Support filtering the /messages API by relation type (MSC3874). (#14148)

Gated behind an experimental configuration flag.

synapse/api/filtering.py

@@ -36,7 +36,7 @@ from jsonschema import FormatChecker
 from synapse.api.constants import EduTypes, EventContentFields
 from synapse.api.errors import SynapseError
 from synapse.api.presence import UserPresenceState
-from synapse.events import EventBase
+from synapse.events import EventBase, relation_from_event
 from synapse.types import JsonDict, RoomID, UserID
 
 if TYPE_CHECKING:
@@ -53,6 +53,12 @@ FILTER_SCHEMA = {
         # check types are valid event types
         "types": {"type": "array", "items": {"type": "string"}},
         "not_types": {"type": "array", "items": {"type": "string"}},
+        # MSC3874, filtering /messages.
+        "org.matrix.msc3874.rel_types": {"type": "array", "items": {"type": "string"}},
+        "org.matrix.msc3874.not_rel_types": {
+            "type": "array",
+            "items": {"type": "string"},
+        },
     },
 }
 
@@ -334,8 +340,15 @@ class Filter:
         self.labels = filter_json.get("org.matrix.labels", None)
         self.not_labels = filter_json.get("org.matrix.not_labels", [])
 
-        self.related_by_senders = self.filter_json.get("related_by_senders", None)
-        self.related_by_rel_types = self.filter_json.get("related_by_rel_types", None)
+        self.related_by_senders = filter_json.get("related_by_senders", None)
+        self.related_by_rel_types = filter_json.get("related_by_rel_types", None)
+
+        # For compatibility with _check_fields.
+        self.rel_types = None
+        self.not_rel_types = []
+        if hs.config.experimental.msc3874_enabled:
+            self.rel_types = filter_json.get("org.matrix.msc3874.rel_types", None)
+            self.not_rel_types = filter_json.get("org.matrix.msc3874.not_rel_types", [])
 
     def filters_all_types(self) -> bool:
         return "*" in self.not_types
@@ -386,11 +399,19 @@ class Filter:
             # check if there is a string url field in the content for filtering purposes
             labels = content.get(EventContentFields.LABELS, [])
 
+            # Check if the event has a relation.
+            rel_type = None
+            if isinstance(event, EventBase):
+                relation = relation_from_event(event)
+                if relation:
+                    rel_type = relation.rel_type
+
             field_matchers = {
                 "rooms": lambda v: room_id == v,
                 "senders": lambda v: sender == v,
                 "types": lambda v: _matches_wildcard(ev_type, v),
                 "labels": lambda v: v in labels,
+                "rel_types": lambda v: rel_type == v,
             }
 
             result = self._check_fields(field_matchers)

synapse/config/experimental.py

@@ -117,3 +117,6 @@ class ExperimentalConfig(Config):
         self.msc3882_token_timeout = self.parse_duration(
             experimental.get("msc3882_token_timeout", "5m")
         )
+
+        # MSC3874: Filtering /messages with rel_types / not_rel_types.
+        self.msc3874_enabled: bool = experimental.get("msc3874_enabled", False)

synapse/rest/client/versions.py

@@ -114,6 +114,8 @@ class VersionsRestServlet(RestServlet):
                     "org.matrix.msc3882": self.config.experimental.msc3882_enabled,
                     # Adds support for remotely enabling/disabling pushers, as per MSC3881
                     "org.matrix.msc3881": self.config.experimental.msc3881_enabled,
+                    # Adds support for filtering /messages by event relation.
+                    "org.matrix.msc3874": self.config.experimental.msc3874_enabled,
                 },
             },
         )

synapse/storage/databases/main/stream.py

@@ -357,6 +357,24 @@ def filter_to_clause(event_filter: Optional[Filter]) -> Tuple[str, List[str]]:
         )
         args.extend(event_filter.related_by_rel_types)
 
+    if event_filter.rel_types:
+        clauses.append(
+            "(%s)"
+            % " OR ".join(
+                "event_relation.relation_type = ?" for _ in event_filter.rel_types
+            )
+        )
+        args.extend(event_filter.rel_types)
+
+    if event_filter.not_rel_types:
+        clauses.append(
+            "((%s) OR event_relation.relation_type IS NULL)"
+            % " AND ".join(
+                "event_relation.relation_type != ?" for _ in event_filter.not_rel_types
+            )
+        )
+        args.extend(event_filter.not_rel_types)
+
     return " AND ".join(clauses), args
 
 
@@ -1278,8 +1296,8 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore):
                 # Multiple labels could cause the same event to appear multiple times.
                 needs_distinct = True
 
-        # If there is a filter on relation_senders and relation_types join to the
-        # relations table.
+        # If there is a relation_senders and relation_types filter join to the
+        # relations table to get events related to the current event.
         if event_filter and (
             event_filter.related_by_senders or event_filter.related_by_rel_types
         ):
@@ -1294,6 +1312,13 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore):
                     LEFT JOIN events AS related_event ON (relation.event_id = related_event.event_id)
                 """
 
+        # If there is a not_rel_types filter join to the relations table to get
+        # the event's relation information.
+        if event_filter and (event_filter.rel_types or event_filter.not_rel_types):
+            join_clause += """
+                LEFT JOIN event_relations AS event_relation USING (event_id)
+            """
+
         if needs_distinct:
             select_keywords += " DISTINCT"