Create a separate filter object to do the actual filtering, so that we can

split the storage and management of filters from the actual filter code
and don't have to load a filter from the db each time we filter an event

1 files changed, 108 insertions, 112 deletions

diff --git a/synapse/api/filtering.py b/synapse/api/filtering.py
index e16c0e559f..b7e5d3222f 100644
--- a/synapse/api/filtering.py
+++ b/synapse/api/filtering.py
@@ -25,127 +25,25 @@ class Filtering(object):
         self.store = hs.get_datastore()
 
     def get_user_filter(self, user_localpart, filter_id):
-        return self.store.get_user_filter(user_localpart, filter_id)
+        result = self.store.get_user_filter(user_localpart, filter_id)
+        result.addCallback(Filter)
+        return result
 
     def add_user_filter(self, user_localpart, user_filter):
         self._check_valid_filter(user_filter)
         return self.store.add_user_filter(user_localpart, user_filter)
 
-    def filter_public_user_data(self, events, user, filter_id):
-        return self._filter_on_key(
-            events, user, filter_id, ["public_user_data"]
-        )
-
-    def filter_private_user_data(self, events, user, filter_id):
-        return self._filter_on_key(
-            events, user, filter_id, ["private_user_data"]
-        )
-
-    def filter_room_state(self, events, user, filter_id):
-        return self._filter_on_key(
-            events, user, filter_id, ["room", "state"]
-        )
-
-    def filter_room_events(self, events, user, filter_id):
-        return self._filter_on_key(
-            events, user, filter_id, ["room", "events"]
-        )
-
-    def filter_room_ephemeral(self, events, user, filter_id):
-        return self._filter_on_key(
-            events, user, filter_id, ["room", "ephemeral"]
-        )
-
     # TODO(paul): surely we should probably add a delete_user_filter or
     #   replace_user_filter at some point? There's no REST API specified for
     #   them however
 
-    @defer.inlineCallbacks
-    def _filter_on_key(self, events, user, filter_id, keys):
-        filter_json = yield self.get_user_filter(user.localpart, filter_id)
-        if not filter_json:
-            defer.returnValue(events)
-
-        try:
-            # extract the right definition from the filter
-            definition = filter_json
-            for key in keys:
-                definition = definition[key]
-            defer.returnValue(self._filter_with_definition(events, definition))
-        except KeyError:
-            # return all events if definition isn't specified.
-            defer.returnValue(events)
-
-    def _filter_with_definition(self, events, definition):
-        return [e for e in events if self._passes_definition(definition, e)]
-
-    def _passes_definition(self, definition, event):
-        """Check if the event passes through the given definition.
-
-        Args:
-            definition(dict): The definition to check against.
-            event(Event): The event to check.
-        Returns:
-            True if the event passes through the filter.
-        """
-        # Algorithm notes:
-        # For each key in the definition, check the event meets the criteria:
-        #   * For types: Literal match or prefix match (if ends with wildcard)
-        #   * For senders/rooms: Literal match only
-        #   * "not_" checks take presedence (e.g. if "m.*" is in both 'types'
-        #     and 'not_types' then it is treated as only being in 'not_types')
-
-        # room checks
-        if hasattr(event, "room_id"):
-            room_id = event.room_id
-            allow_rooms = definition.get("rooms", None)
-            reject_rooms = definition.get("not_rooms", None)
-            if reject_rooms and room_id in reject_rooms:
-                return False
-            if allow_rooms and room_id not in allow_rooms:
-                return False
-
-        # sender checks
-        if hasattr(event, "sender"):
-            # Should we be including event.state_key for some event types?
-            sender = event.sender
-            allow_senders = definition.get("senders", None)
-            reject_senders = definition.get("not_senders", None)
-            if reject_senders and sender in reject_senders:
-                return False
-            if allow_senders and sender not in allow_senders:
-                return False
-
-        # type checks
-        if "not_types" in definition:
-            for def_type in definition["not_types"]:
-                if self._event_matches_type(event, def_type):
-                    return False
-        if "types" in definition:
-            included = False
-            for def_type in definition["types"]:
-                if self._event_matches_type(event, def_type):
-                    included = True
-                    break
-            if not included:
-                return False
-
-        return True
-
-    def _event_matches_type(self, event, def_type):
-        if def_type.endswith("*"):
-            type_prefix = def_type[:-1]
-            return event.type.startswith(type_prefix)
-        else:
-            return event.type == def_type
-
-    def _check_valid_filter(self, user_filter):
+    def _check_valid_filter(self, user_filter_json):
         """Check if the provided filter is valid.
 
