Napari Hub Lite - napari.org/hub-lite

This implementation of the napari hub is developed and managed by the napari team. It is based on an original implementation by Yunha Lee.

The website is deployed daily using GitHub pages and contains plugin information served by npe2api. This GitHub repository contains the Python scripts, javascript, HTML templates and CSS templates required to build the web pages.

How the site works

The site is a static build with a small Python-based generator.

1. build_site.py orchestrates the local build, copies shared assets into _build/, and runs the data and HTML generation steps.
2. fetch_napari_data.py fetches plugin data from npe2api for plugin information including PyPI/conda info and individual plugin manifests, then writes build data into _build/data/.
3. create_static_html_files.py reads _build/data/plugin_page_data.json and generates:
   • _build/plugins/*.html for individual plugin pages
   • _build/plugins_list.html for the homepage plugin list
4. index.html is the homepage shell and static/js/plugin_loading_and_search.js loads the generated plugin list and client-side search behavior.
5. .github/workflows/build_and_deploy.yml runs pixi run all and publishes _build/ to GitHub Pages.

The key implementation surfaces are:

• fetch_napari_data.py: external data fetching and normalization
• create_static_html_files.py: HTML generation and markdown rendering
• templates/each_plugin_template.html: plugin detail page template
• index.html: homepage structure
• static/js/plugin_loading_and_search.js: homepage interactivity
• static/css/: copied styling assets used by the generated pages

Local Development

Quick Start with pixi:

To build the website locally:

# install pixi if you don't have it already
pixi run all

or to serve the site locally in your browser:

pixi run live

Development Workflow

All tasks are available through build_site.py and can be run with pixi or uv. The tasks are:

• clean: deletes the _build/ directory and everything inside it, so use with caution. This is the first step in a full build, but you can also run it separately if you want to start from a clean slate.
• prep: copies static assets into _build/ and prepares the build environment, but does not run the data fetching or HTML generation steps, so it is fast to run and can be used before fetch-data or html for iterative development.
• fetch-data: runs the data fetching step, which can be slow due to the number of plugins and API calls, but is only needed when plugin data changes.
• html: runs the data fetching and HTML generation steps, but does not clean or copy static assets, so it is faster for iterative development.

Recommended: pixi

Pixi is the easiest cross-platform way to build the site locally.

Useful tasks:

• pixi run clean
• pixi run prep
• pixi run fetch-data
• pixi run html
• pixi run all
• pixi run live

Alternative: uv

If you prefer a Python-native workflow without Pixi, use uv against the same pyproject.toml manifest. The underlying Python scripts are the same as with Pixi. E.g.

• uv run python build_site.py all
• uv run python build_site.py live

Acknowledgments

This project makes extensive use of the design and html resources from the original napari hub. All resources are used with permission.