首页 » Python » 正文

爬虫模拟登录破解无原图滑动验证码

模拟登录对象:博客园

验证码类型:无原图滑动验证码

使用工具与模块:python,selenium 

浏览器:Chrome

大体思路:以前的滑动验证码多为有原图的验证码,可以通过Image模块截取两张不同的图,通过对比像素得出移动的距离,无原图验证码也是基于这个原理,只是多了一步找出原图,该操作可以通过driver.execute_script()添加JS代码,改变display显示获得原图,然后就变成了有原图的滑动验证码的操作流程。

具体思路:

第一步:输入账号、密码,然后点击登陆

   from selenium import webdriver
   #为了方便演示与查看结果,在此使用有界面的Chrome浏览器,成功之后可以换成无界面浏览器
    driver=webdriver.Chrome()
    #参数为博客园登录页面
    driver.get('https://account.cnblogs.com/signin')
    #隐式等待3秒
    driver.implicitly_wait(3)
    #找到用户名标签和密码标签,用ID查找
    input_username=driver.find_element_by_id('LoginName')
    input_password=driver.find_element_by_id('Password')
    #输入用户名和密码
    input_username.send_keys('11111111111')
    input_password.send_keys('xxxxxxxxxx')
    #找到提交按钮
    submitBtn=driver.find_element_by_id('submitBtn')
    #点击提交
    submitBtn.click()

效果如图所示:

爬虫模拟登录破解无原图滑动验证码爬虫模拟登录破解无原图滑动验证码

第二步:弹出有缺口的图,并截取

爬虫模拟登录破解无原图滑动验证码

找到该标签,通过xpath查找找到位置,(通过classname查找,可能会报错,原因未知),这个位置不仅是缺口图的位置,还是原图的位置,所以获取原图和缺口图的方式是一样的

先写一个截图函数:

from PIL import Image
def get_snap(driver):
    #创建一个空的图片文件
    driver.save_screenshot('snap.png')
    snap_obj=Image.open('snap.png')
    return snap_obj
def get_image(driver):
    #通过xpath找到元素
    img_element = driver.find_element_by_xpath(
        '//div[@class="geetest_panel_next"]//canvas[@class="geetest_canvas_slice geetest_absolute"]')
    #获得图片的大小和位置
    size = img_element.size
    location = img_element.location
    left=location['x']
    top=location['y']
    right=left+size['width']
    bottom=top+size['height']
    snap_obj=get_snap(driver)
    #注意该参数是元组
    img_obj=snap_obj.crop((left,top,right,bottom))
    return img_obj

爬虫模拟登录破解无原图滑动验证码

 

通过获得的left,top,right,bottom进行截图

第三步:通过JS代码,显示原图

爬虫模拟登录破解无原图滑动验证码

 

 找到该便签,改变style中的display,其值为block时显示的是无缺口图:

爬虫模拟登录破解无原图滑动验证码

现在通过代码改变该标签的值:

driver.execute_script("var x=document.getElementsByClassName('geetest_canvas_fullbg geetest_fade geetest_absolute')[0];"
                          "x.style.display='block';"
                          "x.style.opacity=1"
                          )

测试时,有时候,opacity默认为0,需要变为1才会显示原图。

显示原图之后,因为位置是一样的,同第二步,使用同一个函数进行截图。

第四步:对比两张图片,即滑动的位移

none_img=get_image(driver)#缺口图
    driver.execute_script("var x=document.getElementsByClassName('geetest_canvas_fullbg geetest_fade geetest_absolute')[0];"
                          "x.style.display='block';"
                          "x.style.opacity=1"
                          )
block_img=get_image(driver)#原图

进行图片滑动的距离的计算:

def get_distance(img1,img2):
    start_x=60#初始X
    threhold=60#阈值
    for x in range(start_x,img1.size[0]):
        for y in range(img1.size[1]):
            rgb1=img1.load()[x,y]
            rgb2=img2.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#测试后-7可以提高成功率

关于初始值:

爬虫模拟登录破解无原图滑动验证码

滑动验证码,缺口一定和滑块有距离,所以滑块的所占的X的范围可以排除,测量得出滑块大小约为60像素(包含边距),所以start_x=60。

 

第五步:按照人的行为行为习惯,把总位移切成一段段小的位移

人的习惯为:先加速,再减速,可能有超出的现象。

为了保证更像人,本次有回退步骤

def get_tracks(distance):
    #distance为上一步得出的总距离。20是等会要回退的像素
    distance+=20
    #初速度为0,s是已经走的路程,t是时间
    v0=2
    s=0
    t=0.4
   #mid是进行减速的路程
    mid=distance*3/5
   #存放走的距离
    forward_tracks=[]
    while s<distance:
        if s<mid:
            a=2
        else:
            a=-3
        #高中物理,匀加速路程的计算
        v=v0
        tance=v*t+0.5*a*(t**2)
        tance=round(tance)
        s+=tance
        v0=v+a*t
        forward_tracks.append(tance)
    #因为回退20像素,所以可以手动打出,只要和为20即可
    back_tracks = [-1, -1, -1, -2, -2, -2, -3, -3, -2, -2, -1]  # 20
    return {"forward_tracks": forward_tracks, 'back_tracks': back_tracks}

