selenium实现登录百度（自动识别简单验证码）

需要做的工作

0、工程结构

 

1、代码：

 ①baidu_login.py

import re
import os
import sys
import time
import random
from selenium import webdriver
from PIL import Image, ImageEnhance
import pytesseract
from func import base642str, str2base64


def input_account(account='请传入账号参数', xpath_rule="//input[@id='TANGRAM__PSP_3__userName']"):
    '''模拟输入账号
    :param account: 账号
    :param xpath_rule: 账号输入框的xpath定位规则
    :return:
    '''
    input_box1 = driver.find_element_by_xpath(xpath_rule)
    input_box1.send_keys(account)
    time.sleep(0.5)


def input_pwd(pwd, xpath_rule="//input[@id='TANGRAM__PSP_3__password']"):
    '''模拟输入密码
    :param account: base64后的密码
    :param xpath_rule: 密码输入框的xpath定位规则
    :return:
    '''
    input_box2 = driver.find_element_by_xpath(xpath_rule)
    input_box2.clear()  # 清空密码
    input_box2.send_keys(base642str(pwd))
    time.sleep(0.5)


def input_verify_code(verify_code, xpath_rule="TANGRAM__PSP_3__verifyCode"):
    '''模拟输入验证码
    :param account: base64后的密码
    :param xpath_rule: 密码输入框的xpath定位规则
    :return:
    '''
    driver.find_element_by_id(xpath_rule).send_keys(verify_code.strip())
    time.sleep(0.5)


def identify_verify_code(rootpath, pic_name="screenImg.png"):
    '''tesseract识别百度验证码
    :param rootpath: 验证码图片保存的文件夹路径
    :param pic_name: 验证码图片保存的文件名
    :return: 识别后的验证码文本
    '''
    # 截图或验证码图片保存地址
    screenImg = os.path.join(rootpath, pic_name)
    # 浏览器页面截屏
    time.sleep(3)
    driver.get_screenshot_as_file(screenImg)
    # 定位验证码位置及大小
    location = driver.find_element_by_id('TANGRAM__PSP_3__verifyCodeImg').location
    size = driver.find_element_by_id('TANGRAM__PSP_3__verifyCodeImg').size
    left = location['x']
    top = location['y']
    right = location['x'] + size['width']
    bottom = location['y'] + size['height']
    # 从文件读取截图，截取验证码位置再次保存
    img = Image.open(screenImg).crop((left, top, right, bottom))
    img = img.convert('L')  # 转换模式：L | RGB
    img = ImageEnhance.Contrast(img)  # 增强对比度
    img = img.enhance(2.0)  # 增加饱和度
    img.save(screenImg)
    print("图片验证码以保存：%s" % screenImg)
    # 再次读取识别验证码
    print("开始读取识别图片验证码：%s" % screenImg)
    img = Image.open(screenImg)
    verifycode = pytesseract.image_to_string(img)
    print("识别结果：%s" % verifycode)
    return verifycode


def click_a_link(xpath_rule="//p[@id='TANGRAM__PSP_3__footerULoginBtn']"):
    '''点击一个链接
    :param xpath_rule: 被点击链接的xpath定位规则
    :return:
    '''
    input_box0 = driver.find_element_by_xpath(xpath_rule)
    input_box0.click()


def click_a_id_link(id_rule="TANGRAM__PSP_3__verifyCodeChange"):
    '''点击一个链接
    :param id_rule: 被点击链接的id定位规则
    :return:
    '''
    input_box0 = driver.find_element_by_id(id_rule)
    input_box0.click()


def is_need_verify_code():
    '''判断是否需要验证码
    :return: 需要验证码返回True，否则False
    '''
    imgsrc = driver.find_element_by_id("TANGRAM__PSP_3__verifyCodeImg").get_attribute('src')
    if re.match(r'https://passport.baidu.com/cgi-bin/genimage.*', imgsrc):
        return True
    else:
        return False


def get_id_node_text(id_rule="TANGRAM__PSP_3__error"):
    '''获取id节点提示信息
    :param id_rule:id节点的id匹配规则（id属性的值）
    :return:该id节点中的文本信息
    '''
    one_node = driver.find_element_by_id(id_rule)
    text_info = one_node.text
    return text_info


def is_login_success():
    '''判断登录是否成功
    :return: 登录成功返回True，否则False
    '''
    current_title = driver.title.strip()
    if current_title.startswith("登录"):
        return False
    else:
        return True


