khohols.py

import httplib
import json
import re
import sys


def main():
    if len(sys.argv) < 2:
        print 'Please pass thread url in commandline'
        return

    url = sys.argv[1]
    if url[len(url) - 4:] == 'html':
        url = url.replace('html', 'json')
    else:
        url += '.json'      # this could fail hard

    header = {'User-Agent': 'Opera 9.20'}
    conn = httplib.HTTPConnection('www.2ch.hk', 80, header)
    conn.request('GET', url)
    response = conn.getresponse()
    if 300 <= response.status < 200:
        print response.status + '\n' + response.reason
        return

    thread = response.read()
    conn.close()
    decoded = json.loads(thread)

    trips = {}
    hohols = 0
    regexp = re.compile('.+postertripid\">([\w\d\+/-]{8}).+src=\"/flags/([a-zA-Z]+)\.png')
    for i in decoded['thread']:
        field = i[0][u'name']
        t, c = regexp.match(field).groups()
        if not t in trips:
            if c == u'ua':
                hohols += 1
                trips[t] = 1

    print hohols
    return


if __name__ == '__main__':
    main()