get person names from cited_as, if needed

CenterForOpenScience · Feb 23, 2021 · b4e7142 · b4e7142
1 parent 7f99554
commit b4e7142
Show file tree

Hide file tree

Showing 2 changed files with 19 additions and 1 deletion.
diff --git a/share/regulate/steps/normalize_agent_names.py b/share/regulate/steps/normalize_agent_names.py
@@ -28,14 +28,26 @@ def regulate_node(self, node):
 
     def _normalize_person(self, node):
         name = strip_whitespace(node['name'] or '')
+
         if not name:
+            # try building the name from parts
             name = strip_whitespace(' '.join((
                 node['given_name'] or '',
                 node['additional_name'] or '',
                 node['family_name'] or '',
                 node['suffix'] or '',
             )))
 
+        if not name:
+            # try getting the name from "cited_as"
+            cited_as_names = [
+                relation['cited_as']
+                for relation in node['work_relations']
+                if relation['cited_as']
+            ]
+            if len(cited_as_names) == 1:
+                name = cited_as_names[0]
+
         if not name or self.NULL_RE.match(name):
             self.info('Discarding unnamed person', node.id)
             node.delete()

diff --git a/tests/share/normalize/test_models.py b/tests/share/normalize/test_models.py
@@ -397,7 +397,13 @@ def test_normalize_mixed_agent_relation(self, input, output, Graph, ExpectedGrap
         ([
             Creator(cited_as='', agent=Person(id=0, name='None', identifiers=[AgentIdentifier(1, id=1)])),
         ], [
-        ])
+        ]),
+        # Nameless agent with cited_as
+        ([
+            Creator(cited_as='Magpie', agent=Person(id=0, name='', identifiers=[AgentIdentifier(1, id=1)])),
+        ], [
+            Creator(cited_as='Magpie', agent=Person(id=0, name='Magpie', identifiers=[AgentIdentifier(1, id=1)])),
+        ]),
     ])
     def test_normalize_contributor_creator_relation(self, input, output, Graph, ExpectedGraph):
         graph = Graph(CreativeWork(agent_relations=input))