Allows Jupyter notebook extension writers to make their extension pip installable!
Specifically, if you have a module that you've written which is intended for use in the Jupyter notebook and also has Javascript associated with it, this is designed to solve the distribution problem.
In setup.py
try:
from jupyterpip import cmdclass
except:
import pip, importlib
pip.main(['install', 'jupyter-pip']); cmdclass = importlib.import_module('jupyterpip').cmdclass
and inside the setup(...)
call
install_requires=["jupyter-pip"],
cmdclass=cmdclass('myplugin', 'myplugin/init'),
You don't need to download this to use it! Instead you just need to modify your package's setup.py
to download it at install time.
The following example assumes a folder structure like this:
./myplugin/__init__.py
./myplugin/init.js
./setup.py
Starting with a setup.py file that looks like this:
# -*- coding: utf-8 -*-
from setuptools import setup
setup(
name='myplugin',
packages=['myplugin'],
# ... more setup.py stuff here ...
)
All you need to do is add a try import block and two entries to the setup call to make the project nbextension compatabile:
# -*- coding: utf-8 -*-
from setuptools import setup
try:
from jupyterpip import cmdclass
except:
import pip, importlib
pip.main(['install', 'jupyter-pip']); cmdclass = importlib.import_module('jupyterpip').cmdclass
setup(
name='myplugin',
packages=['myplugin'],
# ... more setup.py stuff here ...
install_requires=["jupyter-pip"],
cmdclass=cmdclass('myplugin', 'myplugin/init'),
)
The second argument to the cmdclass
call is optional. If specified, the JS module will be loaded alongside the notebook automatically. Note that for widget packages, typically this isn't this best idea. Instead the line would read cmdclass=cmdclass('myplugin')
and the widget framework would load the JS when the widget is constructed.
New BSD (see ./LICENSE
file).