|
|
@@ -79,7 +79,7 @@ def _resolve_dataset_id(request_id: str) -> int:
|
|
|
return 12
|
|
|
|
|
|
|
|
|
-def _upload_chunk(text: str, query: str, channel: str = "", dataset_id: int = 12, max_retries: int = 3, backoff_sec: float = 1.0) -> bool:
|
|
|
+def _upload_chunk(text: str, query: str, channel: str = "", dataset_id: int = 12, parsing_id: Optional[str] = None, max_retries: int = 3, backoff_sec: float = 1.0) -> bool:
|
|
|
# ext 需要是字符串 JSON
|
|
|
payload = {
|
|
|
"dataset_id": dataset_id,
|
|
|
@@ -99,6 +99,10 @@ def _upload_chunk(text: str, query: str, channel: str = "", dataset_id: int = 12
|
|
|
except Exception:
|
|
|
logger.info(f"上传chunk返回非JSON: text={resp.text[:500]}")
|
|
|
if resp.json().get("doc_id"):
|
|
|
+ # 取出doc_id，存储到knowledge_extraction_content表的doc_id字段
|
|
|
+ sql = "UPDATE knowledge_extraction_content SET doc_id = %s WHERE parsing_id = %s"
|
|
|
+ MysqlHelper.update_values(sql, (resp.json().get("doc_id"), parsing_id))
|
|
|
+ logger.info(f"更新doc_id成功: parsing_id={parsing_id}, doc_id={resp.json().get('doc_id')}")
|
|
|
return True
|
|
|
logger.warning(f"上传失败，状态码: {resp.status_code}, 第{attempt+1}次重试")
|
|
|
except Exception as e:
|
|
|
@@ -147,7 +151,7 @@ def execute_store_agent(request_id: str) -> Tuple[int, int]:
|
|
|
except Exception as e:
|
|
|
logger.warning(f"获取channel失败: parsing_id={parsing_id}, error={e}")
|
|
|
|
|
|
- ok = _upload_chunk(text, query, channel, dataset_id)
|
|
|
+ ok = _upload_chunk(text, query, channel, dataset_id, parsing_id)
|
|
|
success += 1 if ok else 0
|
|
|
|
|
|
if success == total:
|