def deal_much_pop_up_window():
    '''处理手机验证码认证反复弹窗
    :return:
    '''
    i = 0
    while True:  # 处理手机验证码认证反复弹窗
        try:
            one3_click = driver.find_element_by_id("TANGRAM__%s__header_a" % (22 + i))  # 22+i在应对弹窗的关闭按钮id名称发生变化。
        except Exception as e:
            print("无需手机验证码")
            break
        else:
            print("第 %s 次弹出安全验证，要求获取手机验证码" % (i + 1))
            time.sleep(0.5)
            print("1s后自动选择无需手机验证码")
            time.sleep(1)
            one3_click.click()
            print("1s后自动点击登陆")
            time.sleep(1)
            click_a_link(xpath_rule="//input[@id='TANGRAM__PSP_3__submit']")  # 点击登录，提交表单
            time.sleep(2)
            # 判断是否成功登陆
            current_title = driver.title.strip()
            if current_title.startswith("登录"):
                print('333-登陆失败...')
                time.sleep(0.5)
                print('333-2秒后自动重试...')
                i = i + 1
                time.sleep(2)
                continue
            else:
                print("打印标题")
                print(driver.title)
                print('333-登录成功...')
                sys.exit()  # 程序终止


def deal_a_pop_up_window(xpath_rule="//input[@id='TANGRAM__PSP_27__rebindGuideCancel']"):
    '''处理一次弹窗
    :param xpath_rule:处理弹窗的按钮/链接的xpath匹配规则
    :return:
    '''
    # 判断是否需要手机号绑定确认
    try:
        # 绑定手机号确认
        one_click = driver.find_element_by_xpath(xpath_rule)
    except Exception as e:
        print("无需绑定手机号确认")
    else:
        print("弹出了绑定手机号确认，1s后自动选择不需要")
        time.sleep(1)
        one_click.click()


if __name__ == '__main__':
    # 将mm隐匿化
    # print(str2base64("123456"))
    # str1_base64="MTIzNDU2"
    # print(base642str(str1_base64))
    # sys.exit()

    # 账号和密码准备
    account = random.randint(0, 1000000)  # 随机数字作为账号
    pwd = "MTIzNDU2"
    print("account: %s" % account)
    # 最大登录次数
    max_login = 16
    # 当前目录设置为根路径
    ROOT_PATH = os.getcwd()
    print('000-正在启用selenium...')
    # 调用环境变量指定的PhantomJS浏览器创建浏览器对象
    chromedriver_exe_path = os.path.join(ROOT_PATH, "chromedriver.exe")
    driver = webdriver.Chrome(chromedriver_exe_path)
    print('000-启用OK')

    # 请求登录页面
    url = 'https://passport.baidu.com/v2/?login'
    print('111-selenium正在请求页面：%s' % url)
    driver.get(url)  # get方法请求页面，获取响应
    print('111-请求OK')

    print("打印标题")
    print(driver.title)

    # 点击账号和密码登录
    click_a_link()

    print('222-selenium正在填写表单...')
    time.sleep(1)
    # 第一次尝试登录

    # 模拟填写账号
    input_account(account)

    # 模拟填写密码
    input_pwd(pwd)

    # 判断是否需要验证码
    is_need = is_need_verify_code()
    if is_need:  # 需要验证码
        print("需要验证码")
        # 自动识别和模拟填写验证码
        code = identify_verify_code(rootpath=ROOT_PATH, pic_name="screenImg.png")  # 自动识别验证码
        input_verify_code(code)  # 模拟填写验证码
    else:  # 不需要验证码
        print("不需要验证码")
        print('222-填写表单OK')
        time.sleep(1)

    print('333-selenium提交表单...')
    click_a_link(xpath_rule="//input[@id='TANGRAM__PSP_3__submit']")  # 点击登录，提交表单
    print("第 %s 次尝试登录" % 1)
    time.sleep(3)

