Make the plugin ready for the new data model

[zyzzyxdonta]

Known bug: The new data model uses schema: <http://schema.org/> while rdflib used in the validation code uses schema: <https://schema.org/> which causes validation to be skipped for all non-https things.

[zyzzyxdonta]

Hey @notactuallyfinn, here is another case where @SKernchen and I refactored a HERMES plugin to use the new data model. Would you be willing to take a look at the code to see if this makes sense? Thanks! 😄

[notactuallyfinn]

Looks good to me, only a few very minor things.

[notactuallyfinn]

Why not use the HermesCacheManager the way it was intended to be used?
(validation_file would have to be moved, but is there another reason?)

[zyzzyxdonta]

What would be the intended way? Do you mean with a SoftwareMetadata object?

We tried that but failed because the validation graph is a list and SoftwareMetadata can only be initialized with a dict. It looks like this:

validation.json.txt

The interesting bit in this case is the validation report in ll. 443 - 454. The rest is just SHACL copied into the graph.

[zyzzyxdonta]

I tried working around it, taking a concise bounded description:

        report_refs = list(
            self._validation_graph.subjects(
                predicate=RDF.type, object=SH.ValidationReport
            )
        )
        assert len(report_refs) == 1
        report_ref = report_refs[0]
        self._validation_graph.cbd(report_ref).serialize(
            validation_file, format="json-ld"
        )

But this also creates a list:

[
  {
    "@id": "_:Nd5d1172d2d2d46829819e6fdce9722a9",
    "@type": [
      "http://www.w3.org/ns/shacl#ValidationReport"
    ],
    "http://www.w3.org/ns/shacl#conforms": [
      {
        "@type": "http://www.w3.org/2001/XMLSchema#boolean",
        "@value": true
      }
    ]
  }
]

Maybe the JSON-LD serializer in RDFlib always does this.

I could return the single element from that list. But that won't work for failed validations because in those cases, the report is split into multiple objects rather than nested:

validation-failed.json.txt