ConfigKit

Thread-safe singleton JSON/YAML configuration loader with JSON Schema validation for Python applications.

Features

• JSON & YAML support - Load configuration from .json, .yaml, or .yml files
• Thread-safe singleton - One instance per configuration class
• JSON Schema validation - Draft 2020-12 support via jsonschema
• Dot-notation access - config.get("database.host")
• Runtime reload - Update configuration without restart
• Type-safe - Full type hints with py.typed marker
• Extensible - Custom validation via additional_checks()

Installation

From PyPI

pip install pyConfigKit

From Git repository

pip install git+https://github.com/miichoow/ConfigKit.git

Development mode

git clone https://github.com/miichoow/ConfigKit.git
cd ConfigKit
pip install -e ".[dev]"

Quick Start

1. Create your configuration class

from configkit import ConfigKit


class AppConfig(ConfigKit):
    def additional_checks(self) -> None:
        # Custom validation logic
        if self.data.get("debug") and self.data.get("env") == "production":
            raise ValueError("Debug mode not allowed in production")

    def get_database_url(self) -> str:
        host = self.get("database.host")
        port = self.get("database.port", default=5432)
        name = self.get("database.name")
        return f"postgresql://{host}:{port}/{name}"

2. Create configuration files

config.json

{
  "env": "development",
  "debug": true,
  "database": {
    "host": "localhost",
    "port": 5432,
    "name": "myapp"
  }
}

Or config.yaml

env: development
debug: true
database:
  host: localhost
  port: 5432
  name: myapp

schema.json (schema files are always JSON)

{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "type": "object",
  "required": ["env", "database"],
  "properties": {
    "env": {
      "type": "string",
      "enum": ["development", "staging", "production"]
    },
    "debug": {
      "type": "boolean"
    },
    "database": {
      "type": "object",
      "required": ["host", "name"],
      "properties": {
        "host": { "type": "string" },
        "port": { "type": "integer", "minimum": 1, "maximum": 65535 },
        "name": { "type": "string" }
      }
    }
  }
}

3. Use in your application

# Initialize once at startup (JSON)
config = AppConfig(config_file="config.json", schema_file="schema.json")

# Or with YAML
config = AppConfig(config_file="config.yaml", schema_file="schema.json")

# Access anywhere - returns the same instance
config = AppConfig()
print(config.get_database_url())

API Reference

ConfigKit

Base class for configuration. Subclass and implement additional_checks().

Constructor

ConfigKit(*, config_file: str | Path, schema_file: str | Path)

• config_file - Path to configuration file (.json, .yaml, or .yml)
• schema_file - Path to JSON Schema file

Properties

Methods

ConfigKitMeta

Metaclass providing singleton behavior.

Design Principles

• Fail fast - Invalid configuration raises exceptions immediately
• Single source of truth - One instance per configuration class
• Explicit contracts - Required files on first instantiation
• No magic - Clear, predictable behavior

Publishing

pip install -e ".[dev]"
python -m build
twine upload dist/*

License

MIT License - see LICENSE for details.