Reduce recorder overhead when entity filter is empty (#119631)

homeassistant/components/recorder/__init__.py

@@ -127,15 +127,15 @@ def is_entity_recorded(hass: HomeAssistant, entity_id: str) -> bool:
 
     Async friendly.
     """
-    if DATA_INSTANCE not in hass.data:
-        return False
-    return hass.data[DATA_INSTANCE].entity_filter(entity_id)
+    instance = get_instance(hass)
+    return instance.entity_filter is None or instance.entity_filter(entity_id)
 
 
 async def async_setup(hass: HomeAssistant, config: ConfigType) -> bool:
     """Set up the recorder."""
     conf = config[DOMAIN]
-    entity_filter = convert_include_exclude_filter(conf).get_filter()
+    _filter = convert_include_exclude_filter(conf)
+    entity_filter = None if _filter.empty_filter else _filter.get_filter()
     auto_purge = conf[CONF_AUTO_PURGE]
     auto_repack = conf[CONF_AUTO_REPACK]
     keep_days = conf[CONF_PURGE_KEEP_DAYS]

homeassistant/components/recorder/core.py

@@ -178,7 +178,7 @@ class Recorder(threading.Thread):
         uri: str,
         db_max_retries: int,
         db_retry_wait: int,
-        entity_filter: Callable[[str], bool],
+        entity_filter: Callable[[str], bool] | None,
         exclude_event_types: set[EventType[Any] | str],
     ) -> None:
         """Initialize the recorder."""
@@ -318,7 +318,10 @@ class Recorder(threading.Thread):
             if event.event_type in exclude_event_types:
                 return
 
-            if (entity_id := event.data.get(ATTR_ENTITY_ID)) is None:
+            if (
+                entity_filter is None
+                or (entity_id := event.data.get(ATTR_ENTITY_ID)) is None
+            ):
                 queue_put(event)
                 return
 

homeassistant/components/recorder/purge.py

@@ -645,7 +645,7 @@ def _purge_filtered_data(instance: Recorder, session: Session) -> bool:
         for (metadata_id, entity_id) in session.query(
             StatesMeta.metadata_id, StatesMeta.entity_id
         ).all()
-        if not entity_filter(entity_id)
+        if entity_filter and not entity_filter(entity_id)
     ]
     if excluded_metadata_ids:
         has_more_states_to_purge = _purge_filtered_states(
@@ -765,7 +765,9 @@ def _purge_filtered_events(
 
 @retryable_database_job("purge_entity_data")
 def purge_entity_data(
-    instance: Recorder, entity_filter: Callable[[str], bool], purge_before: datetime
+    instance: Recorder,
+    entity_filter: Callable[[str], bool] | None,
+    purge_before: datetime,
 ) -> bool:
     """Purge states and events of specified entities."""
     database_engine = instance.database_engine
@@ -777,7 +779,7 @@ def purge_entity_data(
             for (metadata_id, entity_id) in session.query(
                 StatesMeta.metadata_id, StatesMeta.entity_id
             ).all()
-            if entity_filter(entity_id)
+            if entity_filter and entity_filter(entity_id)
         ]
         _LOGGER.debug("Purging entity data for %s", selected_metadata_ids)
         if not selected_metadata_ids:

homeassistant/components/sensor/recorder.py

@@ -80,6 +80,7 @@ def _get_sensor_states(hass: HomeAssistant) -> list[State]:
     # We check for state class first before calling the filter
     # function as the filter function is much more expensive
     # than checking the state class
+    entity_filter = instance.entity_filter
     return [
         state
         for state in hass.states.all(DOMAIN)
@@ -88,7 +89,7 @@ def _get_sensor_states(hass: HomeAssistant) -> list[State]:
             type(state_class) is SensorStateClass
             or try_parse_enum(SensorStateClass, state_class)
         )
-        and instance.entity_filter(state.entity_id)
+        and (not entity_filter or entity_filter(state.entity_id))
     ]
 
 
@@ -680,6 +681,7 @@ def validate_statistics(
     sensor_entity_ids = {i.entity_id for i in sensor_states}
     sensor_statistic_ids = set(metadatas)
     instance = get_instance(hass)
+    entity_filter = instance.entity_filter
 
     for state in sensor_states:
         entity_id = state.entity_id
@@ -689,7 +691,7 @@ def validate_statistics(
         state_unit = state.attributes.get(ATTR_UNIT_OF_MEASUREMENT)
 
         if metadata := metadatas.get(entity_id):
-            if not instance.entity_filter(state.entity_id):
+            if entity_filter and not entity_filter(state.entity_id):
                 # Sensor was previously recorded, but no longer is
                 validation_result[entity_id].append(
                     statistics.ValidationIssue(
@@ -739,7 +741,7 @@ def validate_statistics(
                     )
                 )
         elif state_class is not None:
-            if not instance.entity_filter(state.entity_id):
+            if entity_filter and not entity_filter(state.entity_id):
                 # Sensor is not recorded
                 validation_result[entity_id].append(
                     statistics.ValidationIssue(