《python3网络爬虫开发实战》--验证码的识别

1.图形验证码：

中国知网：http://my.cnki.net/elibRegister/CommonRegister.aspx

 1 import tesserocr
 2 from PIL import Image
 3 
 4 image = Image.open('code2.jpg')
 5 image = image.convert('L')
 6 threshold = 180
 7 table = []
 8 for i in range(256):
 9     if i < threshold:
10         table.append(0)
11     else:
12         table.append(1)
13 
14 image = image.point(table, '1')
15 #image = image.convert('1')
16 #image.show()
17 
18 result = tesserocr.image_to_text(image)
19 print(result)

2. 极验滑动验证码的识别

https://www.geetest.com/Sensebot

对于应用了极验验证码的网站如果我们直接模拟表单提交，加密参数的构造是个问题，需要分析其加密和校验逻辑，相对烦琐。所以我们采用直接模拟浏览器动作的方式来完成验证。

可以使用 Selenium来完全模拟人的行为的方式来完成验证，此验证成本相比直接去识别加密算法少很多。

https://account.geetest.com/login

(I)模拟点击验证按钮。

(2)识别附动缺口的位置。

(3)模拟拖动滑块。

  1 import time
  2 from io import BytesIO
  3 from PIL import Image
  4 from selenium import webdriver
  5 from selenium.webdriver import ActionChains
  6 from selenium.webdriver.common.by import By
  7 from selenium.webdriver.support.ui import WebDriverWait
  8 from selenium.webdriver.support import expected_conditions as EC
  9 
 10 EMAIL = 'zcs@163.com'
 11 PASSWORD = '123'
 12 BORDER = 6
 13 #INIT_LEFT = 60
 14 
 15 
 16 class CrackGeetest():
 17     def __init__(self):
 18         self.url = 'https://account.geetest.com/login'
 19         self.browser = webdriver.Chrome()
 20         self.wait = WebDriverWait(self.browser, 20)
 21         self.email = EMAIL
 22         self.password = PASSWORD
 23 
 24     def __del__(self):
 25         self.browser.close()
 26 
 27     def get_geetest_button(self):
 28         """
 29         获取初始验证按钮
 30         :return:
 31         """
 32         button = self.wait.until(EC.element_to_be_clickable((By.CLASS_NAME, 'geetest_radar_tip')))
 33         return button
 34 
 35     def get_position(self):
 36         """
 37         获取验证码位置
 38         :return: 验证码位置元组
 39         """
 40         img = self.wait.until(EC.presence_of_element_located((By.CLASS_NAME, 'geetest_canvas_img')))
 41         time.sleep(2)
 42         location = img.location
 43         size = img.size
 44         top, bottom, left, right = location['y'], location['y'] + size['height'], location['x'], location['x'] + size[
 45             'width']
 46         return (top, bottom, left, right)
 47 
 48     def get_screenshot(self):
 49         """
 50         获取网页截图
 51         :return: 截图对象
 52         """
 53         screenshot = self.browser.get_screenshot_as_png()
 54         screenshot = Image.open(BytesIO(screenshot))
 55         return screenshot
 56 
 57     def get_slider(self):
 58         """
 59         获取滑块
 60         :return: 滑块对象
 61         """
 62         slider = self.wait.until(EC.element_to_be_clickable((By.CLASS_NAME, 'geetest_slider_button')))
 63         return slider
 64 
 65     def get_geetest_image(self, name='captcha.png'):
 66         """
 67         获取验证码图片
 68         :return: 图片对象
 69         """
 70         top, bottom, left, right = self.get_position()
 71         print('验证码位置', top, bottom, left, right)
 72         screenshot = self.get_screenshot()
 73         # crop将图片裁剪
 74         captcha = screenshot.crop((left, top, right, bottom))
 75         captcha.save(name)
 76         return captcha
 77 
 78     def open(self):
 79         """
 80         打开网页输入用户名密码
 81         :return: None
 82         """
 83         self.browser.get(self.url)
 84         email = self.wait.until(EC.presence_of_element_located((By.ID, 'email')))
 85         password = self.wait.until(EC.presence_of_element_located((By.ID, 'password')))
 86         email.send_keys(self.email)
 87         password.send_keys(self.password)
 88 
 89     def get_gap(self, image1, image2):
 90         """
 91         获取缺口偏移量
 92         :param image1: 不带缺口图片
 93         :param image2: 带缺口图片
