Sync with Chromium's fork

Sync with Chromium's fork

* Add the following font metrics fields:
    + fAvgCharWidth
    + fXMin
    + fXMax
    + fXHeight

* Add \#if 0'ed includes for the Freetype headers for Xcode, scons and
  other tools which chase headers.

* Correctly render 1bpp fonts (mostly Chinese fonts at small sizes)

[reed, 2009-04-22 14:09:59]

http://codereview.appspot.com/41081/diff/1/2
File include/core/SkPaint.h (right):

http://codereview.appspot.com/41081/diff/1/2#newcode627
Line 627: SkScalar    fXMin;      //!< The minimum bounding box x value for all
glyphs
I think these min/max values will be useful (assuming they're accurate). Should
we just store a rect, and say its the global bbox for the font, rather than just
the X values?

http://codereview.appspot.com/41081/diff/1/3
File src/ports/SkFontHost_FreeType.cpp (right):

http://codereview.appspot.com/41081/diff/1/3#newcode39
Line 39: #if 0
Will we leave this as #if 0 ?

http://codereview.appspot.com/41081/diff/1/3#newcode768
Line 768: SkFixed myy = fMatrix22.yy;
This happens to work of scalar==float, but the better approach is to call
SkIntToScalar(xmin) / upen. This will work for both fixed and float builds.

[agl, 2009-04-22 18:19:11]

http://codereview.appspot.com/41081/diff/1/2
File include/core/SkPaint.h (right):

http://codereview.appspot.com/41081/diff/1/2#newcode627
Line 627: SkScalar    fXMin;      //!< The minimum bounding box x value for all
glyphs
On 2009/04/22 14:09:59, reed wrote:
> I think these min/max values will be useful (assuming they're accurate).
Should
> we just store a rect, and say its the global bbox for the font, rather than
just
> the X values?

Can I get away with doing that afterwards? I'm afraid that I'm going to mess up
Steven's unforking if I do this now.

http://codereview.appspot.com/41081/diff/1/3
File src/ports/SkFontHost_FreeType.cpp (right):

http://codereview.appspot.com/41081/diff/1/3#newcode39
Line 39: #if 0
On 2009/04/22 14:09:59, reed wrote:
> Will we leave this as #if 0 ?

Yes. The compiler doesn't actually include these files, but build tools will see
the includes and not break. Xcode and scons need this at least.

http://codereview.appspot.com/41081/diff/1/3#newcode768
Line 768: SkFixed myy = fMatrix22.yy;
On 2009/04/22 14:09:59, reed wrote:
> This happens to work of scalar==float, but the better approach is to call
> SkIntToScalar(xmin) / upen. This will work for both fixed and float builds.

Done.

[reed, 2009-04-22 20:29:03]

I'm fine to keep the paint.h as is. We can explore the rect -vs- max/min later
on

http://codereview.appspot.com/41081/diff/1003/2003
File src/ports/SkFontHost_FreeType.cpp (right):

http://codereview.appspot.com/41081/diff/1003/2003#newcode789
Line 789: if (os2 && os2->sxHeight) {
another float -vs- fixed fix

x_height = SkFixedToScalar(SkMulDiv(fScaleX, os2->sxHeight, upem));

http://codereview.appspot.com/41081/diff/1003/2003#newcode797
Line 797: FT_Outline_Get_CBox(&fFace->glyph->outline, &bbox);
x_height = SkIntToScalar(bbox.yMax) / 64;

[agl, 2009-04-22 21:32:55]

http://codereview.appspot.com/41081/diff/1003/2003
File src/ports/SkFontHost_FreeType.cpp (right):

http://codereview.appspot.com/41081/diff/1003/2003#newcode789
Line 789: if (os2 && os2->sxHeight) {
On 2009/04/22 20:29:03, reed wrote:
> another float -vs- fixed fix
> 
> x_height = SkFixedToScalar(SkMulDiv(fScaleX, os2->sxHeight, upem));

Done.

http://codereview.appspot.com/41081/diff/1003/2003#newcode797
Line 797: FT_Outline_Get_CBox(&fFace->glyph->outline, &bbox);
On 2009/04/22 20:29:03, reed wrote:
> x_height = SkIntToScalar(bbox.yMax) / 64;

Done.

[reed, 2009-04-22 21:36:51]

LGTM