[PDFBOX-2855] Allow some flexibility for divergences from the standard on Seq vs Bag in DomXMPParser

• Type: Improvement
• Status: Closed
• Priority: Major
• Resolution: Won't Fix
• Assignee: Unassigned
• Reporter: Tim Allison
• Created: 07/Jul/15 17:45
• Updated: 17/Mar/16 19:14
• Resolved: 07/Jul/15 19:40

Description

While working on TIKA-1285 (migrate to PDFBox 2.0.0), rpialum noticed that the DomXmpParser was failing on some XMP with:

org.apache.xmpbox.xml.XmpParsingException: Invalid array type, expecting Seq and found Bag [prefix=dc; name=creator]
	at org.apache.xmpbox.xml.DomXmpParser.manageArray(DomXmpParser.java:449)
	at org.apache.xmpbox.xml.DomXmpParser.createProperty(DomXmpParser.java:338)
	at org.apache.xmpbox.xml.DomXmpParser.parseChildrenAsProperties(DomXmpParser.java:305)
	at org.apache.xmpbox.xml.DomXmpParser.parseDescriptionRoot(DomXmpParser.java:234)
	at org.apache.xmpbox.xml.DomXmpParser.parse(DomXmpParser.java:198)
	at org.apache.tika.parser.pdf.PDFParser.extractMetadata(PDFParser.java:202)

One file that triggers this is available on Tika-1252 here

The raw xmp for that file includes:

         <dc:creator>
            <rdf:Bag>
               <rdf:li>Single Author</rdf:li>
            </rdf:Bag>
         </dc:creator>

On TIKA-1252, I confirmed that this is against the spec link and [e-mail] confirmed that this was what Acrobat was generating link.

So, would it be easy enough to allow for some divergence from the standard?

Code to reproduce issue in tika setup:

    @Test
    public void oneOffMetadataTest() throws Exception {
        PDDocument doc = PDDocument.load(this.getClass().getResourceAsStream("/test-documents/sampleAcrobat_4_x.pdf"));
        DomXmpParser p = new DomXmpParser();
        p.setStrictParsing(false);
        p.parse(doc.getDocumentCatalog().getMetadata().exportXMPMetadata());
    }

Issue Links

• is depended upon by: TIKA-1285 Upgrade to PDFBox 2.0.0 when available