[TIKA-3701] ZipDetector on a file should back off to streaming detection on failure to open a zipfile

• Type: Task
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 2.4.0
• Assignee: Unassigned
• Reporter: Tim Allison
• Created: 15/Mar/22 23:27
• Updated: 18/Mar/22 20:51
• Resolved: 15/Mar/22 23:29

Description

If a file is passed to Tika wrapped as a TikaInputStream with an underlying file, the DefaultZipDetector tries to open a ZipFile. If there's a truncated file or if that ZipFile open fails, the DefaultZipDetector effectively gives up.

Given that there's still a file available, we should try to do a streaming detect by reopening the file as a regular InputStream.

If we don't do this, we wind up getting different detection for some truncated ooxml if the user sends in a file vs a stream.