#!/usr/bin/env python

# run this script and it will connect to irc://irc.wikimedia.org/en.wikipedia and
# print out live edits that are happening on wikipedia.
#
# you'll need to install python-irclib

import re
import sys
import irclib
import getpass


# now we have two problems, right?
MESSAGE_PATTERN = re.compile('\[\[(.+?)\]\] (.+)? (http:.+?)? (?:\* (.+?) \*)? (?:\(([+-]\d+)\))? (.+)?')

class WikipediaClient(irclib.SimpleIRCClient):

    def on_error(self, connection, event):
        print event

    def on_created(self, connection, event):
        connection.join("#en.wikipedia")

    def on_pubmsg(self, connection, event):
        msg = strip_color(event.arguments()[0])
        match = MESSAGE_PATTERN.search(msg)
        if match:
            page, status, diff_url, user, bytes_changed, msg = match.groups()
            print "page: %s" % page
            print "status: %s" % status
            print "diff: %s" % diff_url
            print "user: %s" % user
            print "bytes: %s" % bytes_changed
            print "msg: %s" % msg
            print
        else:
            print "NO MATCH: %s" % msg
            connection.close()
            sys.exit(-1)


def strip_color(msg):
    # should remove irc color coding
    return re.sub(r"(\x03|\x02)([0-9]{1,2}(,[0-9]{1,2})?)?", "", msg)


def main(username):
    irc = WikipediaClient()
    irc.connect("irc.wikimedia.org", 6667, username)
    irc.start()

if __name__ == "__main__":
    username = getpass.getuser()
    main(username)