A pluggy-based Plugin Architecture Demo

This repository contains a simple, illustrative example of a Python application (hop3-host) that uses a plugin (hop3-plugin) to extend its functionality. The core of this demonstration is the pluggy library, which is the same framework used by the popular testing tool pytest.

Disclaimer: This code is for educational purposes only. It is a simplified demonstration and is not intended for production use. The primary goal is to showcase the mechanics of plugin discovery and integration.

Project Structure

The repository is organized as a workspace with two main packages:

• hop3-host: A host application that defines extension points (hooks) for building and deploying applications. It knows nothing about the specifics of how to build or deploy; it only defines the interfaces (protocols).
• hop3-plugin: An external plugin that provides concrete implementations (strategies) for building and deploying. In this example, it provides a Docker-based build strategy.

Core Concepts Demonstrated

• Plugin Discovery: The host application uses pluggy to discover and load plugins at runtime.
• Hook System: The host defines "hook specifications" (hookspecs) that plugins can implement to provide new functionality.
• Separation of Concerns: The host application is decoupled from the specific implementation details of its plugins. This allows for new build or deployment methods to be added without modifying the host's code.
• Protocols: The use of typing.Protocol ensures that plugins adhere to the expected interface, providing a degree of static analysis and clarity.

How to Run the Demo

This project uses uv for managing dependencies and the workspace.

Setup

1. Install uv: If you don't have it, install uv by following the official instructions.

2. Install Dependencies: From the root of the project, install the packages in editable mode. This will also install pluggy.

   uv sync

Running the Demo Script

The hop3-host package contains a small script, demo.py, to demonstrate the plugin system. It initializes the plugin manager and lists all registered plugins.

To run it:

python -m hop3_host.demo

The expected result is:

Registered plugins:
Total plugins: 1
- docker-plugin (builtins.type)```

Plugin Discovery

To make the host application discover the plugin, an entry point has been added to the plugin's pyproject.toml.

The hop3-plugin/pyproject.toml file contains the following section:

[project.entry-points."hop3"]
smo = "hop3_plugin.plugin:DockerPlugin"

Explanation:

• [project.entry-points."hop3"]: This tells Python's packaging system that this project provides plugins for the "hop3" entry point group.
• smo = "hop3_plugin.plugin:DockerPlugin": This registers the SmoPlugin class from the hop3_plugin.plugin module as a loadable plugin.