diff --git a/synapse/storage/_base.py b/synapse/storage/_base.py
index ab596fa68d..459901ac60 100644
--- a/synapse/storage/_base.py
+++ b/synapse/storage/_base.py
@@ -409,16 +409,15 @@ class SQLBaseStore(object):
i = 0
N = 5
while True:
+ cursor = LoggingTransaction(
+ conn.cursor(),
+ name,
+ self.database_engine,
+ after_callbacks,
+ exception_callbacks,
+ )
try:
- txn = conn.cursor()
- txn = LoggingTransaction(
- txn,
- name,
- self.database_engine,
- after_callbacks,
- exception_callbacks,
- )
- r = func(txn, *args, **kwargs)
+ r = func(cursor, *args, **kwargs)
conn.commit()
return r
except self.database_engine.module.OperationalError as e:
@@ -456,6 +455,40 @@ class SQLBaseStore(object):
)
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
@@ -851,7 +884,7 @@ class SQLBaseStore(object):
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") % (
+ sql = ("INSERT INTO %s (%s) VALUES (%s) ON CONFLICT (%s) DO %s") % (
table,
", ".join(k for k in allvalues),
", ".join("?" for _ in allvalues),
diff --git a/synapse/storage/data_stores/main/appservice.py b/synapse/storage/data_stores/main/appservice.py
index 81babf2029..6635313403 100644
--- a/synapse/storage/data_stores/main/appservice.py
+++ b/synapse/storage/data_stores/main/appservice.py
@@ -34,7 +34,7 @@ def _make_exclusive_regex(services_cache):
exclusive_user_regexes = [
regex.pattern
for service in services_cache
- for regex in service.get_exlusive_user_regexes()
+ for regex in service.get_exclusive_user_regexes()
]
if exclusive_user_regexes:
exclusive_user_regex = "|".join("(" + r + ")" for r in exclusive_user_regexes)
diff --git a/synapse/storage/data_stores/main/deviceinbox.py b/synapse/storage/data_stores/main/deviceinbox.py
index 96cd0fb77a..a23744f11c 100644
--- a/synapse/storage/data_stores/main/deviceinbox.py
+++ b/synapse/storage/data_stores/main/deviceinbox.py
@@ -380,7 +380,7 @@ class DeviceInboxStore(DeviceInboxWorkerStore, DeviceInboxBackgroundUpdateStore)
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 = ?"
+ 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:
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 073412a78d..d8ad59ad93 100644
--- a/synapse/storage/data_stores/main/end_to_end_keys.py
+++ b/synapse/storage/data_stores/main/end_to_end_keys.py
@@ -138,9 +138,9 @@ class EndToEndKeyWorkerStore(SQLBaseStore):
result.setdefault(user_id, {})[device_id] = None
# get signatures on the device
- signature_sql = (
- "SELECT * " " FROM e2e_cross_signing_signatures " " WHERE %s"
- ) % (" OR ".join("(" + q + ")" for q in signature_query_clauses))
+ signature_sql = ("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)
diff --git a/synapse/storage/data_stores/main/events.py b/synapse/storage/data_stores/main/events.py
index 878f7568a6..2737a1d3ae 100644
--- a/synapse/storage/data_stores/main/events.py
+++ b/synapse/storage/data_stores/main/events.py
@@ -713,9 +713,7 @@ class EventsStore(
metadata_json = encode_json(event.internal_metadata.get_dict())
- sql = (
- "UPDATE event_json SET internal_metadata = ?" " WHERE event_id = ?"
- )
+ 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
@@ -732,7 +730,7 @@ class EventsStore(
},
)
- sql = "UPDATE events SET outlier = ?" " WHERE event_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
@@ -929,6 +927,9 @@ 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)
@@ -1479,7 +1480,7 @@ class EventsStore(
# 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("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()
diff --git a/synapse/storage/data_stores/main/filtering.py b/synapse/storage/data_stores/main/filtering.py
index a2a2a67927..f05ace299a 100644
--- a/synapse/storage/data_stores/main/filtering.py
+++ b/synapse/storage/data_stores/main/filtering.py
@@ -55,7 +55,7 @@ class FilteringStore(SQLBaseStore):
if filter_id_response is not None:
return filter_id_response[0]
- sql = "SELECT MAX(filter_id) FROM user_filters " "WHERE user_id = ?"
