update osfmap_json deriver

aaxelb · Oct 21, 2024 · 645aa3c · 645aa3c
1 parent 7558ec1
commit 645aa3c
Show file tree

Hide file tree

Showing 3 changed files with 24 additions and 16 deletions.
diff --git a/tests/trove/derive/_inputs.py b/tests/trove/derive/_inputs.py
@@ -10,6 +10,7 @@
     OSFMAP,
     FOAF,
     OWL,
+    PROV,
     SHAREv2,
 )
 
@@ -166,6 +167,10 @@ class DeriverTestDoc:
             OSFMAP.contains: {'https://osf.example/2ph9b'},
             OSFMAP.hostingInstitution: {'https://cos.example/'},
             OSFMAP.keyword: {rdf.literal('Demo'), rdf.literal('IA'), rdf.literal('IMLS'), rdf.literal('OSF')},
+            PROV.qualifiedAttribution: {rdf.blanknode({
+                DCAT.hadRole: {OSFMAP['admin-contributor']},
+                PROV.agent: {'https://osf.example/bhcjn'},
+            })},
         },
         'https://osf.example/2ph9b': {
             RDF.type: {OSFMAP.File},

diff --git a/tests/trove/derive/test_osfmap_json.py b/tests/trove/derive/test_osfmap_json.py
@@ -203,10 +203,6 @@ def assert_derived_texts_equal(self, expected, actual):
                     ]
                 }
             ],
-            'prov:qualifiedAttribution': [{
-                'prov:agent': [{'@id': 'https://osf.example/bhcjn'}],
-                'dcat:hadRole': [{'@id': 'osf:admin-contributor'}],
-            }],
             "dateCopyrighted": [
                 {"@value": "2021"}
             ],

diff --git a/trove/derive/osfmap_json.py b/trove/derive/osfmap_json.py
@@ -50,7 +50,7 @@ def rdfobject_as_jsonld(self, rdfobject: rdf.RdfObject) -> dict:
                 rdf.twopledict_from_twopleset(rdfobject),
             )
         elif isinstance(rdfobject, rdf.Literal):
-            if not rdfobject.datatype_iris:
+            if not rdfobject.datatype_iris or rdfobject.datatype_iris == {RDF.string}:
                 return {'@value': rdfobject.unicode_value}
             if RDF.JSON in rdfobject.datatype_iris:
                 # NOTE: does not reset jsonld context (is that a problem?)
@@ -62,13 +62,16 @@ def rdfobject_as_jsonld(self, rdfobject: rdf.RdfObject) -> dict:
                     '@language': _language_tag,
                 }
             # datatype iri (or non-standard language iri)
+            _datatype_iris = sorted(
+                (
+                    osfmap_shorthand().compact_iri(_datatype_iri)
+                    for _datatype_iri in rdfobject.datatype_iris
+                ),
+                key=len,
+            )
             return {
                 '@value': rdfobject.unicode_value,
-                '@type': (
-                    list(rdfobject.datatype_iris)
-                    if len(rdfobject.datatype_iris) > 1
-                    else next(iter(rdfobject.datatype_iris))
-                ),
+                '@type': (_datatype_iris if (len(_datatype_iris) > 1) else _datatype_iris[0]),
             }
         elif isinstance(rdfobject, str):
             return {'@id': osfmap_shorthand().compact_iri(rdfobject)}
@@ -111,7 +114,7 @@ def __nested_rdfobject_as_jsonld(
         _nested_obj = (
             {}
             if rdfobject.startswith('_:')  # HACK: non-blank blank nodes (stop that)
-            else {'@id': rdfobject}
+            else {'@id': osfmap_shorthand().compact_iri(rdfobject)}
         )
         for _pred, _objectset in tripledict[rdfobject].items():
             _label = osfmap_shorthand().compact_iri(_pred)
@@ -126,22 +129,26 @@ def __nested_rdfobject_as_jsonld(
         self.__nestvisiting_iris.discard(rdfobject)
         return _nested_obj
 
-    def _list_or_single_value(self, predicate_iri, objectset):
+    def _list_or_single_value(self, predicate_iri, json_list: list):
         _only_one_object = OWL.FunctionalProperty in (
             OSFMAP_THESAURUS
             .get(predicate_iri, {})
             .get(RDF.type, ())
         )
         if _only_one_object:
-            if len(objectset) > 1:
+            if len(json_list) > 1:
                 raise trove_exceptions.OwlObjection((
                     f'expected at most one object for <{predicate_iri}>'
-                    f' (got {objectset})'
+                    f' (got {json_list})'
                 ))
             try:
-                (_only_obj,) = objectset
+                (_only_obj,) = json_list
             except ValueError:
                 return None
             else:
                 return _only_obj
-        return list(objectset)
+        return (
+            sorted(json_list, key=json.dumps)
+            if len(json_list) > 1
+            else json_list
+        )