Snooper Analiser

import json, sys, os, math
from pprint import pprint

def fixMem(data, key):
    if not key in data:
        return

    mem = data[key]
    if not 'MB' in mem:
        data[key] = '%dMB' % int(math.ceil(int(mem) / 1024.0 / 1024.0 / 256.0) * 256)
    else:
        data[key] = '%dMB' % int(math.ceil(int(mem[0:-2]) / 256.0) * 256)

def pad(data, key):
    if not key in data:
        return
    ret = {}
    max = 0
    for item in data[key]:
        if len(str(item)) > max: max = len(str(item))
    for item in data[key]:
        ret[str(item).rjust(max, ' ')] = data[key][item]
    data[key] = ret

f = open('clients.json').readlines()
data = {}
x = 1
for line in f:
    j = json.loads(line)
    for item in j:
        if item == '_id':
            continue

        if not item in data:
            data[item] = {}

        if item == 'memoryMax':   fixMem(j, 'memoryMax')
        if item == 'memoryTotal': fixMem(j, 'memoryTotal')
        if item == 'memoryFree':  fixMem(j, 'memoryFree')

        if item == 'openGlVersion':
            j['openGlVersion'] = j['openGlVersion'].split(' ')[0][0:3]

        if item == 'playersSeen':
            if not 'total' in data['playersSeen']:
                data['playersSeen']['total'] = 0
                data['playersSeen']['average'] = 0

            c = int(j['playersSeen'])
            data['playersSeen']['total'] += c

            if c < 1000: c = math.ceil(c / 100.0) * 100
            else: c = math.ceil(c / 1000.0) * 1000
            j['playersSeen'] = int(c)

        if item == 'playersCurrent':
            if not 'total' in data['playersCurrent']:
                data['playersCurrent']['total'] = 0
                data['playersCurrent']['average'] = 0

            c = int(j['playersCurrent'])
            data['playersCurrent']['total'] += c

            if c < 100: c = math.ceil(c / 10.0) * 10
            else: c = math.ceil(c / 100.0) * 100
            j['playersCurrent'] = int(c)

        if item == 'playersMax':
            if not 'total' in data[item]:
                data[item]['total'] = 0
                data[item]['average'] = 0

            c = int(j[item])
            data[item]['total'] += c

            if c < 100: c = math.ceil(c / 10.0) * 10
            else: c = math.ceil(c / 100.0) * 100
            j[item] = int(c)

        if item == 'osVersion' and 'osName' in j:
            os = j['osName']

            if not os in data[item]:
                data[item][os] = {}

            if  not j[item] in data[item][os]:
                data[item][os][j[item]] = 1
            else:
                data[item][os][j[item]] += 1
        else:
            if  not j[item] in data[item]:
                data[item][j[item]] = 1
            else:
                data[item][j[item]] += 1

    print "%d / %d (%0.0f%%)" % (x, len(f), int(math.floor((x / float(len(f))) * 100.0))), "\r",
    sys.stdout.flush()
    x += 1
print ""

if 'playersSeen' in data:
    data['playersSeen']['average'] = int(data['playersSeen']['total'] / len(f))
if 'playersCurrent' in data:
    data['playersCurrent']['average'] = int(data['playersCurrent']['total'] / len(f))
if 'playersMax' in data:
    data['playersMax']['average'] = int(data['playersMax']['total'] / len(f))
    data['playersMax'].pop('total')

pad(data, 'memoryMax')
pad(data, 'memoryTotal')
pad(data, 'memoryFree')
pad(data, 'playersSeen')
pad(data, 'playersCurrent')
pad(data, 'playersMax')
pprint(data)


five = 0
total = 0
for ver in data['javaVersion']:
    if ver[0:3] == '1.5':
        five += data['javaVersion'][ver]
    total += data['javaVersion'][ver]
print 'Java 5 userbase: %d/%d (%0.000f%%)' % (five, total, float(five) / float(total) * 100.0)