Sejings

Sejings is meant to be a quick and simple tool to rapidly integrate project sejings. The problem I've seen many times with libraries is that it's often hard to have a way for the user to change aspects or functionality deep in a library. Libraries can often have arguments for a function that are unavailable to a developer because they're hidden behind three function calls.

One attempt that evolves as a codebase grows is to pass along keyword arguments, which leads to documentation and maintainability issues. This is evident by the popularity of **kwargs in many data science libraries. The only solution I've seen that tries to solve this problem is to work with dictionaries similar to matplotlib's rcParams. Yet this leads to a tedious and time consuming developer experience by forcing if xxx is None statements at the top of method calls.

To show you my solution let's first start of with a rcParams style function example:

def add(*nums, cache=None, cache_path=None):
    
    if cache is None:
        cache = dict_settings['cache']
    if cache_path is None:
        cache_path = dict_settings['cache.path']
    
    result = sum(nums)
    
    if cache:
        save_to_cache(result, cache_path)
    
    return result

Obviously this would be computationally the fastest way to accomplish sejings but leaves a lot of work to the developer. That's why this project is meant to be friendly to developers consuming packages and to encourage rapid development.

Usage

Import sejings and create the sejings you need:

from sejings import sejings

sejings.cache = True
sejings.cache.path = '/some/dir/path'

To evaluate sejings passed into a function as an argument use the @extract_sejings decorator. This will evaluate all sejings in the function definition and in the arguments being passed into the function:

from sejings import extract_sejings

@extract_sejings
def add(*nums, cache=sejings.cache, cache_path=sejings.cache.path):
    result = sum(nums)
    
    if cache: # True
        save_to_cache(result, cache_path) #'/some/dir/path'
    
    return result

A branch is also evaluated when an endpoint is called.

assert sejings.cache()
assert sejings.cache.path() == '/some/dir/path'

def add(*nums, cache=sejings.cache, cache_path=sejings.cache.path):
    result = sum(nums)
    
    if cache(): # True
        save_to_cache(result, cache_path()) #'/some/dir/path'
    
    return result

In some cases defining arguments as a Sejings object may be desired. This is accomplished by adding the argument name to the @extract_sejings arguments.

@extract_sejings('cache')
def add(*nums, cache=sejings.cache):
    result = sum(nums)
    
    if cache(): # True
        save_to_cache(result, cache.path()) #'/some/dir/path'
    
    return result

Integrating Into Your Project

I chose sejings as a close to settings word that wouldn't conflict with a potentially very common word in programming. Therefore you are encouraged to import Sejings as a name you see fit. To use matplotlibs rcParams again as an example we could look at an fictional file settings.py:

from pathlib import Path
from sejings import (
    extract_sejings as extract_settings,
    Sejings
)

my_settings = Sejings(name='rcParams')

my_settings.cache = True
my_settings.cache.path = Path(r'/some/dir/path')

@extract_settings
def add(*nums, cache=my_settings.cache, cache_path=my_settings.cache.path):
    result = sum(nums)
    
    if cache: # True
        save_to_cache(result, cache_path) #'/some/dir/path'
    
    return result

@TODO

• Context manager
• getitem, setitem