 94         :return:
 95         """
 96         left = 60
 97         for i in range(left, image1.size[0]):
 98             for j in range(image1.size[1]):
 99                 if not self.is_pixel_equal(image1, image2, i, j):
100                     left = i
101                     return left
102         return left
103 
104     def is_pixel_equal(self, image1, image2, x, y):
105         """
106         判断两个像素是否相同
107         :param image1: 图片1
108         :param image2: 图片2
109         :param x: 位置x
110         :param y: 位置y
111         :return: 像素是否相同
112         """
113         # 取两个图片的像素点
114         pixel1 = image1.load()[x, y]
115         pixel2 = image2.load()[x, y]
116         threshold = 60
117         if abs(pixel1[0] - pixel2[0]) < threshold and abs(pixel1[1] - pixel2[1]) < threshold and abs(
118                 pixel1[2] - pixel2[2]) < threshold:
119             return True
120         else:
121             return False
122 
123     def get_track(self, distance):
124         """
125         根据偏移量获取移动轨迹
126         :param distance: 偏移量
127         :return: 移动轨迹
128         """
129         # 移动轨迹
130         track = []
131         # 当前位移
132         current = 0
133         # 减速阈值
134         mid = distance * 4 / 5
135         # 计算间隔
136         t = 0.2
137         # 初速度
138         v = 0
139 
140         while current < distance:
141             if current < mid:
142                 # 加速度为正2
143                 a = 2
144             else:
145                 # 加速度为负3
146                 a = -3
147             # 初速度v0
148             v0 = v
149             # 当前速度v = v0 + at
150             v = v0 + a * t
151             # 移动距离x = v0t + 1/2 * a * t^2
152             move = v0 * t + 1 / 2 * a * t * t
153             # 当前位移
154             current += move
155             # 加入轨迹
156             track.append(round(move))
157         return track
158 
159     def move_to_gap(self, slider, track):
160         """
161         拖动滑块到缺口处
162         :param slider: 滑块
163         :param track: 轨迹
164         :return:
165         """
166         ActionChains(self.browser).click_and_hold(slider).perform()
167         for x in track:
168             ActionChains(self.browser).move_by_offset(xoffset=x, yoffset=0).perform()
169         time.sleep(0.5)
170         ActionChains(self.browser).release().perform()
171 
172     def login(self):
173         """
174         登录
175         :return: None
176         """
177         submit = self.wait.until(EC.element_to_be_clickable((By.CLASS_NAME, 'login-btn')))
178         submit.click()
179         time.sleep(10)
180         print('登录成功')
181 
182     def crack(self):
183         # 输入用户名密码
184         self.open()
185         # 点击验证按钮
186         button = self.get_geetest_button()
187         button.click()
188         # 获取验证码图片
189         image1 = self.get_geetest_image('captcha1.png')
190         # 点按呼出缺口
191         slider = self.get_slider()
192         slider.click()
193         # 获取带缺口的验证码图片
194         image2 = self.get_geetest_image('captcha2.png')
195         # 获取缺口位置
196         gap = self.get_gap(image1, image2)
197         print('缺口位置', gap)
198         # 减去缺口位移
199         gap -= BORDER
200         # 获取移动轨迹
201         track = self.get_track(gap)
202         print('滑动轨迹', track)
203         # 拖动滑块
204         self.move_to_gap(slider, track)
205 
206         success = self.wait.until(
207             EC.text_to_be_present_in_element((By.CLASS_NAME, 'geetest_success_radar_tip_content'), '验证成功'))
208         print(success)
209 
210         # 失败后重试
211         if not success:
212             self.crack()
213         else:
214             self.login()
215 
216 
217 if __name__ == '__main__':
218     crack = CrackGeetest()
219     crack.crack()

但是，当我们截取图片的时候，网站将图片分割为不同的图片随机组合，我们就无法使用这一方法。

3.点触验证码的识别

点触的网址挂了，

4. 微博宫格识别