Merge e3430769b29a0767431b8daf488aa0cae3d6a1fb into 3878699df7a812031da50cfbbb1a76f4b935d045

As we're now performing another db txn to check if the user is an admin.

changelog.d/18238.feature

@@ -0,0 +1 @@
+If enabled by the user, server admins will see [soft failed](https://spec.matrix.org/v1.13/server-server-api/#soft-failure) events over the Client-Server API.

changelog.d/18585.feature

@@ -0,0 +1 @@
+When admins enable themselves to see soft-failed events, they will also see if the cause is due to the policy server flagging them as spam via `unsigned`.

docs/SUMMARY.md

@@ -74,6 +74,7 @@
       - [Users](admin_api/user_admin_api.md)
       - [Server Version](admin_api/version_api.md)
       - [Federation](usage/administration/admin_api/federation.md)
+      - [Client-Server API Extensions](admin_api/client_server_api_extensions.md)
     - [Manhole](manhole.md)
     - [Monitoring](metrics-howto.md)
       - [Reporting Homeserver Usage Statistics](usage/administration/monitoring/reporting_homeserver_usage_statistics.md)

docs/admin_api/client_server_api_extensions.md

@@ -0,0 +1,51 @@
+# Client-Server API Extensions
+
+Server administrators can set special account data to change how the Client-Server API behaves for
+their clients. Setting the account data, or having it already set, as a non-admin has no effect.
+
+All configuration options can be set through the `io.element.synapse.admin_client_config` global 
+account data on the admin's user account. 
+
+Example:
+```
+PUT /_matrix/client/v3/user/{adminUserId}/account_data/io.element.synapse.admin_client_config
+{
+    "return_soft_failed_events": true
+}
+```
+
+## See soft failed events
+
+Learn more about soft failure from [the spec](https://spec.matrix.org/v1.14/server-server-api/#soft-failure).
+
+To receive soft failed events in APIs like `/sync` and `/messages`, set `return_soft_failed_events`
+to `true` in the admin client config. When `false`, the normal behaviour of these endpoints is to
+exclude soft failed events.
+
+**Note**: If the policy server flagged the event as spam and that caused soft failure, that will be indicated
+in the event's `unsigned` content like so:
+
+```json
+{
+  "type": "m.room.message",
+  "other": "event_fields_go_here",
+  "unsigned": {
+    "io.element.synapse.soft_failed": true,
+    "io.element.synapse.policy_server_spammy": true
+  }
+}
+```
+
+Default: `false`
+
+## See events marked spammy by policy servers
+
+Learn more about policy servers from [MSC4284](https://github.com/matrix-org/matrix-spec-proposals/pull/4284).
+
+Similar to `return_soft_failed_events`, clients logged in with admin accounts can see events which were
+flagged by the policy server as spammy (and thus soft failed) by setting `return_policy_server_spammy_events`
+to `true`. If `return_soft_failed_events` is `true`, then `return_policy_server_spammy_events` is implied
+`true`. When `false`, the normal behaviour of Client-Server API endpoints is retained (unless `return_soft_failed_events`
+is `true`, per above).
+
+Default: `false`

rust/src/events/internal_metadata.rs

@@ -54,6 +54,7 @@ enum EventInternalMetadataData {
     RecheckRedaction(bool),
     SoftFailed(bool),
     ProactivelySend(bool),
+    PolicyServerSpammy(bool),
     Redacted(bool),
     TxnId(Box<str>),
     TokenId(i64),
@@ -96,6 +97,13 @@ impl EventInternalMetadataData {
                     .to_owned()
                     .into_any(),
             ),
+            EventInternalMetadataData::PolicyServerSpammy(o) => (
+                pyo3::intern!(py, "policy_server_spammy"),
+                o.into_pyobject(py)
+                    .unwrap_infallible()
+                    .to_owned()
+                    .into_any(),
+            ),
             EventInternalMetadataData::Redacted(o) => (
                 pyo3::intern!(py, "redacted"),
                 o.into_pyobject(py)
@@ -155,6 +163,11 @@ impl EventInternalMetadataData {
                     .extract()
                     .with_context(|| format!("'{key_str}' has invalid type"))?,
             ),
+            "policy_server_spammy" => EventInternalMetadataData::PolicyServerSpammy(
+                value
+                    .extract()
+                    .with_context(|| format!("'{key_str}' has invalid type"))?,
+            ),
             "redacted" => EventInternalMetadataData::Redacted(
                 value
                     .extract()
@@ -427,6 +440,17 @@ impl EventInternalMetadata {
         set_property!(self, ProactivelySend, obj);
     }
 
