feat: added spanish translations for algolia objects

enterprise_catalog/apps/ai_curation/utils/algolia_utils.py

@@ -5,9 +5,7 @@
 
 from django.utils.html import strip_tags
 
-from enterprise_catalog.apps.catalog.algolia_utils import (
-    get_initialized_algolia_client,
-)
+from enterprise_catalog.apps.api_client.algolia import AlgoliaSearchClient
 
 
 @dataclass
@@ -60,6 +58,15 @@ def extract_program_data(hit: dict):
     }
 
 
+def get_initialized_algolia_client():
+    """
+    Initializes and returns an Algolia client for updating search indices
+    """
+    algolia_client = AlgoliaSearchClient()
+    algolia_client.init_index()
+    return algolia_client
+
+
 def fetch_catalog_metadata_from_algolia(enterprise_catalog_query_title: str):
     """
     Returns the ocm_courses, exec_ed_courses, programs, subjects from the

enterprise_catalog/apps/ai_curation/utils/open_ai_utils.py

@@ -81,3 +81,51 @@ def get_keywords_to_prose(query):
         return keywords_to_prose[0]
 
     return ''
+
+
+def translate_to_spanish(text):
+    """
+    Translate the given text to Spanish using OpenAI.
+
+    Args:
+        text (str): The text to translate.
+
+    Returns:
+        str: The translated text in Spanish.
+    """
+    if not text:
+        return ''
+
+    content = (
+        f"Translate the following text to Spanish. "
+        f"Return ONLY the translated text without any comments, "
+        f"explanations, or conversational responses:\n\n{text}"
+    )
+    messages = [
+        {
+            'content': content
+        }
+    ]
+    translated_text = chat_completions(messages=messages)
+
+    if translated_text:
+        return translated_text[0]
+    return ''
+
+
+def translate_object_fields(data, fields):
+    """
+    Translate specified fields in a dictionary to Spanish.
+
+    Args:
+        data (dict): The dictionary containing fields to translate.
+        fields (list): A list of field names to translate.
+
+    Returns:
+        dict: A new dictionary with translated fields.
+    """
+    translated_data = data.copy()
+    for field in fields:
+        if field in data and isinstance(data[field], str):
+            translated_data[field] = translate_to_spanish(data[field])
+    return translated_data

enterprise_catalog/apps/api/tasks.py

@@ -27,6 +27,7 @@
     _algolia_object_from_product,
     configure_algolia_index,
     create_algolia_objects,
+    create_spanish_algolia_object,
     get_algolia_object_id,
     get_initialized_algolia_client,
     get_pathway_course_keys,
@@ -867,6 +868,32 @@ def add_video_to_algolia_objects(
     batched_metadata = _batched_metadata_with_queries(json_metadata, queries)
     _add_in_algolia_products_by_object_id(algolia_products_by_object_id, batched_metadata)
 
+    # Create and index Spanish version
+    json_metadata_es = create_spanish_algolia_object(json_metadata, video)
+
+    if json_metadata_es:
+        # enterprise customer uuids for Spanish
+        batched_metadata_es = _batched_metadata(
+            json_metadata_es,
+            customer_uuids,
+            'enterprise_customer_uuids',
+            '{}-customer-uuids-{}',
+        )
+        _add_in_algolia_products_by_object_id(algolia_products_by_object_id, batched_metadata_es)
+
+        # enterprise catalog uuids for Spanish
+        batched_metadata_es = _batched_metadata(
+            json_metadata_es,
+            catalog_uuids,
+            'enterprise_catalog_uuids',
+            '{}-catalog-uuids-{}',
+        )
+        _add_in_algolia_products_by_object_id(algolia_products_by_object_id, batched_metadata_es)
+
+        # enterprise catalog queries for Spanish
+        batched_metadata_es = _batched_metadata_with_queries(json_metadata_es, queries)
+        _add_in_algolia_products_by_object_id(algolia_products_by_object_id, batched_metadata_es)
+
 
 def add_metadata_to_algolia_objects(
     metadata,
@@ -954,6 +981,32 @@ def add_metadata_to_algolia_objects(
     batched_metadata = _batched_metadata_with_queries(json_metadata, queries)
     _add_in_algolia_products_by_object_id(algolia_products_by_object_id, batched_metadata)
 
+    # Create and index Spanish version
+    json_metadata_es = create_spanish_algolia_object(json_metadata, metadata)
+
+    if json_metadata_es:
+        # enterprise catalog uuids for Spanish
+        batched_metadata_es = _batched_metadata(
+            json_metadata_es,
+            catalog_uuids,
+            'enterprise_catalog_uuids',
+            '{}-catalog-uuids-{}',
+        )
+        _add_in_algolia_products_by_object_id(algolia_products_by_object_id, batched_metadata_es)
+
+        # enterprise customer uuids for Spanish
+        batched_metadata_es = _batched_metadata(
+            json_metadata_es,
+            customer_uuids,
+            'enterprise_customer_uuids',
+            '{}-customer-uuids-{}',
+        )
+        _add_in_algolia_products_by_object_id(algolia_products_by_object_id, batched_metadata_es)
+
+        # enterprise catalog queries for Spanish
+        batched_metadata_es = _batched_metadata_with_queries(json_metadata_es, queries)
+        _add_in_algolia_products_by_object_id(algolia_products_by_object_id, batched_metadata_es)
+
 
