How can I understand fuzzing around with the font?

How can I tel to FOP Serializer to embed only glyphs are used?

Should I write wrappers
of FOPSerializer reader of .ttf files and .ttf.xml files,  which ones
will skip font data of unused symbols for each document?

The problem is the code which embeds the font into the PDF, it
just puts in the whole file summarily. Nobody bothered yet to
write code for extracting definitions of used glyphs and the
associated data (which is the hard part).