Xpath实战三：下载煎蛋网妹子照片

  

   

    

     

    
    

     

      #coding=utf-8 
     
    
   

    

     

    
    

     

      import requests
     
    
   

    

     

    
    

     

      import urllib.request
     
    
   

    

     

    
    

     

      from lxml import etree
     
    
   

    

     

    
    

     
 
     
    
   

    

     

    
    

     

      def getHtml(url):
     
    
   

    

     

    
    

     

       headers = {'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:56.0) Gecko/20100101 Firefox/56.0'}
     
    
   

    

     

    
    

     

       page = requests.get(url,headers = headers)  
     
    
   

    

     

    
    

     

       html =page.text
     
    
   

    

     

    
    

     
 return html
     
    
   

    

     

    
    

     
 
     
    
   

    

     

    
    

     

      def getImg(html):
     
    
   

    

     

    
    

     

       html = etree.HTML(html)
     
    
   

    

     

    
    

     

       img_info = html.xpath('//a[@class = "view_img_link"]//@href') #照片链接都在a标签下，先定位view_img_link，再@href，即可获取所有图片链接
     
    
   

    

     

    
    

     
 for index,img in enumerate(img_info,1):
     
    
   

    

     

    
    

     
 print ("正在下载第{}张图片".format(index))
     
    
   

    

     

    
    

     

       img = "http:" + img
     
    
   

    

     

    
    

     

       urllib.request.urlretrieve(img,'%s%s' % (index,img[-4:]))
     
    
   

    

     

    
    

     
 
     
    
   

    

     

    
    

     

      if __name__=='__main__':
     
    
   

    

     

    
    

     

       url = "http://jandan.net/ooxx"
     
    
   

    

     

    
    

     

       html = getHtml(url)
     
    
   

    

     

    
    

     

       getImg(html)
     
    
   

    

     

    
    

     
 
     
    
   

    

     

    
    

     
 print ("OK!All DownLoad!")
     
    
  
 

文章来源: blog.csdn.net，作者：悦来客栈的老板，版权归原作者所有，如需转载，请联系作者。

原文链接：blog.csdn.net/qq523176585/article/details/78524457