Reusing JupyterLab UI#

The @jupyterlab/ui-components package provides UI elements that are widely used in JupyterLab core, and that can be reused in your own extensions.

For example, all of the icons in JupyterLab core can be reused via LabIcon. You can also use LabIcon to create your own custom icons that will be able to automatically change color to match the current JupyterLab theme.

LabIcon - set up and render icons#

LabIcon is the icon class used by JupyterLab, and is part of the new icon system introduced in JupyterLab v2.0.

How JupyterLab handles icons#

The @jupyterlab/ui-components package provides icons to the rest of JupyterLab, in the form of a set of LabIcon instances (currently about 80). All of the icons in the core JupyterLab packages are rendered using one of these LabIcon instances.

Using the icons in your own code#

You can use any of JupyterLab icons in your own code via an import statement. For example, to use jupyterIcon you would first do:

import { jupyterIcon } from '@jupyterlab/ui-components';

How to render an icon into a DOM node#

Icons can be added as children to any div or span nodes using the icon.element(...) method (where icon is any instance of LabIcon). For example, to render the Jupyter icon you could do:

  container: elem,
  height: '16px',
  width: '16px',
  marginLeft: '2px'

where elem is any HTMLElement with a div or span tag. As shown in the above example, the icon can be styled by passing CSS parameters into .element(...). Any valid CSS parameter can be used (one catch: snake case params do have to be converted to camel case: instead of foo-bar: '8px', you’d need to use fooBar: '8px'.

How to render an icon as a React component#

Icons can also be rendered using React. The icon.react parameter holds a standard React component that will display the icon on render. Like any React component, icon.react can be used in various ways.

For example, here is how you would add the Jupyter icon to the render tree of another React component:

public render() {
  return (
    <div className="outer">
      <div className="inner">
        <jupyterIcon.react tag="span" right="7px" top="5px" />
        "and here's a text node"

Alternatively, you can just render the icon directly into any existing DOM node elem by using the ReactDOM module:

ReactDOM.render(jupyterIcon.react, elem);

If do you use ReactDOM to render, and if the elem node is ever removed from the DOM, you’ll first need to clean it up:


This cleanup step is not a special property of LabIcon, but is instead needed for any React component that is rendered directly at the top level by ReactDOM: failure to call unmountComponentAtNode can result in a memory leak.

How to create your own custom LabIcon#

You can create your own custom icon by constructing a new instance of LabIcon:

export const fooIcon = new LabIcon({
  name: 'barpkg:foo',
  svgstr: '<svg>...</svg>'

where name should be of the form “your-pkg:icon-name”, and svgstr is the raw contents of your icon’s svg file.

How to create a new LabIcon from an external svg file#

Although you can copy-and-paste an svg directly into the LabIcon constructor, the best practice is to keep the svg for each of your icons in its own separate svg file. You will need to have an svg.d.ts file at the root of your project’s src folder:

// svg.d.ts

declare module '*.svg' {
  const value: string;
  export default value;

You can then import the contents of an svg file:

import fooSvgstr from 'path-to-your/foo.svg';

export const fooIcon = new LabIcon({
  name: 'barpkg:foo',
  svgstr: fooSvgstr

Sync icon color to JupyterLab theme#

Example svgs with class annotation can be found in ui-components/style/icons

You can ensure that the colors of your custom LabIcon sync up to the colors of the current JupyterLab theme by adding appropriate class annotations to each colored element of your icon’s svg.

In other words, each element of your svg that a fill="..." or a stroke="..." property should also have a class="jp-icon<whatever>" property.

Available icon classes#

Icon-related CSS classes are defined in ui-components/style/icons.css

All colors shown are for the standard light/dark theme, mouse over for hex values.
jp-iconX: contrast to theme background#
  • jp-icon0: #111 / #fff
  • jp-icon1: #212121 / #fff
  • jp-icon2: #424242 / #eee
  • jp-icon3: #616161 / #bdbdbd
  • jp-icon4: #757575 / #757575

Most one-color icons in JupyterLab (including the sidebar and toolbar icons) are colored using the jp-icon3 class.

For light/dark themes, jp-icon0 corresponds to the darkest/lightest background color, while jp-icon1 is somewhat lighter/darker, and so forth.

jp-icon-accentX: match to theme background#
  • jp-icon-accent0: #fff / #111
  • jp-icon-accent1: #fff / #212121
  • jp-icon-accent2: #eee / #424242
  • jp-icon-accent3: #bdbdbd / #616161
  • jp-icon-accent4: #757575 / #757575

For light/dark themes, jp-icon-accent0 corresponds to the lightest/darkest background color, while jp-icon-accent1 is somewhat darker/lighter, and so forth.

Adding classes to a one-color icon#

For most simple, one-color icons, it is desirable for the icon’s color to strongly contrast with that of the application’s background. You can achieve this using one of the jp-iconX classes.

Example: check icon

svg source:

<svg xmlns="" width="100" viewBox="0 0 24 24">
  <path class="jp-icon3" fill="#616161" d="M9 16.17L4.83 12l-1.42 1.41L9 19 21 7l-1.41-1.41z"/>

rendered icon:

Adding classes to a multi-colored icon#

For more complex icons, each element that needs to match the background should be annotated with a jp-icon-accentX class, while each element that needs to contrast with the background should be annotated with a jp-iconX class.

Example: close-circle icon

svg source:

<svg xmlns="" width="100" viewBox="0 0 24 24">
  <circle class="jp-icon3" fill="#616161" cx="12" cy="12" r="11"/>
  <rect class="jp-icon-accent0" fill="#fff" height="18" width="2" x="11" y="3" transform="rotate(315, 12, 12)"/>
  <rect class="jp-icon-accent0" fill="#fff" height="18" width="2" x="11" y="3" transform="rotate(45, 12, 12)"/>

rendered icon:


Icon handling in Jupyterlab#

Pre JupyterLab 2.0, most icons were created using the icons-as-css-background pattern:

  • Set up the icon’s svg as a background-image in CSS:

    /* CSS */
    .jp-FooIcon {
      background-image: url('path-to-your/foo.svg');
  • Add the icon to the DOM by constructing an otherwise empty DOM node with the appropriate class:

    // typescript
    const e = document.createElement('div');
    e.className = 'jp-FooIcon';

What you end up with is a single DOM node that has the “foo” icon as a background image.

Post JupyterLab 2.0, nearly all icons in core are now created using LabIcon and the icons-as-inline-svg pattern:

  • Construct a new instance of LabIcon from the icon’s name and svg:

    // typescript
    // svgstr is the raw contents of an icon's svg file
    export const fooIcon = new LabIcon({
      name: 'barpkg:foo',
      svgstr: '<svg>...</svg>'
  • Add the icon to the DOM using the appropriate property of your LabIcon instance (either LabIcon.element() to directly create a DOM node, or LabIcon.react to get the icon as a react component):

    // typescript
    const e = fooIcon.element();

What you end up with is a DOM node (by default a ‘div’) that has an inline svg node as a child.

background-image vs inline svg#

The big limitation of the old icon-as-css-background pattern is that svg images rendered as background-image are invisible to CSS. On the other hand, an icon rendered as an inline svg node is fully exposed to the CSS. This allows us to dynamically change icon styling as needed simply by modifying our CSS. Most importantly, this allows us to recolor icons according to Jupyterlab’s current theme.