Add an API for listing threads in a room. (#13394)

Implement the /threads endpoint from MSC3856.

This is currently unstable and behind an experimental configuration
flag.

It includes a background update to backfill data, results from
the /threads endpoint will be partial until that finishes.

synapse/handlers/relations.py

@@ -11,6 +11,7 @@
 # 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.
+import enum
 import logging
 from typing import TYPE_CHECKING, Dict, FrozenSet, Iterable, List, Optional, Tuple
 
@@ -20,7 +21,7 @@ from synapse.api.constants import RelationTypes
 from synapse.api.errors import SynapseError
 from synapse.events import EventBase, relation_from_event
 from synapse.logging.opentracing import trace
-from synapse.storage.databases.main.relations import _RelatedEvent
+from synapse.storage.databases.main.relations import ThreadsNextBatch, _RelatedEvent
 from synapse.streams.config import PaginationConfig
 from synapse.types import JsonDict, Requester, StreamToken, UserID
 from synapse.visibility import filter_events_for_client
@@ -32,6 +33,13 @@ if TYPE_CHECKING:
 logger = logging.getLogger(__name__)
 
 
+class ThreadsListInclude(str, enum.Enum):
+    """Valid values for the 'include' flag of /threads."""
+
+    all = "all"
+    participated = "participated"
+
+
 @attr.s(slots=True, frozen=True, auto_attribs=True)
 class _ThreadAggregation:
     # The latest event in the thread.
@@ -482,3 +490,79 @@ class RelationsHandler:
             results.setdefault(event_id, BundledAggregations()).replace = edit
 
         return results
+
+    async def get_threads(
+        self,
+        requester: Requester,
+        room_id: str,
+        include: ThreadsListInclude,
+        limit: int = 5,
+        from_token: Optional[ThreadsNextBatch] = None,
+    ) -> JsonDict:
+        """Get related events of a event, ordered by topological ordering.
+
+        Args:
+            requester: The user requesting the relations.
+            room_id: The room the event belongs to.
+            include: One of "all" or "participated" to indicate which threads should
+                be returned.
+            limit: Only fetch the most recent `limit` events.
+            from_token: Fetch rows from the given token, or from the start if None.
+
+        Returns:
+            The pagination chunk.
+        """
+
+        user_id = requester.user.to_string()
+
+        # TODO Properly handle a user leaving a room.
+        (_, member_event_id) = await self._auth.check_user_in_room_or_world_readable(
+            room_id, requester, allow_departed_users=True
+        )
+
+        # Note that ignored users are not passed into get_relations_for_event
+        # below. Ignored users are handled in filter_events_for_client (and by
+        # not passing them in here we should get a better cache hit rate).
+        thread_roots, next_batch = await self._main_store.get_threads(
+            room_id=room_id, limit=limit, from_token=from_token
+        )
+
+        events = await self._main_store.get_events_as_list(thread_roots)
+
+        if include == ThreadsListInclude.participated:
+            # Pre-seed thread participation with whether the requester sent the event.
+            participated = {event.event_id: event.sender == user_id for event in events}
+            # For events the requester did not send, check the database for whether
+            # the requester sent a threaded reply.
+            participated.update(
+                await self._main_store.get_threads_participated(
+                    [eid for eid, p in participated.items() if not p],
+                    user_id,
+                )
+            )
+
+            # Limit the returned threads to those the user has participated in.
+            events = [event for event in events if participated[event.event_id]]
+
+        events = await filter_events_for_client(
+            self._storage_controllers,
+            user_id,
+            events,
+            is_peeking=(member_event_id is None),
+        )
+
+        aggregations = await self.get_bundled_aggregations(
+            events, requester.user.to_string()
+        )
+
+        now = self._clock.time_msec()
+        serialized_events = self._event_serializer.serialize_events(
+            events, now, bundle_aggregations=aggregations
+        )
+
+        return_value: JsonDict = {"chunk": serialized_events}
+
+        if next_batch:
+            return_value["next_batch"] = str(next_batch)
+
+        return return_value