Examples

Runnable examples are in the examples/ directory at the root of the repository. Each file is self-contained and can be run directly.

FastAPI integration

examples/fastapi_integration.py

Shows how to use cfx for application settings alongside a FastAPI app. cfx handles the settings layer — database connection parameters, feature flags, env-var overrides, CLI arguments. pydantic handles request/response schemas; the two roles don’t overlap.

Key patterns shown:

  • Environment-variable-backed fields (env="DB_HOST")

  • Cross-field validation in Config.validate()

  • Nested sub-configs via components=[DatabaseConfig, AppSettings]

  • CLI entry point via add_arguments / from_argparse

ML pipeline with views

examples/ml_pipeline.py

Shows composable subsystem configs for an ML training pipeline (preprocessing, model, training) assembled with components=, plus a ConfigView that surfaces the most-tweaked parameters under short names for interactive use in a Jupyter notebook.

Key patterns shown:

  • Multi-component config composition

  • ConfigView with Alias paths for notebook ergonomics

  • Writes through the view go back to the underlying config

  • CLI entry point with nested --model.learning_rate-style flags

Custom field types

examples/custom_fields.py

Demonstrates three patterns for custom ConfigField subclasses:

  • Angle — normalize wraps any numeric input to [0, 360)

  • LogScale — validate enforces a log10-space range; extra __init__ args

  • Percentage — normalize rounds to configurable precision; validate checks [0, 100]