Allow selecting "prejoin" events by state keys (#14642)

* Declare new config

* Parse new config

* Read new config

* Don't use trial/our TestCase where it's not needed

Before:

```
$ time trial tests/events/test_utils.py > /dev/null

real	0m2.277s
user	0m2.186s
sys	0m0.083s
```

After:
```
$ time trial tests/events/test_utils.py > /dev/null

real	0m0.566s
user	0m0.508s
sys	0m0.056s
```

* Helper to upsert to event fields

without exceeding size limits.

* Use helper when adding invite/knock state

Now that we allow admins to include events in prejoin room state with
arbitrary state keys, be a good Matrix citizen and ensure they don't
accidentally create an oversized event.

* Changelog

* Move StateFilter tests

should have done this in #14668

* Add extra methods to StateFilter

* Use StateFilter

* Ensure test file enforces typed defs; alphabetise

* Workaround surprising get_current_state_ids

* Whoops, fix mypy

Author: David Robertson

changelog.d/14642.feature

@@ -0,0 +1 @@
+Allow selecting "prejoin" events by state keys in addition to event types.

docs/usage/configuration/config_documentation.md

@@ -2501,32 +2501,53 @@ Config settings related to the client/server API
 ---
 ### `room_prejoin_state`
 
-Controls for the state that is shared with users who receive an invite
-to a room. By default, the following state event types are shared with users who
-receive invites to the room:
-- m.room.join_rules
-- m.room.canonical_alias
-- m.room.avatar
-- m.room.encryption
-- m.room.name
-- m.room.create
-- m.room.topic
+This setting controls the state that is shared with users upon receiving an
+invite to a room, or in reply to a knock on a room. By default, the following
+state events are shared with users:
+
+- `m.room.join_rules`
+- `m.room.canonical_alias`
+- `m.room.avatar`
+- `m.room.encryption`
+- `m.room.name`
+- `m.room.create`
+- `m.room.topic`
 
 To change the default behavior, use the following sub-options:
-* `disable_default_event_types`: set to true to disable the above defaults. If this
-   is enabled, only the event types listed in `additional_event_types` are shared.
-   Defaults to false.
-* `additional_event_types`: Additional state event types to share with users when they are invited
-   to a room. By default, this list is empty (so only the default event types are shared).
+* `disable_default_event_types`: boolean. Set to `true` to disable the above 
+  defaults. If this is enabled, only the event types listed in
+  `additional_event_types` are shared. Defaults to `false`.
+* `additional_event_types`: A list of additional state events to include in the 
+  events to be shared. By default, this list is empty (so only the default event 
+  types are shared).
+
+  Each entry in this list should be either a single string or a list of two
+  strings. 
+  * A standalone string `t` represents all events with type `t` (i.e.
+    with no restrictions on state keys).
+  * A pair of strings `[t, s]` represents a single event with type `t` and 
+    state key `s`. The same type can appear in two entries with different state
+    keys: in this situation, both state keys are included in prejoin state.
 
 Example configuration:
 ```yaml
 room_prejoin_state:
-   disable_default_event_types: true
+   disable_default_event_types: false
    additional_event_types:
-     - org.example.custom.event.type
-     - m.room.join_rules
+     # Share all events of type `org.example.custom.event.typeA`
+     - org.example.custom.event.typeA
+     # Share only events of type `org.example.custom.event.typeB` whose
+     # state_key is "foo"
+     - ["org.example.custom.event.typeB", "foo"]
+     # Share only events of type `org.example.custom.event.typeC` whose
+     # state_key is "bar" or "baz"
+     - ["org.example.custom.event.typeC", "bar"]
+     - ["org.example.custom.event.typeC", "baz"]
 ```
+
+*Changed in Synapse 1.74:* admins can filter the events in prejoin state based
+on their state key.
+
 ---
 ### `track_puppeted_user_ips`
 

mypy.ini

@@ -89,6 +89,12 @@ disallow_untyped_defs = False
 [mypy-tests.*]
 disallow_untyped_defs = False
 
+[mypy-tests.config.test_api]
+disallow_untyped_defs = True
+
+[mypy-tests.federation.transport.test_client]
+disallow_untyped_defs = True
+
 [mypy-tests.handlers.test_sso]
 disallow_untyped_defs = True
 
@@ -101,7 +107,7 @@ disallow_untyped_defs = True
 [mypy-tests.push.test_bulk_push_rule_evaluator]
 disallow_untyped_defs = True
 
