Vidhi Documentation¶

Vidhi is a configuration management library for Python applications that provides type-safe, immutable configurations with CLI and YAML support.

Note

Vidhi (विधि) means “method” or “procedure” in Sanskrit.

Key Features¶

Type-safe configurations: Build on Python dataclasses with full type checking support.
Immutability: Configurations are frozen after creation, preventing accidental modifications.
Polymorphic configs: Define configuration variants with automatic type-based selection.
CLI integration: Auto-generate CLI arguments from your config classes with --help support.
YAML/JSON loading: Load configurations from files with nested structure support.

Quick Example¶

Define a configuration:

from vidhi import frozen_dataclass, field, parse_cli_args

@frozen_dataclass
class TrainingConfig:
    learning_rate: float = field(0.001, help="Learning rate", name="lr")
    batch_size: int = field(32, help="Batch size")
    epochs: int = field(10, help="Number of epochs")

config = parse_cli_args(TrainingConfig)

Run from command line:

python train.py --lr 0.01 --batch_size 64 --epochs 20