This repository has been archived by the owner on Apr 26, 2024. It is now read-only.
-
-
Notifications
You must be signed in to change notification settings - Fork 2.1k
/
events.py
2276 lines (1925 loc) · 83.3 KB
/
events.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
# -*- 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 ._base import SQLBaseStore
from twisted.internet import defer, reactor
from synapse.events import FrozenEvent, USE_FROZEN_DICTS
from synapse.events.utils import prune_event
from synapse.util.async import ObservableDeferred
from synapse.util.logcontext import (
preserve_fn, PreserveLoggingContext, preserve_context_over_deferred
)
from synapse.util.logutils import log_function
from synapse.util.metrics import Measure
from synapse.api.constants import EventTypes
from synapse.api.errors import SynapseError
from synapse.state import resolve_events
from synapse.util.caches.descriptors import cached
from synapse.types import get_domain_from_id
from canonicaljson import encode_canonical_json
from collections import deque, namedtuple, OrderedDict
from functools import wraps
import synapse.metrics
import logging
import ujson as json
# these are only included to make the type annotations work
from synapse.events import EventBase # noqa: F401
from synapse.events.snapshot import EventContext # noqa: F401
logger = logging.getLogger(__name__)
metrics = synapse.metrics.get_metrics_for(__name__)
persist_event_counter = metrics.register_counter("persisted_events")
event_counter = metrics.register_counter(
"persisted_events_sep", labels=["type", "origin_type", "origin_entity"]
)
def encode_json(json_object):
if USE_FROZEN_DICTS:
# ujson doesn't like frozen_dicts
return encode_canonical_json(json_object)
else:
return json.dumps(json_object, ensure_ascii=False)
# 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
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.
Args:
room_id (str):
events_and_contexts (list[(EventBase, EventContext)]):
backfilled (bool):
"""
queue = self._event_persist_queues.setdefault(room_id, deque())
if queue:
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())
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,1
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 deferres 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)
item.deferred.callback(ret)
except Exception as e:
item.deferred.errback(e)
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)
preserve_fn(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)
defer.returnValue(res)
return f
class EventsStore(SQLBaseStore):
EVENT_ORIGIN_SERVER_TS_NAME = "event_origin_server_ts"
EVENT_FIELDS_SENDER_URL_UPDATE_NAME = "event_fields_sender_url"
def __init__(self, hs):
super(EventsStore, self).__init__(hs)
self._clock = hs.get_clock()
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._event_persist_queue = _EventPeristenceQueue()
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: ?
"""
partitioned = {}
for event, ctx in events_and_contexts:
partitioned.setdefault(event.room_id, []).append((event, ctx))
deferreds = []
for room_id, evs_ctxs in partitioned.iteritems():
d = preserve_fn(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)
return preserve_context_over_deferred(
defer.gatherResults(deferreds, consumeErrors=True)
)
@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 preserve_context_over_deferred(deferred)
max_persisted_id = yield self._stream_id_gen.get_current_token()
defer.returnValue((event.internal_metadata.stream_ordering, max_persisted_id))
def _maybe_start_persisting(self, room_id):
@defer.inlineCallbacks
def persisting_queue(item):
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
if backfilled:
stream_ordering_manager = self._backfill_id_gen.get_next_mult(
len(events_and_contexts)
)
else:
stream_ordering_manager = self._stream_id_gen.get_next_mult(
len(events_and_contexts)
)
with stream_ordering_manager as stream_orderings:
for (event, context), stream, in zip(
events_and_contexts, stream_orderings
):
event.internal_metadata.stream_ordering = stream
chunks = [
events_and_contexts[x:x + 100]
for x in xrange(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.
new_forward_extremeties = {}
current_state_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 events_by_room.iteritems():
# Work out new extremities by recursively adding and removing
# the new events.
latest_event_ids = yield self.get_latest_event_ids_in_room(
room_id
)
new_latest_event_ids = yield self._calculate_new_extremeties(
room_id, ev_ctx_rm, latest_event_ids
)
if new_latest_event_ids == set(latest_event_ids):
# No change in extremities, so no change in state
continue
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_events) == 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 = yield self._calculate_state_delta(
room_id, ev_ctx_rm, new_latest_event_ids
)
if state:
current_state_for_room[room_id] = state
yield self.runInteraction(
"persist_events",
self._persist_events_txn,
events_and_contexts=chunk,
backfilled=backfilled,
delete_existing=delete_existing,
current_state_for_room=current_state_for_room,
new_forward_extremeties=new_forward_extremeties,
)
persist_event_counter.inc_by(len(chunk))
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.inc(event.type, origin_type, origin_entity)
for room_id, (_, _, new_state) in current_state_for_room.iteritems():
self.get_current_state_ids.prefill(
(room_id, ), new_state
)
@defer.inlineCallbacks
def _calculate_new_extremeties(self, room_id, event_contexts, latest_event_ids):
"""Calculates the new forward extremeties for a room given events to
persist.
