Merge remote-tracking branch 'origin/develop' into clokep/thumbnail-info

changelog.d/16162.misc

@@ -0,0 +1 @@
+Bump pyo3 from 0.17.1 to 0.19.2.

changelog.d/16403.bugfix

@@ -0,0 +1 @@
+Remove legacy unspecced `knock_state_events` field returned in some responses.

changelog.d/16419.misc

@@ -0,0 +1 @@
+Update registration of media repository URLs.

changelog.d/16420.doc

@@ -0,0 +1 @@
+Document internal background update mechanism.

changelog.d/16428.misc

@@ -0,0 +1 @@
+Improve code comments.

changelog.d/16435.misc

@@ -0,0 +1 @@
+Remove unused method.

docs/development/database_schema.md

@@ -150,6 +150,67 @@ def run_upgrade(
     ...
 ```
 
+## Background updates
+
+It is sometimes appropriate to perform database migrations as part of a background
+process (instead of blocking Synapse until the migration is done). In particular,
+this is useful for migrating data when adding new columns or tables.
+
+Pending background updates stored in the `background_updates` table and are denoted
+by a unique name, the current status (stored in JSON), and some dependency information:
+
+* Whether the update requires a previous update to be complete.
+* A rough ordering for which to complete updates.
+
+A new background updates needs to be added to the `background_updates` table:
+
+```sql
+INSERT INTO background_updates (ordering, update_name, depends_on, progress_json) VALUES
+  (7706, 'my_background_update', 'a_previous_background_update' '{}');
+```
+
+And then needs an associated handler in the appropriate datastore:
+
+```python
+self.db_pool.updates.register_background_update_handler(
+    "my_background_update",
+    update_handler=self._my_background_update,
+)
+```
+
+There are a few types of updates that can be performed, see the `BackgroundUpdater`:
+
+* `register_background_update_handler`: A generic handler for custom SQL
+* `register_background_index_update`: Create an index in the background
+* `register_background_validate_constraint`: Validate a constraint in the background
+  (PostgreSQL-only)
+* `register_background_validate_constraint_and_delete_rows`: Similar to
+  `register_background_validate_constraint`, but deletes rows which don't fit
+  the constraint.
+
+For `register_background_update_handler`, the generic handler must track progress
+and then finalize the background update:
+
+```python
+async def _my_background_update(self, progress: JsonDict, batch_size: int) -> int:
+    def _do_something(txn: LoggingTransaction) -> int:
+        ...
+        self.db_pool.updates._background_update_progress_txn(
+            txn, "my_background_update", {"last_processed": last_processed}
+        )
+        return last_processed - prev_last_processed
+
+    num_processed = await self.db_pool.runInteraction("_do_something", _do_something)
+    await self.db_pool.updates._end_background_update("my_background_update")
+
+    return num_processed
+```
+
+Synapse will attempt to rate-limit how often background updates are run via the
+given batch-size and the returned number of processed entries (and how long the
+function took to run). See
+[background update controller callbacks](../modules/background_update_controller_callbacks.md).
+
 ## Boolean columns
 
 Boolean columns require special treatment, since SQLite treats booleans the

rust/Cargo.toml

@@ -25,14 +25,14 @@ name = "synapse.synapse_rust"
 anyhow = "1.0.63"
 lazy_static = "1.4.0"
 log = "0.4.17"
-pyo3 = { version = "0.17.1", features = [
+pyo3 = { version = "0.19.2", features = [
     "macros",
     "anyhow",
     "abi3",
-    "abi3-py37",
+    "abi3-py38",
 ] }
 pyo3-log = "0.8.1"
-pythonize = "0.17.0"
+pythonize = "0.19.0"
 regex = "1.6.0"
 serde = { version = "1.0.144", features = ["derive"] }
 serde_json = "1.0.85"

rust/src/push/evaluator.rs

@@ -105,6 +105,17 @@ impl PushRuleEvaluator {
     /// Create a new `PushRuleEvaluator`. See struct docstring for details.
     #[allow(clippy::too_many_arguments)]
     #[new]
+    #[pyo3(signature = (
+        flattened_keys,
+        has_mentions,
+        room_member_count,
+        sender_power_level,
+        notification_power_levels,
+        related_events_flattened,
+        related_event_match_enabled,
