feat: programs ONLY inherit UUIDs for catalogs common to ALL content within.

enterprise_catalog/apps/api/tasks.py

@@ -10,6 +10,7 @@
 from celery import shared_task, states
 from celery.exceptions import Ignore
 from celery_utils.logged_task import LoggedTask
+from django.conf import settings
 from django.db import IntegrityError
 from django.db.models import Prefetch, Q
 from django.db.utils import OperationalError
@@ -676,6 +677,7 @@ def add_metadata_to_algolia_objects(
     _add_in_algolia_products_by_object_id(algolia_products_by_object_id, batched_metadata)
 
 
+# pylint: disable=too-many-statements
 def _get_algolia_products_for_batch(
     batch_num,
     content_keys_batch,
@@ -727,6 +729,9 @@ def _get_algolia_products_for_batch(
     customer_uuids_by_key = defaultdict(set)
     catalog_queries_by_key = defaultdict(set)
 
+    catalog_query_uuid_by_catalog_uuid = defaultdict(set)
+    customer_uuid_by_catalog_uuid = defaultdict(set)
+
     # Create a shared convenience queryset to prefetch catalogs for all metadata lookups below.
     all_catalog_queries = CatalogQuery.objects.prefetch_related('enterprise_catalogs')
 
@@ -789,18 +794,42 @@ def _get_algolia_products_for_batch(
             for catalog in associated_catalogs:
                 catalog_uuids_by_key[content_key].add(str(catalog.uuid))
                 customer_uuids_by_key[content_key].add(str(catalog.enterprise_uuid))
+                # Cache UUIDs related to each catalog.
+                catalog_query_uuid_by_catalog_uuid[str(catalog.uuid)] = (str(catalog_query.uuid), catalog_query.title)
+                customer_uuid_by_catalog_uuid[str(catalog.uuid)] = str(catalog.enterprise_uuid)
 
     # Second pass.  This time the goal is to capture indirect relationships on programs:
     #  * For each program:
     #    - Absorb all UUIDs associated with every associated course.
-    for metadata in content_metadata_to_process:
-        if metadata.content_type != PROGRAM:
-            continue
-        program_content_key = metadata.content_key
-        for metadata in program_to_courses_mapping[program_content_key]:
-            catalog_queries_by_key[program_content_key].update(catalog_queries_by_key[metadata.content_key])
-            catalog_uuids_by_key[program_content_key].update(catalog_uuids_by_key[metadata.content_key])
-            customer_uuids_by_key[program_content_key].update(customer_uuids_by_key[metadata.content_key])
+    if settings.ENABLE_ENT_7729_ONLY_SHOW_COMPLETE_PROGRAMS:
+        for program_metadata in content_metadata_to_process:
+            if program_metadata.content_type != PROGRAM:
+                continue
+            program_content_key = program_metadata.content_key
+            catalog_uuids_for_courses_of_program = [
+                catalog_uuids_by_key[course_metadata.content_key]
+                for course_metadata in program_to_courses_mapping[program_content_key]
+            ]
+            common_catalogs = set()
+            if catalog_uuids_for_courses_of_program:
+                common_catalogs = set.intersection(*catalog_uuids_for_courses_of_program)
+            for course_metadata in program_to_courses_mapping[program_content_key]:
+                catalog_queries_by_key[program_content_key].update(
+                    catalog_query_uuid_by_catalog_uuid[catalog_uuid] for catalog_uuid in common_catalogs
+                )
+                catalog_uuids_by_key[program_content_key].update(common_catalogs)
+                customer_uuids_by_key[program_content_key].update(
+                    customer_uuid_by_catalog_uuid[catalog_uuid] for catalog_uuid in common_catalogs
+                )
+    else:  # Old deprecated code in this else block. Remove as part of ENT-7729.
+        for metadata in content_metadata_to_process:
+            if metadata.content_type != PROGRAM:
+                continue
+            program_content_key = metadata.content_key
+            for metadata in program_to_courses_mapping[program_content_key]:
+                catalog_queries_by_key[program_content_key].update(catalog_queries_by_key[metadata.content_key])
+                catalog_uuids_by_key[program_content_key].update(catalog_uuids_by_key[metadata.content_key])
+                customer_uuids_by_key[program_content_key].update(customer_uuids_by_key[metadata.content_key])
 
     # Third pass.  This time the goal is to capture indirect relationships on pathways:
     #  * For each pathway:

enterprise_catalog/apps/api/tests/test_tasks.py

@@ -719,6 +719,188 @@ def _set_up_factory_data_for_algolia(self):
             'course_metadata_unpublished': self.course_metadata_unpublished,
         }
 
