91porn.py

#-*-coding=utf-8-*-
import requests
import re
import time
import random

page=1  #max=79
home='http://91porn.com/video.php?page=%d&category=rf'
cont_reg=re.compile('<a target=blank href="(http://91porn\.com/view_video\.php.*?)" title="')
vid_reg=re.compile('http://91\.p9p\.co/ev\.php\?VID=(.*?)<')
get_real='http://91.9p91.com/getfile_jw.php?VID='
real_reg=re.compile("<embed id='91' name='91' width='700' height='450' src='(.*?)'")
title_reg=re.compile('<title>([\w\W]*?)-')
vhead='http://91.9p91.com/9e.swf?autoplay=false&video_id=%s&mp4=999'
tr="<tr><td>%s</td><td><form target='_blank' action='hhhhhhhhplayer.html'><input type='hidden' name='url' value='%s'/> <input type='submit' value='view online'/></form></td></tr>\n"
data={'session_language':'cn_CN'
    }



cont_queue=[]
vid_queue=[]
video_list=[]
for i in range(1,page+1):
    url='http://91porn.com/video.php?page=%d&category=rf'%i
    print 'start parse url: ' + url
    try:
        content=requests.post(url,data=data).content
        list_=cont_reg.findall(content)
        vid_queue.extend(list_)
    except Exception,e:
        print e


for i in vid_queue:
    try:
        print 'start parse url: ' + i
        content=requests.post(i,data=data).content
        video=vid_reg.findall(content)[0]
        title=title_reg.findall(content)[0]
        print video,title
        apiurl=get_real+video
        r=requests.get(apiurl).content
        link=re.findall('file=(.*?mp4)&',r)[0]
        print link
        #video_list.append((title,vhead%video))
    except Exception,e:
        print e