[PDFBOX-5879] Regression from PDFBOX-5841: Text extraction with rotation magic fails for PDF with multiple content streams in a page

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.0.32, 3.0.3 PDFBox
• Fix Version/s: 2.0.33, 3.0.4 PDFBox, 4.0.0
• Component/s: Text extraction
• Assignee: Tilman Hausherr
• Reporter: Gábor Stefanik
• Created: 16/Sep/24 15:13
• Updated: 17/Sep/24 11:33
• Resolved: 17/Sep/24 07:34

Description

java -jar pdfbox-app-3.0.3.jar export:text -console -rotationMagic -i="MVM_Aram_augusztus.pdf" 

fails with the following error:

java.lang.ClassCastException: class org.apache.pdfbox.cos.COSObject cannot be cast to class org.apache.pdfbox.cos.COSArray (org.apache.pdfbox.cos.COSObject and org.apache.pdfbox.cos.COSArray are in unnamed module of loader 'app')
        at org.apache.pdfbox.tools.ExtractText.extractPages(ExtractText.java:336)
        at org.apache.pdfbox.tools.ExtractText.call(ExtractText.java:225)
        at org.apache.pdfbox.tools.ExtractText.call(ExtractText.java:62)
        at picocli.CommandLine.executeUserObject(CommandLine.java:2045)
        at picocli.CommandLine.access$1500(CommandLine.java:148)
        at picocli.CommandLine$RunLast.executeUserObjectOfLastSubcommandWithSameParent(CommandLine.java:2465)
        at picocli.CommandLine$RunLast.handle(CommandLine.java:2457)
        at picocli.CommandLine$RunLast.handle(CommandLine.java:2419)
        at picocli.CommandLine$AbstractParseResultHandler.execute(CommandLine.java:2277)
        at picocli.CommandLine$RunLast.execute(CommandLine.java:2421)
        at picocli.CommandLine.execute(CommandLine.java:2174)
        at org.apache.pdfbox.tools.PDFBox.main(PDFBox.java:76) 

The same command succeeds in 3.0.2.

The triggering PDF can be downloaded from https://nagykorosiallatmentok.hu/wp-content/uploads/2023/09/MVM_Aram_augusztus.pdf, and is also attached.

The root cause appears to be this change: https://github.com/apache/pdfbox/commit/b03d12d56dd74e5c52d80cf0b80c5bfb1f3209b2 from PDFBOX-5841

Issue Links

• is related to: TIKA-4311 Avoid potential ClassCastException in angle detection PDF extraction