Switch to chunk events so we can auth via power_levels (MSC2716) (#10432)

Previously, we were using `content.chunk_id` to connect one
chunk to another. But these events can be from any `sender`
and we can't tell who should be able to send historical events.
We know we only want the application service to do it but these
events have the sender of a real historical message, not the
application service user ID as the sender. Other federated homeservers
also have no indicator which senders are an application service on
the originating homeserver.

So we want to auth all of the MSC2716 events via power_levels
and have them be sent by the application service with proper
PL levels in the room.

changelog.d/10432.misc

@@ -0,0 +1 @@
+Connect historical chunks together with chunk events instead of a content field (MSC2716).

synapse/api/constants.py

@@ -120,6 +120,7 @@ class EventTypes:
     SpaceParent = "m.space.parent"
 
     MSC2716_INSERTION = "org.matrix.msc2716.insertion"
+    MSC2716_CHUNK = "org.matrix.msc2716.chunk"
     MSC2716_MARKER = "org.matrix.msc2716.marker"
 
 
@@ -190,9 +191,10 @@ class EventContentFields:
 
     # Used on normal messages to indicate they were historically imported after the fact
     MSC2716_HISTORICAL = "org.matrix.msc2716.historical"
-    # For "insertion" events
+    # For "insertion" events to indicate what the next chunk ID should be in
+    # order to connect to it
     MSC2716_NEXT_CHUNK_ID = "org.matrix.msc2716.next_chunk_id"
-    # Used on normal message events to indicate where the chunk connects to
+    # Used on "chunk" events to indicate which insertion event it connects to
     MSC2716_CHUNK_ID = "org.matrix.msc2716.chunk_id"
     # For "marker" events
     MSC2716_MARKER_INSERTION = "org.matrix.msc2716.marker.insertion"

synapse/rest/client/v1/room.py

@@ -553,9 +553,18 @@ class RoomBatchSendEventRestServlet(TransactionRestServlet):
             ]
 
         # Connect this current chunk to the insertion event from the previous chunk
-        last_event_in_chunk["content"][
+        chunk_event = {
-            EventContentFields.MSC2716_CHUNK_ID
+            "type": EventTypes.MSC2716_CHUNK,
-        ] = chunk_id_to_connect_to
+            "sender": requester.user.to_string(),
+            "room_id": room_id,
+            "content": {EventContentFields.MSC2716_CHUNK_ID: chunk_id_to_connect_to},
+            # Since the chunk event is put at the end of the chunk,
+            # where the newest-in-time event is, copy the origin_server_ts from
+            # the last event we're inserting
+            "origin_server_ts": last_event_in_chunk["origin_server_ts"],
+        }
+        # Add the chunk event to the end of the chunk (newest-in-time)
+        events_to_create.append(chunk_event)
 
         # Add an "insertion" event to the start of each chunk (next to the oldest-in-time
         # event in the chunk) so the next chunk can be connected to this one.
@@ -567,7 +576,7 @@ class RoomBatchSendEventRestServlet(TransactionRestServlet):
             # the first event we're inserting
             origin_server_ts=events_to_create[0]["origin_server_ts"],
         )
-        # Prepend the insertion event to the start of the chunk
+        # Prepend the insertion event to the start of the chunk (oldest-in-time)
         events_to_create = [insertion_event] + events_to_create
 
         event_ids = []