-[mypy-tests.test_server]
+[mypy-tests.rest.*]
 disallow_untyped_defs = True
 
 [mypy-tests.state.test_profile]
@@ -110,10 +116,10 @@ disallow_untyped_defs = True
 [mypy-tests.storage.*]
 disallow_untyped_defs = True
 
-[mypy-tests.rest.*]
+[mypy-tests.test_server]
 disallow_untyped_defs = True
 
-[mypy-tests.federation.transport.test_client]
+[mypy-tests.types.*]
 disallow_untyped_defs = True
 
 [mypy-tests.util.caches.*]

synapse/config/_util.py

@@ -33,6 +33,9 @@ def validate_config(
         config: the configuration value to be validated
         config_path: the path within the config file. This will be used as a basis
            for the error message.
+
+    Raises:
+        ConfigError, if validation fails.
     """
     try:
         jsonschema.validate(config, json_schema)

synapse/config/api.py

@@ -13,29 +13,34 @@
 # limitations under the License.
 
 import logging
-from typing import Any, Iterable
+from typing import Any, Iterable, Optional, Tuple
 
 from synapse.api.constants import EventTypes
 from synapse.config._base import Config, ConfigError
 from synapse.config._util import validate_config
 from synapse.types import JsonDict
+from synapse.types.state import StateFilter
 
 logger = logging.getLogger(__name__)
 
 
 class ApiConfig(Config):
     section = "api"
 
+    room_prejoin_state: StateFilter
+    track_puppetted_users_ips: bool
+
     def read_config(self, config: JsonDict, **kwargs: Any) -> None:
         validate_config(_MAIN_SCHEMA, config, ())
-        self.room_prejoin_state = list(self._get_prejoin_state_types(config))
+        self.room_prejoin_state = StateFilter.from_types(
+            self._get_prejoin_state_entries(config)
+        )
         self.track_puppeted_user_ips = config.get("track_puppeted_user_ips", False)
 
-    def _get_prejoin_state_types(self, config: JsonDict) -> Iterable[str]:
-        """Get the event types to include in the prejoin state
-
-        Parses the config and returns an iterable of the event types to be included.
-        """
+    def _get_prejoin_state_entries(
+        self, config: JsonDict
+    ) -> Iterable[Tuple[str, Optional[str]]]:
+        """Get the event types and state keys to include in the prejoin state."""
         room_prejoin_state_config = config.get("room_prejoin_state") or {}
 
         # backwards-compatibility support for room_invite_state_types
@@ -50,33 +55,39 @@ def _get_prejoin_state_types(self, config: JsonDict) -> Iterable[str]:
 
             logger.warning(_ROOM_INVITE_STATE_TYPES_WARNING)
 
-            yield from config["room_invite_state_types"]
+            for event_type in config["room_invite_state_types"]:
+                yield event_type, None
             return
 
         if not room_prejoin_state_config.get("disable_default_event_types"):
-            yield from _DEFAULT_PREJOIN_STATE_TYPES
+            yield from _DEFAULT_PREJOIN_STATE_TYPES_AND_STATE_KEYS
 
-        yield from room_prejoin_state_config.get("additional_event_types", [])
+        for entry in room_prejoin_state_config.get("additional_event_types", []):
+            if isinstance(entry, str):
+                yield entry, None
+            else:
+                yield entry
 
 
 _ROOM_INVITE_STATE_TYPES_WARNING = """\
 WARNING: The 'room_invite_state_types' configuration setting is now deprecated,
 and replaced with 'room_prejoin_state'. New features may not work correctly
-unless 'room_invite_state_types' is removed. See the sample configuration file for
-details of 'room_prejoin_state'.
+unless 'room_invite_state_types' is removed. See the config documentation at
+    https://matrix-org.github.io/synapse/latest/usage/configuration/config_documentation.html#room_prejoin_state
+for details of 'room_prejoin_state'.
 --------------------------------------------------------------------------------
 """
 
-_DEFAULT_PREJOIN_STATE_TYPES = [
-    EventTypes.JoinRules,
-    EventTypes.CanonicalAlias,
-    EventTypes.RoomAvatar,
-    EventTypes.RoomEncryption,
-    EventTypes.Name,
+_DEFAULT_PREJOIN_STATE_TYPES_AND_STATE_KEYS = [
+    (EventTypes.JoinRules, ""),
+    (EventTypes.CanonicalAlias, ""),
+    (EventTypes.RoomAvatar, ""),
+    (EventTypes.RoomEncryption, ""),
+    (EventTypes.Name, ""),
     # Per MSC1772.
-    EventTypes.Create,
+    (EventTypes.Create, ""),
     # Per MSC3173.
-    EventTypes.Topic,
+    (EventTypes.Topic, ""),
 ]
 