+        room_version_feature_flags,
+        msc3931_enabled,
+    ))]
     pub fn py_new(
         flattened_keys: BTreeMap<String, JsonValue>,
         has_mentions: bool,

synapse/federation/federation_client.py

@@ -1402,7 +1402,7 @@ async def send_knock(self, destinations: List[str], pdu: EventBase) -> JsonDict:
             The remote homeserver return some state from the room. The response
             dictionary is in the form:
 
-            {"knock_state_events": [<state event dict>, ...]}
+            {"knock_room_state": [<state event dict>, ...]}
 
             The list of state events may be empty.
 
@@ -1429,7 +1429,7 @@ async def _do_send_knock(self, destination: str, pdu: EventBase) -> JsonDict:
             The remote homeserver can optionally return some state from the room. The response
             dictionary is in the form:
 
-            {"knock_state_events": [<state event dict>, ...]}
+            {"knock_room_state": [<state event dict>, ...]}
 
             The list of state events may be empty.
         """

synapse/federation/federation_server.py

@@ -850,14 +850,7 @@ async def on_send_knock_request(
                 context, self._room_prejoin_state_types
             )
         )
-        return {
-            "knock_room_state": stripped_room_state,
-            # Since v1.37, Synapse incorrectly used "knock_state_events" for this field.
-            # Thus, we also populate a 'knock_state_events' with the same content to
-            # support old instances.
-            # See https://github.com/matrix-org/synapse/issues/14088.
-            "knock_state_events": stripped_room_state,
-        }
+        return {"knock_room_state": stripped_room_state}
 
     async def _on_send_membership_event(
         self, origin: str, content: JsonDict, membership_type: str, room_id: str

synapse/federation/sender/__init__.py

@@ -67,7 +67,7 @@
 
 When the `PerDestinationQueue` has the catch-up flag set, the *Catch-Up Transmission Loop*
 (`_catch_up_transmission_loop`) is used in lieu of the regular `_transaction_transmission_loop`.
-(Only once the catch-up mode has been exited can the regular tranaction transmission behaviour
+(Only once the catch-up mode has been exited can the regular transaction transmission behaviour
 be resumed.)
 
 *Catch-Up Mode*, entered upon Synapse startup or once a homeserver has fallen behind due to

synapse/federation/transport/client.py

@@ -431,7 +431,7 @@ async def send_knock_v1(
             The remote homeserver can optionally return some state from the room. The response
             dictionary is in the form:
 
-            {"knock_state_events": [<state event dict>, ...]}
+            {"knock_room_state": [<state event dict>, ...]}
 
             The list of state events may be empty.
         """

synapse/handlers/federation.py

@@ -868,19 +868,10 @@ async def do_knock(
         # This is a bit of a hack and is cribbing off of invites. Basically we
         # store the room state here and retrieve it again when this event appears
         # in the invitee's sync stream. It is stripped out for all other local users.
-        stripped_room_state = (
-            knock_response.get("knock_room_state")
-            # Since v1.37, Synapse incorrectly used "knock_state_events" for this field.
-            # Thus, we also check for a 'knock_state_events' to support old instances.
-            # See https://github.com/matrix-org/synapse/issues/14088.
-            or knock_response.get("knock_state_events")
-        )
+        stripped_room_state = knock_response.get("knock_room_state")
 
         if stripped_room_state is None:
-            raise KeyError(
-                "Missing 'knock_room_state' (or legacy 'knock_state_events') field in "
-                "send_knock response"
-            )
+            raise KeyError("Missing 'knock_room_state' field in send_knock response")
 
         event.unsigned["knock_room_state"] = stripped_room_state
 

synapse/http/server.py

@@ -266,7 +266,7 @@ class HttpServer(Protocol):
     def register_paths(
         self,
         method: str,
-        path_patterns: Iterable[Pattern],
+        path_patterns: Iterable[Pattern[str]],
