Chunk purging attributes and data ids for old SQLite versions (#104296)

homeassistant/components/recorder/purge.py

@@ -41,7 +41,7 @@ from .queries import (
     find_statistics_runs_to_purge,
 )
 from .repack import repack_database
-from .util import chunked, retryable_database_job, session_scope
+from .util import chunked_or_all, retryable_database_job, session_scope
 
 if TYPE_CHECKING:
     from . import Recorder
@@ -283,12 +283,16 @@ def _select_event_data_ids_to_purge(
 
 
 def _select_unused_attributes_ids(
-    session: Session, attributes_ids: set[int], database_engine: DatabaseEngine
+    instance: Recorder,
+    session: Session,
+    attributes_ids: set[int],
+    database_engine: DatabaseEngine,
 ) -> set[int]:
     """Return a set of attributes ids that are not used by any states in the db."""
     if not attributes_ids:
         return set()
 
+    seen_ids: set[int] = set()
     if not database_engine.optimizer.slow_range_in_select:
         #
         # SQLite has a superior query optimizer for the distinct query below as it uses
@@ -303,12 +307,17 @@ def _select_unused_attributes_ids(
         #   (136723);
         # ...Using index
         #
-        seen_ids = {
-            state[0]
-            for state in session.execute(
-                attributes_ids_exist_in_states_with_fast_in_distinct(attributes_ids)
-            ).all()
-        }
+        for attributes_ids_chunk in chunked_or_all(
+            attributes_ids, instance.max_bind_vars
+        ):
+            seen_ids.update(
+                state[0]
+                for state in session.execute(
+                    attributes_ids_exist_in_states_with_fast_in_distinct(
+                        attributes_ids_chunk
+                    )
+                ).all()
+            )
     else:
         #
         # This branch is for DBMS that cannot optimize the distinct query well and has
@@ -334,7 +343,6 @@ def _select_unused_attributes_ids(
         # We now break the query into groups of 100 and use a lambda_stmt to ensure
         # that the query is only cached once.
         #
-        seen_ids = set()
         groups = [iter(attributes_ids)] * 100
         for attr_ids in zip_longest(*groups, fillvalue=None):
             seen_ids |= {
@@ -361,29 +369,33 @@ def _purge_unused_attributes_ids(
     database_engine = instance.database_engine
     assert database_engine is not None
     if unused_attribute_ids_set := _select_unused_attributes_ids(
-        session, attributes_ids_batch, database_engine
+        instance, session, attributes_ids_batch, database_engine
     ):
         _purge_batch_attributes_ids(instance, session, unused_attribute_ids_set)
 
 
 def _select_unused_event_data_ids(
-    session: Session, data_ids: set[int], database_engine: DatabaseEngine
+    instance: Recorder,
+    session: Session,
+    data_ids: set[int],
+    database_engine: DatabaseEngine,
 ) -> set[int]:
     """Return a set of event data ids that are not used by any events in the db."""
     if not data_ids:
         return set()
 
+    seen_ids: set[int] = set()
     # See _select_unused_attributes_ids for why this function
     # branches for non-sqlite databases.
     if not database_engine.optimizer.slow_range_in_select:
-        seen_ids = {
-            state[0]
-            for state in session.execute(
-                data_ids_exist_in_events_with_fast_in_distinct(data_ids)
-            ).all()
-        }
+        for data_ids_chunk in chunked_or_all(data_ids, instance.max_bind_vars):
+            seen_ids.update(
+                state[0]
+                for state in session.execute(
+                    data_ids_exist_in_events_with_fast_in_distinct(data_ids_chunk)
+                ).all()
+            )
     else:
-        seen_ids = set()
         groups = [iter(data_ids)] * 100
         for data_ids_group in zip_longest(*groups, fillvalue=None):
             seen_ids |= {
@@ -404,7 +416,7 @@ def _purge_unused_data_ids(
     database_engine = instance.database_engine
     assert database_engine is not None
     if unused_data_ids_set := _select_unused_event_data_ids(
-        session, data_ids_batch, database_engine
+        instance, session, data_ids_batch, database_engine
     ):
         _purge_batch_data_ids(instance, session, unused_data_ids_set)
 
