Turning rdflib jsonld into a "full processor" (a.o. for schema.org compliance)

Cloned from RDFLib/rdflib-jsonld#62 :

The JSON-LD 1.1 draft spec mentions different levels of processing for JSON-LD: https://w3c.github.io/json-ld-syntax/#processor-levels

A pure processor can only parse JSON-LD expressed in JSON directly, but a full processor can also parse JSON-LD embedded in HTML.

It would be great if rdflib-jsonld would support this. It would make rdflib-jsonld a library that could be used for HTML documents following the schema.org guidelines for embedding (meta)data in HTML pages as described in their getting started guide https://schema.org/docs/gs.html.

Together with the RDFa & microdata parsers this can then work as a fully RDF based version of the Structured Data Testing tool from Google: https://search.google.com/structured-data/testing-tool.

I'm looking into implementing this using RDFLib/rdflib-jsonld#63 as a starting point. In that code HTML parsing is attempted after JSON parsing fails, but I'm looking at choosing the parse using the content type of the source.

I see that the the JSON-LD test suite in "test/jsonld/1.1" already provides a number of tests which just need to be enabled.

FYI: I'm occasionally working on this at https://github.com/wallberg-umd/rdflib/tree/issue-2692-embedded-jsonld-draft . I've added the basic functionality and enabled the existing tests. I'm now working through making the tests pass.

[wallberg]

I've completed an initial implementation for this issue, see https://github.com/wallberg/rdflib/tree/issue-2692-embedded-jsonld .

It contains one breaking change: when rdflib.plugins.shared.jsonld.util.source_to_json extracts JSON-LD from HTML, it needs to return the value of the HTML base element in addition to the JSON. I took the simplest path and returned a tuple containing the JSON and the base.

I can think of other ways to return the base without breaking the current return value:

1. Return json when processing a json document and (json, base) when processing an html document.
2. Add an optional parameter to return (json, base) instead of json.
3. Continue returning only json, but add an optional parameter which will receive the value of base.

I'd like to get some feedback on the preferred approach before submitting the PR.

A note on the current status of validation:

1. task test passes
2. task lint passes
3. task mypy fails, I believe to existing problems in main
4. poetry run python -m mypy rdflib/plugins/parsers/jsonld.py rdflib/plugins/shared/jsonld/context.py rdflib/plugins/shared/jsonld/util.py test/jsonld/runner.py test/jsonld/test_context.py test/jsonld/test_onedotone.py passes