    # 处理反复弹窗（手机验证码）：点击关闭按钮
    deal_much_pop_up_window()
    # 处理一次弹窗（绑定手机号确认）：点击不需要修改
    deal_a_pop_up_window()

    for login_i in range(max_login - 1):
        # 判断是否登录成
        is_logined = is_login_success()
        if is_logined:
            print("登录成功")
            break
        else:
            print("第 %s 次登录失败，正在尝试重新登录..." % (login_i + 1))
            # 第二次尝试登录
            print("=>第 %s 次尝试登录" % (login_i + 2))
            error_info = get_id_node_text()
            if ("帐号或密码错误" in error_info) or ("用户名或密码有误"):  # 第二次尝试登录2.1
                print("正在尝试重新输入密码...")
                # 模拟填写密码
                input_pwd(pwd)
                # 判断是否需要验证码
                is_need = is_need_verify_code()
                if is_need:
                    print("需要验证码")
                    # 点击更新验证码
                    click_a_id_link()
                    # 自动识别和模拟填写验证码
                    code = identify_verify_code(rootpath=ROOT_PATH, pic_name="screenImg.png")  # 自动识别验证码
                    input_verify_code(code)  # 模拟填写验证码
                    print("提示：验证码错误，js会使得提交无效")
                else:
                    print("不需要验证码")

            elif ("验证码" in error_info):  # 第二次尝试登录2.2
                print("正在尝试重新输入了验证码和密码...")
                time.sleep(1)
                # 模拟填写密码
                input_pwd(pwd)
                # 点击更新验证码
                click_a_id_link()
                # 自动识别和模拟填写验证码
                code = identify_verify_code(rootpath=ROOT_PATH, pic_name="screenImg.png")  # 自动识别验证码
                input_verify_code(code)  # 模拟填写验证码
                print("提示：验证码错误，js会使得提交无效")
                time.sleep(2)

            else:
                print("其他未知异常：登录失败")
                sys.exit()  # 程序终止

            print('222-填写表单OK')
            time.sleep(1)

            print('333-selenium提交表单...')
            click_a_link(xpath_rule="//input[@id='TANGRAM__PSP_3__submit']")  # 点击登录，提交表单
            time.sleep(3)

    # 打印标题
    print("打印标题")
    print(driver.title)
    time.sleep(1)

    # 判断是否成功登陆
    is_logined = is_login_success()
    if is_logined:
        print('333-登录成功...')
    else:
        print('333-登陆失败...')

②func.py

import base64
import timedef timestamp2datems(timestamp):'''时间戳转为日期字串，精确到ms。单位s:param timestamp:时间戳:return:日期字串'''local_time = time.localtime(timestamp)# data_head = time.strftime("%Y-%m-%d %H:%M:%S", local_time)data_head = time.strftime("%Y-%m-%d_%H-%M-%S", local_time)data_secs = (timestamp - int(timestamp)) * 1000dt_ms = "%s.%03d" % (data_head, data_secs)# print(dt_ms)return dt_msdef bit2humanView(bit_val):'''文件大小bit转为人类易读大小bit、KB、MB:param bit_val:字节数值:return:人类易读大小和单位'''is2kb = int(bit_val / 1042)  # 转换为kb取整is2mb = int(bit_val / 1024 / 1024)  # 转为mb取整is2gb = int(bit_val / 1024 / 1024 / 1024)  # 转为gb取整if is2gb is not 0:gb_val = bit_val / 1024 / 1024 / 1024return "%.2f GB" % gb_valif is2mb is not 0:mb_val = bit_val / 1024 / 1024return "%.2f MB" % mb_valif is2kb is not 0:kb_val = bit_val / 1024return "%.2f KB" % kb_valreturn "%s bit" % bit_valdef str2base64(pwd_decode_str):'''明文str转为base64密文:param pwd_decode_str: 明文str:return: base64密文'''base64_encrypt = base64.b64encode(pwd_decode_str.encode('utf-8'))pwd_encode_str = str(base64_encrypt, 'utf-8')return pwd_encode_strdef base642str(pwd_encode_str):'''base64密文转为明文str:param pwd_encode_str: base64密文:return: 明文str'''base64_decrypt = base64.b64decode(pwd_encode_str.encode('utf-8'))pwd_decode_str = str(base64_decrypt, 'utf-8')return pwd_decode_str

　　

③依赖包：requirements.txt

Pillow==6.0.0
pytesseract==0.2.6
selenium==3.141.0

　　

2、selenium的相关支持：

①google浏览器；

②相应版本的浏览器驱动chromedriver.exe

 

3、PIL安装：pip  install  pillow

 

4、图像识别工具：

①windows安装tesseract.exe 

②python安装pytesseract