I observed the following error. 

févr. 12, 2021 3:20:44 PM org.grobid.core.document.Document addTokenizedDocument
GRAVE: Cannot parse file: my_dir/grobid-0.6.1/grobid-home/tmp/ZW3PpM4E9g.lxml_outline.xml

févr. 12, 2021 3:20:44 PM org.grobid.core.document.Document addTokenizedDocument
GRAVE: Cannot parse file: my_dir/grobid-0.6.1/grobid-home/tmp/ZW3PpM4E9g.lxml_outline.xml

févr. 12, 2021 3:20:46 PM org.grobid.core.engines.ProcessEngine createTraining
INFOS: 2 files processed.

it originates in an illegal nullPointer access in file PDFALTOOutlineSaxHandler.java. 

In understand that the error is not  that "GRAVE". According to @kermitt2, it is simply one of the generated XML file resulting from the PDF parsing which is not XML valid (very frequent) - it has no impact because the outline file (containing the table of content outline if available embedded in the PDF) is not exploited by GROBID for the moment - it's to allow some possible improvements in the future. The error is more a reminder for the developers... the XML parser that classifies it as "GRAVE" but it would be rather INFO for us.

Unfortunately, a side effect might have been overlooked. If father is null, then father.addChild(currentNode) is called with a nullPointer exception.

This exception is caught by catch (Exception e) at line 372 in grobid-core/src/main/java/org/grobid/core/document/Document.java where the error message "Cannot parse" is misleading.

I think an additional else is just missing.

I cleaned up the correction as requested. Luc.

…ing a test case, rename test to match the class name

…items