# SHACL shapes for mahabharata_characters.ttl
#
# Two layers:
#   1. SKOS integrity conditions from the SKOS Reference that SHACL can state
#      (S9, S13, S14, S27, S36, S37, S46), as they apply to this scheme.
#   2. Conventions of this scheme: label languages, the character/grouping
#      split, direction and typing of the mbhx: relations, mapping targets.
#
# Run with pyshacl and RDFS inference so that rdfs:subPropertyOf and
# rdfs:subClassOf apply: python validate.py

@prefix sh:      <http://www.w3.org/ns/shacl#> .
@prefix skos:    <http://www.w3.org/2004/02/skos/core#> .
@prefix rdf:     <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
@prefix rdfs:    <http://www.w3.org/2000/01/rdf-schema#> .
@prefix owl:     <http://www.w3.org/2002/07/owl#> .
@prefix xsd:     <http://www.w3.org/2001/XMLSchema#> .
@prefix foaf:    <http://xmlns.com/foaf/0.1/> .
@prefix mb:      <https://mahabharata.tanay.uk/> .
@prefix mbh:     <https://mahabharata.tanay.uk/characters#> .
@prefix mbhx:    <https://mahabharata.tanay.uk/schema#> .
@prefix mbsh:    <https://mahabharata.tanay.uk/shapes#> .

# ---------------------------------------------------------------------------
# 1. SKOS integrity
# ---------------------------------------------------------------------------

mbsh:ConceptShape a sh:NodeShape ;
    sh:targetClass skos:Concept ;
    rdfs:comment "Labels, documentation and scheme membership of every concept."@en ;
    # S14: at most one prefLabel per language tag; this scheme requires English.
    sh:property [
        sh:path skos:prefLabel ;
        sh:datatype rdf:langString ;
        sh:minCount 1 ;
        sh:uniqueLang true ;
        sh:message "Every concept has language-tagged prefLabels, at most one per language (S14)."@en ;
    ] ;
    sh:property [
        sh:path skos:prefLabel ;
        sh:qualifiedValueShape [ sh:languageIn ( "en" ) ] ;
        sh:qualifiedMinCount 1 ;
        sh:message "Every concept has an English prefLabel."@en ;
    ] ;
    sh:property [
        sh:path skos:prefLabel ;
        sh:languageIn ( "en" "sa-Deva" "sa-Latn" ) ;
        sh:message "prefLabel languages are limited to en, sa-Deva and sa-Latn."@en ;
    ] ;
    sh:property [
        sh:path skos:altLabel ;
        sh:datatype rdf:langString ;
        sh:languageIn ( "en" "sa-Deva" "sa-Latn" ) ;
    ] ;
    sh:property [
        sh:path skos:hiddenLabel ;
        sh:datatype rdf:langString ;
        sh:languageIn ( "en" ) ;
        sh:message "hiddenLabel holds English spelling variants only."@en ;
    ] ;
    # S13: prefLabel, altLabel and hiddenLabel are pairwise disjoint.
    sh:property [ sh:path skos:prefLabel ;   sh:disjoint skos:altLabel ;    sh:message "S13: a prefLabel is also an altLabel."@en ] ;
    sh:property [ sh:path skos:prefLabel ;   sh:disjoint skos:hiddenLabel ; sh:message "S13: a prefLabel is also a hiddenLabel."@en ] ;
    sh:property [ sh:path skos:altLabel ;    sh:disjoint skos:hiddenLabel ; sh:message "S13: an altLabel is also a hiddenLabel."@en ] ;
    sh:property [
        sh:path skos:definition ;
        sh:minCount 1 ;
        sh:datatype rdf:langString ;
        sh:qualifiedValueShape [ sh:languageIn ( "en" ) ] ;
        sh:qualifiedMinCount 1 ;
        sh:message "Every concept has an English definition."@en ;
    ] ;
    sh:property [
        sh:path skos:inScheme ;
        sh:hasValue mb:characters ;
        sh:minCount 1 ; sh:maxCount 1 ;
        sh:message "Every concept is in the scheme mb:characters."@en ;
    ] ;
    # Every concept is either a top concept or has a broader concept, never both.
    sh:xone (
        [ sh:property [ sh:path skos:topConceptOf ; sh:hasValue mb:characters ; sh:minCount 1 ] ]
        [ sh:property [ sh:path skos:broader ; sh:minCount 1 ] ]
    ) ;
    sh:property [
        sh:path skos:broader ;
        sh:class skos:Concept ;
        sh:nodeKind sh:IRI ;
        sh:message "skos:broader points at a concept (S19)."@en ;
    ] ;
    # S9 and S37: a concept is neither a scheme nor a collection.
    sh:not [ sh:class skos:ConceptScheme ] ;
    sh:not [ sh:class skos:Collection ] ;
    # S27: skos:related (and its sub-properties) is disjoint with skos:broaderTransitive.
    sh:sparql [
        sh:message "S27: {$this} is related to {?other}, which is also above or below it in the hierarchy."@en ;
        sh:prefixes mbsh: ;
        sh:select """
            SELECT $this ?other WHERE {
                $this skos:related ?other .
                { $this skos:broader+ ?other } UNION { ?other skos:broader+ $this }
            }
        """ ;
    ] ;
    # S46: exactMatch is disjoint with broadMatch and relatedMatch.
    sh:property [ sh:path skos:exactMatch ; sh:disjoint skos:broadMatch ] ;
    sh:property [ sh:path skos:exactMatch ; sh:disjoint skos:relatedMatch ] .

