How to parse RDF/XML, Turtle and JSON-LD resources with jsonld.js?

[jamsden]

I have had good success reading RDF/XML, Turtle and JSON-LD resources with rdflib, and doing in-memory SPQRQL-like queries on the graphs. However, I would like to explore using jsonld.js so that applications can use native JSON and sift.js instead of the rdflib IndexedFormulas and queries.

Unfortunately jsonld.js only parses application/nquads. But it has a means of registering other parsers. I tried to create a jsonld parser for RDF/XML using rdf-ext:

var rdf = require('rdf-ext')();
var rdfXmlParser = function(input, callback) {
    var parser = new rdf.RdfXmlParser();
    parser.parse(input, function doneParsing(dataset) {
        callback(undefined, {'graph': dataset.toArray()});
    });
}
jsonld.registerRDFParser('application/rdf+xml', rdfXmlParser);
jsonld.fromRDF(rs_xml.toString(), {}, function loadRDFXML(err, doc) {
...

But this doesn't work since the RDF dataset format expected by jsonld requires the triples to be URIs or primitive types, not the objects that are created by the rdf-ext parsers.

Are there any examples of RDF/XML, Turtle, and JSON-LD parsers for jsonld.js? If now, how would one go about creating them? Are there things that could be reused like rdflib, rdf-ext, rdfstore-js, or rdf-interfaces that would be useful starting points?

Or is there a better way to do what I'm trying to do - query RDF resources using JSON in JavaScript?

Thanks for the help.

[nicola]

+1 especially for turtle

[dlongley]

Sorry for the slow response, been swamped over here lately.

@nicola, @jamsden, see https://github.com/warpr/turtle-to-jsonld for an example turtle parser that works with jsonld.js.

@jamsden,

Yes, we need both better documentation for how to implement a parser and we need to support RDF Interfaces (related to #89).

jsonld.js is expecting a JavaScript object that represents a dataset that is formatted like so:

{
  "@default": [{
    subject: {type: "IRI", value: "http://example.com/subject"},
    predicate: {type: "IRI", value: "http://example.com/predicate"},
    object: {type: "IRI", value: "http://example.com/object"}
  }, {
    subject: {type: "blank node", value: "_:b0"},
    predicate: {type: "IRI", value: "http://example.com/predicate"},
    object: {type: "literal", value: "foo", datatype: "http://www.w3.org/2001/XMLSchema#string"}
  }, {
    subject: {type: "IRI", value: "http://example.com/subject"},
    predicate: {type: "IRI", value: "http://example.com/predicate"},
    object: {type: "literal", value: "English", language: "en"}
  }, {
    subject: {type: "IRI", value: "http://example.com/subject"},
    predicate: {type: "IRI", value: "http://example.com/predicate"},
    object: {type: "literal", value: "5", datatype: "http://example.com/number"}
  }],
  "http://example.com/named-graph": [{ ... }]
}

[elf-pavlik]

added to https://github.com/rdfjs/rdfjs.org/wiki/Internal-representations

BTW @dlongley @davidlehn we managed in last days to get gitter channel for rdfjs orga active, I've just send both of you invites to that gh orga! https://twitter.com/RubenVerborgh/status/642628797150445568

[paulwilton]

+1 for turtle

[fontanka16]

Three years down the line, what is the established way to parse rdfxml into jsonld today? This thread ranks high on Google and an example would make a good contribution I think.

[gkellogg]

Ideally, jsonld.js would integrate well with rdfjs. Transforming between different RDF representations then becomes a common matter of transforming to the abstract RDF representation, and then re-serialized to the desired output format, such as JSON-LD.

[mkmelin]

Wanted to mention I ended up doing an rdfxml2jsonld package. Not sure if it's the best way or not, but it works at least for my use case.