SPARQL query parsing fails when the query has nested SERVICE calls

[MathiasVDA]

Hello all,

I'm trying to create a SPARQL syntax checker to be used as github action. I'm running in some trouble for more advanced queries. Below I use a reproducible simple example.

Considering this SPARQL query:

PREFIX ex: <http://example.org/>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>

CONSTRUCT {
    ?subject ex:hasInfo ?info .
    ?subject ex:hasDetail ?detail .
}
WHERE {
    SERVICE <http://example.org/endpoint1> {
        ?subject rdfs:label ?label .
        ?subject ex:hasIdentifier ?id .
        
        SERVICE <http://example.org/endpoint2> {
            ?subject ex:info ?info .
            OPTIONAL {
                ?subject ex:detail ?detail .
            }
        }
    }
}

And this python script:

import os
import sys
from pathlib import Path
from rdflib.plugins.sparql import parser

def validate_sparql(file_path):
    with open(file_path, 'r', encoding='utf-8') as f:
        query = f.read()
    
    try:
        # Parse the query to check syntax
        parser.parseQuery(query)
        print(f"✅ {file_path} is valid")
        return True
    except Exception as e:
        print(f"❌ {file_path} has syntax error: {str(e)}")
        print(f"::error file={file_path}::SPARQL syntax error: {str(e)}")
        return False

def main():
    directory = Path('.')
    
    if not directory.exists():
        print(f"Error: Directory {directory} does not exist")
        return 1
    
    all_valid = True
    
    for file_path in directory.glob('*.rq'):
        print(f"Validating {file_path}")
        is_valid = validate_sparql(file_path)
        if not is_valid:
            all_valid = False
    
    return 0 if all_valid else 1

if __name__ == "__main__":
    sys.exit(main())

The script will report a syntax error "maximum recursion depth exceeded" although there is no error. I believe there is an issue with the parser.parseQuery(query) line not being able to handle the nested SERVICE calls.

[henrieglesorotos]

Might be related to this - #3069. I don't think rdflib actually supports multiple SERVICE keywords?

[MathiasVDA]

That seems to be related indeed!

[vemonet]

The OG issue about this with longest discussion is here :) #2136

It's a major issue, nested SERVICE are not uncommon (wikidata uses service for label resolution, so if you do a service call to wikidata and wants label you get nested service)