#!/usr/bin/python

import gzip, re, time, sys, gdchart
import psyco; psyco.full()

def swizzleDate(d):
    t = time.strptime(d, "%d/%b/%Y")
    return time.strftime("%Y-%m-%d", t)

assert('2004-04-25' == swizzleDate('25/Apr/2004'))

parseRE=re.compile(r'\[(\d+/\w+/\d+).*GET /~nelson/weblog/index\.rss.*Bloglines.*?(\d+) subscribers')

data = {}
def parse(fp):
    for l in fp:
        m = parseRE.search(l)
        if m:
            date, count = m.groups()
            date = swizzleDate(date)
            data[date] = count

parse(file('/var/log/apache/access.log'))
parse(file('/var/log/apache/access.log.1'))
for n in range(2,39):
    parse(gzip.open('/var/log/apache/access.log.%d.gz' % n))

dates = data.keys()
dates.sort()
for d in dates:
    print d, data[d]

gdchart.option(
    bg_color=0xffffff,
    grid=gdchart.GDC_TICK_NONE,
    format=gdchart.GDC_PNG,
    title='Bloglines readers over time',
    set_color=[0x000000]
)

graphDates=[(d[:4] + d[5:7]) for d in dates]
graphData =[int(data[d]) for d in dates]

gdchart.chart(gdchart.GDC_LINE, (350, 225), 'bloglines.png',
              graphDates, graphData)