Hotfix: Issues with EMMO Fixes #518 (#519)

* Issues with EMMO Fixes #518

* Fixed some more tests

* minor changes

osp/core/ontology/attribute.py

@@ -6,6 +6,9 @@
 logger = logging.getLogger(__name__)
 
 
+BLACKLIST = {rdflib.OWL.bottomDataProperty, rdflib.OWL.topDataProperty}
+
+
 class OntologyAttribute(OntologyEntity):
     def __init__(self, namespace, name, iri_suffix):
         super().__init__(namespace, name, iri_suffix)
@@ -30,12 +33,14 @@ def datatype(self):
         Returns:
             URIRef: IRI of the datatype
         """
+        blacklist = [rdflib.RDFS.Literal]
         superclasses = self.superclasses
         datatypes = set()
         for superclass in superclasses:
             triple = (superclass.iri, rdflib.RDFS.range, None)
             for _, _, o in self.namespace._graph.triples(triple):
-                datatypes.add(o)
+                if o not in blacklist:
+                    datatypes.add(o)
         if len(datatypes) == 1:
             return datatypes.pop()
         if len(datatypes) == 0:
@@ -64,17 +69,19 @@ def convert_to_basic_type(self, value):
         return convert_from(value, self.datatype)
 
     def _direct_superclasses(self):
-        return self._directly_connected(rdflib.RDFS.subPropertyOf)
+        return self._directly_connected(rdflib.RDFS.subPropertyOf,
+                                        blacklist=BLACKLIST)
 
     def _direct_subclasses(self):
         return self._directly_connected(rdflib.RDFS.subPropertyOf,
-                                        inverse=True)
+                                        inverse=True, blacklist=BLACKLIST)
 
     def _superclasses(self):
         yield self
-        yield from self._transitive_hull(rdflib.RDFS.subPropertyOf)
+        yield from self._transitive_hull(rdflib.RDFS.subPropertyOf,
+                                         blacklist=BLACKLIST)
 
     def _subclasses(self):
         yield self
         yield from self._transitive_hull(rdflib.RDFS.subPropertyOf,
-                                         inverse=True)
+                                         inverse=True, blacklist=BLACKLIST)

osp/core/ontology/docs/cuba.ttl

@@ -59,4 +59,9 @@ cuba:Class a owl:Class ;
     rdfs:isDefinedBy "The root of the ontology." .
 
 
-
+cuba:Class rdfs:subClassOf owl:Thing .
+cuba:relationship rdfs:subPropertyOf owl:topObjectProperty .
+cuba:attribute rdfs:subPropertyOf owl:topDataProperty .
+owl:Thing rdfs:subClassOf cuba:Class .
+owl:topObjectProperty rdfs:subPropertyOf cuba:relationship .
+owl:topDataProperty rdfs:subPropertyOf cuba:attribute .

osp/core/ontology/entity.py

@@ -126,55 +126,91 @@ def _superclasses(self):
     def _subclasses(self):
         pass
 
-    def _transitive_hull(self, predicate_iri, inverse=False):
+    def _transitive_hull(self, predicate_iri, inverse=False, blacklist=()):
         """Get all the entities connected with the given predicate.
 
         Args:
             predicate_iri (URIRef): The IRI of the predicate
             inverse (bool, optional): Use the inverse instead.
                 Defaults to False.
+            blacklist (collection): A collection of IRIs not to return.
 
         Yields:
             OntologyEntity: The connected entities
         """
-        result = {self.iri}
+        visited = {self.iri}
         frontier = {self.iri}
         while frontier:
             current = frontier.pop()
-            triple = (current, predicate_iri, None)
-            if inverse:
-                triple = (None, predicate_iri, current)
-            for x in self.namespace._graph.triples(triple):
-                o = x[0 if inverse else 2]
-                if o not in result and not isinstance(o, rdflib.BNode) \
-                    and not str(o).startswith((str(rdflib.RDF),
-                                               str(rdflib.RDFS),
-                                               str(rdflib.OWL))):
-                    frontier.add(o)
-                    result.add(o)
-                    yield self.namespace._namespace_registry.from_iri(o)
+            yield from self._directly_connected(predicate_iri=predicate_iri,
+                                                inverse=inverse,
+                                                blacklist=blacklist,
+                                                _frontier=frontier,
+                                                _visited=visited,
+                                                _iri=current)
+
+    def _special_cases(self, triple):
+        """Some supclass statements are often omitted in the ontology.
+        Replace these with safer triple patterns.
+
+        Args:
+            triple (Tuple[rdflib.term]): A triple pattern to possibly replace.
 
-    def _directly_connected(self, predicate_iri, inverse=False):
+        Returns:
+            triple (Tuple[rdflib.term]): Possibly replaced triple.
+        """
+        if triple == (None, rdflib.RDFS.subClassOf, rdflib.OWL.Thing):
+            return (None, rdflib.RDF.type, rdflib.OWL.Class)
+        if triple == (rdflib.OWL.Nothing, rdflib.RDFS.subClassOf, None):
+            return (None, rdflib.RDF.type, rdflib.OWL.Class)
+
+        if triple == (None, rdflib.RDFS.subPropertyOf,
+                      rdflib.OWL.topObjectProperty):
+            return (None, rdflib.RDF.type, rdflib.OWL.ObjectProperty)
+        if triple == (rdflib.OWL.bottomObjectProperty,
+                      rdflib.RDFS.subPropertyOf, None):
+            return (None, rdflib.RDF.type, rdflib.OWL.ObjectProperty)
+
+        if triple == (None, rdflib.RDFS.subPropertyOf,
+                      rdflib.OWL.topDataProperty):
+            return (None, rdflib.RDF.type, rdflib.OWL.DataProperty)
+        if triple == (rdflib.OWL.bottomDataProperty,
+                      rdflib.RDFS.subPropertyOf, None):
+            return (None, rdflib.RDF.type, rdflib.OWL.DataProperty)
+        return triple
+
+    def _directly_connected(self, predicate_iri, inverse=False, blacklist=(),
+                            _frontier=None, _visited=None, _iri=None):
         """Get all the entities directly connected with the given predicate.
 
         Args:
             predicate_iri (URIRef): The IRI of the predicate
             inverse (bool, optional): Use the inverse instead.
                 Defaults to False.
+            blacklist (collection): A collection of IRIs not to return.
+            Others: Helper for _transitive_hull method.
 
         Yields:
             OntologyEntity: The connected entities
         """
-        triple = (self.iri, predicate_iri, None)
+        triple = (_iri or self.iri, predicate_iri, None)
         if inverse:
-            triple = (None, predicate_iri, self.iri)
+            triple = (None, predicate_iri, _iri or self.iri)
+
+        if predicate_iri in [rdflib.RDFS.subClassOf,
+                             rdflib.RDFS.subPropertyOf]:
+            triple = self._special_cases(triple)
         for x in self.namespace._graph.triples(triple):
-            o = x[0 if inverse else 2]
-            if not isinstance(o, rdflib.BNode) \
-                and not str(o).startswith((str(rdflib.RDF),
-                                           str(rdflib.RDFS),
-                                           str(rdflib.OWL))):
-                yield self.namespace._namespace_registry.from_iri(o)
+            o = x[0 if triple[0] is None else 2]
+            if _visited and o in _visited:
+                continue
+            if not isinstance(o, rdflib.BNode):
+                if _visited is not None:
+                    _visited.add(o)
+                if _frontier is not None:
+                    _frontier.add(o)
+                if o not in blacklist:
+                    yield self.namespace._namespace_registry.from_iri(o)
 
     def __hash__(self):
         return hash(self.iri)

osp/core/ontology/namespace_registry.py

@@ -160,8 +160,9 @@ def load(self, path):
 
     def _load_cuba(self):
         """Load the cuba namespace"""
-        path_cuba = os.path.join(os.path.dirname(__file__), "docs", "cuba.ttl")
-        self._graph.parse(path_cuba, format="ttl")
+        for x in ["cuba"]:  #, "rdf", "rdfs", "owl"]:
+            path = os.path.join(os.path.dirname(__file__), "docs", x + ".ttl")
+            self._graph.parse(path, format="ttl")
         self._graph.bind("cuba",
                          rdflib.URIRef("http://www.osp-core.com/cuba#"))
         self.update_namespaces()

osp/core/ontology/oclass.py

@@ -5,6 +5,9 @@
 
 logger = logging.getLogger(__name__)
 
+BLACKLIST = {rdflib.OWL.Nothing, rdflib.OWL.Thing,
+             rdflib.OWL.NamedIndividual}
+
 
 class OntologyClass(OntologyEntity):
     def __init__(self, namespace, name, iri_suffix):
@@ -20,7 +23,10 @@ def attributes(self):
         """
         attributes = dict()
         for superclass in self.superclasses:
-            attributes.update(self._get_attributes(superclass.iri))
+            for attr, v in self._get_attributes(superclass.iri).items():
+                x = attributes.get(attr, (None, None, None))
+                x = (x[0] or v[0], x[1] or v[1], x[2] or v[2])
+                attributes[attr] = x
         return attributes
 
     @property
@@ -43,28 +49,51 @@ def _get_attributes(self, iri):
         """
         graph = self._namespace._graph
         attributes = dict()
+
+        blacklist = [rdflib.OWL.topDataProperty, rdflib.OWL.bottomDataProperty]
         # Case 1: domain of Datatype
         triple = (None, rdflib.RDFS.domain, iri)
         for a_iri, _, _ in self.namespace._graph.triples(triple):
             triple = (a_iri, rdflib.RDF.type, rdflib.OWL.DatatypeProperty)
-            if triple in graph \
-                    and not isinstance(a_iri, rdflib.BNode):
-                a = self.namespace._namespace_registry.from_iri(a_iri)
-                attributes[a] = self._get_default(a_iri, iri)
+            if triple not in graph or isinstance(a_iri, rdflib.BNode) \
+                    or a_iri in blacklist:
+                continue
+            a = self.namespace._namespace_registry.from_iri(a_iri)
+            default = self._get_default(a_iri, iri)
+            triple = (a_iri, rdflib.RDF.type, rdflib.OWL.FunctionalProperty)
+            obligatory = default is None and triple in graph
+            attributes[a] = (self._get_default(a_iri, iri), obligatory, None)
 
         # Case 2: restrictions
         triple = (iri, rdflib.RDFS.subClassOf, None)
         for _, _, o in self.namespace._graph.triples(triple):
-            if (o, rdflib.RDF.type, rdflib.OWL.Restriction) in graph:
-                a_iri = graph.value(o, rdflib.OWL.onProperty)
-                triple = (a_iri, rdflib.RDF.type, rdflib.OWL.DatatypeProperty)
-                if triple in graph \
-                        and not isinstance(a_iri, rdflib.BNode):
-                    a = self.namespace._namespace_registry.from_iri(a_iri)
-                    attributes[a] = self._get_default(a_iri, iri)
+            if (o, rdflib.RDF.type, rdflib.OWL.Restriction) not in graph:
+                continue
+            a_iri = graph.value(o, rdflib.OWL.onProperty)
+            triple = (a_iri, rdflib.RDF.type, rdflib.OWL.DatatypeProperty)
+            if triple not in graph or isinstance(a_iri, rdflib.BNode):
+                continue
+            a = self.namespace._namespace_registry.from_iri(a_iri)
+            default = self._get_default(a_iri, iri)
+            dt, obligatory = self._get_datatype_for_restriction(o)
+            obligatory = default is None and obligatory
+            attributes[a] = (self._get_default(a_iri, iri), obligatory, dt)
+
         # TODO more cases
         return attributes
 
+    def _get_datatype_for_restriction(self, r):
+        obligatory = False
+        dt = None
+        g = self.namespace._graph
+
+        dt = g.value(r, rdflib.OWL.someValuesFrom)
+        obligatory = dt is not None
+        dt = dt or g.value(r, rdflib.OWL.allValuesFrom)
+        obligatory = obligatory or (r, rdflib.OWL.cardinality) != 0
+        obligatory = obligatory or (r, rdflib.OWL.minCardinality) != 0
+        return dt, obligatory
+
     def _get_default(self, attribute_iri, superclass_iri):
         """Get the default of the attribute with the given iri.
 
@@ -96,7 +125,7 @@ def _get_attributes_values(self, kwargs, _force):
         """
         kwargs = dict(kwargs)
         attributes = dict()
-        for attribute, default in self.attributes.items():
+        for attribute, (default, obligatory, dt) in self.attributes.items():
             if attribute.argname in kwargs:
                 attributes[attribute] = kwargs[attribute.argname]
                 del kwargs[attribute.argname]
@@ -112,32 +141,37 @@ def _get_attributes_values(self, kwargs, _force):
                     f"to be ALL_CAPS. You can use the yaml2camelcase "
                     f"commandline tool to transform entity names to CamelCase."
                 )
-            else:
+            elif not _force and obligatory:
+                raise TypeError("Missing keyword argument: %s" % attribute)
+            elif default is not None:
                 attributes[attribute] = default
 
         # Check validity of arguments
         if not _force:
             if kwargs:
                 raise TypeError("Unexpected keyword arguments: %s"
                                 % kwargs.keys())
-            missing = [k.argname for k, v in attributes.items() if v is None]
-            if missing:
-                raise TypeError("Missing keyword arguments: %s" % missing)
         return attributes
 
     def _direct_superclasses(self):
-        return self._directly_connected(rdflib.RDFS.subClassOf)
+        return self._directly_connected(rdflib.RDFS.subClassOf,
+                                        blacklist=BLACKLIST)
 
     def _direct_subclasses(self):
-        return self._directly_connected(rdflib.RDFS.subClassOf, inverse=True)
+        return self._directly_connected(rdflib.RDFS.subClassOf,
+                                        inverse=True, blacklist=BLACKLIST)
 
     def _superclasses(self):
         yield self
-        yield from self._transitive_hull(rdflib.RDFS.subClassOf)
+        yield from self._transitive_hull(
+            rdflib.RDFS.subClassOf,
+            blacklist=BLACKLIST)
 
     def _subclasses(self):
         yield self
-        yield from self._transitive_hull(rdflib.RDFS.subClassOf, inverse=True)
+        yield from self._transitive_hull(
+            rdflib.RDFS.subClassOf, inverse=True,
+            blacklist=BLACKLIST)
 
     def __call__(self, uid=None, session=None, _force=False, **kwargs):
         """Create a Cuds object from this ontology class.

osp/core/ontology/parser.py

@@ -247,18 +247,20 @@ def _parse_rdf(self, **kwargs):
         self.graph.parse(rdf_file, format=file_format)
         default_rels = dict()
         reference_styles = dict()
+        namespace_iris = set()
         for namespace, iri in namespaces.items():
             if not (
                 iri.endswith("#") or iri.endswith("/")
             ):
                 iri += "#"
+            namespace_iris.add(iri)
             logger.info(f"You can now use `from osp.core.namespaces import "
                         f"{namespace}`.")
             self.graph.bind(namespace, rdflib.URIRef(iri))
             default_rels[iri] = default_rel
             reference_styles[iri] = reference_style
 
-        self._check_namespaces()
+        self._check_namespaces(namespace_iris)
         self._add_cuba_triples(active_rels)
         self._add_default_rel_triples(default_rels)
         self._add_reference_style_triples(reference_styles)
@@ -313,9 +315,8 @@ def _add_reference_style_triples(self, reference_styles):
                     rdflib.Literal(True)
                 ))
 
-    def _check_namespaces(self):
-        namespaces = set(x for _, x in self.graph.namespaces()
-                         if not x.startswith("http://www.w3.org/"))
+    def _check_namespaces(self, namespace_iris):
+        namespaces = set(namespace_iris)
         for s, p, o in self.graph:
             pop = None
             for ns in namespaces: