Merge branch 'release-v0.9.1' of github.com:matrix-org/synapse v0.9.1

1 files changed, 230 insertions, 285 deletions

diff --git a/synapse/notifier.py b/synapse/notifier.py
index 78eb28e4b2..078abfc56d 100644
--- a/synapse/notifier.py
+++ b/synapse/notifier.py
@@ -16,7 +16,7 @@
 from twisted.internet import defer
 
 from synapse.util.logutils import log_function
-from synapse.util.logcontext import PreserveLoggingContext
+from synapse.util.async import run_on_reactor
 from synapse.types import StreamToken
 import synapse.metrics
 
@@ -42,63 +42,78 @@ def count(func, l):
 
 class _NotificationListener(object):
     """ This represents a single client connection to the events stream.
-
     The events stream handler will have yielded to the deferred, so to
     notify the handler it is sufficient to resolve the deferred.
-
-    This listener will also keep track of which rooms it is listening in
-    so that it can remove itself from the indexes in the Notifier class.
     """
 
-    def __init__(self, user, rooms, from_token, limit, timeout, deferred,
-                 appservice=None):
-        self.user = user
-        self.appservice = appservice
-        self.from_token = from_token
-        self.limit = limit
-        self.timeout = timeout
+    def __init__(self, deferred):
         self.deferred = deferred
-        self.rooms = rooms
-        self.timer = None
 
     def notified(self):
         return self.deferred.called
 
-    def notify(self, notifier, events, start_token, end_token):
-        """ Inform whoever is listening about the new events. This will
-        also remove this listener from all the indexes in the Notifier
-        it knows about.
+    def notify(self, token):
+        """ Inform whoever is listening about the new events.
         """
-
-        result = (events, (start_token, end_token))
-
         try:
-            self.deferred.callback(result)
-            notified_events_counter.inc_by(len(events))
+            self.deferred.callback(token)
         except defer.AlreadyCalledError:
             pass
 
-        # Should the following be done be using intrusively linked lists?
-        # -- erikj
+
+class _NotifierUserStream(object):
+    """This represents a user connected to the event stream.
+    It tracks the most recent stream token for that user.
+    At a given point a user may have a number of streams listening for
+    events.
+
+    This listener will also keep track of which rooms it is listening in
+    so that it can remove itself from the indexes in the Notifier class.
+    """
+
+    def __init__(self, user, rooms, current_token, time_now_ms,
+                 appservice=None):
+        self.user = str(user)
+        self.appservice = appservice
+        self.listeners = set()
+        self.rooms = set(rooms)
+        self.current_token = current_token
+        self.last_notified_ms = time_now_ms
+
+    def notify(self, stream_key, stream_id, time_now_ms):
+        """Notify any listeners for this user of a new event from an
+        event source.
+        Args:
+            stream_key(str): The stream the event came from.
+            stream_id(str): The new id for the stream the event came from.
+            time_now_ms(int): The current time in milliseconds.
+        """
+        self.current_token = self.current_token.copy_and_advance(
+            stream_key, stream_id
+        )
+        if self.listeners:
+            self.last_notified_ms = time_now_ms
+            listeners = self.listeners
+            self.listeners = set()
+            for listener in listeners:
+                listener.notify(self.current_token)
+
+    def remove(self, notifier):
+        """ Remove this listener from all the indexes in the Notifier
+        it knows about.
+        """
 
         for room in self.rooms:
-            lst = notifier.room_to_listeners.get(room, set())
+            lst = notifier.room_to_user_streams.get(room, set())
             lst.discard(self)
 
-        notifier.user_to_listeners.get(self.user, set()).discard(self)
+        notifier.user_to_user_stream.pop(self.user)
 
         if self.appservice:
-            notifier.appservice_to_listeners.get(
+            notifier.appservice_to_user_streams.get(
                 self.appservice, set()
             ).discard(self)
 
-        # Cancel the timeout for this notifer if one exists.
-        if self.timer is not None:
-            try:
-                notifier.clock.cancel_call_later(self.timer)
-            except:
-                logger.warn("Failed to cancel notifier timer")
-
 
 class Notifier(object):
     """ This class is responsible for notifying any listeners when there are
@@ -107,14 +122,18 @@ class Notifier(object):
     Primarily used from the /events stream.
     """
 
+    UNUSED_STREAM_EXPIRY_MS = 10 * 60 * 1000
+
     def __init__(self, hs):
         self.hs = hs
 
