gistfile1.py

#!/usr/bin/python

import sys
import re
import pygeoip

# Note. Download GeoLite database from http://geolite.maxmind.com/download/geoip/database/GeoLiteCity.dat.gz and extract to /tmp dir
geoip_rawdata = pygeoip.GeoIP('/tmp/GeoLiteCity.dat')
def ipquery(ip):
    data = geoip_rawdata.record_by_name(ip)
    return data['country_name']

from datetime import *

one_day_uuids = {}
seven_day_uuids = {}
alltime_uuids = {}
version_uuids = {}
by_country = {}
current_date = ""
days = 0

print "%32s" % "number of unique uuids"
with open(sys.argv[1]) as f:
  for line in f:
    m = re.search( "^([\d\.]+).+\[(.+) \+0000\] \"GET \/manifest\.json\?uuid=([\d\w\-]+)&version=v([\d\.]+)", line)
    if not m:
      continue
    date = datetime.strptime( m.group(2), "%d/%b/%Y:%H:%M:%S" ).strftime("%d/%m/%y")
    if current_date != date:
      if current_date != "":
        days += 1
        print "%32s %5d" % (current_date, len(one_day_uuids))
      if days % 7 == 0:
        if current_date != "":
          print "%32s %5d" % ("7 days total", len(seven_day_uuids))
        seven_day_uuids.clear()
      one_day_uuids.clear()
      current_date = date
    one_day_uuids[m.group(3)] = 1
    seven_day_uuids[m.group(3)] = 1
    alltime_uuids[m.group(3)] = 1
    if not version_uuids.has_key(m.group(4)):
      version_uuids[m.group(4)] = {}
    version_uuids[m.group(4)][m.group(3)] = 1
    country = ipquery(m.group(1))
    if not by_country.has_key(country):
      by_country[country] = {}
    by_country[country][m.group(3)] = 1

print "%32s %5d" % (current_date, len(one_day_uuids))
if days % 7 and current_date != "":
  print "%32s %5d" % (str((days % 7) + 1) + " days unique", len(seven_day_uuids))

total_unique = len(alltime_uuids)
print "%32s %5d" % ("total unique", total_unique)

print
print "%32s" % "total by version"
for k,v in version_uuids.iteritems():
  uuids = len(v)
  print "%32s %6d %6.2f%%" % (k, uuids, uuids*100.0/total_unique)

print
print "%32s" % "total by country"
for k,v in by_country.iteritems():
  uuids = len(v)
  print "%32s %6d %6.2f%%" % (k, uuids, uuids*100.0/total_unique)