测试Json压缩方式的效果

.gitignore

*.gz
*.json
.*.swp

README.md

What's this

Comparing json compression methods.

Run

make to generate two json files containing the same set of data, but one is compressed using google+-style minification. Those two files are then gzipped.

Result

There is a 22% difference in gzipped sizes.

gen.py

import random
from itertools import chain
import json

ATTR1 = 'some_very_long_name'
ATTR2 = 'another_very_long_age'
ATTR3 = 'we_also_got_some_qq_numbers'
ATTR4 = 'and_the_users_phone'

def mk_object(uniq):
    return {
        ATTR1: 'name_%d' % uniq,
        ATTR2: uniq,
        ATTR3: [mk_number() for _ in xrange(random.randint(0, 10))],
        ATTR4: mk_number()
    }

def mk_number():
    if random.random() > 0.5:
        return random.randint(10000000, 99999999)
    return None

def to_array(o):
    return [o[ATTR1], o[ATTR2], o[ATTR3], o[ATTR4]]

objects = [mk_object(i) for i in xrange(10000)]

# Flattened and stripped
obj_array = list(chain(*[to_array(o) for o in objects]))

with open('objects.json', 'w') as f:
    json.dump(objects, f)

jstr = json.dumps(obj_array).replace(' ', '').replace('null', '')
with open('obj_array.json', 'w') as f:
    f.write(jstr)

Makefile

all : obj_array.json.gz objects.json.gz

%.json: gen.py
	python gen.py

%.json.gz : %.json
	gzip -c $< > $@