Skip to content

Latest commit

 

History

History
 
 

autocomplete-js

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
 
 
 
 
 
 
 
 
 
 

@trevoreyre/autocomplete-js

Demo

Take a look at the documentation page, and the Codepen examples.

Installation

Install the component from npm.

npm install --save @trevoreyre/autocomplete-js

Or using yarn.

yarn add @trevoreyre/autocomplete-js

You can also use the browser bundle in a script tag.

<script src="https://unpkg.com/@trevoreyre/autocomplete-js"></script>

To add the default styling for the component, include the CSS file on your page as well.

<link
  rel="stylesheet"
  href="https://unpkg.com/@trevoreyre/autocomplete-js/dist/style.css"
/>

Usage

The JavaScript component expects a fairly simple HTML structure consisting of a root container element, with an input and ul children.

<div id="autocomplete" class="autocomplete">
  <input class="autocomplete-input" />
  <ul class="autocomplete-result-list"></ul>
</div>

The constructor takes two arguments, the root, and an options object.

new Autocomplete('#autocomplete', options)

Upon initialization, the component will take care of initializing all of the proper ARIA attributes.

Arguments

Argument Type Description
root String | DOM Element (required) Either the container DOM element, or a selector for it
options Object An object of options to pass to the component. See below for more details.

root

The root argument can be either a reference to the container DOM element, or a selector for it.

// root can be a DOM element
const root = document.getElementById('autocomplete')
new Autocomplete(root, options)

// or a string selector
new Autocomplete('#autocomplete', options)

Note that if using a selector, it's expected that the selector only matches one DOM element. In the case of a selector that matches multiple elements, the autocomplete component is only attached to the first match. If you need to attach to multiple elements, you will need to loop through them manually.

// Select all of the elements
const elements = document.querySelectorAll('.autocomplete')

const search = input => {
  if (input.length < 1) {
    return []
  }
  return countries.filter(country => {
    return country.toLowerCase().startsWith(input.toLowerCase())
  })
}

// Attach autocomplete to each element
elements.forEach(el => {
  new Autocomplete(el, { search })
})

Options

Option Type Default Description
search Function (required) The search function to be executed on user input. Can be a synchronous function or a Promise.
onSubmit Function Executed on input submission
baseClass String 'autocomplete' Base class used to create classes and IDs for generated DOM elements
autoSelect Boolean false Controls whether first result should be highlighted after input
getResultValue Function For complex search results, this function is executed to get the value to display in the input
renderResult Function Override default rendering of result items

search

The search function is executed on user input. It is expected to return either an array of results to be rendered, or a Promise that resolves to an array of results.

In the simplest case, search can return an array of strings.

new Autocomplete('#autocomplete', {
  search: input => {
    if (input.length < 1) {
      return []
    }
    return countries.filter(country => {
      return country.toLowerCase().startsWith(input.toLowerCase())
    })
  },
})

The search function can also return a Promise, to make asynchronous calls to an API, for example. The return value doesn't have to be an array of strings, it can also be an array of objects. In this case, you will need to provide a getResultValue function to get the raw value from your result object to display in the input field when a user selects or submits a result.

Below is a more advanced search example showing these options.

const wikiUrl = 'https://en.wikipedia.org'
const params = 'action=query&list=search&format=json&origin=*'

new Autocomplete('#autocomplete', {
  // Search function can return a promise
  // which resolves with an array of
  // results. In this case we're using
  // the Wikipedia search API.
  search: input => {
    const url = `${wikiUrl}/w/api.php?${params}&srsearch=${encodeURI(input)}`

    return new Promise(resolve => {
      if (input.length < 3) {
        return resolve([])
      }

      fetch(url)
        .then(response => response.json())
        .then(data => {
          resolve(data.query.search)
        })
    })
  },

  // Wikipedia returns a format like this:
  //
  // {
  //   pageid: 12345,
  //   title: 'Article title',
  //   ...
  // }
  //
  // We want to display the title
  getResultValue: result => result.title,

  // Open the selected article in
  // a new window
  onSubmit: result => {
    window.open(`${wikiUrl}/wiki/${encodeURI(result.title)}`)
  },
})

onSubmit

The onSubmit function is executed when the user submits their result by either selecting a result from the list, or pressing enter/return. The function receives the selected result as an argument.

new Autocomplete('#autocomplete', {
  search: input => {
    if (input.length < 1) {
      return []
    }
    return countries.filter(country => {
      return country.toLowerCase().startsWith(input.toLowerCase())
    })
  },

  onSubmit: result => {
    alert(`You selected ${result}`)
  },
})

baseClass

The baseClass option is used to derive classes for generated DOM elements in the results list. It's also used to create IDs when necessary for use in ARIA attributes.

