NEERC Standings Filter

#!/usr/bin/env python
import re

team_re = re.compile('<td class="party">(.*?)</td>')
data = open("standings.html").read()

seen_uni = set()
replacements = set()

data = data.replace('../standings-wt.css','http://neerc.ifmo.ru/standings-wt.css')

c = 0
for match in team_re.finditer(data):
    match = match.groups()[0]
    uni, team = match.rsplit('(', 1)
    uni, team = uni.strip(), '(' + team
    su = uni[:-1] if uni[-1].isdigit() else uni
    if su in seen_uni:
        replacements.add((match, '<del>%s</del>' % match))
    else:
        c += 1
        replacements.add((match, '<b>%d:</b> %s' % (c, match)))
        seen_uni.add(su)

for x, y in replacements:
    data = data.replace('<td class="party">%s</td>' % x, '<td class="party">%s</td>' % y)

with open("standings-unique.html", "wb") as f:
    f.write(data)