From 449dc5bd5ad7b16eaee49b3b98aeca40684f965c Mon Sep 17 00:00:00 2001
From: Alex Ioannidis <a.ioannidis@cern.ch>
Date: Thu, 5 Dec 2024 12:35:19 +0100
Subject: [PATCH] names: fix affiliations deduplication

---
 invenio_vocabularies/contrib/names/datastreams.py | 10 ++++------
 tests/contrib/names/test_names_datastreams.py     | 14 ++++++++++++++
 2 files changed, 18 insertions(+), 6 deletions(-)

diff --git a/invenio_vocabularies/contrib/names/datastreams.py b/invenio_vocabularies/contrib/names/datastreams.py
index 007c5cda..2c7c9288 100644
--- a/invenio_vocabularies/contrib/names/datastreams.py
+++ b/invenio_vocabularies/contrib/names/datastreams.py
@@ -260,15 +260,12 @@ 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
@@ -276,7 +273,8 @@ def _extract_affiliations(self, record):
                 if aff_id:
                     aff["id"] = aff_id
 
-                result.append(aff)
+                if aff not in result:
+                    result.append(aff)
         except Exception:
             pass
         return result
diff --git a/tests/contrib/names/test_names_datastreams.py b/tests/contrib/names/test_names_datastreams.py
index 3abd82ee..c9920947 100644
--- a/tests/contrib/names/test_names_datastreams.py
+++ b/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."},
+            ],
+        },
     }