多线程下载sis001的网友自拍贴图版面的图片,这个程序并不能下载全部版面而是只下载第二页的内容。因为第二页是最新的内容哦。你需要一个sis001,5级以上的用户名和密码否则无法访问此版面。
#!/usr/bin/python
# -*- coding: cp936 -*-
#coding utf-8
import urllib
import urllib2
import re
import cookielib
import Queue
import threading
def downPic(tiezi_url):
req = urllib2.Request(q.get(),None,headers)
tiezi_html = opener.open(req).read()
#print tiezi_html
re_img = re.compile(r'\<img src\=\"(http\:\/\/.*?\.jpg|attachments\/.*?.jpg)\"')
img_list = re_img.findall(tiezi_html)
#print img_list
for i in img_list:
if re.match("http",i):
print "%s downloading..."%i
filename = re.split(r'/',i)
try:
req = urllib2.Request(i,None,headers)
res = opener.open(req).read()
open(filename[-1],'wb').write(res)
except:
pass
else:
img_url = "http://38.103.161.185/forum/%s"%i
filename = re.split(r'/',img_url)
print "inner link %s"%img_url
try:
req = urllib2.Request(img_url,None,headers)
res = opener.open(req).read()
open(filename[-1],'wb').write(res)
e