goodconf 7.1.0

Load configuration variables from a file or environment

A thin wrapper over Pydantic’s settings management. Allows you to define configuration variables and load them from environment or JSON/YAML/TOML file. Also generates initial configuration files and documentation for your defined configuration.

Installation

pip install goodconf or pip install goodconf[yaml] / pip install goodconf[toml] if parsing/generating YAML/TOML files is required. When running on Python 3.11+ the [toml] extra is only required for generating TOML files as parsing is supported natively.

Quick Start

Let’s use configurable Django settings as an example.

First, create a conf.py file in your project’s directory, next to settings.py:

import base64
import os

from goodconf import GoodConf, Field
from pydantic import PostgresDsn

class AppConfig(GoodConf):
    "Configuration for My App"
    DEBUG: bool
    DATABASE_URL: PostgresDsn = "postgres://localhost:5432/mydb"
    SECRET_KEY: str = Field(
        initial=lambda: base64.b64encode(os.urandom(60)).decode(),
        description="Used for cryptographic signing. "
        "https://docs.djangoproject.com/en/2.0/ref/settings/#secret-key")

    model_config = {"default_files": ["/etc/myproject/myproject.yaml", "myproject.yaml"]}

config = AppConfig()

Next, use the config in your settings.py file:

import dj_database_url
from .conf import config

config.load()

DEBUG = config.DEBUG
SECRET_KEY = config.SECRET_KEY
DATABASES = {"default": dj_database_url.parse(config.DATABASE_URL)}

In your initial developer installation instructions, give some advice such as:

python -c "import myproject; print(myproject.conf.config.generate_yaml(DEBUG=True))" > myproject.yaml

Better yet, make it a function and entry point so you can install your project and run something like generate-config > myproject.yaml.

Usage

GoodConf

Your subclassed GoodConf object can include a model_config dictionary with the following attributes:

file_env_var

The name of an environment variable which can be used for the name of the configuration file to load.

default_files

If no file is passed to the load method, try to load a configuration from these files in order.

It also has one method:

load

Trigger the load method during instantiation. Defaults to False.

Use plain-text docstring for use as a header when generating a configuration file.

Environment variables always take precedence over variables in the configuration files.

See Pydantic’s docs for examples of loading:

• Dotenv (.env) files

• Docker secrets

Fields

Declare configuration values by subclassing GoodConf and defining class attributes which are standard Python type definitions or Pydantic FieldInfo instances generated by the Field function.

Goodconf can use one extra argument provided to the Field to define an function which can generate an initial value for the field:

initial

Callable to use for initial value when generating a config

Django Usage

Install with the django extra to pull in Django for the helper below:

pip install goodconf[django]

A helper is provided which monkey-patches Django’s management commands to accept a --config argument. Replace your manage.py with the following:

# Define your GoodConf in `myproject/conf.py`
from myproject.conf import config

if __name__ == '__main__':
    config.django_manage()

Why?

I took inspiration from logan (used by Sentry) and derpconf (used by Thumbor). Both, however used Python files for configuration. I wanted a safer format and one that was easier to serialize data into from a configuration management system.

Environment Variables

I don’t like working with environment variables. First, there are potential security issues:

1. Accidental leaks via logging or error reporting services.

2. Child process inheritance (see ImageTragick for an idea why this could be bad).

Second, in practice on deployment environments, environment variables end up getting written to a number of files (cron, bash profile, service definitions, web server config, etc.). Not only is it cumbersome, but also increases the possibility of leaks via incorrect file permissions.

I prefer a single structured file which is explicitly read by the application. I also want it to be easy to run my applications on services like Heroku where environment variables are the preferred configuration method.

This module let’s me do things the way I prefer in environments I control, but still run them with environment variables on environments I don’t control with minimal fuss.

Contribute

Install dependencies.

uv sync

Run the tests against your environment:

uv run pytest

Or run the full matrix with tox (installed with the tox-uv plugin):

uv tool install tox --with tox-uv
tox

Tested versions

CI runs two dependency stacks (see tox.ini): the minimum supported versions (pydantic 2.7, pydantic-settings 2.13, ruamel.yaml 0.17, tomlkit 0.11, Django 5.2) on the older Pythons, and the latest releases on the newer ones.

Python	Django	deps
3.10	5.2	minimum
3.11	5.2	minimum
3.12	6.0	latest
3.13	6.0	latest
3.14	6.0	latest

dev also runs the exact versions pinned in uv.lock; mypy type-checks.

Releases are done with GitHub Actions whenever a new tag is created. For more information, see ./.github/workflows/build.yml