mbsh:CollectionShape a sh:NodeShape ;
    sh:targetClass skos:Collection ;
    sh:property [
        sh:path skos:prefLabel ;
        sh:datatype rdf:langString ;
        sh:minCount 1 ; sh:maxCount 1 ;
        sh:languageIn ( "en" ) ;
    ] ;
    sh:property [ sh:path skos:definition ; sh:minCount 1 ; sh:datatype rdf:langString ] ;
    sh:property [ sh:path skos:inScheme ; sh:hasValue mb:characters ] ;
    sh:property [
        sh:path skos:member ;
        sh:minCount 2 ;
        sh:class mbhx:Character ;
        sh:message "A collection gathers at least two characters; groupings are not collection members."@en ;
    ] ;
    # S37: a collection is not a concept.
    sh:not [ sh:class skos:Concept ] .

mbsh:OrderedCollectionShape a sh:NodeShape ;
    sh:targetClass skos:OrderedCollection ;
    # S36: an ordered collection has exactly one member list, which is an RDF list.
    sh:property [
        sh:path skos:memberList ;
        sh:minCount 1 ; sh:maxCount 1 ;
        sh:nodeKind sh:BlankNodeOrIRI ;
    ] ;
    sh:sparql [
        sh:message "The member list of {$this} is not a well-formed RDF list at {?node}."@en ;
        sh:prefixes mbsh: ;
        sh:select """
            SELECT $this ?node WHERE {
                $this skos:memberList/rdf:rest* ?node .
                FILTER(?node != rdf:nil)
                FILTER( NOT EXISTS { ?node rdf:first ?f } || NOT EXISTS { ?node rdf:rest ?r } )
            }
        """ ;
    ] ;
    sh:sparql [
        sh:message "Member list and skos:member of {$this} disagree."@en ;
        sh:prefixes mbsh: ;
        sh:select """
            SELECT $this ?m WHERE {
                { $this skos:memberList/rdf:rest*/rdf:first ?m . FILTER NOT EXISTS { $this skos:member ?m } }
                UNION
                { $this skos:member ?m . FILTER NOT EXISTS { $this skos:memberList/rdf:rest*/rdf:first ?m } }
            }
        """ ;
    ] .

mbsh:SchemeShape a sh:NodeShape ;
    sh:targetClass skos:ConceptScheme ;
    sh:property [ sh:path skos:hasTopConcept ; sh:minCount 1 ; sh:class mbhx:Grouping ;
                  sh:message "Top concepts are groupings."@en ] ;
    sh:sparql [
        sh:message "hasTopConcept and topConceptOf disagree for {?c}."@en ;
        sh:prefixes mbsh: ;
        sh:select """
            SELECT $this ?c WHERE {
                { $this skos:hasTopConcept ?c . FILTER NOT EXISTS { ?c skos:topConceptOf $this } }
                UNION
                { ?c skos:topConceptOf $this . FILTER NOT EXISTS { $this skos:hasTopConcept ?c } }
            }
        """ ;
    ] .

# ---------------------------------------------------------------------------
# 2. Conventions of this scheme
# ---------------------------------------------------------------------------

mbsh:CharacterShape a sh:NodeShape ;
    sh:targetClass mbhx:Character ;
    sh:not [ sh:class mbhx:Grouping ] ;
    sh:property [
        sh:path skos:broader ;
        sh:minCount 1 ;
        sh:class mbhx:Grouping ;
        sh:message "A character sits under at least one grouping and never under another character."@en ;
    ] ;
    sh:property [
        sh:path skos:topConceptOf ;
        sh:maxCount 0 ;
        sh:message "A character is never a top concept."@en ;
    ] ;
    sh:property [
        sh:path mbhx:fellOnWarDay ;
        sh:datatype xsd:integer ;
        sh:minInclusive 1 ; sh:maxInclusive 18 ;
        sh:maxCount 1 ;
    ] ;
    sh:property [
        sh:path skos:exactMatch ;
        sh:nodeKind sh:IRI ;
        sh:pattern "^http://(www\\.wikidata\\.org/entity/Q[0-9]+|dbpedia\\.org/resource/.+)$" ;
        sh:message "exactMatch targets are Wikidata items or DBpedia resources."@en ;
    ] ;
    sh:property [
        sh:path skos:closeMatch ;
        sh:nodeKind sh:IRI ;
        sh:pattern "^http://(www\\.wikidata\\.org/entity/Q[0-9]+|dbpedia\\.org/resource/.+)$" ;
    ] ;
    sh:property [
        sh:path foaf:isPrimaryTopicOf ;
        sh:nodeKind sh:IRI ;
        sh:pattern "^https://en\\.wikipedia\\.org/wiki/.+$" ;
        sh:maxCount 1 ;
    ] ;
    # Every character has either a mapping or a seeAlso, so that each entry can be checked against a source.
    sh:or (
        [ sh:property [ sh:path skos:exactMatch ; sh:minCount 1 ] ]
        [ sh:property [ sh:path skos:closeMatch ; sh:minCount 1 ] ]
        [ sh:property [ sh:path rdfs:seeAlso ;   sh:minCount 1 ] ]
    ) .

mbsh:GroupingShape a sh:NodeShape ;
    sh:targetClass mbhx:Grouping ;
    sh:not [ sh:class mbhx:Character ] ;
    sh:property [
        sh:path skos:broader ;
        sh:class mbhx:Grouping ;
        sh:message "A grouping is only ever under another grouping."@en ;
    ] ;
    sh:property [
        sh:path [ sh:inversePath skos:broader ] ;
        sh:minCount 1 ;
        sh:message "Every grouping has at least one narrower concept; an empty grouping is a leftover."@en ;
    ] ;
    sh:property [
        sh:path mbhx:characterRelation ;
        sh:maxCount 0 ;
        sh:message "Character relations never attach to a grouping."@en ;
    ] .

# Character relations: both ends are characters (domain and range of mbhx:characterRelation).
mbsh:CharacterRelationSubjectShape a sh:NodeShape ;
    sh:targetSubjectsOf mbhx:characterRelation ;
    sh:class mbhx:Character ;
    sh:property [
        sh:path mbhx:characterRelation ;
        sh:class mbhx:Character ;
        sh:nodeKind sh:IRI ;
        sh:message "The object of a character relation is a character."@en ;
    ] .

# Each relation is stated once in the direction its property names; the
# declared inverses are left to inference.
mbsh:InverseNotMaterialisedShape a sh:NodeShape ;
    sh:targetSubjectsOf mbhx:characterRelation ;
    sh:property [ sh:path mbhx:parentOf ;      sh:maxCount 0 ; sh:message "State childOf on the child, not parentOf."@en ] ;
    sh:property [ sh:path mbhx:fosterParentOf ; sh:maxCount 0 ; sh:message "State fosterChildOf on the child."@en ] ;
    sh:property [ sh:path mbhx:teacherOf ;     sh:maxCount 0 ; sh:message "State studentOf on the student."@en ] ;
    sh:property [ sh:path mbhx:killed ;        sh:maxCount 0 ; sh:message "State killedBy on the one killed."@en ] ;
    sh:property [ sh:path mbhx:hasCharioteer ; sh:maxCount 0 ; sh:message "State charioteerOf on the charioteer."@en ] ;
    sh:property [ sh:path mbhx:rebornAs ;      sh:maxCount 0 ; sh:message "State rebirthOf on the later birth."@en ] ;
    sh:property [ sh:path mbhx:hasAvatar ;     sh:maxCount 0 ; sh:message "State avatarOf on the avatar."@en ] ;
    sh:property [ sh:path mbhx:cursedBy ;      sh:maxCount 0 ; sh:message "State cursed on the one who curses."@en ] .

# Symmetric relations are stated once per pair.
mbsh:SymmetricOnceShape a sh:NodeShape ;
    sh:targetSubjectsOf mbhx:spouseOf, mbhx:siblingOf, mbhx:friendOf, mbhx:rivalOf ;
    sh:sparql [
        sh:message "{$this} and {?o} state {?p} in both directions; state it once."@en ;
        sh:prefixes mbsh: ;
        sh:select """
            SELECT $this ?p ?o WHERE {
                $this ?p ?o . ?o ?p $this .
                FILTER(?p IN (mbhx:spouseOf, mbhx:siblingOf, mbhx:friendOf, mbhx:rivalOf))
            }
        """ ;
    ] .

# Nobody is their own relative, and nobody is both parent and child of the same character.
mbsh:NoSelfOrCycleShape a sh:NodeShape ;
    sh:targetSubjectsOf mbhx:characterRelation ;
    sh:sparql [
        sh:message "{$this} is related to itself by {?p}."@en ;
        sh:prefixes mbsh: ;
        sh:select "SELECT $this ?p WHERE { $this ?p $this . ?p rdfs:subPropertyOf* mbhx:characterRelation . }" ;
    ] ;
    sh:sparql [
        sh:message "{$this} and {?o} are each other's child."@en ;
        sh:prefixes mbsh: ;
        sh:select "SELECT $this ?o WHERE { $this mbhx:childOf ?o . ?o mbhx:childOf+ $this . }" ;
    ] .

# Namespace declarations for the SPARQL constraints above.
mbsh: a owl:Ontology ;
    sh:declare [ sh:prefix "skos" ; sh:namespace "http://www.w3.org/2004/02/skos/core#"^^xsd:anyURI ] ;
    sh:declare [ sh:prefix "rdf"  ; sh:namespace "http://www.w3.org/1999/02/22-rdf-syntax-ns#"^^xsd:anyURI ] ;
    sh:declare [ sh:prefix "rdfs" ; sh:namespace "http://www.w3.org/2000/01/rdf-schema#"^^xsd:anyURI ] ;
    sh:declare [ sh:prefix "mbhx" ; sh:namespace "https://mahabharata.tanay.uk/schema#"^^xsd:anyURI ] .