ReactComponent#

Open this notebook in Jupyterlite | Download this notebook from GitHub (right-click to download).


import panel as pn

pn.extension()

ReactComponent simplifies the creation of custom Panel components by allowing you to write standard React code without the need to pre-compile or requiring a deep understanding of Javascript build tooling.

import panel as pn
import param

from panel.custom import ReactComponent


class CounterButton(ReactComponent):

    value = param.Integer()

    _esm = """
    export function render({model}) {
      const [value, setValue] = model.useState("value");
      return (
        <button onClick={e => setValue(value+1)}>
          count is {value}
        </button>
      )
    }
    """

CounterButton()

Note

ReactComponent extends the JSComponent class, which allows you to create custom Panel components using JavaScript.

ReactComponent bears similarities to AnyWidget and IpyReact, but ReactComponent is specifically optimized for use with Panel and React.

If you are looking to create custom components using Python and Panel component only, check out Viewer.

API#

ReactComponent Attributes#

  • _esm (str | PurePath): This attribute accepts either a string or a path that points to an ECMAScript module. The ECMAScript module should export a render function which returns the HTML element to display. In a development environment such as a notebook or when using --autoreload, the module will automatically reload upon saving changes. You can use JSX and TypeScript. The _esm script is transpiled on the fly using Sucrase. The global namespace contains a React object that provides access to React hooks.

  • _importmap (dict | None): This optional dictionary defines an import map, allowing you to customize how module specifiers are resolved.

  • _stylesheets (List[str | PurePath] | None): This optional attribute accepts a list of CSS strings or paths to CSS files. It supports automatic reloading in development environments.

You may specify a path to a file as a string instead of a PurePath. The path should be specified relative to the file its specified in.

render Function#

The _esm attribute must export the render function. It accepts the following parameters:

  • model: Represents the Parameters of the component and provides methods to add (and remove) event listeners using .on and .off, render child React components using .get_child, get a state hook for a parameter value using .useState and to .send_event back to Python.

  • view: The Bokeh view.

  • el: The HTML element that the component will be rendered into.

Any React component returned from the render function will be appended to the HTML element (el) of the component.

State Hooks#

The recommended approach to build components that depend on parameters in Python is to create useState hooks by calling model.useState('<parameter>'). The model.useState method returns an array with exactly two values:

  1. The current state. During the first render, it will match the initialState you have passed.

  2. The set function that lets you update the state to a different value and trigger a re-render.

Using the state value in your React component will automatically re-render the component when it is updated.

Callbacks#

The model.on and model.off methods allow registering event handlers inside the render function. This includes the ability to listen to parameter changes and register lifecycle hooks.

Change Events#

The following signatures are valid when listening to change events:

  • .on('<parameter>', callback): Allows registering an event handler for a single parameter.

  • .on(['<parameter>', ...], callback): Allows adding an event handler for multiple parameters at once.

  • .on('change:<parameter>', callback): The change: prefix allows disambiguating change events from lifecycle hooks should a parameter name and lifecycle hook overlap.

The change: prefix allows disambiguating change events from lifecycle hooks should a parameter name and lifecycle hook overlap.

Bidirectional Events#

JS -> Python#
  • .send_event('<name>', DOMEvent): Allows sending browser DOMEvent to Python and associating it with a name. An event handler can be registered by name with the .on_event method or by implementing a _handle_<name> method on the class.

  • .send_msg(data): Allows sending arbitrary data to Python. An event handler can be registered with the .on_msg(callback) method on the Python component or by implementing a _handle_msg method on the class.

Python -> JS#
  • ._send_event(ESMEvent, data=msg): Allows sending arbitrary data to the frontend, which can be observed by registering a handler with .on('msg:custom', callback).

Lifecycle Hooks#

  • .on('after_layout', callback): Called whenever the layout around the component is changed.

  • .on('after_render', callback): Called once after the component has been fully rendered.

  • .on('resize', callback): Called after the component has been resized.

  • .on('remove', callback): Called when the component view is being removed from the DOM.

The lifecycle: prefix allows disambiguating lifecycle hooks from change events should a parameter name and lifecycle hook overlap.

Usage#

Styling with CSS#

Include CSS within the _stylesheets attribute to style the component. The CSS is injected directly into the component’s HTML.

import panel as pn
import param

from panel.custom import ReactComponent


class CounterButton(ReactComponent):

    value = param.Integer()

    _stylesheets = [
        """
        button {
            background: #0072B5;
            color: white;
            border: none;
            padding: 10px;
            border-radius: 4px;
        }
        button:hover {
            background: #4099da;
        }
        """
    ]

    _esm = """
    export function render({ model }) {
      const [value, setValue] = model.useState("value");
      return (
        <button onClick={e => setValue(value+1)}>
          count is {value}
        </button>
    );
    }
    """

CounterButton()

Send Events from JavaScript to Python#

Events from JavaScript can be sent to Python using the model.send_event method. Define a handler in Python to manage these events. A handler is a method on the form _handle_<name-of-event>(self, event):

import panel as pn
import param

from panel.custom import ReactComponent

pn.extension()

class EventExample(ReactComponent):

    value = param.Parameter()

