使用cookie登陆人人网

from urllib import request, parse
from http import cookiejar

filename = 'cookie.txt'


class Spider(object):
    headers = {
        'User-Agent': 'Mozilla/4.0 (compatible; MSIE 5.5; Windows NT)'
    }

    def __init__(self, login_url=None, url=None, filename=None, login_page_url=None):
        """

        :param login_url:   登陆页面提交表单url'http://www.renren.com/PLogin.do'
        :param url:         目标个人中心的url 'http://www.renren.com/966969502/profile'
        :param filename:  cookie文件名
        :param login_page_url:  登陆页面url 'http://www.renren.com/PLogin.do'
        """
        self.login_url = login_url
        self.url = url
        self.filename = filename
        self.login_page_url = login_page_url

    def create_cookiejar(self):
        """
        实例化MozillaCookieJar(),存储cookiejar对象到文件
        :param
        :return:
        """
        cookie_jar = cookiejar.MozillaCookieJar(filename=self.filename)
        return cookie_jar

    def create_openor(self, cookie_jar):
        handler = request.HTTPCookieProcessor(cookie_jar)
        opener = request.build_opener(handler)
        return opener

    # 先访问登陆页面获取部分cookie
    def get_login_page(self, opener):
        """
        :param opener: create_opner 的返回值
        :return:
        """
        login_get_request = request.Request(url=self.login_page_url, headers=self.headers)
        login_get_response = opener.open(login_get_request)

    # 第一次登陆获取全部cookie
    def first_login(self, opener, cookie_jar):
        data = {
            'email': '你的email',
            'password': '你的密码'
        }
        data = parse.urlencode(data).encode('utf-8')
        first_login_request = request.Request(url=self.login_url, headers=self.headers, data=data)
        first_login_response = opener.open(first_login_request)
        # 保存获得所有cookie
        cookie_jar.save(ignore_discard=True, ignore_expires=True)
        self.status = first_login_response.getcode()
        return self.status

    # 从文件中获取cookie
    def get_cookie_from_file(self):
        full_cookie = cookiejar.MozillaCookieJar()
        full_cookie.load(self.filename, ignore_discard=True, ignore_expires=True)
        print(full_cookie)
        return full_cookie

    # 携带cookie访问
    def get_data_with_cookie(self, opener):
        personcenter_request = request.Request(url=self.url, headers=self.headers)  # url 访问
        personcenter_response = opener.open(personcenter_request)
        content = personcenter_response.read().decode('utf-8')
        with open('./renren_personal_center.html', 'w', encoding='utf-8') as fp:
            fp.write(content)


def main():
    filename = 'cookie.txt'
    login_url = 'http://www.renren.com/PLogin.do'
    url = 'http://www.renren.com/966969502/profile'
    login_page_url = 'http://www.renren.com/PLogin.do'
    renren_spider = Spider(login_page_url=login_page_url, url=url, login_url=login_url, filename=filename)
    # cookie_jar = renren_spider.create_cookiejar()
    # opener = renren_spider.create_openor(cookie_jar)
    # # 第一次登陆
    # renren_spider.get_login_page(opener)
    # renren_spider.first_login(opener=opener, cookie_jar=cookie_jar)
    # # 携带cookie访问
    full_cookie = renren_spider.get_cookie_from_file()
    newopener = renren_spider.create_openor(full_cookie)
    renren_spider.get_data_with_cookie(newopener)


if __name__ == '__main__':
    main()

# print(cookie)


# print(personcenter_response)
# loginurl = 'http://renren.com'
# s = requests.Session()
# r = s.get(url=loginurl, allow_redirects=True)
#
# datas = {'email': '18813388480', 'password': 'zm636057893'}
# headers = {
#     'User-Agent': 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10.12; rv:61.0) Gecko/20100101 Firefox/61.0',
#
# }
# url = 'http://www.renren.com/PLogin.do'
# response = s.post(url=url, data=datas, allow_redirects=True)
# cookie_dict = requests.utils.dict_from_cookiejar(s.cookies)
# print(cookie_dict)
# print(response.text)