[Android] Add a workaround for setting empty IME compositions.

[Android] Add a workaround for setting empty IME compositions.

This is a tempory workaround where Blink does not cancel the composition when
we send a commit with the empty text.

BUG=373934
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=272615

[aurimas (slooooooooow), 2014-05-21 23:40:56]

Hey Ted,

PTAL.

Aurimas

[Ted C, 2014-05-22 00:42:35]

https://chromiumcodereview.appspot.com/289863013/diff/1/content/public/androi...
File
content/public/android/java/src/org/chromium/content/browser/input/AdapterInputConnection.java
(right):

https://chromiumcodereview.appspot.com/289863013/diff/1/content/public/androi...
content/public/android/java/src/org/chromium/content/browser/input/AdapterInputConnection.java:228:
if (maybePerformEmptyCompositionWorkaround(text)) return true;
Why not just do this in setComposingText?

https://chromiumcodereview.appspot.com/289863013/diff/1/content/public/androi...
content/public/android/java/src/org/chromium/content/browser/input/AdapterInputConnection.java:465:
if (text.toString().isEmpty() && (selectionStart == selectionEnd)
I think you can use TextUtils.isEmpty(text)

https://chromiumcodereview.appspot.com/289863013/diff/1/content/public/androi...
content/public/android/java/src/org/chromium/content/browser/input/AdapterInputConnection.java:467:
|| compositionEnd != INVALID_COMPOSITION) ) {
shouldn't this be &&?  you subtract these values from selection below and this
seems like it could mess up things.

https://chromiumcodereview.appspot.com/289863013/diff/1/content/public/androi...
content/public/android/java/src/org/chromium/content/browser/input/AdapterInputConnection.java:470:
int selection = Selection.getSelectionStart(mEditable);
do you think selection might change from finishComposingText?  Just wondering
why you're not using selectionStart above?

[aurimas (slooooooooow), 2014-05-22 21:13:38]

https://chromiumcodereview.appspot.com/289863013/diff/1/content/public/androi...
File
content/public/android/java/src/org/chromium/content/browser/input/AdapterInputConnection.java
(right):

https://chromiumcodereview.appspot.com/289863013/diff/1/content/public/androi...
content/public/android/java/src/org/chromium/content/browser/input/AdapterInputConnection.java:228:
if (maybePerformEmptyCompositionWorkaround(text)) return true;
On 2014/05/22 00:42:36, Ted C wrote:
> Why not just do this in setComposingText?

In internal bug Google Keyboard team mentioned that in certain cases they can
call either setComposingText() or commitText() with empty text.

https://chromiumcodereview.appspot.com/289863013/diff/1/content/public/androi...
content/public/android/java/src/org/chromium/content/browser/input/AdapterInputConnection.java:465:
if (text.toString().isEmpty() && (selectionStart == selectionEnd)
On 2014/05/22 00:42:36, Ted C wrote:
> I think you can use TextUtils.isEmpty(text)

Done.

https://chromiumcodereview.appspot.com/289863013/diff/1/content/public/androi...
content/public/android/java/src/org/chromium/content/browser/input/AdapterInputConnection.java:467:
|| compositionEnd != INVALID_COMPOSITION) ) {
On 2014/05/22 00:42:36, Ted C wrote:
> shouldn't this be &&?  you subtract these values from selection below and this
> seems like it could mess up things.

Done.

https://chromiumcodereview.appspot.com/289863013/diff/1/content/public/androi...
content/public/android/java/src/org/chromium/content/browser/input/AdapterInputConnection.java:470:
int selection = Selection.getSelectionStart(mEditable);
On 2014/05/22 00:42:36, Ted C wrote:
> do you think selection might change from finishComposingText?  Just wondering
> why you're not using selectionStart above?

It is valid to have a selection anywhere within the composition, calling
finishComposingText commits the composition which moves the selection to the end
of composition.

[Ted C, 2014-05-23 18:05:33]

lgtm

[aurimas (slooooooooow), 2014-05-23 18:23:19]

The CQ bit was checked by aurimas@chromium.org

[commit-bot: I haz the power, 2014-05-23 18:24:36]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/aurimas@chromium.org/289863013/20001

[commit-bot: I haz the power, 2014-05-23 22:17:53]

Change committed as 272615