+    @mock.patch('django.conf.settings.ENABLE_ENT_7729_ONLY_SHOW_COMPLETE_PROGRAMS', True)
+    @mock.patch('enterprise_catalog.apps.api.tasks.get_initialized_algolia_client', return_value=mock.MagicMock())
+    def test_index_algolia_partial_program(self, mock_search_client):
+        """
+        Assert that when a program contains multiple courses, that program only inherits the UUIDs common to all
+        contained courses.
+
+        This DAG represents the complete test environment:
+        ┌────────────┐┌────────────┐┌────────────┐
+        │*course-1   ││*course-2   ││*course-3   │
+        │------------││------------││------------│
+        │in catalog-1││            ││            │
+        │in catalog-2││in catalog-2││            │
+        │in catalog-3││in catalog-3││in catalog-3│
+        │            ││in catalog-4││in catalog-4│
+        │            ││            ││in catalog-5│
+        └┬───────────┘└┬───────────┘└┬───────────┘
+        ┌▽─────────────▽─────────────▽───────────┐
+        │*program-1                              │
+        │----------------------------------------│
+        │(should inherit catalog-3 only)         │
+        └────────────────────────────────────────┘
+        * = indexable
+        """
+        program_1 = ContentMetadataFactory(content_type=PROGRAM, content_key='program-1')
+        course_1 = ContentMetadataFactory(content_type=COURSE, content_key='test-course-1')
+        course_2 = ContentMetadataFactory(content_type=COURSE, content_key='test-course-2')
+        course_3 = ContentMetadataFactory(content_type=COURSE, content_key='test-course-3')
+
+        # Associate all three courses with the program.
+        course_1.associated_content_metadata.set([program_1])
+        course_2.associated_content_metadata.set([program_1])
+        course_3.associated_content_metadata.set([program_1])
+
+        # Create all 5 test catalogs.
+        catalog_queries = [CatalogQueryFactory(uuid=uuid.uuid4()) for _ in range(5)]
+        catalogs = [
+            EnterpriseCatalogFactory(catalog_query=query)
+            for query in catalog_queries
+        ]
+
+        # Associate the 5 catalogs to the 3 courses in a staggering fashion.
+        course_1.catalog_queries.set(catalog_queries[0:3])
+        course_2.catalog_queries.set(catalog_queries[1:4])
+        course_3.catalog_queries.set(catalog_queries[2:5])
+
+        course_1.save()
+        course_2.save()
+        course_3.save()
+
+        actual_algolia_products_sent = []
+
+        # `replace_all_objects` is swapped out for a mock implementation that forces generator evaluation and saves the
+        # result into `actual_algolia_products_sent` for unit testing.
+        def mock_replace_all_objects(products_iterable):
+            nonlocal actual_algolia_products_sent
+            actual_algolia_products_sent = list(products_iterable)
+        mock_search_client().replace_all_objects.side_effect = mock_replace_all_objects
+
+        with mock.patch('enterprise_catalog.apps.api.tasks.ALGOLIA_FIELDS', self.ALGOLIA_FIELDS):
+            with self.assertLogs(level='INFO') as info_logs:
+                tasks.index_enterprise_catalog_in_algolia_task()  # pylint: disable=no-value-for-parameter
+
+        products_found_log_records = [record for record in info_logs.output if ' products found.' in record]
+        assert ' 15 products found.' in products_found_log_records[0]
+
+        # create expected data to be added/updated in the Algolia index.
+        expected_program_1_objects_to_index = []
+        program_uuid = program_1.json_metadata.get('uuid')
+        expected_program_1_objects_to_index.append({
+            'objectID': f'program-{program_uuid}-catalog-uuids-0',
+            'enterprise_catalog_uuids': [str(catalogs[2].uuid)],
+        })
+        expected_program_1_objects_to_index.append({
+            'objectID': f'program-{program_uuid}-customer-uuids-0',
+            'enterprise_customer_uuids': [str(catalogs[2].enterprise_uuid)],
+        })
+        expected_program_1_objects_to_index.append({
+            'objectID': f'program-{program_uuid}-catalog-query-uuids-0',
+            'enterprise_catalog_query_uuids': [str(catalog_queries[2].uuid)],
+            'enterprise_catalog_query_titles': [catalog_queries[2].title],
+        })
+
+        # verify replace_all_objects is called with the correct Algolia object data.
+        expected_program_call_args = sorted(expected_program_1_objects_to_index, key=itemgetter('objectID'))
+        actual_program_call_args = sorted(
+            [product for product in actual_algolia_products_sent if program_uuid in product['objectID']],
+            key=itemgetter('objectID'),
+        )
+        assert expected_program_call_args == actual_program_call_args
+
+    @mock.patch('django.conf.settings.ENABLE_ENT_7729_ONLY_SHOW_COMPLETE_PROGRAMS', False)
+    @mock.patch('enterprise_catalog.apps.api.tasks.get_initialized_algolia_client', return_value=mock.MagicMock())
+    def test_index_algolia_partial_program_disabled(self, mock_search_client):
+        """
+        Assert that when a program contains multiple courses, that program inherits all the UUIDs from contained
