diff --git a/main.py b/main.py
index 61fdd33..e886ddf 100644
--- a/main.py
+++ b/main.py
@@ -1,3 +1,4 @@
+import json
 import random
 import traceback
 from urllib.parse import quote
@@ -227,7 +228,7 @@ def gettoken(proxy, r=2):
             return copy.deepcopy(headers1)
 
 
-def get_searchInfo(keyword, level, headers, proxy_name):
+def get_searchInfo(keyword, level, headers, proxy_name, r=2):
     video_list = []
     max_page = 2
     limit = 30
@@ -589,12 +590,25 @@ def get_searchInfo(keyword, level, headers, proxy_name):
 
         jsondata = response.json()
         try:
-            resinfo = jsondata['data']['search']['stories']['edges']
-            print('resinfo :', len(resinfo))
+            errors = jsondata.get("errors")  # GraphQL errors 数组
+            stories = jsondata.get("data", {}).get("search", {}).get("stories")
+
+            if errors or stories is None:  # 有错误 或 stories 为 null
+                if r == 0:
+                    print("连续 3 次错误或空结果：", json.dumps(jsondata, ensure_ascii=False))
+                    return None
+                time.sleep((3 - r) * 5)
+                return get_searchInfo(keyword, level, headers, proxy_name, r - 1)
+            resinfo = stories["edges"]
+            print("resinfo :", len(resinfo))
         except Exception:
-            resinfo = []
-            print("[搜索接口]", response.text)
-            print("返回字段解析错误!")
+            if r < 0:
+                print("[搜索接口] 未知:未处理", response.text)
+                print("返回字段解析错误!")
+                return None
+            else:
+                time.sleep((3 - r) * 5)
+                return get_searchInfo(keyword, level, headers, proxy_name, r - 1)
         for index, iteminfo in enumerate(resinfo):
             calculated_index = index + 1 + (j - 1) * 100
             node = iteminfo['node']
@@ -631,24 +645,26 @@ proxiesdict = db.get_proxy_agent_dict()
 def search_worker(payload, kitem, flag):
     try:
         gproxies = proxiesdict[kitem['rn']]
-        header   = gettoken(gproxies)
+        header = gettoken(gproxies)
         v_list = get_searchInfo(kitem['keyword'], kitem['level'], header, gproxies)
         if not v_list:
             for i in range(2):
                 time.sleep(i * 5)
-                v_list = get_searchInfo(kitem['keyword'], kitem['level'], header)
+                v_list = get_searchInfo(kitem['keyword'], kitem['level'], header, gproxies)
                 if v_list:
                     break
                 time.sleep(2)
 
-        return True, flag, payload, kitem, v_list      # 成功
+        return True, flag, payload, kitem, v_list  # 成功
     except Exception as e:
         print(f"[线程异常] {kitem['keyword']} → {e}")
         traceback.print_exc()
-        return False, flag, payload, kitem, []         # 失败
+        return False, flag, payload, kitem, []  # 失败
 
 
 executor = concurrent.futures.ThreadPoolExecutor(MAX_WORKERS)
+
+
 def integrate_data_parallel():
     while True:
         tasks, flag = db.item_keyword()
@@ -675,26 +691,26 @@ def integrate_data_parallel():
             for item in v_list:
                 record = {
                     "keyword": kitem["keyword"],
-                    "v_name":  kitem["v_name"],
-                    "v_id":    item["v_id"],
-                    "v_xid":   item["v_xid"],
-                    "link":    item["link"],
-                    "title":   item["title"],
-                    "duration":        format_duration(item["duration"]),
-                    "fans":            clean_dash_to_zero(item["fans"]),
-                    "videos":          clean_dash_to_zero(item["videos"]),
-                    "watch_number":    clean_dash_to_zero(item["view"]),
-                    "create_time":     format_create_time(item["createtime"]),
-                    "cover_pic":       item["pic"],
-                    "index":           item["index"],
-                    "u_id":   item["u_id"],
-                    "u_xid":  item["u_xid"],
+                    "v_name": kitem["v_name"],
+                    "v_id": item["v_id"],
+                    "v_xid": item["v_xid"],
+                    "link": item["link"],
+                    "title": item["title"],
+                    "duration": format_duration(item["duration"]),
+                    "fans": clean_dash_to_zero(item["fans"]),
+                    "videos": clean_dash_to_zero(item["videos"]),
+                    "watch_number": clean_dash_to_zero(item["view"]),
+                    "create_time": format_create_time(item["createtime"]),
+                    "cover_pic": item["pic"],
+                    "index": item["index"],
+                    "u_id": item["u_id"],
+                    "u_xid": item["u_xid"],
                     "u_name": item["u_name"],
-                    "u_pic":  item["u_pic"],
-                    "rn":     kitem["rn"],
-                    "batch":  kitem["batch"],
+                    "u_pic": item["u_pic"],
+                    "rn": kitem["rn"],
+                    "batch": kitem["batch"],
                     "machine_id": MACHINE_ID,
-                    "level":  kitem["level"],
+                    "level": kitem["level"],
                 }
                 db.upsert_video(record)
                 db.flush()