Distributing Javascript#
There are two ways that you can distribute your Custom Javascript Components:
These options are not mutually exclusive though, and it may be beneficial to support both options. For example, if you upload your Javascript components to NPM and also bundle your Javascript inside a Python package, in principle your users can determine which work best for them. Regardless though, either you or, if you give then the choice, your users, will have to consider the tradeoffs of either approach.
Distributing Javascript via CDN - Most useful in production-grade applications where its assumed the user has a network connection. In this scenario a CDN’s edge network can be used to bring the Javascript source closer to the user in order to reduce page load times.
Distributing Javascript via PyPI - This method is ideal for local usage since the user can server all the Javascript components they depend on from their computer without requiring a network connection.
Distributing Javascript via CDN#
Under this approach, to simplify these instructions, we’re going to ignore the problem of distributing the Javascript since that must be handled by your CDN. For open source or personal projects, a CDN like https://unpkg.com/ makes things easy by automatically preparing any package that’s been uploaded to NPM. If you need to roll with your own private CDN, this will likely be more complicated.
In either case though, on the Python side, things are quite simple. You need only pass
the URL where your package can be found to module_from_url()
where you can then load any of its exports:
import reactpy
your_module = ido.web.module_from_url("https://some.cdn/your-module")
YourComponent = reactpy.web.export(your_module, "YourComponent")
Distributing Javascript via PyPI#
This can be most easily accomplished by using the template repository that’s been purpose-built for this. However, to get a better sense for its inner workings, we’ll briefly look at what’s required. At a high level, we must consider how to…
bundle your Javascript into an ECMAScript Module)
include that Javascript bundle in a Python package
use it as a component in your application using ReactPy
In the descriptions to follow we’ll be assuming that:
NPM is the Javascript package manager
The components are implemented with React
Rollup bundles the Javascript module
Setuptools builds the Python package
To start, let’s take a look at the file structure we’ll be building:
your-project
|-- js
| |-- src
| | \-- index.js
| |-- package.json
| \-- rollup.config.js
|-- your_python_package
| |-- __init__.py
| \-- widget.py
|-- Manifest.in
|-- pyproject.toml
\-- setup.py
index.js
should contain the relevant exports (see
Custom Javascript Components for more info):
import * as React from "react";
import * as ReactDOM from "react-dom";
export function bind(node, config) {
return {
create: (component, props, children) =>
React.createElement(component, props, ...children),
render: (element) => ReactDOM.render(element, node),
unmount: () => ReactDOM.unmountComponentAtNode(node),
};
}
// exports for your components
export YourFirstComponent(props) {...};
export YourSecondComponent(props) {...};
export YourThirdComponent(props) {...};
Your package.json
should include the following:
{
"name": "YOUR-PACKAGE-NAME",
"scripts": {
"build": "rollup --config",
...
},
"devDependencies": {
"rollup": "^2.35.1",
"rollup-plugin-commonjs": "^10.1.0",
"rollup-plugin-node-resolve": "^5.2.0",
"rollup-plugin-replace": "^2.2.0",
...
},
"dependencies": {
"react": "^17.0.1",
"react-dom": "^17.0.1",
"@reactpy/client": "^0.8.5",
...
},
...
}
Getting a bit more in the weeds now, your rollup.config.js
file should be designed
such that it drops an ES Module at your-project/your_python_package/bundle.js
since
we’ll be writing widget.py
under that assumption.
Note
Don’t forget to ignore this bundle.js
file when committing code (with a
.gitignore
if you’re using Git) since it can always rebuild from the raw
Javascript source in your-project/js
.
import resolve from "rollup-plugin-node-resolve";
import commonjs from "rollup-plugin-commonjs";
import replace from "rollup-plugin-replace";
export default {
input: "src/index.js",
output: {
file: "../your_python_package/bundle.js",
format: "esm",
},
plugins: [
resolve(),
commonjs(),
replace({
"process.env.NODE_ENV": JSON.stringify("production"),
}),
]
};
Your widget.py
file should then load the neighboring bundle file using
module_from_file()
. Then components from that bundle can be
loaded with export()
.
from pathlib import Path
import reactpy
_BUNDLE_PATH = Path(__file__).parent / "bundle.js"
_WEB_MODULE = reactpy.web.module_from_file(
# Note that this is the same name from package.json - this must be globally
# unique since it must share a namespace with all other javascript packages.
name="YOUR-PACKAGE-NAME",
file=_BUNDLE_PATH,
# What to temporarily display while the module is being loaded
fallback="Loading...",
)
# Your module must provide a named export for YourFirstComponent
YourFirstComponent = reactpy.web.export(_WEB_MODULE, "YourFirstComponent")
# It's possible to export multiple components at once
YourSecondComponent, YourThirdComponent = reactpy.web.export(
_WEB_MODULE, ["YourSecondComponent", "YourThirdComponent"]
)
Note
When reactpy.config.REACTPY_DEBUG_MODE
is active, named exports will be validated.
The remaining files that we need to create are concerned with creating a Python package. We won’t cover all the details here, so refer to the Setuptools documentation for more information. With that said, the first file to fill out is pyproject.toml since we need to declare what our build tool is (in this case Setuptools):
[build-system]
requires = ["setuptools>=40.8.0", "wheel"]
build-backend = "setuptools.build_meta"
Then, we can create the setup.py
file which uses Setuptools. This will differ
substantially from a normal setup.py
file since, as part of the build process we’ll
need to use NPM to bundle our Javascript. This requires customizing some of the build
commands in Setuptools like build
, sdist
, and develop
:
import subprocess
from pathlib import Path
from setuptools import setup, find_packages
from distutils.command.build import build
from distutils.command.sdist import sdist
from setuptools.command.develop import develop
PACKAGE_SPEC = {} # gets passed to setup() at the end
# -----------------------------------------------------------------------------
# General Package Info
# -----------------------------------------------------------------------------
PACKAGE_NAME = "your_python_package"
PACKAGE_SPEC.update(
name=PACKAGE_NAME,
version="0.0.1",
packages=find_packages(exclude=["tests*"]),
classifiers=["Framework :: ReactPy", ...],
keywords=["ReactPy", "components", ...],
# install ReactPy with this package
install_requires=["reactpy"],
# required in order to include static files like bundle.js using MANIFEST.in
include_package_data=True,
# we need access to the file system, so cannot be run from a zip file
zip_safe=False,
)
# ----------------------------------------------------------------------------
# Build Javascript
# ----------------------------------------------------------------------------
# basic paths used to gather files
PROJECT_ROOT = Path(__file__).parent
PACKAGE_DIR = PROJECT_ROOT / PACKAGE_NAME
JS_DIR = PROJECT_ROOT / "js"
def build_javascript_first(cls):
class Command(cls):
def run(self):
for cmd_str in ["npm install", "npm run build"]:
subprocess.run(cmd_str.split(), cwd=str(JS_DIR), check=True)
super().run()
return Command
package["cmdclass"] = {
"sdist": build_javascript_first(sdist),
"build": build_javascript_first(build),
"develop": build_javascript_first(develop),
}
# -----------------------------------------------------------------------------
# Run It
# -----------------------------------------------------------------------------
if __name__ == "__main__":
setup(**package)
Finally, since we’re using include_package_data
you’ll need a MANIFEST.in file that
includes bundle.js
:
include your_python_package/bundle.js
And that’s it! While this might seem like a lot of work, you’re always free to start creating your custom components using the provided template repository so you can get up and running as quickly as possible.