20 Sep
2009
20 Sep
'09
12:06 a.m.
Barry Schwartz
In practice what I see with my method is that discretionary hyphens always get an ActualText, and if the font is older and has names like "Asmall" or "ffl" (which I don't bother handling specially) then the substituted stuff gets an ActualText. I could look at the font's internal encoding the way I think Cairo does, but it doesn't matter a whole lot.
Oops, "ffl" is in the Adobe Glyph List and so would get put into the ToUnicode. Something like "ffh" wouldn't, however, but "f_f_h" would because it can broken down into parts that are in the AGL.