Re: [Freetype] Arabic

[Werner LEMBERG]

> I had assumed that Character Map was creating the unfindable glyphs
> on-the-fly from their component parts (adding diacritics) yet U06ba
> (Arabic Letter Noon Ghunna) is not found and has no diacritical
> marks (unless the absence of a dot is sort of diacritic).

In case OpenType features are active (which is very likely for
Arabic), there are many glyphs not accessible with a cmap file since
the OpenType table lookup is based on glyph indices, happening after
the Cmap->glyph index conversion.

> Can anyone suggest how I might develop the glyphs that Character Map
> can produce but freetype seems not able to?

Only manually.  Get Just's fonttools package (from fonttools.sf.net),
to analyze the font.  Then you can set up a fixed cmap for the
particular font.  Anyway, I doubt this makes sense since those glyphs
are intentionally hidden from the cmap, I think.  You should use
FreeType 1 if you really need GSUB and GPOS support right now.
Otherwise, I suggest you try IBM's ICU, STSF (from stsf.sf.net), or
Pango.


    Werner