canvasxpress-anywidget

CanvasXpress interactive charts as notebook widgets, built on anywidget. One package renders in Jupyter, JupyterLab, Google Colab, VS Code notebooks and marimo — anywhere the anywidget protocol is supported.

The CanvasXpress library (CSS + JS) is loaded from the CanvasXpress CDN in the browser, so there is no JavaScript build step and nothing to bundle.

Install

pip install canvasxpress-anywidget

Use

from canvasxpress_anywidget import CanvasXpress

CanvasXpress(
    data={
        "y": {
            "vars": ["Revenue"],
            "smps": ["Q1", "Q2", "Q3", "Q4"],
            "data": [[10, 14, 9, 17]],
        }
    },
    config={"graphType": "Bar", "title": "Quarterly Revenue"},
)

data, config and events are exactly the arguments of new CanvasXpress({...}) in JavaScript, so any spec from the gallery, the R / Python packages, or the Quarto shortcode works unchanged.

From a shared spec file

The {"data", "config", "events"} JSON object used across CanvasXpress hosts loads directly:

import json
from canvasxpress_anywidget import CanvasXpress

spec = json.load(open("chart.json"))
CanvasXpress.from_spec(spec)

Update in place

Assigning to an attribute re-renders the chart without creating a new cell output:

w = CanvasXpress(data=..., config={"graphType": "Bar"})
w  # display it
w.config = {**w.config, "graphType": "Line"}   # morphs to a line chart

Parameters

License

Dual-licensed. The core library is LGPL-3.0-or-later; commercial terms are available — see https://canvasxpress.org/license.html.