验证码
#破解博客园后台登陆 from selenium import webdriver from selenium.webdriver import ActionChains from selenium.webdriver.common.by import By from selenium.webdriver.common.keys import Keys from selenium.webdriver.support import expected_conditions as EC from selenium.webdriver.support.wait import WebDriverWait from PIL import Image import time def get_snap(): driver.save_screenshot('full_snap.png') page_snap_obj=Image.open('full_snap.png') return page_snap_obj def get_image(): img=driver.find_element_by_class_name('geetest_canvas_img') time.sleep(2) location=img.location size=img.size left=location['x'] top=location['y'] right=left+size['width'] bottom=top+size['height'] page_snap_obj=get_snap() image_obj=page_snap_obj.crop((left,top,right,bottom)) # image_obj.show() return image_obj def get_distance(image1,image2): start=57 threhold=60 for i in range(start,image1.size[0]): for j in range(image1.size[1]): rgb1=image1.load()[i,j] rgb2=image2.load()[i,j] res1=abs(rgb1[0]-rgb2[0]) res2=abs(rgb1[1]-rgb2[1]) res3=abs(rgb1[2]-rgb2[2]) # print(res1,res2,res3) if not (res1 < threhold and res2 < threhold and res3 < threhold): return i-7 return i-7 def get_tracks(distance): distance+=20 #先滑过一点，最后再反着滑动回来 v=0 t=0.2 forward_tracks=[] current=0 mid=distance*3/5 while current < distance: if current < mid: a=2 else: a=-3 s=v*t+0.5*a*(t**2) v=v+a*t current+=s forward_tracks.append(round(s)) #反着滑动到准确位置 back_tracks=[-3,-3,-2,-2,-2,-2,-2,-1,-1,-1] #总共等于-20 return {'forward_tracks':forward_tracks,'back_tracks':back_tracks} try: # 一、输入帐号密码回车 driver = webdriver.Chrome() driver.implicitly_wait(3) driver.get('https://passport.cnblogs.com/user/signin') username = driver.find_element_by_id('input1') pwd = driver.find_element_by_id('input2') signin = driver.find_element_by_id('signin') username.send_keys('******') #用户名 pwd.send_keys('******') #密码 signin.click() # 二、点击按钮，获得没有缺口的图片 button = driver.find_element_by_class_name('geetest_radar_tip') button.click() # 三、获取没有缺口的图片 image1 = get_image() # 四、点击滑动按钮，获得有缺口的图片 button = driver.find_element_by_class_name('geetest_slider_button') button.click() # 五、获取有缺口的图片 image2 = get_image() # 六、对比两种图片的像素点，找出位移 distance = get_distance(image1, image2) # 七、模拟人的行为习惯，根据总位移获得行为轨迹 tracks = get_tracks(distance) print(tracks) # 八、按照行动轨迹先正向滑动，后反滑动 button = driver.find_element_by_class_name('geetest_slider_button') ActionChains(driver).click_and_hold(button).perform() # 正常人类老是自信满满地开始正向滑动，自信地表现是疯狂加速 for track in tracks['forward_tracks']: ActionChains(driver).move_by_offset(xoffset=track, yoffset=0).perform() # 结果傻逼了，正常的人类停顿了一下，回过神来发现，卧槽，滑过了,而后开始反向滑动 time.sleep(0.5) for back_track in tracks['back_tracks']: ActionChains(driver).move_by_offset(xoffset=back_track, yoffset=0).perform() # 小范围震荡一下，进一步迷惑极验后台，这一步能够极大地提升成功率 ActionChains(driver).move_by_offset(xoffset=-3, yoffset=0).perform() ActionChains(driver).move_by_offset(xoffset=3, yoffset=0).perform() # 成功后，骚包人类总喜欢默默地欣赏一下本身拼图的成果，而后依依不舍地松开那只脏手 time.sleep(0.5) ActionChains(driver).release().perform() time.sleep(10) # 睡时间长一点，肯定登陆成功 finally: driver.close()
#总结：测试了几下，感受就是验证码的位置是随机变化的，可是此码给出的信息倒是不变的，so，pass
再来：参考
