lychee.json is an extreamly fast json encoder / decoder package for python. Encoding and decoding output fully compatible with python.json package.

• Extreamly fast (see benchmark results in '/test/benchmark' directory)
• Fully compatible output with Python json package
• Builtin object serialization method __json__ (see below)
• Strict JSON (RFC 4627) expected: Infinity, NaN (JavaScript compatible infinity and not a number symbols)
• UTF-8 encoding / decoding support
• Accurate float encoding / decoding

You can install via pip:

pip install lychee.json

or setuptools:

easy_install lychee.json

Python 2.6-2.7 developer headers and any c or c++ compatible compiler, but only tested with gcc under linux.

Very similar that python.json, let's see some example

Json data to python

code:

from lychee import json

>>> json.loads('"Hello World"')
"Hello World"

Python object to json data

code:

from lychee import json

>>> json.dumps("Hello World")
'"Hello World"'

class Point:
   def __json__(self):
      return {"x":1, "y":2}

>>> json.dumps(Point())
'{"x":1,"y":2}'

• lychee.json.loads (str s [, callable object_hook [, callable parse_float]])

  • s: json string

  • object_hook: this function call on every object that decoded example:

    >>> from lychee import json
    >>> def hook(dict_):
    ...     if "__complex__" in dict_:
    ...         return complex(dict_["real"], dict_["imag"])
    ...
    >>> json.loads('{"__complex__":true, "real":1, "imag":2}',
    >>>     object_hook=hook)
    (1+2j)
    

  • parse_float: call this function when float parsed from json data. If you are need to use decimal.Decimal for float decoding provide this paramater with decimal.Decimal eg.:

    >>> from lychee import json
    >>> from decimal import Decimal
    >>> json.loads("1.2", parse_float=Decimal)
    Decimal('1.2')
    

• lychee.json.dumps (object obj [, callable default [, str tojson [, bool ensure_ascii [, bool dump_date [, int utc_offset]]]]])

  • obj: python object

  • default: default function for non basic python type serialization:

    >>> from lychee import json
    >>> def default_func(o):
    ...     if isinstance(o, complex):
    ...         return {"__complex__":True, "real":1, "imag":2}
    ...
    >>> json.dumps(1 + 2j, default=default_func)
    '{"__complex__":true,"real":1,"imag":2}'
    

  • tojson: method name to use to convert object to json serializable type when need to serialize non supported object, default value is __json__:

    >>> from lychee import json
    >>> class Point(object):
    ...     def __init__(self, x, y):
    ...         self.x = x
    ...         self.y = y
    ...     def __json__(self):
    ...         return {"x":self.x, "y":self.y}
    ...
    >>> json.dumps(Point(10, 20))
    '{"x":10,"y":20}'
    

  • ensure_ascii: ancode all characters are ascii compatible format

  • dump_date: default value is True, if True automatically convert date / datetime objects to ISO 8601 format.

  • utc_offset: If datetime object has no tzinfo this value is use insted. Possible values is:

    1. 'local' string: use machine local utc offset
    2. timedelta
    3. int: utc offset in seconds

• lychee.json.JsonError: base exception class
• lychee.json.JsonEncodeError: exception class for encoding errors
• lychee.json.JsonDecodeError: exception class for decoding errors