+ 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:
diff --git a/synapse/storage/data_stores/main/media_repository.py b/synapse/storage/data_stores/main/media_repository.py
index 84b5f3ad5e..0f2887bdce 100644
--- a/synapse/storage/data_stores/main/media_repository.py
+++ b/synapse/storage/data_stores/main/media_repository.py
@@ -337,7 +337,7 @@ class MediaRepositoryStore(MediaRepositoryBackgroundUpdateStore):
if len(media_ids) == 0:
return
- sql = "DELETE FROM local_media_repository_url_cache" " WHERE media_id = ?"
+ 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])
@@ -365,11 +365,11 @@ class MediaRepositoryStore(MediaRepositoryBackgroundUpdateStore):
return
def _delete_url_cache_media_txn(txn):
- sql = "DELETE FROM local_media_repository" " WHERE media_id = ?"
+ 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 = ?"
+ sql = "DELETE FROM local_media_repository_thumbnails WHERE media_id = ?"
txn.executemany(sql, [(media_id,) for media_id in media_ids])
diff --git a/synapse/storage/data_stores/main/profile.py b/synapse/storage/data_stores/main/profile.py
index e4e8a1c1d6..8accf0c31f 100644
--- a/synapse/storage/data_stores/main/profile.py
+++ b/synapse/storage/data_stores/main/profile.py
@@ -1,5 +1,6 @@
# -*- 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.
@@ -16,9 +17,13 @@
from twisted.internet import defer
from synapse.api.errors import StoreError
+
+from synapse.storage import background_updates
from synapse.storage._base import SQLBaseStore
from synapse.storage.data_stores.main.roommember import ProfileInfo
+BATCH_SIZE = 100
+
class ProfileWorkerStore(SQLBaseStore):
@defer.inlineCallbacks
@@ -57,6 +62,54 @@ class ProfileWorkerStore(SQLBaseStore):
desc="get_profile_avatar_url",
)
+ def get_latest_profile_replication_batch_number(self):
+ def f(txn):
+ txn.execute("SELECT MAX(batch) as maxbatch FROM profiles")
+ rows = self.cursor_to_dict(txn)
+ return rows[0]["maxbatch"]
+
+ return self.runInteraction("get_latest_profile_replication_batch_number", f)
+
+ def get_profile_batch(self, batchnum):
+ return self._simple_select_list(
+ table="profiles",
+ keyvalues={"batch": batchnum},
+ retcols=("user_id", "displayname", "avatar_url", "active"),
+ desc="get_profile_batch",
+ )
+
+ def assign_profile_batch(self):
+ def f(txn):
+ sql = (
+ "UPDATE profiles SET batch = "
+ "(SELECT COALESCE(MAX(batch), -1) + 1 FROM profiles) "
+ "WHERE user_id in ("
+ " SELECT user_id FROM profiles WHERE batch is NULL limit ?"
