Bug 157644 – AX: Wrong CharacterOffset from an upstream VisiblePosition

• Status: RESOLVED
• Resolution: FIXED
• Component: Accessibility

[Nan Wang]

When we are getting a text marker at the line wrap and it has an upstream VisiblePosition, we should adjust the CharacterOffset as well.

<rdar://problem/25993225>

[Nan Wang]

Created attachment 278776 [details]
Initial patch

Spent quite a long time but was not able to create a layout test.
1. In Mail the end text marker of left line range has upstream affinity but I tried left/right and regular line text marker range call in layout test it gave downstream affinity all the time.
2. Although I had the fix to adjust CharacterOffset to match Range's end offset, TextIterator will always include an extra space at the line wrap. So there's no easy way to test this fix in a layout test by using lengthForTextMarkerRange of stringForTextMarkerRange, they all use TextIterator under the hood.

[chris fleizach]

Comment on attachment 278776 [details]
Initial patch

When I do a grep for upstream in LayoutTests, I see 

platform/ios-simulator-wk1/editing/pasteboard/4242293-expected.txt:caret: position 3 of child 0 {#text} of child 1 {SPAN} of child 6 {DIV} of body (upstream affinity)

[chris fleizach]

(In reply to comment #2)
> Comment on attachment 278776 [details]
> Initial patch
> 
> When I do a grep for upstream in LayoutTests, I see 
> 
> platform/ios-simulator-wk1/editing/pasteboard/4242293-expected.txt:caret:
> position 3 of child 0 {#text} of child 1 {SPAN} of child 6 {DIV} of body
> (upstream affinity)

Do you think that test might help you figure out how to create an upstream position?

[Darin Adler]

Comment on attachment 278776 [details]
Initial patch

View in context: https://bugs.webkit.org/attachment.cgi?id=278776&action=review

> Source/WebCore/ChangeLog:11
> +        Not able to create an upstream VisiblePosition in a layout test.

Why is that? Can we please add something to internals so we can test this? This super-subtle behavior seems critical to test; people will break it without realizing it.

[Nan Wang]

(In reply to comment #3)
> (In reply to comment #2)
> > Comment on attachment 278776 [details]
> > Initial patch
> > 
> > When I do a grep for upstream in LayoutTests, I see 
> > 
> > platform/ios-simulator-wk1/editing/pasteboard/4242293-expected.txt:caret:
> > position 3 of child 0 {#text} of child 1 {SPAN} of child 6 {DIV} of body
> > (upstream affinity)
> 
> Do you think that test might help you figure out how to create an upstream
> position?

Yes I can create an upstream position by adding contenteditable="true"

[Nan Wang]

Created attachment 278851 [details]
patch

Added layout test. Thanks to Chris!

[Nan Wang]

Committed r200872: <http://trac.webkit.org/changeset/200872>