org.apache.jena.shared.PrefixMapping$IllegalPrefixException: 😀 about jena HOT 6 CLOSED

Steps to reproduce in a new fuseki dataset:

1. Save data below into a trig file
2. Upload to fuseki using "Add Data" tab in web interface
3. Query DESCRIBE<https://ex.com/Test1> Returns Result
4. Query DESCRIBE<https://ex.com/Test1> FROM <urn:x-arq:DefaultGraph> Error 502

@prefix test: <https://ex.com/ontologies/> .
@prefix  😀: <https://ex.com/> .
@prefix xsd: <http://www.w3.org/2001/XMLSchema#> .

 😀:Test1 a test:Test ;
    test:title "Test"^^xsd:string ;
.

from jena.

Fuseki is behind a proxy?

It is running in a container on kubernetes behind a Traefik ingress

from jena.

I just tried it on 5.0.0 and I get a similar error

from jena.

This bug report is incomplete. As you note it appears to be triggered by your data but you have failed to provide that data

I would also suggest experimenting with sending different Accept headers as it's possible your data has a prefix declaration that is legal for some RDF syntaxes but not others

from jena.

Fuseki is behind a proxy?

org.apache.jena.shared.PrefixMapping$IllegalPrefixException: 😀

FROM triggers DynamicDatasets.dynamicDataset which sets up prefixes as it is created at the start of the request, and does not wait until prefixes are used (which would only delay problems to later in the request execution).

While 😀 is a legal prefix name in Turtle family languages, it is not a legal prefix for RDF/XML. That must be a XML NCName to be used in the predicate position. In RDF/XML properties are qnames.

Jena PrefixMapping add restrictions to ensure that, when used to write RDF/XML data, the writing process won't fail because of prefix names. PrefixMapping is quite old code and we don't want to change the contract.

For most processing there is an internal abstraction that supports TriG.

The code in GraphUnionRead should skip such prefixes.

from jena.

PR #2386 includes allowing "😀" in graph prefix mappings. It turns out that Jena's RDF/XML writing was already quite defensive and converted illegal prefixes (illegal to RDF/XML writing in XML 1.0) to give them a generated legal name. Combine that with ensuring writing RDF/XML follows the more restrictive XML 1.0 with limitation to java char size, then prefixes generally compatible across Turtle TriG etc.

from jena.