returned. """ - self.hs.config.registration_shared_secret = None url = "/_synapse/admin/v2/users/@bob:test" request, channel = self.make_request( @@ -409,7 +458,6 @@ class UserRestTestCase(unittest.HomeserverTestCase): """ Tests that a lookup for a user that does not exist returns a 404 """ - self.hs.config.registration_shared_secret = None request, channel = self.make_request( "GET", @@ -425,7 +473,6 @@ class UserRestTestCase(unittest.HomeserverTestCase): """ Check that a new admin user is created successfully. """ - self.hs.config.registration_shared_secret = None url = "/_synapse/admin/v2/users/@bob:test" # Create user (server admin) @@ -473,7 +520,6 @@ class UserRestTestCase(unittest.HomeserverTestCase): """ Check that a new regular user is created successfully. """ - self.hs.config.registration_shared_secret = None url = "/_synapse/admin/v2/users/@bob:test" # Create user @@ -516,14 +562,114 @@ class UserRestTestCase(unittest.HomeserverTestCase): self.assertEqual(False, channel.json_body["is_guest"]) self.assertEqual(False, channel.json_body["deactivated"]) + @override_config( + {"limit_usage_by_mau": True, "max_mau_value": 2, "mau_trial_days": 0} + ) + def test_create_user_mau_limit_reached_active_admin(self): + """ + Check that an admin can register a new user via the admin API + even if the MAU limit is reached. + Admin user was active before creating user. + """ + + handler = self.hs.get_registration_handler() + + # Sync to set admin user to active + # before limit of monthly active users is reached + request, channel = self.make_request( + "GET", "/sync", access_token=self.admin_user_tok + ) + self.render(request) + + if channel.code != 200: + raise HttpResponseException( + channel.code, channel.result["reason"], channel.result["body"] + ) + + # Set monthly active users to the limit + self.store.get_monthly_active_count = Mock( + return_value=self.hs.config.max_mau_value + ) + # Check that the blocking of monthly active users is working as expected + # The registration of a new user fails due to the limit + self.get_failure( + handler.register_user(localpart="local_part"), ResourceLimitError + ) + + # Register new user with admin API + url = "/_synapse/admin/v2/users/@bob:test" + + # Create user + body = json.dumps({"password": "abc123", "admin": False}) + + request, channel = self.make_request( + "PUT", + url, + access_token=self.admin_user_tok, + content=body.encode(encoding="utf_8"), + ) + self.render(request) + + self.assertEqual(201, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@bob:test", channel.json_body["name"]) + self.assertEqual(False, channel.json_body["admin"]) + + @override_config( + {"limit_usage_by_mau": True, "max_mau_value": 2, "mau_trial_days": 0} + ) + def test_create_user_mau_limit_reached_passive_admin(self): + """ + Check that an admin can register a new user via the admin API + even if the MAU limit is reached. + Admin user was not active before creating user. + """ + + handler = self.hs.get_registration_handler() + + # Set monthly active users to the limit + self.store.get_monthly_active_count = Mock( + return_value=self.hs.config.max_mau_value + ) + # Check that the blocking of monthly active users is working as expected + # The registration of a new user fails due to the limit + self.get_failure( + handler.register_user(localpart="local_part"), ResourceLimitError + ) + + # Register new user with admin API + url = "/_synapse/admin/v2/users/@bob:test" + + # Create user + body = json.dumps({"password": "abc123", "admin": False}) + + request, channel = self.make_request( + "PUT", + url, + access_token=self.admin_user_tok, + content=body.encode(encoding="utf_8"), + ) + self.render(request) + + # Admin user is not blocked by mau anymore + self.assertEqual(201, int(channel.result["code"]), msg=channel.result["body"]) + self.assertEqual("@bob:test", channel.json_body["name"]) + self.assertEqual(False, channel.json_body["admin"]) + + @override_config( + { + "email": { + "enable_notifs": True, + "notif_for_new_users": True, + "notif_from": "test@example.com", + }, + "public_baseurl": "https://example.com", + } + ) def test_create_user_email_notif_for_new_users(self): """ Check that a new regular user is created successfully and got an email pusher. """ - self.hs.config.registration_shared_secret = None - self.hs.config.email_enable_notifs = True - self.hs.config.email_notif_for_new_users = True url = "/_synapse/admin/v2/users/@bob:test" # Create user @@ -554,14 +700,21 @@ class UserRestTestCase(unittest.HomeserverTestCase): self.assertEqual(len(pushers), 1) self.assertEqual("@bob:test", pushers[0]["user_name"]) + @override_config( + { + "email": { + "enable_notifs": False, + "notif_for_new_users": False, + "notif_from": "test@example.com", + }, + "public_baseurl": "https://example.com", + } + ) def test_create_user_email_no_notif_for_new_users(self): """ Check that a new regular user is created successfully and got not an email pusher. """ - self.hs.config.registration_shared_secret = None - self.hs.config.email_enable_notifs = False - self.hs.config.email_notif_for_new_users = False url = "/_synapse/admin/v2/users/@bob:test" # Create user @@ -595,7 +748,6 @@ class UserRestTestCase(unittest.HomeserverTestCase): """ Test setting a new password for another user. """ - self.hs.config.registration_shared_secret = None # Change password body = json.dumps({"password": "hahaha"}) @@ -614,7 +766,6 @@ class UserRestTestCase(unittest.HomeserverTestCase): """ Test setting the displayname of another user. """ - self.hs.config.registration_shared_secret = None # Modify user body = json.dumps({"displayname": "foobar"}) @@ -645,7 +796,6 @@ class UserRestTestCase(unittest.HomeserverTestCase): """ Test setting threepid for an other user. """ - self.hs.config.registration_shared_secret = None # Delete old and add new threepid to user body = json.dumps( @@ -711,7 +861,6 @@ class UserRestTestCase(unittest.HomeserverTestCase): """ Test setting the admin flag on a user. """ - self.hs.config.registration_shared_secret = None # Set a user as an admin body = json.dumps({"admin": True}) @@ -743,7 +892,6 @@ class UserRestTestCase(unittest.HomeserverTestCase): Ensure an account can't accidentally be deactivated by using a str value for the deactivated body parameter """ - self.hs.config.registration_shared_secret = None url = "/_synapse/admin/v2/users/@bob:test" # Create user -- cgit 1.4.1 From 09099313e6d527938013bb46640efc3768960d21 Mon Sep 17 00:00:00 2001 From: Travis Ralston Date: Fri, 5 Jun 2020 11:18:15 -0600 Subject: Add an option to disable autojoin for guest accounts (#6637) Fixes https://github.com/matrix-org/synapse/issues/3177 --- changelog.d/6637.feature | 1 + docs/sample_config.yaml | 7 +++++++ synapse/config/registration.py | 8 ++++++++ synapse/handlers/register.py | 8 +++++++- tests/handlers/test_register.py | 10 ++++++++++ 5 files changed, 33 insertions(+), 1 deletion(-) create mode 100644 changelog.d/6637.feature (limited to 'tests') diff --git a/changelog.d/6637.feature b/changelog.d/6637.feature new file mode 100644 index 0000000000..5228ebc1e5 --- /dev/null +++ b/changelog.d/6637.feature @@ -0,0 +1 @@ +Add an option to disable autojoining rooms for guest accounts. diff --git a/docs/sample_config.yaml b/docs/sample_config.yaml index b06394a2bd..94e1ec698f 100644 --- a/docs/sample_config.yaml +++ b/docs/sample_config.yaml @@ -1223,6 +1223,13 @@ account_threepid_delegates: # #autocreate_auto_join_rooms: true +# When auto_join_rooms is specified, setting this flag to false prevents +# guest accounts from being automatically joined to the rooms. +# +# Defaults to true. +# +#auto_join_rooms_for_guests: false + ## Metrics ### diff --git a/synapse/config/registration.py b/synapse/config/registration.py index a9aa8c3737..fecced2d57 100644 --- a/synapse/config/registration.py +++ b/synapse/config/registration.py @@ -128,6 +128,7 @@ class RegistrationConfig(Config): if not RoomAlias.is_valid(room_alias): raise ConfigError("Invalid auto_join_rooms entry %s" % (room_alias,)) self.autocreate_auto_join_rooms = config.get("autocreate_auto_join_rooms", True) + self.auto_join_rooms_for_guests = config.get("auto_join_rooms_for_guests", True) self.enable_set_displayname = config.get("enable_set_displayname", True) self.enable_set_avatar_url = config.get("enable_set_avatar_url", True) @@ -368,6 +369,13 @@ class RegistrationConfig(Config): # users cannot be auto-joined since they do not exist. # #autocreate_auto_join_rooms: true + + # When auto_join_rooms is specified, setting this flag to false prevents + # guest accounts from being automatically joined to the rooms. + # + # Defaults to true. + # + #auto_join_rooms_for_guests: false """ % locals() ) diff --git a/synapse/handlers/register.py b/synapse/handlers/register.py index ffda09226c..5c7113a3bb 100644 --- a/synapse/handlers/register.py +++ b/synapse/handlers/register.py @@ -244,7 +244,13 @@ class RegistrationHandler(BaseHandler): fail_count += 1 if not self.hs.config.user_consent_at_registration: - yield defer.ensureDeferred(self._auto_join_rooms(user_id)) + if not self.hs.config.auto_join_rooms_for_guests and make_guest: + logger.info( + "Skipping auto-join for %s because auto-join for guests is disabled", + user_id, + ) + else: + yield defer.ensureDeferred(self._auto_join_rooms(user_id)) else: logger.info( "Skipping auto-join for %s because consent is required at registration", diff --git a/tests/handlers/test_register.py b/tests/handlers/test_register.py index 1b7935cef2..ca32f993a3 100644 --- a/tests/handlers/test_register.py +++ b/tests/handlers/test_register.py @@ -135,6 +135,16 @@ class RegistrationTestCase(unittest.HomeserverTestCase): self.handler.register_user(localpart="local_part"), ResourceLimitError ) + def test_auto_join_rooms_for_guests(self): + room_alias_str = "#room:test" + self.hs.config.auto_join_rooms = [room_alias_str] + self.hs.config.auto_join_rooms_for_guests = False + user_id = self.get_success( + self.handler.register_user(localpart="jeff", make_guest=True), + ) + rooms = self.get_success(self.store.get_rooms_for_user(user_id)) + self.assertEqual(len(rooms), 0) + def test_auto_create_auto_join_rooms(self): room_alias_str = "#room:test" self.hs.config.auto_join_rooms = [room_alias_str] -- cgit 1.4.1 From 737b4a936e35daaa839e7296d888041941546b47 Mon Sep 17 00:00:00 2001 From: Patrick Cloke Date: Fri, 5 Jun 2020 14:42:55 -0400 Subject: Convert user directory handler and related classes to async/await. (#7640) --- changelog.d/7640.misc | 1 + synapse/handlers/register.py | 6 +- synapse/handlers/state_deltas.py | 9 +-- synapse/handlers/stats.py | 47 ++++++-------- synapse/handlers/user_directory.py | 118 +++++++++++++--------------------- tests/handlers/test_user_directory.py | 8 +-- 6 files changed, 78 insertions(+), 111 deletions(-) create mode 100644 changelog.d/7640.misc (limited to 'tests') diff --git a/changelog.d/7640.misc b/changelog.d/7640.misc new file mode 100644 index 0000000000..55edc1c781 --- /dev/null +++ b/changelog.d/7640.misc @@ -0,0 +1 @@ +Convert user directory, state deltas, and stats handlers to async/await. diff --git a/synapse/handlers/register.py b/synapse/handlers/register.py index 5c7113a3bb..af812dbda9 100644 --- a/synapse/handlers/register.py +++ b/synapse/handlers/register.py @@ -207,8 +207,10 @@ class RegistrationHandler(BaseHandler): if self.hs.config.user_directory_search_all_users: profile = yield self.store.get_profileinfo(localpart) - yield self.user_directory_handler.handle_local_profile_change( - user_id, profile + yield defer.ensureDeferred( + self.user_directory_handler.handle_local_profile_change( + user_id, profile + ) ) else: diff --git a/synapse/handlers/state_deltas.py b/synapse/handlers/state_deltas.py index f065970c40..8590c1eff4 100644 --- a/synapse/handlers/state_deltas.py +++ b/synapse/handlers/state_deltas.py @@ -15,8 +15,6 @@ import logging -from twisted.internet import defer - logger = logging.getLogger(__name__) @@ -24,8 +22,7 @@ class StateDeltasHandler(object): def __init__(self, hs): self.store = hs.get_datastore() - @defer.inlineCallbacks - def _get_key_change(self, prev_event_id, event_id, key_name, public_value): + async def _get_key_change(self, prev_event_id, event_id, key_name, public_value): """Given two events check if the `key_name` field in content changed from not matching `public_value` to doing so. @@ -41,10 +38,10 @@ class StateDeltasHandler(object): prev_event = None event = None if prev_event_id: - prev_event = yield self.store.get_event(prev_event_id, allow_none=True) + prev_event = await self.store.get_event(prev_event_id, allow_none=True) if event_id: - event = yield self.store.get_event(event_id, allow_none=True) + event = await self.store.get_event(event_id, allow_none=True) if not event and not prev_event: logger.debug("Neither event exists: %r %r", prev_event_id, event_id) diff --git a/synapse/handlers/stats.py b/synapse/handlers/stats.py index d93a276693..149f861239 100644 --- a/synapse/handlers/stats.py +++ b/synapse/handlers/stats.py @@ -16,17 +16,14 @@ import logging from collections import Counter -from twisted.internet import defer - from synapse.api.constants import EventTypes, Membership -from synapse.handlers.state_deltas import StateDeltasHandler from synapse.metrics import event_processing_positions from synapse.metrics.background_process_metrics import run_as_background_process logger = logging.getLogger(__name__) -class StatsHandler(StateDeltasHandler): +class StatsHandler: """Handles keeping the *_stats tables updated with a simple time-series of information about the users, rooms and media on the server, such that admins have some idea of who is consuming their resources. @@ -35,7 +32,6 @@ class StatsHandler(StateDeltasHandler): """ def __init__(self, hs): - super(StatsHandler, self).__init__(hs) self.hs = hs self.store = hs.get_datastore() self.state = hs.get_state_handler() @@ -68,20 +64,18 @@ class StatsHandler(StateDeltasHandler): self._is_processing = True - @defer.inlineCallbacks - def process(): + async def process(): try: - yield self._unsafe_process() + await self._unsafe_process() finally: self._is_processing = False run_as_background_process("stats.notify_new_event", process) - @defer.inlineCallbacks - def _unsafe_process(self): + async def _unsafe_process(self): # If self.pos is None then means we haven't fetched it from DB if self.pos is None: - self.pos = yield self.store.get_stats_positions() + self.pos = await self.store.get_stats_positions() # Loop round handling deltas until we're up to date @@ -96,13 +90,13 @@ class StatsHandler(StateDeltasHandler): logger.debug( "Processing room stats %s->%s", self.pos, room_max_stream_ordering ) - max_pos, deltas = yield self.store.get_current_state_deltas( + max_pos, deltas = await self.store.get_current_state_deltas( self.pos, room_max_stream_ordering ) if deltas: logger.debug("Handling %d state deltas", len(deltas)) - room_deltas, user_deltas = yield self._handle_deltas(deltas) + room_deltas, user_deltas = await self._handle_deltas(deltas) else: room_deltas = {} user_deltas = {} @@ -111,7 +105,7 @@ class StatsHandler(StateDeltasHandler): ( room_count, user_count, - ) = yield self.store.get_changes_room_total_events_and_bytes( + ) = await self.store.get_changes_room_total_events_and_bytes( self.pos, max_pos ) @@ -125,7 +119,7 @@ class