[PDFBOX-1196] Object offsets should be of type long + PATCH

• Type: Improvement
• Status: Closed
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: 1.6.0
• Fix Version/s: 1.7.0
• Component/s: Parsing
• Assignee: Andreas Lehmkühler
• Reporter: Timo Boehme
• Created: 23/Dec/11 15:27
• Updated: 01/Jun/12 11:34
• Resolved: 04/Jan/12 18:10

Description

Currently object offsets are stored as integer (32 bit). However the PDF specification does not limit a PDF document to 2GB. The only restriction is the XREF table which supports 10 digit values for object offsets. Thus we could have offsets >2GB (<10GB). The attached patch changes all locations (as far as I could find them) dealing with object offsets to long type.
Since this is a rather small change which should not have greater performance/space impact I recommend to include it. Otherwise we would have to include checks that the parsed file is smaller then Integer.MAX_VALUE.

Issue Links

• is duplicated by: PDFBOX-1225 java.io.IOException: For input string: "2152829241"