Skip to main content

A common API for multiple serialization formats with support for custom classes

Reason this release was yanked:

importing does not work in all python versions

Project description

Latest Version License Python Versions CI LINTER Coverage

Serialize: A common Python API for multiple serialization formats

There are multiple serialization formats out there ...
    ... and great packages to use them.

But they all have a different API and switching among them is not so simple as it should be. Serialize helps you to do it, including dealing with custom classes. Let’s dump a dict using the pickle format:

>>> from serialize import dumps, loads
>>> dumps(dict(answer=42), fmt='pickle')
b'\x80\x03}q\x00X\x06\x00\x00\x00answerq\x01K*s.'
>>> loads(_, fmt='pickle')
{'answer': 42}

And here comes the cool thing, you can just change the serialization format without having to learn a new API. Let’s now dump it using msgpack:

>>> dumps(dict(answer=42), fmt='msgpack')
b'\x81\xa6answer*'
>>> loads(_, fmt='msgpack')
{'answer': 42}

Serialize currently support 8 different formats: bson, dill, json (builtin or with simplejson package), msgpack, phpserialize, pickle, serpent and yaml. Serialize does not implement these formats but rather relies on established, well tested packages. If they are installed, serialize will use them.

** Serialize allows you to use them all with the same API! **

You can also use the dump and load to write directly to file-like object:

>>> from serialize import dump, load
>>> with open('output.yaml', 'wb') as fp:
...     dump(dict(answer=42), fp, fmt='yaml')
>>> with open('output.yaml', 'rb') as fp:
...     load(fp, fmt='yaml')
{'answer': 42}

or use directly the filename and the format will be inferred:

>>> dump(dict(answer=42), 'output.yaml')
>>> load('output.yaml')
{'answer': 42}

A very common case is to dump and load objects from custom classes such as:

>>> class User:
...     def __init__(self, name, age):
...         self.name = name
...         self.age = age
...
>>> john = User('John Smith', 27)

But some serialization packages do not support this important feature and the rest usually have very different API between them. Serialize provides you a common, simple interface for this. You just need to define a function that is able to convert the object to an instance of a builtin type and the converse:

>>> from serialize import register_class
>>> def user_to_builtin(u):
...     return (u.name, u.age)
...
>>> def user_from_builtin(c):
...     return User(c[0], c[1])
...

>>> register_class(User, user_to_builtin, user_from_builtin)

And that’s all. You can then use it directly without any hassle:

>>> dumps(john, fmt='bson')
b"y\x00\x00\x00\x03__bson_follow__\x00c\x00\x00\x00\x04__dumped_obj__
\x00\x1e\x00\x00\x00\x020\x00\x0b\x00\x00\x00John Smith\x00\x101\x00
\x1b\x00\x00\x00\x00\x02__class_name__\x00\x1c\x00\x00\x00<class '__m
ain__.Username'>\x00\x00\x00"
>>> v = loads(_, fmt='bson')
>>> v.name
'John Smith'
>>> v.age
27

Enjoy!

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

Serialize-0.2.tar.gz (14.6 kB view hashes)

Uploaded Source

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page