itunesu2rss.py

#! /usr/bin/env
import datetime
import sys
import urllib2

import PyRSS2Gen
from lxml import etree
from slugify import slugify


def url2rss(itunesu_url):
    response = urllib2.urlopen(itunesu_url)
    htmlparser = etree.HTMLParser()
    tree = etree.parse(response, htmlparser)
    feed_items = list()
    for row in tree.xpath('//*[@id="content"]/div/div[2]/div/div/table/tbody/tr'):
        episode = dict(row.items())
        # print episode['preview-title'].encode('utf-8')
        feed_items.append(PyRSS2Gen.RSSItem(
            title=episode['preview-title'],
            link=episode['audio-preview-url'],
            enclosure=PyRSS2Gen.Enclosure(episode['audio-preview-url'],
                                          urllib2.urlopen(
                                              episode['audio-preview-url']).info()['Content-Length'], "audio/mpeg"),
            guid=PyRSS2Gen.Guid(episode['audio-preview-url']),
            pubDate=datetime.datetime.strptime(urllib2.urlopen(episode['audio-preview-url']).info()['Date'],
                                               '%a, %d %b %Y %H:%M:%S %Z')))

    rss = PyRSS2Gen.RSS2(
        title=episode['preview-album'],
        link=itunesu_url,
        description=episode['preview-album'],
        lastBuildDate=datetime.datetime.now(),
        items=feed_items)

    rss.write_xml(open(slugify(episode['preview-album']) + ".xml", "w"))


if __name__ == '__main__':
    for link in sys.argv[1:]:
        try:
            print "Working on %s" % link
            url2rss(link)
        except Exception, e:
            print "%s failed: %r" % (link, e)