Assumes that we are only persisting events for one room at a time.
"""
new_latest_event_ids = set(latest_event_ids)
# First, add all the new events to the list
new_latest_event_ids.update(
event.event_id for event, ctx in event_contexts
if not event.internal_metadata.is_outlier() and not ctx.rejected
)
# Now remove all events that are referenced by the to-be-added events
new_latest_event_ids.difference_update(
e_id
for event, ctx in event_contexts
for e_id, _ in event.prev_events
if not event.internal_metadata.is_outlier() and not ctx.rejected
)
# And finally remove any events that are referenced by previously added
# events.
rows = yield self._simple_select_many_batch(
table="event_edges",
column="prev_event_id",
iterable=list(new_latest_event_ids),
retcols=["prev_event_id"],
keyvalues={
"room_id": room_id,
"is_state": False,
},
desc="_calculate_new_extremeties",
)
new_latest_event_ids.difference_update(
row["prev_event_id"] for row in rows
)
defer.returnValue(new_latest_event_ids)
@defer.inlineCallbacks
def _calculate_state_delta(self, room_id, events_context, new_latest_event_ids):
"""Calculate the new state deltas for a room.
Assumes that we are only persisting events for one room at a time.
Returns:
3-tuple (to_delete, to_insert, new_state) where both are state dicts,
i.e. (type, state_key) -> event_id. `to_delete` are the entries to
first be deleted from current_state_events, `to_insert` are entries
to insert. `new_state` is the full set of state.
May return None if there are no changes to be applied.
"""
# Now we need to work out the different state sets for
# each state extremities
state_sets = []
state_groups = set()
missing_event_ids = []
was_updated = False
for event_id in new_latest_event_ids:
# First search in the list of new events we're adding,
# and then use the current state from that
for ev, ctx in events_context:
if event_id == ev.event_id:
if ctx.current_state_ids is None:
raise Exception("Unknown current state")
# If we've already seen the state group don't bother adding
# it to the state sets again
if ctx.state_group not in state_groups:
state_sets.append(ctx.current_state_ids)
if ctx.delta_ids or hasattr(ev, "state_key"):
was_updated = True
if ctx.state_group:
# Add this as a seen state group (if it has a state
# group)
state_groups.add(ctx.state_group)
break
else:
# If we couldn't find it, then we'll need to pull
# the state from the database
was_updated = True
missing_event_ids.append(event_id)
if missing_event_ids:
# Now pull out the state for any missing events from DB
event_to_groups = yield self._get_state_group_for_events(
missing_event_ids,
)
groups = set(event_to_groups.itervalues()) - state_groups
if groups:
group_to_state = yield self._get_state_for_groups(groups)
state_sets.extend(group_to_state.itervalues())
if not new_latest_event_ids:
current_state = {}
elif was_updated:
if len(state_sets) == 1:
# If there is only one state set, then we know what the current
# state is.
current_state = state_sets[0]
else:
# We work out the current state by passing the state sets to the
# state resolution algorithm. It may ask for some events, including
# the events we have yet to persist, so we need a slightly more
# complicated event lookup function than simply looking the events
# up in the db.
events_map = {ev.event_id: ev for ev, _ in events_context}
@defer.inlineCallbacks
def get_events(ev_ids):
