[PDFBOX-5833] NullPointerException when calling Splitter.split() with pdf files generated by Keynote

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Duplicate
• Affects Version/s: 3.0.2 PDFBox
• Component/s: Utilities
• Labels: bug regression
• Environment: Happens both on our server environment (Linux) and when testing locally (Windows)
• Assignee: Unassigned
• Reporter: Lucas Welscher
• Created: 04/Jun/24 14:12
• Updated: 04/Jun/24 15:43
• Resolved: 04/Jun/24 15:43

Description

After upgrading from PDFBox 3.0.1 to version 3.0.2 I experienced a NullPointerException when using the split() method:

java.lang.NullPointerException: Cannot invoke "org.apache.pdfbox.pdmodel.common.PDNumberTreeNode.getNumbers()" because "tree" is null
 at org.apache.pdfbox.multipdf.PDFMergerUtility.getNumberTreeAsMap(PDFMergerUtility.java:1156)
 at org.apache.pdfbox.multipdf.Splitter.cloneStructureTree(Splitter.java:192)
 at org.apache.pdfbox.multipdf.Splitter.split(Splitter.java:135)

 

When testing with different inputs, I noticed that the issue only occurred when using a PDF generated from Keynote. I attached an example below to reproduce the issue.

Downgrading to 3.0.1 again solves the issue.

This is proably caused by PDFBOX-2725 and this commit. The issue happens in this segment of Splitter.java:

PDNumberTreeNode srcParentTree = srcStructureTreeRoot.getParentTree();
// srcParentTree is null here
Map<Integer, COSObjectable> srcNumberTreeAsMap = PDFMergerUtility.getNumberTreeAsMap(srcParentTree);

 

Issue Links

• duplicates: PDFBOX-5809 PDDocument#importPage slowed down by factor 1300