+    #[getter]
+    fn get_policy_server_spammy(&self) -> PyResult<bool> {
+        Ok(get_property_opt!(self, PolicyServerSpammy)
+            .copied()
+            .unwrap_or(false))
+    }
+    #[setter]
+    fn set_policy_server_spammy(&mut self, obj: bool) {
+        set_property!(self, PolicyServerSpammy, obj);
+    }
+
     #[getter]
     fn get_redacted(&self) -> PyResult<bool> {
         let bool = get_property!(self, Redacted)?;

synapse/api/constants.py

@@ -290,6 +290,9 @@ class AccountDataTypes:
     MSC4155_INVITE_PERMISSION_CONFIG: Final = (
         "org.matrix.msc4155.invite_permission_config"
     )
+    # Synapse-specific behaviour. See "Client-Server API Extensions" documentation
+    # in Admin API for more information.
+    SYNAPSE_ADMIN_CLIENT_CONFIG: Final = "io.element.synapse.admin_client_config"
 
 
 class HistoryVisibility:

synapse/appservice/api.py

@@ -555,6 +555,9 @@ class ApplicationServiceApi(SimpleHttpClient):
                         )
                         and service.is_interested_in_user(e.state_key)
                     ),
+                    # Appservices are considered 'trusted' by the admin and should have
+                    # applicable metadata on their events.
+                    include_admin_metadata=True,
                 ),
             )
             for e in events

synapse/events/utils.py

@@ -421,11 +421,31 @@ class SerializeEventConfig:
     # False, that state will be removed from the event before it is returned.
     # Otherwise, it will be kept.
     include_stripped_room_state: bool = False
+    # When True, sets unsigned flags to help clients identify events which
+    # only server admins can see through other configuration. For example,
+    # whether an event was soft failed by the server.
+    include_admin_metadata: bool = False
+
+    # Developer note: when adding properties, update make_config_for_admin() below.
 
 
 _DEFAULT_SERIALIZE_EVENT_CONFIG = SerializeEventConfig()
 
 
+def make_config_for_admin(existing: SerializeEventConfig) -> SerializeEventConfig:
+    # Developer note: when adding properties, update test_make_serialize_config_for_admin_retains_other_fields
+    return SerializeEventConfig(
+        # Set the options which are only available to server admins
+        include_admin_metadata=True,
+        # And copy the rest
+        as_client_event=existing.as_client_event,
+        event_format=existing.event_format,
+        requester=existing.requester,
+        only_event_fields=existing.only_event_fields,
+        include_stripped_room_state=existing.include_stripped_room_state,
+    )
+
+
 def serialize_event(
     e: Union[JsonDict, EventBase],
     time_now_ms: int,
@@ -528,6 +548,12 @@ def serialize_event(
             d["content"] = dict(d["content"])
             d["content"]["redacts"] = e.redacts
 
+    if config.include_admin_metadata:
+        if e.internal_metadata.is_soft_failed():
+            d["unsigned"]["io.element.synapse.soft_failed"] = True
+        if e.internal_metadata.policy_server_spammy:
+            d["unsigned"]["io.element.synapse.policy_server_spammy"] = True
+
     only_event_fields = config.only_event_fields
     if only_event_fields:
         if not isinstance(only_event_fields, list) or not all(
@@ -576,6 +602,15 @@ class EventClientSerializer:
         if not isinstance(event, EventBase):
             return event
 
