graphviz-anywidget

Interactive Graphviz visualization widget for Jupyter notebooks using anywidget. Graphviz is provided via WASM (hpcc-js-wasm) and the rendering is done using graphvizsvg and d3-graphviz, inspired by the VS Code extension Graphviz Interactive Preview.

https://github.com/user-attachments/assets/74cf39c5-2d64-4c98-b3ee-cf7308753da6

Features

• 🎨 Interactive SVG visualization of Graphviz DOT graphs
• 🔍 Search functionality with regex support
• 🎯 Node and edge highlighting
• ↔️ Directional graph traversal
• 🔄 Zoom reset functionality
• 📱 Responsive design
• 🎨 Smooth animations and transitions
• 💻 Works in JupyterLab and Jupyter Notebook

Installation

pip install graphviz-anywidget

or with uv:

uv add graphviz-anywidget

Usage

from graphviz_anywidget import graphviz_widget

# Create a widget with a DOT string
dot_source = """
digraph {
    a -> b;
    b -> c;
    c -> a;
}
"""
widget = graphviz_widget(dot_source)
widget

Features

1. Search: Use the search box to find nodes and edges

   • Supports exact match, substring, and regex search
   • Case-sensitive option available

2. Direction Selection: Choose how to traverse the graph

   • Bidirectional: Show connections in both directions
   • Downstream: Show only outgoing connections
   • Upstream: Show only incoming connections
   • Single: Show only the selected node

3. Zoom Reset: Reset the graph to its original position and scale

API

graphviz_widget

def graphviz_widget(dot_source: str = "digraph { a -> b; }") -> widgets.VBox:
    """Create an interactive Graphviz widget.

    Parameters
    ----------
    dot_source
        The DOT string representing the graph

    Returns
    -------
    widgets.VBox
        The widget containing the graph and controls
    """

Dependencies

• anywidget
• ipywidgets
• graphvizsvg (npm package)
• d3-graphviz (npm package)
• hpcc-js-wasm (npm package)

Development

We recommend using uv for development. It will automatically manage virtual environments and dependencies for you.

uv run jupyter lab example.ipynb

Alternatively, create and manage your own virtual environment:

python -m venv .venv
source .venv/bin/activate
pip install -e ".[dev]"
jupyter lab example.ipynb

The widget front-end code bundles it's JavaScript dependencies. After setting up Python, make sure to install these dependencies locally:

npm install

While developing, you can run the following in a separate terminal to automatically rebuild JavaScript as you make changes:

npm run dev

Open example.ipynb in JupyterLab, VS Code, or your favorite editor to start developing. Changes made in js/ will be reflected in the notebook.

Implementation Notes

• The WASM binary is embedded in the JavaScript bundle as base64
• We override the fetch API to intercept WASM file requests
• Web Worker mode is disabled to ensure consistent WASM loading
• This approach works in any Jupyter environment without needing a separate file server

License

MIT

Credits

Built with:

• anywidget
• graphvizsvg
• d3-graphviz
• The WASM binary comes from @hpcc-js/wasm (via d3-graphviz)
• Inspired by the VS Code extension Graphviz Interactive Preview.