-        self.room_to_listeners = {}
-        self.user_to_listeners = {}
-        self.appservice_to_listeners = {}
+        self.user_to_user_stream = {}
+        self.room_to_user_streams = {}
+        self.appservice_to_user_streams = {}
 
         self.event_sources = hs.get_event_sources()
+        self.store = hs.get_datastore()
+        self.pending_new_room_events = []
 
         self.clock = hs.get_clock()
 
@@ -122,45 +141,80 @@ class Notifier(object):
             "user_joined_room", self._user_joined_room
         )
 
+        self.clock.looping_call(
+            self.remove_expired_streams, self.UNUSED_STREAM_EXPIRY_MS
+        )
+
         # This is not a very cheap test to perform, but it's only executed
         # when rendering the metrics page, which is likely once per minute at
         # most when scraping it.
         def count_listeners():
-            all_listeners = set()
+            all_user_streams = set()
 
-            for x in self.room_to_listeners.values():
-                all_listeners |= x
-            for x in self.user_to_listeners.values():
-                all_listeners |= x
-            for x in self.appservice_to_listeners.values():
-                all_listeners |= x
+            for x in self.room_to_user_streams.values():
+                all_user_streams |= x
+            for x in self.user_to_user_stream.values():
+                all_user_streams.add(x)
+            for x in self.appservice_to_user_streams.values():
+                all_user_streams |= x
 
-            return len(all_listeners)
+            return sum(len(stream.listeners) for stream in all_user_streams)
         metrics.register_callback("listeners", count_listeners)
 
         metrics.register_callback(
             "rooms",
-            lambda: count(bool, self.room_to_listeners.values()),
+            lambda: count(bool, self.room_to_user_streams.values()),
         )
         metrics.register_callback(
             "users",
-            lambda: count(bool, self.user_to_listeners.values()),
+            lambda: len(self.user_to_user_stream),
         )
         metrics.register_callback(
             "appservices",
-            lambda: count(bool, self.appservice_to_listeners.values()),
+            lambda: count(bool, self.appservice_to_user_streams.values()),
         )
 
     @log_function
     @defer.inlineCallbacks
-    def on_new_room_event(self, event, extra_users=[]):
+    def on_new_room_event(self, event, room_stream_id, max_room_stream_id,
+                          extra_users=[]):
         """ Used by handlers to inform the notifier something has happened
         in the room, room event wise.
 
         This triggers the notifier to wake up any listeners that are
         listening to the room, and any listeners for the users in the
         `extra_users` param.
+
+        The events can be peristed out of order. The notifier will wait
+        until all previous events have been persisted before notifying
+        the client streams.
+        """
+        yield run_on_reactor()
+
+        self.pending_new_room_events.append((
+            room_stream_id, event, extra_users
+        ))
+        self._notify_pending_new_room_events(max_room_stream_id)
+
+    def _notify_pending_new_room_events(self, max_room_stream_id):
+        """Notify for the room events that were queued waiting for a previous
+        event to be persisted.
+        Args:
+            max_room_stream_id(int): The highest stream_id below which all
+                events have been persisted.
         """
+        pending = self.pending_new_room_events
+        self.pending_new_room_events = []
+        for room_stream_id, event, extra_users in pending:
+            if room_stream_id > max_room_stream_id:
+                self.pending_new_room_events.append((
+                    room_stream_id, event, extra_users
+                ))
+            else:
+                self._on_new_room_event(event, room_stream_id, extra_users)
+
+    def _on_new_room_event(self, event, room_stream_id, extra_users=[]):
+        """Notify any user streams that are interested in this room event"""
         # poke any interested application service.
         self.hs.get_handlers().appservice_handler.notify_interested_services(
             event
@@ -168,192 +222,129 @@ class Notifier(object):
 
         room_id = event.room_id
 
-        room_source = self.event_sources.sources["room"]
-
-        room_listeners = self.room_to_listeners.get(room_id, set())
-
-        _discard_if_notified(room_listeners)
+        room_user_streams = self.room_to_user_streams.get(room_id, set())
 
-        listeners = room_listeners.copy()
+        user_streams = room_user_streams.copy()
 
         for user in extra_users:
-            user_listeners = self.user_to_listeners.get(user, set())
+            user_stream = self.user_to_user_stream.get(str(user))
+            if user_stream is not None:
+                user_streams.add(user_stream)
 
