Merge pull request #3541 from matrix-org/rav/optimize_filter_events_for_server

Refactor and optimze filter_events_for_server
This commit is contained in:
Richard van der Hoff 2018-07-17 14:01:39 +01:00 committed by GitHub
commit 9c04b4abf9
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
4 changed files with 409 additions and 138 deletions

View file

@ -43,7 +43,6 @@ from synapse.crypto.event_signing import (
add_hashes_and_signatures,
compute_event_signature,
)
from synapse.events.utils import prune_event
from synapse.events.validator import EventValidator
from synapse.state import resolve_events_with_factory
from synapse.types import UserID, get_domain_from_id
@ -52,8 +51,8 @@ from synapse.util.async import Linearizer
from synapse.util.distributor import user_joined_room
from synapse.util.frozenutils import unfreeze
from synapse.util.logutils import log_function
from synapse.util.metrics import measure_func
from synapse.util.retryutils import NotRetryingDestination
from synapse.visibility import filter_events_for_server
from ._base import BaseHandler
@ -501,137 +500,6 @@ class FederationHandler(BaseHandler):
user = UserID.from_string(event.state_key)
yield user_joined_room(self.distributor, user, event.room_id)
@measure_func("_filter_events_for_server")
@defer.inlineCallbacks
def _filter_events_for_server(self, server_name, room_id, events):
"""Filter the given events for the given server, redacting those the
server can't see.
Assumes the server is currently in the room.
Returns
list[FrozenEvent]
"""
# First 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 self.store.get_state_ids_for_events(
frozenset(e.event_id for e in events),
types=(
(EventTypes.RoomHistoryVisibility, ""),
)
)
visibility_ids = set()
for sids in event_to_state_ids.itervalues():
hist = sids.get((EventTypes.RoomHistoryVisibility, ""))
if hist:
visibility_ids.add(hist)
# If we failed to find any history visibility events then the default
# is "shared" visiblity.
if not visibility_ids:
defer.returnValue(events)
event_map = yield self.store.get_events(visibility_ids)
all_open = all(
e.content.get("history_visibility") in (None, "shared", "world_readable")
for e in event_map.itervalues()
)
if all_open:
defer.returnValue(events)
# Ok, so we're dealing with events that have non-trivial visibility
# rules, so we need to also get the memberships of the room.
event_to_state_ids = yield self.store.get_state_ids_for_events(
frozenset(e.event_id for e in events),
types=(
(EventTypes.RoomHistoryVisibility, ""),
(EventTypes.Member, None),
)
)
# We only want to pull out member events that correspond to the
# server's domain.
def check_match(id):
try:
return server_name == get_domain_from_id(id)
except Exception:
return False
# Parses mapping `event_id -> (type, state_key) -> state event_id`
# to get all state ids that we're interested in.
event_map = yield self.store.get_events([
e_id
for key_to_eid in list(event_to_state_ids.values())
for key, e_id in key_to_eid.items()
if key[0] != EventTypes.Member or check_match(key[1])
])
event_to_state = {
e_id: {
key: event_map[inner_e_id]
for key, inner_e_id in key_to_eid.iteritems()
if inner_e_id in event_map
}
for e_id, key_to_eid in event_to_state_ids.iteritems()
}
erased_senders = yield self.store.are_users_erased(
e.sender for e in events,
)
def redact_disallowed(event, state):
# if the sender has been gdpr17ed, always return a redacted
# copy of the event.
if erased_senders[event.sender]:
logger.info(
"Sender of %s has been erased, redacting",
event.event_id,
)
return prune_event(event)
if not state:
return event
history = state.get((EventTypes.RoomHistoryVisibility, ''), None)
if history:
visibility = history.content.get("history_visibility", "shared")
if visibility in ["invited", "joined"]:
# We now loop through all state events looking for
# membership states for the requesting server to determine
# if the server is either in the room or has been invited
# into the room.
for ev in state.itervalues():
if ev.type != EventTypes.Member:
continue
try:
domain = get_domain_from_id(ev.state_key)
except Exception:
continue
if domain != server_name:
continue
memtype = ev.membership
if memtype == Membership.JOIN:
return event
elif memtype == Membership.INVITE:
if visibility == "invited":
return event
else:
return prune_event(event)
return event
defer.returnValue([
redact_disallowed(e, event_to_state[e.event_id])
for e in events
])
@log_function
@defer.inlineCallbacks
def backfill(self, dest, room_id, limit, extremities):
@ -1558,7 +1426,7 @@ class FederationHandler(BaseHandler):
limit
)
events = yield self._filter_events_for_server(origin, room_id, events)
events = yield filter_events_for_server(self.store, origin, events)
defer.returnValue(events)
@ -1605,8 +1473,8 @@ class FederationHandler(BaseHandler):
if not in_room:
raise AuthError(403, "Host not in room.")
events = yield self._filter_events_for_server(
origin, event.room_id, [event]
events = yield filter_events_for_server(
self.store, origin, [event],
)
event = events[0]
defer.returnValue(event)
@ -1896,8 +1764,8 @@ class FederationHandler(BaseHandler):
min_depth=min_depth,
)
missing_events = yield self._filter_events_for_server(
origin, room_id, missing_events,
missing_events = yield filter_events_for_server(
self.store, origin, missing_events,
)
defer.returnValue(missing_events)