Fix is_missing state updates for asset cache states on startup

- Add bulk_update_is_missing() to efficiently update is_missing flag
- Update sync_cache_states_with_filesystem() to mark non-existent files as is_missing=True
- Call restore_cache_states_by_paths() in batch_insert_seed_assets() to restore
  previously-missing states when files reappear during scanning

Amp-Thread-ID: https://ampcode.com/threads/T-019c3177-e591-7666-ac6b-7e05c71c8ebf
Co-authored-by: Amp <amp@ampcode.com>

app/assets/database/queries/__init__.py

@@ -25,6 +25,7 @@ from app.assets.database.queries.asset_info import (
 from app.assets.database.queries.cache_state import (
     CacheStateRow,
     bulk_insert_cache_states_ignore_conflicts,
+    bulk_update_is_missing,
     bulk_update_needs_verify,
     delete_assets_by_ids,
     delete_cache_states_by_ids,
@@ -65,6 +66,7 @@ __all__ = [
     "bulk_insert_assets",
     "bulk_insert_cache_states_ignore_conflicts",
     "bulk_insert_tags_and_meta",
+    "bulk_update_is_missing",
     "bulk_update_needs_verify",
     "delete_asset_info_by_id",
     "delete_assets_by_ids",

app/assets/database/queries/cache_state.py

@@ -261,6 +261,21 @@ def bulk_update_needs_verify(session: Session, state_ids: list[int], value: bool
     return result.rowcount
 
 
+def bulk_update_is_missing(session: Session, state_ids: list[int], value: bool) -> int:
+    """Set is_missing flag for multiple cache states.
+
+    Returns: Number of rows updated
+    """
+    if not state_ids:
+        return 0
+    result = session.execute(
+        sa.update(AssetCacheState)
+        .where(AssetCacheState.id.in_(state_ids))
+        .values(is_missing=value)
+    )
+    return result.rowcount
+
+
 def delete_cache_states_by_ids(session: Session, state_ids: list[int]) -> int:
     """Delete cache states by their IDs.
 

app/assets/scanner.py

@@ -7,6 +7,7 @@ from typing import Literal, TypedDict
 import folder_paths
 from app.assets.database.queries import (
     add_missing_tag_for_asset_id,
+    bulk_update_is_missing,
     bulk_update_needs_verify,
     delete_cache_states_by_ids,
     delete_orphaned_seed_asset,
@@ -154,6 +155,7 @@ def sync_cache_states_with_filesystem(
     to_set_verify: list[int] = []
     to_clear_verify: list[int] = []
     stale_state_ids: list[int] = []
+    to_mark_missing: list[int] = []
     survivors: set[str] = set()
 
     for aid, acc in by_asset.items():
@@ -164,6 +166,7 @@ def sync_cache_states_with_filesystem(
 
         for s in states:
             if not s["exists"]:
+                to_mark_missing.append(s["sid"])
                 continue
             if s["fast_ok"] and s["needs_verify"]:
                 to_clear_verify.append(s["sid"])
@@ -195,6 +198,9 @@ def sync_cache_states_with_filesystem(
                 survivors.add(os.path.abspath(s["fp"]))
 
     delete_cache_states_by_ids(session, stale_state_ids)
+    stale_set = set(stale_state_ids)
+    to_mark_missing = [sid for sid in to_mark_missing if sid not in stale_set]
+    bulk_update_is_missing(session, to_mark_missing, value=True)
     bulk_update_needs_verify(session, to_set_verify, value=True)
     bulk_update_needs_verify(session, to_clear_verify, value=False)
 

app/assets/services/bulk_ingest.py

@@ -18,6 +18,7 @@ from app.assets.database.queries import (
     get_cache_states_by_paths_and_asset_ids,
     get_unreferenced_unhashed_asset_ids,
     mark_cache_states_missing_outside_prefixes,
+    restore_cache_states_by_paths,
 )
 from app.assets.helpers import get_utc_now
 
@@ -201,6 +202,7 @@ def batch_insert_seed_assets(
 
     bulk_insert_assets(session, asset_rows)
     bulk_insert_cache_states_ignore_conflicts(session, cache_state_rows)
+    restore_cache_states_by_paths(session, absolute_path_list)
     winning_paths = get_cache_states_by_paths_and_asset_ids(session, path_to_asset_id)
 
     all_paths_set = set(absolute_path_list)