Page Comparison

Expected behaviour to ensure a modal window is accessible for keyboard and screen reader users

Keyboard:

Keyboard access is limited to only interacting with the modal dialog once it is visible
When the modal window is opened, focused is moved to the first element in the window that can receive focus.
As a user tabs through all focusable elements, they are cycled back to the beginning of the window
The user is able to close the modal window through a clearly indicated close or OK button. In addition, ESC may also be set to close the window.

For screen reader as above plus:

The window must have role=dialog (or role=alertdialog if appropriate). This informs the user that a dialog has opened.
When the dialog is opened the screen reader announces the title or accessible name of the dialog
1. If text appears before first element to receive focus or additional instructions are needed for screen reader intereaction with the dialog this is added to additional off-screen text and referenced this using aria-describedby (see https://developer.mozilla.org/en-US/docs/Web/Accessibility/ARIA/ARIA_Techniques/Using_the_aria-describedby_attribute) .
For dialogs with text, use headings to provide focus point and structure for the information.
When the user tabs through each element its name and role is annouced. So each focusable elements must have:
1. roles defined to be <button>, < a href="..."> or using role=... (see https://www.w3.org/TR/wai-aria/roles for definitions of roles)
2. accessible names if there is no text in the element, set using aria-label or aria-labelledby

Adding a semantic-ui-react modal with a self state

...

Add the dependencies to package.json...if they are not:
Code Block
language xml
theme DJango
"semantic-ui-react":"^0.67.1", "focus-trap-react":"^2.0.0",

Import all the semantic-ui-modal components needed, the focus trap and react moduls in the new component:

Code Block

language	js
theme	DJango

import React from 'react';
import { Button, Icon, Modal, Container, Segment, Menu,Label,Input,Divider} from 'semantic-ui-react';
import FocusTrap from 'focus-trap-react';

Create the modal class, extending from React.Component as usual:
Code Block
language js
theme DJango
class AccesibleModal extends React.Component{ ... }

Prepare the constructor with the initial state, which will control if the modal will be displaying or not, and also, if the focus will be trapped to the modal or not, from the begining. For example, in this example the modal will be opened after pressing a button which is located in the same component.

Code Block

language	js
theme	DJango

class AccesibleModal extends React.Component{
    constructor(props) {
        super(props);

        this.state = {
            openModal: false,           
            activeTrap: false
        };

    }
}

Create the methods to open and close the modal. Also, the method needed to unmount the trap focus component.

The method which opens the modal should change the state of the modal component and also set the aria-hidden label of the application to true. In fact, the element #app is the div element which contains all the SlideWiki page. It is defined in the 'components/DefaultHTMLLayout.js' component, at line 26:

Code Block

language	xml
theme	DJango
title	DeafultHTMLLayout.js

...
<body>
	<div id="app" aria-hidden = "false" dangerouslySetInnerHTML={{__html: this.props.markup}}></div>
....

Thus, the following method will be enough to open the modal:

Code Block

language	js
theme	DJango

handleOpen(){
        $('#app').attr('aria-hidden','true');
        this.setState({
            modalOpen:true,
            activeTrap:true
        });
}

We can bind it at the constructor also:

Code Block

language	js
theme	DJango

 constructor(props) {
        super(props);

        this.state = {
            openModal: false,
            activeTrap: false
        };
        this.handleOpen = this.handleOpen.bind(this);
        
  }

The method which close the modal should change the state of the modal component and also set the aria-hidden label of the application to false.
Code Block
language js
theme DJango
handleClose(){ $('#app').attr('aria-hidden','false'); this.setState({ modalOpen:false, activeTrap: false }); }
The method which unmounts the focus trap compoenent component should also change the state of the component. If we don't use it, if the user exits pressing outside the modal, the state of the component could be inconsistent.
Code Block
language js
theme DJango
unmountTrap(){ if(this.state.activeTrap){ this.setState({ activeTrap: false }); $('#app').attr('aria-hidden','false'); } }

Final constructor method:

Code Block

language	js
theme	DJango

constructor(props) {
        super(props);

        this.state = {
            openModal: false,
            activeItem: 'MyDecks',
            activeTrap: false
        };

        this.handleOpen = this.handleOpen.bind(this);
        this.handleClose = this.handleClose.bind(this);
        this.unmountTrap = this.unmountTrap.bind(this);
   }

Add the modal to the render method, adding all the aria-labels required

Code Block

language	js
theme	DJango

<Modal trigger={
                    <Button as="button" 
                      type="button" aria-label="Open Button" data-tooltip="Open Button" aria-hidden={this.state.modalOpen}
                      basic  onClick={this.handleOpen} > Open Button
                    </Button>
                   }
                open={this.state.modalOpen}
                onClose={this.handleClose}
                size="large"
                role="dialog"
                id="exampleModal"
                aria-labelledby="exampleModalHeader"
                aria-describedby="exampleModalDescription"              
                tabIndex="0">
</Modal>

Trigger: Defines the button which will open the modal.You can see different attributes needed for making it fully accesibleaccessible:
- as="button", because if not use it, semantic-ui-react render it as a div, and it is problematicwhich is not focusable. This also exposes the role of the element.
- aria-label="Open Button", it this is required . Should contain the name of the button displayed, or it its explanation in case of a icon if there is no text on the button.It should contain explanation in case of a icon button. of the action of the button but it does not need to say that it is a button.
- data-tooltip="Open Button", extra information displayed when you pass over the button
- aria-hidden={this.state.modalOpen}, hides the button for reader when modal is open
- onClick={this.handleOpen}, associaction association to the previously handleOpen method (which is binded in the constructor)
open={this.state.modalOpen}, it constrols controls if the modal is displayed (true) or not (false). Notice that in this case there is a difference with recpect respect semantic-ui. when the modal is not open, also is not rendered in the final DOM, so the components it has inside are not available.

