Merge branch 'develop' of github.com:matrix-org/synapse into erikj-perf

1 files changed, 132 insertions, 15 deletions

diff --git a/synapse/storage/stream.py b/synapse/storage/stream.py
index 8ac2adab05..09bc522210 100644
--- a/synapse/storage/stream.py
+++ b/synapse/storage/stream.py
@@ -36,6 +36,7 @@ what sort order was used:
 from twisted.internet import defer
 
 from ._base import SQLBaseStore
+from synapse.api.constants import EventTypes
 from synapse.api.errors import SynapseError
 from synapse.util.logutils import log_function
 
@@ -82,10 +83,10 @@ class _StreamToken(namedtuple("_StreamToken", "topological stream")):
     def parse(cls, string):
         try:
             if string[0] == 's':
-                return cls(None, int(string[1:]))
+                return cls(topological=None, stream=int(string[1:]))
             if string[0] == 't':
                 parts = string[1:].split('-', 1)
-                return cls(int(parts[1]), int(parts[0]))
+                return cls(topological=int(parts[0]), stream=int(parts[1]))
         except:
             pass
         raise SynapseError(400, "Invalid token %r" % (string,))
@@ -94,7 +95,7 @@ class _StreamToken(namedtuple("_StreamToken", "topological stream")):
     def parse_stream_token(cls, string):
         try:
             if string[0] == 's':
-                return cls(None, int(string[1:]))
+                return cls(topological=None, stream=int(string[1:]))
         except:
             pass
         raise SynapseError(400, "Invalid token %r" % (string,))
@@ -127,6 +128,85 @@ class _StreamToken(namedtuple("_StreamToken", "topological stream")):
 
 
 class StreamStore(SQLBaseStore):
+
+    @defer.inlineCallbacks
+    def get_appservice_room_stream(self, service, from_key, to_key, limit=0):
+        # NB this lives here instead of appservice.py so we can reuse the
+        # 'private' StreamToken class in this file.
+        if limit:
+            limit = max(limit, MAX_STREAM_SIZE)
+        else:
+            limit = MAX_STREAM_SIZE
+
+        # From and to keys should be integers from ordering.
+        from_id = _StreamToken.parse_stream_token(from_key)
+        to_id = _StreamToken.parse_stream_token(to_key)
+
+        if from_key == to_key:
+            defer.returnValue(([], to_key))
+            return
+
+        # select all the events between from/to with a sensible limit
+        sql = (
+            "SELECT e.event_id, e.room_id, e.type, s.state_key, "
+            "e.stream_ordering FROM events AS e LEFT JOIN state_events as s ON "
+            "e.event_id = s.event_id "
+            "WHERE e.stream_ordering > ? AND e.stream_ordering <= ? "
+            "ORDER BY stream_ordering ASC LIMIT %(limit)d "
+        ) % {
+            "limit": limit
+        }
+
+        def f(txn):
+            # pull out all the events between the tokens
+            txn.execute(sql, (from_id.stream, to_id.stream,))
+            rows = self.cursor_to_dict(txn)
+
+            # Logic:
+            #  - We want ALL events which match the AS room_id regex
+            #  - We want ALL events which match the rooms represented by the AS
+            #    room_alias regex
+            #  - We want ALL events for rooms that AS users have joined.
+            # This is currently supported via get_app_service_rooms (which is
+            # used for the Notifier listener rooms). We can't reasonably make a
+            # SQL query for these room IDs, so we'll pull all the events between
+            # from/to and filter in python.
+            rooms_for_as = self._get_app_service_rooms_txn(txn, service)
+            room_ids_for_as = [r.room_id for r in rooms_for_as]
+
+            def app_service_interested(row):
+                if row["room_id"] in room_ids_for_as:
+                    return True
+
+                if row["type"] == EventTypes.Member:
+                    if service.is_interested_in_user(row.get("state_key")):
+                        return True
+                return False
+
+            ret = self._get_events_txn(
+                txn,
+                # apply the filter on the room id list
+                [
+                    r["event_id"] for r in rows
+                    if app_service_interested(r)
+                ],
+                get_prev_content=True
+            )
+
+            self._set_before_and_after(ret, rows)
+
+            if rows:
+                key = "s%d" % max(r["stream_ordering"] for r in rows)
+            else:
+                # Assume we didn't get anything because there was nothing to
+                # get.
+                key = to_key
+
+            return ret, key
+
+        results = yield self.runInteraction("get_appservice_room_stream", f)
+        defer.returnValue(results)
+
     @log_function
     def get_room_events_stream(self, user_id, from_key, to_key, room_id,
                                limit=0, with_feedback=False):
@@ -181,8 +261,10 @@ class StreamStore(SQLBaseStore):
                 get_prev_content=True
             )
 