+        # Force-enable server admin metadata because the only time an event with
+        # relevant metadata will be when the admin requested it via their admin
+        # client config account data. Also, it's "just" some `unsigned` flags, so
+        # shouldn't cause much in terms of problems to downstream consumers.
+        if config.requester is not None and await self._store.is_server_admin(
+            config.requester.user
+        ):
+            config = make_config_for_admin(config)
+
         serialized_event = serialize_event(event, time_now, config=config)
 
         new_unsigned = {}

synapse/federation/federation_base.py

@@ -174,6 +174,7 @@ class FederationBase:
                 "Event not allowed by policy server, soft-failing %s", pdu.event_id
             )
             pdu.internal_metadata.soft_failed = True
+            pdu.internal_metadata.policy_server_spammy = True
             # Note: we don't redact the event so admins can inspect the event after the
             # fact. Other processes may redact the event, but that won't be applied to
             # the database copy of the event until the server's config requires it.

synapse/handlers/message.py

@@ -1111,6 +1111,9 @@ class EventCreationHandler:
 
                 policy_allowed = await self._policy_handler.is_event_allowed(event)
                 if not policy_allowed:
+                    # We shouldn't need to set the metadata because the raise should
+                    # cause the request to be denied, but just in case:
+                    event.internal_metadata.policy_server_spammy = True
                     logger.warning(
                         "Event not allowed by policy server, rejecting %s",
                         event.event_id,

synapse/storage/admin_client_config.py

@@ -0,0 +1,22 @@
+import logging
+from typing import Optional
+
+from synapse.types import JsonMapping
+
+logger = logging.getLogger(__name__)
+
+
+class AdminClientConfig:
+    """Class to track various Synapse-specific admin-only client-impacting config options."""
+
+    def __init__(self, account_data: Optional[JsonMapping]):
+        self.return_soft_failed_events: bool = False
+        self.return_policy_server_spammy_events: bool = False
+
+        if account_data:
+            self.return_soft_failed_events = account_data.get(
+                "return_soft_failed_events", False
+            )
+            self.return_policy_server_spammy_events = account_data.get(
+                "return_policy_server_spammy_events", False
+            )

synapse/storage/databases/main/account_data.py

@@ -37,6 +37,7 @@ from synapse.api.constants import AccountDataTypes
 from synapse.api.errors import Codes, SynapseError
 from synapse.replication.tcp.streams import AccountDataStream
 from synapse.storage._base import db_to_json
+from synapse.storage.admin_client_config import AdminClientConfig
 from synapse.storage.database import (
     DatabasePool,
     LoggingDatabaseConnection,
@@ -578,6 +579,21 @@ class AccountDataWorkerStore(PushRulesWorkerStore, CacheInvalidationWorkerStore)
         )
         return InviteRulesConfig(data)
 
+    async def get_admin_client_config_for_user(self, user_id: str) -> AdminClientConfig:
+        """
+        Get the admin client configuration for the specified user.
+
+        The admin client config contains Synapse-specific settings that clients running
+        server admin accounts can use. They have no effect on non-admin users.
+
+        Args:
+            user_id: The user ID to get config for.
+        """
+        data = await self.get_global_account_data_by_type_for_user(
+            user_id, AccountDataTypes.SYNAPSE_ADMIN_CLIENT_CONFIG
+        )
+        return AdminClientConfig(data)
+
     def process_replication_rows(
         self,
         stream_name: str,

synapse/synapse_rust/events.pyi

@@ -33,6 +33,9 @@ class EventInternalMetadata:
     proactively_send: bool
     redacted: bool
 
+    policy_server_spammy: bool
+    """whether the policy server indicated that this event is spammy"""
+
     txn_id: str
     """The transaction ID, if it was set when the event was created."""
     token_id: int

synapse/visibility.py

@@ -48,7 +48,13 @@ from synapse.logging.opentracing import trace
 from synapse.storage.controllers import StorageControllers
 from synapse.storage.databases.main import DataStore
 from synapse.synapse_rust.events import event_visible_to_server
-from synapse.types import RetentionPolicy, StateMap, StrCollection, get_domain_from_id
+from synapse.types import (
+    RetentionPolicy,
+    StateMap,
+    StrCollection,
+    UserID,
+    get_domain_from_id,
+)
 from synapse.types.state import StateFilter
 from synapse.util import Clock
 
