The issue, as it turns out, is that Adobe added a section to the standard when crypt filters were added. In my case, the StrF and StmF use the same algorithm as would be used without crypt filters, so that's not an issue. The real issue here is that "EncryptMetadata", which was added with crypt filters, passes 0xFFFFFFFF to the MD5 hash following the first element of the file's identifier array, when EncryptMetadata is false (it defaults to true).
I've developed a fix and verified it for my use case. I've never submitted code to a project in my life; I'm guessing I should bring it up on the dev list. Gary On Fri, Aug 27, 2010 at 3:19 PM, Andreas Lehmkuehler <[email protected]>wrote: > Hi, > > Am 27.08.2010 21:10, schrieb Gary LosHuertos: > > The encryption dictionary is as follows (I omitted the user and owner >> strings): >> 132 0 obj<</R 4/Length 128/Filter/Standard/O(...)/P -1324/U(...)/V >> 4/CF<</StdCF<</Length >> 16/CFM/V2/AuthEvent/DocOpen>>>>/StmF/StdCF/StrF/StdCF/EncryptMetadata >> false>> >> endobj >> > Looks like the "new" encryption algorithm which isn't supported yet. > See [1] for further details. > > BR > Andreas Lehmkühler > > [1] https://issues.apache.org/jira/browse/PDFBOX-697 >
