Deploy RAG Tourism API

.idea/workspace.xml

@@ -157,7 +157,7 @@
       <workItem from="1780806816923" duration="5651000" />
       <workItem from="1781019152997" duration="1850000" />
       <workItem from="1784951212519" duration="2143000" />
-      <workItem from="1785045303244" duration="7532000" />
+      <workItem from="1785045303244" duration="8380000" />
     </task>
     <servers />
   </component>

app/main.py

@@ -141,6 +141,15 @@ def _same_category(value: str, target: str) -> bool:
     return _strip_diacritics(value) == _strip_diacritics(target)
 
 
+def _kg_location_count(kg) -> Optional[int]:
+    """Số Location trong KG — để so với số trong vector index, tự rebuild khi lệch."""
+    try:
+        with kg.driver.session() as s:
+            return s.run("MATCH (l:Location) RETURN count(l) AS c").single()["c"]
+    except Exception:
+        return None
+
+
 # Cache of known old district/city names for fuzzy matching (filled lazily)
 _district_names_cache = None
 _province_names_cache = None
@@ -232,8 +241,13 @@ async def startup_event():
                     for m in metadatas
                     if isinstance(m, dict)
                 )
-                if not has_location_fields and kg_service:
-                    logger.warning("ChromaDB index loaded but appears to lack location metadata; rebuilding from Neo4j KG...")
+                # Rebuild khi thiếu metadata HOẶC số lượng trong index lệch với KG
+                # (vd vừa thêm khách sạn/nhà hàng vào KG mà index chưa cập nhật).
+                _idx_n = vector_service.collection.count() if vector_service.collection else 0
+                _kg_n = _kg_location_count(kg_service) if kg_service else None
+                _stale = (_kg_n is not None and _idx_n < _kg_n)
+                if (not has_location_fields or _stale) and kg_service:
+                    logger.warning(f"ChromaDB index cần rebuild (index={_idx_n}, KG={_kg_n}); rebuilding từ Neo4j KG...")
                     kg_locations = kg_service.search_locations_by_name('', limit=5000)
                     if kg_locations and vector_service.build_index(kg_locations):
                         logger.info("ChromaDB index rebuilt successfully from KG data")
@@ -268,8 +282,11 @@ async def startup_event():
                 sample = phobert_service.collection.get(limit=5, include=["metadatas"]) if phobert_service.collection else None
                 metas = (sample or {}).get("metadatas") or []
                 ok = any(isinstance(m, dict) and (m.get("ward") or "").strip() for m in metas)
-                if not ok and kg_service:
-                    logger.warning("PhoBERT index thiếu metadata ward/province; rebuilding từ KG...")
+                _idx_n = phobert_service.collection.count() if phobert_service.collection else 0
+                _kg_n = _kg_location_count(kg_service) if kg_service else None
+                _stale = (_kg_n is not None and _idx_n < _kg_n)
+                if (not ok or _stale) and kg_service:
+                    logger.warning(f"PhoBERT index cần rebuild (index={_idx_n}, KG={_kg_n}); rebuilding từ KG...")
                     kg_locations = kg_service.search_locations_by_name('', limit=5000)
                     if kg_locations and phobert_service.build_index(kg_locations):
                         logger.info("PhoBERT index rebuilt successfully")