爬取最新电影下载网每个电影链接

赵乾舟

1. import requests
   
2. from pyquery import PyQuery as pq
   
3. url = 'https://www.993dy.com/vod-type-id-1-pg-{pn}.html'
   
4. headers = {
   
5. 
   
6.     'Referer':'https://www.993dy.com/vod-type-id-1-pg-1.html',
   
7.     'User-Agent':'Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:88.0) Gecko/20100101 Firefox/88.0'
   
8. }
   
9. 
   
10. for page in range(1,3):
    
11. 
    
12.     listurl = url.format(pn=page)
    
13.     r = requests.get(listurl,headers=headers).text
    
14. 
    
15. 
    
16.     d = pq(r)
    
17. 
    
18. 
    
19.     for i in d('.img-list li'):
    
20.         sub_d = pq(i)
    
21.         print(sub_d('h5').text(),end=' ')
    
22.         print('https://www.993dy.com'+sub_d('a').attr('href'))
    

复制代码