Custom elements provide a component model for the web. The custom elements specification provides:

  • A mechanism for associating a class with a custom element name.
  • A set of lifecycle callbacks invoked when an instance of the custom element changes state (for example, added or removed from the document).
  • A callback invoked whenever one of a specified set of attributes changes on the instance.

Put together, these features let you build an element with its own public API that reacts to state changes.

This document provides an overview of custom elements as they relate to Polymer. For a more detailed overview of custom elements, see: Custom Elements v1: Reusable Web Components on Web Fundamentals.

To define a custom element, you create an ES6 class and associate it with the custom element name.

// Create a class that extends HTMLElement (directly or indirectly)
class MyElement extends HTMLElement { … };

// Associate the new class with an element name
window.customElements.define('my-element', MyElement);

You can use a custom element just like you'd use a standard element:

<my-element></my-element>

Or:

const myEl = document.createElement('my-element');

Or:

const myEl = new MyElement();

The element's class defines its behavior and public API. The class must extend HTMLElement or one of its subclasses (for example, another custom element).

Custom element names. By specification, the custom element's name must start with a lower-case ASCII letter and must contain a dash (-). There's also a short list of prohibited element names that match existing names. For details, see the Custom elements core concepts section in the HTML specification.

Polymer provides a set of features on top of the basic custom element specification. To add these features to your element, extend Polymer's base element class, Polymer.Element:

<link rel="import" href="/bower_components/polymer/polymer-element.html">

<script>
  class MyPolymerElement extends Polymer.Element {
    ...
  }

  customElements.define('my-polymer-element', MyPolymerElement);
</script>

Polymer adds a set of features to the basic custom element:

  • Instance methods to handle common tasks.
  • Automation for handling properties and attributes, such as setting a property based on the corresponding attribute.
  • Creating shadow DOM trees for element instances based on a supplied <template>.
  • A data system that supports data binding, property change observers, and computed properties.

The custom element spec provides a set of callbacks called "custom element reactions" that allow you to run user code in response to certain lifecycle changes.

Reaction Description
constructor Called when the element is upgraded (that is, when an element is created, or when a previously-created element becomes defined).
connectedCallback Called when the element is added to a document.
disconnectedCallback Called when the element is removed from a document.
attributeChangedCallback Called when any of the element's attributes are changed, appended, removed, or replaced,

For each reaction, the first line of your implementation must be a call to the superclass constructor or reaction. For the constructor, this is simply the super() call.

constructor() {
  super();
  // …
}

For other reactions, call the superclass method. This is required so Polymer can hook into the element's lifecycle.

connectedCallback() {
  super.connectedCallback();
  // …
}

The element constructor has a few special limitations:

  • The first statement in the constructor body must be a parameter-less call to the super method.
  • The constructor can't include a return statement, unless it is a simple early return (return or return this).
  • The constructor can't examine the element's attributes or children, and the constructor can't add attributes or children.

For a complete list of limitations, see Requirements for custom element constructors in the WHATWG HTML Specification.

Whenever possible, defer work until the connectedCallback or later instead of performing it in the constructor.

The custom elements specification doesn't provide a one-time initialization callback. Polymer provides a ready callback, invoked the first time the element is added to the DOM.

ready() {
  super.ready();
  // When possible, use afterNextRender to defer non-critical
  // work until after first paint.
  Polymer.RenderStatus.afterNextRender(this, function() {
    ...
  });
}

The Polymer.Element class initializes your element's template and data system during the ready callback, so if you override ready, you must call super.ready() at some point.

When the superclass ready method returns, the element's template has been instantiated and initial property values have been set. However, light DOM elements may not have been distributed when ready is called.

Don't use ready to initialize an element based on dynamic values, like property values or an element's light DOM children. Instead, use observers to react to property changes, and observeNodes or the slotchange event to react to children being added and removed from the element.

Related topics:

By specification, custom elements can be used before they're defined. Adding a definition for an element causes any existing instances of that element to be upgraded to the custom class.

For example, consider the following code:

<my-element></my-element>
<script>
  class MyElement extends HTMLElement { ... };

  // ...some time much later...
  customElements.define('my-element', MyElement);
</script>

When parsing this page, the browser will create an instance of <my-element> before parsing and executing the script. In this case, the element is created as an instance of HTMLElement, not MyElement. After the element is defined, the <my-element> instance is upgraded so it has the correct class (MyElement). The class constructor is called during the upgrade process, followed by any pending lifecycle callbacks.

Element upgrades allow you to place elements in the DOM while deferring the cost of initializing them. It's a progressive enhancement feature.

Elements have a custom element state that takes one of the following values:

  • "uncustomized". The element does not have a valid custom element name. It is either a built-in element (<p>, <input>) or an unknown element that cannot become a custom element (<nonsense>)
  • "undefined". The element has a valid custom element name (such as "my-element"), but has not been defined.
  • "custom". The element has a valid custom element name and has been defined and upgraded.
  • "failed". An attempt to upgrade the element failed (for example, because the class was invalid).

The custom element state isn't exposed as a property, but you can style elements depending on whether they're defined or undefined.

