Yandex Python Parser

# coding: utf-8
import urllib
import sys
from BeautifulSoup import BeautifulSoup

#"""Fake class for serialization"""
class A(object):pass

def A2XML(A):
    result = u'<train><time>%s</time><url>%s</url><note>%s</note></train>' % (A.time, u'not realized', A.note, )
    return result

def dumpXML(url):
    archives_url = url
    archives_html = urllib.urlopen(archives_url)
    soup = BeautifulSoup(unicode(archives_html.read(),'utf8'))

    direction = ''.join(soup.find("h2", { "class" : "b-holster b-title"}).contents[0])

    rasp = []
    for i in soup.findAll("span", { "class" : "time " }):
        aaa = A()
        aaa.href = i.a['href']
        aaa.time = ''.join(i.a.contents[0])
        aaa.note = ''.join(i.nextSibling.contents[0])
        rasp.append(aaa)
    raspstring = [ u'<rasp name="%s">' % (direction,) ,]
    for aaa in rasp: raspstring.append(A2XML(aaa))
    raspstring.append('</rasp')


    resultSoup = BeautifulSoup(''.join(raspstring))
    return resultSoup.prettify()

if __name__ == "__main__":
    if (len(sys.argv) < 2):
        print u'Необходимо задать путь к расписанию в качестве единственного аргумента скрипта.'
        exit(1)
    print dumpXML(str(sys.argv[1]))