voici 0.5.0

Voici turns Jupyter notebooks into static web applications

Voici

Voici is a tool for generating static dashboards from Jupyter Notebooks. It can be used as a drop-in replacement for Voilà and it has the same commands and supports most of Voila's configuration options.

Unlike Voila, which renders interactive dashboards using server-side execution, Voici uses WebAssembly (Wasm) kernels to render notebooks in the browser, making the resulting dashboard entirely self-contained and distributable.

This is made possible thanks to the amazing work done in the JupyterLite project.

https://user-images.githubusercontent.com/591645/222892327-2a5b1341-640d-49c2-9e95-1f2d3ec122be.mp4

Features 🚀

• Generates self-contained HTML files with embedded Wasm kernels.
• Works offline, without requiring a server to run the dashboard.
• Supports custom templates for styling dashboards, powered by Jinja2.
• Supports all programming languages that have JupyterLite kernels available. e.g. the default JavaScript and Python kernels JupyterLite provides, python with xeus-python, lua, nelson

Getting Started 🏁

To use Voici, you'll need to install it first:

pip install voici

Then, you can generate static dashboards from a notebook or a directory of Notebooks like this:

# Build a single dashboard
voici my-notebook.ipynb
# Build a directory of notebooks
voici notebooks/

Once your dashboards are built, you can simply serve them with a simple static server, e.g.:

cd _output
python -m http.server

Advanced usage

The voici command line interface is a mix between voila and jupyter lite. For most cases, users can rely on the voici command by using the voila CLI syntax.

Voici runs the build sub-command by default, the voici my-notebook.ipynb command is a shortcut for voici build --contents my-notebook.ipynb For advanced usage, users can invoke voici with the jupyter lite CLI syntax, e.g.:

voici build --contents my-notebook.ipynb

The difference between voici build and jupyter lite build commands is that the voici one will only generate Voici dashboards, excluding the full JupyterLab interface from the output. Running voici build --contents . is equivalent to running jupyter lite build --contents . --apps voici.

You can generate the classic jupyter lite output alongside your dashboards by specifying the additional apps you want:

voici build --contents . --apps lab --apps retro

In order to get some help on how to use the voici command, you can run:

voici --help

We'd also suggest looking into the JupyterLite documentation for more insights on how to configure your voici deployment.

Build the demo site yourself

Before you can build the demo site from this repository, you will need to install jupyterlite-xeus-python:

pip install jupyterlite-xeus-python

You will also need either micromamba, mamba or conda installed in order to build the emscripten environment.

The demo directory contains:

• notebooks/: The directory of Notebooks that will be served by Voici
• environment.yml: The file specifying the Emscripten environment that will be used for rendering the dashboards, this must contain all your Notebook dependencies
• jupyter-lite.json: The JupyterLite config file, this config will disable the default Python kernel JupyterLite provides, allowing to use xeus-python instead.

Run the following command to build the demo site:

git clone https://github.com/voila-dashboards/voici
cd voici/demo

voici notebooks

Then serve it!

cd _output
python -m http.server

Make your own Github pages deployment

Please follow this guide for making your own Github pages deployment.

Limitations ⚠️

Because Voici uses Wasm kernels to execute notebooks, there are some limitations to the types of notebooks that can be rendered: Some notebook features may not work correctly or may have limited functionality when rendered in Voici.

Contributing 👋

If you find a bug or have a feature request, please open an issue on the GitHub repository. If you'd like to contribute code, please fork the repository and submit a pull request. We welcome contributions from anyone!