滑动验证码验证

Posted 2021-03-24 alex-13

selenium +chrome+ firefox + webdriver 遇到的坑

lunix中启动webdriver时报错一：

测试代码为：

1.  
   #!/usr/bin/python
2.  
   # -*- coding: utf-8 -*-
3.  
    
4.  
    
5.  
   from selenium import webdriver
6.  
    
7.  
   driver = webdriver.Firefox()
8.  
   driver.get("https://www.baidu.com")

运行报错信息如下：

1.  
   Traceback (most recent call last):
2.  
   File "maimai_web.py", line 14, in <module>
3.  
   driver = webdriver.Firefox()
4.  
   File "/usr/local/python3.6/lib/python3.6/site-packages/selenium/webdriver/firefox/webdriver.py", line 152, in __init__
5.  
   keep_alive=True)
6.  
   File "/usr/local/python3.6/lib/python3.6/site-packages/selenium/webdriver/remote/webdriver.py", line 98, in __init__
7.  
   self.start_session(desired_capabilities, browser_profile)
8.  
   File "/usr/local/python3.6/lib/python3.6/site-packages/selenium/webdriver/remote/webdriver.py", line 188, in start_session
9.  
   response = self.execute(Command.NEW_SESSION, parameters)
10.  
    File "/usr/local/python3.6/lib/python3.6/site-packages/selenium/webdriver/remote/webdriver.py", line 256, in execute
11.  
    self.error_handler.check_response(response)
12.  
    File "/usr/local/python3.6/lib/python3.6/site-packages/selenium/webdriver/remote/errorhandler.py", line 194, in check_response
13.  
    raise exception_class(message, screen, stacktrace)
14.  
    selenium.common.exceptions.WebDriverException: Message: Process unexpectedly closed with status 1

处理方法：

1.  
   #!/usr/bin/python
2.  
   # -*- coding: utf-8 -*-
3.  
    
4.  
    
5.  
   from pyvirtualdisplay import Display
6.  
   from selenium import webdriver
7.  
    
8.  
    
9.  
   display = Display(visible=0, size=(1920, 1080))
10.  
    display.start()
11.  
    driver = webdriver.Firefox()
12.  
    driver.get("https://www.baidu.com")

结果：

运行ok，搞定！

坑二、webdriver实例化报错

采用多线程调用webdriver时候，偶尔会出现这样的错：selenium.common.exceptions.WebDriverException: Message: connection refused

1.  
   Exception in thread Thread-2:
2.  
   Traceback (most recent call last):
3.  
   File "/usr/local/python3.6/lib/python3.6/threading.py", line 916, in _bootstrap_inner
4.  
   self.run()
5.  
   File "/usr/local/python3.6/lib/python3.6/threading.py", line 864, in run
6.  
   self._target(*self._args, **self._kwargs)
7.  
   File "maimai_tran_account_driver.py", line 591, in debug
8.  
   t = TrainAccount(count,lock)
9.  
   File "maimai_tran_account_driver.py", line 32, in __init__
10.  
    self.chrome = webdriver.Firefox()
11.  
    File "/usr/local/python3.6/lib/python3.6/site-packages/selenium/webdriver/firefox/webdriver.py", line 152, in __init__
12.  
    keep_alive=True)
13.  
    File "/usr/local/python3.6/lib/python3.6/site-packages/selenium/webdriver/remote/webdriver.py", line 98, in __init__
14.  
    self.start_session(desired_capabilities, browser_profile)
15.  
    File "/usr/local/python3.6/lib/python3.6/site-packages/selenium/webdriver/remote/webdriver.py", line 188, in start_session
16.  
    response = self.execute(Command.NEW_SESSION, parameters)
17.  
    File "/usr/local/python3.6/lib/python3.6/site-packages/selenium/webdriver/remote/webdriver.py", line 256, in execute
18.  
    self.error_handler.check_response(response)
19.  
    File "/usr/local/python3.6/lib/python3.6/site-packages/selenium/webdriver/remote/errorhandler.py", line 194, in check_response
20.  
    raise exception_class(message, screen, stacktrace)
21.  
    selenium.common.exceptions.WebDriverException: Message: connection refused

