Fix touch adjustment to never return pseudo elements

Fix touch adjustment to never return pseudo elements

Hit testing which lands on a pseudo element (eg. :before/:after) is
normally modified (in HitTestResult.setInnerNode) to return the
parent element.  This CL makes touch adjustment consistent with this.
We still adjust towards any pseudo elements, but the final node will
be the pseudo element's parent.

This fixes an issues where checkboxes covered by a pseudo element
(as for chrome://settings custom checkboxes) weren't being activated
on tap.  CheckboxInputType::willDispatchClick is invoked only when
the click event is targetted directly at the <input> element, not
when it bubbles up from a descendant such as this pseudo element.

Prior to 
http://src.chromium.org/viewvc/blink?revision=178098&view=revision
this issue was masked because we did a redundant non-rect hit test
 for the mouseup event using the adjusted position.  Now that we
share the original hit test results the difference of behavior was
exposed to mouse event handlers.

BUG=395942
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=178965

[Rick Byers, 2014-07-25 19:56:30]

Kevin,
I know it's been a long time since you've worked in this code but can you do a
little code review for me please?  Zeeshan and Ojan are unavailable at the
moment.

Thanks,
   Rick

[Rick Byers, 2014-07-25 19:59:14]

https://codereview.chromium.org/418133010/diff/1/LayoutTests/touchadjustment/...
File LayoutTests/touchadjustment/pseudo-element-expected.txt (left):

https://codereview.chromium.org/418133010/diff/1/LayoutTests/touchadjustment/...
LayoutTests/touchadjustment/pseudo-element-expected.txt:8: PASS adjustedNode.id
is element.id
Note that the failure mode of this test before my fix is to FAIL here (and at
each ID check in the cases below) because adjustedNode.id is empty (since it's
the pseudo element node).

The co-ordinates were already correct.  I'm testing them here to verify I didn't
break consideration of pseudo elements in touch adjustment decisions.

[kevers, 2014-07-25 20:46:56]

lgtm with nits.

https://codereview.chromium.org/418133010/diff/1/LayoutTests/touchadjustment/...
File LayoutTests/touchadjustment/pseudo-element.html (right):

https://codereview.chromium.org/418133010/diff/1/LayoutTests/touchadjustment/...
LayoutTests/touchadjustment/pseudo-element.html:46: }
nit: missing trailing semicolon.

https://codereview.chromium.org/418133010/diff/1/LayoutTests/touchadjustment/...
LayoutTests/touchadjustment/pseudo-element.html:98: if (window.internals) {
Should we be testing for window.testRunner as well?

[Rick Byers, 2014-07-25 20:55:16]

On 2014/07/25 20:46:56, kevers wrote:
> lgtm with nits.
> 
>
https://codereview.chromium.org/418133010/diff/1/LayoutTests/touchadjustment/...
> File LayoutTests/touchadjustment/pseudo-element.html (right):
> 
>
https://codereview.chromium.org/418133010/diff/1/LayoutTests/touchadjustment/...
> LayoutTests/touchadjustment/pseudo-element.html:46: }
> nit: missing trailing semicolon.

Fixed, thanks.

>
https://codereview.chromium.org/418133010/diff/1/LayoutTests/touchadjustment/...
> LayoutTests/touchadjustment/pseudo-element.html:98: if (window.internals) {
> Should we be testing for window.testRunner as well?

I don't think it's necessary.  Neither I nor touchadjustment.js call any
testRunner APIs directly, and js-test.js is designed to work with or without
testRunner.  This is nice because it means I can (in principle - I didn't try in
this case) run the test interactively with content_shell
--expose-internals-for-testing (stepping through with the devtools debugger,
etc.).

[Rick Byers, 2014-07-25 20:55:53]

The CQ bit was checked by rbyers@chromium.org

[commit-bot: I haz the power, 2014-07-25 20:56:23]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/rbyers@chromium.org/418133010/20001

[commit-bot: I haz the power, 2014-07-25 22:04:18]

Change committed as 178965