Release v1.0.2 (What’s new?).

Documentation Status https://github.com/MacHu-GWU/superjson-project/workflows/CI/badge.svg https://codecov.io/gh/MacHu-GWU/superjson-project/branch/master/graph/badge.svg https://img.shields.io/pypi/v/superjson.svg https://img.shields.io/pypi/l/superjson.svg https://img.shields.io/pypi/pyversions/superjson.svg PyPI - Downloadshttps://img.shields.io/badge/STAR_Me_on_GitHub!--None.svg?style=social
https://img.shields.io/badge/Link-Document-orange.svg https://img.shields.io/badge/Link-API-blue.svg https://img.shields.io/badge/Link-Source_Code-blue.svg https://img.shields.io/badge/Link-Install-blue.svg https://img.shields.io/badge/Link-GitHub-blue.svg https://img.shields.io/badge/Link-Submit_Issue-blue.svg https://img.shields.io/badge/Link-Request_Feature-blue.svg https://img.shields.io/badge/Link-Download-blue.svg

Welcome to superjson Documentation

Features:

  1. Support for date, datetime, set, OrderedDict, deque, that the original json module can not serialize.

  2. Easy to extend to support any custom data type, SEE HOW.

  3. Allow // double slash started comments, # Pound key started comments style comment in json file (Good for human/machine readable config file).

  4. Use .gz for file extension, data will be automatically compressed.

  5. Atomic Write is ensured, operation of overwriting an existing json file is atomic.

Click HERE for full documentation.

Install

superjson is released on PyPI, so all you need is:

$ pip install superjson

To upgrade to latest version:

$ pip install --upgrade superjson

Extend

Custom serializer for your Python Object is easy.

  1. suppose you have a class User: ... object defined in model.py

>>> # it doesn't matter that how's the instance been constructed
>>> class User(object):
...     def __init__(self, id, name):
...         self.id = id
...         self.name = name
...
...     def __repr__(self):
...         return "User(id=%r, name=%r)" % (self.id, self.name)
  1. Make your MySuperJson inherit from superjson.SuperJson.

# Add support to xxx.model.User
>>> from superjson import SuperJson
>>> class MySuperJson(SuperJson): pass
  1. Get the full class name for User.

>>> from superjson import get_class_name
>>> user_class_name = get_class_name(User(id=1, name="Alice")) # a dummy user

# or you can just do
>>> user_class_name = "xxx.model.User"
  1. Define encode method and decode method in this name convention dump_xxx, load_xxx. You just need to manually transform the instance to a SuperJson serializable object, a combination of dict, list, tuple, set, str, integer, float, datetime, bytes, etc. And just construct the instance from the SuperJson serializable object we just defined. In the User example, we dump a user to {"id": user.id, "name": user.name}, and load a user from User(**dict_data).

# Add support to xxx.model.User
>>> from xxx.model import User
>>> from superjson import SuperJson, get_class_name

>>> user_class_name = get_class_name(User(id=1, name="Alice")) # a dummy user

>>> class MySuperJson(SuperJson):
...     # dumper method has three input argument, self, obj and class_name
...     def dump_User(self, obj, class_name=user_class_name):
...         return {"$" + class_name: {"id": obj.id, "name": obj.name}}
...
...     # loader method has three input argument, self, dct and class_name
...     def load_User(self, dct, class_name=user_class_name):
...         return User(**dct["$" + class_name])
  1. The final code looks like:

# Add support to xxx.model.User
>>> from xxx.model import User
>>> from superjson import SuperJson, get_class_name

>>> user_class_name = get_class_name(User(id=1, name="Alice")) # a dummy user

>>> class MySuperJson(SuperJson):
...     # dumper method has three input argument, self, obj and class_name
...     def dump_User(self, obj, class_name=user_class_name):
...         return {"$" + class_name: {"id": obj.id, "name": obj.name}}
...
...     # loader method has three input argument, self, dct and class_name
...     def load_User(self, dct, class_name=user_class_name):
...         return User(**dct["$" + class_name])

>>> json = MySuperJson()

>>> a_complex_data = {"users": [User(id=1, name="Alice"), User(id=2, name="Bob")]}
>>> s = json.dumps(p)
>>> s
{"users": [{"$xxx.model.User": {"id": 1, "name": "Alice"}}, {"$xxx.model.User": {"id": 2, "name": "Bob"}}]}
>>> data = json.loads(s)
>>> data
{'users': [User(id=1, name='Alice'), User(id=2, name='Bob')]}

Comment

You can add comments to your json file, and superjson can still read it!

>>> s= \
"""
{
    # This is host
    "host": "127.0.0.1",
    "port": 8080 // This is port
}
"""
>>> config = json.loads(s, ignore_comments=True)
>>> config
{"host": "127.0.0.1", "port": 8080}

Compression

Compress your json file is easy.

>>> import sys
>>> data = {str(i): i for i in range(1000)}

# Compress when dump to string
>>> text = json.dumps(data, compress=True)

# Decompress when load from compressed string
>>> data1 = json.loads(text, decompress=True)

# Auto compress when dump to file
>>> json.dump(data, "data.gz") # "data.json" will not been compressed

# Auto decompress when load from file
>>> json.load("data.gz")

# compare
>>> text1 = json.dumps(data)
>>> text2 = json.dumps(data, compress=True)
>>> sys.getsizeof(text1)
11829

>>> sys.getsizeof(text2)
5809

Advance file I/O utility method

If your program is interrupted while writing, you got an incomplete file, and you also lose the original file! To solve this issue, json.dump(data, abspath, overwrite=True) method first write json to a temporary file, then rename to what you expect, and silently overwrite old one. This can guarantee atomic write operation.

>>> data = dict(a=1, b=2, c=3)
# it first write to "data.gz.tmp", when it's done, overwrite the
# original "data.gz" file
>>> json.dump(data, abspath="data.gz", overwrite=True)

More options for dump, load can be found HERE.

About the Author

(\ (\
( -.-)o    I am a lovely Rabbit!
o_(")(")

Sanhe Hu is a very active Python Developer Since 2010. Research area includes Machine Learning, Big Data Infrastructure, Block Chain, Business Intelligent, AWS, Distributive System. Love photography, outdoor, arts, game, and also the best Python.

API Document