airyconf

Lightweight configuration manager

Inspired by zifeo/dataconf, implemented with no required dependencies for Python 3.6+ (except Python 3.6 requires dataclasses package to support the corresponding functionality). Supports typing, dataclasses and custom classes:

from typing import List
from dataclasses import dataclass
from airyconf import parse


class CustomObject(object):
    def __init__(self, a: float):
        self.a = a
    
    def __eq__(self, other):
        return isinstance(other, CustomObject) and other.a == self.a

@dataclass
class SomeDataclass:
    b: str
    c: List[CustomObject]

json_config = '{"SomeDataclass": {"b": "str", "c": [{"a": 1.0}, {"a": 2.0}]}}'

expected = SomeDataclass('str', [CustomObject(1.), CustomObject(2.)])

assert parse(json_config, SomeDataclass) == expected

Optionally, supports yaml if installed:

from airyconf import parse_config
parse_config('file/to/config.yaml', SomeDataclass)

Thanks to rossmacarthur/typing-compat by Ross MacArthur, typing is supported for Python 3.6+ (the single-file package is copied to avoid dependencies).

Package is under development. TODO:

• Add documentation
• Github actions: tests & coverage & package update
• Multiple typing arguments (Tuple[int, CustomClass, str])
• Nested typing arguments (List[Dict[str, List[int]]])
• Specify classes in config (a: SomeCustomClass({param: 1}))