diff --git a/api/db/services/connector_service.py b/api/db/services/connector_service.py index 40f0b7b5c..9f7b0e6de 100644 --- a/api/db/services/connector_service.py +++ b/api/db/services/connector_service.py @@ -100,7 +100,7 @@ class ConnectorService(CommonService): return 0, [] source_type = f"{conn.source}/{conn.id}" - retain_doc_ids = {hash128(file.id) for file in file_list} + retain_doc_ids = {hash128(f"{connector_id}:{file.id}") for file in file_list} existing_docs = DocumentService.list_doc_headers_by_kb_and_source_type( kb_id, source_type, diff --git a/rag/svr/sync_data_source.py b/rag/svr/sync_data_source.py index b5801905d..9a60701e7 100644 --- a/rag/svr/sync_data_source.py +++ b/rag/svr/sync_data_source.py @@ -202,7 +202,7 @@ class SyncBase: docs = [] for doc in document_batch: d = { - "id": hash128(doc.id), + "id": hash128(f"{task['connector_id']}:{doc.id}"), "connector_id": task["connector_id"], "source": self.SOURCE_NAME, "semantic_identifier": doc.semantic_identifier,