-            _discard_if_notified(user_listeners)
-
-            listeners |= user_listeners
-
-        for appservice in self.appservice_to_listeners:
+        for appservice in self.appservice_to_user_streams:
             # TODO (kegan): Redundant appservice listener checks?
-            # App services will already be in the room_to_listeners set, but
+            # App services will already be in the room_to_user_streams set, but
             # that isn't enough. They need to be checked here in order to
             # receive *invites* for users they are interested in. Does this
-            # make the room_to_listeners check somewhat obselete?
+            # make the room_to_user_streams check somewhat obselete?
             if appservice.is_interested(event):
-                app_listeners = self.appservice_to_listeners.get(
+                app_user_streams = self.appservice_to_user_streams.get(
                     appservice, set()
                 )
+                user_streams |= app_user_streams
 
-                _discard_if_notified(app_listeners)
-
-                listeners |= app_listeners
-
-        logger.debug("on_new_room_event listeners %s", listeners)
-
-        # TODO (erikj): Can we make this more efficient by hitting the
-        # db once?
-
-        @defer.inlineCallbacks
-        def notify(listener):
-            events, end_key = yield room_source.get_new_events_for_user(
-                listener.user,
-                listener.from_token.room_key,
-                listener.limit,
-            )
-
-            if events:
-                end_token = listener.from_token.copy_and_replace(
-                    "room_key", end_key
-                )
+        logger.debug("on_new_room_event listeners %s", user_streams)
 
-                listener.notify(
-                    self, events, listener.from_token, end_token
+        time_now_ms = self.clock.time_msec()
+        for user_stream in user_streams:
+            try:
+                user_stream.notify(
+                    "room_key", "s%d" % (room_stream_id,), time_now_ms
                 )
-
-        def eb(failure):
-            logger.exception("Failed to notify listener", failure)
-
-        with PreserveLoggingContext():
-            yield defer.DeferredList(
-                [notify(l).addErrback(eb) for l in listeners],
-                consumeErrors=True,
-            )
+            except:
+                logger.exception("Failed to notify listener")
 
     @defer.inlineCallbacks
     @log_function
-    def on_new_user_event(self, users=[], rooms=[]):
+    def on_new_user_event(self, stream_key, new_token, users=[], rooms=[]):
         """ Used to inform listeners that something has happend
         presence/user event wise.
 
         Will wake up all listeners for the given users and rooms.
         """
-        # TODO(paul): This is horrible, having to manually list every event
-        # source here individually
-        presence_source = self.event_sources.sources["presence"]
-        typing_source = self.event_sources.sources["typing"]
-
-        listeners = set()
+        yield run_on_reactor()
+        user_streams = set()
 
         for user in users:
-            user_listeners = self.user_to_listeners.get(user, set())
-
-            _discard_if_notified(user_listeners)
-
-            listeners |= user_listeners
+            user_stream = self.user_to_user_stream.get(str(user))
+            if user_stream is not None:
+                user_streams.add(user_stream)
 
         for room in rooms:
-            room_listeners = self.room_to_listeners.get(room, set())
-
-            _discard_if_notified(room_listeners)
-
-            listeners |= room_listeners
-
-        @defer.inlineCallbacks
-        def notify(listener):
-            presence_events, presence_end_key = (
-                yield presence_source.get_new_events_for_user(
-                    listener.user,
-                    listener.from_token.presence_key,
-                    listener.limit,
-                )
-            )
-            typing_events, typing_end_key = (
-                yield typing_source.get_new_events_for_user(
-                    listener.user,
-                    listener.from_token.typing_key,
-                    listener.limit,
-                )
-            )
-
-            if presence_events or typing_events:
-                end_token = listener.from_token.copy_and_replace(
-                    "presence_key", presence_end_key
-                ).copy_and_replace(
-                    "typing_key", typing_end_key
-                )
+            user_streams |= self.room_to_user_streams.get(room, set())
 
-                listener.notify(
-                    self,
-                    presence_events + typing_events,
-                    listener.from_token,
-                    end_token
-                )
-
-        def eb(failure):
-            logger.error(
-                "Failed to notify listener",
-                exc_info=(
-                    failure.type,
-                    failure.value,
-                    failure.getTracebackObject())
-            )
-
-        with PreserveLoggingContext():
-            yield defer.DeferredList(
-                [notify(l).addErrback(eb) for l in listeners],
-                consumeErrors=True,
-            )
+        time_now_ms = self.clock.time_msec()
+        for user_stream in user_streams:
+            try:
+                user_stream.notify(stream_key, new_token, time_now_ms)
+            except:
+                logger.exception("Failed to notify listener")
 
