feat(osf): update migration script to remove unmapped crossref funders

Author: sh-andriy

osf/management/commands/migrate_funder_ids_to_ror.py

@@ -97,7 +97,7 @@ def handle(self, *args, **options):
         self.stdout.write(f"  Records updated: {stats['updated']}")
         self.stdout.write(f"  Records re-indexed: {stats['reindexed']}")
         self.stdout.write(f"  Funders migrated: {stats['funders_migrated']}")
-        self.stdout.write(f"  Funders not in mapping: {stats['not_in_mapping']}")
+        self.stdout.write(f"  Unmapped funders removed: {stats['not_in_mapping']}")
         if stats['errors']:
             self.stdout.write(self.style.ERROR(f"  Errors: {stats['errors']}"))
 
@@ -307,14 +307,15 @@ def migrate_record(self, record, mapping, dry_run, update_funder_name):
                     f'{funder_identifier} -> {ror_info["ror_id"]}'
                 )
             else:
-                # No mapping found, keep original
-                updated_funding_info.append(funder)
+                # No mapping found, remove unmapped Crossref funder
+                record_modified = True
                 funder_stats['not_found'] += 1
                 funder_stats['unmapped_ids'].add(funder_identifier)
 
                 logger.warning(
-                    f'No ROR mapping found for Crossref Funder ID: {funder_identifier} '
-                    f'in record {record.guid._id}'
+                    f'{"[DRY RUN] " if dry_run else ""}'
+                    f'Removing unmapped Crossref Funder ID: {funder_identifier} '
+                    f'from record {record.guid._id}'
                 )
 
         # Warn about duplicate ROR IDs that would result from migration

osf_tests/management_commands/test_migrate_funder_ids_to_ror.py

@@ -201,8 +201,8 @@ def test_migrate_multiple_funders(self, record_with_multiple_funders, csv_mappin
         assert funders[2]['funder_identifier'] == 'https://ror.org/021nxhr62'
         assert funders[2]['funder_identifier_type'] == 'ROR'
 
-    def test_unmapped_funder_preserved(self, record_with_unmapped_funder, csv_mapping_file):
-        """Test that funders not in mapping are preserved unchanged."""
+    def test_unmapped_funder_removed(self, record_with_unmapped_funder, csv_mapping_file):
+        """Test that funders not in mapping are removed."""
         command = Command()
         command.stdout = type('MockStdout', (), {'write': lambda self, x: None})()
 
@@ -214,17 +214,13 @@ def test_unmapped_funder_preserved(self, record_with_unmapped_funder, csv_mappin
             update_funder_name=False
         )
 
-        assert updated is False
+        assert updated is True
         assert stats['migrated'] == 0
         assert stats['not_found'] == 1
         assert 'http://dx.doi.org/10.13039/999999999' in stats['unmapped_ids']
 
         record_with_unmapped_funder.refresh_from_db()
-        funder = record_with_unmapped_funder.funding_info[0]
-
-        # Should be unchanged
-        assert funder['funder_identifier'] == 'http://dx.doi.org/10.13039/999999999'
-        assert funder['funder_identifier_type'] == 'Crossref Funder ID'
+        assert record_with_unmapped_funder.funding_info == []
 
     def test_load_mapping_various_id_formats(self, csv_mapping_file):
         """Test that mapping handles various ID formats."""
@@ -347,17 +343,17 @@ def test_reindex_not_triggered_with_skip_flag(self, record_with_crossref_funder,
         assert funder['funder_identifier'] == 'https://ror.org/01cwqze88'
         assert funder['funder_identifier_type'] == 'ROR'
 
-    def test_reindex_not_triggered_for_unmapped_records(self, record_with_unmapped_funder, csv_mapping_file, mock_reindex):
-        """Test that re-indexing is NOT triggered for records that weren't updated."""
+    def test_reindex_triggered_for_unmapped_records(self, record_with_unmapped_funder, csv_mapping_file, mock_reindex):
+        """Test that re-indexing IS triggered when unmapped funders are removed."""
         mock_update_search, mock_request_identifier_update = mock_reindex
 
         call_command(
             'migrate_funder_ids_to_ror',
             '--csv-file', csv_mapping_file,
         )
 
-        mock_update_search.assert_not_called()
-        mock_request_identifier_update.assert_not_called()
+        mock_update_search.assert_called()
+        mock_request_identifier_update.assert_called_with('doi')
 
     def test_end_to_end_call_command(self, record_with_crossref_funder, record_with_multiple_funders, csv_mapping_file, mock_reindex):
         """Test the full management command end-to-end via call_command."""