Shellex Blog Here

#!/usr/bin/python
# -*- coding: utf-8 -*-
# by shellex [www.sxnsx.com]
import re
import os
import sys
import gtk
import gtk.glade
import cookielib
import urllib, urllib2
import threading
 
class xn_photo_downloader(threading.Thread):
	def __init__(self):
		threading.Thread.__init__(self, name='downloader')
		self.cookies = cookielib.CookieJar()
		self.opener = urllib2.build_opener(urllib2.HTTPCookieProcessor(self.cookies))
		self.save_path = '/home/shellex/Desktop/xnphotosd/d'
		self.target_url = ''
		self.usr = ''
		self.pwd = ''
		self.page_cnt = 0
		self.pbar = None
 
	def verify(self):
		verify_url = 'http://login.xiaonei.com/Login.do'
		verify_data= urllib.urlencode(
			{'email':self.usr,
			 'password': self.pwd,
			 'origURL':'http://www.xiaonei.com/SysHome.do'
			 })
		self.opener.open(verify_url, verify_data)
		print '[!]verify finish.'
 
	def go(self, target):
		rep = self.opener.open(target)
		return rep.read()
 
	def enmu_page_url(self):
		reg1 = '<div class="pager-bottom">[\s\S]*</ol';
		m = re.compile(reg1).search(self.go(self.target_url))
		if m == None or m.group().find('li') == -1:
			return [self.target_url]
		else:
			reg2 = '<li.*href="(.*?)"'
			params = re.findall(reg2, m.group())
			params = params[:-1]
			base_url = 'http://photo.xiaonei.com/'
			return [ base_url + each for each in params]
 
 
 
	def enmu_photo_url(self, url):
		reg = '<td class="photoPan">\s*<a href="(.*)">'
		result = re.compile(reg).findall(self.go(url))
		return result
 
	def download_photo(self, url):
		reg = '<div class="photo">\s*<img .* src="(.*?)" '
		photo_url = re.compile(reg).search(self.go(url)).group(1)
		print '[!]download:' + photo_url
		buf = self.opener.open(photo_url, 'wb').read()
		head, tail = os.path.split(photo_url)
		file = open(self.save_path + '/' + tail, 'w+')
		file.write(buf)
		file.close()
 
	def run(self):
		self.verify()
		pages = self.enmu_page_url()
		self.page_cnt = len(pages)
 
		loop1, loop2 = 1, 1
		for page_url in pages:
			print "[!]enter page:" +  page_url
			photos = self.enmu_photo_url(page_url)
			self.photo_cnt = len(photos)
			loop2 = 1
			for url in photos:
				progress_txt = 'Page %s/%s: Photo %s/%s' % (loop1, self.page_cnt, loop2, self.photo_cnt) 
				self.download_photo(url)
				self.pbar.set_text(progress_txt)
				self.pbar.set_fraction((0.0 + loop2)/self.photo_cnt)
				loop2 += 1
			loop1 += 1
		self.pbar.set_text("Done!")
 
class main_win():
	def start_btn_clicked_cb(self, widget):
		downloader = xn_photo_downloader()
		downloader.save_path = self.ui.get_widget('brower_btn').get_current_folder()
		downloader.usr = self.ui.get_widget('usr_box').get_text()
		downloader.pwd = self.ui.get_widget('pwd_box').get_text()
		downloader.target_url = self.ui.get_widget('album_url_box').get_text()
		downloader.pbar = self.ui.get_widget('progressbar_main')
		downloader.setDaemon(True)
		downloader.start()
 
	def __init__(self):
		# load ui and connect signals
		self.ui = gtk.glade.XML('ui.glade')
		self.ui.signal_autoconnect(self)
		self.win = self.ui.get_widget('main_win')
		self.win.show()
		self.win.connect("destroy", gtk.main_quit)
		self.ui.get_widget('exit_btn').connect("clicked", gtk.main_quit)
		try:
			gtk.main()
		except:
			print "Shit. A fucking exception caught. Dying a death."
			sys.exit(1)
 
if __name__ == '__main__':
	gtk.gdk.threads_init()
	mw = main_win()