@@ -106,8 +112,29 @@ async def filter_events_for_client(
         of `user_id` at each event.
     """
     # Filter out events that have been soft failed so that we don't relay them
-    # to clients.
+    # to clients, unless they're a server admin and want that to happen.
     events_before_filtering = events
+    client_config = await storage.main.get_admin_client_config_for_user(user_id)
+    if (
+        filter_send_to_client
+        and (
+            client_config.return_soft_failed_events
+            or client_config.return_policy_server_spammy_events
+        )
+        and await storage.main.is_server_admin(UserID.from_string(user_id))
+    ):
+        # `return_soft_failed_events` implies `return_policy_server_spammy_events`, so
+        # we want to check when they've asked for *just* `return_policy_server_spammy_events`
+        if not client_config.return_soft_failed_events:
+            events = [
+                e
+                for e in events
+                if not e.internal_metadata.is_soft_failed()
+                or e.internal_metadata.policy_server_spammy
+            ]
+        else:
+            events = events_before_filtering
+    else:
         events = [e for e in events if not e.internal_metadata.is_soft_failed()]
     if len(events_before_filtering) != len(events):
         if filtered_event_logger.isEnabledFor(logging.DEBUG):

tests/events/test_utils.py

@@ -34,11 +34,13 @@ from synapse.events.utils import (
     _split_field,
     clone_event,
     copy_and_fixup_power_levels_contents,
+    format_event_raw,
+    make_config_for_admin,
     maybe_upsert_event_field,
     prune_event,
     serialize_event,
 )
-from synapse.types import JsonDict
+from synapse.types import JsonDict, create_requester
 from synapse.util.frozenutils import freeze
 
 
@@ -49,7 +51,13 @@ def MockEvent(**kwargs: Any) -> EventBase:
         kwargs["type"] = "fake_type"
     if "content" not in kwargs:
         kwargs["content"] = {}
-    return make_event_from_dict(kwargs)
+
+    # Move internal metadata out so we can call make_event properly
+    internal_metadata = kwargs.get("internal_metadata")
+    if internal_metadata is not None:
+        kwargs.pop("internal_metadata")
+
+    return make_event_from_dict(kwargs, internal_metadata_dict=internal_metadata)
 
 
 class TestMaybeUpsertEventField(stdlib_unittest.TestCase):
@@ -639,9 +647,18 @@ class CloneEventTestCase(stdlib_unittest.TestCase):
 
 
 class SerializeEventTestCase(stdlib_unittest.TestCase):
-    def serialize(self, ev: EventBase, fields: Optional[List[str]]) -> JsonDict:
+    def serialize(
+        self,
+        ev: EventBase,
+        fields: Optional[List[str]],
+        include_admin_metadata: bool = False,
+    ) -> JsonDict:
         return serialize_event(
-            ev, 1479807801915, config=SerializeEventConfig(only_event_fields=fields)
+            ev,
+            1479807801915,
+            config=SerializeEventConfig(
+                only_event_fields=fields, include_admin_metadata=include_admin_metadata
+            ),
         )
 
     def test_event_fields_works_with_keys(self) -> None:
@@ -760,6 +777,78 @@ class SerializeEventTestCase(stdlib_unittest.TestCase):
                 ["room_id", 4],  # type: ignore[list-item]
             )
 
