pyjson.py

from __future__ import print_function
import re

ws_match = re.compile(r'[ \t\n\r]*').match
number_match = re.compile(
    r'(-?(?:0|[1-9]\d*))(\.\d+)?([eE][-+]?\d+)?[ \t\n\r]*',
    (re.VERBOSE | re.MULTILINE | re.DOTALL)).match
nan = float('nan')
inf = float('inf')

def decode(s, pos=0):

    def parse_any(pos):
        c = s[pos]

        if c == '"':
            return parse_string(pos)

        if c == '{':
            return parse_dict(ws_match(s, pos + 1).end())

        if c == '[':
            return parse_list(ws_match(s, pos + 1).end())

        if c in 'ntfNI-':
            # constants
            if s.startswith('null', pos):
                return None, ws_match(s, pos + 4).end()
            if s.startswith('true', pos):
                return True, ws_match(s, pos + 4).end()
            if s.startswith('false', pos):
                return False, ws_match(s, pos + 5).end()

            # numbers, but constants
            if s.startswith('NaN', pos):
                return nan, ws_match(s, pos + 5).end()
            if s.startswith('Infinity', pos):
                return inf, ws_match(s, pos + 8).end()
            if s.startswith('-Infinity', pos):
                return -inf, ws_match(s, pos + 9).end()

        num = number_match(s, pos)
        if num is not None:
            integer, frac, exp = num.groups()
            if frac or exp:
                res = float(integer + (frac or '') + (exp or ''))
            else:
                res = int(integer)
            return res, num.end()

        raise ValueError('Unknown token.')

    def parse_dict(pos):
        if s[pos] == '}':
            return {}, pos + 1

        d = {}
        while True:
            if s[pos] != '"':
                raise ValueError('Objects keys should be strings.')
            key, pos = parse_string(pos)
            if s[pos] != ':':
                raise ValueError('Not colon.')
            value, pos = parse_any(ws_match(s, pos + 1).end())
            d[key] = value

            if s[pos] != ',':
                break
            pos = ws_match(s, pos + 1).end()

        if s[pos] != '}':
            raise ValueError('Object not closed.')
        return d, ws_match(s, pos + 1).end()

    def parse_list(pos):
        if s[pos] == ']':
            return [], pos + 1

        l = []
        _append = l.append
        while True:
            value, pos = parse_any(pos)
            l.append(value)
            if s[pos] != ',':
                break
            pos = ws_match(s, pos + 1).end()

        if s[pos] != ']':
            raise ValueError('Array not closed.')
        return l, ws_match(s, pos + 1).end()

    def parse_string(pos):
        first = pos + 1
        while True:
            pos = s.index('"', pos + 1)
            if s[pos - 1] != '\\':
                break
        return s[first:pos], ws_match(s, pos + 1).end()

    return parse_any(ws_match(s, pos).end())[0]

if __name__ == '__main__':
    test = """
{
    "json\\\"on\\\" on"  : "yes"  ,
    "om-nom-nom":null,"empty obj":{},
    "arr": [null, true , false, [], null, true , false, []],
    "numbers = ": [2, 3,5,66e4],
    "longobj": {
        "key1": "testdata",
        "key2": "testdata",
        "key3": "testdata",
        "key4": "testdata",
        "key5": "testdata",
        "key6": "testdata",
        "key7": "testdata",
        "key8": "testdata",
        "key9": "testdata",
        "key10": "testdata",
        "key11": "testdata",
        "key12": "testdata",
        "key13": "testdata",
        "key14": "testdata"
    }
}
"""
    from timeit import repeat
    number = 1000
    # test = open("go/src/pkg/encoding/json/testdata/code.json").read()

    print()
    print('>>> pyjson')
    print(decode(test))
    print(repeat(lambda: decode(test), number=number, repeat=3))

    from json import loads
    print()
    print('>>> json')
    print(loads(test))
    print(repeat(lambda: loads(test), number=number, repeat=3))

    try:
        from demjson import decode
    except ImportError:
        pass
    else:
        print()
        print('>>> demjson')
        print(decode(test))
        print(repeat(lambda: decode(test), number=number, repeat=3))

    try:
        from ujson import loads
    except ImportError:
        pass
    else:
        print()
        print('>>> ujson')
        print(loads(test))
        print(repeat(lambda: loads(test), number=number, repeat=3))