 

第六步:按照距离移动

 

#获得滑块元素
geetest_slider_button=driver.find_element_by_class_name('geetest_slider_button')
    #获得距离
    distance=get_distance(block_img,none_img)
    #获得步数
    tracks_dic=get_tracks(distance)
   #点击并按住    ActionChains(driver).click_and_hold(geetest_slider_button).perform()
    forword_tracks=tracks_dic['forward_tracks']
    back_tracks=tracks_dic['back_tracks']
    for forword_track in forword_tracks:
        ActionChains(driver).move_by_offset(xoffset=forword_track,yoffset=0).perform()
    #停顿一会,更像人
    time.sleep(0.2)
    for back_tracks in back_tracks:
        ActionChains(driver).move_by_offset(xoffset=back_tracks, yoffset=0).perform()
    print(forword_tracks)
    ActionChains(driver).move_by_offset(xoffset=-3, yoffset=0).perform()
    ActionChains(driver).move_by_offset(xoffset=3, yoffset=0).perform()
    time.sleep(0.3)
    #松开鼠标
    ActionChains(driver).release().perform()

 

 完整代码:

from selenium import webdriver
from selenium.webdriver import ActionChains
from selenium.webdriver.common.keys import Keys
from PIL import Image
import time
driver=webdriver.Chrome()

def get_snap(driver):
    driver.save_screenshot('snap.png')
    snap_obj=Image.open('snap.png')
    return snap_obj
def get_image(driver):
    img_element = driver.find_element_by_xpath(
        '//div[@class="geetest_panel_next"]//canvas[@class="geetest_canvas_slice geetest_absolute"]')
    size = img_element.size
    location = img_element.location
    left=location['x']
    top=location['y']
    right=left+size['width']
    bottom=top+size['height']
    snap_obj=get_snap(driver)
    img_obj=snap_obj.crop((left,top,right,bottom))
    return img_obj
# try:
#     driver.get('https://www.baidu.com')
#     driver.implicitly_wait(5)
#     r1=driver.find_element_by_link_text('登录').click()
#     driver.find_element_by_id('TANGRAM__PSP_10__footerULoginBtn').click()
#     input_username=driver.find_element_by_id('TANGRAM__PSP_10__userName')
#     input_username.send_keys('17396876501')
#     input_password=driver.find_element_by_id('TANGRAM__PSP_10__password')
#     input_password.send_keys('dfcver')
#     driver.find_element_by_id('TANGRAM__PSP_10__submit').click()
#     time.sleep(5)
# finally:
#     driver.close()
def get_distance(img1,img2):
    start_x=60
    threhold=60#阈值
    for x in range(start_x,img1.size[0]):
        for y in range(img1.size[1]):
            rgb1=img1.load()[x,y]
            rgb2=img2.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
def get_tracks(distance):
    distance+=20
    v0=2
    s=0
    t=0.4
    mid=distance*3/5
    forward_tracks=[]
    while s<distance:
        if s<mid:
            a=2
        else:
            a=-3
        v=v0
        tance=v*t+0.5*a*(t**2)
        tance=round(tance)
        s+=tance
        v0=v+a*t
        forward_tracks.append(tance)
    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.get('https://account.cnblogs.com/signin')
    driver.implicitly_wait(3)
    input_username=driver.find_element_by_id('LoginName')
    input_password=driver.find_element_by_id('Password')
    input_username.send_keys('928480709')
    input_password.send_keys('dfcver1112223334')
    submitBtn=driver.find_element_by_id('submitBtn')
    submitBtn.click()
    time.sleep(2)#等待验证码加载
    none_img=get_image(driver)
    driver.execute_script("var x=document.getElementsByClassName('geetest_canvas_fullbg geetest_fade geetest_absolute')[0];"
                          "x.style.display='block';"
                          "x.style.opacity=1"
                          )
    block_img=get_image(driver)
    geetest_slider_button=driver.find_element_by_class_name('geetest_slider_button')

    distance=get_distance(block_img,none_img)
    tracks_dic=get_tracks(distance)
    ActionChains(driver).click_and_hold(geetest_slider_button).perform()
    forword_tracks=tracks_dic['forward_tracks']
    back_tracks=tracks_dic['back_tracks']
    for forword_track in forword_tracks:
        ActionChains(driver).move_by_offset(xoffset=forword_track,yoffset=0).perform()
    time.sleep(0.2)
    for back_tracks in back_tracks:
        ActionChains(driver).move_by_offset(xoffset=back_tracks, yoffset=0).perform()
    print(forword_tracks)
    ActionChains(driver).move_by_offset(xoffset=-3, yoffset=0).perform()
    ActionChains(driver).move_by_offset(xoffset=3, yoffset=0).perform()
    time.sleep(0.3)
    ActionChains(driver).release().perform()

    time.sleep(60)
finally:
    driver.close()

完整代码

本文来源于互联网:爬虫模拟登录破解无原图滑动验证码

发表评论