Merge pull request #6186 from matrix-org/erikj/disable_sql_bytes

Disable bytes usage with postgres

5 files changed, 61 insertions, 12 deletions

diff --git a/synapse/storage/engines/postgres.py b/synapse/storage/engines/postgres.py
index 601617b21e..d670286fa5 100644
--- a/synapse/storage/engines/postgres.py
+++ b/synapse/storage/engines/postgres.py
@@ -22,6 +22,13 @@ class PostgresEngine(object):
     def __init__(self, database_module, database_config):
         self.module = database_module
         self.module.extensions.register_type(self.module.extensions.UNICODE)
+
+        # Disables passing `bytes` to txn.execute, c.f. #6186. If you do
+        # actually want to use bytes than wrap it in `bytearray`.
+        def _disable_bytes_adapter(_):
+            raise Exception("Passing bytes to DB is disabled.")
+
+        self.module.extensions.register_adapter(bytes, _disable_bytes_adapter)
         self.synchronous_commit = database_config.get("synchronous_commit", True)
         self._version = None  # unknown as yet
 
diff --git a/synapse/storage/events_bg_updates.py b/synapse/storage/events_bg_updates.py
index 5717baf48c..e77a7e28af 100644
--- a/synapse/storage/events_bg_updates.py
+++ b/synapse/storage/events_bg_updates.py
@@ -71,6 +71,19 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore):
             "redactions_received_ts", self._redactions_received_ts
         )
 
+        # This index gets deleted in `event_fix_redactions_bytes` update
+        self.register_background_index_update(
+            "event_fix_redactions_bytes_create_index",
+            index_name="redactions_censored_redacts",
+            table="redactions",
+            columns=["redacts"],
+            where_clause="have_censored",
+        )
+
+        self.register_background_update_handler(
+            "event_fix_redactions_bytes", self._event_fix_redactions_bytes
+        )
+
     @defer.inlineCallbacks
     def _background_reindex_fields_sender(self, progress, batch_size):
         target_min_stream_id = progress["target_min_stream_id_inclusive"]
@@ -458,3 +471,33 @@ class EventsBackgroundUpdatesStore(BackgroundUpdateStore):
             yield self._end_background_update("redactions_received_ts")
 
         return count
+
+    @defer.inlineCallbacks
+    def _event_fix_redactions_bytes(self, progress, batch_size):
+        """Undoes hex encoded censored redacted event JSON.
+        """
+
+        def _event_fix_redactions_bytes_txn(txn):
+            # This update is quite fast due to new index.
+            txn.execute(
+                """
+                UPDATE event_json
+                SET
+                    json = convert_from(json::bytea, 'utf8')
+                FROM redactions
+                WHERE
+                    redactions.have_censored
+                    AND event_json.event_id = redactions.redacts
+                    AND json NOT LIKE '{%';
+                """
+            )
+
+            txn.execute("DROP INDEX redactions_censored_redacts")
+
+        yield self.runInteraction(
+            "_event_fix_redactions_bytes", _event_fix_redactions_bytes_txn
+        )
+
+        yield self._end_background_update("event_fix_redactions_bytes")
+
+        return 1
diff --git a/synapse/storage/filtering.py b/synapse/storage/filtering.py
index 23b48f6cea..7c2a7da836 100644
--- a/synapse/storage/filtering.py
+++ b/synapse/storage/filtering.py
@@ -51,7 +51,7 @@ class FilteringStore(SQLBaseStore):
                 "SELECT filter_id FROM user_filters "
                 "WHERE user_id = ? AND filter_json = ?"
             )
-            txn.execute(sql, (user_localpart, def_json))
+            txn.execute(sql, (user_localpart, bytearray(def_json)))
             filter_id_response = txn.fetchone()
             if filter_id_response is not None:
                 return filter_id_response[0]
@@ -68,7 +68,7 @@ class FilteringStore(SQLBaseStore):
                 "INSERT INTO user_filters (user_id, filter_id, filter_json)"
                 "VALUES(?, ?, ?)"
             )
-            txn.execute(sql, (user_localpart, filter_id, def_json))
+            txn.execute(sql, (user_localpart, filter_id, bytearray(def_json)))
 
             return filter_id
 
diff --git a/synapse/storage/pusher.py b/synapse/storage/pusher.py
index 3e0e834a62..b12e80440a 100644
--- a/synapse/storage/pusher.py
+++ b/synapse/storage/pusher.py
@@ -241,7 +241,7 @@ class PusherStore(PusherWorkerStore):
                     "device_display_name": device_display_name,
                     "ts": pushkey_ts,
                     "lang": lang,
-                    "data": encode_canonical_json(data),
+                    "data": bytearray(encode_canonical_json(data)),
                     "last_stream_ordering": last_stream_ordering,
                     "profile_tag": profile_tag,
                     "id": stream_id,
diff --git a/synapse/storage/schema/delta/56/redaction_censor3_fix_update.sql.postgres b/synapse/storage/schema/delta/56/redaction_censor3_fix_update.sql.postgres
index f7bcc5e2f2..67471f3ef5 100644
--- a/synapse/storage/schema/delta/56/redaction_censor3_fix_update.sql.postgres
+++ b/synapse/storage/schema/delta/56/redaction_censor3_fix_update.sql.postgres
@@ -15,12 +15,11 @@
 
 
 -- There was a bug where we may have updated censored redactions as bytes,
--- which can (somehow) cause json to be inserted hex encoded. This goes and
--- undoes any such hex encoded JSON.
-UPDATE event_json SET json = convert_from(json::bytea, 'utf8')
-WHERE event_id IN (
-  SELECT event_json.event_id
-  FROM event_json
-  INNER JOIN redactions ON (event_json.event_id = redacts)
-  WHERE have_censored AND json NOT LIKE '{%'
-);
+-- which can (somehow) cause json to be inserted hex encoded. These updates go
+-- and undoes any such hex encoded JSON.
+
+INSERT into background_updates (update_name, progress_json)
+  VALUES ('event_fix_redactions_bytes_create_index', '{}');
+
+INSERT into background_updates (update_name, progress_json, depends_on)
+  VALUES ('event_fix_redactions_bytes', '{}', 'event_fix_redactions_bytes_create_index');