     @defer.inlineCallbacks
-    def wait_for_events(self, user, rooms, filter, timeout, callback):
+    def wait_for_events(self, user, rooms, timeout, callback,
+                        from_token=StreamToken("s0", "0", "0")):
         """Wait until the callback returns a non empty response or the
         timeout fires.
         """
 
         deferred = defer.Deferred()
+        time_now_ms = self.clock.time_msec()
+
+        user = str(user)
+        user_stream = self.user_to_user_stream.get(user)
+        if user_stream is None:
+            appservice = yield self.store.get_app_service_by_user_id(user)
+            current_token = yield self.event_sources.get_current_token()
+            rooms = yield self.store.get_rooms_for_user(user)
+            rooms = [room.room_id for room in rooms]
+            user_stream = _NotifierUserStream(
+                user=user,
+                rooms=rooms,
+                appservice=appservice,
+                current_token=current_token,
+                time_now_ms=time_now_ms,
+            )
+            self._register_with_keys(user_stream)
+        else:
+            current_token = user_stream.current_token
 
-        from_token = StreamToken("s0", "0", "0")
+        listener = [_NotificationListener(deferred)]
 
-        listener = [_NotificationListener(
-            user=user,
-            rooms=rooms,
-            from_token=from_token,
-            limit=1,
-            timeout=timeout,
-            deferred=deferred,
-        )]
+        if timeout and not current_token.is_after(from_token):
+            user_stream.listeners.add(listener[0])
 
-        if timeout:
-            self._register_with_keys(listener[0])
+        if current_token.is_after(from_token):
+            result = yield callback(from_token, current_token)
+        else:
+            result = None
 
-        result = yield callback()
         timer = [None]
 
+        if result:
+            user_stream.listeners.discard(listener[0])
+            defer.returnValue(result)
+            return
+
         if timeout:
             timed_out = [False]
 
             def _timeout_listener():
                 timed_out[0] = True
                 timer[0] = None
-                listener[0].notify(self, [], from_token, from_token)
+                user_stream.listeners.discard(listener[0])
+                listener[0].notify(current_token)
 
             # We create multiple notification listeners so we have to manage
             # canceling the timeout ourselves.
             timer[0] = self.clock.call_later(timeout/1000., _timeout_listener)
 
             while not result and not timed_out[0]:
-                yield deferred
+                new_token = yield deferred
                 deferred = defer.Deferred()
-                listener[0] = _NotificationListener(
-                    user=user,
-                    rooms=rooms,
-                    from_token=from_token,
-                    limit=1,
-                    timeout=timeout,
-                    deferred=deferred,
-                )
-                self._register_with_keys(listener[0])
-                result = yield callback()
+                listener[0] = _NotificationListener(deferred)
+                user_stream.listeners.add(listener[0])
+                result = yield callback(current_token, new_token)
+                current_token = new_token
 
         if timer[0] is not None:
             try:
@@ -363,125 +354,79 @@ class Notifier(object):
 
         defer.returnValue(result)
 
+    @defer.inlineCallbacks
     def get_events_for(self, user, rooms, pagination_config, timeout):
         """ For the given user and rooms, return any new events for them. If
         there are no new events wait for up to `timeout` milliseconds for any
         new events to happen before returning.
         """
-        deferred = defer.Deferred()
-
-        self._get_events(
-            deferred, user, rooms, pagination_config.from_token,
-            pagination_config.limit, timeout
-        ).addErrback(deferred.errback)
-
-        return deferred
-
-    @defer.inlineCallbacks
-    def _get_events(self, deferred, user, rooms, from_token, limit, timeout):
+        from_token = pagination_config.from_token
         if not from_token:
             from_token = yield self.event_sources.get_current_token()
 
-        appservice = yield self.hs.get_datastore().get_app_service_by_user_id(
-            user.to_string()
-        )
+        limit = pagination_config.limit
 