For example, if you passed a baseClass of 'search':

new Autocomplete('.search', { baseClass: 'search' })

You would get the following DOM (simplified for demonstration purposes):

<!-- The root, input, and ul element classes are unaffected by the baseClass option.
   - It's expected that you provide these classes in your HTML. The baseClass is used
   - to generate an ID for the ul element only if you didn't provide one.
   -->
<div class="search">
  <input class="search-input" />
  <ul id="search-result-list-1" class="search-result-list">
    <!-- The ID and class for result list items are generated from the baseClass option -->
    <li id="search-result-0" class="search-result">
      First result
    </li>
    <li id="search-result-1" class="search-result">
      Second result
    </li>
  </ul>
</div>

This option can be useful if you need a certain class in your DOM for styling purposes. See the section on Styling and customization for more info.

autoSelect

If the autoSelect option is set to true, the first result in the list will automatically be highlighted after user input.

new Autocomplete('#autocomplete', {
  search: input => {
    if (input.length < 1) {
      return []
    }
    return countries.filter(country => {
      return country.toLowerCase().startsWith(input.toLowerCase())
    })
  },

  autoSelect: true,
})

getResultValue

If your search function returns more complex results like an array of objects, you can use the getResultValue function to tell the autocomplete component what value to display in the results list and input. The function receives a result as an argument, and is expected to return a String to display for that result.

new Autocomplete('#autocomplete', {
  search: input => {
    if (input.length < 1) {
      return []
    }
    return animals.filter(animal => {
      return animal.name.toLowerCase().startsWith(input.toLowerCase())
    })
  },

  // Results come back in the format:
  //
  // {
  //   name: 'lion',
  //   collateral_adjectives: [
  //     'leonine',
  //     'lionish'
  //   ]
  // }
  //
  // We want to display the name
  getResultValue: result => result.name,
})

renderResult

You can use the renderResult function to override the default rendering of items in your result list. This function takes the following arguments:

  • result - The result value returned from your search function
  • props - An object containing generated attributes for the result item, which are expected to be set on your li element. The object has a custom toString function which lets you easily serialize it to a String of HTML attributes in the form attribute1="value1" attribute2="value2". This way, you don't have to worry about generating the proper IDs, classes, and ARIA attributes yourself.

The renderResult function should return either a DOM element or an HTML string.

const wikiUrl = 'https://en.wikipedia.org'
const params = 'action=query&list=search&format=json&origin=*'

new Autocomplete('#autocomplete', {
  // Search function can return a promise
  // which resolves with an array of
  // results. In this case we're using
  // the Wikipedia search API.
  search: input => {
    const url = `${wikiUrl}/w/api.php?${params}&srsearch=${encodeURI(input)}`

    return new Promise(resolve => {
      if (input.length < 3) {
        return resolve([])
      }

      fetch(url)
        .then(response => response.json())
        .then(data => {
          resolve(data.query.search)
        })
    })
  },

  // Control the rendering of the result
  // items. Let's show the title and snippet
  // from the Wikipedia results
  renderResult: (result, props) => `
    <li ${props}>
      <div class="wiki-title">
        ${result.title}
      </div>
      <div class="wiki-snippet">
        ${result.snippet}
      </div>
    </li>
  `,

  // Wikipedia returns a format like this:
  //
  // {
  //   pageid: 12345,
  //   title: 'Article title',
  //   ...
  // }
  //
  // We want to display the title
  getResultValue: result => result.title,

  // Open the selected article in
  // a new window
  onSubmit: result => {
    window.open(`${wikiUrl}/wiki/${encodeURI(result.title)}`)
  },
})

Styling and customization

To include the default styling of the autocomplete component that you see here in the docs, include the CSS file on your page.

<link
  rel="stylesheet"
  href="https://unpkg.com/@trevoreyre/autocomplete-js/dist/style.css"
/>

This styling is intentionally opinionated, however, it's relatively easy to write your own CSS if you want a different style. All positional styling is handled inline, so you don't have to worry about positioning the results list in your CSS.

You can provide the IDs and classes for the root element, input, and ul elements yourself in your HTML template. IDs and classes for the li elements are generated for you, but can be customized using the baseClass option. If you need more control than the baseClass option can provide, you can also take full control of the rendering of items in your results list using the renderResult option.

Below is an example of a typical DOM structure, and all the properties that might be relevant for styling.

<div
  id="autocomplete"
  class="autocomplete"
  data-expanded="true"
  data-loading="false"
  data-position="below"
>
  <input class="autocomplete-input" aria-expanded="true" />
  <ul id="autocomplete-result-list-1" class="autocomplete-result-list">
    <li
      id="autocomplete-result-0"
      class="autocomplete-result"
      data-result-index="0"
      aria-selected="true"
    >
      First result
    </li>
    <li
      id="autocomplete-result-1"
      class="autocomplete-result"
      data-result-index="1"
    >
      Second result
    </li>
  </ul>
</div>

There are a few data attributes that are added to the root element as well to show the current state of the component.

  • data-expanded="true" - This is added when the results list is open
  • data-loading="true" - This is added if your search function is a Promise, and hasn't resolved yet
  • data-position="below" - This shows if the results list is positioned above or below the input element

In addition, an aria-expanded attribute is added to the input element, and aria-selected is added to the currently selected li element.

Below is an example of how you could use these attributes in your CSS.

/* Change border if results are above input */
[data-position='above'] .autocomplete-result-list {
  border-bottom: none;
  border-radius: 8px 8px 0 0;
}