RDFLib / rdflib / 6017747210 / 5

90%

master: 91%

push

github

fix: `queryGraph` selection for `query` and `update` (#2546)

`Store.query` and `Store.update` have a parameter `queryGraph`, which
presumably specifies the initial [active
graph](https://www.w3.org/TR/sparql11-query/#sparqlDataset).

When SPARQL is evaluated against a `Dataset`, the initial active graph
should be the default graph
[[ref](https://www.w3.org/TR/sparql11-query/#sparqlAlgebraEval)],
however, when calling `rdflib.graph.Graph.query` or
`rdflib.graph.Graph.update`, they were passing
`rdflib.graph.Dataset.identifier` as the value of `queryGraph` to
`Store.query` and `Store.update`. The passed value would not have been
the identifier of the default graph of the Dataset, and would instead
have been an identifier that has really no meaning whatsoever (see
<https://github.com/RDFLib/rdflib/issues/2537>).

This change fixes the selection of the initial active graph by passing
`ConjunctiveGraph.default_context.identifier` if the `Graph.query` or
`Graph.update` method is called on a `ConjunctiveGraph` and
`ConjunctiveGraph.default_union` is `False`.

Related issues:
- Fixes <https://github.com/RDFLib/rdflib/issues/2538>.

6507 of 8112 branches covered (0.0%)

18384 of 20807 relevant lines covered (88.35%)

0.88 hits per line