baukasten 0.0.1

A plugin framework.

Python Baukasten

Python Baukasten is an easy to use pluggable plugin framework for python. It is a result of a long period of comparing numerous python plugin frameworks, identifying their key characteristics and designing an interface which allows to develop your own plugin manager with all the features you need while providing a sane default implementation supporting the following features:

• Import of plugins from entry-points, folders, modules, packages, and classes
• Configurable automatic initialization of plugins
• Filterable custom plugin metadata definition
• Easy to manage persistent plugin manager and plugin configuration
• Dependency-, version- and update-management

Usage

This section illustrates how to use the default implementation of Python Baukasten.

app_dir = os.path.abspath(os.path.dirname(__file__))
context = baukasten.Context(app_dir=app_dir, app_name="my-app")
plugin_manager = baukasten.PluginManager(context)

By default the PluginManager searches for plugins at the following locations:

• app_dir/plugins (e.g. /path/to/application/plugins)
• xdg_data_dir/plugins (e.g. /home/user/.config/app_name/plugins)
• entry-points

Plugins are detected via the @plugin decorator which can be used on functions, classes, and methods. It accepts arbitrary arguments which can be used to enrich the plugin with metadata.

@plugin(spam="egg", egg="spam")
def my_function():
    return 42

@plugin(egg="spam")
class MyClass:
    pass
    
class OtherClass:
    
    @plugin(spam="egg")
    def my_method(self):
        return 42

TODO: derive from yet unknown class (e.g. depends on other plugin) To retrieve the plugins the filter method can be used:

# Gather all plugins
plugins = plugin_manager.filter()

# Gather plugins filtering on metadata
plugins = plugin_manager.filter(spam="egg")

Advanced Usage

Configuration

The default implementation of Python Baukasten supports two types of persistent configurations. The plugin manager configuration used for storing settings related how plugins are loaded and processed and plugin configuration where individual settings of a plugin can be stored.

Hooks

When initializing the PluginManager a list of hooks can be passed:

plugin_manager = baukasten.PluginManager(context, hooks=[
    PluginPreLoadHook(lambda plugins: filter(lambda plugin: plugin.__plugin_metadata.get("egg") == "spam", plugins)),
    PluginInitHook(lambda plugin: plugin("parameter-1", "parameter-2", ...))
])