Skip to main content

The classic ``compose``, with all the Pythonic features.

Project description

The classic compose, with all the Pythonic features.

This compose follows the lead of functools.partial and returns callable compose objects which:

  • have a regular and unambiguous repr,

  • retain correct signature introspection,

  • allow introspection of the composed callables,

  • can be type-checked,

  • can be weakly referenced,

  • can have attributes,

  • will merge when nested, and

  • can be pickled (if all composed callables can be pickled).

This compose also throws a TypeError when called with no arguments or with any non-callable arguments.

For async/await support, the right behavior of function composition depends on what you are doing, so variants of compose are included for those cases.

Versioning

This library’s version numbers follow the SemVer 2.0.0 specification.

Installation

pip install compose

Usage

Basics

Import compose:

from compose import compose

All the usual function composition you know and love:

>>> def double(x):
...     return x * 2
...
>>> def increment(x):
...     return x + 1
...
>>> double_then_increment = compose(increment, double)
>>> double_then_increment(1)
3

Of course any number of functions can be composed:

>>> def double(x):
...     return x * 2
...
>>> times_eight = compose(douple, double, double)
>>> times_16 = compose(double, double, double, double)

We still get the correct signature introspection:

>>> def f(a, b, c=0, **kwargs):
...     pass
...
>>> def g(x):
...     pass
...
>>> g_of_f = compose(g, f)
>>> import inspect
>>> inspect.signature(g_of_f)
<Signature (a, b, c=0, **kwargs)>

And we can inspect all the composed callables:

>>> g_of_f.functions  # in order of execution:
(<function f at 0x4048e6f0>, <function g at 0x405228e8>)

compose instances flatten when nested:

>>> times_eight_times_two = compose(double, times_eight)
>>> times_eight_times_two.functions == times_16.functions
True

When programmatically inspecting arbitrary callables, we can check if we are looking at a compose instance:

>>> isinstance(g_of_f, compose)
True

async/await

We can compose async code by using acompose or sacompose (they are mostly the same):

>>> import asyncio
>>> from compose import acompose
>>>
>>> async def get_data():
...     # pretend this data is fetched from some async API
...     await asyncio.sleep(0)
...     return 42
...
>>> get_and_double_data = acompose(double, get_data)
>>> asyncio.run(get_and_double_data())
84

acompose and sacompose can compose any number of async and regular functions, in any order:

>>> async def async_double(x):
...     await asyncio.sleep(0)
...     return x * 2
...
>>> async_times_16 = acompose(async_double, double, async_double, double)
>>> asyncio.run(async_times_16(1))
16

sacompose provides a different way of handling a corner case that arises when composing functions that we get from users or other code: what if every function we receive to compose is regular, not async, but we want to support async?

  • acompose handles that case by returning an awaitable anyway - so we can just write simple code that calls await in all cases. This is the best choice for function composition that we know will be used in async code.

  • sacompose handles that case by returning a callable which will sometimes behave in an async way, by returning an awaitable only if any of the composed functions return an awaitable. This is needed to simplify reusable helper code that can’t know if it is composing for regular or async code:

    >>> from compose import sacompose
    >>>
    >>> regular_times_4 = sacompose(double, double)
    >>> awaitable_times_4 = sacompose(double, async_double)
    >>>
    >>> # Right:
    >>> regular_times_4(1) == 4
    >>> await awaitable_times_4(1) == 4
    >>>
    >>> # Wrong (TypeError from the `==`, and coroutine not awaited):
    >>> awaitable_times_4(1) == 4
    >>> # Wrong (TypeError from the `await`):
    >>> await regular_times_4(1) == 4

acompose and sacompose instances flatten when nested:

>>> acompose(f, acompose(f, f)).functions == (f, f, f)
True
>>> acompose(sacompose(f, f), f).functions == (f, f, f)
True
>>> sacompose(acompose(f, f), f).functions == (f, f, f)
True
>>> sacompose(f, sacompose(f, f)).functions == (f, f, f)
True

But compose instances don’t flatten when nested into acompose and sacompose, and vice versa:

>>> acompose(g_of_f).functions
(compose(<function f at 0x4048e6f0>, <function g at 0x405228e8>),)
>>> sacompose(g_of_f).functions
(compose(<function f at 0x4048e6f0>, <function g at 0x405228e8>),)
>>> compose(acompose(g, f)).functions
(acompose(<function f at 0x4048e6f0>, <function g at 0x405228e8>),)
>>> compose(sacompose(g, f)).functions
(sacompose(<function f at 0x4048e6f0>, <function g at 0x405228e8>),)

compose, acompose, and sacompose instances are all distinct types:

>>> isinstance(g_of_f, compose)
True
>>> isinstance(g_of_f, (acompose, sacompose))
False
>>> isinstance(async_times_16, acompose)
True
>>> isinstance(async_times_16, (compose, sacompose))
False
>>> isinstance(awaitable_times_4, sacompose)
True
>>> isinstance(awaitable_times_4, (compose, acompose))
False

Recipes

  • If you want composing zero functions to be the identity function:

    from functools import partial
    
    def identity(x):
        return x
    
    icompose = partial(compose, identity)
  • To compose arguments in reverse order:

    def rcompose(*functions):
        return compose(*reversed(functions))
  • When you need composition to return a normal function:

    def fcompose(*functions):
        composed = compose(*functions)
        return lambda *args, **kwargs: composed(*args, **kwargs)

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

compose-1.4.2.tar.gz (6.0 kB view hashes)

Uploaded Source

Built Distributions

compose-1.4.2-py38-none-any.whl (5.3 kB view hashes)

Uploaded Python 3.8

compose-1.4.2-py35-none-any.whl (5.3 kB view hashes)

Uploaded Python 3.5

compose-1.4.2-py2.py30-none-any.whl (5.0 kB view hashes)

Uploaded Python 2 Python 3.0

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