'''
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

import wikipedia as pywikibot

bot_edit_rates = {}
user_edit_rates = {}
secs_per_min = 60
default_max_bot_edit_rate_per_minute = 6
recentedits = 1850

try:
    site = pywikibot.getSite()
    for i, item in enumerate(site.recentchanges(number=recentedits, rcshow="bot")):
        utc_sec = timegm(strptime(item[1], "%Y-%m-%dT%H:%M:%SZ"))
        bot_str = item[4]
        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 site.recentchanges(number=5000, rcshow="!bot"):
        utc_sec = timegm(strptime(item[1], "%Y-%m-%dT%H:%M:%SZ"))
        if utc_sec > last_bot_edit:
            continue
        if utc_sec < first_bot_edit:
            print "BREAK"
            break 
        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()) > default_max_bot_edit_rate_per_minute:
        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)

accum_edit_rate = []
other_bots_edit_rates = []

figure(figsize=(12,9))

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
    for bot_str, ber in bot_edit_rates.iteritems():
        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)
plot(minutes_after_first_bot_edit, accum_edit_rate, '-', c='green', label="All edits", lw=2, alpha = 0.8)

edit_rate_arr = [user_edit_rates.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, c='k', label="User edits", lw=1.5, alpha=0.8)

for bot_str in bot_edit_rates:
    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=0.8)

axhline(default_max_bot_edit_rate_per_minute, ls='--', c="red", label="Default max bot edit rate")
plot(minutes_after_first_bot_edit, other_bots_edit_rates, c="b", label="Other bots (accumulated)", alpha=0.8)


xlabel('Elapsed minutes')
ylabel('Edit rate [edits/min]')
title("Edit rates on Wikimedia Commons from %s" % strftime("%Y-%m-%d %H:%M (UTC)", gmtime(utc_min_minimum * secs_per_min)))
legend()
grid()
show()