+    def test_default_serialize_config_excludes_admin_metadata(self) -> None:
+        # We just really don't want this to be set to True accidentally
+        self.assertFalse(SerializeEventConfig().include_admin_metadata)
+
+    def test_event_flagged_for_admins(self) -> None:
+        # Default behaviour should be *not* to include it
+        self.assertEqual(
+            self.serialize(
+                MockEvent(
+                    type="foo",
+                    event_id="test",
+                    room_id="!foo:bar",
+                    content={"foo": "bar"},
+                    internal_metadata={"soft_failed": True},
+                ),
+                [],
+            ),
+            {
+                "type": "foo",
+                "event_id": "test",
+                "room_id": "!foo:bar",
+                "content": {"foo": "bar"},
+                "unsigned": {},
+            },
+        )
+
+        # When asked though, we should set it
+        self.assertEqual(
+            self.serialize(
+                MockEvent(
+                    type="foo",
+                    event_id="test",
+                    room_id="!foo:bar",
+                    content={"foo": "bar"},
+                    internal_metadata={"soft_failed": True},
+                ),
+                [],
+                True,
+            ),
+            {
+                "type": "foo",
+                "event_id": "test",
+                "room_id": "!foo:bar",
+                "content": {"foo": "bar"},
+                "unsigned": {"io.element.synapse.soft_failed": True},
+            },
+        )
+
+    def test_make_serialize_config_for_admin_retains_other_fields(self) -> None:
+        non_default_config = SerializeEventConfig(
+            include_admin_metadata=False,  # should be True in a moment
+            as_client_event=False,  # default True
+            event_format=format_event_raw,  # default format_event_for_client_v1
+            requester=create_requester("@example:example.org"),  # default None
+            only_event_fields=["foo"],  # default None
+            include_stripped_room_state=True,  # default False
+        )
+        admin_config = make_config_for_admin(non_default_config)
+        self.assertEqual(
+            admin_config.as_client_event, non_default_config.as_client_event
+        )
+        self.assertEqual(admin_config.event_format, non_default_config.event_format)
+        self.assertEqual(admin_config.requester, non_default_config.requester)
+        self.assertEqual(
+            admin_config.only_event_fields, non_default_config.only_event_fields
+        )
+        self.assertEqual(
+            admin_config.include_stripped_room_state,
+            admin_config.include_stripped_room_state,
+        )
+        self.assertTrue(admin_config.include_admin_metadata)
+
 
 class CopyPowerLevelsContentTestCase(stdlib_unittest.TestCase):
     def setUp(self) -> None:

tests/rest/client/test_relations.py

@@ -1181,7 +1181,7 @@ class BundledAggregationsTestCase(BaseRelationsTestCase):
                 bundled_aggregations,
             )
 
-        self._test_bundled_aggregations(RelationTypes.REFERENCE, assert_annotations, 6)
+        self._test_bundled_aggregations(RelationTypes.REFERENCE, assert_annotations, 8)
 
     def test_thread(self) -> None:
         """
@@ -1226,21 +1226,21 @@ class BundledAggregationsTestCase(BaseRelationsTestCase):
 
         # The "user" sent the root event and is making queries for the bundled
         # aggregations: they have participated.
-        self._test_bundled_aggregations(RelationTypes.THREAD, _gen_assert(True), 6)
+        self._test_bundled_aggregations(RelationTypes.THREAD, _gen_assert(True), 9)
         # The "user2" sent replies in the thread and is making queries for the
         # bundled aggregations: they have participated.
         #
         # Note that this re-uses some cached values, so the total number of
         # queries is much smaller.
         self._test_bundled_aggregations(
-            RelationTypes.THREAD, _gen_assert(True), 3, access_token=self.user2_token
+            RelationTypes.THREAD, _gen_assert(True), 6, access_token=self.user2_token
         )
 
         # A user with no interactions with the thread: they have not participated.
         user3_id, user3_token = self._create_user("charlie")
         self.helper.join(self.room, user=user3_id, tok=user3_token)
         self._test_bundled_aggregations(
-            RelationTypes.THREAD, _gen_assert(False), 3, access_token=user3_token
+            RelationTypes.THREAD, _gen_assert(False), 6, access_token=user3_token
         )
 
     def test_thread_with_bundled_aggregations_for_latest(self) -> None:
@@ -1287,7 +1287,7 @@ class BundledAggregationsTestCase(BaseRelationsTestCase):
                 bundled_aggregations["latest_event"].get("unsigned"),
             )
 
-        self._test_bundled_aggregations(RelationTypes.THREAD, assert_thread, 6)
+        self._test_bundled_aggregations(RelationTypes.THREAD, assert_thread, 9)
 
     def test_nested_thread(self) -> None:
         """