fix subject deduping bug

aaxelb · Mar 10, 2021 · dba4d51 · dba4d51
1 parent f470c98
commit dba4d51
Show file tree

Hide file tree

Showing 4 changed files with 25 additions and 3 deletions.
diff --git a/share/disambiguation/strategies/graph.py b/share/disambiguation/strategies/graph.py
@@ -60,6 +60,7 @@ def match_subjects(self, nodes):
             matches = [
                 n for n in graph_nodes
                 if n != node
+                and n['parent'] == node['parent']
                 and n['central_synonym'] == node['central_synonym']
                 and (equal_not_none(n['uri'], node['uri']) or equal_not_none(n['name'], node['name']))
             ]

diff --git a/share/schema/schema-spec.yaml b/share/schema/schema-spec.yaml
@@ -261,6 +261,7 @@
     - name: central_synonym
       relation_shape: many_to_one
       related_concrete_type: Subject
+      inverse_relation: custom_synonyms
 
 - concrete_type: ThroughSubjects
   attributes:

diff --git a/share/util/graph.py b/share/util/graph.py
@@ -560,13 +560,13 @@ def delete(self, cascade=True):
         self.graph.remove_node(self.id, cascade)
         self.__graph = None
 
-    def to_jsonld(self, ref=False, in_edges=True):
+    def to_jsonld(self, ref=False, in_edges=False):
         ld_node = {
             '@id': self.id,
             '@type': self.type,
         }
         if not ref:
-            ld_node.update(self.relations(in_edges=False, jsonld=True))
+            ld_node.update(self.relations(in_edges=in_edges, jsonld=True))
             ld_node.update(self.attrs())
         return ld_node
 

diff --git a/tests/share/regulate/steps/test_deduplicate.py b/tests/share/regulate/steps/test_deduplicate.py
@@ -32,7 +32,27 @@ def test_no_change(self, Graph, input):
                 WorkIdentifier(4),
                 WorkIdentifier(uri='http://osf.io/guidguid'),
             ]),
-        ])
+        ]),
+        ([
+            Registration(0, subjects=[
+                Subject(
+                    0,
+                    name='custom-child',
+                    central_synonym=Subject(1, name='central-child', parent=Subject(3, name='central-parent')),
+                    parent=Subject(2, name='custom-parent', central_synonym=Subject(3, name='central-parent')),
+                )
+                for _ in range(3)
+            ]),
+        ], [
+            Registration(0, subjects=[
+                Subject(
+                    0,
+                    name='custom-child',
+                    central_synonym=Subject(1, name='central-child', parent=Subject(3, id='central-parent', name='central-parent')),
+                    parent=Subject(2, name='custom-parent', central_synonym=Subject(id='central-parent')),
+                )
+            ]),
+        ]),
     ])
     def test_deduplicate(self, Graph, input, output):
         graph = Graph(input)