+ ")"
+ )
+ txn.execute(sql, (BATCH_SIZE,))
+ return txn.rowcount
+
+ return self.runInteraction("assign_profile_batch", f)
+
+ def get_replication_hosts(self):
+ def f(txn):
+ txn.execute(
+ "SELECT host, last_synced_batch FROM profile_replication_status"
+ )
+ rows = self.cursor_to_dict(txn)
+ return {r["host"]: r["last_synced_batch"] for r in rows}
+
+ return self.runInteraction("get_replication_hosts", f)
+
+ def update_replication_batch_for_host(self, host, last_synced_batch):
+ return self._simple_upsert(
+ table="profile_replication_status",
+ keyvalues={"host": host},
+ values={"last_synced_batch": last_synced_batch},
+ desc="update_replication_batch_for_host",
+ )
+
def get_from_remote_profile_cache(self, user_id):
return self._simple_select_one(
table="remote_profile_cache",
@@ -66,29 +119,53 @@ class ProfileWorkerStore(SQLBaseStore):
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(
+ def set_profile_displayname(self, user_localpart, new_displayname, batchnum):
+ return self._simple_upsert(
table="profiles",
keyvalues={"user_id": user_localpart},
- updatevalues={"displayname": new_displayname},
+ values={"displayname": new_displayname, "batch": batchnum},
desc="set_profile_displayname",
+ lock=False, # we can do this because user_id has a unique index
)
- def set_profile_avatar_url(self, user_localpart, new_avatar_url):
- return self._simple_update_one(
+ def set_profile_avatar_url(self, user_localpart, new_avatar_url, batchnum):
+ return self._simple_upsert(
table="profiles",
keyvalues={"user_id": user_localpart},
- updatevalues={"avatar_url": new_avatar_url},
+ values={"avatar_url": new_avatar_url, "batch": batchnum},
desc="set_profile_avatar_url",
+ lock=False, # we can do this because user_id has a unique index
+ )
+
+ def set_profile_active(self, user_localpart, active, hide, batchnum):
+ values = {"active": int(active), "batch": batchnum}
+ if not active and not hide:
+ # we are deactivating for real (not in hide mode)
+ # so clear the profile.
+ values["avatar_url"] = None
+ values["displayname"] = None
+ return self._simple_upsert(
+ table="profiles",
+ keyvalues={"user_id": user_localpart},
+ values=values,
+ desc="set_profile_active",
+ lock=False, # we can do this because user_id has a unique index
)
-class ProfileStore(ProfileWorkerStore):
+class ProfileStore(ProfileWorkerStore, background_updates.BackgroundUpdateStore):
+ def __init__(self, db_conn, hs):
+
+ super(ProfileStore, self).__init__(db_conn, hs)
+
+ self.register_background_index_update(
+ "profile_replication_status_host_index",
+ index_name="profile_replication_status_idx",
+ table="profile_replication_status",
+ columns=["host"],
+ unique=True,
+ )
+
def add_remote_profile_cache(self, user_id, displayname, avatar_url):
"""Ensure we are caching the remote user's profiles.
diff --git a/synapse/storage/data_stores/main/receipts.py b/synapse/storage/data_stores/main/receipts.py
index 0c24430f28..8b17334ff4 100644
--- a/synapse/storage/data_stores/main/receipts.py
+++ b/synapse/storage/data_stores/main/receipts.py
@@ -280,7 +280,7 @@ class ReceiptsWorkerStore(SQLBaseStore):
args.append(limit)
txn.execute(sql, args)
- return (r[0:5] + (json.loads(r[5]),) for r in txn)
+ return list(r[0:5] + (json.loads(r[5]),) for r in txn)
return self.runInteraction(
"get_all_updated_receipts", get_all_updated_receipts_txn
diff --git a/synapse/storage/data_stores/main/registration.py b/synapse/storage/data_stores/main/registration.py
index ee1b2b2bbf..c3c54ca859 100644
--- a/synapse/storage/data_stores/main/registration.py
+++ b/synapse/storage/data_stores/main/registration.py
@@ -157,6 +157,28 @@ class RegistrationWorkerStore(SQLBaseStore):
)
@defer.inlineCallbacks
+ def get_expired_users(self):
+ """Get IDs of all expired users
+
+ Returns:
+ Deferred[list[str]]: List of expired user IDs
+ """
+
+ def get_expired_users_txn(txn, now_ms):
+ sql = """
+ SELECT user_id from account_validity
+ WHERE expiration_ts_ms <= ?
