NamespaceManager.expand_curie() does not handle the default prefix

[jclerman]

I have found that rdflib's handling of cURIEs for entities within an ontology's own namespace is somewhat difficult.

One of the issues is already documented (#2077) - namely, that NamesapceManager does not handle mapping multiple prefixes to a single URI.

I also see that a cURIE like :my_class is rejected by expand_curie() as invalid. It seems like it should be valid though, based on https://www.w3.org/TR/curie/#s_syntax, in particular

A host language MAY declare a default prefix value, or MAY provide a mechanism for defining a default prefix value. In such a host language, when the prefix is omitted from a CURIE, the default prefix value MUST be used.

rdflib does record the default prefix value, mapping it to the empty string, as can be seen by examining NamespaceManager.namespaces() - but that default value is never used in expand_curie(). The problematic code appears to be

rdflib/rdflib/namespace/__init__.py

Line 618 in 5c3c78a

if len(parts) != 2 or len(parts[0]) < 1:

, which requires that the prefix in CURIE to be expanded has non-zero length - but that's in conflict with the (correct) storage of the default prefix in namespaces().

The two issues make it difficult (impossible?) to expand a CURIE for an entity in the default namespace of an ontology:

• The default namespace prefix is inaccessible per the above
• A substitute prefix, mapping to the same namespace, is dropped by NamespaceManager, per rdflib 6.2.0 SPARQL parsing regression: multiple prefixes for same IRI not supported? #2077.

Is there a workaround via which I could write a CURIE that maps to an item in a parsed ontology (e.g. one stored in RDF/XML), such that rdflib could expand it to the correct URI?

[aucampia]

Is there a workaround via which I could write a CURIE that maps to an item in a parsed ontology (e.g. one stored in RDF/XML), such that rdflib could expand it to the correct URI?

I'm not aware of any, I will look at the multiple prefixes issue this month if I get time, though I'm going to try get our test-suite for JSON-LD up to scratch first to prepare for PyLD integration with RDFLib.

I also see that a cURIE like :my_class is rejected by expand_curie() as invalid. It seems like it should be valid though, based on https://www.w3.org/TR/curie/#s_syntax, in particular

It seems like it should be valid, and think just fixing expand_curie to allow this should be uncontroversial. I'm a bit unsure if the way RDFLib stores the default prefix is optimal, but it is what it is, and making it work for expanding curies seems pretty reasonable.