root/confluence/feedfilters/shfiction.py

#!/usr/bin/env python

# This filter converts the Strange Horizons new fiction page to an rss feed.
# Don't be a jerk; set the update period to at least one day.  It only updates once a week.
#
# Written by Adrianna Pinska
# Licence: GPLv3

import sys
import cgi
import re
import time

storypattern = re.compile('<p class="contents-title"><a href="([^"]*)">(.*?)</a></p>\n *<p class="contents-author">([^<]*)</p>\n *<p class="contents-date">([^<]*)</p>\n *<p class="contents-pullquote">(.*?)</p>\n *', re.DOTALL)
text = sys.stdin.read()
stories = storypattern.findall(text)

print """<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0">
  <channel>
    <title>Strange Horizons Fiction</title>
    <link>http://www.strangehorizons.com/fiction.shtml</link>
    <description>New fiction from Strange Horizons</description>
    <language>en-us</language>
    <ttl>1440</ttl>"""

for link, title, attribution, date, blurb in stories:
    feed_title = "%s %s" % (cgi.escape(title), attribution)
    feed_date = time.strftime("%a, %d %B %Y %H:%M:%S", time.strptime(date, "%d %B %Y"))
    text = cgi.escape(blurb)
    text = text.replace("\x92","&apos;")
    print """  <item>
    <title>%s</title>
    <description>%s</description>
    <pubDate>%s</pubDate>
    <guid>%s</guid>
    <link>%s</link>
  </item>""" % (feed_title, text, feed_date, link, link)

print """  </channel>
</rss>"""