[XeTeX] Several suggestions and bug reports concerning Asian language support

[Jjgod Jiang]

Hi Yin, Jonathan,

2007/10/18, Yin Dian <yindian at gmail.com>:
> 2. Wrong width for CJK Ext-B characters.
>
> This bug is introduced in revision 93 of xdvipdfmx. Revision 92 generates
> the correct result, while the following revisions not.

After some debugging, it turns out this bug is introduced by
do_glyph_array() in
dvi.c of xdvipdfmx, after r93, it use FreeType version to determinate whether
to retrieve the horiAdvance data directly from FreeType or load it by
itself, but
there is a little difference between these two cases, which results the
miscalculation for Sun-ExtB.ttf font.

Let's see the calling sequence of case 1:

xdvipdfmx: dvi.c: FT_Load_Glyph
freetype: truetype/ttgload.c: TT_Load_Glyph
freetype: truetype/ttgload.c: load_truetype_glyph
freetype: truetype/ttgload.c: Get_HMetrics

In Get_HMetrics(), if flag FT_LOAD_IGNORE_GLOBAL_ADVANCE_WIDTH
is not set, and face->postscript.isFixedPitch (Sun-ExtB.ttf is declared as
fixed pitch), it will use face->horizontal.advance_Width_Max as the advance
width. (which is 299, but that glyph advance should be 256, as recorded in
SFNT table).

So here is a simple patch of xdvipdfmx to fix this:

Index: src/dvi.c
===================================================================
--- src/dvi.c	(revision 96)
+++ src/dvi.c	(working copy)
@@ -1961,7 +1961,7 @@
             font->glyph_widths[glyph_id] = 0;
         }
         else {
-          FT_Load_Glyph(font->ft_face, glyph_id, FT_LOAD_NO_SCALE);
+          FT_Load_Glyph(font->ft_face, glyph_id, FT_LOAD_NO_SCALE |
FT_LOAD_IGNORE_GLOBAL_ADVANCE_WIDTH);
           font->glyph_widths[glyph_id] = (font->layout_dir == 0)
                                           ?
font->ft_face->glyph->metrics.horiAdvance
                                           :
font->ft_face->glyph->metrics.vertAdvance;


- Jiang