sparqlconnector sends correct accept header in case of construct query

[smeckler]

resolves #3332

Summary of changes

For SPARQL construct queries, send mime types for RDF instead of for SPARQL results in the header of the HTTP request.
Some SPARQL servers (in my case Anzo) respond with a 400 'Bad Request' response if the Accept header does not describe RDF for construct or describe queries.

sparqlconnector.py

• get the type of SPARQL query from the passed query string
• if it's a construct or describe query, use an HTTP Accept header that represents an RDF serialization

plugin.py

• register the TurtleParser for the mime type "application/x-turtle" (for compatibility with legacy systems)

Checklist

• Checked that there aren't other open pull requests for
  the same change.

update:

• Checked that all tests and type checking passes.
  7600 passed, 592 skipped, 331 xfailed, 36 xpassed, 36012 warnings in 139.91s (0:02:19)

[WhiteGobo]

I suggest also adding a test like test/test_store/test_store_sparqlstore_query.test_query_construct_format, where you check, that turtle is in request.headers['Accept'].

[WhiteGobo]

As far as i remember we have different parsers available for Construct-queries and Select-queries. So i would rather see use of a variable self.returnFormatConstruct than reuse self.returnFormat. Of course, you would have to add it to __init__ too.

[edmondchuc]

@smeckler Are there plans to address this?

[WhiteGobo]

Because of the failing testrun, you can take a look into developers guide
you can run black and mypy for the exact errors and the autoformatting

poetry run mypy
poetry run black .

Apart from that, this looks fine, to me at least. Sorry for my misunderstanding in your issue description.

[coveralls]

coverage: 90.335%. first build
when pulling e4c2290 on smeckler:improvement/sparqlconnector-construct-query
into 72a98ed on RDFLib:main.

[edmondchuc]

@smeckler Is this PR backwards compatible with existing v7 releases? If so, we should rebase onto 7.x and target the 7.x branch.