fix: skip custom subjects that add nothing

if a custom subject has the exact name of its bepress synonym, do not bother including both in metadata records (they result in wrong-looking duplicates in the search interface and add no value to the metadata)
aaxelb · Aug 24, 2023 · a83d506 · a83d506
1 parent adaaf54
commit a83d506
Show file tree

Hide file tree

Showing 4 changed files with 18 additions and 21 deletions.
diff --git a/osf/metadata/osf_gathering.py b/osf/metadata/osf_gathering.py
@@ -523,21 +523,24 @@ def gather_subjects(focus):
 
 
 def _subject_triples(dbsubject, *, child_ref=None, related_ref=None):
-    _subject_ref = rdflib.URIRef(dbsubject.absolute_api_v2_subject_url)
-    yield (DCTERMS.subject, _subject_ref)
-    yield (_subject_ref, RDF.type, SKOS.Concept)
-    yield (_subject_ref, SKOS.prefLabel, dbsubject.text)
-    yield from _subject_scheme_triples(dbsubject, _subject_ref)
-    if dbsubject.text != dbsubject.bepress_text:
-        yield (_subject_ref, SKOS.altLabel, dbsubject.bepress_text)
-    if child_ref is not None:
-        yield (child_ref, SKOS.broader, _subject_ref)
-    if related_ref is not None:
-        yield (related_ref, SKOS.related, _subject_ref)
-    if dbsubject.bepress_subject and (dbsubject != dbsubject.bepress_subject):
-        yield from _subject_triples(dbsubject.bepress_subject, related_ref=_subject_ref)
-    if dbsubject.parent and (dbsubject != dbsubject.parent):
-        yield from _subject_triples(dbsubject.parent, child_ref=_subject_ref)
+    _is_bepress = (not dbsubject.bepress_subject)
+    _is_distinct_from_bepress = (dbsubject.text != dbsubject.bepress_text)
+    if _is_bepress or _is_distinct_from_bepress:
+        _subject_ref = rdflib.URIRef(dbsubject.absolute_api_v2_subject_url)
+        yield (DCTERMS.subject, _subject_ref)
+        yield (_subject_ref, RDF.type, SKOS.Concept)
+        yield (_subject_ref, SKOS.prefLabel, dbsubject.text)
+        yield from _subject_scheme_triples(dbsubject, _subject_ref)
+        if _is_distinct_from_bepress:
+            yield from _subject_triples(dbsubject.bepress_subject, related_ref=_subject_ref)
+        if child_ref is not None:
+            yield (child_ref, SKOS.broader, _subject_ref)
+        if related_ref is not None:
+            yield (related_ref, SKOS.related, _subject_ref)
+        if dbsubject.parent and (dbsubject != dbsubject.parent):
+            yield from _subject_triples(dbsubject.parent, child_ref=_subject_ref)
+    else:  # if the custom subject adds nothing of value, just include the bepress subject
+        yield from _subject_triples(dbsubject.bepress_subject, child_ref=child_ref, related_ref=related_ref)
 
 
 def _subject_scheme_triples(dbsubject, subject_ref):

diff --git a/osf_tests/metadata/expected_metadata_files/preprint_basic.turtle b/osf_tests/metadata/expected_metadata_files/preprint_basic.turtle
@@ -51,7 +51,6 @@
     owl:sameAs <https://doi.org/10.70102/FK2osf.io/w2ibb> .
 
 <http://localhost:8000/v2/subjects/subjwobb/> a skos:Concept ;
-    skos:altLabel "wobbble" ;
     skos:broader <http://localhost:8000/v2/subjects/subjwibb/> ;
     skos:inScheme <http://localhost:8000/v2/providers/preprints/preprovi/subjects/> ;
     skos:prefLabel "wobble" ;
@@ -77,7 +76,6 @@
     dcterms:title "preprovi" .
 
 <http://localhost:8000/v2/subjects/subjwibb/> a skos:Concept ;
-    skos:altLabel "wibbble" ;
     skos:inScheme <http://localhost:8000/v2/providers/preprints/preprovi/subjects/> ;
     skos:prefLabel "wibble" ;
     skos:related <http://localhost:8000/v2/subjects/subjwibbb/> .

diff --git a/osf_tests/metadata/expected_metadata_files/preprint_full.turtle b/osf_tests/metadata/expected_metadata_files/preprint_full.turtle
@@ -54,7 +54,6 @@
     osf:hasFunding <https://moneypockets.example/millions> .
 
 <http://localhost:8000/v2/subjects/subjwobb/> a skos:Concept ;
-    skos:altLabel "wobbble" ;
     skos:broader <http://localhost:8000/v2/subjects/subjwibb/> ;
     skos:inScheme <http://localhost:8000/v2/providers/preprints/preprovi/subjects/> ;
     skos:prefLabel "wobble" ;
@@ -91,7 +90,6 @@
     dcterms:title "preprovi" .
 
 <http://localhost:8000/v2/subjects/subjwibb/> a skos:Concept ;
-    skos:altLabel "wibbble" ;
     skos:inScheme <http://localhost:8000/v2/providers/preprints/preprovi/subjects/> ;
     skos:prefLabel "wibble" ;
     skos:related <http://localhost:8000/v2/subjects/subjwibbb/> .

diff --git a/osf_tests/metadata/test_osf_gathering.py b/osf_tests/metadata/test_osf_gathering.py
@@ -394,8 +394,6 @@ def test_gather_subjects(self):
             (_child_iri, SKOS.prefLabel, Literal('Child')),
             (_customparent_iri, SKOS.prefLabel, Literal('Custom-parent')),
             (_customchild_iri, SKOS.prefLabel, Literal('Custom-child')),
-            (_customparent_iri, SKOS.altLabel, Literal('Parent')),
-            (_customchild_iri, SKOS.altLabel, Literal('Child')),
             (_child_iri, SKOS.broader, _parent_iri),
             (_customchild_iri, SKOS.broader, _customparent_iri),
             (_customchild_iri, SKOS.related, _child_iri),