gistfile1.txt

import binascii
import hashlib
import random
import json
import bz2


def mbid(pos):
    h = hashlib
    return h.sha224(str(pos)).hexdigest()[:10]


def sim_config(name, strings=86, doms_per_str=60, domfrac=1.0):
    return {'config': name,
            'strings': [{'str': s + 1,
                         'doms': [{mbid(d + 1): d + 1}
                                  for d in range(doms_per_str)
                                  if random.random() < domfrac]}
                        for s in range(strings)]}


def test_config_barf():
    cfg = sim_config('sim')
    dumped = json.dumps(cfg)
    compressed = bz2.compress(dumped)
    hexed = binascii.hexlify(compressed)
    unhexed = binascii.unhexlify(hexed)
    uncompressed = bz2.decompress(unhexed)
    undumped = json.loads(uncompressed)
    assert undumped == cfg
    print len(dumped), len(compressed), len(hexed)