知乎登录脚本

import time
from PIL import Image
from numpy import argmax
from selenium import webdriver
from selenium.webdriver import ActionChains
from selenium.webdriver.common.keys import Keys


class Zhihu:
    def __init__(self,
                 url='https://www.zhihu.com/',
                 exec_path="c:/users/python_package/chromedriver.exe"
                 ):
        self._login_failed = 1  # flag
        self._sig_err = 0  # flag
        self._init_webdriver(exec_path)
        try:
            self._login(url)
        finally:
            while 1:
                try:
                    handles = self.browser.window_handles
                    # time.sleep(5)
                except:
                    break
            self.browser.quit()

    def _init_webdriver(self, exec_path, other_options=None):
        options = webdriver.ChromeOptions()
        # 添加其他webdriver配置选项
        if isinstance(other_options, list):
            for each in other_options:
                options.add_argument(each)
        if isinstance(other_options, str):
            options.add_argument(other_options)
        # 隐藏window.navigator.webdriver(Chromedriver)
        options.add_experimental_option('excludeSwitches', ['enable-automation'])
        # options.add_argument('--headless') #隐藏浏览器界面
        # 实例化Chrome对象
        self.browser = webdriver.Chrome(executable_path=exec_path, options=options)

    def _login(self, url):
        self.browser.get(url)
        # 切换至密码登录界面
        self.browser.find_element_by_xpath('//div[@class="SignFlow-tabs"]/div[2]').click()

        while self._login_failed:
            current_page = self.browser.current_url
            if self._sig_err == 0:
                self._input()
            self._click_login_button()
            err = self._err_catch()
            print(err)
            if '验证码' in err or '倒立的文字' in err:
                self._sig_err = 1
                self._sign_process()
            elif len(err) > 0:
                self._sig_err = 0
                print('请重新输入账号密码.\n')
                continue
            else:
                pass
            if current_page != self.browser.current_url: 
                self._login_failed = 0
        print('Redirecting..')

    def _sign_process(self):
        # time.sleep(1)  # 睡眠1s等待验证码加载
        try:
            signimg = self.browser.find_element_by_xpath('//img[@alt="图形验证码"]')  # 查看是否正常加载验证码
            ce = signimg.get_attribute("class")  # 验证码类型
            self.browser.save_screenshot('capture.png')  # 保存登录页截图
            self.capture_sign(signimg.location['x'], signimg.location['y'], ce)  # 截取验证码区域用于后续处理
            if ce == 'Captcha-englishImg':  # 输入英文验证码
                signinput = self.browser.find_element_by_xpath('//div[@class="SignFlowInput"]//input[@name="captcha"]')
                x = input('输入验证码:\n')
                signinput.send_keys(x)
            else:  # Captcha-chineseImg 点击反向中文文字
                space = 200 // 8
                x = input('输入倒立文字序号(base1),多个则用逗号分隔\n')
                if ',' in x or '，' in x:
                    list_err = x.replace('，', ',').split(',')
                    mv_x = signimg.location['x'] + space * int(list_err[0]) - 1 / space
                    x_total = mv_x
                    mv_y = signimg.location['y'] + 22
                    y_total = mv_y
                    ActionChains(self.browser).move_by_offset(mv_x, mv_y).click().perform()
                    for i, each in enumerate(list_err):
                        if i == 0: 
                            continue
                        mv_x = space * (int(list_err[i]) - int(list_err[0]))
                        x_total += mv_x
                        ActionChains(self.browser).move_by_offset(mv_x, 0).click().perform()
                else:
                    mv_x = signimg.location['x'] + space * int(x) - 1 / space
                    mv_y = signimg.location['y'] + 22
                    x_total = mv_x
                    y_total = mv_y
                    ActionChains(self.browser).move_by_offset(mv_x, mv_y).click().perform()
                ActionChains(self.browser).move_by_offset(x_total, y_total).perform()
        except:
            print('未能成功加载验证码.')

    def _input(self):
        usr = input("username: ")
        pwd = input("password: ")
        u = self.browser.find_element_by_xpath('//input[@name="username"]')
        u.send_keys(Keys.CONTROL, "a")  # clear input form
        u.send_keys(Keys.DELETE)
        u.send_keys(usr)
        p = self.browser.find_element_by_xpath('//input[@name="password"]')
        p.send_keys(Keys.CONTROL, "a")  # clear input form
        p.send_keys(Keys.DELETE)
        p.send_keys(pwd)

    def sig_cv(self):
        pass

    def _click_login_button(self):
        self.browser.find_element_by_xpath('//button[contains(@class,"SignFlow-submitButton")]').click()

    def _err_catch(self):
        def _try_find(porp, cname):
            time.sleep(1)
            try:
                xpath_str = '//{}[contains(@class,"{}")]'.format(porp, cname)
                err = self.browser.find_elements_by_xpath(xpath_str)
                for each in err:
                    if len(each.text) > 0:
                        return each.text
                return ''
            except:
                return ''

        input_err = _try_find('div', 'SignFlowInput-errorMask')  # 账号密码错误
        sign_err_e = _try_find('div', 'Captcha-errorMessage')  # 英文验证码错误-请提交正确的验证码 :(
        sign_err_c = _try_find('span', 'Captcha-error')  # 中文验证码错误-请提交正确的验证码 :(
        errs = [input_err, sign_err_e, sign_err_c]
        index = argmax([len(each) for each in errs])
        return errs[int(index)]

    def capture_sign(self, x, y, sign_type, path='capture.png'):
        '''
        验证码处理
        :param x: sign loc_x
        :param y: sign loc_y
        :param sign_type: english(input form) or chinese(click inverse words)
        :param path: capture path
        :return: None
        '''
        img = Image.open(path)
        if sign_type == 'Captcha-englishImg':
            box = (x, y, x + 75, y + 30)  # 左、上、右、下
        else:  # Captcha-chineseImg
            box = (x, y, x + 200, y + 44)
        region = img.crop(box)
        region.save('sign.png')
        Image.open('sign.png').show()


a = Zhihu()