encryption string/unicode bug

[Mathieu Fenniak]

Originally reported via e-mail from Geoff Gerrietts:

I had to make a "fix" to pyPdf -- on lines 1344 and 1444 of pdf.py I had to change id1_entry to id1_entry.original_bytes.

For some of my PDFs, md5 was trying to convert unicode to ascii and failing during this step of the process.

This seems to have corrected my problem, but there's some lingering suspicion that I have not discovered all the places this needs attention.

[Emanuele Rocca (ema)]

Hello Mathieu,

I can confirm that changing m.update(id1_entry) with m.update(id1_entry.original_bytes) in _alg32 fixes the problem for my encrypted files as well.

[KLEIN Stéphane (stephane-harobed)]

This my patch to fix this bug.