+            self._set_before_and_after(ret, rows)
+
             if rows:
-                key = "s%d" % max([r["stream_ordering"] for r in rows])
+                key = "s%d" % max(r["stream_ordering"] for r in rows)
             else:
                 # Assume we didn't get anything because there was nothing to
                 # get.
@@ -260,22 +342,44 @@ class StreamStore(SQLBaseStore):
                 get_prev_content=True
             )
 
+            self._set_before_and_after(events, rows)
+
             return events, next_token,
 
         return self.runInteraction("paginate_room_events", f)
 
     def get_recent_events_for_room(self, room_id, limit, end_token,
-                                   with_feedback=False):
+                                   with_feedback=False, from_token=None):
         # TODO (erikj): Handle compressed feedback
 
-        sql = (
-            "SELECT stream_ordering, topological_ordering, event_id FROM events "
-            "WHERE room_id = ? AND stream_ordering <= ? AND outlier = 0 "
-            "ORDER BY topological_ordering DESC, stream_ordering DESC LIMIT ? "
-        )
+        end_token = _StreamToken.parse_stream_token(end_token)
 
-        def f(txn):
-            txn.execute(sql, (room_id, end_token, limit,))
+        if from_token is None:
+            sql = (
+                "SELECT stream_ordering, topological_ordering, event_id"
+                " FROM events"
+                " WHERE room_id = ? AND stream_ordering <= ? AND outlier = 0"
+                " ORDER BY topological_ordering DESC, stream_ordering DESC"
+                " LIMIT ?"
+            )
+        else:
+            from_token = _StreamToken.parse_stream_token(from_token)
+            sql = (
+                "SELECT stream_ordering, topological_ordering, event_id"
+                " FROM events"
+                " WHERE room_id = ? AND stream_ordering > ?"
+                " AND stream_ordering <= ? AND outlier = 0"
+                " ORDER BY topological_ordering DESC, stream_ordering DESC"
+                " LIMIT ?"
+            )
+
+        def get_recent_events_for_room_txn(txn):
+            if from_token is None:
+                txn.execute(sql, (room_id, end_token.stream, limit,))
+            else:
+                txn.execute(sql, (
+                    room_id, from_token.stream, end_token.stream, limit
+                ))
 
             rows = self.cursor_to_dict(txn)
 
@@ -291,9 +395,9 @@ class StreamStore(SQLBaseStore):
                 toke = rows[0]["stream_ordering"] - 1
                 start_token = str(_StreamToken(topo, toke))
 
-                token = (start_token, end_token)
+                token = (start_token, str(end_token))
             else:
-                token = (end_token, end_token)
+                token = (str(end_token), str(end_token))
 
             events = self._get_events_txn(
                 txn,
@@ -301,9 +405,13 @@ class StreamStore(SQLBaseStore):
                 get_prev_content=True
             )
 
+            self._set_before_and_after(events, rows)
+
             return events, token
 
-        return self.runInteraction("get_recent_events_for_room", f)
+        return self.runInteraction(
+            "get_recent_events_for_room", get_recent_events_for_room_txn
+        )
 
     def get_room_events_max_id(self):
         return self.runInteraction(
@@ -325,3 +433,12 @@ class StreamStore(SQLBaseStore):
 
         key = res[0]["m"]
         return "s%d" % (key,)
+
+    @staticmethod
+    def _set_before_and_after(events, rows):
+        for event, row in zip(events, rows):
+            stream = row["stream_ordering"]
+            topo = event.depth
+            internal = event.internal_metadata
+            internal.before = str(_StreamToken(topo, stream - 1))
+            internal.after = str(_StreamToken(topo, stream))