Merge pull request #922 from matrix-org/erikj/file_api2

Feature: Add filter to /messages. Add 'contains_url' to filter.

1 files changed, 82 insertions, 0 deletions

diff --git a/synapse/storage/events.py b/synapse/storage/events.py
index 9d74fd159d..6610549281 100644
--- a/synapse/storage/events.py
+++ b/synapse/storage/events.py
@@ -152,6 +152,7 @@ _EventCacheEntry = namedtuple("_EventCacheEntry", ("event", "redacted_event"))
 
 class EventsStore(SQLBaseStore):
     EVENT_ORIGIN_SERVER_TS_NAME = "event_origin_server_ts"
+    EVENT_FIELDS_SENDER_URL_UPDATE_NAME = "event_fields_sender_url"
 
     def __init__(self, hs):
         super(EventsStore, self).__init__(hs)
@@ -159,6 +160,10 @@ class EventsStore(SQLBaseStore):
         self.register_background_update_handler(
             self.EVENT_ORIGIN_SERVER_TS_NAME, self._background_reindex_origin_server_ts
         )
+        self.register_background_update_handler(
+            self.EVENT_FIELDS_SENDER_URL_UPDATE_NAME,
+            self._background_reindex_fields_sender,
+        )
 
         self._event_persist_queue = _EventPeristenceQueue()
 
@@ -576,6 +581,11 @@ class EventsStore(SQLBaseStore):
                     "content": encode_json(event.content).decode("UTF-8"),
                     "origin_server_ts": int(event.origin_server_ts),
                     "received_ts": self._clock.time_msec(),
+                    "sender": event.sender,
+                    "contains_url": (
+                        "url" in event.content
+                        and isinstance(event.content["url"], basestring)
+                    ),
                 }
                 for event, _ in events_and_contexts
             ],
@@ -1116,6 +1126,78 @@ class EventsStore(SQLBaseStore):
         defer.returnValue(ret)
 
     @defer.inlineCallbacks
+    def _background_reindex_fields_sender(self, progress, batch_size):
+        target_min_stream_id = progress["target_min_stream_id_inclusive"]
+        max_stream_id = progress["max_stream_id_exclusive"]
+        rows_inserted = progress.get("rows_inserted", 0)
+
+        INSERT_CLUMP_SIZE = 1000
+
+        def reindex_txn(txn):
+            sql = (
+                "SELECT stream_ordering, event_id, json FROM events"
+                " INNER JOIN event_json USING (event_id)"
+                " WHERE ? <= stream_ordering AND stream_ordering < ?"
+                " ORDER BY stream_ordering DESC"
+                " LIMIT ?"
+            )
+
+            txn.execute(sql, (target_min_stream_id, max_stream_id, batch_size))
+
+            rows = txn.fetchall()
+            if not rows:
+                return 0
+
+            min_stream_id = rows[-1][0]
+
+            update_rows = []
+            for row in rows:
+                try:
+                    event_id = row[1]
+                    event_json = json.loads(row[2])
+                    sender = event_json["sender"]
+                    content = event_json["content"]
+
+                    contains_url = "url" in content
+                    if contains_url:
+                        contains_url &= isinstance(content["url"], basestring)
+                except (KeyError, AttributeError):
+                    # If the event is missing a necessary field then
+                    # skip over it.
+                    continue
+
+                update_rows.append((sender, contains_url, event_id))
+
+            sql = (
+                "UPDATE events SET sender = ?, contains_url = ? WHERE event_id = ?"
+            )
+
+            for index in range(0, len(update_rows), INSERT_CLUMP_SIZE):
+                clump = update_rows[index:index + INSERT_CLUMP_SIZE]
+                txn.executemany(sql, clump)
+
+            progress = {
+                "target_min_stream_id_inclusive": target_min_stream_id,
+                "max_stream_id_exclusive": min_stream_id,
+                "rows_inserted": rows_inserted + len(rows)
+            }
+
+            self._background_update_progress_txn(
+                txn, self.EVENT_FIELDS_SENDER_URL_UPDATE_NAME, progress
+            )
+
+            return len(rows)
+
+        result = yield self.runInteraction(
+            self.EVENT_FIELDS_SENDER_URL_UPDATE_NAME, reindex_txn
+        )
+
+        if not result:
+            yield self._end_background_update(self.EVENT_FIELDS_SENDER_URL_UPDATE_NAME)
+
+        defer.returnValue(result)
+
+    @defer.inlineCallbacks
     def _background_reindex_origin_server_ts(self, progress, batch_size):
         target_min_stream_id = progress["target_min_stream_id_inclusive"]
         max_stream_id = progress["max_stream_id_exclusive"]