Nov 25, 2006 : Javascript RDFParser from Tabulator

David Sheets introduces himself and the new RDF/XML parser he added to the Tabulator all-javascript RDF browser. I've been using the RDF parser from Jim Ley in a generic JavaScript RDF Editor (to work with any OWL/schema files; no public release yet, but let me know if you're interested, especially in helping out). So recently I worked on using Tabulator's rdf parser to see if it was better. It seems to be designed a bit better (separate data store from parser from XML loading) and David's post says it is more accurate and often faster.

I had been hoping to find some additional functionality to use in my JsRdfEditor. Namely, modelling the RDF document instead of just getting all the triples. For example, knowing where a nodeID was used, and the XML arrangement of the triples. That way the JsRdfEditor could parse a file, add some triples, and output a new document that is arranged much like the one that was loaded. But no luck there; sounds like I'll have to add that capability to the parser myself, if I want to have elegant RDF/XML output.

So here's how you can use Tabulator's RDF Parser in your own projects:

// depends on:
//  dig.csail.mit.edu/2005/ajar/ajaw/rdf/term.js
//  dig.csail.mit.edu/2005/ajar/ajaw/uri.js
//  dig.csail.mit.edu/2005/ajar/ajaw/rdf/rdfparser.js

// TestStore implementation from dig.csail.mit.edu/2005/ajar/ajaw/test/rdf/rdfparser.test.html
// see also RDFIndexedFormula from dig.csail.mit.edu/2005/ajar/ajaw/rdf/identity.js
//  (extends RDFFormula from dig.csail.mit.edu/2005/ajar/ajaw/rdf/term.js no indexing and smushing)
// for the real implementation used by Tabulator which uses indexing and smushing
var store = new TestStore()
var parser = new RDFParser(store);
parser.reify = parser.forceRDF = true;
// forceRDF isn't used??


var url = 'http://something.or/other';

// get the XML
var xhr = XMLHTTPFactory(); // returns a new XMLHttpRequest, or ActiveX XMLHTTP object
if (xhr.overrideMimeType) {
    xhr.overrideMimeType("text/xml");
}
// the "data/" path and encoding is just how I store files locally
xhr.open("GET", "data/" + encodeURIComponent(encodeURIComponent(url)), false);
xhr.send("");
var nodeTree = xhr.responseXML;
if (nodeTree === null && xhr.responseText !== null) {
    // Only if the server fails to set Content-Type: text/xml AND xmlhttprequest doesn't have the overrideMimeType method
    console.debug("no responseXML, parsing responseText");
    nodeTree = (new DOMParser()).parseFromString(xhr.responseText, 'text/xml');
}
// must be an XML document node tree
parser.parse(nodeTree,url);

// use FireBug extension to inspect console.debug'd objects
// Using TestStore you can access store.triples
console.debug('store',store);