###############优化后的代码（将功能封装成函数调用）####### from selenium import webdriver from selenium.webdriver import ActionChains from selenium.webdriver.common.by import By #按照什么方式查找，By.ID,By.CSS_SELECTOR from selenium.webdriver.common.keys import Keys #键盘按键操做 from selenium.webdriver.support import expected_conditions as EC from selenium.webdriver.support.wait import WebDriverWait #等待页面加载某些元素 from PIL import Image #pip3 install pillow import time def get_snap(driver): driver.save_screenshot('snap.png')#截图 snap_obj=Image.open('snap.png')#保存 return snap_obj def get_image(driver): img=driver.find_element_by_class_name('geetest_canvas_img') time.sleep(2) #等待图片加载完毕 size=img.size location=img.location #获取图片位置 left=location['x'] top=location['y'] right=left+size['width'] bottom=top+size['height'] snap_obj=get_snap(driver) image_obj=snap_obj.crop((left,top,right,bottom))#截图操做 # image_obj.show() return image_obj def get_distance(image1,image2): start_x=58#滑块最左侧 threhold=60#去除伪影响 # print(image1.size) # print(image2.size) for x in range(start_x,image1.size[0]): for y in range(image1.size[1]): rgb1=image1.load()[x,y] rgb2=image2.load()[x,y] res1=abs(rgb1[0]-rgb2[0]) res2=abs(rgb1[1]-rgb2[1]) res3=abs(rgb1[2]-rgb2[2]) if not (res1 < threhold and res2 < threhold and res3 < threhold): return x-7#偏差范围 def get_tracks(distance): distance+=20#故意划过头20像素 #v=v0+a*t #s=v*t+0.5*a*(t**2) v0=0 s=0 t=0.2 mid=distance*3/5 forward_tracks=[] while s < distance: if s < mid: a=2 else: a=-3 v=v0 track=v*t+0.5*a*(t**2) track=round(track)#取整数 v0=v+a*t s+=track forward_tracks.append(track) back_tracks=[-1,-1,-1,-2,-2,-2,-3,-3,-2,-2,-1] #20 return {"forward_tracks":forward_tracks,'back_tracks':back_tracks} def crack(driver):#封装滑动的函数 # 二、点击验证人机按钮，弹出没有缺口的图 button = driver.find_element_by_class_name('geetest_radar_tip_content') button.click() # 三、针对没有缺口的图片进行截图 image1 = get_image(driver) # 四、点击滑动按钮，弹出有缺口的图 slider_button = driver.find_element_by_class_name('geetest_slider_button') slider_button.click() # 五、针对有缺口的图片进行截图 image2 = get_image(driver) # 六、对比两张图片，找出缺口，即滑动的位移 distance = get_distance(image1, image2) # print(distance) # 七、按照人的行为行为习惯，把总位移切成一段段小的位移 traks_dic = get_tracks(distance) # 八、按照位移移动 slider_button = driver.find_element_by_class_name('geetest_slider_button') ActionChains(driver).click_and_hold(slider_button).perform() # 按住不放手 # 先向前移动 forward_tracks = traks_dic["forward_tracks"] back_tracks = traks_dic["back_tracks"] for forward_track in forward_tracks: ActionChains(driver).move_by_offset(xoffset=forward_track, yoffset=0).perform() # 短暂停顿，发现傻逼，移过了 time.sleep(0.2) # 先向后移动 for back_track in back_tracks: ActionChains(driver).move_by_offset(xoffset=back_track, yoffset=0).perform() # 抖一抖 ActionChains(driver).move_by_offset(xoffset=-4, yoffset=0).perform() ActionChains(driver).move_by_offset(xoffset=3, yoffset=0).perform() time.sleep(0.1) ActionChains(driver).move_by_offset(xoffset=-2, yoffset=0).perform() ActionChains(driver).move_by_offset(xoffset=3, yoffset=0).perform() time.sleep(0.3) ActionChains(driver).release().perform() # 松开鼠标 def login_cnblogs(username,pwd): driver = webdriver.Chrome() # 谷歌浏览器driver = webdriver.Chrome()#谷歌浏览器 try: driver.get('https://passport.cnblogs.com/user/signin')#博客园 driver.implicitly_wait(10)#隐形等待10秒 #一、输入帐号、密码，而后点击登录 input_user=driver.find_element_by_id('input1') input_pwd=driver.find_element_by_id('input2') login_button=driver.find_element_by_id('signin') input_user.send_keys(username)#输入帐号 input_pwd.send_keys(pwd)#输入密码 login_button.click()#点击登陆按钮 # 调用 封装滑动的函数 crack(driver) time.sleep(10) finally: driver.close() if __name__ == '__main__': login_cnblogs(username='脚本小孩',pwd='*****')
仍是出错，缘由在第4个def
“
def get_tracks(distance): distance+=20#故意划过头20像素
” 待解惑
继续啊
#首先要安装Pillow pip3 install pillow #Pillow:基于PIL，处理python 3.x的图形图像库.由于PIL只能处理到python 2.x，而这个模块能处理Python3.x，目前用它作图形的不少. # 破解滑动验证码自动登陆博客园 ###########思路整理########## from selenium import webdriver from selenium.webdriver import ActionChains from selenium.webdriver.common.by import By #按照什么方式查找，By.ID,By.CSS_SELECTOR from selenium.webdriver.common.keys import Keys #键盘按键操做 from selenium.webdriver.support import expected_conditions as EC from selenium.webdriver.support.wait import WebDriverWait #等待页面加载某些元素 from PIL import Image #pip3 install pillow import time def get_snap(driver): driver.save_screenshot('snap.png')#截图 snap_obj=Image.open('snap.png')#保存 return snap_obj def get_image(driver): img=driver.find_element_by_class_name('geetest_canvas_img') time.sleep(2) #等待图片加载完毕 size=img.size location=img.location #获取图片位置 left=location['x'] top=location['y'] right=left+size['width'] bottom=top+size['height'] snap_obj=get_snap(driver) image_obj=snap_obj.crop((left,top,right,bottom))#截图操做 # image_obj.show() return image_obj def get_distance(image1,image2): start_x=58#滑块最左侧 threhold=60#去除伪影响 # print(image1.size) # print(image2.size) for x in range(start_x,image1.size[0]): for y in range(image1.size[1]): rgb1=image1.load()[x,y] rgb2=image2.load()[x,y] res1=abs(rgb1[0]-rgb2[0]) res2=abs(rgb1[1]-rgb2[1]) res3=abs(rgb1[2]-rgb2[2]) if not (res1 < threhold and res2 < threhold and res3 < threhold): return x-7#偏差范围 def get_tracks(distance): distance+=20#故意划过头20像素 #v=v0+a*t #s=v*t+0.5*a*(t**2) v0=0 s=0 t=0.2 mid=distance*3/5 forward_tracks=[] while s < distance: if s < mid: a=2 else: a=-3 v=v0 track=v*t+0.5*a*(t**2) track=round(track)#取整数 v0=v+a*t s+=track forward_tracks.append(track) back_tracks=[-1,-1,-1,-2,-2,-2,-3,-3,-2,-2,-1] #20 return {"forward_tracks":forward_tracks,'back_tracks':back_tracks} try: driver = webdriver.Chrome()#谷歌浏览器 driver.get('https://passport.cnblogs.com/user/signin')#博客园 driver.implicitly_wait(10)#隐形等待10秒 #一、输入帐号、密码，而后点击登录 input_user=driver.find_element_by_id('input1') input_pwd=driver.find_element_by_id('input2') login_button=driver.find_element_by_id('signin') input_user.send_keys('脚本小孩')#输入帐号 input_pwd.send_keys('********')#输入密码 login_button.click()#点击登陆按钮 #二、点击验证人机按钮，弹出没有缺口的图 button=driver.find_element_by_class_name('geetest_radar_tip_content') button.click() #三、针对没有缺口的图片进行截图 image1=get_image(driver) #四、点击滑动按钮，弹出有缺口的图 slider_button=driver.find_element_by_class_name('geetest_slider_button') slider_button.click() #五、针对有缺口的图片进行截图 image2=get_image(driver) #六、对比两张图片，找出缺口，即滑动的位移 distance=get_distance(image1,image2) # print(distance) #七、按照人的行为行为习惯，把总位移切成一段段小的位移 traks_dic=get_tracks(distance) #八、按照位移移动 slider_button=driver.find_element_by_class_name('geetest_slider_button') ActionChains(driver).click_and_hold(slider_button).perform()#按住不放手 #先向前移动 forward_tracks=traks_dic["forward_tracks"] back_tracks=traks_dic["back_tracks"] for forward_track in forward_tracks: ActionChains(driver).move_by_offset(xoffset=forward_track,yoffset=0).perform() #短暂停顿，发现傻逼，移过了 time.sleep(0.2) # 先向后移动 for back_track in back_tracks: ActionChains(driver).move_by_offset(xoffset=back_track,yoffset=0).perform() # 抖一抖 ActionChains(driver).move_by_offset(xoffset=-4,yoffset=0).perform() ActionChains(driver).move_by_offset(xoffset=3,yoffset=0).perform() time.sleep(0.1) ActionChains(driver).move_by_offset(xoffset=-2,yoffset=0).perform() ActionChains(driver).move_by_offset(xoffset=3,yoffset=0).perform() time.sleep(0.3) ActionChains(driver).release().perform()#松开鼠标 time.sleep(10) finally: driver.close()
补充：html
