Please provide a link to such an unreadable PDF.

Then try pdftohtml -xml on it.

And please keep in mind that PDFs which don't use a standard font are not necessarily parsable, cause they might embed an own font with random code-points for the glyphs.

In this case deciphering is only possible with character and word recognition. Either automatic (OCR) or human (by populating a hash $glyph{codepoint} for each unknown font)

HTH! (Inshallah =)