Double Post Point Calculator

# Calculates player scores for the Double Post thread at the Forum Games section fo the xkcd fora.
# Usage: python doublepostpoints.py [flags]
# Use flag -l to include a log of all post authors, timestamps, and how many points each post scored.
# Use flag -t to only count posts up to the supplied number
# Use flag -i to specify a topic id (to use a different thread)

import urllib.request, sys, re, datetime, getopt, time as mtime

postPattern = re.compile('<p class="author"><a[^>]*><img[^>]*></a>by <strong><a[^>]*>([^>]*)</a></strong> &raquo; ([^>]*)</p>')
datetimePattern = re.compile('... (...) ([0-9]{2}), ([0-9]{4}) ([0-9]{1,2}):([0-9]{2}) ([ap]m) UTC')
pageNumberPattern = re.compile('Page <strong>([0-9]+)</strong> of <strong>([0-9]+)</strong>')
months = {'Jan':1, 'Feb':2, 'Mar':3, 'Apr':4, 'May':5, 'Jun':6, 'Jul':7, 'Aug':8, 'Sep':9, 'Oct':10, 'Nov':11, 'Dec':12}

doneLastPage = False
pagesDone = 0
postsDone = 0
lastPoster = None
lastTime = None
playerPoints = {}

log, upto, topicId = False, None, '108854'
flags, _ = getopt.getopt(sys.argv[1:], 'lt:i:')
for flag, value in flags:
    if flag == '-t':
        upto = int(value)
    elif flag == '-l':
        log = True
    elif flag == '-i':
        topicId = (value == 'count' or value == 'million') and '109251' or value

log and print('==== Log ====')
try:
    while not doneLastPage:
        response=None
        attempts=0
        while response==None and attempts < 3:
            try:
                response = urllib.request.urlopen('http://forums.xkcd.com/viewtopic.php?t=' + topicId + '&start=' + str(40*pagesDone), timeout=15)
            except urllib.error.URLError:
                pass
            attempts += 1
        if not response:
            sys.exit("Unable to load page " + str(pagesDone+1) + " after 3 attempts")
        pageHTML = str(response.read())
        response.close()
        log and print('Page ' + str(pagesDone+1) + ':')
        for (poster, time) in re.findall(postPattern, pageHTML):
            log and print('\t' + poster + ': ' + time)
            timeParts = re.search(datetimePattern, time).groups()
            time = datetime.datetime(int(timeParts[2]), months[timeParts[0]], int(timeParts[1]), int(timeParts[3]) % 12 + (timeParts[5] == 'pm' and 12 or 0), int(timeParts[4]))
            if poster == lastPoster:
                points = int(((time - lastTime).total_seconds()/60)**2)
                playerPoints[poster] = (poster in playerPoints and playerPoints[poster] or 0) + points
                log and print('\t\tScored: ' + str(points))
            postsDone += 1
            if upto and postsDone == upto:
                raise Exception
            lastPoster, lastTime = poster, time
        (currentPage, lastPage) = re.search(pageNumberPattern, pageHTML).group(1, 2)
        pagesDone += 1
        doneLastPage = currentPage == lastPage
        mtime.sleep(1) #Pause to prevent spamming the server with requests.
except Exception:
    pass

log and print('')
print('==== Points ====')
for name in sorted(playerPoints, key=playerPoints.get, reverse=True):
    print(name + ': ' + str(playerPoints[name]))