    _esm = """
    export function render({ model }) {
    return (
        <button onClick={e => model.send_event('click', e) }>
          Click me
        </button>
      );
    }
    """

    def _handle_click(self, event):
        self.value = event.data

button = EventExample()
event_json = pn.pane.JSON(button.param.value)

pn.Column(button, event_json)

You can also define and send arbitrary data using the .send_msg() API and by implementing a _handle_msg method on the component:

import datetime

import panel as pn
import param

from panel.custom import ReactComponent


class CustomEventExample(ReactComponent):

    value = param.String()

    _esm = """
    function send_event(model) {
      const currentDate = new Date();
      model.send_msg(currentDate.getTime())
    }

    export function render({ model }) {
    return (
        <button onClick={e => send_event(model) }>
          Click me
        </button>
    );
    }
    """

    def _handle_msg(self, msg):
        unix_timestamp = msg/1000
        python_datetime = datetime.datetime.fromtimestamp(unix_timestamp)
        self.value = str(python_datetime)

button = CustomEventExample()

pn.Column(button, button.param.value)

Dependency Imports#

JavaScript dependencies can be directly imported via URLs, such as those from esm.sh.

import panel as pn

from panel.custom import ReactComponent


class ConfettiButton(ReactComponent):

    _esm = """
    import confetti from "https://esm.sh/canvas-confetti@1.6.0";

    export function render() {
      return (
        <button onClick={e => confetti()}>
          Click Me
        </button>
      );
    }
    """

ConfettiButton()

Use the _importmap attribute for more concise module references.

import panel as pn

from panel.custom import ReactComponent


class ConfettiButton(ReactComponent):
    _importmap = {
        "imports": {
            "canvas-confetti": "https://esm.sh/canvas-confetti@1.6.0",
        }
    }

    _esm = """
    import confetti from "canvas-confetti";

    export function render() {
      return (
        <button onClick={(e) => confetti()}>
          Click Me
        </button>
      );
    }
    """

ConfettiButton()

See import map for more info about the import map format.

External Files#

You can load JSX and CSS from files by providing the paths to these files.

Create the file counter_button.py.

from pathlib import Path

import param
import panel as pn

from panel.custom import ReactComponent

pn.extension()

class CounterButton(ReactComponent):

    value = param.Integer()

    _esm = "counter_button.jsx"
    _stylesheets = [Path("counter_button.css")]

CounterButton().servable()

Now create the file counter_button.jsx.

export function render({ model }) {
  const [value, setValue] = model.useState("value");
  return (
    <button onClick={e => setValue(value+1)}>
      count is {value}
    </button>
  );
}

Now create the file counter_button.css.

button {
    background: #0072B5;
    color: white;
    border: none;
    padding: 10px;
    border-radius: 4px;
}
button:hover {
    background: #4099da;
}

Serve the app with panel serve counter_button.py --autoreload.

You can now edit the JSX or CSS file, and the changes will be automatically reloaded.

  • Try changing count is {value} to COUNT IS {value} and observe the update.

  • Try changing the background color from #0072B5 to #008080.

Displaying A Single Child#

You can display Panel components (Viewables) by defining a Child parameter.

Lets start with the simplest example

import panel as pn

from panel.custom import Child, ReactComponent

class Example(ReactComponent):

    child = Child()

    _esm = """
    export function render({ model }) {
      return <button>{model.get_child("child")}</button>
    }
    """

Example(child=pn.panel("A **Markdown** pane!"))

If you provide a non-Viewable child it will automatically be converted to a Viewable by pn.panel:

Example(child="A **Markdown** pane!")

If you want to allow a certain type of Panel components only you can specify the specific type in the class_ argument.

import panel as pn

from panel.custom import Child, ReactComponent

class Example(ReactComponent):

    child = Child(class_=pn.pane.Markdown)

    _esm = """
    export function render({ model }) {
      return <button>{model.get_child("child")}</button>
    }
    """

Example(child=pn.panel("A **Markdown** pane!"))

The class_ argument also supports a tuple of types:

    child = Child(class_=(pn.pane.Markdown, pn.pane.HTML))

Displaying a List of Children#

You can also display a List of Viewable objects using the Children parameter type:

import panel as pn

from panel.custom import Children, ReactComponent

class Example(ReactComponent):

    objects = Children()

    _esm = """
    export function render({ model }) {
      return <div>{model.get_child("objects")}</div>
    }"""


Example(
    objects=[
        pn.panel("A **Markdown** pane!"),
        pn.widgets.Button(name="Click me!"),
        {"text": "I'm shown as a JSON Pane"}
    ]
)

You can change the item_type to a specific subtype of Viewable or a tuple of Viewable subtypes.

Using React Hooks#

The global namespace also contains a React object that provides access to React hooks. Here is an example of a simple counter button using the useState hook:

import panel as pn

from panel.custom import ReactComponent

class CounterButton(ReactComponent):

    _esm = """
    let { useState } = React;

    export function render() {
      const [value, setValue] = useState(0);
      return (
        <button onClick={e => setValue(value+1)}>
          count is {value}
        </button>
      );
    }
    """

CounterButton()

References#

Tutorials#

How-To Guides#

Reference Guides#


Open this notebook in Jupyterlite | Download this notebook from GitHub (right-click to download).