 
@@ -90,7 +101,17 @@ def _get_prejoin_state_types(self, config: JsonDict) -> Iterable[str]:
                 "disable_default_event_types": {"type": "boolean"},
                 "additional_event_types": {
                     "type": "array",
-                    "items": {"type": "string"},
+                    "items": {
+                        "oneOf": [
+                            {"type": "string"},
+                            {
+                                "type": "array",
+                                "items": {"type": "string"},
+                                "minItems": 2,
+                                "maxItems": 2,
+                            },
+                        ],
+                    },
                 },
             },
         },

synapse/events/utils.py

@@ -28,8 +28,14 @@
 )
 
 import attr
+from canonicaljson import encode_canonical_json
 
-from synapse.api.constants import EventContentFields, EventTypes, RelationTypes
+from synapse.api.constants import (
+    MAX_PDU_SIZE,
+    EventContentFields,
+    EventTypes,
+    RelationTypes,
+)
 from synapse.api.errors import Codes, SynapseError
 from synapse.api.room_versions import RoomVersion
 from synapse.types import JsonDict
@@ -674,3 +680,27 @@ def validate_canonicaljson(value: Any) -> None:
     elif not isinstance(value, (bool, str)) and value is not None:
         # Other potential JSON values (bool, None, str) are safe.
         raise SynapseError(400, "Unknown JSON value", Codes.BAD_JSON)
+
+
+def maybe_upsert_event_field(
+    event: EventBase, container: JsonDict, key: str, value: object
+) -> bool:
+    """Upsert an event field, but only if this doesn't make the event too large.
+
+    Returns true iff the upsert took place.
+    """
+    if key in container:
+        old_value: object = container[key]
+        container[key] = value
+        # NB: here and below, we assume that passing a non-None `time_now` argument to
+        # get_pdu_json doesn't increase the size of the encoded result.
+        upsert_okay = len(encode_canonical_json(event.get_pdu_json())) <= MAX_PDU_SIZE
+        if not upsert_okay:
+            container[key] = old_value
+    else:
+        container[key] = value
+        upsert_okay = len(encode_canonical_json(event.get_pdu_json())) <= MAX_PDU_SIZE
+        if not upsert_okay:
+            del container[key]
+
+    return upsert_okay

synapse/handlers/message.py

@@ -50,6 +50,7 @@
 from synapse.events import EventBase, relation_from_event
 from synapse.events.builder import EventBuilder
 from synapse.events.snapshot import EventContext
+from synapse.events.utils import maybe_upsert_event_field
 from synapse.events.validator import EventValidator
 from synapse.handlers.directory import DirectoryHandler
 from synapse.logging import opentracing
@@ -1739,12 +1740,15 @@ async def persist_and_notify_client_events(
 
             if event.type == EventTypes.Member:
                 if event.content["membership"] == Membership.INVITE:
-                    event.unsigned[
-                        "invite_room_state"
-                    ] = await self.store.get_stripped_room_state_from_event_context(
-                        context,
-                        self.room_prejoin_state_types,
-                        membership_user_id=event.sender,
+                    maybe_upsert_event_field(
+                        event,
+                        event.unsigned,
+                        "invite_room_state",
+                        await self.store.get_stripped_room_state_from_event_context(
+                            context,
+                            self.room_prejoin_state_types,
+                            membership_user_id=event.sender,
+                        ),
                     )
 
                     invitee = UserID.from_string(event.state_key)
@@ -1762,11 +1766,14 @@ async def persist_and_notify_client_events(
                         event.signatures.update(returned_invite.signatures)
 
                 if event.content["membership"] == Membership.KNOCK:
-                    event.unsigned[
-                        "knock_room_state"
-                    ] = await self.store.get_stripped_room_state_from_event_context(
-                        context,
-                        self.room_prejoin_state_types,
+                    maybe_upsert_event_field(
+                        event,
+                        event.unsigned,
+                        "knock_room_state",
+                        await self.store.get_stripped_room_state_from_event_context(
+                            context,
+                            self.room_prejoin_state_types,
+                        ),
                     )
 
             if event.type == EventTypes.Redaction: