suggested-edit-delay-stats

#! /usr/bin/env python
import json, sys, time, urllib

filter_code = '!*MKIiKJWje.bjTxT'
topbar_rollout_dates = {
    'meta.stackoverflow': 1384466400, # 2013-11-14 22:00 UTC
    None: 1386183600, # 2013-12-04 19:00 UTC
    'askubuntu': 1386633600, # 2013-12-10 00:00 UTC
}
seconds_per_week = 604800
delay_between_sites = 1

def retrieve(site, from_date, to_date):
    page = 1
    url = 'http://api.stackexchange.com/2.1/suggested-edits?page=%d&pagesize=100&fromdate=%d&todate=%d&order=desc&sort=creation&site=%s&filter=%s' % (page, from_date, to_date, site, filter_code)
    f = urllib.urlopen(url)
    data = json.load(f)
    f.close()
    return data['items']

def average_delay(se_list):
    total = 0
    count = 0
    for se in se_list:
        creation_date = se['creation_date']
        if se.has_key('approval_date'):
            total += se['approval_date'] - creation_date
            count += 1
        elif se.has_key('rejection_date'):
            total += se['rejection_date'] - creation_date
            count += 1
    if count == 0: return (0, float('nan'))
    return (count, total / 60.0 / count)

if __name__ == '__main__':
    print 'site                   before^2      before       after'
    print '                      avg(mn) cnt  avg(mn) cnt  avg(mn) cnt'
    for site in sys.argv[1:]:
        topbar_rollout_date = \
            topbar_rollout_dates[site if topbar_rollout_dates.has_key(site) else None]
        after_list = retrieve(site, topbar_rollout_date, topbar_rollout_date + seconds_per_week)
        (after_count, after_delay) = average_delay(after_list)
        before_list = retrieve(site, topbar_rollout_date - seconds_per_week, topbar_rollout_date)
        (before_count, before_delay) = average_delay(before_list)
        before2_list = retrieve(site, topbar_rollout_date - 2 * seconds_per_week, topbar_rollout_date - seconds_per_week)
        (before2_count, before2_delay) = average_delay(before2_list)
        print '%20s  %6.0f %4d  %6.0f %4d  %6.0f %4d  %4d%%' % (
            site,
            before2_delay, before2_count,
            before_delay, before_count,
            after_delay, after_count,
            int(200.0 * after_delay / (before2_delay + before_delay)))
        time.sleep(delay_between_sites)