Updated marketing site school loader

The UUID is now used as the primary identifier when refreshing schools. Usage of the title as the key poses issues with schools whose titles and keys differ on LMS (e.g. UC BerkeleyX vs. BerkeleyX). This change ensures that we can still pull changes to images and descriptions, while also supporting manual edits to keys.

EDUCATOR-1531

course_discovery/apps/course_metadata/data_loaders/marketing_site.py

@@ -174,20 +174,36 @@ class SchoolMarketingSiteDataLoader(AbstractMarketingSiteDataLoader):
         return 'school'
 
     def process_node(self, data):
-        key = data['title']
+        # NOTE: Some titles in Drupal have the form "UC BerkeleyX" however, course keys (for which we use the
+        # organization key) cannot contain spaces.
+        key = data['title'].replace(' ', '')
+        uuid = UUID(data['uuid'])
+
         defaults = {
-            'uuid': data['uuid'],
             'name': data['field_school_name'],
             'description': self.clean_html(data['field_school_description']['value']),
             'logo_image_url': self._get_nested_url(data.get('field_school_image_logo')),
             'banner_image_url': self._get_nested_url(data.get('field_school_image_banner')),
             'marketing_url_path': 'school/' + data['field_school_url_slug'],
+            'partner': self.partner,
         }
-        school, __ = Organization.objects.update_or_create(key=key, partner=self.partner, defaults=defaults)
+
+        try:
+            school = Organization.objects.get(uuid=uuid, partner=self.partner)
+            Organization.objects.filter(pk=school.pk).update(**defaults)
+            logger.info('Updated school with key [%s].', school.key)
+        except Organization.DoesNotExist:
+            # NOTE: Some organizations' keys do not match the title. For example, "UC BerkeleyX" courses use
+            # BerkeleyX as the key. Those fixes will be made manually after initial import, and we don't want to
+            # override them with subsequent imports. Thus, we only set the key when creating a new organization.
+            defaults['key'] = key
+            defaults['uuid'] = uuid
+            school = Organization.objects.create(**defaults)
+            logger.info('Created school with key [%s].', school.key)
 
         self.set_tags(school, data)
 
-        logger.info('Processed school with key [%s].', key)
+        logger.info('Processed school with key [%s].', school.key)
         return school
 
     def set_tags(self, school, data):

course_discovery/apps/course_metadata/data_loaders/tests/test_marketing_site.py

@@ -194,10 +194,9 @@ class SchoolMarketingSiteDataLoaderTests(AbstractMarketingSiteDataLoaderTestMixi
     mocked_data = mock_data.MARKETING_SITE_API_SCHOOL_BODIES
 
     def assert_school_loaded(self, data):
-        key = data['title']
-        school = Organization.objects.get(key=key, partner=self.partner)
+        school = Organization.objects.get(uuid=UUID(data['uuid']), partner=self.partner)
         expected_values = {
-            'uuid': UUID(data['uuid']),
+            'key': data['title'],
             'name': data['field_school_name'],
             'description': self.loader.clean_html(data['field_school_description']['value']),
             'logo_image_url': data['field_school_image_logo']['url'],
@@ -220,6 +219,21 @@ class SchoolMarketingSiteDataLoaderTests(AbstractMarketingSiteDataLoaderTestMixi
         for school in schools:
             self.assert_school_loaded(school)
 
+        # If the key of an organization changes, the data loader should continue updating the
+        # organization by matching on the UUID.
+        school = Organization.objects.get(key='MITx', partner=self.partner)
+        # NOTE (CCB): As an MIT alum, this makes me feel dirty. IHTFT(est)!
+        modified_key = 'MassTechX'
+        school.key = modified_key
+        school.save()
+
+        count = Organization.objects.count()
+        self.loader.ingest()
+        school.refresh_from_db()
+
+        assert Organization.objects.count() == count
+        assert school.key == modified_key
+
 
 class SponsorMarketingSiteDataLoaderTests(AbstractMarketingSiteDataLoaderTestMixin, TestCase):
     loader_class = SponsorMarketingSiteDataLoader