冗余公司名字

app/services/job_clean_service.py

@@ -143,7 +143,7 @@ async def _do_clean(data: dict) -> None:
                 return
 
     # 第一次AI：结构化提取
-    user_message = _build_user_message(data)
+    user_message = await _build_user_message(data)
     result = await ai_chat_json(JobCleanModel.STRUCTURE, JOB_STRUCTURE_SYSTEM, user_message, scene="structure")
     if result is None or not result.get("valid", False):
         log.info("[id={}] 丢弃：AI判定无效", data_id)
@@ -373,14 +373,18 @@ async def _update_pg_status(data_id: int, status: str) -> None:
         await pg.commit()
 
 
-def _build_user_message(data: dict) -> str:
+async def _build_user_message(data: dict) -> str:
     """构建第一次AI的用户消息"""
+    # 从 app_url_list 取来源公司名，作为 companyShortName 的参考
+    input_company_name = await _get_input_company_name(data.get("urllistid"))
+
     parts = [
         "【原始数据】",
         f"岗位名称: {data.get('job_title') or ''}",
         f"薪资: {data.get('salary') or ''}",
         f"工作地点: {data.get('location') or ''}",
         f"公司: {data.get('company') or ''}",
+        f"来源公司名(参考): {input_company_name or ''}",
         f"经验要求: {data.get('experience') or ''}",
         f"学历要求: {data.get('education') or ''}",
         f"岗位详情: {data.get('description') or ''}",
@@ -390,3 +394,15 @@ def _build_user_message(data: dict) -> str:
         f"【行业列表】\n{dict_cache.industry_text}",
     ]
     return "\n".join(parts)
+
+
+async def _get_input_company_name(urllistid: int | None) -> str | None:
+    """从 PG app_url_list 按 urllistid 读取 input_company_name"""
+    if not urllistid:
+        return None
+    async with PgSession() as pg:
+        row = await pg.execute(
+            text("SELECT input_company_name FROM app_url_list WHERE id = :id"),
+            {"id": urllistid},
+        )
+        return row.scalar()