SPARQL result parsing

[white-gecko]

Summary of changes

• Remove plugin registration for "application/sparql-results+xml; charset=UTF-8"
  • The ResultParser plugin with the name "application/sparql-results+xml; charset=UTF-8" since it is never called in internal code. The only place where ResultParser plugins are requested is in "rdflib/query.py:276". In case of a content-type the charset is split off already.
  • cf. #941
• Resolve TODO: "Pull in these from the result implementation plugins?" for _response_mime_types in sparqlconnector
  • The _response_mime_types dict is replaced by a method response_mime_types that generates an Accept filed containing the supported mime types. Per default these are all mime types for which ResultParser plugins are registered.
• Dynamically register the rdflib.plugins.sparql.results.graph.GraphResultParser plugin for all names of the Parser plugins since it calls the Parsers internally.

Eventually, these changes could be superseded by a more flexible plugin system, that would distinguish between plugin name, supported format, and supported mime-type.

The changes should maintain backward compatibility.

But this API is now more flexible. The returnFormat for SPARQLConnector and SPARQLStore is now optional and None per default. Also the default Accept header lists all mime-types supported by parsers. Maybe this should be mentioned in the change notes.

Checklist

• [x] Checked that there aren't other open pull requests for the same change.
• [x] Checked that all tests and type checking passes.
• [x] Considered granting push permissions to the PR branch, so maintainers can fix minor issues and keep your PR up to date.

[ashleysommer]

@white-gecko this looks great, thanks. This addresses an issue I had only last month, I was trying to dynamically get the correct SPARQL result parser based on a mimetype, though the RDFLib plugin system, and I couldn't get it to work. Looks like now there is a way to do it.

[ashleysommer]

@white-gecko I've clearned up the commit pipeline, so we don't have Dependabot conflicts, mypy issues, black updates and ruff updates in the loop to prevent the proper testing of PRs. I've just re-ran the test suite on your PR, and it is failing one real test:

test_example[sparqlstore_example.py]

[white-gecko]

@ashleysommer I'm sorry, I don't really understand the test output, since it is successful for some runs but the output of the other test doesn't show the failing test.

Ok, I got it, these are failing:

    graph = Graph("SPARQLStore", identifier="http://dbpedia.org")
    graph.open("http://dbpedia.org/sparql")

    pop = graph.value(URIRef("http://dbpedia.org/resource/Berlin"), dbo.populationTotal)

This can be due to the content type, mimetype handling.

[ashleysommer]

Weirdly I'm seeing that exact same test_example[sparqlstore_example.py] failure on this unrelated PR: https://github.com/RDFLib/rdflib/pull/2818 Do you know what could be causing that?