StatsHandler(StateDeltasHandler): logger.debug("user_deltas: %s", user_deltas) # Always call this so that we update the stats position. - yield self.store.bulk_update_stats_delta( + await self.store.bulk_update_stats_delta( self.clock.time_msec(), updates={"room": room_deltas, "user": user_deltas}, stream_id=max_pos, @@ -137,13 +131,12 @@ class StatsHandler(StateDeltasHandler): self.pos = max_pos - @defer.inlineCallbacks - def _handle_deltas(self, deltas): + async def _handle_deltas(self, deltas): """Called with the state deltas to process Returns: - Deferred[tuple[dict[str, Counter], dict[str, counter]]] - Resovles to two dicts, the room deltas and the user deltas, + tuple[dict[str, Counter], dict[str, counter]] + Two dicts: the room deltas and the user deltas, mapping from room/user ID to changes in the various fields. """ @@ -162,7 +155,7 @@ class StatsHandler(StateDeltasHandler): logger.debug("Handling: %r, %r %r, %s", room_id, typ, state_key, event_id) - token = yield self.store.get_earliest_token_for_stats("room", room_id) + token = await self.store.get_earliest_token_for_stats("room", room_id) # If the earliest token to begin from is larger than our current # stream ID, skip processing this delta. @@ -184,7 +177,7 @@ class StatsHandler(StateDeltasHandler): sender = None if event_id is not None: - event = yield self.store.get_event(event_id, allow_none=True) + event = await self.store.get_event(event_id, allow_none=True) if event: event_content = event.content or {} sender = event.sender @@ -200,16 +193,16 @@ class StatsHandler(StateDeltasHandler): room_stats_delta["current_state_events"] += 1 if typ == EventTypes.Member: - # we could use _get_key_change here but it's a bit inefficient - # given we're not testing for a specific result; might as well - # just grab the prev_membership and membership strings and - # compare them. + # we could use StateDeltasHandler._get_key_change here but it's + # a bit inefficient given we're not testing for a specific + # result; might as well just grab the prev_membership and + # membership strings and compare them. # We take None rather than leave as a previous membership # in the absence of a previous event because we do not want to # reduce the leave count when a new-to-the-room user joins. prev_membership = None if prev_event_id is not None: - prev_event = yield self.store.get_event( + prev_event = await self.store.get_event( prev_event_id, allow_none=True ) if prev_event: @@ -301,6 +294,6 @@ class StatsHandler(StateDeltasHandler): for room_id, state in room_to_state_updates.items(): logger.debug("Updating room_stats_state for %s: %s", room_id, state) - yield self.store.update_room_state(room_id, state) + await self.store.update_room_state(room_id, state) return room_to_stats_deltas, user_to_stats_deltas diff --git a/synapse/handlers/user_directory.py b/synapse/handlers/user_directory.py index 722760c59d..12423b909a 100644 --- a/synapse/handlers/user_directory.py +++ b/synapse/handlers/user_directory.py @@ -17,14 +17,11 @@ import logging from six import iteritems, iterkeys -from twisted.internet import defer - import synapse.metrics from synapse.api.constants import EventTypes, JoinRules, Membership from synapse.handlers.state_deltas import StateDeltasHandler from synapse.metrics.background_process_metrics import run_as_background_process from synapse.storage.roommember import ProfileInfo -from synapse.types import get_localpart_from_id from synapse.util.metrics import Measure logger = logging.getLogger(__name__) @@ -103,43 +100,39 @@ class UserDirectoryHandler(StateDeltasHandler): if self._is_processing: return - @defer.inlineCallbacks - def process(): + async def process(): try: - yield self._unsafe_process() + await self._unsafe_process() finally: self._is_processing = False self._is_processing = True run_as_background_process("user_directory.notify_new_event", process) - @defer.inlineCallbacks - def handle_local_profile_change(self, user_id, profile): + async def handle_local_profile_change(self, user_id, profile): """Called to update index of our local user profiles when they change irrespective of any rooms the user may be in. """ # FIXME(#3714): We should probably do this in the same worker as all # the other changes. - is_support = yield self.store.is_support_user(user_id) + is_support = await self.store.is_support_user(user_id) # Support users are for diagnostics and should not appear in the user directory. if not is_support: - yield self.store.update_profile_in_user_dir( + await self.store.update_profile_in_user_dir( user_id, profile.display_name, profile.avatar_url ) - @defer.inlineCallbacks - def handle_user_deactivated(self, user_id): + async def handle_user_deactivated(self, user_id): """Called when a user ID is deactivated """ # FIXME(#3714): We should probably do this in the same worker as all # the other changes. - yield self.store.remove_from_user_dir(user_id) + await self.store.remove_from_user_dir(user_id) - @defer.inlineCallbacks - def _unsafe_process(self): + async def _unsafe_process(self): # If self.pos is None then means we haven't fetched it from DB if self.pos is None: - self.pos = yield self.store.get_user_directory_stream_pos() + self.pos = await self.store.get_user_directory_stream_pos() # If still None then the initial background update hasn't happened yet if self.pos is None: @@ -155,12 +148,12 @@ class UserDirectoryHandler(StateDeltasHandler): logger.debug( "Processing user stats %s->%s", self.pos, room_max_stream_ordering ) - max_pos, deltas = yield self.store.get_current_state_deltas( + max_pos, deltas = await self.store.get_current_state_deltas( self.pos, room_max_stream_ordering ) logger.debug("Handling %d state deltas", len(deltas)) - yield self._handle_deltas(deltas) + await self._handle_deltas(deltas) self.pos = max_pos @@ -169,10 +162,9 @@ class UserDirectoryHandler(StateDeltasHandler): max_pos ) - yield self.store.update_user_directory_stream_pos(max_pos) + await self.store.update_user_directory_stream_pos(max_pos) - @defer.inlineCallbacks - def _handle_deltas(self, deltas): + async def _handle_deltas(self, deltas): """Called with the state deltas to process """ for delta in deltas: @@ -187,11 +179,11 @@ class UserDirectoryHandler(StateDeltasHandler): # For join rule and visibility changes we need to check if the room # may have become public or not and add/remove the users in said room if typ in (EventTypes.RoomHistoryVisibility, EventTypes.JoinRules): - yield self._handle_room_publicity_change( + await self._handle_room_publicity_change( room_id, prev_event_id, event_id, typ ) elif typ == EventTypes.Member: - change = yield self._get_key_change( + change = await self._get_key_change( prev_event_id, event_id, key_name="membership", @@ -201,7 +193,7 @@ class UserDirectoryHandler(StateDeltasHandler): if change is False: # Need to check if the server left the room entirely, if so # we might need to remove all the users in that room - is_in_room = yield self.store.is_host_joined( + is_in_room = await self.store.is_host_joined( room_id, self.server_name ) if not is_in_room: @@ -209,40 +201,41 @@ class UserDirectoryHandler(StateDeltasHandler): # Fetch all the users that we marked as being in user # directory due to being in the room and then check if # need to remove those users or not - user_ids = yield self.store.get_users_in_dir_due_to_room( + user_ids = await self.store.get_users_in_dir_due_to_room( room_id ) for user_id in user_ids: - yield self._handle_remove_user(room_id, user_id) + await self._handle_remove_user(room_id, user_id) return else: logger.debug("Server is still in room: %r", room_id) - is_support = yield self.store.is_support_user(state_key) + is_support = await