补充文档，修改清洗方案

app/services/job_clean_service.py

@@ -86,6 +86,26 @@ async def _do_clean(data: dict) -> None:
         await _update_pg_status(data_id, "discarded")
         return
 
+    # 哈希去重续命：content_hash 已存在于 bg_job → 说明是重复刷新，直接续命
+    content_hash = data.get("content_hash")
+    if content_hash:
+        async with MysqlSession() as mysql:
+            row = await mysql.execute(
+                text("SELECT id FROM bg_job WHERE content_hash = :hash LIMIT 1"),
+                {"hash": content_hash},
+            )
+            existing_id = row.scalar()
+            if existing_id:
+                now = datetime.now()
+                await mysql.execute(
+                    text("UPDATE bg_job SET status = 0, create_time = :now, update_time = :now WHERE id = :id"),
+                    {"now": now, "id": existing_id},
+                )
+                await mysql.commit()
+                await _update_pg_status(data_id, "cleaned")
+                log.info("[id={}] 续命：hash命中，岗位ID={}", data_id, existing_id)
+                return
+
     # 第一次AI：结构化提取
     user_message = _build_user_message(data)
     result = await ai_chat_json(JobCleanModel.STRUCTURE, JOB_STRUCTURE_SYSTEM, user_message)
@@ -94,18 +114,6 @@ async def _do_clean(data: dict) -> None:
         await _update_pg_status(data_id, "discarded")
         return
 
-    # 去重检查
-    source_id = str(data_id)
-    async with MysqlSession() as mysql:
-        existing = await mysql.execute(
-            text("SELECT COUNT(*) AS cnt FROM bg_job WHERE source_id = :sid"),
-            {"sid": source_id},
-        )
-        if existing.scalar() > 0:
-            log.info("[id={}] 跳过：已入库(去重)", data_id)
-            await _update_pg_status(data_id, "cleaned")
-            return
-
     # 公司处理
     company_short_name = result.get("companyShortName") or data.get("company") or ""
     company_id = await _find_or_create_company(company_short_name)
@@ -139,6 +147,7 @@ async def _do_clean(data: dict) -> None:
                 required_industry_id=result.get("requiredIndustryId"),
                 recruit_category=data.get("recruit_category", 3),
                 expire_at=data.get("expire_at"),
+                content_hash=data.get("content_hash"),
                 source_url=data.get("detail_url"),
                 source_id=source_id,
                 status=0,