Reduce size of load query to prime event_types and states_meta at startup (#89677)

homeassistant/components/recorder/core.py

@@ -793,13 +793,13 @@ class Recorder(threading.Thread):
         until its primed.
         """
         assert self.event_session is not None
-        if hashes := [
+        if hashes := {
             StateAttributes.hash_shared_attrs_bytes(shared_attrs_bytes)
             for event in events
             if (
                 shared_attrs_bytes := self._serialize_state_attributes_from_event(event)
             )
-        ]:
+        }:
             with self.event_session.no_autoflush:
                 for hash_chunk in chunked(hashes, SQLITE_MAX_BIND_VARS):
                     for id_, shared_attrs in self.event_session.execute(
@@ -815,11 +815,11 @@ class Recorder(threading.Thread):
         the data in the database for every event until its primed.
         """
         assert self.event_session is not None
-        if hashes := [
+        if hashes := {
             EventData.hash_shared_data_bytes(shared_event_bytes)
             for event in events
             if (shared_event_bytes := self._serialize_event_data_from_event(event))
-        ]:
+        }:
             with self.event_session.no_autoflush:
                 for hash_chunk in chunked(hashes, SQLITE_MAX_BIND_VARS):
                     for id_, shared_data in self.event_session.execute(

homeassistant/components/recorder/table_managers/event_types.py

@@ -9,8 +9,10 @@ from sqlalchemy.orm.session import Session
 
 from homeassistant.core import Event
 
+from ..const import SQLITE_MAX_BIND_VARS
 from ..db_schema import EventTypes
 from ..queries import find_event_type_ids
+from ..util import chunked
 
 CACHE_SIZE = 2048
 
@@ -27,7 +29,7 @@ class EventTypeManager:
     def load(self, events: list[Event], session: Session) -> None:
         """Load the event_type to event_type_ids mapping into memory."""
         self.get_many(
-            (event.event_type for event in events if event.event_type is not None),
+            {event.event_type for event in events if event.event_type is not None},
             session,
         )
 
@@ -51,12 +53,13 @@ class EventTypeManager:
             return results
 
         with session.no_autoflush:
-            for event_type_id, event_type in session.execute(
+            for missing_chunk in chunked(missing, SQLITE_MAX_BIND_VARS):
-                find_event_type_ids(missing)
+                for event_type_id, event_type in session.execute(
-            ):
+                    find_event_type_ids(missing_chunk)
-                results[event_type] = self._id_map[event_type] = cast(
+                ):
-                    int, event_type_id
+                    results[event_type] = self._id_map[event_type] = cast(
-                )
+                        int, event_type_id
+                    )
 
         return results
 

homeassistant/components/recorder/table_managers/states_meta.py

@@ -9,8 +9,10 @@ from sqlalchemy.orm.session import Session
 
 from homeassistant.core import Event
 
+from ..const import SQLITE_MAX_BIND_VARS
 from ..db_schema import StatesMeta
 from ..queries import find_all_states_metadata_ids, find_states_metadata_ids
+from ..util import chunked
 
 CACHE_SIZE = 8192
 
@@ -27,11 +29,11 @@ class StatesMetaManager:
     def load(self, events: list[Event], session: Session) -> None:
         """Load the entity_id to metadata_id mapping into memory."""
         self.get_many(
-            (
+            {
                 event.data["new_state"].entity_id
                 for event in events
                 if event.data.get("new_state") is not None
-            ),
+            },
             session,
         )
 
@@ -60,10 +62,13 @@ class StatesMetaManager:
             return results
 
         with session.no_autoflush:
-            for metadata_id, entity_id in session.execute(
+            for missing_chunk in chunked(missing, SQLITE_MAX_BIND_VARS):
-                find_states_metadata_ids(missing)
+                for metadata_id, entity_id in session.execute(
-            ):
+                    find_states_metadata_ids(missing_chunk)
-                results[entity_id] = self._id_map[entity_id] = cast(int, metadata_id)
+                ):
+                    results[entity_id] = self._id_map[entity_id] = cast(
+                        int, metadata_id
+                    )
 
         return results