# We get the events by first looking at the list of events we
# are trying to persist, and then fetching the rest from the DB.
db = []
to_return = {}
for ev_id in ev_ids:
ev = events_map.get(ev_id, None)
if ev:
to_return[ev_id] = ev
else:
db.append(ev_id)
if db:
evs = yield self.get_events(
ev_ids, get_prev_content=False, check_redacted=False,
)
to_return.update(evs)
defer.returnValue(to_return)
current_state = yield resolve_events(
state_sets,
state_map_factory=get_events,
)
else:
return
existing_state = yield self.get_current_state_ids(room_id)
existing_events = set(existing_state.itervalues())
new_events = set(ev_id for ev_id in current_state.itervalues())
changed_events = existing_events ^ new_events
if not changed_events:
return
to_delete = {
key: ev_id for key, ev_id in existing_state.iteritems()
if ev_id in changed_events
}
events_to_insert = (new_events - existing_events)
to_insert = {
key: ev_id for key, ev_id in current_state.iteritems()
if ev_id in events_to_insert
}
defer.returnValue((to_delete, to_insert, current_state))
@defer.inlineCallbacks
def get_event(self, event_id, check_redacted=True,
get_prev_content=False, allow_rejected=False,
allow_none=False):
"""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 an exception.
Returns:
Deferred : A FrozenEvent.
"""
events = yield self._get_events(
[event_id],
check_redacted=check_redacted,
get_prev_content=get_prev_content,
allow_rejected=allow_rejected,
)
if not events and not allow_none:
raise SynapseError(404, "Could not find event %s" % (event_id,))
defer.returnValue(events[0] if events else None)
@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(
event_ids,
check_redacted=check_redacted,
get_prev_content=get_prev_content,
allow_rejected=allow_rejected,
)
defer.returnValue({e.event_id: e for e in events})
@log_function
def _persist_events_txn(self, txn, events_and_contexts, backfilled,
delete_existing=False, current_state_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.
current_state_for_room (dict[str, (list[str], list[str])]):
The current-state delta for each room. For each room, a tuple
(to_delete, to_insert), being a list of event ids to be removed
from the current state, and a list of event ids 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.
"""
max_stream_order = events_and_contexts[-1][0].internal_metadata.stream_ordering
self._update_current_state_txn(txn, current_state_for_room, max_stream_order)
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 the state_groups, state_groups_state, and
# event_to_state_groups tables.
self._store_mult_state_groups_txn(txn, events_and_contexts)
# _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,
backfilled=backfilled,
)
def _update_current_state_txn(self, txn, state_delta_by_room, max_stream_order):
for room_id, current_state_tuple in state_delta_by_room.iteritems():
to_delete, to_insert, _ = current_state_tuple
txn.executemany(
"DELETE FROM current_state_events WHERE event_id = ?",
[(ev_id,) for ev_id in to_delete.itervalues()],
)
self._simple_insert_many_txn(
txn,
table="current_state_events",
values=[
{
"event_id": ev_id,
"room_id": room_id,
"type": key[0],
"state_key": key[1],
}
for key, ev_id in to_insert.iteritems()
],
)
state_deltas = {key: None for key in to_delete}
state_deltas.update(to_insert)
self._simple_insert_many_txn(
txn,
table="current_state_delta_stream",
values=[
{
"stream_id": max_stream_order,
"room_id": room_id,
"type": key[0],
"state_key": key[1],
"event_id": ev_id,
"prev_event_id": to_delete.get(key, None),
}
for key, ev_id in state_deltas.iteritems()
]
)
self._curr_state_delta_stream_cache.entity_has_changed(
room_id, max_stream_order,
)
# 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 state_deltas
if ev_type == EventTypes.Member
)
for member in members_changed:
self._invalidate_cache_and_stream(
txn, self.get_rooms_for_user, (member,)
)
for host in set(get_domain_from_id(u) for u in members_changed):
self._invalidate_cache_and_stream(
txn, self.is_host_joined, (room_id, host)
)
self._invalidate_cache_and_stream(
txn, self.get_users_in_room, (room_id,)
)
self._invalidate_cache_and_stream(
txn, self.get_current_state_ids, (room_id,)
)
def _update_forward_extremities_txn(self, txn, new_forward_extremities,
max_stream_order):
for room_id, new_extrem in new_forward_extremities.iteritems():
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 new_forward_extremities.iteritems()
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 new_forward_extremities.iteritems()
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 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 depth_updates.iteritems():
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 the state_group, state_groups_state and
# event_to_state_groups tables.
try:
self._store_mult_state_groups_txn(txn, ((event, context),))
except Exception:
logger.exception("")
raise
metadata_json = encode_json(
event.internal_metadata.get_dict()
).decode("UTF-8")
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_content_hashes",
"event_destinations",
"event_edge_hashes",
"event_edges",
"event_forward_extremities",
"event_push_actions",
"event_reference_hashes",