Add the focus-trap component. Modals from semantic-ui and semantic-ui-react do not trap the focus, so when users taps over the last element of the modal, the focus goes outside the modal. This make them no accesiblesnot accessible. In order to avoid that, we should use the focus-trap-react component. The only matter is that If incorrectly placed this component destroys a little some of the final layout. After many tests, we found a good way to introduce it with less impact: adding it under <Modal>, and surrounding the rest of the elements of the modal:

Code Block

language	js
theme	DJango

<Modal trigger={
                    <Button as="button" 
                      type="button" aria-label="Open Button" data-tooltip="Open Button" aria-hidden={this.state.modalOpen}
                      basic  onClick={this.handleOpen} > Open Button
                    </Button>
                   }
                open={this.state.modalOpen}
                onClose={this.handleClose}
                size="large"
                role="dialog"
                id="exampleModal"
                aria-labelledby="exampleModalHeader"
                aria-describedby="exampleModalDescription"              
                tabIndex="0">


  		<FocusTrap
                        id='focus-trap-exampleModal'
                        className = "header"
                        active={this.state.activeTrap}
                        focusTrapOptions={{
                            onDeactivate: this.unmountTrap,
                            clickOutsideDeactivates: true,
                            initialFocus: '#firstElement',
                        }}>


		{/*elements of the modal here*/}


	</FocusTrap>
</Modal>

- className="header": using header as extra class, the impact in the final layout is reduced
- active={this.state.activeTrap}, controls if the focus is trap or not
- aria-label="Open Button", it is required. Should contain the name of the button displayed, or its explanation in case of an icon button.
- focusTrapOptions.onDeactivate:this.unmountTrap, In this property we add the call to the method which ensures the trap component is unmount, even the user exits the modal pressing outside
- focusTrapOptions.clickOutsidesDeactivates:true, also, it is the default value..
- initialFocus:'#firsElement', the id of the element which will receive the focus. I notice that f you don't indicate it, the modal does not receive the focus well

Add Modal Elements. As mentioned above, we need to ad add a header for the modal and an explanation about it. Off Of course, all the content needed and, most probably , some action buttons. We can add them as semantic-ui-react documentation suggest us, but taking into account some tips.

...

Code Block

language	js
theme	DJango
linenumbers	true

<Modal trigger={
                    <Button as="button" 
                      type="button" aria-label="Open Button" data-tooltip="Open Button" aria-hidden={this.state.modalOpen}
                      basic  onClick={this.handleOpen} > Open Button
                    </Button>
                   }
                open={this.state.modalOpen}
                onClose={this.handleClose}
                size="large"
                role="dialog"
                id="exampleModal"
                aria-labelledby="exampleModalHeader"
                aria-describedby="exampleModalDescription"              
                tabIndex="0">


  		<FocusTrap
                        id='focus-trap-exampleModal'
                        className = "header"
                        active={this.state.activeTrap}
                        focusTrapOptions={{
                            onDeactivate: this.unmountTrap,
                            clickOutsideDeactivates: true,
                            initialFocus: '#firstElement',
                        }}>


		{/*elements of the modal here*/}
			<Modal.Header className="ui center aligned" as="h1" id="exampleModalHeader">
                     Accesible Modal 
            </Modal.Header>
            <Modal.Content>
                    <Container>
                         <Segment color="blue" textAlign="center" padded>                           
                            <Segment attached='bottom' textAlign="left">
                               <TextArea className="sr-only" id="exampleModalDescription" value="This modal only is a an example of how make a modal fully accesibleaccessible" />
                               <Label htmlFor="firstElement" as="label" basic color="blue" pointing="right">Input some text</Label>
                               <Input type="text" id="firstElement" placeholder="You should input something.." tabIndex="0" />
                              </Segment>;
                            </Segment>
                            <Modal.Actions>
                              <Button color="green" tabIndex="0" type="button" aria-label="Accept" data-tooltip="Accept">
                                  Accept
                              </Button>
                              <Button color='red' tabIndex="0" type="button" aria-label="Cancel" data-tooltip="Cancel" onClick={this.handleClose} >
                                Cancel
                              </Button>
                            </Modal.Actions>

                      </Segment>
                   </Container>

              </Modal.Content>

	</FocusTrap>
</Modal>

...

Versions Compared

Old Version 21

New Version 22

Key

Expected behaviour to ensure a modal window is accessible for keyboard and screen reader users

Adding a semantic-ui-react modal with a self state