Hey! this project moved to area 17, go wild there.

After looking over the choices for autocomplete libs on npm i felt that the choices for typeahead/autocomplete UI lacking one or more in the following areas:

• vetted acessibility standards
• simplicity, with clear documentation
• a modern event bus system
• flexible styling
• doesn't jquery (im taking about you typeahead.js),
• small footprint (<4kb)

This library in essence is porting all the good work done at on https://github.com/alphagov/accessible-autocomplete from their react component. The end result will be 100% a11y parity with https://github.com/alphagov/accessible-autocomplete :)

Any features beyond that, id be happy to include, leave an issue and some reasoning behind it.

While this library is every thing I need usually, it might not be everything you need. If you need image rendering, custom html injections in the list, need sub-sections, sub-titles ect, this isnt it. Checkout @algolia/autocomplete-js This beast will do it all.

Have you every tried to search "autocomplete" on github or npm. Its the wild west out there

On to the docs!

To initialize flotsam in a project import it in the desired file!

import flotsam from 'flotsam-autocomplete'

You'll also need some very basic styles, imported however you like to do that in your project:

import './node_modules/flotsam-autocomplete/dist/flotsam.css'

and initailize it on the page with

const typeahead = new flotsam({
    el: document.querySelector('input'),
    data: [!!data here!!]
})

Now, when interacting with the input flotsom will render an absolutely positioned box under the input for best results set the parent element to relaitive position so flotsom knows where to go!

The rendered html should look a lil' something like this

<div class="autocomplete-modal">
    <div class="autocomplete-modal__inner">
        <ul class="autocomplete-modal__list" role="combobox">
            <li role="option">Rendered value 1</li>
            <li role="option">Rendered value 2</li>
            <li role="option">Rendered value 3</li>
            ... ect
        </ul>
    </div>
</div>

There are no opinionated styles on flotsom, thats up to you!

// a complex instance of flotsam
const typeahead = new flotsam({
    // input element you want to attach to
    el: document.querySelector('input'),

    // static data formatted in an array, this is the data that will render on interaction
    data: ['lorem ipsum', 'lipsum', 'hello world', 'foo', 'bar', 'foo bar'],

    // the ajax implimentation -
    // this allows you flexablity to pass your own promise to the render function
    // textValue is the inputed value :)
    onAjax: (textValue) => {
        return axios
            .get(`https://dummyjson.com/users/${textValue}`)
            .then((d) => {
                // after we get data we format it into an array and pass it back to flotsom THANKS!
                const users = d.data.users
                return users.map((user) => {
                    return user.firstName
                })
            })
    }

    // this sets the minimum count of input characters before we will render the box
    minChars: 2,

    // this is an interaction that google input does that i thought i should include:
    // when key-ing up or down, the input will fill with the previewed string
    inputPreview: true,

})

Flotsom comes with an event bus to trigger your own crazy ideas! When you have access to the const typeahead variable you can listen to events with it

typeahead.on('init', (instanceData) => {
    console.log('modal init!')
})

typeahead.on('openModal', (instanceData) => {
    console.log('modal opened')
})

typeahead.on('closeModal', (instanceData) => {
    console.log('modal closed')
})

typeahead.on('selectKey', (instanceData) => {
    console.log('modal item keyed')
})

typeahead.on('disabled', (instanceData) => {
    console.log('modal item keyed')
})

InstanceData has access to the current, modal, input, options (readonly), and any other useful information i could think of!

Flotsom also has some manual controls if you need them

// force close the flotsom
typeahead.triggerClose()

// disables flotsom
typeahead.Triggerdisable()

Flotsom hijacks several keys while opened to provide a holistic feature set much like the google autocomplete UI.

• Arrow Up and Arrow Down keys allow you to select previous and next items while open
• Tab will set the value, close the modal, but NOT submit (google-like)
• Escape will set the value, close the modal, but NOT submit (google-like)
• Enter will set the value, close the modal AND submit (google-like)

If above was a bit too conversational for you, here are tables of the above information! (TODO!)

I'm not done with this project! While the goal is to keep it lean i keep having shower ideas. Here's what's in for sure:

• 100% a11y parity with https://github.com/alphagov/accessible-autocomplete
• Custom empty state option
• Assistive text for screen readers (using aira-describeby)
• Optional empty state modal
• Outside click close (with option to disable that)

• Custom filterBy and sortBy functions before we generate autocomplete list items
• Custom regex for the text highlight on autocomplete (make you do all the work!)
• Allow for custom html to be put in the dropdown (this would make some current features a bit more annoying - more docs ect)