Add support for transformed, stoked, shadowed, filled text. We do this by...

Add support for transformed, stoked, shadowed, filled text. We do this by drawing the text using Skia after getting the outlines from Windows. A cache of these outlines is maintained because the outlines are very slow to retrieve.  This new functionality is in SkiaFontWin.*, and is called when the context detects text attributes that are not supported by Windows.


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=7223

[brettw, 2008-10-28 16:02:12]

No huge rush on this patch, I will wait for the merge branch to land before
submitting anyway.

[M-A Ruel, 2008-10-28 16:46:21]

Mostly nits. I haven't reviewed ThemeHelperWin.* yet.

http://codereview.chromium.org/8615/diff/29/43
File webkit/build/port/port.vcproj (left):

http://codereview.chromium.org/8615/diff/29/43#oldcode663
Line 663: RelativePath="..\..\port\platform\GKURL.cpp"
is this change related?

http://codereview.chromium.org/8615/diff/29/39
File webkit/port/platform/graphics/ImageSkia.cpp (right):

http://codereview.chromium.org/8615/diff/29/39#newcode271
Line 271: /*FIXME(brettw) why did I add this?
I wonder. You wanted to double-draw?

http://codereview.chromium.org/8615/diff/29/30
File webkit/port/platform/graphics/SkGraphicsContext.cpp (right):

http://codereview.chromium.org/8615/diff/29/30#newcode184
Line 184: gfx::PlatformCanvas* canvas =
helper.context()->platformContext()->canvas();
How is it different from canvas_ ?

http://codereview.chromium.org/8615/diff/29/32
File webkit/port/platform/graphics/SkGraphicsContext.h (right):

http://codereview.chromium.org/8615/diff/29/32#newcode70
Line 70: // TODO(brettw) these functions should not have to take GraphicsContext
Maybe this class should be annihilated first?

http://codereview.chromium.org/8615/diff/29/31
File webkit/port/platform/graphics/SkPaintContext.h (right):

http://codereview.chromium.org/8615/diff/29/31#newcode57
Line 57: // Returns rtue if the current graphics state is compatible with
Windows GDI
true

http://codereview.chromium.org/8615/diff/29/36
File webkit/port/platform/graphics/SkiaFontWin.cpp (right):

http://codereview.chromium.org/8615/diff/29/36#newcode1
Line 1: #include <windows.h>
header?

http://codereview.chromium.org/8615/diff/29/37
File webkit/port/platform/graphics/SkiaFontWin.h (right):

http://codereview.chromium.org/8615/diff/29/37#newcode22
Line 22: // HFONTs are not unique for all time).
HFONTs handles are recycled).

http://codereview.chromium.org/8615/diff/29/37#newcode29
Line 29: // from Windows
.

[brettw, 2008-10-28 17:16:29]

http://codereview.chromium.org/8615/diff/29/30
File webkit/port/platform/graphics/SkGraphicsContext.cpp (right):

http://codereview.chromium.org/8615/diff/29/30#newcode184
Line 184: gfx::PlatformCanvas* canvas =
helper.context()->platformContext()->canvas();
On 2008/10/28 16:46:21, M-A wrote:
> How is it different from canvas_ ?

It may or may not be a different canvas. This is the entire point of the
ThemeHelper, is that you draw to the canvas it says and it flips things around
to make it magically work.

http://codereview.chromium.org/8615/diff/29/32
File webkit/port/platform/graphics/SkGraphicsContext.h (right):

http://codereview.chromium.org/8615/diff/29/32#newcode70
Line 70: // TODO(brettw) these functions should not have to take GraphicsContext
If I worked on that it would cause a lot of conflicts with this patch which
would mean more work, so I was going to do it afterwards. If you can refactor
our graphics code before I can check this patch in, though, I won't complain :)

[M-A Ruel, 2008-11-01 00:32:00]

Can you sync with relation to your other changes since some classes disapeared?

[brettw, 2008-12-14 16:47:14]

New patch uploaded (only took 5 weeks!).

[M-A Ruel, 2008-12-15 19:54:51]

looks good.

http://codereview.chromium.org/8615/diff/1401/1203
File webkit/port/platform/graphics/chromium/FontChromiumWin.cpp (right):

http://codereview.chromium.org/8615/diff/1401/1203#newcode60
Line 60: if (context->platformContext()->getDrawLooper())
shadow should be ignored when printing on Windows. :/

http://codereview.chromium.org/8615/diff/1401/1203#newcode157
Line 157: bool canUseGDI = windowsCanHandleTextDrawing(graphicsContext);
I wonder if canUseGDI should always be true when printing.

http://codereview.chromium.org/8615/diff/1401/1203#newcode187
Line 187: // Skia's text draing origin is the baseline, like WebKit, not
drawing

http://codereview.chromium.org/8615/diff/1401/1203#newcode192
Line 192: reinterpret_cast<const WORD*>(&glyphs[0]),
80 cols, if you care.

http://codereview.chromium.org/8615/diff/1401/1208
File webkit/port/platform/graphics/skia/SkiaFontWin.cpp (right):

http://codereview.chromium.org/8615/diff/1401/1208#newcode100
Line 100: // Need to flip the y coordinates since the font's coordinate system
is
80 cols, if you care.

http://codereview.chromium.org/8615/diff/1401/1208#newcode255
Line 255: Vector<CachedOutlineKey> outlinesToDelete;
I guess Vector<OutlineCache::iterator> would break awfully?