sliveshow

A JupyterLab extension that turns Jupyter notebooks into live, animated Reveal.js slideshows — with scrollable slides, SVG and math (MathJax 4) animations that also play in the normal notebook view, and MyST directive support for Jupyter Book 2.

Install:

pip install sliveshow

PyPI: https://pypi.org/project/sliveshow/
Demo: https://alyanaamirahmedani.github.io/sliveshow/lab/index.html?path=sliveshow_demo.ipynb
(Run all cells before starting the slideshow to display animations correctly.)

Requirements

• JupyterLab >= 4.0.0

Install

pip install sliveshow

Uninstall

pip uninstall sliveshow

Usage

Slideshow

The slideshow uses the Reveal.js framework. Set the slide type of each cell via COMMON TOOLS > Slide Type. Available types: Slide, Sub-Slide, Fragment, Skip.

Transition type and duration are set per-cell in SLIDESHOW TOOLS. Available types: Slide, Fade, Zoom. The default can be changed in Settings.

For Slide transitions, direction is Horizontal (Slides) or Vertical (Sub-Slides) by default.

Code cell input can be hidden via Hide Code Cell, showing only the output.

To start a slideshow, use the Sliveshow menu → Start from first cell (full screen) or Start from current cell (full screen). To exit, press Escape or use Exit slideshow.

Slides beside the notebook

Sliveshow → Open beside notebook puts the slides in a panel next to the notebook instead of taking over the screen, so you can keep working on the notebook while the deck is up:

Nothing is exported and there is no preview to reload, because the panel is a second view of the same notebook: one document, one model, one kernel. Both views show the same live cells, so a change on one side is already on the other.

Sliveshow → Open beside notebook from current cell does the same but opens on the slide holding the selected cell, and Rebuild slides rebuilds the deck by hand if you ever need it.

Close the panel, or use Exit slideshow, to stop. The notebook itself is never modified: the panel is the copy, so closing it leaves your notebook exactly as it was.

Keyboard shortcuts drive the slides only while the slides panel is the active tab, so typing in the notebook can never navigate the deck. The deck's tab is marked with a ▶ so you can tell the two views apart.

SVG Animation

SVG animations use the Reveal.js Animate plugin by Asvin Goel. Animations load after the slideshow starts.

Method A — Raw HTML in a markdown cell

Add a <div data-animate> block containing your SVG and a JSON config comment directly in a markdown cell:

<div data-animate>
<svg xmlns="http://www.w3.org/2000/svg" width="300" height="300">
  <circle id="c1" cx="150" cy="150" r="60" style="fill:steelblue"/>
</svg>
<!--
{
  "setup": [{
    "element": "#c1",
    "modifier": "attr",
    "parameters": [{ "r": 100 }],
    "duration": 1500,
    "begin": 0
  }]
}
-->
</div>

Method B — {svg-animate} MyST directive (recommended)

Use the {svg-animate} directive in a markdown cell. The same cell works in both the live Reveal.js slideshow and a static Jupyter Book 2 / mystmd export:

:::{svg-animate} Circle animation
:height: 320px
:background: white
<svg xmlns='http://www.w3.org/2000/svg' width='300' height='300'>
  <circle id='c1' cx='150' cy='150' r='60' style='fill:steelblue'/>
</svg>
<!--
{
  "setup": [{ "element": "#c1", "modifier": "attr", "parameters": [{ "r": 100 }], "duration": 1500, "begin": 0 }]
}
-->
:::

Note for hubs with jupyterlab-myst installed (e.g. DIVE): jupyterlab-myst parses ::: blocks itself and shows “svg-animate — Unknown Directive” for directives it doesn’t know. sliveshow replaces that error block with the animation, so the cell still animates — but if you only care about the notebook and the slideshow, Method A (<div data-animate>) is the simpler choice. Method B exists so the same cell also renders in a static Jupyter Book 2 build.

