Re: [Freetype] Certain characters of pbf fonts don't render?

[Joris van der Hoeven]

On Fri, 23 Jan 2004, Joris van der Hoeven wrote:
> On Fri, 23 Jan 2004, Jan-Åke Larsson wrote:
> > Joris van der Hoeven wrote:
> > > I just implemented the support for pfb font rendering inside GNU TeXmacs,
> > > but it seems that certain characters are not rendered at all...
> > > For example, nor the "fi" ligature (character 12) of cmr10,
> > > nor most symbols in cmsy10 are rendered. Is this a known problem,
> > > or did I do something wrong?
> >
> > This is most probably an encoding problem. You _do_ translate the
> > character code into a glyph index? Usually this is done by
> >
> >    FT_Get_Char_Index( face, c )
> >
> > These glyphs are not (as far as I know) accessible in any of the ADOBE
> > encodings (you probably selected one of those). Well, ISTR that
> > FT_ENCODING_ADOBE_CUSTOM works in some cases but not in others.
> > Anyway, the TeX fonts have their own encodings, usually present on
> > disk in an .enc file. My solution to this was to load such an encoding
> > from disk and use it. This is a map from character code to glyph
> > _name_. I then get the glyph index by
> >
> >    FT_Get_Name_Index( face, enc_charname[c] )
> >
> > I would have liked FT2 to load the encoding, but I could not find
> > anything in the API docs for it. There is a possibility to attach .afm
> > (metric) files to a font, but not custom encodings like this. It is
> > not difficult to code this, of course. If you want to see an example,
> > look at the dvipng code (a fast DVI-to-PNG translater), accessible at
> >
> >   http://sourceforge.net/projects/preview-latex
> >
> > The CVS is at
> >
> >   http://cvs.sourceforge.net/viewcvs.py/preview-latex/preview/dvipng/
> >
> > You will want to look at the files ft.c and enc.c (and possibly fontmap.c)
>
> Thanks for your answer. You are probably right; I forgot about this.
> In fact, I already have my own names for all characters and
> I already know the character indices for all TeX fonts.
> Can I somehow tell FreeType to get the character by its raw index?

After a lot of trying, it seems that

        FT_Select_Charmap (ft_face, ft_encoding_adobe_custom);

indeed does the trick... Strange, since this has nothing to do
with Adobe encodings. Maybe FreeType should provide

        ft_encoding_custom

for all fonts and use ft_encoding_adobe_custom as an alias to
this encoding when it applies.

I also noticed a small bug in the documentation:
you often write CharMap instead of Charmap.

Best wishes, Joris