New Forum | Previous | Next | (P-PDF) Developers
Topic: Difference in Cross-Reference
Conf: (P-PDF) Developers, Msg: 52805
Date: 5/29/2002 04:54 PM
This is my first post in this forum.
We are creating a PDF/FDF parser in java using the PJ library.
We came across some differences between Acrobat 4.05 Windows and Acrobat 4.05 Mac.
According to the PDF Reference manual version 1.3: one cross-reference table entry is *EXACTLY* 20 bytes long
On Windows they are *19* bytes long!!!! And our parser crashes on that file.
Strangely this file can be read with the Mac version of Acrobat, and If we do a "Save as" the cross-reference entries are 20 bytes long and our parser can successfully decode the file.
Conclusion: In order to parse Windows Acrobat generated PDF file we should:
1) test the length of the cross-reference entry
2) if the length is 19 bytes subtract the number of entries found (i.e. number of missing bytes in the xref) on each offset address in the cross-reference this, for each offset greater than the number of bytes missing in the xref.
Is this a bug or a feature?