批量下载图片代码,Python代码库,德仔网

代码语言

代码分类

【Python】批量下载图片

作者:figer1 / 发布于2013/3/1/ 1249

批量下载图片


	#!/usr/bin/env python
	#-*-coding:utf-8-*-'
	#Filename:download_file.py
	import os,sys
	import re
	import urllib
	import urllib2

	
	base_url = 'xxx'

	array_url = list()
	pic_url = list()
	inner_url = list()

	def get_array_url(array_url,base_url):
	 content = urllib.urlopen(base_url).read()
	 array_url_a = re.findall(r'/rihan.*?.html',content)
	 for url in array_url_a:
	 url_a = 'xxx'+url
	 #print url_a
	 array_url.append(url_a)
	
	


	def get_inner_url(array_url,inner_url):
	 inner_url.append(array_url)
	 content = urllib.urlopen(array_url[10]).read()
	 content = content.replace(" ","")
	 url_a = re.findall(r'<li>.*?</li>',content)
	

	 for i in url_a:
	 url = re.findall(r'ahref=\'.*?.html\'target',i)
	 if len(url)>0:
	 # print url[0]
	
	 url_b = re.sub(r'ahref=\'','',url[0])
	 # print url_b
	
	 url_c = re.sub(r'\'target','',url_b)
	 url_c = 'http://xxx/'+re.sub(r'/.*/','',url_c)
	 inner_url.append(url_c)


	 del inner_url[1]
	 # print inner_url

	

	def get_pic_url(pic_url,inner_url,array_url):

	 content = urllib.urlopen(array_url).read()
	 pic_url_a = re.findall(r'center.*?.jpg',content)
	 print 'bbbbbbbbb',len(pic_url_a)
	 pic_url_a = re.findall(r'http://.*.jpg',pic_url_a[0])
	 pic_url.append(pic_url_a[0])

	 j=2

	

	 for i in inner_url:
	 jj = '/'+str(j)+'.jpg'
	 pic = re.sub(r'/1.jpg',jj,pic_url_a[0])

	 pic_url.append(pic)
	
	
	 j = j+1
	
	 del pic_url[-1]
	 for i in pic_url:
	 print i

	


	def urlcallback(a,b,c):
	 """
	 call back function
	 a,已下载的数据块
	 b,数据块的大小
	 c,远程文件的大小
	 """
	 print "callback"
	 prec=100.0*a*b/c
	 if 100 < prec:
	 prec=100
	 print "%.2f%%"%(prec,)

	
	def download(img_url,file_num):
	
	
	 for img in img_url:
	 print img
	 img_name = re.sub(r'http://.*/','',img)
	 path = 'C:/'+str(file_num)+'/'+img_name
	 urllib.urlretrieve(img,path,urlcallback)

	

	get_array_url(array_url,base_url)

	
	file_num = 3
	#download(pic_url,file_num)
	get_inner_url(url,inner_url)
	get_pic_url(pic_url,inner_url,url)


	'''

	for url in array_url:
	 print url
	 # get_inner_url(url,inner_url)
	 # get_pic_url(pic_url,inner_url)
	 get_inner_url(url,inner_url)
	 get_pic_url(pic_url,inner_url,url)
	 download(pic_url,file_num)
	 file_num = file_num+1
	 del inner_url[:]
	 del pic_url[:]
	
	'''

试试其它关键字

　批量下载图片　

同语言下

可能有用的

figer1贡献的其它代码(8)