typesys

Typesys

Intro

Typesys is a module that is meant to make it more easy to manage types. It contains four decorators; type_hints, type_corrector, return_type and returns.

The type_hints decorator lets the user specify what the arguements of a functions should be. If arguments of another type than specified in the decorator are passed in, a TypeError will be raised.

The type_corrector decorator lets the user specify what types the arguments to a function should have, but not necessarily are passed in as, which means it allows some margin of error. It's not 100% safe to use as it still might result in a ValueError or TypeError if the user isn't careful enough.

The return_type decorator lets the user specify what type the decoratred function should return. A TypeError will be raised if the function tries to return a value of another type.

The returns decorator lets the user specify what type a function's return type has to be (if possible to return this a value of this type).

The motivation behing this module was to abstract some of the type checking and type casting to a higher level. I wanted to find a way that makes it easier for the programmer to see what types the arguments should be, what a type a function should return or must return, and at the same time allow some margin of error (in case of type_corrector). I'm not sure whether this is a good idea or not, or if it's a good approach. It was mostly developed for fun while playing around with decorators.

Installation

pip install typesys

Build from source

clone and run python setup.py install

Usage

First import the decorators from the typesys module:

.. code:: python

from typesys import type_hints, type_corrector, return_type, returns

Then you are ready to start decorating your functions.

type_hints ''''''''''

Decorate your functions with the types you want the arguments to be, as shown in the examples below

.. code:: python

# a and b must be integers
@type_hints(int, int)
def add(a, b):
    return a+b


# also work with default arguments
@type_hints(int, float)
def add(a, b=0.0):
    return a+b


# accepts both integers and floating 
# point numbers as arguments
@type_hints(int, float)
def mult(*numbers):
    result = 1
    for num in numbers:
        result *= num
    return result


# Only accept integer arguments
@type_hints(int)
def mult(**kwargs):
    x = kwargs.get('x', 1)
    y = kwargs.get('y', 1)
    z = kwargs.get('z', 1)
    return x * y * z

type_corrector ''''''''''''''

Decorate your functions with the types you want the arguments to be treated as, but not necessarily are passed in as, as shown in the examples below.

.. code:: python

@type_corrector(int, int)
def add(x,y):
    return x+y


@type_corrector(float, float)
def div(x,y):
    return x/y
   

A call to add(1,'2') will cast '2' to an int, since that is what we specified as the type of the second paramater in the decorator. We can also call div as div('10', '3'), and div will return 3.3333333333333335 as expected.

This decorator also works with *args and **kwargs

.. code:: python

@type_corrector(int)
def mult(*numbers):
    result = 1
    for num in numbers:
        result *= num
    return result


@type_corrector(int)
def mult(**kwargs):
    x = kwargs.get('x', 1)
    y = kwargs.get('y', 1)
    z = kwargs.get('z', 1)
    return x * y * z   

This allows us to call the functions like this:

• mult(2, '3', '4')
• kw_mult(x=2, y='3', z='4')

When looking at the function definitions of add, mult and kw_mult we can easily see that the arguments are supposed to be integers. By decorating the functions like this it should also be a clear hint what types we want the arguments to be passed in as, even though it allows some margin of error.

return_type '''''''''''

Decorate your functions with the type or types you want your functions to return, as shown in the examples below.

.. code:: python

# accepts both integers, floatint point numbers 
# and complex numbers to be returned
@return_type(int, float, complex)
def add(x,y):
    return x+y


# only accept integers to be returned
@return_type(int)
def strict_add(x,y):
    return x+y

The same applies for functions defined with *args and/or **kwargs

.. code:: python

# accepts both integers and floating point numbers
# to be returned
@return_type(int, float)
def mult(*numbers):
    res = 1
    for number in numbers:
        res *= number
    return res


# only accepts integers to be returned
@return_type(int)
def stric_kw_mult(**kwargs):
    x = kwargs.get('x', 1)
    y = kwargs.get('y', 1)
    z = kwargs.get('z', 1)
    return x * y * z

returns '''''''

Decorate your functions with the type your funcitons must return, as long as it's possible.

.. code:: python

# returns x+y as a string
@returns(str)
def add(x,y):
    return x+y

A call to add(1,2) will return the number 3 as a string.

Known issues

• When calling help on a decorated function the parameters are not shown correctly, instead it will just say (*args, **kwargs). Thanks to the functools.wraps decorator the docstring of a decorated function will still be shown correctly.
• When using the inspect module to get the argument specification with inspect.getargspec or getting the source code from inspect.getsourcelines it will fail and show the wrapped function instead.

Bugs, problems and new features

If you find any bugs, have any problems, or maybe you just want to request a new feature, then use the issue tracker <https://github.com/fredgj/typesys/issues>_.