diff options
Diffstat (limited to 'synapse/handlers/sync.py')
| -rw-r--r-- | synapse/handlers/sync.py | 367 |
1 files changed, 324 insertions, 43 deletions
diff --git a/synapse/handlers/sync.py b/synapse/handlers/sync.py index 51ec727df0..ef20c2296c 100644 --- a/synapse/handlers/sync.py +++ b/synapse/handlers/sync.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- -# Copyright 2015 - 2016 OpenMarket Ltd +# 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. @@ -13,24 +14,34 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.api.constants import Membership, EventTypes -from synapse.util.async import concurrently_execute +import collections +import itertools +import logging + +from six import iteritems, itervalues + +from twisted.internet import defer + +from synapse.api.constants import EventTypes, Membership +from synapse.push.clientformat import format_push_rules_for_user +from synapse.types import RoomStreamToken +from synapse.util.async_helpers import concurrently_execute +from synapse.util.caches.expiringcache import ExpiringCache +from synapse.util.caches.lrucache import LruCache +from synapse.util.caches.response_cache import ResponseCache from synapse.util.logcontext import LoggingContext from synapse.util.metrics import Measure, measure_func -from synapse.util.caches.response_cache import ResponseCache -from synapse.push.clientformat import format_push_rules_for_user from synapse.visibility import filter_events_for_client -from synapse.types import RoomStreamToken -from twisted.internet import defer - -import collections -import logging -import itertools +logger = logging.getLogger(__name__) -from six import itervalues, iteritems +# Store the cache that tracks which lazy-loaded members have been sent to a given +# client for no more than 30 minutes. +LAZY_LOADED_MEMBERS_CACHE_MAX_AGE = 30 * 60 * 1000 -logger = logging.getLogger(__name__) +# Remember the last 100 members we sent to a client for the purposes of +# avoiding redundantly sending the same lazy-loaded members to the client +LAZY_LOADED_MEMBERS_CACHE_MAX_SIZE = 100 SyncConfig = collections.namedtuple("SyncConfig", [ @@ -64,6 +75,7 @@ class JoinedSyncResult(collections.namedtuple("JoinedSyncResult", [ "ephemeral", "account_data", "unread_notifications", + "summary", ])): __slots__ = [] @@ -145,7 +157,7 @@ class SyncResult(collections.namedtuple("SyncResult", [ "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 chanegd + "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", @@ -173,6 +185,7 @@ class SyncResult(collections.namedtuple("SyncResult", [ class SyncHandler(object): def __init__(self, hs): + self.hs_config = hs.config self.store = hs.get_datastore() self.notifier = hs.get_notifier() self.presence_handler = hs.get_presence_handler() @@ -180,20 +193,35 @@ class SyncHandler(object): self.clock = hs.get_clock() self.response_cache = ResponseCache(hs, "sync") self.state = hs.get_state_handler() + self.auth = hs.get_auth() + + # ExpiringCache((User, Device)) -> LruCache(state_key => event_id) + self.lazy_loaded_members_cache = ExpiringCache( + "lazy_loaded_members_cache", self.clock, + max_len=0, expiry_ms=LAZY_LOADED_MEMBERS_CACHE_MAX_AGE, + ) + @defer.inlineCallbacks 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 wait for new data to arrive on the server. If the timeout expires, then return an empty sync result. Returns: - A Deferred SyncResult. + Deferred[SyncResult] """ - return self.response_cache.wrap( + # 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 + # auth_blocking will occur) + user_id = sync_config.user.to_string() + yield self.auth.check_auth_blocking(user_id) + + res = yield self.response_cache.wrap( sync_config.request_key, self._wait_for_sync_for_user, sync_config, since_token, timeout, full_state, ) + defer.returnValue(res) @defer.inlineCallbacks def _wait_for_sync_for_user(self, sync_config, since_token, timeout, @@ -416,29 +444,44 @@ class SyncHandler(object): )) @defer.inlineCallbacks - def get_state_after_event(self, event): + def get_state_after_event(self, event, types=None, filtered_types=None): """ Get the room state after the given event Args: event(synapse.events.EventBase): event of interest + types(list[(str, str|None)]|None): List of (type, state_key) tuples + which are used to filter the state fetched. If `state_key` is None, + all events are returned of the given type. + May be None, which matches any key. + filtered_types(list[str]|None): Only apply filtering via `types` to this + list of event types. Other types of events are returned unfiltered. + If None, `types` filtering is applied to all events. Returns: A Deferred map from ((type, state_key)->Event) """ - state_ids = yield self.store.get_state_ids_for_event(event.event_id) + state_ids = yield self.store.get_state_ids_for_event( + event.event_id, types, filtered_types=filtered_types, + ) if event.is_state(): state_ids = state_ids.copy() state_ids[(event.type, event.state_key)] = event.event_id defer.returnValue(state_ids) @defer.inlineCallbacks - def get_state_at(self, room_id, stream_position): + def get_state_at(self, room_id, stream_position, types=None, filtered_types=None): """ 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 + types(list[(str, str|None)]|None): List of (type, state_key) tuples + which are used to filter the state fetched. If `state_key` is None, + all events are returned of the given type. + filtered_types(list[str]|None): Only apply filtering via `types` to this + list of event types. Other types of events are returned unfiltered. + If None, `types` filtering is applied to all events. Returns: A Deferred map from ((type, state_key)->Event) @@ -453,7 +496,9 @@ class SyncHandler(object): if last_events: last_event = last_events[-1] - state = yield self.get_state_after_event(last_event) + state = yield self.get_state_after_event( + last_event, types, filtered_types=filtered_types, + ) else: # no events in this room - so presumably no state @@ -461,9 +506,141 @@ class SyncHandler(object): defer.returnValue(state) @defer.inlineCallbacks + 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 + 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 + """ + + # FIXME: this promulgates https://github.com/matrix-org/synapse/issues/3305 + last_events, _ = yield self.store.get_recent_event_ids_for_room( + room_id, end_token=now_token.room_key, limit=1, + ) + + if not last_events: + defer.returnValue(None) + return + + last_event = last_events[-1] + state_ids = yield self.store.get_state_ids_for_event( + last_event.event_id, [ + (EventTypes.Member, None), + (EventTypes.Name, ''), + (EventTypes.CanonicalAlias, ''), + ] + ) + + member_ids = { + state_key: event_id + for (t, state_key), event_id in state_ids.iteritems() + if t == EventTypes.Member + } + name_id = state_ids.get((EventTypes.Name, '')) + canonical_alias_id = state_ids.get((EventTypes.CanonicalAlias, '')) + + summary = {} + + # FIXME: it feels very heavy to load up every single membership event + # just to calculate the counts. + member_events = yield self.store.get_events(member_ids.values()) + + joined_user_ids = [] + invited_user_ids = [] + + for ev in member_events.values(): + if ev.content.get("membership") == Membership.JOIN: + joined_user_ids.append(ev.state_key) + elif ev.content.get("membership") == Membership.INVITE: + invited_user_ids.append(ev.state_key) + + # TODO: only send these when they change. + summary["m.joined_member_count"] = len(joined_user_ids) + summary["m.invited_member_count"] = len(invited_user_ids) + + if name_id or canonical_alias_id: + defer.returnValue(summary) + + # FIXME: order by stream ordering, not alphabetic + + me = sync_config.user.to_string() + if (joined_user_ids or invited_user_ids): + summary['m.heroes'] = sorted( + [ + user_id + for user_id in (joined_user_ids + invited_user_ids) + if user_id != me + ] + )[0:5] + else: + summary['m.heroes'] = sorted( + [user_id for user_id in member_ids.keys() if user_id != me] + )[0:5] + + if not sync_config.filter_collection.lazy_load_members(): + defer.returnValue(summary) + + # ensure we send membership events for heroes if needed + cache_key = (sync_config.user.to_string(), sync_config.device_id) + cache = self.get_lazy_loaded_members_cache(cache_key) + + # track which members the client should already know about via LL: + # Ones which are already in state... + existing_members = set( + 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: + if ev.type == EventTypes.Member: + existing_members.add(ev.state_key) + + # ...and