From c41f697b1807334a1c078d85a9142e962fd74127 Mon Sep 17 00:00:00 2001
From: ErnestaP <ernesta.petraityte@yahoo.com>
Date: Thu, 7 Dec 2023 13:11:44 +0100
Subject: [PATCH] APS: duplicated affids removal

---
 hepcrawl/extractors/aps_parser.py            |  2 +-
 tests/responses/aps/aps_single_response.json |  1 +
 tests/test_aps.py                            | 40 ++++++++++++--------
 3 files changed, 27 insertions(+), 16 deletions(-)

diff --git a/hepcrawl/extractors/aps_parser.py b/hepcrawl/extractors/aps_parser.py
index b90ea3d..bb6cb3a 100644
--- a/hepcrawl/extractors/aps_parser.py
+++ b/hepcrawl/extractors/aps_parser.py
@@ -110,7 +110,7 @@ def _get_authors_and_collab(self, article, dois):
                 author_affiliations = []
                 if 'affiliations' in article and 'affiliationIds' in author:
                     affiliations = build_dict(article['affiliations'], 'id')
-                    for aff_id in author['affiliationIds']:
+                    for aff_id in set(author['affiliationIds']):
                         if aff_id in affiliations:
                             author_affiliations.append({'value': affiliations[aff_id]['name']})
 
diff --git a/tests/responses/aps/aps_single_response.json b/tests/responses/aps/aps_single_response.json
index f32611a..655a1f9 100644
--- a/tests/responses/aps/aps_single_response.json
+++ b/tests/responses/aps/aps_single_response.json
@@ -64,6 +64,7 @@
             {
                "surname":"Alemi",
                "affiliationIds":[
+                  "a1",
                   "a1"
                ],
                "type":"Person",
diff --git a/tests/test_aps.py b/tests/test_aps.py
index 5c6af6c..a210e4f 100644
--- a/tests/test_aps.py
+++ b/tests/test_aps.py
@@ -58,7 +58,8 @@ def test_abstract(results):
 
 def test_title(results):
     """Test extracting title."""
-    titles = ("You can run, you can hide: The epidemiology and statistical mechanics of zombies",)
+    titles = (
+        "You can run, you can hide: The epidemiology and statistical mechanics of zombies",)
     for title, record in zip(titles, results):
         assert 'title' in record
         assert record['title'] == title
@@ -151,20 +152,29 @@ def test_publication_info(results):
 
 def test_authors(results):
     """Test authors."""
-    expected_results = (
-        dict(
-            affiliation='Laboratory of Atomic and Solid State Physics, Cornell University, Ithaca, New York 14853, USA',
-            author_full_names=['Alemi, Alexander A.', 'Bierbaum, Matthew', 'Myers, Christopher R.', 'Sethna, James P.']
-        ),
-    )
-    for expected, record in zip(expected_results, results):
-        assert 'authors' in record
-        assert len(record['authors']) == len(expected['author_full_names'])
-
-        record_full_names = [author['full_name'] for author in record['authors']]
-        assert set(expected['author_full_names']) == set(record_full_names)  # assert that we have the same list of authors
-        for author in record['authors']:
-            assert author['affiliations'][0]['value'] == expected['affiliation']
+    expected_results = [{'affiliations': [{'value': u'Laboratory of Atomic and Solid State Physics, Cornell University, Ithaca, New York 14853, USA'}],
+                         'full_name': u'Alemi, Alexander A.',
+                         'given_names': u'Alexander A.',
+                         'raw_name': u'Alexander A. Alemi',
+                         'surname': u'Alemi'},
+                        {'affiliations': [{'value': u'Laboratory of Atomic and Solid State Physics, Cornell University, Ithaca, New York 14853, USA'}],
+                         'full_name': u'Bierbaum, Matthew',
+                         'given_names': u'Matthew',
+                         'raw_name': u'Matthew Bierbaum',
+                         'surname': u'Bierbaum'},
+                        {'affiliations': [{'value': u'Laboratory of Atomic and Solid State Physics, Cornell University, Ithaca, New York 14853, USA'},
+                                          {'value': u'Institute of Biotechnology, Cornell University, Ithaca, New York 14853, USA'}],
+                         'full_name': u'Myers, Christopher R.',
+                         'given_names': u'Christopher R.',
+                         'raw_name': u'Christopher R. Myers',
+                         'surname': u'Myers'},
+                        {'affiliations': [{'value': u'Laboratory of Atomic and Solid State Physics, Cornell University, Ithaca, New York 14853, USA'}],
+                         'full_name': u'Sethna, James P.',
+                         'given_names': u'James P.',
+                         'raw_name': u'James P. Sethna',
+                         'surname': u'Sethna'}]
+
+    assert results[0]['authors'] == expected_results
 
 
 def test_copyrights(results):