Xemph: XMP Parser

Abstract

This project is just a new XMP parser. It has been inspired of the xmpbox module in Apache PDFBox project. On the contrary of Xmpbox, I did not try keep compatibility with the Jempbox.

Most of the namespaces described in the XMP specification are implemented. Namespaces helpers are generated from a configuration file. It is possible to create new namespaces with a simple configuration file and using the xemph generator plugin for maven..

Examples

Here are some examples based on PDFBox

Extracting XMP object from PDF

How to extract xmp packet from PDF file

// open PDF file
File input = new File ("files/XMPSpecificationPart1.pdf");
RandomAccessFile raf = new RandomAccessFile(input,"r");
PDFParser parser = new PDFParser(raf);
parser.parse();
PDDocument document = parser.getPDDocument();

// retrieve serialized xmp from document catalog
InputStream inputstream = document.getDocumentCatalog().getMetadata().exportXMPMetadata();

// deserialize
XmpReader xmpReader = new XmpReader();
Packet packet = xmpReader.parse(inputstream);

// display
for (Map.Entry<Name,Value> entry: packet.getProperties().entrySet()) {
    System.out.println(entry.getKey()+" : "+entry.getValue().getClass()+" / "+entry.getValue());
}
System.out.println(Xmp.getMetadataDate(packet).asDate().getTime());
System.out.println(XmpMM.getInstanceID(packet).asUUID());

OrderedArray<Value> creators = DublinCore.getCreator(packet);
for (Value sv : creators.getItems()) {
    System.out.println(">> "+sv.getClass().getSimpleName()+" : "+sv.toString());
}
System.out.println("Contains DC:title = "+DublinCore.containsTitle(packet));

Creating a PDF with XMP packet

This example is extracted from xemph-example (PDFCreator.java)

// create the PDF Document
PDDocument document = new PDDocument();
PDPage blankPage = new PDPage();
document.addPage( blankPage );

// create the xmp packet
Packet packet = new Packet();
DublinCore.setCreator(packet,"Guillaume Bailleul");
// There is no helper for this one
packet.addProperty(DublinCore.DATE, new Date());
DublinCore.setPublisher(packet,UnorderedArray.parse("publisher1","publisher2","publisher3","publisher4"));

// serialize the xmp packet
XmpWriter writer = new XmpWriter();
byte [] xmpContent = writer.write(packet);

// add the serialized xmp in the document catalog
PDMetadata metadata = new PDMetadata(document);
metadata.importXMPMetadata(xmpContent);
document.getDocumentCatalog().setMetadata(metadata);
document.save("example.pdf");

Creating a specific namespace

Current namespaces are defined in yaml files (see core/namespaces).

It is possible to generate new namespaces with a simple yaml file and with a maven plugin. Here is an example of namespace; default prefix will be ec, its url will be http://example.com/my_ns/1.0 and it will have 5 properties (yes this is an extract of Dublin Core).

default-prefix: ec
uri: http://example.com/my_ns/1.0
classname: ExampleNamespace
properties:
  - name: contributor
    type: Unordered
    element: Text
  - name: coverage
    type: Text
  - name: creator
    type: Ordered
    element: ProperName
  - name: date
    type: Ordered
    element: Date
  - name: description
    type: LangAlternative

In your project pom, add this maven plugin:

<build>
        <plugins>
        ...
            <plugin>
                <groupId>xemph</groupId>
                <artifactId>generator</artifactId>
                <version>${project.version}</version>
                <executions>
                    <execution>
                        <goals>
                            <goal>genNamespaces</goal>
                        </goals>
                        <configuration>
                            <packageName>com.example.namespaces</packageName>
                        </configuration>
                    </execution>
                </executions>
            </plugin>
         ...
        </plugins>
    </build>

The classes will be generated in target/generated-sources/namespaces

Format

Only XML parsing and writing is implemented