[PDFBOX-5127] Multithreading issue in JempBox's DateConverter

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.8.16
• Fix Version/s: 1.8.17
• Component/s: JempBox
• Labels: multithreading
• Assignee: Tilman Hausherr
• Reporter: Tim Allison
• Created: 12/Mar/21 12:55
• Updated: 15/Sep/22 17:21
• Resolved: 13/Mar/21 10:27

Description

tilman recently found an exception thrown from here (https://github.com/apache/pdfbox/blob/1.8/jempbox/src/main/java/org/apache/jempbox/impl/DateConverter.java#L186) in one run of tika-eval but not in another. 

 

This is a multithreading issue caused by https://github.com/apache/pdfbox/blob/1.8/jempbox/src/main/java/org/apache/jempbox/impl/DateConverter.java#L43 SimpleDateFormat is not threadsafe.  I'm surprised we haven't seen this earlier, but so it goes.

 

Many, many thanks to Tilman for finding this!