root/confluence/feedfilters/twitterfilter.py @ 587

#!/usr/bin/env python

# This filter removes posts which contain any of the specified keywords
# from a Twitter feed.  It also makes urls and @usernames into clickable
# links, and shortens item titles to poster usernames.
#
# Written by Adrianna Pinska
# Licence: GPLv3

import sys
import re
import xml.dom.minidom

# Add crap you don't want to see to this list:
noise = ["#postcrossing", "[Nn]et[Pp]rophet", "#[Aa]frica09"]

noisepattern = re.compile("(%s)" % "|".join(noise))
titlepattern = re.compile(r"^([^:]*):.*$", re.DOTALL)
linkurls = re.compile(r"(https?://[^ ]*)")
linknames = re.compile(r"@([a-zA-Z0-9_]+)")
hashtags = re.compile(r"#([a-zA-Z0-9_]+)")
statusidpattern = re.compile(r"http://twitter.com/.*/statuses/([0-9]*)")

xml = xml.dom.minidom.parseString(sys.stdin.read())

channel = xml.getElementsByTagName("rss")[0].getElementsByTagName("channel")[0]
for item in channel.getElementsByTagName("item"):
    description = item.getElementsByTagName("description")[0].firstChild
    text = description.data

    if noise and noisepattern.search(text):
        channel.removeChild(item)
        item.unlink()
        continue

    title = item.getElementsByTagName("title")[0].firstChild
    title.data = titlepattern.sub(r"\1", title.data)
    username = title.data.strip(":")

    guid = item.getElementsByTagName("guid")[0].firstChild
    statusid = statusidpattern.match(guid.data).group(1)

    text = linkurls.sub(r"<a href='\1'>\1</a>", text)
    text = linknames.sub(r"<a href='http://twitter.com/\1'>@\1</a>", text)
    text = hashtags.sub(r"<a href='http://search.twitter.com/search?q=%23\1'>#\1</a>", text)
    text += " <a href='http://twitter.com/home?status=@%s%%20&in_reply_to_status_id=%s&in_reply_to=%s'>Reply</a>" % (username, statusid, username)
    description.data = text

print xml.toxml(encoding="UTF-8")