allcinema

cinema.py

#!/usr/lib/python
#vim:fileencoding=utf-8
import urllib2
import re
from BeautifulSoup import BeautifulSoup

class Cinema:
    def __init__(self):
        url = 'http://www.allcinema.net/prog/index2.php'
        self.soup = BeautifulSoup(urllib2.urlopen(url))

    def getTitle(self):
        node = self.soup.find('ul', {'class': 'toprankinglist'})
        res = re.search(u'href="(.*?)".*?>(.*?)</a>', str(node))
        return 'http://www.allcinema.net/prog/' + res.group(1), res.group(2)

if __name__ == '__main__':
    obj = Cinema()
    obj.getTitle()