政策法规脚本维护

comData/policylaw/gwysasac.py

@@ -168,10 +168,10 @@ def get_content3():
                     href = f'http://www.sasac.gov.cn{href_.replace("../../..", "")}'
                     # 判断是否已经爬取过
                     is_href = baseTool.db_storage.find_one({'网址': href})
-                    # if is_href:
-                    #     num += 1
-                    #     log.info('已采集----------跳过')
-                    #     continue
+                    if is_href:
+                        num += 1
+                        log.info('已采集----------跳过')
+                        continue
                     title = doc_item('a').attr('title')
                     pub_time = doc_item('span').text().replace('[', '').replace(']', '')
                 except: