免费个人网站模板_常德软件开发_网站建设心得_宝安做网站的

cookies = {'TYCID': 'c5ed9bd0960011e891962b12a1f9a1f1', 'undefined': 'c5ed9bd0960011e891962b12a1f9a1f1', 'ssuid': '7456483413', '_ga': 'GA1.2.64663727.1537443593', '_gid': 'GA1.2.1799316349.1537443593', 'RTYCID': '0c517c4c030d4eb482a941a5383b8107', 'CT_TYCID': 'c716946a3f384978a185529cecd59277', 'tyc-user-info': '%257B%2522token%2522%253A%2522eyJhbGciOiJIUzUxMiJ9.eyJzdWIiOiIxNTcxMzcxMDk0NCIsImlhdCI6MTUzNzQ5Mjg1NiwiZXhwIjoxNTUzMDQ0ODU2fQ.xO88A8fKq9ztOGxO0F-to_jdZ6cbBKg3DMsUE9kJvxWXm5qaQaE9JaGKrr_5vmLuWJGdXDImK1hbZmkLU3aLcQ%2522%252C%2522integrity%2522%253A%25220%2525%2522%252C%2522state%2522%253A%25220%2522%252C%2522redPoint%2522%253A%25220%2522%252C%2522vipManager%2522%253A%25220%2522%252C%2522vnum%2522%253A%25220%2522%252C%2522monitorUnreadCount%2522%253A%25221%2522%252C%2522onum%2522%253A%25220%2522%252C%2522mobile%2522%253A%252215713710944%2522%257D', 'auth_token': 'eyJhbGciOiJIUzUxMiJ9.eyJzdWIiOiIxNTcxMzcxMDk0NCIsImlhdCI6MTUzNzQ5Mjg1NiwiZXhwIjoxNTUzMDQ0ODU2fQ.xO88A8fKq9ztOGxO0F-to_jdZ6cbBKg3DMsUE9kJvxWXm5qaQaE9JaGKrr_5vmLuWJGdXDImK1hbZmkLU3aLcQ', 'Hm_lvt_e92c8d65d92d534b0fc290df538b4758': '1537443593,1537493356','Hm_lpvt_e92c8d65d92d534b0fc290df538b4758': '1537510340', 'aliyungf_tc': 'AQAAAOh4sTFdjxwUAtymFt9gzyUSvNWrC', 'csrfToken': 'iXGG1ldNy8JcVC4KCJXn0z9D', 'Hm_lvt_d5ceb643638c8ee5fbf79d207b00f07e': '1537513104','_gat_gtag_UA_123487620_1': '1','Hm_lpvt_d5ceb643638c8ee5fbf79d207b00f07e':str(timestrap)}for conli in company_list:url = "https://m.tianyancha.com/search?key={}&checkFrom=searchBox".format(str(conli))print(url)# /html/body/div[3]/div[3]/div[1]/div[1]/div[1]/a# time_cookie = {'Hm_lpvt_d5ceb643638c8ee5fbf79d207b00f07e':timestrap}# cookies = dict(cookies,**time_cookie)# print(cookies)# 第二步:根据获得的信息来得到其response# 通过response.content来获取其中的信息,再将content的内容进行decode解码来观察其中的内容response = requests.get(url=url,headers=headers,cookies=cookies)con = response.contentcon = con.decode("utf-8")con_html = etree.HTML(con)new_url = con_html.xpath("/html/body/div[3]/div[3]/div[1]/div[1]/div[1]/a/@href")new_url =new_url[0]response_con = requests.get(url=new_url,headers=headers,cookies=cookies)content = response_con.contentcontent = content.decode("utf-8")html = etree.HTML(content)# 公司名称company = html.xpath('//*[@id="wap_header_top"]/div[1]/div[1]/div[1]/text()')# 法定代表人people = html.xpath('/html/body/div[3]/div[1]/div[6]/div/div[1]/span[2]/a/text()')# 经营状态status = html.xpath('/html/body/div[3]/div[1]/div[6]/div/div[2]/span[2]/text()')# 工商号gs_num = html.xpath('/html/body/div[3]/div[1]/div[6]/div/div[7]/span[2]/text()')# 注册资本money = html.xpath('/html/body/div[3]/div[1]/div[6]/div/div[4]/span[2]/text/text()')# 纳税人识别号people_num = html.xpath('/html/body/div[3]/div[1]/div[6]/div/div[10]/span[2]/text()')# print(content)# print("公司名称:"+company[0]+'\n'+"纳税人识别号:"+people_num[0])print(company[0]+" 法定代表人:"+people[0]+" 工商号:"+gs_num[0]+" 注册资本:"+money[0]+" 经营状态:"+status[0]+" 纳税人识别号:"+people_num[0])