File:Edit rates by users and bots on wikimedia Commons July1 July27 2012.png

'''
Created on 09/07/2012

@author: Kim
'''

from calendar import timegm
from time import strptime, strftime, gmtime

from numpy import arange
from pylab import plot, xlabel, ylabel, legend, grid, title, show, axhline, figure, scatter, xlim, ylim

import wikipedia as pywikibot

bot_edit_rates = {}
user_edit_rates = {}
secs_per_min = 60
default_max_bot_edit_rate_per_minute = 6
datefmtstr = "%Y-%m-%dT%H:%M:%SZ"
rcstart = "2012-07-27T00:00:00Z"
rcend   = "2012-07-01T00:00:00Z"

def recent_changes_iter(rcstart, rcend, rcshow):
    interval = 60 * 60
    rcstart_utc_sec = timegm(strptime(rcstart, datefmtstr))
    rcend_utc_sec = timegm(strptime(rcend, datefmtstr))
    rc1 = rcstart_utc_sec
    rc2 = max(rcend_utc_sec, rc1 - interval)
    while True:
        rc1str, rc2str = strftime(datefmtstr, gmtime(rc1)), strftime(datefmtstr, gmtime(rc2))
        print rcshow, ":", rc2str,'-',rc1str
        for item in site.recentchanges(number=5000, rcshow=rcshow, 
                                       rcstart=rc1str,
                                       rcend=rc2str):
            yield timegm(strptime(item[1], datefmtstr)), item[4]
        if rc2 == rcend_utc_sec:
            break
        rc1 -= interval
        rc2 = max(rcend_utc_sec, rc1 - interval)        

try:
    site = pywikibot.getSite()
    for i, (utc_sec, bot_str) in enumerate(recent_changes_iter(rcstart, rcend, rcshow="bot")):
        utc_min = utc_sec / secs_per_min
        edit_rates = bot_edit_rates.setdefault(bot_str, {})
        if edit_rates.has_key(utc_min):
            edit_rates[utc_min] += 1
        else:
            edit_rates[utc_min] = 1
        if i == 0:
            last_bot_edit = utc_sec
    first_bot_edit = utc_sec

    for item in recent_changes_iter(rcstart, rcend, rcshow="!bot"):
        utc_sec = item[0]
        utc_min = utc_sec / secs_per_min
        if user_edit_rates.has_key(utc_min):
            user_edit_rates[utc_min] += 1
        else:
            user_edit_rates[utc_min] = 1
    
finally:
    pywikibot.stopme()
    
fast_bots = []
for bot_str, edit_rates in bot_edit_rates.iteritems(): 
    if max(edit_rates.itervalues()) > 7:
        fast_bots.append(bot_str)
fast_bots = frozenset(fast_bots)    
    
utc_min_minimum = first_bot_edit / 60 + 1
utc_min_maximum = last_bot_edit / 60 - 1
minutes_after_first_bot_edit = arange(utc_min_maximum + 1 - utc_min_minimum)
days_after_first_bot_edit = minutes_after_first_bot_edit / (60.0 * 24)

accum_edit_rate = []
other_bots_edit_rates = []
all_bots_edit_rates = []

for utc_min in xrange(utc_min_minimum, utc_min_maximum + 1):
    edit_rate = user_edit_rates.get(utc_min, 0)
    other_bots_edit_rate = 0
    all_bots_edit_rate = 0
    for bot_str, ber in bot_edit_rates.iteritems():
        edit_rate += ber.get(utc_min, 0)
        all_bots_edit_rate += ber.get(utc_min, 0)
        if bot_str not in fast_bots:
            other_bots_edit_rate += ber.get(utc_min, 0)
    accum_edit_rate.append(edit_rate)
    other_bots_edit_rates.append(other_bots_edit_rate)
    all_bots_edit_rates.append(all_bots_edit_rate)

all_user_edit_rates = [user_edit_rates.get(utc_min, 0) for utc_min in xrange(utc_min_minimum, utc_min_maximum + 1)]

max_user_edit_rate = max(all_user_edit_rates)
max_bot_edit_rate = max(all_bots_edit_rates)
max_edit_rate = max(max_user_edit_rate, max_bot_edit_rate)
    


#plot(minutes_after_first_bot_edit, accum_edit_rate, '-', label="All edits", lw=2, alpha = 0.7)

for bot_str in sorted(bot_edit_rates.iterkeys()):
    if bot_str in fast_bots:
        d = bot_edit_rates[bot_str]
        edit_rate_arr = [d.get(utc_min, 0) for utc_min in xrange(utc_min_minimum, utc_min_maximum + 1)]
        #plot(minutes_after_first_bot_edit, edit_rate_arr, ls="-", lw=2.5, label=repr(bot_str)[2:-1], alpha=0.7)

figure(figsize=(12, 9))
plot(days_after_first_bot_edit, all_user_edit_rates, 'b.', label="All user edits", lw=1, alpha=0.5)
plot(days_after_first_bot_edit, all_bots_edit_rates, 'r.', label="All bot edits", lw=1, alpha=0.5)
xlabel('Days after %s' % rcend)
ylabel('Edit rate [edits/min]')
title("Edit rates on Wikimedia Commons between %s and %s" % (rcend, rcstart))
xlim((0, days_after_first_bot_edit[-1]))
legend()
grid()

figure(figsize=(11, 11))
scatter(all_user_edit_rates, all_bots_edit_rates)
xlabel('Accumulated user edit rate [edits/min]')
ylabel('Accumulated bot edit rate [edits/min]')
#axhline(default_max_bot_edit_rate_per_minute, ls='--', c="red", label="Default max bot edit rate")
#legend()
grid()
xlim((0, max_edit_rate))
ylim((0, max_edit_rate))
title("Correlation between user and bot edit rates\nbetween %s and %s" % (rcend, rcstart))

#plot(minutes_after_first_bot_edit, other_bots_edit_rates, label="Other bots (accumulated)", alpha=0.7)




show()