[TIKA-1118] OOXML parser throws when relationship points to 0 byte embedded part

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.3
• Fix Version/s: 1.7
• Component/s: parser

Environment

Tested on MAC and Ubuntu server

• Assignee: Unassigned
• Reporter: Lee Graber
• Created: 14/May/13 20:53
• Updated: 05/Aug/14 12:15
• Resolved: 04/Aug/14 16:05

Description

I have a test document (pptx) where there is a 0 byte embedded part which is referenced in a relationship. I don't really know how the document got like this but Office can open it without any issues. The problem is in AbstractOOXMLExtractor::handleEmbeddedOle. It attempts to create a POIFSFileSystem:
POIFSFileSystem fs = new POIFSFileSystem(part.getInputStream());
using the part stream in the constructor but fails because of insufficient data. Given that the whole function except for this one line is in a try -> catch -> ignore, it seems like the best option is to put this line inside the try / catch. There is no metadata to extract from the 0 byte part.