Refactor processing logic and enhance error handling

- Cleaned up code in processing.py by removing inline semicolons and improving readability.
- Updated upsert_file_relation calls to ensure consistent handling of file relations.
- Enhanced query_file_relations in db_schema.py to support filtering by file existence.
- Improved API error handling in index.ts with user-friendly messages for 401 and 403 errors.
- Added online/offline status tracking in Layout.vue.
- Implemented debounced search functionality across multiple views to optimize performance.
- Introduced loading skeletons in Dashboard.vue for better user experience during data fetching.
- Enhanced file preview cleanup logic in Images.vue, Orders.vue, and Tables.vue to prevent memory leaks.
- Updated global styles to include new loading and notification animations.

web/backend/routers/memory.py

@@ -53,6 +53,7 @@ class MemoryListResponse(BaseModel):
     total: int
     page: int
     page_size: int
+    stats: Optional[Dict] = None
 
 
 def _get_db():
@@ -96,11 +97,17 @@ async def list_memory(
     start = (page - 1) * page_size
     page_items = results[start:start + page_size]
 
+    # Compute confidence stats from all results (not just current page)
+    high = sum(1 for r in results if r.get("confidence", 0) > 50)
+    medium = sum(1 for r in results if 10 <= r.get("confidence", 0) <= 50)
+    low = sum(1 for r in results if r.get("confidence", 0) < 10)
+
     return MemoryListResponse(
         items=[_row_to_item(r) for r in page_items],
         total=total,
         page=page,
         page_size=page_size,
+        stats={"high": high, "medium": medium, "low": low, "total": total},
     )