Unverified Commit 60134aef authored by Maxime Lefrançois's avatar Maxime Lefrançois
Browse files

check: support vocabularies (TS 103 673 V1.2.1), exit code 1 on errors 

- clause 9.4.3.1: owl:imports may be Vocabulary Version IRIs and Pattern
  Version IRIs; vocabulary documents are checked against their Vocabulary
  IRI and Vocabulary Version IRI (clause 9.8)
- clause 9.4.3.2: check dcterms:title on the document's declaration IRI
- terms only declared in an imported document (e.g., another vocabulary
  of the same project) are no longer counted as declared by the document
- clause 9.6.3: dcterms:conformsTo to standard ontologies outside
  saref.etsi.org no longer produces a warning
- clauses 9.4.5/9.6.4: declare the schema.org metadata terms of clause
  9.4.3.3 before the OWL 2 DL profile check
- saref-dev exits with code 1 when some error is reported
- explicit error when no SAREF project can be deduced from the directory
  name or its git remote origin; use the absolute directory name
parent 708552a3

src/saref_pypeline/__main__.py

+4 −0
Changes for src/saref_pypeline/__main__.py: 4 added lines, 0 removed lines.
Original line number Diff line number Diff line
@@ -184,6 +184,10 @@ def main(): 
    duration = end - start

    print(f"Duration: {duration:.3f} seconds")



    if pipeline.error_count:

        print(f"{pipeline.error_count} error(s) reported.", file=sys.stderr)

        sys.exit(1)





if __name__ == "__main__":

    main()

src/saref_pypeline/check/TS103673_Clause_9_4_3_1_Checker.py

+37 −27
Changes for src/saref_pypeline/check/TS103673_Clause_9_4_3_1_Checker.py: 37 added lines, 27 removed lines.
Original line number Diff line number Diff line 
from saref_pypeline.check.BaseGraphDocumentSHACLChecker import BaseGraphDocumentSHACLChecker

from rdflib import Graph, RDF, OWL

from saref_pypeline.entities import SAREFGraphDocument

from saref_pypeline.entities import SAREFGraphDocument, SAREFGraphDocumentType



from typing import TYPE_CHECKING
@@ -16,52 +16,61 @@ class Checker(BaseGraphDocumentSHACLChecker): 
        super().__init__(__file__, pipeline, document)



    def update_shapes_graph(self, shapes_graph: Graph):

        # clauses 9.8 and 10.3: vocabulary documents and pattern ontological

        # definitions conform to this clause, with the substitutions of the

        # document name, the Ontology IRI, and the Ontology Version IRI

        doc_name = self.document.type.display_name

        version_iri_name = self.document.type.version_iri_name

        shapes_graph += Graph(namespace_manager=shapes_graph.namespace_manager).parse(

            data=f"""

@prefix sh: <http://www.w3.org/ns/shacl#> .

@prefix sarefsh: <https://saref.etsi.org/shape#> .



sarefsh:versionInfo sh:pattern "^{self.version}$" ;

  sh:message "The ontology document shall contain exactly one metadata `owl:versionInfo`, with value `\\"{self.version}\\"`" .

  sh:message "The {doc_name} shall contain exactly one metadata `owl:versionInfo`, with value `\\"{self.version}\\"`" .



sarefsh:versionIRI sh:pattern "^{self.document.version_iri}$" ;

  sh:message "The ontology document shall contain exactly one metadata `owl:versionIRI`, with value `<{self.document.version_iri}>`" .

  sh:message "The {doc_name} shall contain exactly one metadata `owl:versionIRI`, with value `<{self.document.version_iri}>` (the {version_iri_name})" .



sarefsh:preferredNamespacePrefix sh:pattern "^{self.project.prefix}$" ;

  sh:message "The ontology document shall contain exactly one metadata `vann:preferredNamespacePrefix`, with value `\\"{self.project.prefix}\\"`" .

  sh:message "The {doc_name} shall contain exactly one metadata `vann:preferredNamespacePrefix`, with value `\\"{self.project.prefix}\\"`" .



sarefsh:preferredNamespaceUri sh:message "The ontology document shall contain exactly one metadata `vann:preferredNamespaceUri`, with value `\\"{self.project.namespace}\\"`" .

sarefsh:preferredNamespaceUri sh:message "The {doc_name} shall contain exactly one metadata `vann:preferredNamespaceUri`, with value `\\"{self.project.namespace}\\"`" .

""",

            format="turtle",

        )



        prior_version = self.pipeline.get_prior_version(self.project, self.version)

        if prior_version:

            prior_version_iri = str(self.document.version_iri).replace(

                f"{self.project.namespace}{self.version}/",

                f"{self.project.namespace}{prior_version.version}/",

                1,

            )

        if not prior_version:

            shapes_graph += Graph(

                namespace_manager=shapes_graph.namespace_manager

            ).parse(

                data=f"""

@prefix sh: <http://www.w3.org/ns/shacl#> .

@prefix sarefsh: <https://saref.etsi.org/shape#> .



            shape = f"""

sarefsh:priorVersion sh:maxCount 0 ;

    sh:message "There exists no prior version to {self.version}, so the ontology document shall not contain a metadata `owl:priorVersion`" .

""",

                format="turtle",

            )

    sh:message "There exists no prior version to {self.version}, so the {doc_name} shall not contain a metadata `owl:priorVersion`" .

"""

        elif self.document.type == SAREFGraphDocumentType.ONTOLOGY:

            shape = f"""

sarefsh:priorVersion sh:minCount 1 ;

    sh:maxCount 1 ;

    sh:pattern "^{prior_version_iri}$" ;

    sh:message "There exists a prior version {prior_version.version}, so the {doc_name} shall contain exactly one metadata `owl:priorVersion`, with value `<{prior_version_iri}>`" .

"""

        else:

            shapes_graph += Graph(

                namespace_manager=shapes_graph.namespace_manager

            ).parse(

            # the document may not exist in the prior version

            shape = f"""

sarefsh:priorVersion sh:maxCount 1 ;

    sh:pattern "^{prior_version_iri}$" ;

    sh:message "If the {doc_name} contains a metadata `owl:priorVersion`, then it shall contain exactly one, with value `<{prior_version_iri}>`" .

"""

        shapes_graph += Graph(namespace_manager=shapes_graph.namespace_manager).parse(

            data=f"""

@prefix sh: <http://www.w3.org/ns/shacl#> .

@prefix sarefsh: <https://saref.etsi.org/shape#> .



sarefsh:priorVersion sh:minCount 1 ;

    sh:maxCount 1 ;

    sh:pattern "^{self.project.namespace}{prior_version.version}/$" ;

    sh:message "There exists a prior version {prior_version.version}, so the ontology document shall contain exactly one metadata `owl:priorVersion`, with value `<{self.project.namespace}{prior_version.version}/>`" .

""",

{shape}""",

            format="turtle",

        )
@@ -70,9 +79,10 @@ sarefsh:priorVersion sh:minCount 1 ; 
    def check_clause(self):

        graph = self.pipeline.dataset.graph(self.document.version_iri)

        ontologies = list(graph.subjects(RDF.type, OWL.Ontology))

        if len(ontologies) != 1 or str(ontologies[0]) != self.project.namespace:

        declaration_iri = self.document.declaration_iri

        if len(ontologies) != 1 or ontologies[0] != declaration_iri:

            self.logger.error(

                f"The ontology document shall contain exactly one ontology declaration, with value `<{self.project.namespace}>`"

                f"The {self.document.type.display_name} shall contain exactly one ontology declaration, with value `<{declaration_iri}>` (the {self.document.type.iri_name})"

            )

            return

        super().check_clause()

src/saref_pypeline/check/TS103673_Clause_9_4_3_2_Checker.py

+1 −1
Changes for src/saref_pypeline/check/TS103673_Clause_9_4_3_2_Checker.py: 1 added line, 1 removed line.
Original line number Diff line number Diff line
@@ -47,7 +47,7 @@ sarefsh:source sh:message "The ontology document shall contain exactly one metad 
        if not any(

            filter(

                lambda o: isinstance(o, Literal) and o.language == "en",

                graph.objects(self.project.ontology_IRI, DCTERMS.title),

                graph.objects(self.document.declaration_iri, DCTERMS.title),

            )

        ):

            self.logger.warning(

src/saref_pypeline/check/TS103673_Clause_9_4_5_Checker.py

+29 −1
Changes for src/saref_pypeline/check/TS103673_Clause_9_4_5_Checker.py: 29 added lines, 1 removed line.
Original line number Diff line number Diff line
@@ -7,7 +7,7 @@ import tempfile 
from typing import TYPE_CHECKING

from urllib.request import urlretrieve



from rdflib import Graph, URIRef, DCMITYPE, DCTERMS, RDF, OWL

from rdflib import Graph, Namespace, URIRef, DCMITYPE, DCTERMS, RDF, OWL

import owlready2

import owlready2.reasoning
@@ -36,6 +36,33 @@ def java_major_version(java_cmd="java"): 


major = java_major_version(JAVA_CMD)



SCHEMA = Namespace("http://schema.org/")



# terms used for metadata about creators and contributors (TS103673 Clause 9.4.3.3)

METADATA_TERMS = {

    SCHEMA.Person: OWL.Class,

    SCHEMA.Organization: OWL.Class,

    SCHEMA.Affiliation: OWL.Class,

    SCHEMA.givenName: OWL.AnnotationProperty,

    SCHEMA.familyName: OWL.AnnotationProperty,

    SCHEMA.affiliation: OWL.AnnotationProperty,

    SCHEMA.name: OWL.AnnotationProperty,

}





def declare_metadata_terms(graph: Graph):

    """Declare the metadata terms that are used but not declared in the graph."""

    for term, typ in METADATA_TERMS.items():

        if (term, RDF.type, None) in graph:

            continue

        used = (

            (None, term, None) in graph

            if typ == OWL.AnnotationProperty

            else (None, RDF.type, term) in graph

        )

        if used:

            graph.add((term, RDF.type, typ))



class Checker(BaseGraphDocumentChecker):



    def __init__(self, pipeline: "SAREFPipeline", document: SAREFGraphDocument):
@@ -180,6 +207,7 @@ class Checker(BaseGraphDocumentChecker): 
                to_serialize = Graph()

                to_serialize += graph_with_imports

                to_serialize.remove((None, OWL.imports, None))

                declare_metadata_terms(to_serialize)

                to_serialize.serialize(

                    destination=temp_file.name, format="ttl", encoding="utf-8"

                )

src/saref_pypeline/check/resources/shapes/TS103673_Clause_9_4_3_1_Checker.ttl

+2 −2
Changes for src/saref_pypeline/check/resources/shapes/TS103673_Clause_9_4_3_1_Checker.ttl: 2 added lines, 2 removed lines.
Original line number Diff line number Diff line
@@ -39,8 +39,8 @@ sarefsh:priorVersion sh:path owl:priorVersion ; 
sarefsh:imports sh:path owl:imports ;

  sh:severity sh:Violation ;

  sh:nodeKind sh:IRI ;

  sh:pattern "^https://saref.etsi.org/(core|saref4[a-z]{4})/v[0-9]+\\.[0-9]+\\.[0-9]+/$" ;

  sh:message "If the ontology document contains some `owl:imports` declaration, then each value of the declaration shall be an Ontology Version IRI of a SAREF project version" .

  sh:pattern "^https://saref.etsi.org/(core|saref4[a-z]{4})/v[0-9]+\\.[0-9]+\\.[0-9]+/((vocab|patterns?)/[a-zA-Z0-9_-]+/)?$" ;

  sh:message "If the ontology document contains some `owl:imports` declaration, then each value of the declaration shall be an Ontology Version IRI of a SAREF project version, a Vocabulary Version IRI of a SAREF project version, or a Pattern Version IRI of a reference ontology pattern" .



sarefsh:preferredNamespacePrefix sh:path vann:preferredNamespacePrefix ;

  sh:severity sh:Violation ;