爬虫验证码识别-01--古诗词登录

Stubborn

看了代码，你也许会的好傻，哈哈，我也觉得是，请跳过，嘿，看下验证码啊！  这个辣眼的提示

1. # -*- coding: utf-8 -*-
   
2. # @Time    : 2019-04-10 14:47
   
3. # @Author  : Ein
   
4. # @File    : tesserocr验证码识别-古诗文.py
   
5. # @Software: PyCharm
   
6. '''
   
7. g1RRZ+h02qqY4rNct0PPw/ELkgPaoF82SoPcA8lIEM2q0bQsHXhB6ANpGwCX25N5r3jqUq9XpNkweWhcQD7j1iaAp4kG4xG48jx3iow1bBtRhU+9fqw1Y/Tp3g0=
   
8. C93BE1AE
   
9. '''
   
10. 
    
11. import requests,os
    
12. from lxml import etree
    
13. def yzm_file_path():
    
14.     currrent_path = os.path.realpath(__file__)  # 文件绝对路径
    
15.     current_dir = os.path.split(currrent_path)[0]
    
16.     return current_dir + r"/yzm.jpg"
    
17. def down_html():
    
18.     currrent_path = os.path.realpath(__file__)  # 文件绝对路径
    
19.     current_dir = os.path.split(currrent_path)[0]
    
20.     return current_dir + r"/htj.html"
    
21. 
    
22. headers = {
    
23.     "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64)"
    
24. }
    
25. session = requests.Session()
    
26. 
    
27. def download_yzm():
    
28.     url = 'https://so.gushiwen.org/user/login.aspx?from='
    
29.     resp = session.get(url=url,headers=headers).text
    
30.     tree = etree.HTML(resp)
    
31.     #图片地址：
    
32.     img_url = 'https://so.gushiwen.org' + tree.xpath("//img[@id='imgCode']/@src")[0]
    
33.     #获取验证码令牌
    
34.     VIEWSTATE = tree.xpath("//input[@type='hidden'][1]/@value")[0]
    
35.     VIEWSTATEGENERATOR = tree.xpath("//input[@type='hidden'][1]/@value")[1]
    
36.     #下载到本地
    
37.     response = session.get(url=img_url,headers=headers)
    
38.     yzm_path = yzm_file_path()
    
39.     with open(yzm_path, "wb") as f:
    
40.         f.write(response.content)
    
41.     return VIEWSTATE,VIEWSTATEGENERATOR
    
42. 
    
43. def poss_pwd(vi,vie):
    
44.     pwd_url = 'https://so.gushiwen.org/user/login.aspx?from=http%3a%2f%2fso.gushiwen.org%2fuser%2fcollect.aspx'
    
45.     yzm = input("嘿，看下验证码啊！=")
    
46.     data = {
    
47.         '__VIEWSTATE': vi,
    
48.         '__VIEWSTATEGENERATOR': vie,
    
49.         'from':'http://so.gushiwen.org/user/collect.aspx',
    
50.         'email':'812745685@qq.com',
    
51.         'pwd': 'asd123..',
    
52.         'code': yzm,
    
53.         'denglu': '登录',
    
54.     }
    
55.     path = down_html()
    
56.     response = session.post(url=pwd_url,data=data)
    
57.     with open(path,'w',encoding='utf8') as fp:
    
58.         fp.write(response.text)
    
59. 
    
60. 
    
61. def main():
    
62.     vi,vie = download_yzm()
    
63.     poss_pwd(vi,vie)
    
64. 
    
65. if __name__ == '__main__':
    
66.     main()

复制代码

goolming

你好，可以加个好友交流一下吗