         This inspects all definitions contained within the filter.
 
         Args:
-            user_filter(dict): The filter
+            user_filter_json(dict): The filter
         Raises:
             SynapseError: If the filter is not valid.
         """
@@ -162,13 +60,13 @@ class Filtering(object):
         ]
 
         for key in top_level_definitions:
-            if key in user_filter:
-                self._check_definition(user_filter[key])
+            if key in user_filter_json:
+                self._check_definition(user_filter_json[key])
 
-        if "room" in user_filter:
+        if "room" in user_filter_json:
             for key in room_level_definitions:
-                if key in user_filter["room"]:
-                    self._check_definition(user_filter["room"][key])
+                if key in user_filter_json["room"]:
+                    self._check_definition(user_filter_json["room"][key])
 
     def _check_definition(self, definition):
         """Check if the provided definition is valid.
@@ -237,3 +135,101 @@ class Filtering(object):
         if ("bundle_updates" in definition and
                 type(definition["bundle_updates"]) != bool):
             raise SynapseError(400, "Bad bundle_updates: expected bool.")
+
+
+class Filter(object):
+    def __init__(self, filter_json):
+        self.filter_json = filter_json
+
+    def filter_public_user_data(self, events):
+        return self._filter_on_key(events, ["public_user_data"])
+
+    def filter_private_user_data(self, events):
+        return self._filter_on_key(events, ["private_user_data"])
+
+    def filter_room_state(self, events):
+        return self._filter_on_key(events, ["room", "state"])
+
+    def filter_room_events(self, events):
+        return self._filter_on_key(events, ["room", "events"])
+
+    def filter_room_ephemeral(self, events):
+        return self._filter_on_key(events, ["room", "ephemeral"])
+
+    def _filter_on_key(self, events, keys):
+        filter_json = self.filter_json
+        if not filter_json:
+            return events
+
+        try:
+            # extract the right definition from the filter
+            definition = filter_json
+            for key in keys:
+                definition = definition[key]
+            return self._filter_with_definition(events, definition)
+        except KeyError:
+            # return all events if definition isn't specified.
+            return events
+
+    def _filter_with_definition(self, events, definition):
+        return [e for e in events if self._passes_definition(definition, e)]
+
+    def _passes_definition(self, definition, event):
+        """Check if the event passes through the given definition.
+
+        Args:
+            definition(dict): The definition to check against.
+            event(Event): The event to check.
+        Returns:
+            True if the event passes through the filter.
+        """
+        # Algorithm notes:
+        # For each key in the definition, check the event meets the criteria:
+        #   * For types: Literal match or prefix match (if ends with wildcard)
+        #   * For senders/rooms: Literal match only
+        #   * "not_" checks take presedence (e.g. if "m.*" is in both 'types'
+        #     and 'not_types' then it is treated as only being in 'not_types')
+
+        # room checks
+        if hasattr(event, "room_id"):
+            room_id = event.room_id
+            allow_rooms = definition.get("rooms", None)
+            reject_rooms = definition.get("not_rooms", None)
+            if reject_rooms and room_id in reject_rooms:
+                return False
+            if allow_rooms and room_id not in allow_rooms:
+                return False
+
+        # sender checks
+        if hasattr(event, "sender"):
+            # Should we be including event.state_key for some event types?
+            sender = event.sender
+            allow_senders = definition.get("senders", None)
+            reject_senders = definition.get("not_senders", None)
+            if reject_senders and sender in reject_senders:
+                return False
+            if allow_senders and sender not in allow_senders:
+                return False
+
+        # type checks
+        if "not_types" in definition:
+            for def_type in definition["not_types"]:
+                if self._event_matches_type(event, def_type):
+                    return False
+        if "types" in definition:
+            included = False
+            for def_type in definition["types"]:
+                if self._event_matches_type(event, def_type):
+                    included = True
+                    break
+            if not included:
+                return False
+
+        return True
+
+    def _event_matches_type(self, event, def_type):
+        if def_type.endswith("*"):
+            type_prefix = def_type[:-1]
+            return event.type.startswith(type_prefix)
+        else:
+            return event.type == def_type