 查看geckodriver.log具体报错信息。

坑三、模拟器被反爬

原因是在webdriver发送请求的时候，会有webdriver的js判断，当检测到此字段时会被作为爬虫处理，应对策略如下。

工具：mitmproxy做代理，替换掉请求里面的webdriver为别的字段

部分代码如下：

1.  
   if "/_next/static/js/common_pdd" in flow.request.url:
2.  
   flow.response.text = flow.response.text.replace("webdriver", "userAgent")

坑四、滑动验证码验证失败

同样的代码，chromedriver验证码通过，firefox滑动到正常位置报失败，最后发现原因是firefox在滑动模块的时候速度太慢被机器识别出来，解决方法，增大滑动的速度，附上滑动验证的部分代码，如下：

1.  
   def crack_geetest(self, max_retry=10):
2.  
   driver = self.driver
3.  
   l = self.logger
4.  
   l.info("process handle geetest captcha...")
5.  
    
6.  
   def get_position():
7.  
   """
8.  
   获取验证码位置
9.  
   :return: 验证码位置元组
10.  
    """
11.  
    img = driver.find_element_by_xpath(‘//div[@class="geetest_canvas_img geetest_absolute"]‘)
12.  
    time.sleep(2)
13.  
    location = img.location
14.  
    size = img.size
15.  
    top, bottom, left, right = location[‘y‘], location[‘y‘] + size[‘height‘], location[‘x‘], location[‘x‘] +
16.  
    size[‘width‘]
17.  
    return (top, bottom, left, right)
18.  
     
19.  
    def get_geetest_image(name):
20.  
    """
21.  
    获取验证码图片
22.  
    :return: 图片对象
23.  
    """
24.  
    full_img_path = ‘./zhilian_screenshot_{}.png‘.format(self.account[‘user_id‘])
25.  
    driver.save_screenshot(filename=full_img_path)
26.  
    image = Image.open(fp=full_img_path, mode=‘r‘)
27.  
    top, bottom, left, right = get_position()
28.  
    print(‘验证码位置:({},{},{},{})‘.format(left, top, right, bottom))
29.  
    t = driver.execute_script(‘var q=document.documentElement.scrollTop; return q;‘)
30.  
    print(‘验证码位置:({},{},{},{})‘.format(left, top - int(t), right, bottom - int(t)))
31.  
    print(‘p--->>>‘, t)
32.  
    captcha = image.crop((left, top - int(t), right, bottom - int(t)))
33.  
    captcha_file_name = ‘./zhilian_captcha_{}_{}.png‘.format(self.account[‘user_id‘], name)
34.  
    captcha.save(captcha_file_name)
35.  
    return captcha, captcha_file_name
36.  
     
37.  
    def get_slider():
38.  
    """
39.  
    获取滑块
40.  
    :return: 滑块对象
41.  
    """
42.  
    slider = driver.find_element_by_xpath(‘//div[@class="geetest_slider_button"]‘)
43.  
    return slider
44.  
     
45.  
    def get_gap(captcha_file_name):
46.  
    """
47.  
    获取缺口偏移量
48.  
    :param image1: 不带缺口图片
49.  
    :param image2: 带缺口图片
50.  
    :return:
51.  
    """
52.  
    res = self.dama2.decode_captcha(6137, captcha_file_name)
53.  
    print(res)
54.  
    # (‘b800b4f6-0d9a-40e2-a972-d87c91582b46‘, [(176, 101)])
55.  
    return int(res[1][0][0])
56.  
     
57.  
    def calculate_tracks(distance):
58.  
    def generate_rand(n, sum_v): # 随机生成n个总和为sum_v的list
59.  
    Vector = [random.randint(1, 3) for _ in range(n)]
60.  
    Vector = [int(i / sum(Vector) * sum_v) for i in Vector]
61.  
    if sum(Vector) < sum_v:
62.  
    res = sum_v - sum(Vector)
63.  
    for i in range(res):
64.  
    Vector[random.randint(0, n - 1)] += 1
65.  
    return [0 - i for i in Vector]
66.  
     
67.  
    back_dis = random.randint(16, 26)
68.  
    distance += back_dis # 先滑过一点，最后再反着滑动回来
69.  
    v = 0
70.  
    t = 0.2
71.  
    forward_tracks = []
72.  
     
