#!/usr/bin/python

import re
import urllib

def getHtml(url):
    return urllib.urlopen(url).read()

def getImgUrl(html):
    reg = r'src="(.*?\.jpg)"'
    imgRe = re.compile(reg)
    imgList = re.findall(imgRe, html)
    return imgList

def getImg(imgUrlList):
    x = 1
    imgUrlCount = len(imgUrlList)
    for imgUrl in imgUrlList:
        urllib.urlretrieve(imgUrl, '%s.jpg' % x)
        jindu = (x*100)/imgUrlCount
        print '#'*jindu
        print 'end %s' % jindu
        x += 1
    else:
        print 'endding!'

html =  getHtml('http://image.baidu.com')
imgUrlList = getImgUrl(html)
getImg(imgUrlList)

Logo

CSDN联合极客时间,共同打造面向开发者的精品内容学习社区,助力成长!

更多推荐