Bug 97164 – [Chromium] Improve glyph selection of HarfBuzzShaper

• Status: RESOLVED
• Resolution: FIXED
• Component: Platform

[Kenichi Ishibashi]

In complex text, a character can correspond with zero, one, or multiple glyphs. Similarly, a glyph can correspond with zero, one, or multiple characters. I noticed that HarfBuzzShaper doesn't handle selection correctly for some cases.

It seems that we should return character index like below:

       +--- cluster 0 --+--- cluster 1 --+-----+---- cluster n-2 ---+---- cluster n-1 ---+
Glyphs | c_0 | .. | c_0 | c_1 | .. | c_1 | ... | c_n-2 | .. | c_n-2 | c_n-1 | .. | c_n-1 |
       +----------------+----------------+-----+--------------------+--------------------+
CharacterIndex to be returned:
LTR         0   |      c_1      |    ...    |    c_n-2   |       c_n-1        |     n
RTL         n   |      c_0      |    c_1    |     ...    |       c_n-2        |     0

(A cluster is the character index which correspond with a glyph)

[Kenichi Ishibashi]

Created attachment 164833 [details]
Patch

[Kenichi Ishibashi]

Hi Tony,
could you take a look?

[WebKit Review Bot]

Comment on attachment 164833 [details]
Patch

Attachment 164833 [details] did not pass chromium-ews (chromium-xvfb):
Output: http://queues.webkit.org/results/13951080

New failing tests:
fast/text/international/hebrew-selection.html
http/tests/css/link-css-disabled-value-with-slow-loading-sheet.html

[Kenichi Ishibashi]

Created attachment 164852 [details]
Patch

[Tony Chang]

Comment on attachment 164852 [details]
Patch

I like your ref test.

[Kenichi Ishibashi]

Created attachment 164994 [details]
Patch for landing

[Kenichi Ishibashi]

(In reply to comment #5)
> (From update of attachment 164852 [details])
> I like your ref test.

Thank you for review!

[WebKit Review Bot]

Comment on attachment 164994 [details]
Patch for landing

Clearing flags on attachment: 164994

Committed r129175: <http://trac.webkit.org/changeset/129175>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.

[Yury Semikhatsky]

The test is failing with image diff on Windows too. Looking at the image output I didn't see expected selection so I updated the expectations to mark the test as failing on Win instead of providing Win-specific expectations:
http://trac.webkit.org/changeset/129194


Link to the flakiness dashboard:
http://test-results.appspot.com/dashboards/flakiness_dashboard.html#tests=fast%2Ftext%2Finternational%2Fhebrew-selection.html

Also reopening the bug as it was specified in TestExpectations. You should either file another bug for tracking the failure or keep this one open.

[Kenichi Ishibashi]

(In reply to comment #10)
> The test is failing with image diff on Windows too. Looking at the image output I didn't see expected selection so I updated the expectations to mark the test as failing on Win instead of providing Win-specific expectations:
> http://trac.webkit.org/changeset/129194
> 
> 
> Link to the flakiness dashboard:
> http://test-results.appspot.com/dashboards/flakiness_dashboard.html#tests=fast%2Ftext%2Finternational%2Fhebrew-selection.html
> 
> Also reopening the bug as it was specified in TestExpectations. You should either file another bug for tracking the failure or keep this one open.

Thanks for the heads-up. I created a bug for tracking the failure.

[Yury Semikhatsky]

(In reply to comment #11)
> (In reply to comment #10)
> > The test is failing with image diff on Windows too. Looking at the image output I didn't see expected selection so I updated the expectations to mark the test as failing on Win instead of providing Win-specific expectations:
> > http://trac.webkit.org/changeset/129194
> > 
> > 
> > Link to the flakiness dashboard:
> > http://test-results.appspot.com/dashboards/flakiness_dashboard.html#tests=fast%2Ftext%2Finternational%2Fhebrew-selection.html
> > 
> > Also reopening the bug as it was specified in TestExpectations. You should either file another bug for tracking the failure or keep this one open.
> 
> Thanks for the heads-up. I created a bug for tracking the failure.

Thanks!