- #!/usr/bin/env pyhton
- #coding:utf-8
- import urllib2
- import re
- import urllib
- import time
- def dl(url,mh,cao):
- req=urllib2.Request(url)
- req.add_header('Referer',cao)
- res=urllib2.urlopen(req)
- f=open(mh,'w')
- f.write(res.read())
- f.close()
- res.close()
- def jishu(url,url1,num,ri):
- req=urllib2.Request(url)
- req.add_header('Referer',url1)
- fi=urllib2.urlopen(req)
- down='http://c4.mangafiles.com/pictures/46/'+str(num)+'/'
- mat=re.compile(r'<option value="')
- for i in fi:
- a=len(re.findall(mat,i))
- for i in range(1,a+1):
- if i<10:
- mh="00"+str(i)+'.jpg'
- elif i<100:
- mh="0"+str(i)+'.jpg'
- else:
- mh=str(i)+'.jpg'
- dl(down+mh,mh,ri)
- time.sleep(0.9)
- fi.close()
- try:
- l=[]
- n=[]
- sum_page=0
- a=0
- tmp="http://www.imanhua.com/"
- url="http://www.imanhua.com/comic/46/" #这里可以改为用raw_input输入
- f=urllib2.urlopen(url)
- mat=re.compile(r'<a href="(/comic/\\d+/list_\\d+\\.html)"')
- for i in f:
- l=re.findall(mat,i)
- for i in l:
- mat=re.compile(r'.*?list_(\\d+).*?')
- match=mat.search(i)
- if match:
- n.append(match.group(1))
- sum_page=sum_page+1
- cao=tmp+i
- jishu(cao,url,n[a],tmp+i)
- a+=1
- break
- f.close()
- except Exception as e:
- print e
- #该片段来自于http://www.codesnippet.cn/detail/141020136406.html
来源: http://www.codesnippet.cn/detail/141020136406.html