lml 0.1.0

Load me later. A lazy plugin management system.

lml seamlessly finds the lml based plugins from your current python environment but loads your plugins on demand. It is designed to support plugins that have external dependencies, especially bulky and/or memory hungry ones. lml provides the plugin management system only and the plugin interface is on your shoulder.

lml enabled applications helps your customers [1] in two ways:

1. Your customers could cherry-pick the plugins from pypi per python environment. They could remove a plugin using pip uninstall command.
2. Only the plugins used at runtime gets loaded into computer memory.

When you would use lml to refactor your existing code, it aims to flatten the complexity and to shrink the size of your bulky python library by distributing the similar functionalities across its plugins. However, you as the developer need to do the code refactoring by yourself and lml would lend you a hand.

[1]	the end developers who uses your library and packages achieve their objectives.

Quick start

The following code tries to get you started quickly with non-lazy loading.

from lml.plugin import PluginInfo, PluginManager


@PluginInfo("cuisine", tags=["Portable Battery"])
class Boost(object):
    def make(self, food=None, **keywords):
        print("I can cook %s for robots" % food)


class CuisineManager(PluginManager):
    def __init__(self):
        PluginManager.__init__(self, "cuisine")

    def get_a_plugin(self, food_name=None, **keywords):
        return PluginManager.get_a_plugin(self, key=food_name, **keywords)


if __name__ == '__main__':
    manager = CuisineManager()
    chef = manager.get_a_plugin("Portable Battery")
    chef.make()

At a glance, above code simply replaces the Factory pattern should you write them without lml. What’s not obvious is, that once you got hands-on with it, you can start work on how to do lazy loading.

Installation

You can install lml via pip:

$ pip install lml

or clone it and install it:

$ git clone https://github.com/python-lml/lml.git
$ cd lml
$ python setup.py install

lml enabled project

Beyond the documentation above, here is a list of projects using lml:

1. pyexcel
2. pyecharts
3. moban

lml is available on these distributions:

1. ARCH linux
2. Conda forge
3. OpenSuse

License

New BSD

Change log

0.1.0 - 21/10/2020

Updated

1. non class object can be a plugin too
2. #20: When a plugin was not installed, it now calls raise_exception method

0.0.9 - 7/1/2019

Updated

1. #11: more test contents for OpenSuse package validation

0.0.8 - 4/1/2019

Updated

1. #9: include tests, docs for OpenSuse package validation

0.0.7 - 17/11/2018

Fixed

1. #8: get_primary_key will fail when a module is loaded later
2. deprecated old style plugin scanner: scan_plugins

0.0.6 - 07/11/2018

Fixed

1. Revert the version 0.0.5 changes. Raise Import error and log the exception

0.0.5 - 06/11/2018

Fixed

1. #6: Catch and Ignore ModuleNotFoundError

0.0.4 - 07.08.2018

Added

1. #4: to find plugin names with different naming patterns

0.0.3 - 12/06/2018

Added

1. dict can be a pluggable type in addition to function, class
2. get primary tag of your tag, helping you find out which category of plugins your tag points to

0.0.2 - 23/10/2017

Updated

1. pyexcel#103: include LICENSE in tar ball

0.0.1 - 30/05/2017

Added

1. First release