[TIKA-2726] Handle truncated ooxml more robustly

• Type: Task
• Status: Resolved
• Priority: Major
• Resolution: Duplicate
• Fix Version/s: 2.0.0, 1.21
• Assignee: Tim Allison
• Reporter: Tim Allison
• Created: 14/Sep/18 13:20
• Updated: 03/Jan/19 20:33
• Resolved: 03/Jan/19 20:33

Description

With the move from Tika 1.18 to 1.19, we're now failing to detect some truncated ooxmls more specifically than tika-ooxml.  In the attached example, which Excel is able to fix, in 1.18, this file is identified as application/vnd.ms-excel.sheet.macroenabled.12, and it is parsed without exception by the ooxml parser.  However, in 1.19, this is identified as tika-ooxml and then parsed by the Package parser.

Issue Links

• is related to: TIKA-2765 Regression extracting text from corrupted docx files