Elements in the "custom" and "uncustomized" state are considered "defined". In CSS you can use the :defined pseudo-class selector to target elements that are defined. You can use this to provide placeholder styles for elements before they're upgraded:

my-element:not(:defined) {
  background-color: blue;
}

In addition to HTMLElement, a custom element can extend another custom element:

class ExtendedElement extends MyElement {
  static get is() { return 'extended-element'; }

  static get properties() {
    return {
      thingCount: {
        value: 0,
        observer: '_thingCountChanged'
      }
    }
  }
  _thingCountChanged() {
    console.log(`thing count is ${this.thingCount}`);
  }
};

customElements.define(ExtendedElement.is, ExtendedElement);

Polymer does not currently support extending built-in elements. The custom elements spec provides a mechanism for extending built-in elements, such as <button> and <input>. The spec calls these elements customized built-in elements. Customized built-in elements provide many advantages (for example, being able to take advantage of built-in accessibility features of UI elements like <button> and <input>). However, not all browser makers have agreed to support customized built-in elements, so Polymer does not support them at this time.

When you extend custom elements, Polymer treats the properties object and observers array specially: when instantiating an element, Polymer walks the prototype chain and flattens these objects. So the properties and observers of a subclass are added to those defined by the superclass.

A subclass can also inherit a template from its superclass. For details, see Inherited templates.

ES6 classes allow single inheritance, which can make it challenging to share code between unrelated elements. Class expression mixins let you share code between elements without adding a common superclass.

A class expression mixin is basically a function that operates as a class factory. You pass in a superclass, and the function generates a new class that extends the superclass with the mixin's methods.

const fancyDogClass = FancyMixin(dogClass);
const fancyCatClass = FancyMixin(catClass);

Add a mixin to your element like this:

class MyElement extends MyMixin(Polymer.Element) {
  static get is() { return 'my-element' }
}

If that isn't clear, it may help to see it in two steps:

// Create new base class that adds MyMixin's methods to Polymer.Element
const polymerElementPlusMixin = MyMixin(Polymer.Element);

// Extend the new base class
class MyElement extends polymerElementPlusMixin {
  static get is() { return 'my-element' }
}

So the inheritance hierarchy is:

MyElement <= polymerElementPlusMixin <= Polymer.Element

You can apply mixins to any element class, not just Polymer.Element:

class MyExtendedElement extends SomeMixin(MyElement) {
  ...
}

You can also apply multiple mixins in sequence:

class AnotherElement extends AnotherMixin(MyMixin(Polymer.Element)) { … }

A mixin is simply a function that takes a class and returns a subclass:

MyMixin = function(superClass) {
  return class extends superClass {
    constructor() {
      super();
      this.addEventListener('keypress', e => this.handlePress(e));
    }

    static get properties() {
      return {
        bar: {
          type: Object
        }
      };
    }

    static get observers() {
      return [ '_barChanged(bar.*)' ];
    }

    _barChanged(bar) { ... }

    handlePress(e) { console.log('key pressed: ' + e.charCode); }
  }
}

Or using an ES6 arrow function:

MyMixin = (superClass) => class extends superClass {
  ...
}

The mixin class can define properties, observers, and methods just like a regular element class. In addition, a mixin can incorporate other mixins:

MyCompositeMixin = (base) => class extends MyMixin2(MyMixin1(base)) {
  ...
}

Because mixins are simply adding classes to the inheritance chain, all of the usual rules of inheritance apply. For example, mixin classes can define constructors, can call superclass methods with super, and so on.

Document your mixins. The Polymer build and lint tools require some extra documentation tags to property analyze mixins and elements that use them. Without the documentation tags, the tools will log warnings. For details on documenting mixins, see Class mixins in Document your elements.

When creating a mixin that you intend to share with other groups or publish, a couple of additional steps are recommended:

  • Use the Polymer.dedupingMixin function to produce a mixin that can only be applied once.

  • Create a unique namespace for your mixins, to avoid colliding with other mixins or classes that might have similar names.

The dedupingMixin function is useful because a mixin that's used by other mixins may accidentally be applied more than once. For example if MixinA includes MixinB and MixinC, and you create an element that uses MixinA but also uses MixinB directly:

class MyElement extends MixinB(MixinA(Polymer.Element)) { ... }

At this point, your element contains two copies of MixinB in its prototype chain. dedupingMixin takes a mixin function as an argument, and returns a new, deduplicating mixin function:

dedupingMixinB = Polymer.dedupingMixin(mixinB);

The deduping mixin has two advantages: first, whenever you use the mixin, it memoizes the generated class, so any subsequent uses on the same base class return the same class object—a minor optimization.

More importantly, the deduping mixin checks whether this mixin has already been applied anywhere in the base class's prototype chain. If it has, the mixin simply returns the base class. In the example above, if you used dedupingMixinB instead of mixinB in both places, the mixin would only be applied once.

The following example shows one way you might create a namespaced, deduping mixin:

// Create my namespace, if it doesn't exist
if (!window.MyNamespace) {
  window.MyNamespace = {};
}

MyNamespace.MyMixin = Polymer.dedupingMixin((base) =>

  // the mixin class
  class extends base {
    ...
  }
);

More information: Custom elements v1: reusable web components on Web Fundamentals.