python爬虫脚本ie=utf-8_分享一个Python爬虫小脚本

此Python小脚本为抓取此页面：/p/2108681777 下的所有jpg图像

'''

Created on -4-2

@author: Administrator

'''

import re

import urllib

def getHtml(url):

page = urllib.urlopen(url)

html = page.read()

return html

def getImg(html):

reg = r'src="(.*?\.jpg)" width'

imgreg = pile(reg)

imglist = re.findall(imgreg, html)

x = 0

for imgurl in imglist:

urllib.urlretrieve(imgurl, '%s.jpg' % x)

x += 1

html = getHtml("/p/2108681777")

getImg(html)

阅读(1960) | 评论(0) | 转发(0) |