phx-class-registry 5.2.1

Factory+Registry pattern for Python classes

ClassRegistry

At the intersection of the Registry and Factory patterns lies the ClassRegistry:

• Define global factories that generate new class instances based on configurable keys.

• Seamlessly create powerful service registries.

• Integrate with setuptools’s entry_points system to make your registries infinitely extensible by 3rd-party libraries!

• And more!

Upgrading from ClassRegistry v4

Important

ClassRegistry v5 introduces some changes that can break code that was previously using ClassRegistry v4. If you are upgrading from ClassRegistry v4 to ClassRegistry v5, please read Upgrading to ClassRegistry v5.

Getting Started

Create a registry using the class_registry.ClassRegistry class, then decorate any classes that you wish to register with its register method:

from class_registry import ClassRegistry

pokedex = ClassRegistry()

@pokedex.register('fire')
class Charizard(Pokemon):
  ...

@pokedex.register('grass')
class Bulbasaur(Pokemon):
  ...

@pokedex.register('water')
class Squirtle(Pokemon):
  ...

To create a class instance from a registry, use the subscript operator:

# Charizard, I choose you!
fighter1 = pokedex['fire']

# CHARIZARD fainted!
# How come my rival always picks the type that my pokémon is weak against??
fighter2 = pokedex['grass']

Tip

If a ClassRegistry always returns objects derived from a particular base class, you can provide a type parameter to help with type checking, autocomplete, etc.:

# Add type parameter ``[Pokemon]``:
pokedex = ClassRegistry[Pokemon]()

# Your IDE will automatically infer that ``fighter1`` is a ``Pokemon``.
fighter1 = pokedex['fire']

Advanced Usage

There’s a whole lot more you can do with ClassRegistry, including:

• Provide args and kwargs to new class instances.

• Automatically register non-abstract classes.

• Integrate with setuptools’s entry_points system so that 3rd-party libraries can add their own classes to your registries.

• Wrap your registry in an instance cache to create a service registry.

• And more!

For more advanced usage, check out the documentation on ReadTheDocs!

Requirements

ClassRegistry is known to be compatible with the following Python versions:

• 3.14

• 3.13

• 3.12

Note

I’m only one person, so to keep from getting overwhelmed, I’m only committing to supporting the 3 most recent versions of Python. ClassRegistry’s code is pretty simple, so it’s likely to be compatible with versions not listed here; there just won’t be any test coverage to prove it 😇

Installation

Install the latest stable version via pip:

pip install phx-class-registry

Important

Make sure to install phx-class-registry, not class-registry. I created the latter at a previous job years ago, and after I left they never touched that project again and stopped responding to my emails — so in the end I had to fork it 🤷

Maintainers

To install the distribution for local development, some additional setup is required:

1. Install uv (only needs to be done once).

2. Run the following command to install additional dependencies:

   uv sync --group=dev

3. Activate pre-commit hook:

   uv run autohooks activate --mode=pythonpath

Running Unit Tests and Type Checker

Run the tests for all supported versions of Python using tox:

uv run tox -p

The -p flag runs tests for each version of Python in parallel. Omit it if you want to see the tests run for one Python version at a time.

Note

The first time this runs, it will take awhile, as mypy needs to build up its cache. Subsequent runs should be much faster.

If you just want to run unit tests in the current virtualenv (using pytest):

uv run pytest

If you just want to run type checking in the current virtualenv (using mypy):

uv run mypy src test

Documentation

To build the documentation locally:

1. Switch to the docs directory:

   cd docs

2. Build the documentation:

   uv run make html

Releases

Steps to build releases are based on Packaging Python Projects Tutorial.

Important

Make sure to build releases off of the main branch, and check that all changes from develop have been merged before creating the release!

One-time Setup

1. Install the keyring tool and add it to your PATH:

   uv tool install keyring
   uv tool update-shell

   Restart your shell after running update-shell.

2. Create a PyPI API token and store it in the OS keychain:

   keyring set https://upload.pypi.org/legacy/ __token__

   Paste the pypi-... token when prompted.

1. Build the Project

1. Delete artefacts from previous builds, if applicable:

   rm dist/*

2. Run the build:

   uv build

3. The build artefacts will be located in the dist directory at the top level of the project.

2. Upload to PyPI

1. Bump the version (also updates uv.lock):

   uv version <version>

2. Upload build artefacts to PyPI:

   uv publish --username __token__

3. Create GitHub Release

1. Create a tag and push to GitHub:

   git tag -a <version> -m "Release <version>"
   git push origin <version>

2. Go to the Releases page for the repo.

3. Click Draft a new release.

4. Select the tag that you created in step 1.

5. Specify the title of the release (e.g., ClassRegistry v1.2.3).

6. Write a description for the release. Make sure to include: - Credit for code contributed by community members. - Significant functionality that was added/changed/removed. - Any backwards-incompatible changes and/or migration instructions. - SHA256 hashes of the build artefacts.

7. GPG-sign the description for the release (ASCII-armoured).

8. Attach the build artefacts to the release.

9. Click Publish release.