-        listener = _NotificationListener(
-            user,
-            rooms,
-            from_token,
-            limit,
-            timeout,
-            deferred,
-            appservice=appservice
-        )
+        @defer.inlineCallbacks
+        def check_for_updates(before_token, after_token):
+            events = []
+            end_token = from_token
+            for name, source in self.event_sources.sources.items():
+                keyname = "%s_key" % name
+                before_id = getattr(before_token, keyname)
+                after_id = getattr(after_token, keyname)
+                if before_id == after_id:
+                    continue
+                stuff, new_key = yield source.get_new_events_for_user(
+                    user, getattr(from_token, keyname), limit,
+                )
+                events.extend(stuff)
+                end_token = end_token.copy_and_replace(keyname, new_key)
 
-        def _timeout_listener():
-            # TODO (erikj): We should probably set to_token to the current
-            # max rather than reusing from_token.
-            # Remove the timer from the listener so we don't try to cancel it.
-            listener.timer = None
-            listener.notify(
-                self,
-                [],
-                listener.from_token,
-                listener.from_token,
-            )
+            if events:
+                defer.returnValue((events, (from_token, end_token)))
+            else:
+                defer.returnValue(None)
 
-        if timeout:
-            self._register_with_keys(listener)
+        result = yield self.wait_for_events(
+            user, rooms, timeout, check_for_updates, from_token=from_token
+        )
 
-        yield self._check_for_updates(listener)
+        if result is None:
+            result = ([], (from_token, from_token))
 
-        if not timeout:
-            _timeout_listener()
-        else:
-            # Only add the timer if the listener hasn't been notified
-            if not listener.notified():
-                listener.timer = self.clock.call_later(
-                    timeout/1000.0, _timeout_listener
-                )
-        return
+        defer.returnValue(result)
 
     @log_function
-    def _register_with_keys(self, listener):
-        for room in listener.rooms:
-            s = self.room_to_listeners.setdefault(room, set())
-            s.add(listener)
+    def remove_expired_streams(self):
+        time_now_ms = self.clock.time_msec()
+        expired_streams = []
+        expire_before_ts = time_now_ms - self.UNUSED_STREAM_EXPIRY_MS
+        for stream in self.user_to_user_stream.values():
+            if stream.listeners:
+                continue
+            if stream.last_notified_ms < expire_before_ts:
+                expired_streams.append(stream)
+
+        for expired_stream in expired_streams:
+            expired_stream.remove(self)
 
-        self.user_to_listeners.setdefault(listener.user, set()).add(listener)
-
-        if listener.appservice:
-            self.appservice_to_listeners.setdefault(
-                listener.appservice, set()
-            ).add(listener)
-
-    @defer.inlineCallbacks
     @log_function
-    def _check_for_updates(self, listener):
-        # TODO (erikj): We need to think about limits across multiple sources
-        events = []
+    def _register_with_keys(self, user_stream):
+        self.user_to_user_stream[user_stream.user] = user_stream
 
-        from_token = listener.from_token
-        limit = listener.limit
+        for room in user_stream.rooms:
+            s = self.room_to_user_streams.setdefault(room, set())
+            s.add(user_stream)
 
-        # TODO (erikj): DeferredList?
-        for name, source in self.event_sources.sources.items():
-            keyname = "%s_key" % name
-
-            stuff, new_key = yield source.get_new_events_for_user(
-                listener.user,
-                getattr(from_token, keyname),
-                limit,
-            )
-
-            events.extend(stuff)
-
-            from_token = from_token.copy_and_replace(keyname, new_key)
-
-        end_token = from_token
-
-        if events:
-            listener.notify(self, events, listener.from_token, end_token)
-
-        defer.returnValue(listener)
+        if user_stream.appservice:
+            self.appservice_to_user_stream.setdefault(
+                user_stream.appservice, set()
+            ).add(user_stream)
 
     def _user_joined_room(self, user, room_id):
-        new_listeners = self.user_to_listeners.get(user, set())
-
-        listeners = self.room_to_listeners.setdefault(room_id, set())
-        listeners |= new_listeners
-
-        for l in new_listeners:
-            l.rooms.add(room_id)
-
-
-def _discard_if_notified(listener_set):
-    """Remove any 'stale' listeners from the given set.
-    """
-    to_discard = set()
-    for l in listener_set:
-        if l.notified():
-            to_discard.add(l)
-
-    listener_set -= to_discard
+        user = str(user)
+        new_user_stream = self.user_to_user_stream.get(user)
+        if new_user_stream is not None:
+            room_streams = self.room_to_user_streams.setdefault(room_id, set())
+            room_streams.add(new_user_stream)
+            new_user_stream.rooms.add(room_id)