DirectWrite to use aliased if ClearType is empty.

DirectWrite to use aliased if ClearType is empty.

Some CJK fonts with some versions of DirectWrite return valid
data for bitmaps, but not for cleartype data.

For reference, two screenshots. M37 Stable and then with this patch:

http://imgur.com/9pf3rB9,EiTb6Li

See https://code.google.com/p/chromium/issues/detail?id=396624#c10 for
content of test html file.

R=eae@chromium.org, reed@google.com, shrikant@chromium.org, bungeman@chromium.org, cpu@chromium.org
BUG=chromium:407945
Committed: https://skia.googlesource.com/skia/+/b2f7fce

[bungeman-skia, 2014-08-27 21:58:16]

A somewhat cleaner version of https://codereview.chromium.org/510563003/ .

[eae, 2014-08-27 22:03:41]

LGTM

Thanks Ben!

[reed1, 2014-08-27 22:07:33]

reed@google.com changed reviewers:
+ reed@google.com

[reed1, 2014-08-27 22:07:33]

https://codereview.chromium.org/504343007/diff/1/src/ports/SkScalerContext_wi...
File src/ports/SkScalerContext_win_dw.cpp (right):

https://codereview.chromium.org/504343007/diff/1/src/ports/SkScalerContext_wi...
src/ports/SkScalerContext_win_dw.cpp:458: if (bbox.left != bbox.right &&
bbox.top != bbox.bottom) {
I'd add a comment inside the if...

// We concede that, from here out for this strike, we are force ourselves to
return BW glyphs.
// If it turns out that DW version alone is sufficient to know that LCD requests
will fail, we could
//    move the test into our constructor.
// If instead, we discover that this failure is not global per strike, then
overwriting fTextureType
//    here is overkill, and we should find a way to do this per-glyph (perhaps
checking for failure
//    in generateImage).

[reed1, 2014-08-27 22:46:54]

seems very readable. lgtm (assuming it works)

https://codereview.chromium.org/504343007/diff/20001/src/ports/SkScalerContex...
File src/ports/SkScalerContext_win_dw.cpp (right):

https://codereview.chromium.org/504343007/diff/20001/src/ports/SkScalerContex...
src/ports/SkScalerContext_win_dw.cpp:458: if (bbox.left != bbox.right &&
bbox.top != bbox.bottom) {
/*
 * DWRITE 3x1 seems to have a bug/feature where requesting LCD fails (gives
empty bounds)
 * but if we ask for BW, we succeed. If we detect this behavior, then signal to
ourselves to
 * request BW inside our generateImage().
 */

[scottmg, 2014-08-27 23:10:07]

Confirmed fixes the repro I had. LGTM, thanks.

Result here: http://imgur.com/GOFQh2s (text after Japanese text looks better
with this version as that was all the same run).

[Shrikant Kelkar, 2014-08-27 23:11:37]

On 2014/08/27 23:10:07, scottmg wrote:
> Confirmed fixes the repro I had. LGTM, thanks.
> 
> Result here: http://imgur.com/GOFQh2s (text after Japanese text looks better
> with this version as that was all the same run).

LGTM

Thanks

[bungeman-skia, 2014-08-27 23:18:50]

Committed patchset #3 (id:40001) manually as b2f7fce.