@crui/core

The aim of this package is to be the foundation on top of which the rest is built. For this reason, we value composability and performance over everything else.

@crui/core has three main concepts:

Components: represent a piece of UI
Elements: any function that generates a Component
Setups: used by Elements to configure the many aspects of a Component

All three are based on functions and therefore are immutable and composable by definition.

Setups

child: append a child to an Element
children: append many children to an Element
ctext: append a Text. Equivalent of child + text
on: setup an event handler
onClick: setup a 'click' event handler
props: setup Element's properties
style: setup inline styling
cleanup: hook to cleanup resources once the Element is disposed

We also provide two utilities to combine setups:

sc2: SetupCombine2, combine two setups into one
sc: SetupCombine, combine a list of setups

Elements

h: the main building block for HTML elements
text: create a Text element, which is the way to securely display texts
br: the br tag
ht: shortcut for h + children setup
empty: an empty element, useful as a placeholder for dynamic content
e: escape hatch to create special elements

We also have higher-order elements, ie: elements that build on top of other elements:

setup: apply a new setup to a Component
useContext: allow the use of Context to generate a new Component
withContext: adapt Context to a Component

Mounting

Generating components will do nothing at all, remember that they are just definition of what needs to be done. To materialize a definition, we need to mount it.

The most common mounting operation is to render it in the browser DOM:

import { text } from '@crui/core/elems/text'
import { mount } from '@crui/core/dom/browser'

mount(
    document.getElementById('root')!,
    text('Hello World'),
    {}
)

The first element is just a valid HTML Node to which we will append the final result of the materialized component, which in this case is a simple Text node.
The third and final argument is the Context.

Supposing our server rendered an HTML with a body like this:

<body>
    <div id="root"></div>
</body>

Once the above typescript code runs in the browser, it will become:

<body>
    <div id="root">Hello World</div>
</body>

Learn by Examples

All the following example will omit the mount part.

Unless specified otherwise, all function can be imported from @crui/core. Each function can also be imported singularly to reduce boundle size; have a look at source code for more information.

Element with Text

import { h, child, text } from '@crui/core'
import { mount } from '@crui/core/dom/browser'

const title = h('h1', child(
    text('<Hello World>')
))

This will generate an HTML equivalent to:

<h1>&lt;Hello World&gt;</h1>

The title component could be more easily written as:

import { ht, text } from '@crui/core'

const title = ht('h1', '<Hello World>')

List of elements

import { hc } from '@crui/core/elems/children'
import { ht } from '@crui/core/elems/ht'

const list = hc('ul', [
    ht('li', 'a')
    ht('li', 'b')
    ht('li', 'c')
])

Shouter

const shouter = (str: string) => (
    h('button', sc2(
        child(text(str)),
        onClick(() => alert(str.toUpperCase()))
    ))
)

const shouters = hc('div', [
    shouter('a'),
    shouter('b'),
    shouter('c')
])

As a reminder:

sc2: combine two Setup into one, in this case child and onClick
child: append an element
onClick: setup a click event handler

Therefore it's equivalent to:

<div>
    <button onClick="alert('a'.toUpperCase())">a<button>
    <button onClick="alert('b'.toUpperCase())">b<button>
    <button onClick="alert('c'.toUpperCase())">c<button>
</ul>

CRUI is highly based on functions, so it should be no surprise that the way to generate sharable and reusable Components is by wrapping Elements in just another function :)

The best part is that we can use all normal means of abstraction, eg:

const shouter = (str: string) => (...)
const shouters = hc('div', ['a', 'b', 'c'].map(shouter))

Simple Form

This example further demonstrate how easy is it to abstract concepts and how readable it can be.

const form = h('form', sc2(
    props({ action: '/signup', method: 'POST' }),
    children([
        field('First name', 'text', 'first-name'),
        field('Last name', 'text', 'last-name'),
        field('E-Mail', 'email', 'email'),
        field('Password', 'password', 'pass'),
        field('Confirm', 'password', 'confirm'),
    ])
))

type Type = 'text'|'email'|'password'

const field = (text: string, type: Type, name: string) => 
    hc('div', [
        label(name, text)
        input(type, name)
    ])

const input = (type: Type, name: string) => 
    h('input', props({ id: name, name, type }))

const label = (target: string, text: string) => 
    h('label', sc([
        props({ labelFor: target }),
        style({ fontWeight: 'bold', marginRight: '1rem' }),
        ctext(text)
    ]))

