Sha256: 207a4460ea9453fbd62b8244bd2e6c62234c725c69f7e16fb3a252d9032f5e04
Contents?: true
Size: 1.83 KB
Versions: 375
Compression:
Stored size: 1.83 KB
Contents
/* global aria, dom, text */
/**
* Get the accessible name based on aria-labelledby
*
* @param {VirtualNode} element
* @param {Object} context
* @property {Bool} inLabelledByContext Whether or not the lookup is part of aria-labelledby reference
* @property {Bool} inControlContext Whether or not the lookup is part of a native label reference
* @property {Element} startNode First node in accessible name computation
* @property {Bool} debug Enable logging for formControlValue
* @return {string} Cancatinated text value for referenced elements
*/
aria.arialabelledbyText = function arialabelledbyText(node, context = {}) {
node = node.actualNode || node;
/**
* Note: The there are significant difference in how many "leads" browsers follow.
* - Firefox stops after the first IDREF, so it
* doesn't follow aria-labelledby after a for:>ID ref.
* - Chrome seems to just keep iterating no matter how many levels deep.
* - AccName-AAM 1.1 suggests going one level deep, but to treat
* each ref type separately.
*
* Axe-core's implementation behaves most closely like Firefox as it seems
* to be the common denominator. Main difference is that Firefox
* includes the value of form controls in addition to aria-label(s),
* something no other browser seems to do. Axe doesn't do that.
*/
if (
node.nodeType !== 1 ||
context.inLabelledByContext ||
context.inControlContext
) {
return '';
}
const refs = dom.idrefs(node, 'aria-labelledby').filter(elm => elm);
return refs.reduce((accessibleName, elm) => {
const accessibleNameAdd = text.accessibleText(elm, {
// Prevent the infinite reference loop:
inLabelledByContext: true,
startNode: context.startNode || node,
...context
});
if (!accessibleName) {
return accessibleNameAdd;
} else {
return `${accessibleName} ${accessibleNameAdd}`;
}
}, '');
};
Version data entries
375 entries across 375 versions & 1 rubygems