[PDFBOX-4817] Generated XMP Metadata with other XSLT processor are XML invalid

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.0.19
• Fix Version/s: 2.0.20, 3.0.0 PDFBox
• Component/s: XmpBox
• Assignee: Andreas Lehmkühler
• Reporter: Miroslav Holubec
• Created: 20/Apr/20 08:39
• Updated: 15/Jun/20 04:54
• Resolved: 14/May/20 17:55

Description

In our system I cannot affect javax.xml.transform.TransformerFactory property, it is per default set to org.apache.xalan.processor.TransformerFactoryImpl (Xalan v2.7.2).

When I generate XMP metadata DublinCoreSchema with title, I will get following piece of XML:

<rdf:Description xmlns:dc="http://purl.org/dc/elements/1.1/" rdf:about="">
  <dc:title>
   <rdf:Alt>
     <rdf:li lang="x-default">some title</rdf:li>
   </rdf:Alt>
  </dc:title>
</rdf:Description>

Problem is, that before lang attribute is xml: attribute prefix missing, and then is XML invalid as per specification. With org.apache.xalan.xsltc.trax.TransformerFactoryImpl is everything OK.

Solution to the problem will be providing DocumentBuilderFactory and TransformerFactory from outside via constructor.