XCRI-CAP to RDF and back again

This repository contains XSL stylesheets to convert XCRI-CAP 1.2 XML into RDF, a Python script to turn such RDF back into XML, and SKOS concept schemes for a few relevant controlled vocabularies.

Stylesheets

These stylesheets convert XCRI-CAP 1.2 into RDF/XML using appropriate vocabularies.

You will probably want to extend and override parts of the main stylesheet (xcri2rdf.xsl) in order to model locally-important concepts and provide identifiers for your courses, presentations, and so forth.

Using

Make sure you have an XSLT 2.0 processor installed (we'll assume saxon).

Then, simply:

$ saxon your-xcri-feed.xml xcri2rdf.xsl > your-xcri-feed.rdf

Overriding

The daisy2rdf.xsl stylesheet should give you some clues. Note the <xsl:import /> line, and how we have overriden the rdf-about template to check some special case before deferring to the implementation in xcri2rdf.xsl using <xsl:apply-imports />.

Python script

Prerequisites

You'll need the rdflib module, which you can install using either easy_install or pip, or using a Linux distribution package manager.

API

python/xcri_rdf.py contains an XCRICAPSerializer class which takes an rdflib.Graph instance as its only argument. The graph must contain a subject of type xcri:catalog:

import rdflib
from xcri_rdf import XCRICAPSerializer

graph = rdflib.ConjunctiveGraph()
graph.parse('course-data.rdf')
serializer = XCRICAPSerializer(graph)

You can then serialize the graph to XML using either its serialize or generator methods. The former takes a file-like object (e.g. a file, or a socket), whereas the later yield strings:

import sys
serializer.serialize(sys.stdout)

# or

for str in serializer.generator():
    print str

As a script

The Python module may also be invoked as a script:

$ python -m xcri_rdf file-one.rdf thesauri/*/*.rdf

The script is invoked with some number of RDF/XML files as arguments. Include the thesauri files so that things like studyMode and attendancePattern are suitably serialized.

Feedback

Feedback is very welcome to [email protected].