Nano settings

Creates simple python config from environment variables. Smaller analog of pydantic-settings.

In comparison with pydantic-settings

Pros:

• Easier to understand and support.
• Relies only on standard library.
• Uses dataclasses, not polluting your code with external types (if you're using approaches similar to clean architecture).
• Almost no configuration.

Cons:

• No validation, you have to do it yourself.
• Almost no configuration.

Installation

pip install nano-settings

Basic usage

from dataclasses import dataclass

import nano_settings as ns


@dataclass
class DbSetup(ns.BaseConfig):
    max_sessions: int
    autocommit: bool = True


@dataclass
class Database(ns.BaseConfig):
    url: str
    timeout: int
    setup: DbSetup


# export MY_VAR__URL=https://site.com
# export MY_VAR__TIMEOUT=10
# export MY_VAR__SETUP__MAX_SESSIONS=2
config = ns.from_env(Database, env_prefix='my_var')
print(config)
# Database(timeout=10, url='https://site.com', setup=DbSetup(max_sessions=2, autocommit=True))

Supported variants

Straightforward - when your annotations are callable

from dataclasses import dataclass

import nano_settings as ns


@dataclass
class Config(ns.BaseConfig):
    variable: int
    # equivalent of
    # variable = int(os.getenv('VARIABLE'))

Annotated - when your annotations are complex and cannot be called

from dataclasses import dataclass
from typing import Annotated
import json

import nano_settings as ns


@dataclass
class Config(ns.BaseConfig):
    variable: Annotated[list[int], lambda x: x['key'], json.loads]
    # equivalent of
    # variable = json.loads(os.getenv('VARIABLE'))['key']

Nested - when your annotations are models

Example is listed in Basic usage

Aliases - when you want to get value by different name

Normal - try default and then alternatives

from dataclasses import dataclass
from typing import Annotated

import nano_settings as ns


@dataclass
class Config(ns.BaseConfig):
    variable: Annotated[str, ns.EnvAlias('OTHER')]
    # will try to get `VARIABLE` and then `OTHER`

Strict - try only alternatives

from dataclasses import dataclass
from typing import Annotated

import nano_settings as ns


@dataclass
class Config(ns.BaseConfig):
    variable: Annotated[str, ns.EnvAliasStrict('OTHER')]
    # will only try to get `OTHER`

Nullable - if null is also a valid value

from dataclasses import dataclass
from typing import Annotated

import nano_settings as ns


@dataclass
class Config(ns.BaseConfig):
    variable: Annotated[int | None, ns.Nullable(int)]

Choices - if you have a set of valid variants

from dataclasses import dataclass
from typing import Annotated

import nano_settings as ns


@dataclass
class Config(ns.BaseConfig):
  variable: Annotated[str, ns.Choices('one', 'two')]

Interval - when you have minimum and maximum (including)

from dataclasses import dataclass
from typing import Annotated

import nano_settings as ns


@dataclass
class Config(ns.BaseConfig):
  variable_1: Annotated[int, ns.Interval(1, 15)]
  variable_2: Annotated[float, ns.Interval(0.0, 0.6, cast=float)]

Boolean - when you're using true or false

from dataclasses import dataclass
from typing import Annotated

import nano_settings as ns


@dataclass
class Config(ns.BaseConfig):
  variable_1: Annotated[bool, ns.Boolean()]