[SLING-9153] MIME Type detection leaks temporary files when given a ZIP file

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: Content Detection Support 1.0.2
• Fix Version/s: Content Detection Support 1.0.4
• Component/s: Commons
• Assignee: Konrad Windszus
• Reporter: Ankita Agarwal
• Created: 26/Feb/20 11:08
• Updated: 09/Mar/20 12:08
• Resolved: 27/Feb/20 09:37

Description

Apache Sling ContentAwareMimeTypeServiceImpl seems to use the tika API improperly:
https://github.com/apache/sling-org-apache-sling-commons-contentdetection/blob/master/src/main/java/org/apache/sling/commons/contentdetection/internal/ContentAwareMimeTypeServiceImpl.java#L55

Per the Tika javadoc, it is using the wrong method. Instead it should be passing in a TemporaryResources object:
https://github.com/apache/tika/blob/1.22/tika-core/src/main/java/org/apache/tika/io/TikaInputStream.java#L91

 

Temporary disk ran out of space because of a large amount of files named "apache-tika-xxx.tmp".   It seems the bug resides in ContentAwareMimeTypeService OSGi service.  Calls to this service seem to generate the temporary files.

 

Issue Links

• links to: GitHub Pull Request #2