+ """
+ txn.execute(sql, (now_ms,))
+ rows = txn.fetchall()
+ return [row[0] for row in rows]
+
+ res = yield self.runInteraction(
+ "get_expired_users", get_expired_users_txn, self.clock.time_msec()
+ )
+ defer.returnValue(res)
+
+ @defer.inlineCallbacks
def set_renewal_token_for_user(self, user_id, renewal_token):
"""Defines a renewal token for a given user.
@@ -377,9 +399,7 @@ class RegistrationWorkerStore(SQLBaseStore):
"""
def f(txn):
- sql = (
- "SELECT name, password_hash FROM users" " WHERE lower(name) = lower(?)"
- )
+ sql = "SELECT name, password_hash FROM users WHERE lower(name) = lower(?)"
txn.execute(sql, (user_id,))
return dict(txn)
diff --git a/synapse/storage/data_stores/main/room.py b/synapse/storage/data_stores/main/room.py
index 67bb1b6f60..edc8b10ab1 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
@@ -271,6 +274,24 @@ class RoomWorkerStore(SQLBaseStore):
desc="is_room_blocked",
)
+ @defer.inlineCallbacks
+ def is_room_published(self, room_id):
+ """Check whether a room has been published in the local public room
+ directory.
+
+ Args:
+ room_id (str)
+ Returns:
+ bool: Whether the room is currently published in the room directory
+ """
+ # Get room information
+ room_info = yield self.get_room(room_id)
+ if not room_info:
+ defer.returnValue(False)
+
+ # Check the is_public value
+ defer.returnValue(room_info.get("is_public", False))
+
@cachedInlineCallbacks(max_entries=10000)
def get_ratelimit_for_user(self, user_id):
"""Check if there are any overrides for ratelimiting for the given
@@ -300,8 +321,146 @@ class RoomWorkerStore(SQLBaseStore):
else:
return None
+ @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.
+ """
+ # If the room retention feature is disabled, return a policy with no minimum nor
+ # maximum, in order not to filter out events we should filter out when sending to
+ # the client.
+ if not self.config.retention_enabled:
+ defer.returnValue({"min_lifetime": None, "max_lifetime": None})
+
+ 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)
+
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 +661,35 @@ 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 +871,89 @@ 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)
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..023f493f7d 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
@@ -658,10 +658,19 @@ CREATE TABLE presence_stream (
+CREATE TABLE profile_replication_status (
+ host text NOT NULL,
+ last_synced_batch bigint NOT NULL
+);
+
+
+
CREATE TABLE profiles (
user_id text NOT NULL,
displayname text,
- avatar_url text
+ avatar_url text,
+ batch bigint,
+ active smallint DEFAULT 1 NOT NULL
);
@@ -1828,6 +1837,10 @@ CREATE INDEX presence_stream_user_id ON presence_stream USING btree (user_id);
+CREATE INDEX profiles_batch_idx ON profiles USING btree (batch);
+
+
+
CREATE INDEX public_room_index ON rooms USING btree (is_public);
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..b3112c24a9 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
@@ -6,7 +6,7 @@ CREATE TABLE presence_allow_inbound( observed_user_id TEXT NOT NULL, observer_us
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 profiles( user_id TEXT NOT NULL, displayname TEXT, avatar_url TEXT, batch BIGINT DEFAULT NULL, active SMALLINT DEFAULT 1 NOT NULL, 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) );
@@ -207,6 +207,8 @@ 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 INDEX profiles_batch_idx ON profiles(batch);
+CREATE TABLE profile_replication_status ( host TEXT NOT NULL, last_synced_batch BIGINT NOT NULL );
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/stream.py b/synapse/storage/data_stores/main/stream.py
index 8780fdd989..9ae4a913a1 100644
--- a/synapse/storage/data_stores/main/stream.py
+++ b/synapse/storage/data_stores/main/stream.py
@@ -616,7 +616,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore):
def _get_max_topological_txn(self, txn, room_id):
txn.execute(
- "SELECT MAX(topological_ordering) FROM events" " WHERE room_id = ?",
+ "SELECT MAX(topological_ordering) FROM events WHERE room_id = ?",
(room_id,),
)
diff --git a/synapse/storage/data_stores/main/tags.py b/synapse/storage/data_stores/main/tags.py
index 10d1887f75..aa24339717 100644
--- a/synapse/storage/data_stores/main/tags.py
+++ b/synapse/storage/data_stores/main/tags.py
@@ -83,9 +83,7 @@ class TagsWorkerStore(AccountDataWorkerStore):
)
def get_tag_content(txn, tag_ids):
- sql = (
- "SELECT tag, content" " FROM room_tags" " WHERE user_id=? AND room_id=?"
