'commit'

dsLightRag/WxGzh/T2_CollectArticle.py

@@ -161,6 +161,7 @@ if __name__ == '__main__':
 
     # 爬取文章
     for item in gzlist:
+        cnt = 0
         account_name = item["account_name"]
         account_id = item["account_id"]
         id = item["id"]
@@ -205,6 +206,7 @@ if __name__ == '__main__':
         query_fakeid_response = requests.get(appmsg_url, cookies=cookies, headers=header, params=query_id_data)
         fakeid_list = query_fakeid_response.json().get('app_msg_list')
 
+
         for item in fakeid_list:
             article_url = item.get('link')
             article_title = item.get('title')
@@ -225,10 +227,13 @@ if __name__ == '__main__':
                 pool = loop.run_until_complete(init_postgres_pool())
                 loop.run_until_complete(
                     save_article_to_db(pool, article_title, account_name, article_url, publish_time, content, id))
+                cnt = cnt + 1
             finally:
                 loop.run_until_complete(pool.close())
                 loop.close()
-
+            # 休息1秒，防止频繁访问被封
             time.sleep(1)
+        logger.info(f"成功获取公众号: {account_name} {cnt}篇文章。")
         # 关闭浏览器
     driver.quit()
+