Merge pull request #1047 from matrix-org/erikj/state_ids

Avoid pulling the full state of a room out so often.

1 files changed, 84 insertions, 51 deletions

diff --git a/synapse/state.py b/synapse/state.py
index ef1bc470be..78461215ca 100644
--- a/synapse/state.py
+++ b/synapse/state.py
@@ -93,8 +93,30 @@ class StateHandler(object):
         if not latest_event_ids:
             latest_event_ids = yield self.store.get_latest_event_ids_in_room(room_id)
 
-        res = yield self.resolve_state_groups(room_id, latest_event_ids)
-        state = res[1]
+        _, state = yield self.resolve_state_groups(room_id, latest_event_ids)
+
+        if event_type:
+            event_id = state.get((event_type, state_key))
+            event = None
+            if event_id:
+                event = yield self.store.get_event(event_id, allow_none=True)
+            defer.returnValue(event)
+            return
+
+        state_map = yield self.store.get_events(state.values(), get_prev_content=False)
+        state = {
+            key: state_map[e_id] for key, e_id in state.items() if e_id in state_map
+        }
+
+        defer.returnValue(state)
+
+    @defer.inlineCallbacks
+    def get_current_state_ids(self, room_id, event_type=None, state_key="",
+                              latest_event_ids=None):
+        if not latest_event_ids:
+            latest_event_ids = yield self.store.get_latest_event_ids_in_room(room_id)
+
+        _, state = yield self.resolve_state_groups(room_id, latest_event_ids)
 
         if event_type:
             defer.returnValue(state.get((event_type, state_key)))
@@ -123,27 +145,27 @@ class StateHandler(object):
             # state. Certainly store.get_current_state won't return any, and
             # persisting the event won't store the state group.
             if old_state:
-                context.current_state = {
-                    (s.type, s.state_key): s for s in old_state
+                context.current_state_ids = {
+                    (s.type, s.state_key): s.event_id for s in old_state
                 }
             else:
-                context.current_state = {}
+                context.current_state_ids = {}
             context.prev_state_events = []
             context.state_group = None
             defer.returnValue(context)
 
         if old_state:
-            context.current_state = {
-                (s.type, s.state_key): s for s in old_state
+            context.current_state_ids = {
+                (s.type, s.state_key): s.event_id for s in old_state
             }
             context.state_group = None
 
             if event.is_state():
                 key = (event.type, event.state_key)
-                if key in context.current_state:
-                    replaces = context.current_state[key]
-                    if replaces.event_id != event.event_id:  # Paranoia check
-                        event.unsigned["replaces_state"] = replaces.event_id
+                if key in context.current_state_ids:
+                    replaces = context.current_state_ids[key]
+                    if replaces != event.event_id:  # Paranoia check
+                        event.unsigned["replaces_state"] = replaces
 
             context.prev_state_events = []
             defer.returnValue(context)
@@ -159,18 +181,18 @@ class StateHandler(object):
                 event.room_id, [e for e, _ in event.prev_events],
             )
 
-        group, curr_state, prev_state = ret
+        group, curr_state = ret
 
-        context.current_state = curr_state
+        context.current_state_ids = curr_state
         context.state_group = group if not event.is_state() else None
 
         if event.is_state():
             key = (event.type, event.state_key)
-            if key in context.current_state:
-                replaces = context.current_state[key]
-                event.unsigned["replaces_state"] = replaces.event_id
+            if key in context.current_state_ids:
+                replaces = context.current_state_ids[key]
+                event.unsigned["replaces_state"] = replaces
 
-        context.prev_state_events = prev_state
+        context.prev_state_events = []
         defer.returnValue(context)
 
     @defer.inlineCallbacks
@@ -187,72 +209,83 @@ class StateHandler(object):
         """
         logger.debug("resolve_state_groups event_ids %s", event_ids)
 
-        state_groups = yield self.store.get_state_groups(
+        state_groups_ids = yield self.store.get_state_groups_ids(
             room_id, event_ids
         )
 
         logger.debug(
             "resolve_state_groups state_groups %s",
-            state_groups.keys()
+            state_groups_ids.keys()
         )
 
-        group_names = frozenset(state_groups.keys())
+        group_names = frozenset(state_groups_ids.keys())
         if len(group_names) == 1:
-            name, state_list = state_groups.items().pop()
-            state = {
-                (e.type, e.state_key): e
-                for e in state_list
-            }
-            prev_state = state.get((event_type, state_key), None)
-            if prev_state:
-                prev_state = prev_state.event_id
-                prev_states = [prev_state]
-            else:
-                prev_states = []
+            name, state_list = state_groups_ids.items().pop()
 
-            defer.returnValue((name, state, prev_states))
+            defer.returnValue((name, state_list,))
 
         if self._state_cache is not None:
             cache = self._state_cache.get(group_names, None)
             if cache:
                 cache.ts = self.clock.time_msec()
 
-                event_dict = yield self.store.get_events(cache.state.values())
-                state = {(e.type, e.state_key): e for e in event_dict.values()}
-
-                prev_state = state.get((event_type, state_key), None)
-                if prev_state:
-                    prev_state = prev_state.event_id
-                    prev_states = [prev_state]
-                else:
-                    prev_states = []
                 defer.returnValue(
-                    (cache.state_group, state, prev_states)
+                    (cache.state_group, cache.state,)
                 )
 
-        logger.info("Resolving state for %s with %d groups", room_id, len(state_groups))
-
-        new_state, prev_states = self._resolve_events(
-            state_groups.values(), event_type, state_key
+        logger.info(
+            "Resolving state for %s with %d groups", room_id, len(state_groups_ids)
         )
 
+        state = {}
+        for st in state_groups_ids.values():
+            for key, e_id in st.items():
+                state.setdefault(key, set()).add(e_id)
+
+        conflicted_state = {
+            k: list(v)
+            for k, v in state.items()
+            if len(v) > 1
+        }
+
+        if conflicted_state:
+            logger.info("Resolving conflicted state for %r", room_id)
+            state_map = yield self.store.get_events(
+                [e_id for st in state_groups_ids.values() for e_id in st.values()],
+                get_prev_content=False
+            )
+            state_sets = [
+                [state_map[e_id] for key, e_id in st.items() if e_id in state_map]
+                for st in state_groups_ids.values()
+            ]
+            new_state, _ = self._resolve_events(
+                state_sets, event_type, state_key
+            )
+            new_state = {
+                key: e.event_id for key, e in new_state.items()
+            }
+        else:
+            new_state = {
+                key: e_ids.pop() for key, e_ids in state.items()
+            }
+
         state_group = None
-        new_state_event_ids = frozenset(e.event_id for e in new_state.values())
-        for sg, events in state_groups.items():
-            if new_state_event_ids == frozenset(e.event_id for e in events):
+        new_state_event_ids = frozenset(new_state.values())
+        for sg, events in state_groups_ids.items():
+            if new_state_event_ids == frozenset(e_id for e_id in events):
                 state_group = sg
                 break
 
         if self._state_cache is not None:
             cache = _StateCacheEntry(
-                state={key: event.event_id for key, event in new_state.items()},
+                state=new_state,
                 state_group=state_group,
                 ts=self.clock.time_msec()
             )
 
             self._state_cache[group_names] = cache
 
-        defer.returnValue((state_group, new_state, prev_states))
+        defer.returnValue((state_group, new_state,))
 
     def resolve_events(self, state_sets, event):
         logger.info(