nt2dc

There should be one—and preferably only one—obvious way to do it. The zen of Python

Python objects

Currently, python offers different ways of modelling a dead-simple objects. The most complex way is simply a plain old python object:

class Complex:
    def __init__(self, real: float = 0.0, imaginary: float = 0.0) -> None:
        self.__real = real
        self.__imaginary = imaginary

    def __set_real(self, value: float) -> None:
        self.__real = value

    def __get_real(self) -> float:
        return self.__real

    real = property(__get_real, __set_real)

    def __set_imaginary(self, value: float) -> None:
        self.__imaginary = value

    def __get_imaginary(self) -> float:
        return self.__imaginary

    imaginary = property(__get_imaginary, __set_imaginary)

    def __repr__(self) -> str:
        return "{} + {}i".format(self.__real, self.__imaginary)

Dataclasses

With newer versions of python 3, the dataclasses have been introduced, that can be created in two ways:

Complex = make_dataclass('Complex', [('real', float), ('imaginary', float)])

or

@dataclass
class Complex:
    real: float = field()
    imaginary: float = field()

Dataclasses are a great way of handling python data in a dead-simple way. Constructor, __repr__ function, getters, setters and introspection is possible and make it more easy to create a serialization and deserialization engine. They hence can easily be combined with multiple frameworks like sqlalchemy, marshmallow and entire web frameworks like bottle, flask etc. to create rest-ful APIs.

Named tuples

On the other hand, there are legacy modelling ways like named tuples, that can be create using two ways:

Complex = namedtuple('Complex', ['real', 'imaginary'])

or

class Complex(NamedTuple):
    real: float
    imaginary: float

Please note, that the latter has type hints while the former does not support type hinting yet.

NamedTuples are like tuples: A read-only combination of multiple values, just like regular tuples, but each value has a unique name.

Conversion

What if, one API provides or requests NamedTuples but you need a dataclass for further processing or vice versa?

This package provides a set of functions for the conversion from one notation (NamedTuple) to the other (Dataclass) and back.

Please note, the type hinted variant should be used.

Licensing

This library is published under BSD-3-Clause license.

Versioning

This library follows semantic versioning 2.0. Any breaking change will produce a new major release. Versions below 1.0 are considered to have a unstable interface.