Pinto

Pinto is a Java framework for transforming Java beans into RDF (and back).

The current version is 2.0 as of 2016-06-14. We're using git flow for development.

Inspired by Jackson and Empire, it aims to be simple and easy to use. No annotations or configuration are required. If you have a compliant Java bean Pinto will turn it into RDF.

The rdf4j framework is used to represent RDF primitives such as Graph and Statement.

Building

To create the artifacts:

$ gradle jar

And to run the tests:

$ gradle test

Example Usage

Given this simple Java Bean:

public static final class Person {
    private String mName;

    public Person() {
    }

    public Person(final String theName) {
        mName = theName;
    }

    public String getName() {
        return mName;
    }

    public void setName(final String theName) {
        mName = theName;
    }

    @Override
    public int hashCode() {
        return Objects.hashCode(mName);
    }

    @Override
    public boolean equals(final Object theObj) {
        if (theObj == this) {
            return true;
        }
        else if (theObj instanceof Person) {
            return Objects.equal(mName, ((Person) theObj).mName);
        }
        else {
            return false;
        }
    }
}

You can easily serialize it as RDF:

Graph aGraph = RDFMapper.create().writeValue(new Person("Michael Grove"));

And aGraph serialized as NTriples:

<tag:complexible:pinto:f97658c7048377a026111c7806bd7280> <tag:complexible:pinto:name> "Michael Grove"^^<http://www.w3.org/2001/XMLSchema#string> .

And if you have that RDF, you can turn it back into a Person:

final Person aPerson RDFMapper.create().readValue(aGraph, Person.class)

This is the quick and dirty example, but for more detailed examples, check out the tests.

Annotations

Pinto does not require annotations to serialize Beans as RDF, but does support a few basic annotations so you can control how the object is serialized.

@RdfId

An annotation which specifies the properties to be used for generating the URI of the object. By default, a hash of the object itself is used to generate a URI. But when present on a getter or setter of one or more properties on the bean, the values of those properties will be used as the seed for the URI.

Note: There is a secondary mechanism for controlling the URI of an object. If the object implements Identifiable the mapper will use the URI returned by #id ignoring any @RdfId annotated properties.

@RdfProperty

An annotation which can be applied to a property on a bean, either the getter or the setter, which specifies the URI of the property when generating RDF for the bean. Normally, a URI for the property is auto-generated, but when this annotation is present, the URI specified in the annotation is used instead. The value of the annotation can also be a QName.

@RdfsClass

An annotation which can be applied to a class to specify the rdf:type of the class when generating the RDF. Can be a QName or a URI. When not present, no rdf:type assertion is generated for the object.

@Iri

Annotation which can be used to control the URI assigned to an Enum. Normally the URI's are generated by Pinto, but if you want to map them to specific constants in your ontology, you can use Iri to explicitly identify the objects.

public enum TestEnum {
    @Iri("urn:my_ontology:test:Foo")
    Foo,

    @Iri("urn:my_ontology:test:Bar")
    Bar,
}

Configuration

By default, RDFMapper does not require any configuration, it's meant to generate reasonable RDF out of the box. There are a couple (de)serialization options which are specified by MappingOptions:

• REQUIRE_IDS - By default, Pinto will auto-generate URIs for objects when @RdfId is not specified. By setting this property to true the mapper will not auto-generate URIs, they must be specified explicitly. (default: false)
• SERIALIZE_COLLECTIONS_AS_LISTS - When true, collections are serialized as RDF lists. Otherwise, they're serialized using Collection#size separate property assertions. (default: false)
• IGNORE_INVALID_ANNOTATIONS - Whether or not to ignore an annotation which is invalid, such as @RdfProperty which defines a property with an invalid URI. Properties with invalid/ignored annotations are simply not used when generating a Bean or RDF. (default: true)

Beyond these configuration options, RDFMapper has a few other configuration mechanisms that can be specified on its Builder when creating the mapper:

• #map(URI, Class) - Specify the provided type corresponds to instances of the given Java class. Functions like the @RdfsClass annotation.
• #namespace(...) - Methods to specify namespace mappings which are used to expand any QNames used in the annotations
• #valueFactory(ValueFactory) - Provide the ValueFactory to be used when creating RDF from a bean
• #collectionFactory(CollectionFactory) - The factory to be used for creating instances of java.util.Collection. Defaults to DefaultCollectionFactory
• #mapFactory(MapFactory) - The factory to be used for creating instances of java.util.Map. Defaults to DefaultMapFactory

Custom serialization

In some cases, an object won't adhere to the Java Bean specification, or it's a third-party class that you don't control so you cannot add annotations, but you need a specific serialization. For these cases RDFCodec can be used. It's a small plugin to RDFMapper which will handle transforming a Java object to/from RDF. Pinto includes an example implementation of a codec for java.util.UUID called UUIDCodec.

Codecs are registered when the RDFMapper is created via its builder: Builder.codec(Class<T>, RDFCodec<T>)

Why Pinto?

Why create Pinto when there are similar frameworks available? Well, the other frameworks, like Empire or Alibaba are focused on more than just transforming Beans into RDF and back. Neither are a good fit for just round-tripping between beans and RDF.

A good example is if you're building a JAX-RS based web service and you have some bean in your domain that you'd like to serialize as RDF, or accept as RDF, that's normally done with a custom implementation of MessageBodyReader/MessageBodyWriter. But that implementation is not straight-forward with the heavier-weight frameworks. With Pinto, it's a single line of code.