def qyjcxx(self): try: a = Mysql.qiyexx_url(bh='1')[0] # 从复爬表 if a == None: print('没有数据可以爬取') time.sleep(10) else: self.qyid = a[0] #eid self.z = a[2] #公司名字 qw = self.gx_qyid() #这个东西可以优化,在失败或者加载不出东西可以尝试更新,不用每次加载 self.qyid1 = qw #qyid if a[7] == '1': self.jichu12() # 基础信息的爬取 else: print('基础信息爬取完毕') if a[8] == '1': self.qyzz() # 资质信息的爬取 else: print('资质信息爬取完毕') if a[9] == '1': self.qy_user() # 人员信息的爬取 else: print('人员信息爬取完毕') # self.gcxmxx() a = Mysql.qiyexx_url(bh='1')[0] if a[7] == '0' and (a[8] == '0' or a[8] == '404') and (a[9] == '0' or a[9] == '404'): Mysql.gxqy_fupa(cx_state='0', eid=self.qyid) print('状态更新完毕') except Exception as e: print(e, 'jgfufh')
driver.refresh() break except Exception as e: print(e) driver.quit() break def ipz(): # 设置代理连接 while True: resp = requests.get('代理连接').text if resp.find('data') != -1: resp1 = json.loads(resp)['data'] http = str(resp1[0]["ip"]) + ":" + str(resp1[0]["port"]) return http else: time.sleep(5) while True: a = Mysql.qiyexx_url(bh='1') for x in a: try: qyid = x[0] # 公司eid z = x[2] # 公司名字 qyid1 = x[3] #qyid qw = gx_qyid(z=z, eid=qyid) #这个东西可以优化,在失败或者加载不出东西可以尝试更新,不用每次加载 selenu(qw, z, ipz()) except Exception as E: print(E)