- )
+ 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))
diff --git a/synapse/storage/prepare_database.py b/synapse/storage/prepare_database.py
index 2e7753820e..731e1c9d9c 100644
--- a/synapse/storage/prepare_database.py
+++ b/synapse/storage/prepare_database.py
@@ -447,7 +447,7 @@ def _apply_module_schema_files(cur, database_engine, modname, names_and_streams)
# Mark as done.
cur.execute(
database_engine.convert_param_style(
- "INSERT INTO applied_module_schemas (module_name, file)" " VALUES (?,?)"
+ "INSERT INTO applied_module_schemas (module_name, file) VALUES (?,?)"
),
(modname, name),
)
diff --git a/synapse/storage/schema/delta/48/profiles_batch.sql b/synapse/storage/schema/delta/48/profiles_batch.sql
new file mode 100644
index 0000000000..e744c02fe8
--- /dev/null
+++ b/synapse/storage/schema/delta/48/profiles_batch.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.
+ */
+
+/*
+ * Add a batch number to track changes to profiles and the
+ * order they're made in so we can replicate user profiles
+ * to other hosts as they change
+ */
+ALTER TABLE profiles ADD COLUMN batch BIGINT DEFAULT NULL;
+
+/*
+ * Index on the batch number so we can get profiles
+ * by their batch
+ */
+CREATE INDEX profiles_batch_idx ON profiles(batch);
+
+/*
+ * A table to track what batch of user profiles has been
+ * synced to what profile replication target.
+ */
+CREATE TABLE profile_replication_status (
+ host TEXT NOT NULL,
+ last_synced_batch BIGINT NOT NULL
+);
diff --git a/synapse/storage/schema/delta/50/profiles_deactivated_users.sql b/synapse/storage/schema/delta/50/profiles_deactivated_users.sql
new file mode 100644
index 0000000000..c8893ecbe8
--- /dev/null
+++ b/synapse/storage/schema/delta/50/profiles_deactivated_users.sql
@@ -0,0 +1,23 @@
+/* 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 flag saying whether the user owning the profile has been deactivated
+ * This really belongs on the users table, not here, but the users table
+ * stores users by their full user_id and profiles stores them by localpart,
+ * so we can't easily join between the two tables. Plus, the batch number
+ * realy ought to represent data in this table that has changed.
+ */
+ALTER TABLE profiles ADD COLUMN active SMALLINT DEFAULT 1 NOT NULL;
diff --git a/synapse/storage/schema/delta/55/profile_replication_status_index.sql b/synapse/storage/schema/delta/55/profile_replication_status_index.sql
new file mode 100644
index 0000000000..18a0f7e10c
--- /dev/null
+++ b/synapse/storage/schema/delta/55/profile_replication_status_index.sql
@@ -0,0 +1,17 @@
+/* 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.
+ */
+
+INSERT INTO background_updates (update_name, progress_json) VALUES
+ ('profile_replication_status_host_index', '{}');
diff --git a/synapse/storage/schema/delta/55/room_retention.sql b/synapse/storage/schema/delta/55/room_retention.sql
new file mode 100644
index 0000000000..ee6cdf7a14
--- /dev/null
+++ b/synapse/storage/schema/delta/55/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', '{}');
|