73.  
    current = 0
74.  
    mid = distance * 3 / 5
75.  
    while current < distance:
76.  
    if current < mid:
77.  
    a = 2
78.  
    else:
79.  
    a = -3
80.  
     
81.  
    s = v * t + 0.5 * a * (t ** 2)
82.  
    v = v + a * t
83.  
    current += s
84.  
    forward_tracks.append(round(s))
85.  
     
86.  
    # 反着滑动到准确位置
87.  
    back_tracks = generate_rand(15, back_dis) # 总共等于 back_dis
88.  
    return {‘forward_tracks‘: forward_tracks, ‘back_tracks‘: back_tracks}
89.  
     
90.  
    def move_to_gap(slider, tracks):
91.  
    """
92.  
    拖动滑块到缺口处
93.  
    :param slider: 滑块
94.  
    :param track: 轨迹
95.  
    :return:
96.  
    """
97.  
    ActionChains(driver).click_and_hold(slider).perform()
98.  
     
99.  
    # 往后移动
100.  
     for i in tracks[‘forward_tracks‘]:
101.  
     ActionChains(driver).move_by_offset(i, 0).perform()
102.  
      
103.  
     # 往回移动
104.  
     time.sleep(0.5)
105.  
     for i in tracks[‘back_tracks‘]:
106.  
     ActionChains(driver).move_by_offset(i, 0).perform()
107.  
      
108.  
     # 小范围震荡一下
109.  
     # time.sleep(0.3)
110.  
     random_sc = random.randint(3, 8)
111.  
     ActionChains(driver).move_by_offset(0-random_sc, 0).perform()
112.  
     time.sleep(0.5)
113.  
     ActionChains(driver).move_by_offset(random_sc, 0).perform()
114.  
      
115.  
     # 释放
116.  
     time.sleep(0.5)
117.  
     ActionChains(driver).release().perform()
118.  
      
119.  
     def crack(retry=0):
120.  
     # 输入用户名密码
121.  
     # 点击验证按钮
122.  
     # 获取验证码图片
123.  
     print(‘get_geetest_image‘)
124.  
     captcha_obj, captcha_file_name = get_geetest_image(‘2‘)
125.  
     gap = get_gap(captcha_file_name)
126.  
     l.info(‘缺口位置:{}‘.format(gap))
127.  
     print(‘缺口位置:{}‘.format(gap))
128.  
     # 减去起始缺口位移
129.  
     BORDER = 29
130.  
     gap -= BORDER
131.  
     # 获取移动轨迹
132.  
     track = calculate_tracks(gap)
133.  
     l.info(‘滑动轨迹:{}‘.format(track))
134.  
     print(‘滑动轨迹:{}‘.format(track))
135.  
     # # 拖动滑块
136.  
     slider = get_slider()
137.  
     move_to_gap(slider, track)
138.  
     driver.save_screenshot(‘./zhilian_capresult_{}_{}.png‘.format(self.account[‘user_id‘], retry))
139.  
     #
140.  
     time.sleep(3)
141.  
     # #
142.  
     result = driver.find_element_by_xpath(‘//div[@class="geetest_result_title"]‘).get_attribute(‘textContent‘)
143.  
     l.info(result)
144.  
     print(result)
145.  
     return result
146.  
      
147.  
     retry = 1
148.  
     while True:
149.  
     l.info(f‘{retry}/{max_retry} crack geetest.‘)
150.  
     if retry == max_retry:
151.  
     l.info("max retry reached, return False")
152.  
     return False
153.  
     success = crack(retry)
154.  
     if ‘秒的速度超过‘ in success or ‘passport.lagou.com/login/login‘ not in driver.current_url:
155.  
     l.info("crack succeeded!")
156.  
     print("crack succeeded!")
157.  
     return True
158.  
     elif ‘拖动滑块将悬浮图像正确拼合‘ in success:
159.  
     retry += 1
160.  
     l.info("crack failed, retry:{}/{}".format(retry, max_retry))
161.  
     driver.find_element_by_xpath(‘//a[@class="geetest_refresh_1"]‘).click()
162.  
     time.sleep(5)
163.  
     continue
164.  
     else:
165.  
     time.sleep(5)
166.  
     retry += 1
167.  
     l.info("crack failed, retry:{}/{}".format(retry, max_retry))
168.  
     continue

来源：https://blog.csdn.net/wenq_yang/article/details/81258932