sharev2_elastic format: handle "deleted" subjects

when OSF pushes metadata containing subjects, it will often include "deleted" subjects that should be removed -- because SHARE historically didn't understand what it meant to remove relations, and we added a special-cased hack to allow removing subjects. this is (will be) no longer necessary using FormattedMetadataRecords, since older metadata will be replaced by , not merged with, new updates
CenterForOpenScience · Feb 4, 2021 · fec9537 · fec9537
1 parent 4b2f25c
commit fec9537
Show file tree

Hide file tree

Showing 2 changed files with 121 additions and 5 deletions.
diff --git a/share/metadata_formats/sharev2_elastic.py b/share/metadata_formats/sharev2_elastic.py
@@ -141,15 +141,23 @@ def _get_related_agent_names(self, work_node, relation_types):
 
     def _get_subjects(self, work_node, source_name):
         return [
-            self._serialize_subject(subject_node, source_name)
-            for subject_node in work_node['subjects']
+            self._serialize_subject(through_subject['subject'], source_name)
+            for through_subject in work_node['subject_relations']
+            if (
+                not through_subject['is_deleted']
+                and not through_subject['subject']['is_deleted']
+            )
         ]
 
     def _get_subject_synonyms(self, work_node):
         return [
-            self._serialize_subject(subject_node['central_synonym'])
-            for subject_node in work_node['subjects']
-            if subject_node['central_synonym']
+            self._serialize_subject(through_subject['subject']['central_synonym'])
+            for through_subject in work_node['subject_relations']
+            if (
+                not through_subject['is_deleted']
+                and not through_subject['subject']['is_deleted']
+                and through_subject['subject']['central_synonym']
+            )
         ]
 
     def _serialize_subject(self, subject_node, source_name=None):

diff --git a/tests/share/metadata_formats/test_sharev2_elastic.py b/tests/share/metadata_formats/test_sharev2_elastic.py
@@ -242,6 +242,107 @@
                             '@type': 'registration'
                         }
                     },
+                    {
+                        '@id': '_:through-subj-architecture',
+                        '@type': 'throughsubjects',
+                        'creative_work': {
+                            '@id': '_:4058232c-106f-4a2f-8700-d8c14a6c6ece',
+                            '@type': 'registration'
+                        },
+                        'subject': {
+                            '@id': '_:subj-architecture',
+                            '@type': 'subject'
+                        },
+                    },
+                    {
+                        '@id': '_:subj-architecture',
+                        '@type': 'subject',
+                        'name': 'Architecture',
+                    },
+                    {
+                        '@id': '_:through-subj-business',
+                        '@type': 'throughsubjects',
+                        'is_deleted': True,  # back-compat with a prior hack
+                        'creative_work': {
+                            '@id': '_:4058232c-106f-4a2f-8700-d8c14a6c6ece',
+                            '@type': 'registration'
+                        },
+                        'subject': {
+                            '@id': '_:subj-business',
+                            '@type': 'subject'
+                        },
+                    },
+                    {
+                        '@id': '_:subj-business',
+                        '@type': 'subject',
+                        'name': 'Business',
+                    },
+                    {
+                        '@id': '_:through-subj-education',
+                        '@type': 'throughsubjects',
+                        'creative_work': {
+                            '@id': '_:4058232c-106f-4a2f-8700-d8c14a6c6ece',
+                            '@type': 'registration'
+                        },
+                        'subject': {
+                            '@id': '_:subj-education',
+                            '@type': 'subject'
+                        },
+                    },
+                    {
+                        '@id': '_:subj-education',
+                        '@type': 'subject',
+                        'name': 'Education',
+                        'is_deleted': True,  # back-compat with a prior hack
+                    },
+                    {
+                        '@id': '_:through-subj-custom-biology',
+                        '@type': 'throughsubjects',
+                        'creative_work': {
+                            '@id': '_:4058232c-106f-4a2f-8700-d8c14a6c6ece',
+                            '@type': 'registration'
+                        },
+                        'subject': {
+                            '@id': '_:subj-custom-biology',
+                            '@type': 'subject'
+                        },
+                    },
+                    {
+                        '@id': '_:subj-custom-biology',
+                        '@type': 'subject',
+                        'name': 'Custom biologyyyy',
+                        'parent': {
+                            '@id': '_:subj-custom-life-sciences',
+                            '@type': 'subject',
+                        },
+                        'central_synonym': {
+                            '@id': '_:subj-central-biology',
+                            '@type': 'subject',
+                        },
+                    },
+                    {
+                        '@id': '_:subj-custom-life-sciences',
+                        '@type': 'subject',
+                        'name': 'Custom life sciencesssss',
+                        'central_synonym': {
+                            '@id': '_:subj-central-life-sciences',
+                            '@type': 'subject',
+                        },
+                    },
+                    {
+                        '@id': '_:subj-central-biology',
+                        '@type': 'subject',
+                        'name': 'Biology',
+                        'parent': {
+                            '@id': '_:subj-central-life-sciences',
+                            '@type': 'subject',
+                        },
+                    },
+                    {
+                        '@id': '_:subj-central-life-sciences',
+                        '@type': 'subject',
+                        'name': 'Life Sciences',
+                    },
                 ],
             },
         },
@@ -257,6 +358,13 @@
             'registration_type': 'Open-Ended Registration',
             'retracted': False,
             'sources': ['osf reg'],
+            'subject_synonyms': [
+                'bepress|Life Sciences|Biology',
+            ],
+            'subjects': [
+                'bepress|Architecture',
+                'osf reg|Custom life sciencesssss|Custom biologyyyy',
+            ],
             'title': 'Assorted chair',
             'type': 'registration',
             'types': ['registration', 'publication', 'creative work'],