浙大python读者验证码_Python爬虫学习笔记之极限滑动验证码的识别

1 importtime2 from io importBytesIO3 from PIL importImage4 from selenium importwebdriver5 from selenium.webdriver importActionChains6 from selenium.webdriver.common.by importBy7 from selenium.webdriver.support.ui importWebDriverWait8 from selenium.webdriver.support importexpected_conditions as EC9

10 EMAIL = '' # 邮箱 密码需要自己注册

11 PASSWORD = ''

12 BORDER = 6

13 INIT_LEFT = 60

14

15

16 classCrackGeetest():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 =EMAIL22 self.password =PASSWORD23

24 def __del__(self):25 self.browser.close()26

27 defget_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 returnbutton34

35 defget_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.location43 size =img.size44 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 defget_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 returnscreenshot56

57 defget_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 returnslider64

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 captcha =screenshot.crop((left, top, right, bottom))74 captcha.save(name)75 returncaptcha76

77 defopen(self):78 """

79 打开网页输入用户名密码80 :return: None81 """

82 self.browser.get(self.url)83 email = self.wait.until(EC.presence_of_element_located((By.ID, 'email')))84 password = self.wait.until(EC.presence_of_element_located((By.ID, 'password')))85 email.send_keys(self.email)86 password.send_keys(self.password)87

88 defget_gap(self, image1, image2):89 """

90 获取缺口偏移量91 :param image1: 不带缺口图片92 :param image2: 带缺口图片93 :return:94 """

95 left = 60

96 for i inrange(left, image1.size[0]):97 for j in range(image1.size[1]):98 if notself.is_pixel_equal(image1, image2, i, j):99 left =i100 returnleft101 returnleft102

103 defis_pixel_equal(self, image1, image2, x, y):104 """

105 判断两个像素是否相同106 :param image1: 图片1107 :param image2: 图片2108 :param x: 位置x109 :param y: 位置y110 :return: 像素是否相同111 """

112 #取两个图片的像素点

113 pixel1 =image1.load()[x, y]114 pixel2 =image2.load()[x, y]115 threshold = 60

116 if abs(pixel1[0] - pixel2[0]) < threshold and abs(pixel1[1] - pixel2[1]) < threshold andabs(117 pixel1[2] - pixel2[2]) <118 returntrue119 else:120 returnfalse121>

122 defget_track(self, distance):123 """

124 根据偏移量获取移动轨迹125 :param distance: 偏移量126 :return: 移动轨迹127 """

128 #移动轨迹

129 track =[]130 #当前位移

131 current =0132 #减速阈值

133 mid = distance * 4 / 5

134 #计算间隔

135 t = 0.2

136 #初速度

137 v =0138

139 while current <140 if current>

142 a = 2

143 else:144 #加速度为负3

145 a = -3

146 #初速度v0

147 v0 =v148 #当前速度v = v0 + at

149 v = v0 + a *t150 #移动距离x = v0t + 1/2 * a * t^2

151 move = v0 * t + 1 / 2 * a * t *t152 #当前位移

153 current +=move154 #加入轨迹

155 track.append(round(move))156 returntrack157

158 defmove_to_gap(self, slider, track):159 """

160 拖动滑块到缺口处161 :param slider: 滑块162 :param track: 轨迹163 :return:164 """

165 ActionChains(self.browser).click_and_hold(slider).perform()166 for x intrack:167 ActionChains(self.browser).move_by_offset(xoffset=x, yoffset=0).perform()168 time.sleep(0.5)169 ActionChains(self.browser).release().perform()170

171 deflogin(self):172 """

173 登录174 :return: None175 """

176 submit = self.wait.until(EC.element_to_be_clickable((By.CLASS_NAME, 'login-btn')))177 submit.click()178 time.sleep(10)179 print('登录成功')180

181 defcrack(self):182 #输入用户名密码

183 self.open()184 #点击验证按钮

185 button =self.get_geetest_button()186 button.click()187 #获取验证码图片

188 image1 = self.get_geetest_image('captcha1.png')189 #点按呼出缺口

190 slider =self.get_slider()191 slider.click()192 #获取带缺口的验证码图片

193 image2 = self.get_geetest_image('captcha2.png')194 #获取缺口位置

195 gap =self.get_gap(image1, image2)196 print('缺口位置', gap)197 #减去缺口位移

198 gap -=BORDER199 #获取移动轨迹

200 track =self.get_track(gap)201 print('滑动轨迹', track)202 #拖动滑块

203 self.move_to_gap(slider, track)204

205 success =self.wait.until(206 EC.text_to_be_present_in_element((By.CLASS_NAME, 'geetest_success_radar_tip_content'), '验证成功'))207 print(success)208

209 #失败后重试

210 if notsuccess:211 self.crack()212 else:213 self.login()214

215

216 if __name__ == '__main__':217 crack =CrackGeetest()218 crack.crack()

140>118>