self.store.is_support_user(state_key) if not is_support: if change is None: # Handle any profile changes - yield self._handle_profile_change( + await self._handle_profile_change( state_key, room_id, prev_event_id, event_id ) continue if change: # The user joined - event = yield self.store.get_event(event_id, allow_none=True) + event = await self.store.get_event(event_id, allow_none=True) profile = ProfileInfo( avatar_url=event.content.get("avatar_url"), display_name=event.content.get("displayname"), ) - yield self._handle_new_user(room_id, state_key, profile) + await self._handle_new_user(room_id, state_key, profile) else: # The user left - yield self._handle_remove_user(room_id, state_key) + await self._handle_remove_user(room_id, state_key) else: logger.debug("Ignoring irrelevant type: %r", typ) - @defer.inlineCallbacks - def _handle_room_publicity_change(self, room_id, prev_event_id, event_id, typ): + async def _handle_room_publicity_change( + self, room_id, prev_event_id, event_id, typ + ): """Handle a room having potentially changed from/to world_readable/publically joinable. @@ -255,14 +248,14 @@ class UserDirectoryHandler(StateDeltasHandler): logger.debug("Handling change for %s: %s", typ, room_id) if typ == EventTypes.RoomHistoryVisibility: - change = yield self._get_key_change( + change = await self._get_key_change( prev_event_id, event_id, key_name="history_visibility", public_value="world_readable", ) elif typ == EventTypes.JoinRules: - change = yield self._get_key_change( + change = await self._get_key_change( prev_event_id, event_id, key_name="join_rule", @@ -278,7 +271,7 @@ class UserDirectoryHandler(StateDeltasHandler): # There's been a change to or from being world readable. - is_public = yield self.store.is_room_world_readable_or_publicly_joinable( + is_public = await self.store.is_room_world_readable_or_publicly_joinable( room_id ) @@ -293,11 +286,11 @@ class UserDirectoryHandler(StateDeltasHandler): # ignore the change return - users_with_profile = yield self.state.get_current_users_in_room(room_id) + users_with_profile = await self.state.get_current_users_in_room(room_id) # Remove every user from the sharing tables for that room. for user_id in iterkeys(users_with_profile): - yield self.store.remove_user_who_share_room(user_id, room_id) + await self.store.remove_user_who_share_room(user_id, room_id) # Then, re-add them to the tables. # NOTE: this is not the most efficient method, as handle_new_user sets @@ -306,26 +299,9 @@ class UserDirectoryHandler(StateDeltasHandler): # being added multiple times. The batching upserts shouldn't make this # too bad, though. for user_id, profile in iteritems(users_with_profile): - yield self._handle_new_user(room_id, user_id, profile) - - @defer.inlineCallbacks - def _handle_local_user(self, user_id): - """Adds a new local roomless user into the user_directory_search table. - Used to populate up the user index when we have an - user_directory_search_all_users specified. - """ - logger.debug("Adding new local user to dir, %r", user_id) - - profile = yield self.store.get_profileinfo(get_localpart_from_id(user_id)) - - row = yield self.store.get_user_in_directory(user_id) - if not row: - yield self.store.update_profile_in_user_dir( - user_id, profile.display_name, profile.avatar_url - ) + await self._handle_new_user(room_id, user_id, profile) - @defer.inlineCallbacks - def _handle_new_user(self, room_id, user_id, profile): + async def _handle_new_user(self, room_id, user_id, profile): """Called when we might need to add user to directory Args: @@ -334,18 +310,18 @@ class UserDirectoryHandler(StateDeltasHandler): """ logger.debug("Adding new user to dir, %r", user_id) - yield self.store.update_profile_in_user_dir( + await self.store.update_profile_in_user_dir( user_id, profile.display_name, profile.avatar_url ) - is_public = yield