tamahagane 0.2.0

Forge your application registries.

Tamahagane

Forge your application registries.

Tamahagane is a library designed for constructing application registries, the essential foundation for building decoupled software architectures and enabling unit testing on application handlers.

Motivation

Tamahagane essentially serves as a modern alternative to Venusian, fulfilling the same core functionality.

The decision to create a new library stems from Venusian's limitations: it was originally built for Python 2, relies heavily on dynamic behavior, and lacks type support.

Tamahagane reuse the same vocabular, a Scanner and an attach function, but the API is not fully compatible, in order to get a simpler version.

Usage

To use Tamahagane you need to create a registries class that hold all registries your app may load. The definition of the registry is free and depends of the usage.

from dataclasses import dataclass

@dataclass
class Registries:
    app_registry: ...

After what, the registries is filled out using the scan of the application code.

import tamahagane as th

scanner = th.Scanner[Registries](Registries(app_registry=...))
scanner.scan("app.service_handlers")

At this time, the app.service_handlers is a module, or a package containing submodules, that will be recursibely loaded. It contains decorated function, where the decorator has been created in the application code to create the callback.

from collections.abc import Callable
from typing import Any

import tamahagane as th


CommandHandler = Callable[..., Any]

def command_handler(wrapped: CommandHandler) -> CommandHandler:

    def callback(scanner: th.Scanner[Registries]) -> None:
        scanner.registry.app_registry.do_something_with(wrapped, ...)

    th.attach(callback, category="app_registry")
    return wrapped

Now, you have a command_handler decorator that can be used an be filled out an application registry with the decorated method.

@command_handler
def handle_stuff(...):
    ...

handle_stuff is unmodified by its decorator and is purely unit testable. No overhead.

Installation

Tamahagane is available on PyPI

So you can installing using your favorite packaging tool, mine is uv:

uv add tamahagane

See also

• Documentation - https://mardiros.github.io/tamahagane/
• Venusian - https://docs.pylonsproject.org/projects/venusian/en/