names: fix affiliations deduplication

invenio_vocabularies/contrib/names/datastreams.py

@@ -260,23 +260,21 @@ def _extract_affiliations(self, record):
                 employment.get("employment-summary", {}) for employment in employments
             ]
 
-            history = set()
             for employment in employments:
                 terminated = employment.get("end-date")
-                org = employment["organization"]
-
-                if terminated or org["name"] in history:
+                if terminated:
                     continue
 
-                history.add(org["name"])
+                org = employment["organization"]
                 aff = {"name": org["name"]}
 
                 # Extract the org ID, to link to the affiliation vocabulary
                 aff_id = self._extract_affiliation_id(org)
                 if aff_id:
                     aff["id"] = aff_id
 
-                result.append(aff)
+                if aff not in result:
+                    result.append(aff)
         except Exception:
             pass
         return result

tests/contrib/names/test_names_datastreams.py

@@ -240,6 +240,12 @@ def orcid_data():
                     "employments": {"affiliation-group": employments[0]}
                 },
             },
+            "duplicate_affiliations": {
+                **base,
+                "activities-summary": {
+                    "employments": {"affiliation-group": employments + employments},
+                },
+            },
         },
     }
 
@@ -267,6 +273,14 @@ def expected_from_xml():
             **base,
             "affiliations": [{"id": "01ggx4157", "name": "CERN"}],
         },
+        "duplicate_affiliations": {
+            **base,
+            "affiliations": [
+                # Affiliations are deduplicated
+                {"id": "01ggx4157", "name": "CERN"},
+                {"name": "ACME Inc."},
+            ],
+        },
     }