then ensure any missing ones get included in state. + missing_hero_event_ids = [ + member_ids[hero_id] + for hero_id in summary['m.heroes'] + if ( + cache.get(hero_id) != member_ids[hero_id] and + hero_id not in existing_members + ) + ] + + missing_hero_state = yield self.store.get_events(missing_hero_event_ids) + missing_hero_state = missing_hero_state.values() + + for s in missing_hero_state: + cache.set(s.state_key, s.event_id) + state[(EventTypes.Member, s.state_key)] = s + + defer.returnValue(summary) + + def get_lazy_loaded_members_cache(self, cache_key): + cache = self.lazy_loaded_members_cache.get(cache_key) + if cache is None: + logger.debug("creating LruCache for %r", cache_key) + cache = LruCache(LAZY_LOADED_MEMBERS_CACHE_MAX_SIZE) + self.lazy_loaded_members_cache[cache_key] = cache + else: + logger.debug("found LruCache for %r", cache_key) + return cache + + @defer.inlineCallbacks def compute_state_delta(self, room_id, batch, sync_config, since_token, now_token, full_state): - """ Works out the differnce in state between the start of the timeline + """ Works out the difference in state between the start of the timeline and the previous sync. Args: @@ -477,7 +654,7 @@ class SyncHandler(object): full_state(bool): Whether to force returning the full state. Returns: - A deferred new event dictionary + A deferred dict of (type, state_key) -> Event """ # TODO(mjark) Check if the state events were received by the server # after the previous sync, since we need to include those state @@ -485,59 +662,130 @@ class SyncHandler(object): # TODO(mjark) Check for new redactions in the state events. with Measure(self.clock, "compute_state_delta"): + + types = None + filtered_types = None + + lazy_load_members = sync_config.filter_collection.lazy_load_members() + include_redundant_members = ( + sync_config.filter_collection.include_redundant_members() + ) + + if lazy_load_members: + # We only request state for the members needed to display the + # timeline: + + types = [ + (EventTypes.Member, state_key) + for state_key in set( + event.sender # FIXME: we also care about invite targets etc. + for event in batch.events + ) + ] + + # only apply the filtering to room members + filtered_types = [EventTypes.Member] + + timeline_state = { + (event.type, event.state_key): event.event_id + for event in batch.events if event.is_state() + } + if full_state: if batch: current_state_ids = yield self.store.get_state_ids_for_event( - batch.events[-1].event_id + batch.events[-1].event_id, types=types, + filtered_types=filtered_types, ) state_ids = yield self.store.get_state_ids_for_event( - batch.events[0].event_id + batch.events[0].event_id, types=types, + filtered_types=filtered_types, ) + else: current_state_ids = yield self.get_state_at( - room_id, stream_position=now_token + room_id, stream_position=now_token, types=types, + filtered_types=filtered_types, ) state_ids = current_state_ids - timeline_state = { - (event.type, event.state_key): event.event_id - for event in batch.events if event.is_state() - } - state_ids = _calculate_state( timeline_contains=timeline_state, timeline_start=state_ids, previous={}, current=current_state_ids, + lazy_load_members=lazy_load_members, ) elif batch.limited: state_at_previous_sync = yield self.get_state_at( - room_id, stream_position=since_token + room_id, stream_position=since_token, types=types, + filtered_types=filtered_types, ) current_state_ids = yield self.store.get_state_ids_for_event( - batch.events[-1].event_id + batch.events[-1].event_id, types=types, + filtered_types=filtered_types, ) state_at_timeline_start = yield self.store.get_state_ids_for_event( - batch.events[0].event_id + batch.events[0].event_id, types=types, + filtered_types=filtered_types, ) - timeline_state = { - (event.type, event.state_key): event.event_id - for event in batch.events if event.is_state() - } - state_ids = _calculate_state( timeline_contains=timeline_state, timeline_start=state_at_timeline_start, previous=state_at_previous_sync, current=current_state_ids, + lazy_load_members=lazy_load_members, ) else: state_ids = {} + if lazy_load_members: + if types: + # We're returning an incremental sync, with no "gap" since + # the previous sync, so normally there would be