         callback: ServletCallback,
         servlet_classname: str,
     ) -> None:

synapse/media/_base.py

@@ -26,11 +26,11 @@
 from twisted.protocols.basic import FileSender
 from twisted.web.server import Request
 
-from synapse.api.errors import Codes, SynapseError, cs_error
+from synapse.api.errors import Codes, cs_error
 from synapse.http.server import finish_request, respond_with_json
 from synapse.http.site import SynapseRequest
 from synapse.logging.context import make_deferred_yieldable
-from synapse.util.stringutils import is_ascii, parse_and_validate_server_name
+from synapse.util.stringutils import is_ascii
 
 logger = logging.getLogger(__name__)
 
@@ -84,52 +84,12 @@
 ]
 
 
-def parse_media_id(request: Request) -> Tuple[str, str, Optional[str]]:
-    """Parses the server name, media ID and optional file name from the request URI
-
-    Also performs some rough validation on the server name.
-
-    Args:
-        request: The `Request`.
-
-    Returns:
-        A tuple containing the parsed server name, media ID and optional file name.
-
-    Raises:
-        SynapseError(404): if parsing or validation fail for any reason
-    """
-    try:
-        # The type on postpath seems incorrect in Twisted 21.2.0.
-        postpath: List[bytes] = request.postpath  # type: ignore
-        assert postpath
-
-        # This allows users to append e.g. /test.png to the URL. Useful for
-        # clients that parse the URL to see content type.
-        server_name_bytes, media_id_bytes = postpath[:2]
-        server_name = server_name_bytes.decode("utf-8")
-        media_id = media_id_bytes.decode("utf8")
-
-        # Validate the server name, raising if invalid
-        parse_and_validate_server_name(server_name)
-
-        file_name = None
-        if len(postpath) > 2:
-            try:
-                file_name = urllib.parse.unquote(postpath[-1].decode("utf-8"))
-            except UnicodeDecodeError:
-                pass
-        return server_name, media_id, file_name
-    except Exception:
-        raise SynapseError(
-            404, "Invalid media id token %r" % (request.postpath,), Codes.UNKNOWN
-        )
-
-
 def respond_404(request: SynapseRequest) -> None:
+    assert request.path is not None
     respond_with_json(
         request,
         404,
-        cs_error("Not found %r" % (request.postpath,), code=Codes.NOT_FOUND),
+        cs_error("Not found '%s'" % (request.path.decode(),), code=Codes.NOT_FOUND),
         send_cors=True,
     )
 

synapse/media/media_repository.py

@@ -48,6 +48,7 @@
 from synapse.media.media_storage import MediaStorage
 from synapse.media.storage_provider import StorageProviderWrapper
 from synapse.media.thumbnailer import Thumbnailer, ThumbnailError
+from synapse.media.url_previewer import UrlPreviewer
 from synapse.metrics.background_process_metrics import run_as_background_process
 from synapse.types import UserID
 from synapse.util.async_helpers import Linearizer
@@ -114,7 +115,7 @@ def __init__(self, hs: "HomeServer"):
             )
             storage_providers.append(provider)
 
-        self.media_storage = MediaStorage(
+        self.media_storage: MediaStorage = MediaStorage(
             self.hs, self.primary_base_path, self.filepaths, storage_providers
         )
 
@@ -142,6 +143,13 @@ def __init__(self, hs: "HomeServer"):
                 MEDIA_RETENTION_CHECK_PERIOD_MS,
             )
 
+        if hs.config.media.url_preview_enabled:
+            self.url_previewer: Optional[UrlPreviewer] = UrlPreviewer(
+                hs, self, self.media_storage
+            )
+        else:
+            self.url_previewer = None
+
     def _start_update_recently_accessed(self) -> Deferred:
         return run_as_background_process(
             "update_recently_accessed_media", self._update_recently_accessed

synapse/replication/tcp/commands.py

@@ -446,8 +446,6 @@ class RemoteServerUpCommand(_SimpleCommand):
     """Sent when a worker has detected that a remote server is no longer
     "down" and retry timings should be reset.
 
-    If sent from a client the server will relay to all other workers.
-
     Format::
 
         REMOTE_SERVER_UP <server>