OntopVirtualRepository fails to initialize for Direct Mapping of Table or View with No Primary Key

[thomasjtaylor]

Description

May be similar to #395 (Ontop bootstrap generate mapping without primary keys).

I have a database that has Tables/Views without a primary key. MySQL CREATE TABLE

MySQL CREATE TABLE

-- MySQL 8
CREATE DATABASE `ONTOP_KEYS`;

CREATE TABLE `NoPK` (
  `Column1` varchar(100) DEFAULT NULL,
  `Column2` varchar(100) DEFAULT NULL
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_general_ci;


INSERT INTO `ONTOP_KEYS`.`NoPK` (Column1,Column2) VALUES
	 (NULL,NULL),
	 ('2',NULL),
	 (NULL,'3'),
	 ('4','4');

I have a tool that generates a R2RML Direct Mapping for this table. Following the R2RML specs, this mapping should define r2rml:termType r2rml:BlankNode in the r2rml:subjectMap,

• https://www.w3.org/TR/r2rml/#dfn-default-mapping
• https://www.w3.org/2001/sw/rdb2rdf/directMapping/LC/#no-pk

R2RML Direct Mapping

#  R2RML Mapping Created Thu Aug 24 18:16:29 EDT 2023
@prefix rdb: <urn:its:r2rml-mapping#> .
@prefix rr: <http://www.w3.org/ns/r2rml#> .

rdb:NoPK a rr:TriplesMap;
  rr:logicalTable [
      rr:tableName "\"ONTOP_KEYS\".\"NoPK\""
    ];
  rr:subjectMap [
      rr:graph    rr:defaultGraph;
      rr:termType rr:BlankNode;
      rr:class    <rdb:schemas/ONTOP_KEYS/NoPK>
    ];
  rr:predicateObjectMap [
      rr:graph     rr:defaultGraph;
      rr:predicate <rdb:/schemas/ONTOP_KEYS/NoPK#Column2>;
      rr:objectMap [
          rr:column "\"Column2\"";
          rr:datatype <http://www.w3.org/2001/XMLSchema#string>
        ]
    ], [
      rr:graph     rr:defaultGraph;
      rr:predicate <rdb:schemas/ONTOP_KEYS/NoPK#Column1>;
      rr:objectMap [
          rr:column "\"Column1\"";
          rr:datatype <http://www.w3.org/2001/XMLSchema#string>
        ]
    ] .

ODBA Mapping

(after modifying r2rml-api-core:0.9.0 to permit rr:BlankNode in rr:subjectMap)
>ontop-cli mapping to-odba

[PrefixDeclaration]
rdb:		urn:its:r2rml-mapping#
rr:		    http://www.w3.org/ns/r2rml#
rdf:		http://www.w3.org/1999/02/22-rdf-syntax-ns#
rdfs:		http://www.w3.org/2000/01/rdf-schema#
owl:		http://www.w3.org/2002/07/owl#
xsd:		http://www.w3.org/2001/XMLSchema#
obda:		https://w3id.org/obda/vocabulary#

[MappingDeclaration] @collection [[
mappingId	mapping--1104171867
target		rr:BlankNode a <rdb:schemas/ONTOP_KEYS/NoPK> ; <rdb:schemas/ONTOP_KEYS/NoPK#Column1> {"Column1"}^^xsd:string ; <rdb:schemas/ONTOP_KEYS/NoPK#Column2> {"Column2"}^^xsd:string . 
source		SELECT * FROM "ONTOP_KEYS"."NoPK"
]]

However, Ontop:5.1 (r2rml-api-core:0.9.0) is throwing an NullPointerException when I try to create an OntopVirtualRepository connection.
It looks like eu.optique-project:r2rml-api-core:0.9.0 doesn't properly support Term Type Blank Nodes in the Subject Map.
eu.optique.r2rml.api.model.impl.R2RMLMappingCollectionImpl github

If I update R2RMLMappingCollectionImpl to allow blank nodes, it fixes the NullPointerException. However, Ontop produces the wrong result.

R2RMLMappingCollectionImpl

// look for RDF star triple
if (resource == null) {
  resource = readObjectInMappingGraph(node, getRDF().createIRI(R2RMLVocabulary.PROP_TERM_TYPE));
  if ((resource instanceof IRI) && resource.toString().equals(RDFStarVocabulary.TERM_STAR_TRIPLE)) {
    termMapType = TermMap.TermMapType.RDF_STAR_VALUED; 
  }	// TODO: Change comparison
// BEGIN-FIX
  else if ((resource instanceof IRI) && resource.toString().equals(R2RMLVocabulary.TERM_BLANK_NODE)) {
    return mfact.createSubjectMap((IRI) resource);
 }
// END-FIX

Running a simple query: SELECT * FROM NoPK produces incorrect results

SPARQL query:
SELECT  (?noPK_1_column1 AS ?Column1) (?noPK_1_column2 AS ?Column2)
WHERE
  { ?_cls_NoPK  a  <rdb:schemas/ONTOP_KEYS/NoPK> ;
      <rdb:schemas/ONTOP_KEYS/NoPK#Column1>  ?noPK_1_column1 ;
      <rdb:schemas/ONTOP_KEYS/NoPK#Column2>  ?noPK_1_column2
  }

Resulting native query:
ans1(Column1, Column2)
CONSTRUCT [Column1, Column2] [Column2/RDF(VARCHARToTEXT(Column21m4),xsd:string), Column1/RDF(VARCHARToTEXT(Column11m5),xsd:string)]
   NATIVE [Column11m5, Column21m4]
SELECT DISTINCT v4.`Column1` AS `Column11m5`, v5.`Column2` AS `Column21m4`
FROM (SELECT 1 AS uselessVariable
FROM `NoPK` v1
LIMIT 1) v3, `NoPK` v4, `NoPK` v5
WHERE (v4.`Column1` IS NOT NULL AND v5.`Column2` IS NOT NULL)

Expected:

Column1	Column2
4	4

Actual:

Column1	Column2
2	3
4	3
2	4
4	4

NullPointerException

java.lang.NullPointerException: A TriplesMap must have a SubjectMap.
	at eu.optique.r2rml.api.model.impl.TriplesMapImpl.setSubjectMap(TriplesMapImpl.java:85) ~[r2rml-api-core-0.9.0.jar:?]
	at eu.optique.r2rml.api.model.impl.TriplesMapImpl.<init>(TriplesMapImpl.java:56) ~[r2rml-api-core-0.9.0.jar:?]
	at eu.optique.r2rml.api.model.impl.MappingFactoryImpl.createTriplesMap(MappingFactoryImpl.java:40) ~[r2rml-api-core-0.9.0.jar:?]
	at eu.optique.r2rml.api.model.impl.R2RMLMappingCollectionImpl.readTriplesMap(R2RMLMappingCollectionImpl.java:157) ~[r2rml-api-core-0.9.0.jar:?]
	at eu.optique.r2rml.api.model.impl.R2RMLMappingCollectionImpl.readTriplesMaps(R2RMLMappingCollectionImpl.java:138) ~[r2rml-api-core-0.9.0.jar:?]
	at eu.optique.r2rml.api.model.impl.R2RMLMappingCollectionImpl.load(R2RMLMappingCollectionImpl.java:103) ~[r2rml-api-core-0.9.0.jar:?]
	at eu.optique.r2rml.api.model.impl.R2RMLMappingCollectionImpl.<init>(R2RMLMappingCollectionImpl.java:80) ~[r2rml-api-core-0.9.0.jar:?]
	at eu.optique.r2rml.api.model.impl.R2RMLMappingManagerImpl.extractR2RMLMapping(R2RMLMappingManagerImpl.java:105) ~[r2rml-api-core-0.9.0.jar:?]
	at eu.optique.r2rml.api.model.impl.R2RMLMappingManagerImpl.importMappings(R2RMLMappingManagerImpl.java:66) ~[r2rml-api-core-0.9.0.jar:?]
	at it.unibz.inf.ontop.spec.mapping.parser.impl.R2RMLMappingParser.parse(R2RMLMappingParser.java:102) ~[ontop-mapping-r2rml-5.1.0.jar:5.1.0]
	at it.unibz.inf.ontop.spec.mapping.parser.impl.R2RMLMappingParser.parse(R2RMLMappingParser.java:78) ~[ontop-mapping-r2rml-5.1.0.jar:5.1.0]
	at it.unibz.inf.ontop.spec.mapping.impl.SQLMappingExtractor.extractPPMapping(SQLMappingExtractor.java:128) ~[ontop-mapping-sql-core-5.1.0.jar:5.1.0]
	at it.unibz.inf.ontop.spec.mapping.impl.SQLMappingExtractor.extract(SQLMappingExtractor.java:108) ~[ontop-mapping-sql-core-5.1.0.jar:5.1.0]
	at it.unibz.inf.ontop.spec.impl.DefaultOBDASpecificationExtractor.extract(DefaultOBDASpecificationExtractor.java:51) ~[ontop-mapping-core-5.1.0.jar:5.1.0]
	at it.unibz.inf.ontop.injection.impl.OntopMappingConfigurationImpl.loadSpecification(OntopMappingConfigurationImpl.java:140) ~[ontop-mapping-core-5.1.0.jar:5.1.0]
	at it.unibz.inf.ontop.injection.impl.OntopMappingSQLConfigurationImpl.loadSpecification(OntopMappingSQLConfigurationImpl.java:85) ~[ontop-mapping-sql-core-5.1.0.jar:5.1.0]
	at it.unibz.inf.ontop.injection.impl.OntopMappingSQLAllConfigurationImpl.loadSpecification(OntopMappingSQLAllConfigurationImpl.java:47) ~[ontop-mapping-sql-all-5.1.0.jar:5.1.0]
	at it.unibz.inf.ontop.injection.impl.OntopSQLOWLAPIConfigurationImpl.loadOBDASpecification(OntopSQLOWLAPIConfigurationImpl.java:35) ~[ontop-system-sql-owlapi-5.1.0.jar:5.1.0]
	at it.unibz.inf.ontop.injection.impl.OntopOBDAConfigurationImpl.loadSpecification(OntopOBDAConfigurationImpl.java:44) ~[ontop-obda-core-5.1.0.jar:5.1.0]
	at it.unibz.inf.ontop.injection.OntopSystemConfiguration.loadQueryEngine(OntopSystemConfiguration.java:20) ~[ontop-system-core-5.1.0.jar:5.1.0]
	at it.unibz.inf.ontop.rdf4j.repository.impl.OntopVirtualRepository.initializeInternal(OntopVirtualRepository.java:72) ~[ontop-rdf4j-5.1.0.jar:5.1.0]
	at org.eclipse.rdf4j.repository.base.AbstractRepository.init(AbstractRepository.java:36) ~[rdf4j-repository-api-4.2.2.jar:4.2.2]
	at it.unibz.inf.ontop.rdf4j.repository.impl.OntopVirtualRepository.getConnection(OntopVirtualRepository.java:53) ~[ontop-rdf4j-5.1.0.jar:5.1.0]
	at it.unibz.inf.ontop.rdf4j.repository.impl.OntopVirtualRepository.getConnection(OntopVirtualRepository.java:22) ~[ontop-rdf4j-5.1.0.jar:5.1.0]

Steps to Reproduce

1. Configure Ontop with an R2RML Direct Mapping configuration for a table with no primary key
2. Instantiate an OntopVirtualRepository with the R2RML Mapping

Versions

Ontop:5.1.0

Additional Information

SEE: Example rdb2rdf test w/o PK: #D012
SEE: #395

[thomasjtaylor]

Attached Ontop query reformulation log. ontop-query-plan.txt