xi / aria-api Goto Github PK

If you consider the following HTML:

<img src="..." title="title">
<img src="..." alt="" title="title">
<img src="..." alt="alt" title="title">

then the getName() results for these images should be 'title', '', and 'alt' respectively. However the returned results are 'title', 'title', and 'alt'.

This is explicitly defined by this specification: https://www.w3.org/TR/html-aam-1.0/#img-element-accessible-name-computation

getName() on an input element will return the placeholder content, which does not appear to be behaviour specified by Accessible Name and Description Computation 1.1 or draft 1.2 as far as I can tell (although it is what Chrome does).

aria-api should be ignoring the contents of <noscript> elements. These elements turn their contents into text nodes, even though their contents is in fact HTML. So the accessibility name/description end up being a load of gibberish HTML for any element that contains a noscript element.

I'm looking to build a tool to evaluate gaps w.r.t 2.5.3: Label in Name in a html page.

I need to get the Visual label matches the Accessible Name, to validate the 2.5.3: Label in Name success criterion.

While I can use the getName(element) to get the element's Accessible Name, I'm looking if there is an api to get the (visual) label for my comparison ?

At the moment, implicit aria-level for headings is simply calculated by using the number from the tag name. This ignores the HTML5 document outline algorithm.

The outline algorithm has never been implemented. But three days ago a new proposal has given new hope to the idea:

https://github.com/whatwg/html/issues/83#issuecomment-359871505

It seems like this is the final attempt. Either the outline algorithm will be removed from the spec (in which case nothing needs to be changed about aria-api) or it will be replaced by a new version that will be easier to implement.

Hi, thanks for the work on this code. I want to use the getName API for something else that I am building, so i was testing some of the cases and realised that 5.2 was not working.

https://www.w3.org/TR/html-aam-1.0/#input-type-button-input-type-submit-and-input-type-reset-accessible-name-computation

my tests were

  describe('5.2 button input should behave like buttons', () => {
    test('it uses value', () => {
      document.body.innerHTML = `
        <input id="el" type="button" value="the world! dog!"/>
      `;
      const foundInput = findControl('the world! dog!');
      expect(foundInput).toEqual(document.getElementById('el'));
    });

    test('deafults submit if no content', () => {
      document.body.innerHTML = `
        <input id="el" type="submit"></input>
      `;
      const foundInput = findControl('submit');
      expect(foundInput).toEqual(document.getElementById('el'));
    });

    test('deafults reset if no content', () => {
      document.body.innerHTML = `
        <input id="el" type="reset"></input>
      `;
      const foundInput = findControl('reset');
      expect(foundInput).toEqual(document.getElementById('el'));
    });
  });

would love to get your thoughts if this should be handled by the tool or if it shouldn't be?