get_icons_from_macx.py

#!/usr/bin/python
# -*- coding:utf-8 -*-

import re
import urllib
import urllib2
from BeautifulSoup import *

SAVE_PATH = "/Users/jeswang/Desktop/ICONS/"

def get_icons_from_macx(page_number):
  # page_number: 软件列表的页号
    c = urllib2.urlopen("http://soft.macx.cn/index.htm?page=" + page_number)
    soup = BeautifulSoup(c.read())
    c.close()
    items =  soup.findAll('li', {"class":"item"})

    # 单线程存储图片
    for i in items:
		imgURL = i.find('img')['src']
		title_with_version = i.find('a',{"class":"title"}).contents[0]
		title = re.search(ur"^([\w\W]*?)(\s[0-9|.]*)?$", title_with_version).group(1)
		title = title.replace("/"," ") # 可能会出现系统不支持的文件名，此处过滤的不完全

		print title, imgURL
		urllib.urlretrieve(imgURL, SAVE_PATH+title+".jpg")

if __name__ == "__main__":
	for i in range(0, 220):
		print i
		get_icons_from_macx(str(i))