@@ -519,7 +531,7 @@ def _purge_batch_attributes_ids(
     instance: Recorder, session: Session, attributes_ids: set[int]
 ) -> None:
     """Delete old attributes ids in batches of max_bind_vars."""
-    for attributes_ids_chunk in chunked(attributes_ids, instance.max_bind_vars):
+    for attributes_ids_chunk in chunked_or_all(attributes_ids, instance.max_bind_vars):
         deleted_rows = session.execute(
             delete_states_attributes_rows(attributes_ids_chunk)
         )
@@ -533,7 +545,7 @@ def _purge_batch_data_ids(
     instance: Recorder, session: Session, data_ids: set[int]
 ) -> None:
     """Delete old event data ids in batches of max_bind_vars."""
-    for data_ids_chunk in chunked(data_ids, instance.max_bind_vars):
+    for data_ids_chunk in chunked_or_all(data_ids, instance.max_bind_vars):
         deleted_rows = session.execute(delete_event_data_rows(data_ids_chunk))
         _LOGGER.debug("Deleted %s data events", deleted_rows)
 
@@ -694,7 +706,10 @@ def _purge_filtered_states(
     # we will need to purge them here.
     _purge_event_ids(session, filtered_event_ids)
     unused_attribute_ids_set = _select_unused_attributes_ids(
-        session, {id_ for id_ in attributes_ids if id_ is not None}, database_engine
+        instance,
+        session,
+        {id_ for id_ in attributes_ids if id_ is not None},
+        database_engine,
     )
     _purge_batch_attributes_ids(instance, session, unused_attribute_ids_set)
     return False
@@ -741,7 +756,7 @@ def _purge_filtered_events(
         _purge_state_ids(instance, session, state_ids)
     _purge_event_ids(session, event_ids_set)
     if unused_data_ids_set := _select_unused_event_data_ids(
-        session, set(data_ids), database_engine
+        instance, session, set(data_ids), database_engine
     ):
         _purge_batch_data_ids(instance, session, unused_data_ids_set)
     return False

homeassistant/components/recorder/util.py

@@ -1,7 +1,7 @@
 """SQLAlchemy util functions."""
 from __future__ import annotations
 
-from collections.abc import Callable, Generator, Iterable, Sequence
+from collections.abc import Callable, Collection, Generator, Iterable, Sequence
 from contextlib import contextmanager
 from datetime import date, datetime, timedelta
 import functools
@@ -857,6 +857,20 @@ def chunked(iterable: Iterable, chunked_num: int) -> Iterable[Any]:
     return iter(partial(take, chunked_num, iter(iterable)), [])
 
 
+def chunked_or_all(iterable: Collection[Any], chunked_num: int) -> Iterable[Any]:
+    """Break *collection* into iterables of length *n*.
+
+    Returns the collection if its length is less than *n*.
+
+    Unlike chunked, this function requires a collection so it can
+    determine the length of the collection and return the collection
+    if it is less than *n*.
+    """
+    if len(iterable) <= chunked_num:
+        return (iterable,)
+    return chunked(iterable, chunked_num)
+
+
 def get_index_by_name(session: Session, table_name: str, index_name: str) -> str | None:
     """Get an index by name."""
     connection = session.connection()

tests/components/recorder/test_util.py

@@ -25,6 +25,7 @@ from homeassistant.components.recorder.models import (
     process_timestamp,
 )
 from homeassistant.components.recorder.util import (
+    chunked_or_all,
     end_incomplete_runs,
     is_second_sunday,
     resolve_period,
@@ -1023,3 +1024,24 @@ async def test_resolve_period(hass: HomeAssistant) -> None:
             }
         }
     ) == (now - timedelta(hours=1, minutes=25), now - timedelta(minutes=25))
+
+
+def test_chunked_or_all():
+    """Test chunked_or_all can iterate chunk sizes larger than the passed in collection."""
+    all = []
+    incoming = (1, 2, 3, 4)
+    for chunk in chunked_or_all(incoming, 2):
+        assert len(chunk) == 2
+        all.extend(chunk)
+    assert all == [1, 2, 3, 4]
+
+    all = []
+    incoming = (1, 2, 3, 4)
+    for chunk in chunked_or_all(incoming, 5):
+        assert len(chunk) == 4
+        # Verify the chunk is the same object as the incoming
+        # collection since we want to avoid copying the collection
+        # if we don't need to
+        assert chunk is incoming
+        all.extend(chunk)
+    assert all == [1, 2, 3, 4]