VeraPdf does not remove Temporary Files

[manuelpache]

Hi,
I have problems with the VeraPdf-PDFAParser. The Parser does not delete the Temp-Files of the validated PdfA-Document. Manually closing the Parser does not help. This Problem appears both locally on Mac and on our Linux-Servers.

The Dependency:

<dependency>
   <groupId>org.verapdf</groupId>
   <artifactId>validation-model</artifactId>
   <version>1.14.104</version>
 </dependency>

The Code:

VeraGreenfieldFoundryProvider.initialise();
PDFValidatorAnswer answer = new PDFValidatorAnswer();
PDFAFlavour flavour = PDFAFlavour.valueOf(pdfNorm);
PDFAValidator validator = Foundries.defaultInstance().createValidator(flavour, false);

try (PDFAParser parser = Foundries.defaultInstance().createParser(new ByteArrayInputStream(Base64.getDecoder().decode(base64String)), flavour)) {
  ValidationResult result = validator.validate(parser);

  answer.setIsCompliant(result.isCompliant());
  answer.setMessages(result.getTestAssertions().stream().map(TestAssertion::getMessage).collect(Collectors.toList()));
}

catch (ValidationException | EncryptedPdfException | ModelParsingException | IOException e) {
  answer.addMessagesItem(e.getMessage());
}

If you require any further information, let me know.
Thanks in advance

** I was able to narrow the problem a little bit down. The Temp-File gets correctly deleted, if the validation ends without problems. By debugging the Code I found that the InternalInputStream::closeResource() Method is the problem. I case that there are problems, the numOfFileUsers-Variable never reaches 0. So the File never gets deleted. I hopes this helps you a little bit. **

[bdoubrov]

We were not able to reproduce the issue with your code and a random set of documents. As soon as the parser is closed, all temp files are correctly removed.

Would you please provide more details:

• if the issue is reproduced only on some of the documents, it would be great if you could share any of these
• if the issue is reproduced on your side for any PDF document, then please provide more info on your runtime environment

[bdoubrov]

Could you provide any PDF document when this issue is reproduced?

[nico1a]

attached are two test documents. One passes validation, the other does not. Temp files remain with both documents.

• test_document_pdf1.pdf
• test_document_pdfa1b.pdf

As far as I understand the problem, the buffers in the class ASMemoryInStream are not closed correctly (buffer = zero; in the close() method is never reached). I think this causes the variable numOfFileUsers not to be decremented correctly and the tmp-files remain.

[bdoubrov]

The issue is fixed in the current dev version (1.15.33 - see https://software.verapdf.org/develop/1.15/verapdf-greenfield-1.15.33-installer.zip) and will be included into the coming stable version 1.16.