Configs in Python made smooth and simple

Betterconf (better config) is a Python library for project configuration management. It allows you define your config like a regular Python class.

Features:

• Easy to hack.
• Less boilerplate.
• Minimal code to do big things.
• Zero dependencies. Only vanilla Python >=3.11

Betterconf is heavily typed, so your editors and typecheckers will be happy with it.

It's not that huge as Pydantic, so takes like 5 minutes to dive into.

Betterconf is highly customizable, so you can do anything with it.

Installation

I recommend you to use poetry:

poetry add betterconf

However, you can use pip:

pip install betterconf

How to?

Betterconf is very intuitive, so you don't need special knowledge to use it. We'll cover the basics.

The most simple config will look like this:

from betterconf import betterconf

@betterconf
class Config:
    LOGIN: str
    PASSWORD: str

cfg = Config()
print(config.LOGIN)

Let's dive into what it does. By default, betterconf uses EnvironmentProvider, getting values with os.getenv, so you can run this example with LOGIN=vasya PASSWORD=admin python config.py and simply get your login.

There is a very usable thing in our fancy-web-world, called .envs. Betterconf, since 4.5.0 automatically supports them out-of-the-box! See it:

from betterconf import betterconf, DotenvProvider

# here betterconf gets values from `.env` file, you can change name or the path passing it as the first
# argument to provider.
# also, auto_load lets you not to write `provider.load_into_env` (loading variables to os.environ)
# or `provider.load_into_provider()` (loading them into the provider's inner storage).
@betterconf(provider=DotenvProvider(auto_load=True))
class Config:
    val1: int
    val2: str
    name: str

cfg = Config()
print(cfg.val1, cfg.val2, cfg.name)

Your .env then looks like this:

name="John Wicked"
val1=12
val2="testing value"

But what if you need a different provider? Betterconf lets you set providers as for config itself and for each field respectively.

import json
from betterconf import Alias
from betterconf import betterconf, field
from betterconf import JSONProvider, AbstractProvider

sample_json_config = json.dumps({"field": {"from": {"json": 123}}})

# our provider, that just gives the name of field back
class NameProvider(AbstractProvider):
    def get(self, name: str) -> str:
        return name

@betterconf(provider=NameProvider())
class Config:
    # value will be got from NameProvider and will simply be `my_fancy_name`
    my_fancy_name: str
    # here we get value from JSONProvider; the default nested_access is '.'
    field_from_json: Alias[int, "field::from::json"] = field(provider=JSONProvider(sample_json_config, nested_access="::"))

Betterconf casts primitive types itself, they include list, float, str, int. But if you need specific caster, say for complex object, you can write your own.

from betterconf import betterconf
from betterconf import AbstractCaster, field

class DashesToDotsCaster(AbstractCaster):
    def cast(self, val: str) -> str:
        return val.replace("_", ".")

@betterconf
class Config:
    simple_int: int
    value: str = field(caster=DashesToDotsCaster())

print(Config(value="privet_mir", simple_int="55666").value)

Subconfigs and referencing one field in another declaration is also available. Check out examples folder.

License

This project is licensed under MIT License.

See LICENSE for details.

Made with :heart: by prostomarkeloff and our contributors.