Guard against selecting all invalid entity_ids in history (#89929)

If all the entity_ids that were provided do not exist we would
end up passing an empty list of ids to the SQL query which
would do an unbounded select

homeassistant/components/logbook/processor.py

@@ -14,6 +14,7 @@ from homeassistant.components.recorder import get_instance
 from homeassistant.components.recorder.filters import Filters
 from homeassistant.components.recorder.models import (
     bytes_to_uuid_hex_or_none,
+    extract_metadata_ids,
     process_datetime_to_timestamp,
     process_timestamp_to_utc_isoformat,
 )
@@ -154,14 +155,11 @@ class EventProcessor:
             metadata_ids: list[int] | None = None
             if self.entity_ids:
                 instance = get_instance(self.hass)
-                entity_id_to_metadata_id = instance.states_meta_manager.get_many(
+                metadata_ids = extract_metadata_ids(
-                    self.entity_ids, session, False
+                    instance.states_meta_manager.get_many(
+                        self.entity_ids, session, False
+                    )
                 )
-                metadata_ids = [
-                    metadata_id
-                    for metadata_id in entity_id_to_metadata_id.values()
-                    if metadata_id is not None
-                ]
             stmt = statement_for_request(
                 start_day,
                 end_day,

homeassistant/components/recorder/history/modern.py

@@ -23,7 +23,12 @@ import homeassistant.util.dt as dt_util
 from ... import recorder
 from ..db_schema import RecorderRuns, StateAttributes, States, StatesMeta
 from ..filters import Filters
-from ..models import LazyState, process_timestamp, row_to_compressed_state
+from ..models import (
+    LazyState,
+    extract_metadata_ids,
+    process_timestamp,
+    row_to_compressed_state,
+)
 from ..util import execute_stmt_lambda_element, session_scope
 from .const import (
     IGNORE_DOMAINS_ENTITY_ID_LIKE,
@@ -232,14 +237,12 @@ def get_significant_states_with_session(
     entity_id_to_metadata_id: dict[str, int | None] | None = None
     if entity_ids:
         instance = recorder.get_instance(hass)
-        entity_id_to_metadata_id = instance.states_meta_manager.get_many(
+        if not (
-            entity_ids, session, False
+            entity_id_to_metadata_id := instance.states_meta_manager.get_many(
-        )
+                entity_ids, session, False
-        metadata_ids = [
+            )
-            metadata_id
+        ) or not (metadata_ids := extract_metadata_ids(entity_id_to_metadata_id)):
-            for metadata_id in entity_id_to_metadata_id.values()
+            return {}
-            if metadata_id is not None
-        ]
     stmt = _significant_states_stmt(
         start_time,
         end_time,
@@ -569,14 +572,9 @@ def _get_rows_with_session(
     # We have more than one entity to look at so we need to do a query on states
     # since the last recorder run started.
     if entity_ids:
-        if not entity_id_to_metadata_id:
+        if not entity_id_to_metadata_id or not (
-            return []
+            metadata_ids := extract_metadata_ids(entity_id_to_metadata_id)
-        metadata_ids = [
+        ):
-            metadata_id
-            for metadata_id in entity_id_to_metadata_id.values()
-            if metadata_id is not None
-        ]
-        if not metadata_ids:
             return []
         stmt = _get_states_for_entities_stmt(
             run.start, utc_point_in_time, metadata_ids, no_attributes

homeassistant/components/recorder/models/__init__.py

@@ -8,7 +8,7 @@ from .context import (
     uuid_hex_to_bytes_or_none,
 )
 from .database import DatabaseEngine, DatabaseOptimizer, UnsupportedDialect
-from .state import LazyState, row_to_compressed_state
+from .state import LazyState, extract_metadata_ids, row_to_compressed_state
 from .statistics import (
     CalendarStatisticPeriod,
     FixedStatisticPeriod,
@@ -43,6 +43,7 @@ __all__ = [
     "bytes_to_ulid_or_none",
     "bytes_to_uuid_hex_or_none",
     "datetime_to_timestamp_or_none",
+    "extract_metadata_ids",
     "process_datetime_to_timestamp",
     "process_timestamp",
     "process_timestamp_to_utc_isoformat",

homeassistant/components/recorder/models/state.py

@@ -24,6 +24,17 @@ from .time import process_timestamp
 _LOGGER = logging.getLogger(__name__)
 
 
+def extract_metadata_ids(
+    entity_id_to_metadata_id: dict[str, int | None],
+) -> list[int]:
+    """Extract metadata ids from entity_id_to_metadata_id."""
+    return [
+        metadata_id
+        for metadata_id in entity_id_to_metadata_id.values()
+        if metadata_id is not None
+    ]
+
+
 class LazyState(State):
     """A lazy version of core State after schema 31."""