Prefixes Don't Appear in XML Serialization of rdf:datatype

[temcdrm]

I'm trying to use prefixes in rdf:datatype attributes, and they appear in turtle serializations but not pretty-xml nor xml. Test code:

import rdflib

CIM_NS = 'http://www.ucaiug.org/profile#'

if __name__ == '__main__':
  g = rdflib.Graph()
  CIM = rdflib.Namespace (CIM_NS)
  g.bind('cim', CIM)

  ac = rdflib.URIRef ('placeholderForID')
  g.add ((ac, rdflib.RDF.type, rdflib.URIRef (CIM_NS + 'ACLineSegment')))
  g.add ((ac, rdflib.URIRef (CIM_NS + 'Equipment.inService'), rdflib.Literal (True, datatype='cim:Boolean')))
  g.add ((ac, rdflib.URIRef (CIM_NS + 'Conductor.length'), rdflib.Literal (1.0, datatype=CIM.Length)))
  g.add ((ac, rdflib.URIRef (CIM_NS + 'ACLineSegment.r'), rdflib.Literal (1.0)))

  g.serialize (destination='test.xml', format='pretty-xml', max_depth=1)

Output test.xml:

<?xml version="1.0" encoding="utf-8"?>
<rdf:RDF
  xmlns:cim="http://www.ucaiug.org/profile#"
  xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
>
  <cim:ACLineSegment rdf:about="placeholderForID">
    <cim:Equipment.inService rdf:datatype="cim:Boolean">true</cim:Equipment.inService>
    <cim:Conductor.length rdf:datatype="http://www.ucaiug.org/profile#Length">1.0</cim:Conductor.length>
    <cim:ACLineSegment.r rdf:datatype="http://www.w3.org/2001/XMLSchema#double">1.0</cim:ACLineSegment.r>
  </cim:ACLineSegment>
</rdf:RDF>

Expected test.xml:

<?xml version="1.0" encoding="utf-8"?>
<rdf:RDF
  xmlns:cim="http://www.ucaiug.org/profile#"
  xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
  xmlns:xsd="http://www.w3.org/2001/XMLSchema#"
>
  <cim:ACLineSegment rdf:about="placeholderForID">
    <cim:Equipment.inService rdf:datatype="cim:Boolean">true</cim:Equipment.inService>
    <cim:Conductor.length rdf:datatype="cim:Length">1.0</cim:Conductor.length>
    <cim:ACLineSegment.r rdf:datatype="xsd:double">1.0</cim:ACLineSegment.r>
  </cim:ACLineSegment>
</rdf:RDF>

[WhiteGobo]

I have a patch, that produces your expected test.xml but some other issues came up:

xml parser seems to not understand attributes with:

<ex:prop rdf:datatype="xsd:integer">1.0</ex:prop>

It is interpreted as Literal("1.0", "xsd:integer") instead of Literal(1).

Also i havent found where this abbreviation of rdf:datatype="ex:short_datatype" is described.
But i dont know xml or xml/rdf good enough to know if that expected syntax is correct or not.

[niklasl]

I'm afraid RDF/XML does not expand values in attributes like that. You need to use URIs in rdf:datatype (which can be relative to @xml:base, although it is rarely desired to set that to e.g. the xsd namespace).

One reason for this is because, formally, XML parsers aren't required to preserve the prefix name itself, so it's not formally accessible after parsing. For example. <x:e xmlns:a="urn:x-test:a"/> is allowed to be changed to something like <ns0:e xmlns:ns0="urn:x-test:a"/>. Here's an old article by James Clark with some details. RDFa 1.1 deprecated the use of xmlns in favour of a prefix attribute to avoid this formal limitation.

(Notably, in practise, this is not the case, since for example, xpaths in XSLT wouldn't work! Also XML C14N requires preservation of prefix names. But RDF/XML was designed before that, and this is unlikely to change.)

Historically, for these cases, entities defined in a local DTD declaration have been used to shorten attribute values e.g. in @rdf:datatype. Here's an example I grabbed from some lecture slides:

<?xml version="1.0"?>
<!DOCTYPE rdf:RDF [
    <!ENTITY xsd "http://www.w3.org/2001/XMLSchema#">
]>
<rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
                 xmlns:mon="http://www.semwebtech.org/mondial/10/meta#"
                 xml:base="http://www.semwebtech.org/mondial/10/">
    <mon:Country rdf:about="countries/D">
        <mon:name rdf:datatype="&xsd;string">Germany</mon:name>
        <mon:population rdf:datatype="&xsd;int">83536115</mon:population>
    </mon:Country>
</rdf:RDF>

It's not pretty but it works. (In compliant XML parsers that is, such as the one RDFLib uses. In JS libraries I've found this to be more hit-and-miss.)

[temcdrm]

Okay, thanks for the information from @niklasl and @WhiteGobo. I will close this issue because 1) it seems like full URI is the preferred and safest approach, and 2) it seems like rdflib doesn't output a serialization with DOCTYPEs. I'm relying on rdflib to make my RDF/XML files.