Count duplicates

lms/djangoapps/program_enrollments/management/commands/migrate_saml_uids.py

@@ -48,7 +48,7 @@ class Command(BaseCommand):
             uid_mappings = json.load(f)
         slug = options['saml_provider_slug']
 
-        email_map = { m['email']: {'uid': m['student_key'], 'updated': False} for m in uid_mappings }
+        email_map = { m['email']: {'uid': m['student_key'], 'updated': False, 'counted': False } for m in uid_mappings }
         user_queryset = User.objects.prefetch_related('social_auth').filter(social_auth__uid__startswith=slug + ':')
         users = [u for u in user_queryset]
 
@@ -70,8 +70,16 @@ class Command(BaseCommand):
             updated += 1
 
         not_previously_linked = 0
+        updated = 0
+        duplicated_in_mapping = 0
         for mapping in uid_mappings:
-            not_previously_linked += not email_map[mapping['email']]['updated']
+            info_for_email = email_map[mapping['email']]
+            if not info_for_email['counted']:
+                not_previously_linked += not info_for_email['updated']
+                updated += info_for_email['updated']
+                info_for_email['counted'] = True
+            else:
+                duplicated_in_mapping += 1
 
         log.info(
             'Number of users with {slug} UserSocialAuth records for which there was no mapping in the provided file: {missed}'.format(
@@ -83,5 +91,8 @@ class Command(BaseCommand):
                 slug=slug,
                 not_previously_linked=not_previously_linked
         ))
+        log.info('Number of mappings in the mapping file where the identified user has already been processed: {duplicated_in_mapping}'.format(
+            duplicated_in_mapping=duplicated_in_mapping
+        ))
 
 

lms/djangoapps/program_enrollments/management/commands/tests/test_migrate_saml_uids.py

@@ -149,3 +149,16 @@ class TestMigrateSamlUids(TestCase):
         for ind, auth in enumerate(auths):
             auth.refresh_from_db()
             assert auth.uid == self._format_slug_urn_pair(self.provider_slug, new_urn + six.text_type(ind))
+
+    @patch(_COMMAND_PATH + '.log')
+    def test_learner_duplicated_in_mapping(self, mock_log):
+        auth = UserSocialAuthFactory()
+        email = auth.user.email
+        new_urn = '9001'
+
+        mock_info = mock_log.info
+
+        self._call_command('[{}]'.format(
+            ','.join([self._format_email_uid_pair(email, new_urn) for _ in range(5)])
+        ))
+        mock_info.assert_any_call('Number of mappings in the mapping file where the identified user has already been processed: 4')