Equivalent to:

<form action="/signup" method="POST">
    <div>
        <label for="first-name">First name</label>
        <input type="text" id="first-name" name="first-name">
    </div>
    <div>
        <label for="last-name">Last name</label>
        <input type="text" id="last-name" name="last-name">
    </div>
    <div>
        <label for="email">E-Mail</label>
        <input type="email" id="email" name="email">
    </div>
    <div>
        <label for="pass">Password</label>
        <input type="password" id="pass" name="pass">
    </div>
    <div>
        <label for="confirm">Confirm</label>
        <input type="password" id="confirm" name="confirm">
    </div>
</form>

What is a Component?

A Component is just a function that receives a DOM and Context and returns a rendered Node plus cleanup functions and meta. This is all we need to make a composable interface.

Abstract DOM

As we have seen, Components receive a DOM, but why? Can't they just use document directly?

Given that we want to be composable, bounding ourselves to a global variable and a specific environment makes things a lot harder. What we do instead is to abstract out what a DOM and Node is and do so while making it totally transparent to the library user.

The first big benefit is that each Component can be tested in isolation and in a controlled environment:

import { e } from '@crui/core/elems/elem'

type Node = { tag: string }
const mockDOM = {
    create(tag): Node {
        return { tag }
    }
    // ... mock all other methods ...
}

describe(e, () => {
    it('generates a node with the right tag', () => {
        expect(
            e('div')(mockDOM, {})
        ).toEquals({ tag: 'div' })
    })
})

Given that all functions rely on the passed DOM, you can already see how easy would be to support snapshot testing, which is something will be provided as a library. For this reason is very important that each Element respects the contract and never use global or environment specific code.

Testing is not the only area were this abstraction will be useful: SSR (Server Side Rendering) should be possible, even though it's not as straightforward once reactivity is mixed in the bag.

Context API

Sometimes we need to pass a particular piece of information down in the components tree. We could thread this information through all the components, but it becomes unpractical and tedious very soon.
An example is internationalization (i18n). Let's say we need a component that can render a translated message given a key:

type I18n = {
    t: (key: string) => string
}
const t = (i18n: I18n, key: string) => 
    h('span', sc2(
        props({ className: 'i18n' }),
        ctext(msg)
    ))

We would then pass down the i18n object each time we need to use it, eg:

const homeTitle = (i18n: I18n) => 
    hc('h1', t(i18n, 'home.title'))

Even when you don't really care about translations per se:

const home = (i18n: I18n) => hc(
    homeTitle(i18n),
    homeContent
)

This works, but it's tedious to write and the number of args needed could explode depending on how much central state we have.

A better approach would be to let only the t component care about i18n and nothing else. To do so, we can use the Context API: a piece of information that will be automatically threaded through our elements by CRUI.

const t = (key: string) => useContext(
    ({ t }) => (
        h('span', sc2(
            props({ className: 'i18n' }),
            ctext(t(key))
        ))
    )
)

The component will now capture the context, which in this case is just I18n, use it to get the translated key and then pass it down to text component.

Now any other component don't need to know about i18n anymore and code is cleaner:

const homeTitle = hc('h1', [t('home-title')])

const home = hc(
    homeTitle,
    homeContent
)

The actual Context is provided on mounting:

import { mount } from '@crui/core/dom/browser'

mount(
    document.getElementById('root'),
    home,
    makeI18n()
)

One nice gain of using typescript, is that now mount will complain if we don't pass the right Context, telling as exactly what is needed to satisfy the whole tree.

@crui/core

@crui/core

Setups

Elements

Mounting

Learn by Examples

Element with Text

List of elements

Shouter

Simple Form

What is a Component?

Abstract DOM

Context API

Readme

Keywords

Package Sidebar

Install

Repository

Weekly Downloads

Version

License

Unpacked Size

Total Files

Last publish

Collaborators

@crui/core

@crui/core

Setups

Elements

Mounting

Learn by Examples

Element with Text

List of elements

Shouter

Simple Form

What is a Component?

Abstract DOM

Context API

Readme

Keywords

Package Sidebar

Install

Repository

DownloadsWeekly Downloads

Version

License

Unpacked Size

Total Files

Last publish

Collaborators

Weekly Downloads