B站滑动验证图片下载

import json

import requests

from selenium import webdriver

from selenium.webdriver.common.desired_capabilities import DesiredCapabilities

import time

import re

headers = {

    "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/76.0.3809.132 Safari/537.36"

}

chrome_options = webdriver.ChromeOptions()

chrome_options.add_experimental_option('w3c', False)

caps = DesiredCapabilities.CHROME

caps['loggingPrefs'] = {'performance': 'ALL'}

driver = webdriver.Chrome(desired_capabilities=caps,options=chrome_options)

driver.get('https://passport.bilibili.com/login')

def input_click_01():

    input_name = driver.find_element_by_xpath("//input[@id='login-username']")

    input_pwd = driver.find_element_by_xpath("//input[@id='login-passwd']")

    input_name.send_keys("username")

    input_pwd.send_keys("passport")

    time.sleep(3)

    login_btn = driver.find_element_by_class_name("btn-login")

    login_btn.click()

    time.sleep(5)

def browser_log_02():

    browser_log_list = driver.get_log("performance")

    # 先保存到文件，利于测试，和后面的正则匹配

    logs = [json.loads(log['message'])['message'] for log in browser_log_list]

    with open('devtools.json', 'w') as f:

        json.dump(logs, f, indent=4, ensure_ascii=False)

    with open('devtools.json', 'r') as f:

        browser_log = f.read()

    print("浏览器日志获取完成")

    return browser_log

def get_response_img_url_03(browser_log):

    # 获取requestId

    # 获取到的有两种，取前者，暂时没出错，出现异常再进行筛选

    pat = r"""https://api\.geetest\.com/get\.php\?is_next.*?\".*?\"requestId\": \"(\d+?\.\d+?)\","""

    requestId = re.findall(pat, browser_log, re.S)[0]

    # print(requestId)

    # 最重要的一步：调用接口，通过requestId获取请求的响应

    response_dict = driver.execute_cdp_cmd('Network.getResponseBody', {'requestId': requestId})

    body = response_dict["body"]

    # print(body)

    # 从响应中获取图片链接

    fullbg = re.findall(r"fullbg\":.\"(.*?)\",",body)

    bg = re.findall(r"\"bg\":.\"(.*?)\",",body)

    fullbg_url = "https://static.geetest.com/" + fullbg[0]

    bg_url = "https://static.geetest.com/" + bg[0]

    return fullbg_url,bg_url

def get_img_04(fullbg_url,bg_url):

    # 请求

    origin_img_data = requests.get(fullbg_url, headers=headers).content

    fix_img_data = requests.get(bg_url, headers=headers).content

    # 先保存图片

    with open("full_image.jpg", "wb") as f:

        f.write(origin_img_data)

    with open("fix_image.jpg", "wb") as f:

        f.write(fix_img_data)

    print("保存图片完成")

    return "full_image.jpg","fix_image.jpg"

def main():

    input_click_01()

    log_data = browser_log_02()

    url_tuple = get_response_img_url_03(log_data)

    get_img_04(*url_tuple)

    driver.close()

if __name__ == '__main__':

    main()