Organizing multiple web components with seperation of concerns
Question
I'm building a small application, and planning to use web components (rather than use a UI library). I don't plan to use any bundlers etc., as this is going to be a small personal site.
I would like to store each web component in a separate ES6 JS module file, and here is an example of my setup:
hello-pl.net.mjs
export class HelloPlanet extends HTMLElement {
constructor() {
super();
}
connectedCallback(){
// Attach a shadow root to the element.
let shadowRoot = this.attachShadow({mode: 'open'});
let planet = this.getAttribute('planet')
shadowRoot.innerHTML = `<p>hello ${planet}</p>`;
}
}
hello-world.mjs
export class HelloWorld extends HTMLElement {
constructor(){
super()
}
connectedCallback(){
// Attach a shadow root to the element.
let shadowRoot = this.attachShadow({mode: 'open'});
['Mercury','Venus','Earth','Mars','Jupiter','Saturn','Uranus','Neptune'].forEach(planet=>{
let el = document.createElement('hello-planet')
el.setAttribute('planet', planet);
shadowRoot.appendChild(el)
})
}
}
main.mjs
// ordering of imports does not matter
import {HelloPlanet} from './hello-planet.mjs';
import {HelloWorld} from './hello-world.mjs';
customElements.define('hello-world', HelloWorld);
customElements.define('hello-planet', HelloPlanet);
// this will typically be handled by a client side router (e.g. Navigo)
let el = document.createElement('hello-world');
let body = document.querySelector('body');
body.appendChild(el);
index.html (only calls main.mjs, browser will download the rest of the scripts)
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta http-equiv="X-UA-Compatible" content="IE=edge">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<script src="main.mjs" type="module"></script>
<title>Web components test</title>
</head>
<body>
</body>
</html>
Questions:
- I've not seen this approach in any examples I've encountered so far, so wondering if is a good approach? Or there is a better approach than this in terms of organizing web components.
- When I need template+styles, how can this approach be extended to read those from different files ie html and css in separate files (so we have separation of concerns)? I've seen this but not sure how to adapt to my kind of setup. I've also gone through this - but it seems too complex already, and not even sure if it can handle complex scripting.
Thank you!
3 answers
solution1
1 ACCPTED 2022-10-04 12:13:28
- I've not seen this approach in any examples I've encountered so far, so wondering if is a good approach? Or there is a better approach than this in terms of organizing web components.
It's perfectly fine. Creating your elements programmatically has many advantages, mainly there is no need to query your own shadow root to get access to child elements/components. If need be, you can directly hold references or even create those in class properties, eg:
export class HelloWorld extends HTMLElement {
planets = ['Mercury', 'Venus', 'Earth', 'Mars', 'Jupiter', 'Saturn', 'Uranus', 'Neptune']
.map(
planet => Object.assign(document.createElement('hello-planet'), { planet })
)
)
constructor() {
super().attachShadow({ mode: 'open' }).append(...this.planets);
}
}
Sidenote: Creating the shadow root can and should safely be done in the constructor.
- When I need template+styles, how can this approach be extended to read those from different files ie html and css in separate files (so we have separation of concerns)?
For CSS, we have CSS module scripts :
import styles from './hello-world.css' assert { type: 'css' }
then, in your constructor, do
constructor() {
// ...
this.shadowRoot.adoptedStylesheets.push(styles);
}
For HTML, this importing feature unfortunately is still work in progress .
solution2
0 2022-10-03 12:51:55
The way they do it in react is you import all your components into one file and then you just insert that component into a statice div with an id of root in your HTML page.
So your index file:
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta http-equiv="X-UA-Compatible" content="IE=edge">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<script src="main.mjs" type="module"></script>
<title>Web components test</title>
</head>
<body>
<div id="root"></div>
</body>
</html>
Than anything you want to display in the DOM you insert into that div via JS.
I believe this might help your problem, honestly wasn't a 100% clear on your isssue
solution3
0 2023-01-03 14:40:08
Using import assert to load the component's HTML and CSS may be the generic approach moving forward, however, browser support is still limited. To create web components with a separation of concerns and limited build tools (like bundlers) you can use standard fetch with top level await to ensure the component's bootstrapped appropriately before using.
I similarly wanted to separate concerns and minimize tooling while developing modern web components as ES modules . I've started documentation (WIP) on web component best practices that I wanted to follow while developing tts-element .
This is the relevant setup work to bootstrap the component during its import :
const setup = async () => {
const url = new URL(import.meta.url)
const directory = url.pathname.substring(0, url.pathname.lastIndexOf('/'))
const baseUrl = `${url.origin}${directory}`
const [html, css] = await Promise.all([
fetch(`${baseUrl}/template.html`).then((resp) => resp.text()),
fetch(`${baseUrl}/styles.css`).then((resp) => resp.text())
])
const parser = new DOMParser()
const template = parser
.parseFromString(html, 'text/html')
.querySelector('template') as HTMLTemplateElement
const style = document.createElement('style')
style.textContent = css
template.content.prepend(style)
return class TextToSpeech extends HTMLElement {/* implementation */}
}
Here is how the component will be exported to allow top level await to be used:
export default await setup()
Here is an example of how tts-element can be loaded in various ways (all using top level await) to control the timing of when the component is define d.
- Loading of
defined.jswill automatically register the component under the default nametext-to-speech. - Using the
namequery parameter while loadingdefined.js?name=custom-namewill register the component with the provided custom name. - Loading of
element.jswill require manual registration of the components definition, ie the consumer will be responsible for callingdefine().
Check the.network tab in the dev console to see how the HTML/CSS/JS is loaded:
<:DOCTYPE html> <html lang="en-US"> <head> <meta charset="utf-8" /> <title>tts-element combined example</title> <style> text-to-speech:not(,defined): my-tts:not(,defined): speech-synth:not(:defined) { display; none: } </style> <script type="module" src="https.//unpkg.com/tts-element/dist/text-to-speech/defined:js"></script> <script type="module" src="https.//cdn.jsdelivr.net/npm/tts-element@0.0.3-beta/dist/text-to-speech/defined?js:name=my-tts"></script> <script type="module"> import ctor from 'https.//unpkg.com/tts-element/dist/text-to-speech/element.js' customElements,define('speech-synth'. ctor) </script> </head> <body> <text-to-speech>Your run-of-the-mill text-to-speech example.</text-to-speech> <my-tts>Example using the "name" query parameter.</my-tts> <speech-synth>Example using element.js.</speech-synth> </body> </html>
The technical post webpages of this site follow the CC BY-SA 4.0 protocol. If you need to reprint, please indicate the site URL or the original address.Any question please contact:yoyou2525@163.com.