Parsing apidoc as RDF graph

[cr3a7ure]

Continued from this issue #142 .
As jsonld spec states (https://www.w3.org/TR/json-ld/#identifying-blank-nodes), if there is no '@id' key, then we might end up in a BlankNode. I'm not sure if this is the right place to describe the issue, but i think we should think about it. Have a look at this example:

 "@id": "http://schema.org/Flight",
"@type": "hydra:Class",
"rdfs:label": "Flight",
"hydra:title": "Flight",
"hydra:description": "An airline flight.",
"hydra:supportedProperty": [
{ /blank_node_start
    "@id": <----------------missing id--------------------------------?
    "@type": "hydra:SupportedProperty",
    "hydra:property": {
        "@id": "#Flight/id", <----------
        "@type": "rdf:Property",
        "rdfs:label": "id",
        "domain": "http://schema.org/Flight",
       "range": "xmls:integer"
    },
"hydra:title": "id",
"hydra:required": false,
"hydra:readable": false,
"hydra:writable": true
},/blank_node_end

so what we have is BlankNode_23 ---> '@type' ---> "hydra:SupportedProperty"
etc, meaning that it would be difficult to check the hydra properties [title,required,readable,writable].

The same is true for supportedOperations and others. I believe that this has to do with Hydra itself.
Below is some js code, in order to remove blank nodes, but it may refer the same nodes twice, i'm not sure. I'm editing the jsonld and then parse it through rdflib.js . You can see the parts that have no '@id' .

hydraObject = JSON.parse(hydraDoc);

hydraObject['hydra:supportedClass'].forEach(function(supClass){
    supClass['hydra:supportedProperty'].forEach(function(obj){
        obj['@id'] = obj['hydra:property']['@id']+'/hydra'
        // console.log(obj['hydra:property']['@id'])
    })
    // console.log(supClass["hydra:supportedOperation"])
    if (Array.isArray(supClass["hydra:supportedOperation"])) {
        supClass["hydra:supportedOperation"].forEach(function(obj){
            obj['@id'] = supClass['@id'] + '/' + obj["hydra:method"]
            // console.log(obj['@id'])
        })
    }
    if (supClass['@id'].includes("Entrypoint")) {
        supClass['hydra:supportedOperation']['@id'] = supClass['@id'] + '/' + supClass['hydra:supportedOperation']['hydra:method']
        console.log(supClass['hydra:supportedProperty'])
        supClass['hydra:supportedProperty'].forEach(function(obj){
            // console.log(obj)
            obj['hydra:property']['hydra:supportedOperation'].forEach(function(propObj){
                propObj['@id'] = obj['@id'] + '/' + propObj['hydra:method']
            })
        })
    }

})
console.log(hydraObject)
hydraDoc = JSON.stringify(hydraObject);

I'm posting this for thoughts rather than a solution

[dunglas]

I don't really get why blank nodes are a problem.

Our implementation is inspired of the output of https://github.com/lanthaler/HydraBundle and works well with Hydra console (https://github.com/lanthaler/HydraConsole).

Maybe the Hydra mailing list is a better place to discuss that?

/cc @lanthaler

[lanthaler]

Yeah. @cr3a7ure, why do you think the blank nodes are problematic?

[Simperfit]

@cr3a7ure any update on this ?