+        courses. This is the old behavior prior to ENT-7729.  Remove this unit test as part of that ticket.
+
+        This DAG represents the complete test environment:
+        ┌────────────┐┌────────────┐┌────────────┐
+        │*course-1   ││*course-2   ││*course-3   │
+        │------------││------------││------------│
+        │in catalog-1││            ││            │
+        │in catalog-2││in catalog-2││            │
+        │in catalog-3││in catalog-3││in catalog-3│
+        │            ││in catalog-4││in catalog-4│
+        │            ││            ││in catalog-5│
+        └┬───────────┘└┬───────────┘└┬───────────┘
+        ┌▽─────────────▽─────────────▽───────────┐
+        │*program-1                              │
+        │----------------------------------------│
+        │(should inherit all catalogs)           │
+        └────────────────────────────────────────┘
+        * = indexable
+        """
+        program_1 = ContentMetadataFactory(content_type=PROGRAM, content_key='program-1')
+        course_1 = ContentMetadataFactory(content_type=COURSE, content_key='test-course-1')
+        course_2 = ContentMetadataFactory(content_type=COURSE, content_key='test-course-2')
+        course_3 = ContentMetadataFactory(content_type=COURSE, content_key='test-course-3')
+
+        # Associate all three courses with the program.
+        course_1.associated_content_metadata.set([program_1])
+        course_2.associated_content_metadata.set([program_1])
+        course_3.associated_content_metadata.set([program_1])
+
+        # Create all 5 test catalogs.
+        catalog_queries = [CatalogQueryFactory(uuid=uuid.uuid4()) for _ in range(5)]
+        catalogs = [
+            EnterpriseCatalogFactory(catalog_query=query)
+            for query in catalog_queries
+        ]
+
+        # Associate the 5 catalogs to the 3 courses in a staggering fashion.
+        course_1.catalog_queries.set(catalog_queries[0:3])
+        course_2.catalog_queries.set(catalog_queries[1:4])
+        course_3.catalog_queries.set(catalog_queries[2:5])
+
+        course_1.save()
+        course_2.save()
+        course_3.save()
+
+        actual_algolia_products_sent = []
+
+        # `replace_all_objects` is swapped out for a mock implementation that forces generator evaluation and saves the
+        # result into `actual_algolia_products_sent` for unit testing.
+        def mock_replace_all_objects(products_iterable):
+            nonlocal actual_algolia_products_sent
+            actual_algolia_products_sent = list(products_iterable)
+        mock_search_client().replace_all_objects.side_effect = mock_replace_all_objects
+
+        with mock.patch('enterprise_catalog.apps.api.tasks.ALGOLIA_FIELDS', self.ALGOLIA_FIELDS):
+            with self.assertLogs(level='INFO') as info_logs:
+                tasks.index_enterprise_catalog_in_algolia_task()  # pylint: disable=no-value-for-parameter
+
+        products_found_log_records = [record for record in info_logs.output if ' products found.' in record]
+        assert ' 15 products found.' in products_found_log_records[0]
+
+        # create expected data to be added/updated in the Algolia index.
+        expected_program_1_objects_to_index = []
+        program_uuid = program_1.json_metadata.get('uuid')
+        expected_program_1_objects_to_index.append({
+            'objectID': f'program-{program_uuid}-catalog-uuids-0',
+            'enterprise_catalog_uuids': sorted([str(catalog.uuid) for catalog in catalogs]),
+        })
+        expected_program_1_objects_to_index.append({
+            'objectID': f'program-{program_uuid}-customer-uuids-0',
+            'enterprise_customer_uuids': sorted([str(catalog.enterprise_uuid) for catalog in catalogs]),
+        })
+        expected_program_1_objects_to_index.append({
+            'objectID': f'program-{program_uuid}-catalog-query-uuids-0',
+            'enterprise_catalog_query_uuids': sorted([str(catalog_query.uuid) for catalog_query in catalog_queries]),
+            'enterprise_catalog_query_titles': sorted([catalog_query.title for catalog_query in catalog_queries]),
+        })
+
+        # verify replace_all_objects is called with the correct Algolia object data.
+        expected_program_call_args = sorted(expected_program_1_objects_to_index, key=itemgetter('objectID'))
+        actual_program_call_args = sorted(
+            [product for product in actual_algolia_products_sent if program_uuid in product['objectID']],
+            key=itemgetter('objectID'),
+        )
+        assert expected_program_call_args == actual_program_call_args
+
     def test_index_content_keys_in_algolia(self):
         """
         Test the _index_content_keys_in_algolia helper function to make sure it creates a generator to support batching

enterprise_catalog/settings/base.py

@@ -429,3 +429,10 @@
     'VERSION': '1.0.0',
     'SERVE_INCLUDE_SCHEMA': False,
 }
+
+# (ENT-7729) When indexing programs in Algolia, only attach catalog query/catalog/customer UUIDs common to all content
+# within the program.  This should have the outcome of only showing completely accessible programs in the catalog search
+# page: https://enterprise.edx.org/<customer>/search
+#
+# Enable this on stage first.
+ENABLE_ENT_7729_ONLY_SHOW_COMPLETE_PROGRAMS = False