A script for turning an ugly class webpage into a podcast rss feed

make-feed.py

from pyquery import PyQuery as pq
import urllib2
from datetime import datetime
import PyRSS2Gen as rss

d = pq(url='http://www.ssc.wisc.edu/~wright/Sociology-125-podcasts-2012.htm')
URL_ROOT = 'http://www.ssc.wisc.edu/~wright/'

items = []

for tr in reversed(d('table tr')):
	try:
		slides_link = URL_ROOT + tr[4].find('a').get('href')
	except (AttributeError, IndexError):
		slides_link = ''
	try:
		date_elem = tr[1].iter('font').next()
		date = date_elem.text.split('/')
	except AttributeError:
		date = date_elem.find('b').text.split('/')
	for n in reversed(tr[3]):
		try:
			url = n.get('href')
		except:
			continue
		f = urllib2.urlopen(URL_ROOT + url)
		items.append(rss.RSSItem(
			title = ' '.join(tr[2].iter('font').next().text.strip().split()),
			link = slides_link,
			description = n.text,
			guid = rss.Guid(URL_ROOT + url),
			enclosure = rss.Enclosure(
				url = URL_ROOT + url,
				length = f.headers['Content-Length'],
				type = 'audio/m4a-latm'
			),
			pubDate = datetime(2012, int(date[0]), int(date[1]))
		))

feed = rss.RSS2(
	title = "Sociology 125 Podcast - Fall 2012",
	link = 'http://www.ssc.wisc.edu/~wright/Sociology-125-podcasts-2012.htm',
	description = "Lectures and slides from Sociology 125, taught by Erik Olin Wright",
	lastBuildDate = datetime.now(),
	items = items
)

feed.write_xml(open('soc125-feed.xml', 'w'))