Sfoglia il codice sorgente

Merge remote-tracking branch 'origin/master' into feature_20250402_supeng_add_subCrawlerSrcCode_param

supeng 3 settimane fa
parent
commit
0ee2920fcd
2 ha cambiato i file con 7 aggiunte e 0 eliminazioni
  1. 4 0
      utils/dy_ks_get_url.py
  2. 3 0
      workers/consumption_work.py

+ 4 - 0
utils/dy_ks_get_url.py

@@ -119,15 +119,19 @@ class Dy_KS:
     def get_video_url(cls, data, principal):
         try:
             url = data['video_url']
+            logger.info(f"[+] url=={url}")
             if "&vid=" in url or "?modal_id=" in url:
                 host = urlparse(url).netloc
+                logger.info(f"[+] host=={host}")
             else:
                 msg = html.unescape(url).split('?')[0]
                 pattern = re.search(r'https?://[^\s<>"\'\u4e00-\u9fff]+', msg)
+                logger.info(f"[+] pattern == {pattern}")
                 if not pattern:
                     return "重新处理",None,None,None
                 url = pattern.group()
                 host = urlparse(url).netloc
+                logger.info(f"[+] url == {url}  host=={host}")
             if host in ['v.douyin.com', 'www.douyin.com', 'www.iesdouyin.com']:
                 tag_transport_channel = "抖音"
                 logger.info(f"[+] {url}开始获取抖音视频链接")

+ 3 - 0
workers/consumption_work.py

@@ -479,9 +479,12 @@ class ConsumptionRecommend(object):
         fs_data = os.getenv("FS_DATA")
         # fs_data = '周仙琴,2WIcBU,task:carry_data_redis_zxq'
         fs_data_list = fs_data.split(',')
+        logger.info(f"[+] fs_data_list=={fs_data_list}")
         redis_name = fs_data_list[2]
         studio_key = fs_data_list[3]
+        logger.info(f"[+] redis_name == {redis_name} studio_key == {studio_key}")
         data = RedisHelper().get_client().rpop(name=redis_name)
+        logger.info(f"[+] 本次处理的数据=={data}")
         if not data:
             logger.info('[处理] 无待执行的扫描任务')
             return