no state to return + # But we're lazy-loading, so the client might need some more + # member events to understand the events in this timeline. + # 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( + batch.events[0].event_id, types=types, + filtered_types=None, # we only want members! + ) + + if lazy_load_members and not include_redundant_members: + cache_key = (sync_config.user.to_string(), sync_config.device_id) + cache = self.get_lazy_loaded_members_cache(cache_key) + + # if it's a new sync sequence, then assume the client has had + # amnesia and doesn't want any recent lazy-loaded members + # de-duplicated. + if since_token is None: + logger.debug("clearing LruCache for %r", cache_key) + cache.clear() + else: + # only send members which aren't in our LruCache (either + # because they're new to this client or have been pushed out + # of the cache) + logger.debug("filtering state from %r...", state_ids) + state_ids = { + t: event_id + for t, event_id in state_ids.iteritems() + if cache.get(t[1]) != event_id + } + logger.debug("...to %r", state_ids) + + # add any member IDs we are about to send into our LruCache + for t, event_id in itertools.chain( + state_ids.items(), + timeline_state.items(), + ): + if t[0] == EventTypes.Member: + cache.set(t[1], event_id) state = {} if state_ids: @@ -620,7 +868,7 @@ class SyncHandler(object): since_token is None and sync_config.filter_collection.blocks_all_presence() ) - if not block_all_presence_data: + if self.hs_config.use_presence and not block_all_presence_data: yield self._generate_sync_entry_for_presence( sync_result_builder, newly_joined_rooms, newly_joined_users ) @@ -1312,7 +1560,6 @@ class SyncHandler(object): if events == [] and tags is None: return - since_token = sync_result_builder.since_token now_token = sync_result_builder.now_token sync_config = sync_result_builder.sync_config @@ -1355,6 +1602,18 @@ class SyncHandler(object): full_state=full_state ) + summary = {} + if ( + sync_config.filter_collection.lazy_load_members() and + ( + any(ev.type == EventTypes.Member for ev in batch.events) or + since_token is None + ) + ): + summary = yield self.compute_summary( + room_id, sync_config, batch, state, now_token + ) + if room_builder.rtype == "joined": unread_notifications = {} room_sync = JoinedSyncResult( @@ -1364,6 +1623,7 @@ class SyncHandler(object): ephemeral=ephemeral, account_data=account_data_events, unread_notifications=unread_notifications, + summary=summary, ) if room_sync or always_include: @@ -1448,7 +1708,9 @@ def _action_has_highlight(actions): return False -def _calculate_state(timeline_contains, timeline_start, previous, current): +def _calculate_state( + timeline_contains, timeline_start, previous, current, lazy_load_members, +): """Works out what state to include in a sync response. Args: @@ -1457,6 +1719,9 @@ def _calculate_state(timeline_contains, timeline_start, previous, current): previous (dict): 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 + or not. assumes that timeline_start has already been filtered to + include only the members the client needs to know about. Returns: dict @@ -1472,9 +1737,25 @@ def _calculate_state(timeline_contains, timeline_start, previous, current): } c_ids = set(e for e in current.values()) - tc_ids = set(e for e in timeline_contains.values()) - p_ids = set(e for e in previous.values()) ts_ids = set(e for e in timeline_start.values()) + p_ids = set(e for e in previous.values()) + tc_ids = set(e for e in timeline_contains.values()) + + # 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, + # as we may not have sent them to the client before. We find these membership + # events by filtering them out of timeline_start, which has already been filtered + # to only include membership events for the senders in the timeline. + # In practice, we can do this by removing them from the p_ids list, + # which is the list of relevant state we know we have already sent to the client. + # see https://github.com/matrix-org/synapse/pull/2970 + # /files/efcdacad7d1b7f52f879179701c7e0d9b763511f#r204732809 + + if lazy_load_members: + p_ids.difference_update( + e for t, e in timeline_start.iteritems() + if t[0] == EventTypes.Member + ) state_ids = ((c_ids | ts_ids) - p_ids) - tc_ids |