self.store.is_room_world_readable_or_publicly_joinable( + is_public = await self.store.is_room_world_readable_or_publicly_joinable( room_id ) # Now we update users who share rooms with users. - users_with_profile = yield self.state.get_current_users_in_room(room_id) + users_with_profile = await self.state.get_current_users_in_room(room_id) if is_public: - yield self.store.add_users_in_public_rooms(room_id, (user_id,)) + await self.store.add_users_in_public_rooms(room_id, (user_id,)) else: to_insert = set() @@ -376,10 +352,9 @@ class UserDirectoryHandler(StateDeltasHandler): to_insert.add((other_user_id, user_id)) if to_insert: - yield self.store.add_users_who_share_private_room(room_id, to_insert) + await self.store.add_users_who_share_private_room(room_id, to_insert) - @defer.inlineCallbacks - def _handle_remove_user(self, room_id, user_id): + async def _handle_remove_user(self, room_id, user_id): """Called when we might need to remove user from directory Args: @@ -389,24 +364,23 @@ class UserDirectoryHandler(StateDeltasHandler): logger.debug("Removing user %r", user_id) # Remove user from sharing tables - yield self.store.remove_user_who_share_room(user_id, room_id) + await self.store.remove_user_who_share_room(user_id, room_id) # Are they still in any rooms? If not, remove them entirely. - rooms_user_is_in = yield self.store.get_user_dir_rooms_user_is_in(user_id) + rooms_user_is_in = await self.store.get_user_dir_rooms_user_is_in(user_id) if len(rooms_user_is_in) == 0: - yield self.store.remove_from_user_dir(user_id) + await self.store.remove_from_user_dir(user_id) - @defer.inlineCallbacks - def _handle_profile_change(self, user_id, room_id, prev_event_id, event_id): + async def _handle_profile_change(self, user_id, room_id, prev_event_id, event_id): """Check member event changes for any profile changes and update the database if there are. """ if not prev_event_id or not event_id: return - prev_event = yield self.store.get_event(prev_event_id, allow_none=True) - event = yield self.store.get_event(event_id, allow_none=True) + prev_event = await self.store.get_event(prev_event_id, allow_none=True) + event = await self.store.get_event(event_id, allow_none=True) if not prev_event or not event: return @@ -421,4 +395,4 @@ class UserDirectoryHandler(StateDeltasHandler): new_avatar = event.content.get("avatar_url") if prev_name != new_name or prev_avatar != new_avatar: - yield self.store.update_profile_in_user_dir(user_id, new_name, new_avatar) + await self.store.update_profile_in_user_dir(user_id, new_name, new_avatar) diff --git a/tests/handlers/test_user_directory.py b/tests/handlers/test_user_directory.py index 572df8d80b..c15bce5bef 100644 --- a/tests/handlers/test_user_directory.py +++ b/tests/handlers/test_user_directory.py @@ -14,6 +14,8 @@ # limitations under the License. from mock import Mock +from twisted.internet import defer + import synapse.rest.admin from synapse.api.constants import UserTypes from synapse.rest.client.v1 import login, room @@ -75,18 +77,16 @@ class UserDirectoryTestCase(unittest.HomeserverTestCase): ) ) - self.store.remove_from_user_dir = Mock() - self.store.remove_from_user_in_public_room = Mock() + self.store.remove_from_user_dir = Mock(return_value=defer.succeed(None)) self.get_success(self.handler.handle_user_deactivated(s_user_id)) self.store.remove_from_user_dir.not_called() - self.store.remove_from_user_in_public_room.not_called() def test_handle_user_deactivated_regular_user(self): r_user_id = "@regular:test" self.get_success( self.store.register_user(user_id=r_user_id, password_hash=None) ) - self.store.remove_from_user_dir = Mock() + self.store.remove_from_user_dir = Mock(return_value=defer.succeed(None)) self.get_success(self.handler.handle_user_deactivated(r_user_id)) self.store.remove_from_user_dir.called_once_with(r_user_id) -- cgit 1.4.1