增加rufts

app/amazon/detail_spider.py

@@ -390,19 +390,20 @@ class SpiderTask:
                     runing_task[task_id]["status"] = "failed"
                     runing_task[task_id]["error"] = str(e)
 
-    def post_result(self, task_id: int, chunkIndex:int,chunkTotal: int, asin: str, error:str="",
-                    item_data:dict={},
-                    is_done: bool = False):
+    def post_result(self, task_id: int, chunkIndex:int,chunkTotal: int, asin: str, error:str="",
+                    item_data:dict={},
+                    is_done: bool = False):
         """回传处理结果到API
         """
 
         url = f"{DELETE_BRAND_API_BASE}/api/appearance-patent/tasks/{task_id}/result"
 
-        payload ={
-          "submissionId": f"{int(time.time())}",
-          "chunkIndex": chunkIndex,
-          "chunkTotal": chunkTotal,
-          "error": error,
+        submission_id = f"appearance-patent:{task_id}"
+        payload ={
+          "submissionId": submission_id,
+          "chunkIndex": chunkIndex,
+          "chunkTotal": chunkTotal,
+          "error": error,
           "items": [
             item_data
           ],
@@ -410,8 +411,9 @@ class SpiderTask:
         }
 
         max_retries = 3
-        for retry in range(max_retries):
-            try:
+        for retry in range(max_retries):
+            try:
+                request_timeout = 300 if is_done else 30
                 print("================【详情采集】=====================")
                 self.log(f"尝试回传结果 (第 {retry + 1}/{max_retries} 次)")
                 self.log(f"回传URL: {url}")
@@ -420,7 +422,7 @@ class SpiderTask:
                     url,
                     json=payload,
                     headers={"Content-Type": "application/json"},
-                    timeout=30,
+                    timeout=request_timeout,
                     verify=False
                 )
                 self.log(f"回传结果: {response.text}")
@@ -6904,4 +6906,4 @@ if __name__ == '__main__':
     },
     "code": None
 }
-    spide.process_task(task_data)
+    spide.process_task(task_data)