Plugins
The purpose of a Skelebot plugin is to add some extra functionality into Skelebot that doesn’t already exist. This can be anything, but usually it is more useful when the plugin is trying to accomplish a specific task that most projects would not have a use for. If the functionality you are trying to implement seems general to all machine learning projects, consider instead contributing it directly to Skelebot instead of making a plugin.
Install
Installing new plugins is very simple, all that is needed is a zip file of plugin code structured properly and named after the name of the plugin itself.
> skelebot plugin testPlugin.zip
Once installed the plugin can be used immediately. Some plugins required config in the skelebot.yaml to be present in order to work, as demonstrated in the example below, so be sure to consult any documentation for the plugin to find out how to properly make use of it. Since scaffolding only applies to new Skelebot projects, the details in the config file may need to be entered manually for existing projects.
Structure
A Skelebot plugin should be placed in a folder with the same name as the component class object python script. This folder should be zipped into a file of the same name (jupyter.zip) when it is ready to be installed or distributed.
Example Jupyter Plugin:
- jupyter/
- jupyter.py
A plugin is nothing more than a component that is loaded dynamically by Skelebot at runtime. To create a plugin simply create a class that extends the Component object and ensure it is placed in a script that is named the same (albeit with the first letter lower-case). The name of the script should follow camel case (thisIsCamelCase), and the name of the class object should follow pascal case (ThisIsPascalCase).
Example Plugin Script (jupyter.py):
from skelebot.objects.component import Compenent, Activation
from skelebot.systems.execution import docker
# This component provides the ability to spin up Jupyter in Docker for any project
class Jupyter(Component):
activation = Activation.PROJECT
commands = ["jupyter"]
port = None
folder = None
# Allows for intelligent defaults in the constructor so projects without config can still use the component
def __init__(self, port=8888, folder="."):
self.port = port
self.folder = folder
# Parser for the command that spins up Jupyter inside the Docker Container based on the given port and folder
def addParsers(self, subparsers):
helpMessage = "Spin up Jupyter in a Docker Container (port = {port}, folder = {folder})".format(port=self.port, folder=self.folder)
parser = subparsers.add_parser("jupyter", help=helpMessage)
return subparsers
# Build the docker image and then run the container with the Jupyter command, port mapped, and folder volume mapped
def execute(self, config, args):
status = docker.build(config)
if (status == 0):
command = "jupyter notebook --ip=0.0.0.0 --port=8888 --allow-root --notebook-dir={folder}".format(folder=self.folder)
ports = ["{port}:8888".format(port=self.port)]
return docker.run(config, command, "i", ports, ".", "jupyter")
Activation
The activation field of a Component specifies at which point the plugin Component will be available for use. There are four possible options for this field.
-
EMPTY Only available when run outside of a Skelebot project (no skelebot.yaml file present) -
CONFIG Only available when the Skelebot project has the specified component attributes present in the config -
PROJECT Available when run inside any Skelebot project -
ALWAYS Always available no matter what
If the field is not specifed, it will default to the CONFIG activation level.
Attributes
Any attributes you need can be added to the plugin class as long as you specify them in the __init__ function as well. These attributes will be configurable via the corresponding section of the skelebot.yaml components feild.
The name of the section under components will be the camel case version of the plugin’s name.
Example:
components:
jupyter:
port: 1127
folder: .
API
There are numerous places that a plugin can hook into the Skelebot Systems to augment the tool with new features.
All of the details for the hooks as well as several other Skelebot System functions can be found in the API section of the documentation.