[Webkit-unassigned] [Bug 242401] AX: Expose suggestion, insertion, deletion roles and attributes

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=242401

--- Comment #17 from Joshua Hoffman <jhoffman23 at apple.com> ---
(In reply to Andres Gonzalez from comment #16)
> (In reply to Joshua Hoffman from comment #15)
> > Created attachment 460835 [details]
> > Patch
> 
> --- a/Source/WebCore/accessibility/ios/WebAccessibilityObjectWrapperIOS.mm
> +++ a/Source/WebCore/accessibility/ios/WebAccessibilityObjectWrapperIOS.mm
> 
> +- (BOOL)accessibilityIsFirstItemInSuggestion
> +{
> ...
> +    auto children = suggestion->children();
> +    auto parent = self.axBackingObject->parentObject();
> +    BOOL isFirstChildOfParent = parent && parent->children().size() &&
> parent->children()[0] == self.axBackingObject;
> +    return children.size() && (children[0] == self.axBackingObject ||
> (isFirstChildOfParent &&
> children[0]->isAncestorOfObject(self.axBackingObject)));
> 
> I think this logic fails in the following case:
> 
> <suggestion>
>     <grandparent>
>         <uncle>
>             <child1>
>         <parent>
>             <child2>
> 
> it will return true for child2.

I'd think a traversal up, where in each iteration we check both that a child is always the first element and check for a suggestion ancestor, could fix this? (not using FindAncestor to allow us to break early) Something like:

for (AXCoreObject* obj = self.axBackingObject->parentObject(); obj; obj = obj->parentObject())
{
    if (auto parent = obj->parentObject()) {
        if (parent->children().size() && parent->children()[0] != self.axBackingObject)
            return NO;
        if (parent->roleValue() == AccessibilityRole::Suggestion)
            return YES;
    }
}

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-unassigned/attachments/20220713/bc95427e/attachment-0001.htm>