Introduce use counters for shadow DOM handling in plainText()

Introduce use counters for shadow DOM handling in plainText()

Since http://crrev.com/98493009, |Element.innerText|, |Selection.toString()|,
|Window.find()| and |execCommand("find")| JavaScript API returns text including
text in shadow DOM tree. This behavior is unexpected and incompatible with other
browsers. We would like to make these API behave as same as other browsers.
However, we aren't sure that real world users depend on this incompatible
behavior.

If real world usages of very small, we'll changes these API to compatible other
browsers.

Note: IE and Firefox haven't supported shadow DOM yet. Safari supports user
agent shadow DOM, but these APIs returns text from DOM tree.

This patch introduces use counters for these APIs contains text from shadow
DOM tree for a preparation of collecting real world usage to determine
user affect when change these APIs behavior as other browsers.

BUG=n/a
TEST=n/a; no behavior changes

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=198268

[yosin_UTC9, 2015-07-02 06:13:10]

yosin@chromium.org changed reviewers:
+ hayato@chromium.org

[yosin_UTC9, 2015-07-02 06:19:12]

PTAL

[yoichio, 2015-07-02 06:28:38]

yoichio@chromium.org changed reviewers:
+ yoichio@chromium.org

[yoichio, 2015-07-02 06:28:39]

https://codereview.chromium.org/1213213004/diff/20001/Source/core/dom/Element...
File Source/core/dom/Element.cpp (right):

https://codereview.chromium.org/1213213004/diff/20001/Source/core/dom/Element...
Source/core/dom/Element.cpp:2500: return plainText(Position(this,
Position::PositionIsBeforeChildren), Position(this,
Position::PositionIsAfterChildren), TextIteratorForInnerText);
Why do't you call UseCounter::count here directly?

[yosin_UTC9, 2015-07-02 06:40:29]

Update description to make an intention of this patch clear.

https://codereview.chromium.org/1213213004/diff/20001/Source/core/dom/Element...
File Source/core/dom/Element.cpp (right):

https://codereview.chromium.org/1213213004/diff/20001/Source/core/dom/Element...
Source/core/dom/Element.cpp:2500: return plainText(Position(this,
Position::PositionIsBeforeChildren), Position(this,
Position::PositionIsAfterChildren), TextIteratorForInnerText);
On 2015/07/02 06:28:39, yoichio wrote:
> Why do't you call UseCounter::count here directly?

I would like to count only if |plainText()| handles shadow DOM tree.

[yoichio, 2015-07-02 07:10:32]

https://codereview.chromium.org/1213213004/diff/20001/Source/core/dom/Element...
File Source/core/dom/Element.cpp (right):

https://codereview.chromium.org/1213213004/diff/20001/Source/core/dom/Element...
Source/core/dom/Element.cpp:2500: return plainText(Position(this,
Position::PositionIsBeforeChildren), Position(this,
Position::PositionIsAfterChildren), TextIteratorForInnerText);
On 2015/07/02 06:40:29, Yosi_UTC9 wrote:
> On 2015/07/02 06:28:39, yoichio wrote:
> > Why do't you call UseCounter::count here directly?
> 
> I would like to count only if |plainText()| handles shadow DOM tree.

Can't we check that here?

[yosin_UTC9, 2015-07-02 08:08:31]

PTAL
It's ready!

https://codereview.chromium.org/1213213004/diff/20001/Source/core/dom/Element...
File Source/core/dom/Element.cpp (right):

https://codereview.chromium.org/1213213004/diff/20001/Source/core/dom/Element...
Source/core/dom/Element.cpp:2500: return plainText(Position(this,
Position::PositionIsBeforeChildren), Position(this,
Position::PositionIsAfterChildren), TextIteratorForInnerText);
On 2015/07/02 07:10:32, yoichio wrote:
> On 2015/07/02 06:40:29, Yosi_UTC9 wrote:
> > On 2015/07/02 06:28:39, yoichio wrote:
> > > Why do't you call UseCounter::count here directly?
> > 
> > I would like to count only if |plainText()| handles shadow DOM tree.
> 
> Can't we check that here?

Yes, we can. But, we need to traverse tree here. We don't want to traverse
twice, once for here and another is |plaintText()|.

[yoichio, 2015-07-02 08:17:21]

On 2015/07/02 08:08:31, Yosi_UTC9 wrote:
> PTAL
> It's ready!
> 
>
https://codereview.chromium.org/1213213004/diff/20001/Source/core/dom/Element...
> File Source/core/dom/Element.cpp (right):
> 
>
https://codereview.chromium.org/1213213004/diff/20001/Source/core/dom/Element...
> Source/core/dom/Element.cpp:2500: return plainText(Position(this,
> Position::PositionIsBeforeChildren), Position(this,
> Position::PositionIsAfterChildren), TextIteratorForInnerText);
> On 2015/07/02 07:10:32, yoichio wrote:
> > On 2015/07/02 06:40:29, Yosi_UTC9 wrote:
> > > On 2015/07/02 06:28:39, yoichio wrote:
> > > > Why do't you call UseCounter::count here directly?
> > > 
> > > I would like to count only if |plainText()| handles shadow DOM tree.
> > 
> > Can't we check that here?
> 
> Yes, we can. But, we need to traverse tree here. We don't want to traverse
> twice, once for here and another is |plaintText()|.

I see.
lgtm.

[yosin_UTC9, 2015-07-03 01:18:32]

yosin@chromium.org changed reviewers:
+ tkent@chromium.org

[yosin_UTC9, 2015-07-03 01:18:32]

ping...

[hayato, 2015-07-03 05:11:14]

lgtm

[yosin_UTC9, 2015-07-03 05:41:43]

The CQ bit was checked by yosin@chromium.org

[yosin_UTC9, 2015-07-03 05:41:44]

The patchset sent to the CQ was uploaded after l-g-t-m from
yoichio@chromium.org, hayato@chromium.org
Link to the patchset: https://codereview.chromium.org/1213213004/#ps60001
(title: "2015-07-03T14:40:27 Rebase")

[commit-bot: I haz the power, 2015-07-03 05:41:49]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1213213004/60001

[commit-bot: I haz the power, 2015-07-03 08:22:13]

Committed patchset #4 (id:60001) as
https://src.chromium.org/viewvc/blink?view=rev&revision=198268