Namespaces and Namespace Prefixes

Questions frequently come up in the Hydra community about XML namespaces, namespace prefixes, and XPath. This page will attempt to resolve some of the confusion that comes up without diving too deep into the guts of XML.

A namespace prefix is a shorthand placeholder for a thing, not the thing itself. The prefix only exists within a certain scope, while the namespace itself is durable. For example, the following two RDF documents are equivalent:

<rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
  xmlns:ns0="http://projecthydra.org/ns/relations#"
  xmlns:ns1="info:fedora/fedora-system:def/model#"
  xmlns:ns2="info:fedora/fedora-system:def/relations-external#">
    <rdf:Description rdf:about="info:fedora/avalon:1212">
        <ns0:hasModelVersion>R3</ns0:hasModelVersion>
        <ns1:hasModel rdf:resource="info:fedora/afmodel:MasterFile"/>
        <ns2:isPartOf rdf:resource="info:fedora/avalon:1211"/>
    </rdf:Description>
</rdf:RDF>

<rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">
    <rdf:Description rdf:about="info:fedora/avalon:1212">
        <hasModelVersion xmlns="http://projecthydra.org/ns/relations#">R3</hasModelVersion>
        <hasModel xmlns="info:fedora/fedora-system:def/model#"
            rdf:resource="info:fedora/afmodel:MasterFile"/>
        <isPartOf xmlns="info:fedora/fedora-system:def/relations-external#"
            rdf:resource="info:fedora/avalon:1211"/>
    </rdf:Description>
</rdf:RDF>

while this third one is not:

<rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">
    <rdf:Description rdf:about="info:fedora/avalon:1212">
        <hasModelVersion>R3</hasModelVersion>
        <hasModel rdf:resource="info:fedora/afmodel:MasterFile"/>
        <isPartOf rdf:resource="info:fedora/avalon:1211"/>
    </rdf:Description>
</rdf:RDF>

In the first example, all four XML namespaces are assigned prefixes on the root element of the document. In the second, each predicate is assigned its namespace as a scoped default (applying only to that element and its children). In the third, all predicate-specific namespaces are removed, destroying the semantic meaning of those predicates, because an XML element's identity is a combination of its namespace URI (not prefix!) and its name.

All of the following are true:

1. The XPath query //ns1:hasModel will raise an error against any of these three documents unless the XPath engine (not the document) knows that ns1 is a placeholder for the namespace URI info:fedora/fedora-system:def/model#.
2. The XPath query //fedora-model:hasModel will find the hasModel assertions in the first two documents, provided the XPath engine has defined fedora-model as a prefix for info:fedora/fedora-system:def/model#.
3. The XPath query //hasModel will do one of three things:

• If the XPath engine's default namespace has been set to info:fedora/fedora-system:def/model#, it will find the hasModel assertions in the first two documents as expected.
• If the XPath engine's default namespace is undefined, it will find only the hasModel assertion in the third document, but it won't mean the same thing to ActiveFedora because the namespace is missing.
• If the XPath engine's default namespace is set to something else, it will return an empty result set.

4. The XPath query //*[local-name()="hasModel"] tells the XPath engine to ignore namespaces and find all elements with the local name hasModel. It will find all three hasModel assertions, but the third one won't mean the same thing as the first two. Trust me, you don't want to do this.
5. The XPath query //*[local-name()="hasModel" and namespace-uri()="info:fedora/fedora-system:def/model#"] will work exactly like example 2 above, regardless of which namespaces the XPath engine knows about.