與網上的其餘滑塊驗證碼不一樣,騰訊的驗證碼能夠直接經過url請求獲得,只須要對url進一步分析,提取出驗證碼原圖的地址,並將圖片下載便可。web
但據我觀察,該url彷佛是有兩種不一樣類型的地址格式,須要具體分析。這裏,選擇其中一種進行實驗,源碼在文章末尾。chrome
def get_img(self): """ 獲取驗證碼陰影圖和原圖 :return: """ self.driver.switch_to.frame('tcaptcha_iframe') time.sleep(3) # 獲取有陰影的圖片 src = self.driver.find_element_by_id('slideBg').get_attribute('src') # 分析圖片地址,發現原圖地址能夠經過陰影圖地址改動獲取 src_bg = re.sub('_1_', '_0_', src) urlretrieve(src, 'img1.png') urlretrieve(src_bg, 'img2.png')
有一個須要注意的問題,下載到本地的圖片對原圖進行了放大,因此要對圖片尺寸進行調整還原,保證後面計算出的偏移值的正確性瀏覽器
def resize_img(self, img): """ 下載的圖片把網頁中的圖片進行了放大,因此將圖片還原成原尺寸 :param img: 圖片 :return: 返回還原後的圖片 """ a = 2.428 # 經過本地圖片與原網頁圖片的比較,計算出的縮放比例 (x, y) = img.size x_resize = int(x // a) y_resize = int(y // a) img = img.resize((x_resize, y_resize), Image.ANTIALIAS)
此時,已經拿到原圖和陰影圖,只須要進行像素比較便可,網上有其餘相關教程,可參考dom
def is_pixel_equal(self, img1, img2, x, y): """ 比較兩張圖片同一點上的像數值,差距大於設置標準返回False :param img1: 陰影圖 :param img2: 原圖 :param x: 橫座標 :param y: 縱座標 :return: 是否相等 """ pixel1, pixel2 = img1.load()[x, y], img2.load()[x, y] sub_index = 100 if abs(pixel1[0] - pixel2[0]) < sub_index and abs(pixel1[1] - pixel2[1]) < sub_index and abs( pixel1[2] - pixel2[2]) < sub_index: return True else: return False def get_gap_offset(self, img1, img2): ''' 獲取缺口的偏移量 ''' offset = None distance = 70 for i in range(distance, img1.size[0]): for j in range(img1.size[1]): # 兩張圖片對比,(i,j)像素點的RGB差距,過大則該x爲偏移值 if not self.is_pixel_equal(img1, img2, i, j): offset = i return offset return offset
最後,經過selenuim的動做鏈,模擬滑塊拖動ide
def operate_slider(self, track): """ 拖動滑塊 :param track: 運動軌跡 :return: """ # 定位到拖動按鈕 slider_bt = self.driver.find_element_by_xpath('//div[@class="tc-drag-thumb"]') # 點擊拖動按鈕不放 ActionChains(self.driver).click_and_hold(slider_bt).perform() # 按正向軌跡移動 for i in track: ActionChains(self.driver).move_by_offset(xoffset=i, yoffset=0).perform() time.sleep(random.random() / 100) # 每移動一次隨機停頓0-1/100秒之間騙過了極驗,經過率很高 time.sleep(random.random()) # 按逆向軌跡移動 back_tracks = [-1, -0.5, -1] for i in back_tracks: time.sleep(random.random() / 100) ActionChains(self.driver).move_by_offset(xoffset=i, yoffset=0).perform() # 模擬人手抖動 self.shake_mouse() time.sleep(random.random()) # 鬆開滑塊按鈕 ActionChains(self.driver).release().perform()
下面附上完整的代碼,僅供參考,不足之處,歡迎你們指正ui
import time import re import random from selenium import webdriver from urllib.request import urlretrieve from PIL import Image from selenium.webdriver.common.action_chains import ActionChains class Tencent(): def __init__(self): """ 初始化屬性,傳入url地址,驅動路徑,瀏覽器窗口最大化,僞造ua """ self.url = 'https://qzone.qq.com/' driver_path = r'C:\Users\xiaodengtang\AppData\Local\Google\Chrome\Application\chromedriver.exe' self.driver = webdriver.Chrome(executable_path=driver_path) self.driver.maximize_window() self.headers = { 'user-agent': 'Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/79.0.3945.130 Safari/537.36'} def input_username_password(self, account, password): """ 打開瀏覽器,傳入帳號、密碼,定位到登陸窗口,切換登錄方式 :param account: :param password: :return: """ self.driver.get(self.url) time.sleep(1) self.driver.switch_to.frame('login_frame') self.driver.find_element_by_id('switcher_plogin').click() time.sleep(1) self.driver.find_element_by_id('u').send_keys(account) time.sleep(0.5) self.driver.find_element_by_id('p').send_keys(password) time.sleep(0.5) self.driver.find_element_by_class_name('login_button').click() def get_img(self): """ 獲取驗證碼陰影圖和原圖 :return: """ self.driver.switch_to.frame('tcaptcha_iframe') time.sleep(3) # 獲取有陰影的圖片 src = self.driver.find_element_by_id('slideBg').get_attribute('src') # 分析圖片地址,發現原圖地址能夠經過陰影圖地址改動獲取 src_bg = re.sub('_1_', '_0_', src) urlretrieve(src, 'img1.png') urlretrieve(src_bg, 'img2.png') captcha1 = Image.open('img1.png') captcha2 = Image.open('img2.png') return captcha1, captcha2 def resize_img(self, img): """ 下載的圖片把網頁中的圖片進行了放大,因此將圖片還原成原尺寸 :param img: 圖片 :return: 返回還原後的圖片 """ a = 2.428 # 經過本地圖片與原網頁圖片的比較,計算出的縮放比例 (x, y) = img.size x_resize = int(x // a) y_resize = int(y // a) img = img.resize((x_resize, y_resize), Image.ANTIALIAS) return img def is_pixel_equal(self, img1, img2, x, y): """ 比較兩張圖片同一點上的像數值,差距大於設置標準返回False :param img1: 陰影圖 :param img2: 原圖 :param x: 橫座標 :param y: 縱座標 :return: 是否相等 """ pixel1, pixel2 = img1.load()[x, y], img2.load()[x, y] sub_index = 100 if abs(pixel1[0] - pixel2[0]) < sub_index and abs(pixel1[1] - pixel2[1]) < sub_index and abs( pixel1[2] - pixel2[2]) < sub_index: return True else: return False def get_gap_offset(self, img1, img2): ''' 獲取缺口的偏移量 ''' offset = None distance = 70 for i in range(distance, img1.size[0]): for j in range(img1.size[1]): # 兩張圖片對比,(i,j)像素點的RGB差距,過大則該x爲偏移值 if not self.is_pixel_equal(img1, img2, i, j): offset = i return offset return offset def get_track(self, offset): ''' 計算滑塊的移動軌跡 ''' offset -= 30 # 滑塊並非從0開始移動,有一個初始值 a = offset / 4 track = [a, a, a, a] return track def shake_mouse(self): """ 模擬人手釋放鼠標抖動 :return: None """ ActionChains(self.driver).move_by_offset(xoffset=-2, yoffset=0).perform() ActionChains(self.driver).move_by_offset(xoffset=2, yoffset=0).perform() def operate_slider(self, track): """ 拖動滑塊 :param track: 運動軌跡 :return: """ # 定位到拖動按鈕 slider_bt = self.driver.find_element_by_xpath('//div[@class="tc-drag-thumb"]') # 點擊拖動按鈕不放 ActionChains(self.driver).click_and_hold(slider_bt).perform() # 按正向軌跡移動 for i in track: ActionChains(self.driver).move_by_offset(xoffset=i, yoffset=0).perform() time.sleep(random.random() / 100) # 每移動一次隨機停頓0-1/100秒之間騙過了極驗,經過率很高 time.sleep(random.random()) # 按逆向軌跡移動 back_tracks = [-1, -0.5, -1] for i in back_tracks: time.sleep(random.random() / 100) ActionChains(self.driver).move_by_offset(xoffset=i, yoffset=0).perform() # 模擬人手抖動 self.shake_mouse() time.sleep(random.random()) # 鬆開滑塊按鈕 ActionChains(self.driver).release().perform() def login(self, account, password): ''' 實現主要的登錄邏輯 ''' self.input_username_password(account, password) time.sleep(2) a, b = self.get_img() a = self.resize_img(a) b = self.resize_img(b) distance = self.get_gap_offset(a, b) track = self.get_track(distance) self.operate_slider(track) if __name__ == '__main__': qq = Tencent() account = '123548658' password = 'yanzhengma' qq.login(account, password)