利用selenium和ocr破解点击验证码模拟登录52pojie并自动签到
本帖最后由 frankyxu 于 2019-2-22 14:41 编辑# 利用selenium和ocr破解文字点击验证码
### hello，下午好，这次给大家带来的是selenium和ocr破解文字点击验证码，希望大家会喜欢，项目采用的是selenium模拟人进行点击和滑动，通过ocr计算出要点击文字的位置，然后利用selenium完成最后的剪辑，主要难点是计算距离,[项目地址](https://github.com/guapier/52pojie)
## 主要用到的技术
- selenium
---
## 项目截图
!(http://cdn.guapier.net/QQ20190222-125335-HD.gifhttps://github.com/guapier/52pojie/blob/master/QQ20190222-125335-HD.gif)
---
## 核心源码 :beers:
```python
# -*- coding: utf-8 -*-
# @Time : 2018/01/23 10:02
# @AuThor: xuzy
# @Software: PyCharm
# @Function:
from selenium import webdriver
from selenium.webdriver import ActionChains
from selenium.webdriver.support.wait import WebDriverWait
import time
import base64
import apiutil
import random
class CrackSlider():
"""
通过浏览器截图，识别验证码中缺口位置，获取需要滑动距离，并模仿人类行为破解滑动验证码
"""
def __init__(self):
super(CrackSlider, self).__init__()
# 实际地址
self.url = 'https://www.52pojie.cn/member.php?mod=logging&action=login'
self.driver = webdriver.Chrome()
self.wait = WebDriverWait(self.driver, 30)
self.zoom = 1
def open(self):
self.driver.get(self.url)
self.driver.maximize_window()
def recognize_image(self):
#https://ai.qq.com/product/ocr.shtml#identify优图ocr地址，免费申请
app_id = 'appid'
app_key = 'appkey'
with open('./captcha.jpg', 'rb') as bin_data:
image_data = bin_data.read()
ai_obj = apiutil.AiPlat(app_id, app_key)
print('----------------------SEND REQ----------------------')
rsp = ai_obj.getOcrGeneralocr(image_data)
print(rsp)
return rsp
def cal_location(self, word, rsp):
try:
if rsp['ret'] == 0:
for i in rsp['data']['item_list']:
result = i['itemstring'].find(word)
if result == -1:
continue
else:
x = i['itemcoord']['x']
y = i['itemcoord']['y']
width = i['itemcoord']['width']
height = i['itemcoord']['height']
location_x = int(x + (width / 4) * (result + 0.5))
location_y = int(y + height / 2)
return int(location_x * 23 / 20), int(location_y * 23 / 20)
return 50, 50
except Exception as e:
print(e)
return 100, 100
def get_tracks(self, distance):
v = 0
t = random.uniform(0.2, 0.3)
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))
return {'forward_tracks': forward_tracks}
def crack_slider(self):
try:
time.sleep(3)
self.open()
captcha = 'captcha.jpg'
time.sleep(1)
self.driver.find_element_by_xpath('//input[@name="username"]').send_keys('你的用户名')
time.sleep(1)
self.driver.find_element_by_xpath('//input[@name="password"]').send_keys('你的密码')
distance = 260
tracks = self.get_tracks((distance + 2) * self.zoom)# 对位移的缩放计算
time.sleep(2)
# slider = self.wait.until(EC.element_to_be_clickable((By.CLASS_NAME, 'nc_iconfont btn_slide')))
slider = self.driver.find_element_by_xpath("//*[@id='nc_1_n1z']")
time.sleep(1)
ActionChains(self.driver).click_and_hold(slider).perform()
for track in tracks['forward_tracks']:
ActionChains(self.driver).move_by_offset(xoffset=track, yoffset=0).perform()
time.sleep(0.5)
while True:
print('进入while循环')
print('*' * 50)
# print(location_x,location_y)
print('*' * 50)
time.sleep(0.5)
# self.driver.find_element_by_xpath('//i[@id="nc_1__btn_2"]').click()
captcha_image = self.driver.find_element_by_xpath('//div[@class="clickCaptcha_img"]/img')
captcha = captcha_image.get_attribute('src')
print(captcha)
# 下载图片
fh = open("captcha.jpg", "wb")
fh.write(base64.b64decode(captcha.split(',')))
fh.close()
# 获取要识别的文字
word_ele = self.driver.find_element_by_xpath('//div[@id="nc_1__scale_text"]/i')
word = word_ele.text.replace('”', '').replace('“', '').strip()
print('请点击图中的{}字'.format(word))
# 识别图中的文字
rsp = self.recognize_image()
# 计算要点击的文字在图中的位置
location_x, location_y = self.cal_location(word, rsp)
print(location_x, location_y)
ActionChains(self.driver).move_to_element_with_offset(captcha_image, location_x,
location_y).click().perform()
time.sleep(1)
if '验证通过' in self.driver.page_source:
break
# try:
# menu = self.driver.find_element_by_xpath('//a[@class="showmenu"]')
# except Exception as e:
# menu = None
# if menu:
# break
# 刷新页面
self.driver.find_element_by_xpath('//button[@name="loginsubmit"]').click()
time.sleep(2)
print('已经跳出while循环')
self.driver.get('https://www.52pojie.cn/home.php?mod=task&do=apply&id=2')
# qd = WebDriverWait(self.driver, 10).until(EC.visibility_of(self.find_element(by=By.CLASS_NAME, value='qq_bind')))
#
# qd.click()
time.sleep(5)
self.driver.quit()
except Exception as e:
print(e)
exit(0)
c.crack_slider()
if __name__ == '__main__':
c = CrackSlider()
c.crack_slider()
```
---
## 如何使用
- pip install selenium
- python3 52.py
- 顺便来点热心值呗 tanghengvip 发表于 2019-2-22 14:42
https://www.freebuf.com/articles/web/194628.html
今天正好看到这篇文章~ 楼主的成品就来了。感谢分享。
哈哈哈，你说的这篇文章我也看过，还有一种思路是利用opencv的模板匹配方法来计算距离，你如果有兴趣的话可以自己尝试一下 时空之外 发表于 2019-2-22 16:16
这个优图识别好像不太准确，我换了个百度，依旧不行。还有就是建议self.driver.get(url)和self.maximize_wi ...
这个是基于优图的，返回的数据中有坐标信息，百度的ocr不是太适合，优图的ocr一般三次以内吧 谢谢！！！ 前排合影，优秀作品~ 左三圈 发表于 2019-2-22 13:41
前排合影，优秀作品~
{:1_918:}{:1_918:}{:1_918:},感谢，还有些地方不够完善 https://www.freebuf.com/articles/web/194628.html
今天正好看到这篇文章~ 楼主的成品就来了。感谢分享。 不错 学习学习 本帖最后由 时空之外 于 2019-2-22 16:18 编辑
这个优图识别好像不太准确，我换了个百度，依旧不行。还有就是建议self.driver.get(url)和self.maximize_window互换位置，不然每次加载会被提示挡住拖动按钮 frankyxu 发表于 2019-2-22 16:21
这个是基于优图的，返回的数据中有坐标信息，百度的ocr不是太适合，优图的ocr一般三次以内吧
楼主知道selenium怎么过检测嘛{:1_918:}