爬虫---知乎登录

# 英文验证码的登录方式

# 中文登录（点击倒立文字）

import requests,time,json

from requests.packages.urllib3.exceptions import InsecureRequestWarning
requests.packages.urllib3.disable_warnings(InsecureRequestWarning)

headers = {
    "User-Agent":"Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:60.0) Gecko/20100101 Firefox/60.0",
    "Referer": "https://www.zhihu.com/signup?next=%2F",
    "origin": "https://www.zhihu.com",
    "Authorization": "oauth c3cef7c66a1843f8b3a9e6a1e3160e20",
}

# cookies的自动化管理。
# 获取的服务器的Set-Cookie用session直接自动解析并保存，在后续的请求中，会在请求头中自动携带这些cookie
# LWPCookieJar:对cookie进行自动操作，load() save()
from http.cookiejar import LWPCookieJar

session = requests.Session()
session.cookies = LWPCookieJar(filename='zhihucookie.txt')

try:
    session.cookies.load(filename='zhihucookie.txt', ignore_expires=True, ignore_discard=True)
except Exception as e:
    print('加载失败')

res = session.get('https://www.zhihu.com/', headers=headers, verify=False)
print(res)


def zhihu_login():
    global session

    cap = ''

    has_captcha = is_captcha()
    if has_captcha:
        # 获取验证码
        cap = get_captcha()
        # 在提交登陆之前，还需要对输入的验证码的正确性进行单独验证
        is_true = check_captcha(cap)
        if is_true == False:
            return

    # 1528450244046.0112
    print(time.time())
    login_url = "https://www.zhihu.com/api/v3/oauth/sign_in"
    post_params = {
        "client_id":"c3cef7c66a1843f8b3a9e6a1e3160e20",
        "grant_type": "password",
        "timestamp": "1528450006722",
        "source": "com.zhihu.web",
        "signature": "dbaccdbeeade95fa9ec65ad72f3cdc1473a5a245",
        "username": "", # 您的账户
        "password": "", # 您的密码
        "captcha": cap,
        "lang": "cn",
        "ref_source": "homepage",
        "utm_source": "",
    }

    try:
        response = session.post(login_url, data=post_params, headers=headers, verify=False)
        if response.status_code == 201:
            session.cookies.save(ignore_discard=True, ignore_expires=True)
            print(response.text)
        else:
            print('登陆失败')
    except Exception as e:
        print('请求失败',e)


def is_captcha():
    global COOKIE
    captcha_url = 'https://www.zhihu.com/api/v3/oauth/captcha?lang=cn'
    try:
        response = session.get(url=captcha_url, headers=headers,verify=False)
        if response.status_code == 200:
            show_captcha = json.loads(response.text)['show_captcha']
            if show_captcha:
                print('有验证码')
                return True
            else:
                print('没有验证码')
                return False
    except Exception as e:
        print('')

import base64
from PIL import Image
from io import BytesIO


all_points = [[16.5,25.933349609375],[37.5,16.933349609375],[62.5,21.933349609375],[85.5,21.933349609375],[113.5,25.933349609375],[139.5,19.933349609375],[165.5,20.933349609375]]


def get_captcha():
    captcha_url = 'https://www.zhihu.com/api/v3/oauth/captcha?lang=cn'
    try:
        # 索取验证码图片，在保证有验证码的前提下才会发送PUT
        response = session.put(url=captcha_url, headers=headers,verify=False)
        if response.status_code == 202:
            captcha_url = json.loads(response.text)['img_base64']
            # 解码captcha_url

            url = base64.b64decode(captcha_url)
            url = BytesIO(url)
            image = Image.open(url)
            image.show()

            captcha = input('请输入倒立汉字所在的位置：13')
            result_captcha = ''
            # 倒立汉字可能是1个或者2个
            if len(captcha) == 1:
                # 一个倒立汉字
                point = all_points[int(captcha)-1]
                result_captcha = '{"img_size":[200,44], "input_points":[%s]}' % point
            else:
                # 两个倒立汉字
                first_char = captcha[0]
                second_char = captcha[1]
                first_point = all_points[int(first_char) - 1]
                second_point = all_points[int(second_char) - 1]
                result_captcha = '{"img_size":[200,44], "input_points":[%s, %s]}' % (first_point, second_point)

            return result_captcha
    except Exception as e:
        print('')


def check_captcha(captcha):
    captcha_url = 'https://www.zhihu.com/api/v3/oauth/captcha?lang=cn'
    post_params = {
        'input_text': captcha
    }

    # verify=False: 在发送https请求的时候，关闭证书认证
    response = session.post(url=captcha_url, data=post_params, headers=headers, verify=False)
    json_obj = json.loads(response.text)
    if 'success' in json_obj:
        print('输入验证码正确')
        return True
    else:
        print('输入验证码不正确')
        return False


if __name__ == '__main__':
    zhihu_login()