To enable {svg-animate} in a mystmd/Jupyter Book 2 build, add the plugin to your myst.yml:

project:
  plugins:
    - svg-animate.mjs

The svg-animate.mjs plugin file is included in the Demo/ folder.

The JSON config controls the animation. setup runs on load; animation runs per fragment. See the Animate plugin README for full options.

Animations in the notebook view

Since 0.1.8, the same data-animate / {svg-animate} cells also animate in the normal notebook view — no slideshow required. One source, works in both.

• Animations autoplay when the cell scrolls into view, and double-click replays them.
• "loop": true in the config makes the animation cycle continuously (notebook and slideshow).
• In the notebook, Reveal fragments don't exist, so all animation stages play back-to-back on one timeline.
• Tip: keep one animation block per markdown cell. Text before/after the block in the same cell is preserved.

Reveal.js plugins (chalkboard, ...)

sliveshow can load any Reveal.js plugin at runtime — no rebuild required. Add entries under Settings → Settings Editor → Sliveshow → Reveal.js plugins:

[
  {
    "name": "RevealChalkboard",
    "script": "https://cdn.jsdelivr.net/npm/reveal.js-plugins@latest/chalkboard/plugin.js",
    "css": [
      "https://cdn.jsdelivr.net/npm/reveal.js-plugins@latest/chalkboard/style.css"
    ],
    "config": { "chalkboard": { "theme": "whiteboard" } },
    "enabled": true
  }
]

• name — the global the plugin registers (e.g. RevealChalkboard)
• script / css — where to fetch it from (a CDN, or a file served by Jupyter)
• config — merged into Reveal's configuration
• enabled — set false to keep an entry without loading it

A plugin that fails to load is reported in the browser console and skipped, so an unreachable CDN can never stop a lecture from starting.

Chalkboard is preconfigured (disabled by default): enable it and press C to annotate the current slide or B for a full chalkboard — ideal for working through an example live. Drag to draw, right-drag to erase, DEL to clear.

Animating math (MathJax 4)

sliveshow replaces JupyterLab's default math renderer with MathJax 4 SVG output, so every formula in a markdown cell is an SVG that can be animated like any other — in the slideshow and in the notebook view.

Select formula parts either by their TeX source via g[data-latex='x'] (shortened to mj['x']), or tag them explicitly with \class{name}{...} / \cssId{id}{...} and select with .name / #id:

## The Gaussian integral

<div data-animate>

$$\class{lhs}{\int_{0}^{\infty} e^{-x^2}\,dx} = \class{rhs}{\frac{\sqrt{\pi}}{2}}$$

<!--
{
  "animation": [
    { "element": ".lhs", "modifier": "opacity", "parameters": [0.15], "duration": 600 },
    { "element": ".lhs", "modifier": "opacity", "parameters": [1], "duration": 600 },
    { "element": ".rhs", "modifier": "opacity", "parameters": [0.15], "duration": 600 },
    { "element": ".rhs", "modifier": "opacity", "parameters": [1], "duration": 600 }
  ],
  "loop": true
}
-->
</div>

Math inside a data-animate block is typeset by MathJax 4 when the block is injected (both in the notebook and when the slideshow starts).

Note: sliveshow disables the built-in @jupyterlab/mathjax-extension (MathJax 3) while installed, so its MathJax 4 SVG typesetter is the one JupyterLab uses. Uninstalling sliveshow restores the default renderer.

Contributing

Development install

You will need Node.js and JupyterLab installed.

# Clone the repo
git clone https://github.com/AlyanAamirAhmedani/sliveshow
cd sliveshow

# Install in development mode
pip install -e "."

# Link with JupyterLab
jupyter labextension develop . --overwrite

# Build TypeScript
jlpm build

Watch mode (auto-rebuild on save):

# Terminal 1
jlpm watch

# Terminal 2
jupyter lab

Development uninstall

pip uninstall sliveshow

Remove the symlink created by jupyter labextension develop:

jupyter labextension list  # find labextensions folder
# remove the sliveshow symlink from that folder

Packaging

See RELEASE.