python爬取贴吧图片 


 1 #!/usr/bin/env python
 2 # -*- coding: utf-8 -*-
 3 # @Time    : 2017/7/19 16:21
 4 # @Author  : wqj
 5 # @Contact : wqjhky@gmail.com
 6 # @Site    : 
 7 # @File    : img.py
 8 # @Software: PyCharm Community Edition
 9 import re
10 import urllib
11 import sys
12 import os
13 
14 reload(sys)
15 sys.setdefaultencoding('utf8')
16 
17 
18 def getHtml(url):
19     page = urllib.urlopen(url)
20     html = page.read()
21     return html
22 
23 
24 def getImg(html, x):
25     reg = r'src="(http://img.*?\.jpg)"'
26     imgre = re.compile(reg)
27     imList = re.findall(reg, html)
28 
29     print(imList)
30     for i in imList:
31         print(i)
32         print x
33         urllib.urlretrieve(i, '%s.jpg' % x)
34         x += 1
35     return x
36 
37 
38 x = 1
39 url = "http://tieba.baidu.com/f?ie=utf-8&kw=%E6%B2%B3%E5%8D%97%E7%A7%91%E6%8A%80%E5%AD%A6%E9%99%A2&fr=search?pn="
40 for k in range(1, 28):
41     ul = url + str(k)
42     print ul
43     html = getHtml(ul)
44     # print html
45     x = getImg(html, x)
46     dirpath = r'D:\img'
47     filename = os.path.join(dirpath, str(x) + ".jpg")

 以上是代码，程序较为简单

posted @ 2017-07-20 09:48  没时间看海  阅读(213)  评论(0)  编辑  收藏  举报