Fix llm cache save problem in json_kv storage

lightrag/kg/json_doc_status_impl.py

@@ -96,12 +96,12 @@ class JsonDocStatusStorage(DocStatusStorage):
             write_json(data_dict, self._file_name)
 
     async def upsert(self, data: dict[str, dict[str, Any]]) -> None:
-        logger.info(f"Inserting {len(data)} to {self.namespace}")
         if not data:
             return
-
+        logger.info(f"Inserting {len(data)} to {self.namespace}")
         async with self._storage_lock:
             self._data.update(data)
+
         await self.index_done_callback()
 
     async def get_by_id(self, id: str) -> Union[dict[str, Any], None]:

lightrag/kg/json_kv_impl.py

@@ -98,11 +98,9 @@ class JsonKVStorage(BaseKVStorage):
     async def upsert(self, data: dict[str, dict[str, Any]]) -> None:
         if not data:
             return
+        logger.info(f"Inserting {len(data)} to {self.namespace}")
         async with self._storage_lock:
-            left_data = {k: v for k, v in data.items() if k not in self._data}
+            self._data.update(data)
-            if left_data:
-                logger.info(f"Process {os.getpid()} KV inserting {len(left_data)} to {self.namespace}")
-                self._data.update(left_data)
 
     async def delete(self, ids: list[str]) -> None:
         async with self._storage_lock:

lightrag/operate.py

@@ -403,6 +403,7 @@ async def extract_entities(
             else:
                 _prompt = input_text
 
+            # TODO： add cache_type="extract"
             arg_hash = compute_args_hash(_prompt)
             cached_return, _1, _2, _3 = await handle_cache(
                 llm_response_cache,
@@ -431,7 +432,6 @@ async def extract_entities(
                     cache_type="extract",
                 ),
             )
-            logger.info(f"Extract: saved cache for {arg_hash}")
             return res
 
         if history_messages: