Changes for new SPDX 3 Model

The new SPDX 3 model has exposed some problems with the way the ontology
is parsed.

src/shacl2code/lang/templates/jsonschema.j2

@@ -147,6 +147,14 @@
                     "$ref": "#/$defs/SemVer"
                 {%- elif prop.datatype == "https://spdx.org/rdf/v3/Core/Extension" %}
                     "$ref": "#/$defs/Extension"
+                {%- elif prop.datatype == "https://rdf.spdx.org/v3/Core/DateTime" %}
+                    "$ref": "#/$defs/DateTime"
+                {%- elif prop.datatype == "https://rdf.spdx.org/v3/Core/MediaType" %}
+                    "$ref": "#/$defs/MediaType"
+                {%- elif prop.datatype == "https://rdf.spdx.org/v3/Core/SemVer" %}
+                    "$ref": "#/$defs/SemVer"
+                {%- elif prop.datatype == "https://rdf.spdx.org/v3/Core/Extension" %}
+                    "$ref": "#/$defs/Extension"
                 {%- elif prop.datatype == "http://www.w3.org/2001/XMLSchema#positiveInteger" %}
                     "type": "integer",
                     "minimum": 0

src/shacl2code/lang/templates/python.j2

@@ -803,6 +803,10 @@ DATATYPE_CLASSES = {
     "https://spdx.org/rdf/v3/Core/MediaType": "MediaTypeProp",
     "https://spdx.org/rdf/v3/Core/SemVer": "SemVerProp",
     "https://spdx.org/rdf/v3/Core/Extension": "ExtensionProp",
+    "https://rdf.spdx.org/v3/Core/DateTime": "DateTimeProp",
+    "https://rdf.spdx.org/v3/Core/MediaType": "MediaTypeProp",
+    "https://rdf.spdx.org/v3/Core/SemVer": "SemVerProp",
+    "https://rdf.spdx.org/v3/Core/Extension": "ExtensionProp",
     "http://www.w3.org/2001/XMLSchema#positiveInteger": "PositiveIntegerProp",
     "http://www.w3.org/2001/XMLSchema#nonNegativeInteger": "NonNegativeIntegerProp",
     "http://www.w3.org/2001/XMLSchema#boolean": "BooleanProp",

src/shacl2code/model.py

@@ -30,6 +30,35 @@ def to_var_name(name):
     return name
 
 
+def is_type(obj, *typ):
+    if "@type" not in obj:
+        return False
+
+    if isinstance(obj["@type"], list):
+        return all(t in obj["@type"] for t in typ)
+
+    return all(obj["@type"] == t for t in typ)
+
+
+def common_prefix(*s):
+    if not s:
+        return ""
+
+    if len(s) == 1:
+        return s[0]
+
+    p1 = common_prefix(*s[: len(s) // 2])
+    p2 = common_prefix(*s[len(s) // 2 :])
+    for idx in range(len(p1)):
+        if idx >= len(p2):
+            return p2
+
+        if p1[idx] != p2[idx]:
+            return p2[:idx]
+
+    return p1
+
+
 @dataclass
 class EnumValue:
     _id: str
@@ -70,50 +99,64 @@ class Model(object):
     def __init__(self, model_data, context=None):
         self.model = jsonld.expand(model_data)
         self.context = context
-        self.compact = {}
+        self.compact_ids = {}
         self.objects = {}
         self.enums = []
         self.classes = []
         classes = []
-        enums = []
+        model_context = {}
 
-        context = model_data.get("@context", {})
+        if isinstance(model_data, dict) and "@context" in model_data:
+            model_context = model_data["@context"]
 
         for obj in self.model:
-            self.compact[obj["@id"]] = jsonld.compact(obj, context)
-            del self.compact[obj["@id"]]["@context"]
+            if model_context:
+                self.compact_ids[obj["@id"]] = jsonld.compact(obj, model_context)["@id"]
 
             self.objects[obj["@id"]] = obj
 
-            if "http://www.w3.org/2002/07/owl#oneOf" in obj:
-                enums.append(obj)
-            elif "@type" in obj:
-                if "http://www.w3.org/2002/07/owl#Class" in obj["@type"]:
-                    classes.append(obj)
-
-        self.class_ids = set(c["@id"] for c in classes)
-        self.enum_ids = set(e["@id"] for e in enums)
+            if is_type(obj, "http://www.w3.org/2002/07/owl#Class"):
+                classes.append(obj)
+
+        for c in classes:
+            enum_values = []
+
+            for _id, obj in self.objects.items():
+                if not is_type(
+                    obj,
+                    "http://www.w3.org/2002/07/owl#NamedIndividual",
+                    c["@id"],
+                ):
+                    continue
+
+                v = EnumValue(
+                    _id=_id,
+                    varname=get_prop(
+                        obj,
+                        "http://www.w3.org/2000/01/rdf-schema#label",
+                        "@value",
+                        to_var_name(_id.split("/")[-1]),
+                    ),
+                    comment=self.get_comment(obj),
+                )
+                enum_values.append(v)
 
-        for obj in enums:
-            e = Enum(
-                _id=obj["@id"],
-                clsname=self.get_class_name(obj),
-                comment=self.get_comment(obj),
-                values=[],
-            )
+            if enum_values:
+                enum_values.sort(key=lambda v: v._id)
 
-            for v in get_prop(obj, "http://www.w3.org/2002/07/owl#oneOf", "@list"):
-                e.values.append(
-                    EnumValue(
-                        _id=v["@id"],
-                        varname=to_var_name(v["@id"].split("/")[-1]),
-                        comment=self.get_comment(self.objects[v["@id"]]),
-                    )
+                e = Enum(
+                    _id=c["@id"],
+                    clsname=self.get_class_name(c),
+                    comment=self.get_comment(c),
+                    values=enum_values,
                 )
+                self.enums.append(e)
 
-            e.values.sort(key=lambda v: v._id)
+        self.enum_ids = set(e._id for e in self.enums)
 
-            self.enums.append(e)
+        classes = [c for c in classes if not self.is_enum(c["@id"])]
+
+        self.class_ids = set(c["@id"] for c in classes)
 
         for obj in classes:
             c = Class(
@@ -131,47 +174,69 @@ def __init__(self, model_data, context=None):
             )
 
             for prop_id in obj.get("http://www.w3.org/ns/shacl#property", []):
-                prop = self.objects[prop_id["@id"]]
-                name = prop["http://www.w3.org/ns/shacl#name"][0]["@value"]
-                prop_path = get_prop(
-                    prop,
-                    "http://www.w3.org/ns/shacl#path",
-                    "@id",
-                )
+                class_prop = self.objects[prop_id["@id"]]
+                prop = self.objects[
+                    get_prop(
+                        class_prop,
+                        "http://www.w3.org/ns/shacl#path",
+                        "@id",
+                    )
+                ]
+                name = get_prop(class_prop, "http://www.w3.org/ns/shacl#name", "@value")
+                if name is None:
+                    prefix = common_prefix(prop["@id"], obj["@id"])
+                    name = prop["@id"][len(prefix) :]
 
                 p = Property(
                     varname=to_var_name(name),
-                    path=prop_path,
-                    comment=self.get_comment(self.objects[prop_path]),
+                    path=prop["@id"],
+                    comment=self.get_comment(prop),
                     max_count=get_prop(
-                        prop,
+                        class_prop,
                         "http://www.w3.org/ns/shacl#maxCount",
                         "@value",
                         None,
                     ),
                     min_count=get_prop(
-                        prop,
+                        class_prop,
                         "http://www.w3.org/ns/shacl#minCount",
                         "@value",
                         None,
                     ),
                 )
 
-                prop_cls_id = get_prop(prop, "http://www.w3.org/ns/shacl#class", "@id")
-                if prop_cls_id:
-                    if self.is_enum(prop_cls_id):
-                        p.enum_id = prop_cls_id
+                range_id = get_prop(
+                    prop,
+                    "http://www.w3.org/2000/01/rdf-schema#range",
+                    "@id",
+                )
+                if range_id is not None:
+                    if self.is_enum(range_id):
+                        p.enum_id = range_id
+
+                    elif self.is_class(range_id):
+                        p.class_id = range_id
 
-                    elif self.is_class(prop_cls_id):
-                        p.class_id = prop_cls_id
                     else:
-                        raise ModelException(f"Unknown type '{prop_cls_id}'")
+                        p.datatype = range_id
                 else:
-                    p.datatype = get_prop(
-                        prop,
-                        "http://www.w3.org/ns/shacl#datatype",
-                        "@id",
+                    prop_cls_id = get_prop(
+                        class_prop, "http://www.w3.org/ns/shacl#class", "@id"
                     )
+                    if prop_cls_id:
+                        if self.is_enum(prop_cls_id):
+                            p.enum_id = prop_cls_id
+
+                        elif self.is_class(prop_cls_id):
+                            p.class_id = prop_cls_id
+                        else:
+                            raise ModelException(f"Unknown type '{prop_cls_id}'")
+                    else:
+                        p.datatype = get_prop(
+                            class_prop,
+                            "http://www.w3.org/ns/shacl#datatype",
+                            "@id",
+                        )
 
                 c.properties.append(p)
 
@@ -203,21 +268,21 @@ def is_enum(self, _id):
     def is_class(self, _id):
         return _id in self.class_ids
 
-    def get_compact(self, obj, *path):
+    def get_compact_id(self, _id):
         """
         Returns the "compacted" name of an object, that is the name of the
         object with the context applied
         """
-        v = self.compact[obj["@id"]]
-        for p in path:
-            v = v[p]
-        return v
+        if _id not in self.compact_ids:
+            self.compact_ids[_id] = self.context.compact(_id)
+
+        return self.compact_ids[_id]
 
     def get_class_name(self, c):
         """
         Returns the name for a class that should be used in Code
         """
-        return self.get_compact(c, "@id").replace(":", "_")
+        return to_var_name(self.get_compact_id(c["@id"]).replace(":", "_"))
 
     def get_comment(self, obj, indent=0):
         """

tests/data/bad-reference.jsonld

@@ -44,9 +44,6 @@
       "rdfs:comment": "CreationInfo provides information about the creation of the Element.",
       "rdfs:domain": {
         "@id": "core:Element"
-      },
-      "rdfs:range": {
-        "@id": "core:CreationInfo"
       }
     }
   ]

tests/expect/jsonschema/spdx3-context.json

@@ -2152,7 +2152,7 @@
                 "oneOf": [
                     { "$ref": "#/$defs/idRef" },
                     {
-                        "$ref": "#/$defs/expandedlicensing_ExtendableLicense",
+                        "$ref": "#/$defs/expandedlicensing_License",
                         "unevaluatedProperties": false
                     }
                 ]

tests/expect/jsonschema/spdx3.json

@@ -2148,7 +2148,7 @@
                 "oneOf": [
                     { "$ref": "#/$defs/idRef" },
                     {
-                        "$ref": "#/$defs/expandedlicensing_ExtendableLicense",
+                        "$ref": "#/$defs/expandedlicensing_License",
                         "unevaluatedProperties": false
                     }
                 ]

tests/expect/python/spdx3-context.py

@@ -3689,7 +3689,7 @@ def __init__(self, **kwargs):
         # (OrLaterOperator) or a 'with additional text' effect (WithAdditionOperator).
         self._add_property(
             "subjectLicense",
-            ObjectProp(expandedlicensing_ExtendableLicense, True),
+            ObjectProp(expandedlicensing_License, True),
             json_name="https://spdx.org/rdf/v3/ExpandedLicensing/subjectLicense",
             min_count=1,
             vocab="https://spdx.org/rdf/v3/ExpandedLicensing/subjectLicense",

tests/expect/python/spdx3.py

@@ -2936,7 +2936,7 @@ def __init__(self, **kwargs):
         # (OrLaterOperator) or a 'with additional text' effect (WithAdditionOperator).
         self._add_property(
             "subjectLicense",
-            ObjectProp(expandedlicensing_ExtendableLicense, True),
+            ObjectProp(expandedlicensing_License, True),
             json_name="https://spdx.org/rdf/v3/ExpandedLicensing/subjectLicense",
             min_count=1,
             vocab="https://spdx.org/rdf/v3/ExpandedLicensing/subjectLicense",