Use sh:maxLength/sh:length Instead of SPARQL

[VladimirAlexiev]

https://github.com/Sveino/Inst4CIM-KG/tree/develop/shacl-improved#use-shmaxlength-instead-of-sparql:
Consider this overcomplicated shape in CGMES-NC/r2.3/ap-con/ttl/NC-AP-Con-Complex-IdentifiedObjecStringLength-SHACL.ttl:

io:IdentifiedObjectStringLength  a sh:NodeShape ;
        sh:property     io:IdentifiedObject.name-stringLength;
        sh:targetNode   cim:IdentifiedObjectStringLength .

io:IdentifiedObject.name-stringLength
        a               sh:PropertyShape ;
        sh:description  "The string IdentifiedObject.name has a maximum of 128 characters." ;
        sh:sparql       io:IdentifiedObject.name-stringLengthSparql ;
        sh:path         rdf:type ;
        sh:group        io:IOstringLength ;
        sh:name         "C:452:ALL:IdentifiedObject.name:stringLength" ;
        sh:order        1 ;
        sh:severity     sh:Violation .

io:IdentifiedObject.name-stringLengthSparql
    a         sh:SPARQLConstraint ;
    sh:message      "String length is greater than 128 characters." ;
    sh:prefixes cim: ;
    sh:select """
      SELECT  $this ?value
      WHERE {
        ?s cim:IdentifiedObject.name ?value
        FILTER (STRLEN(?value)>128) .
      }""" .

We don't need the fake node, the misleading sh:path rdf:type, or in fact any SPARQL. We can target by class or property and simplify it to:

io:IdentifiedObjectStringLength  a          sh:NodeShape ;
        sh:property         io:IdentifiedObject.name-stringLength;
        sh:targetSubjectsOf cim:IdentifiedObject.name .

io:IdentifiedObject.name-stringLength
        a               sh:PropertyShape ;
        sh:description  "The string IdentifiedObject.name has a maximum of 128 characters." ;
        sh:path         cim:IdentifiedObject.name ;
        sh:maxLength    128;
        sh:group        io:IOstringLength ;
        sh:name         "C:452:ALL:IdentifiedObject.name:stringLength" ;
        sh:order        1 ;
        sh:severity     sh:Violation .

• https://github.com/entsoe/application-profiles-library/blob/main/validate/shacl-sparql-stats.tsv#L182-L185 shows 4 shapes with that defect
• Don't Overuse SHACL SPARQL in 2024 lists several more cases of SPARQL over-use
  Alternative or Disjunction Instead of SPARQL
  Use sh:maxCount Instead of SPARQL
  Use sh:maxLength Instead of SPARQL
  Use sh:equals Instead of SPARQL
  Generalize Shapes
  Use targetObjectsOf Instead of SPARQL Subquery

@HarisVranaj can you add @nataschake to this repo? She's made a fix and I want to assign to her.

[griddigit-ci]

TargetNode issue was already fixed, but there is something more to be fixed later this week

[VladimirAlexiev]

@nataschake Nice!

• there's one more to fix:

cd application-profiles-library\validate\shacl-sparql
grep -i strlen *
Model.modelingAuthoritySet-marp10-12.rq:          FILTER(STRLEN(str(?value))=0).

• please after this one submit Pull Requests so we can review them line by line.

[griddigit-ci]

A question. Did you do positive/negative test with some engine?
I was planning to check but wanted to know if you already did.

[VladimirAlexiev]

We are waiting for negative tests from @griddigit-ci or @HarisVranaj (so we can start expressing them semantically).
We haven't tested with any negative data.

But we have tested ReliCapGrid with all positive shapes: https://nikolatulechki.github.io/relicap-ap-shacl-eval/validation-results.html.
(Notice there are also counts by constraint component).

[nataschake]

evaluate against relicapgrid data that is loaded into the GraphDB repository relicapgrid on cim.ontotext.com.
More - here

[nataschake]

A question. Did you do positive/negative test with some engine? I was planning to check but wanted to know if you already did.

There are only positive tests, no wrong data.

[griddigit-ci]

I did negative testing and as expected it is OK. However I only take the changes for NCP 2.5. The pull request has also some other changes that we should not apply.
So the remaining actions are for CGMES 3.x