That part of the python code you cited it should cover all the characters of my texts.
What do you mean with "junk PDF"?. I don't know about the inner pdf characteristics although my files seem to be right. (As an example, here one of them:
http://www.archive.org/details/Cetasikas )
I don't know if the pdf needs some inner definition already implemented before to be converted.
thanks for the help,