from selenium import webdriver from selenium.webdriver import ActionChains from selenium.webdriver.common.by import By from selenium.webdriver.common.keys import Keys from selenium.webdriver.support import expected_conditions as EC from selenium.webdriver.support.wait import WebDriverWait from PIL import Image import time def get_snap(driver): driver.save_screenshot('full_snap.png') page_snap_obj=Image.open('full_snap.png') return page_snap_obj def get_image(driver): img=driver.find_element_by_class_name('geetest_canvas_img') time.sleep(2) location=img.location size=img.size left=location['x'] top=location['y'] right=left+size['width'] bottom=top+size['height'] page_snap_obj=get_snap(driver) image_obj=page_snap_obj.crop((left,top,right,bottom)) # image_obj.show() return image_obj def get_distance(image1,image2): start=57 threhold=60 for i in range(start,image1.size[0]): for j in range(image1.size[1]): rgb1=image1.load()[i,j] rgb2=image2.load()[i,j] res1=abs(rgb1[0]-rgb2[0]) res2=abs(rgb1[1]-rgb2[1]) res3=abs(rgb1[2]-rgb2[2]) # print(res1,res2,res3) if not (res1 < threhold and res2 < threhold and res3 < threhold): return i-7 return i-7 def get_tracks(distance): distance+=20 #先滑过一点，最后再反着滑动回来 v=0 t=0.2 forward_tracks=[] current=0 mid=distance*3/5 while current < distance: if current < mid: a=2 else: a=-3 s=v*t+0.5*a*(t**2) v=v+a*t current+=s forward_tracks.append(round(s)) #反着滑动到准确位置 back_tracks=[-3,-3,-2,-2,-2,-2,-2,-1,-1,-1] #总共等于-20 return {'forward_tracks':forward_tracks,'back_tracks':back_tracks} def crack(driver): #破解滑动认证 # 一、点击按钮，获得没有缺口的图片 button = driver.find_element_by_class_name('geetest_radar_tip') button.click() # 二、获取没有缺口的图片 image1 = get_image(driver) # 三、点击滑动按钮，获得有缺口的图片 button = driver.find_element_by_class_name('geetest_slider_button') button.click() # 四、获取有缺口的图片 image2 = get_image(driver) # 五、对比两种图片的像素点，找出位移 distance = get_distance(image1, image2) # 六、模拟人的行为习惯，根据总位移获得行为轨迹 tracks = get_tracks(distance) print(tracks) # 七、按照行动轨迹先正向滑动，后反滑动 button = driver.find_element_by_class_name('geetest_slider_button') ActionChains(driver).click_and_hold(button).perform() # 正常人类老是自信满满地开始正向滑动，自信地表现是疯狂加速 for track in tracks['forward_tracks']: ActionChains(driver).move_by_offset(xoffset=track, yoffset=0).perform() # 结果傻逼了，正常的人类停顿了一下，回过神来发现，卧槽，滑过了,而后开始反向滑动 time.sleep(0.5) for back_track in tracks['back_tracks']: ActionChains(driver).move_by_offset(xoffset=back_track, yoffset=0).perform() # 小范围震荡一下，进一步迷惑极验后台，这一步能够极大地提升成功率 ActionChains(driver).move_by_offset(xoffset=-3, yoffset=0).perform() ActionChains(driver).move_by_offset(xoffset=3, yoffset=0).perform() # 成功后，骚包人类总喜欢默默地欣赏一下本身拼图的成果，而后依依不舍地松开那只脏手 time.sleep(0.5) ActionChains(driver).release().perform() def login_cnblogs(username,password): driver = webdriver.Chrome() try: # 一、输入帐号密码回车 driver.implicitly_wait(3) driver.get('https://passport.cnblogs.com/user/signin') input_username = driver.find_element_by_id('input1') input_pwd = driver.find_element_by_id('input2') signin = driver.find_element_by_id('signin') input_username.send_keys(username) input_pwd.send_keys(password) signin.click() # 二、破解滑动认证 crack(driver) time.sleep(10) # 睡时间长一点，肯定登陆成功 finally: driver.close() if __name__ == '__main__': login_cnblogs(username='linhaifeng',password='xxxx') 修订版
参考网址：https://www.cnblogs.com/linhaifeng/articles/7802150.html#toppython
有一句话感触颇深，由于我也一直是这样认为的，好比说，，，数学，我一直都认为数学只是一种学习的工具
我学习数学不少时候只是想去作某件事，或者想去了解这件事的原理，想把他弄通透罢了程序员
那么网上的学习软件，或者编程也好，本质不变，可是当程序员的性质变了，目的不变，web
引用：编程
也就不修改说明说明了
ps:破解图片验证码的核心在于模拟人的行为， 自笔者在老男孩授课以来，上述的破解思路已经分享给不少人， 相应地网络上也已经有不少copy版， 极验后台的也在不断学习用户的破解行为， 但归根结底只要咱们将破解行为模拟地足够像人，极验就拿咱们没有办法，
上面引用的话也在说，核心在于模拟人的行为，那么计算机的本质是什么呢？不就是解法生产力。。。。解放思想嘛canvas
那么爬虫的本质是什么????? 浏览器
我以为如今我自学这些东西的目的在于什么？？网络
或者说我想要达到什么高度。。。。。app
若是不知道，我想能够先放下了，想搞搞有目的有性质的东西——好比数学ide