通过session爬取我要自学网会员中心的账号信息

import requests
from fake_useragent import UserAgent
from lxml import etree


class Spider(object):
    """爬取我要自学网会员中心的账号信息
       第一步：从登录界面的源码中提取token的值，用于后面发送post请求；
       第二步：输入账号和密码，完善data表单数据，发送post请求，使session对象保存登录的cookie信息
       第三步：发送get请求，解析会员中心数据
    """

    def __init__(self):
        self.login_url = 'https://www.51zxw/login'  # 登录界面url
        self.post_url = 'https://www.51zxw/login/NewLogin/AjaxForlogin'  # 发送post请求url
        self.url = 'https://www.51zxw/Login/UserCenter'  # 会员中心url
        self.headers = {'User-Agent': UserAgent().random}
        self.s = requests.session()  # 创建session对象

    def get_token(self):
        """获取登录界面中的token的值"""
        r = self.s.get(self.login_url, headers=self.headers).content
        html = etree.HTML(r)
        return html.xpath(r'//input[@name="__RequestVerificationToken"]/@value')[0]  # 得到token的值

    def get_cookie(self):
        """发送post请求，使session对象保存cookie信息"""
        user = input('输入账号：')
        password = input('输入密码：')
        data = {'loginStr': user,
                'pwd': password,
                '__RequestVerificationToken': self.get_token(),
                'isRememberlogin': 'false'
                }
        self.s.post(self.post_url, data=data, headers=self.headers)

    def get_data(self):
        """根据cookie获取会员中心的账号数据"""
        r = self.s.get(self.url, headers=self.headers).content
        html = etree.HTML(r)
        item = {}
        item['member_name'] = html.xpath(r'//div[@class="Others"]/div[1]/b/text()')[0]
        item['member_type'] = html.xpath(r'//div[@class="Others"]/div[2]/text()')[0][5:]
        item['register_date'] = html.xpath(r'//div[@class="Others"]/div[4]/text()')[0][4:]
        item['consume_V'] = html.xpath(r'//div[@class="webCount"]//li[3]//div[2]/text()')[0]
        item['remain_V'] = html.xpath(r'//div[@class="webCount"]//li[3]//div[3]/text()')[0]
        print(item)


def main():
    spider = Spider()
    spider.get_token()
    spider.get_cookie()
    spider.get_data()


if __name__ == '__main__':
    main()