# python
# -*- coding: utf8 -*-

import urllib2
import json
from datetime import datetime
import matplotlib.pyplot
import matplotlib.dates

page = 'Wiktionary:Löschkandidaten'
totalcount = 12000

remcount = totalcount
rvcontinue = ''
revlist = []
while remcount > 0:
    if remcount > 500:
        querycount = 500
        remcount -= querycount
    else:
        querycount = remcount
        remcount = 0
    url = 'http://de.wiktionary.org/w/api.php?action=query&prop=revisions&format=json&rvprop=timestamp|size&rvlimit='+str(querycount)+'&rvdir=older'
    if rvcontinue != '':
        url += '&rvcontinue='+rvcontinue
    url += '&titles='+page
    response = urllib2.urlopen(url)
    rawdata = response.read()
    jsondata = json.loads(rawdata)
    revlist += jsondata['query']['pages'].values()[0]['revisions']
    try:
        rvcontinue = str(jsondata['query-continue']['revisions']['rvcontinue'])
    except:
        print 'Page history is complete.'
        break

timestructs = []
size = []

for rev in revlist:
    timestruct = datetime.strptime(rev['timestamp'],"%Y-%m-%dT%H:%M:%SZ")
    # hack to exclude extreme case of vandalism
    if timestruct < datetime.strptime("2009-12-30T23:20:00Z","%Y-%m-%dT%H:%M:%SZ") or timestruct > datetime.strptime("2009-12-31T00:40:00Z","%Y-%m-%dT%H:%M:%SZ"):
        timestructs.append(timestruct)
        size.append(rev['size'])

dates = matplotlib.dates.date2num(timestructs)
matplotlib.pyplot.plot_date(dates,size,'b-')
matplotlib.pyplot.tight_layout()
matplotlib.pyplot.show()