 def get_algolia_objects_from_course_content_metadata(content_metadata):
     content_key = content_metadata.content_key
@@ -1059,6 +1112,7 @@ def _get_algolia_products_for_batch(
         )
     ).prefetch_related(
         Prefetch('catalog_queries', queryset=all_catalog_queries),
+        'translations',
     )
     if getattr(settings, 'SHOULD_INDEX_COURSES_WITH_RESTRICTED_RUNS', False):
         # Make the courses that we index actually contain restricted runs in the payload.
@@ -1081,6 +1135,7 @@ def _get_algolia_products_for_batch(
         parent_content_key__in=course_content_keys
     ).prefetch_related(
         Prefetch('catalog_queries', queryset=all_catalog_queries),
+        'translations',
     )
     course_run_content_keys = [cm.content_key for cm in content_metadata_courseruns]
     videos = Video.objects.filter(

enterprise_catalog/apps/api/tests/test_algolia_integration.py

@@ -0,0 +1,150 @@
+"""
+Integration tests for Algolia indexing.
+"""
+from unittest import mock
+
+from django.db import connection
+from django.test import TestCase
+from django.test.utils import CaptureQueriesContext
+
+from enterprise_catalog.apps.api.tasks import _reindex_algolia
+from enterprise_catalog.apps.catalog.constants import COURSE
+from enterprise_catalog.apps.catalog.models import ContentTranslation
+from enterprise_catalog.apps.catalog.tests.factories import (
+    CatalogQueryFactory,
+    ContentMetadataFactory,
+    EnterpriseCatalogFactory,
+)
+
+
+class AlgoliaIntegrationTests(TestCase):
+    """
+    Integration tests for Algolia indexing logic.
+    """
+
+    @mock.patch('enterprise_catalog.apps.api.tasks._retrieve_inactive_tmp_indices')
+    @mock.patch('enterprise_catalog.apps.api.tasks._delete_indices')
+    @mock.patch('enterprise_catalog.apps.api.tasks.configure_algolia_index')
+    @mock.patch('enterprise_catalog.apps.api.tasks.get_initialized_algolia_client')
+    def test_algolia_indexing_with_spanish_translation(
+        self,
+        mock_get_client,
+        _mock_configure,
+        _mock_delete,
+        _mock_retrieve
+    ):
+        """
+        Test that _reindex_algolia pushes double the objects when Spanish translation exists.
+        """
+        # Setup
+        mock_client = mock.Mock()
+        mock_get_client.return_value = mock_client
+
+        # Create a course
+        course = ContentMetadataFactory(content_type=COURSE, content_key='course-v1:Test+Course')
+
+        # Create catalog query and associate with course
+        catalog_query = CatalogQueryFactory()
+        course.catalog_queries.add(catalog_query)
+
+        # Create enterprise catalog associated with query
+        EnterpriseCatalogFactory(catalog_query=catalog_query)
+
+        content_keys = [course.content_key]
+
+        # --- Execution 1: No Translation ---
+        _reindex_algolia(
+            indexable_content_keys=content_keys,
+            nonindexable_content_keys=[],
+            dry_run=False
+        )
+
+        # Verify 3 objects pushed (catalog_uuids, customer_uuids, catalog_queries)
+        self.assertTrue(mock_client.replace_all_objects.called)
+        # Get the generator passed to replace_all_objects
+        args, _ = mock_client.replace_all_objects.call_args
+        generator = args[0]
+        # Consume generator to count objects
+        objects = list(generator)
+        self.assertEqual(len(objects), 3, "Should have 3 objects (English only)")
+        self.assertFalse(any(obj['objectID'].endswith('-es') for obj in objects))
+
+        # Reset mock
+        mock_client.reset_mock()
+
+        # --- Execution 2: With Translation ---
+        ContentTranslation.objects.create(
+            content_metadata=course,
+            language_code='es',
+            title='Curso de Prueba'
+        )
+
+        _reindex_algolia(
+            indexable_content_keys=content_keys,
+            nonindexable_content_keys=[],
+            dry_run=False
+        )
+
+        # Verify 6 objects pushed (3 English + 3 Spanish)
+        self.assertTrue(mock_client.replace_all_objects.called)
+        args, _ = mock_client.replace_all_objects.call_args
+        generator = args[0]
+        objects = list(generator)
+        self.assertEqual(len(objects), 6, "Should have 6 objects (English + Spanish)")
+
+        # Verify IDs
+        english_obj = next(obj for obj in objects if '-es-' not in obj['objectID'])
+        spanish_obj = next(obj for obj in objects if '-es-' in obj['objectID'])
+
+        self.assertIsNotNone(english_obj)
+        self.assertIsNotNone(spanish_obj)
+        self.assertEqual(spanish_obj['title'], 'Curso de Prueba')
+
+    @mock.patch('enterprise_catalog.apps.api.tasks._retrieve_inactive_tmp_indices')
+    @mock.patch('enterprise_catalog.apps.api.tasks._delete_indices')
+    @mock.patch('enterprise_catalog.apps.api.tasks.configure_algolia_index')
+    @mock.patch('enterprise_catalog.apps.api.tasks.get_initialized_algolia_client')
+    def test_algolia_indexing_performance(
+        self,
+        mock_get_client,
+        _mock_configure,
+        _mock_delete,
+        _mock_retrieve
+    ):
+        """
+        Test that indexing performance (DB queries) does not scale linearly with number of items.
+        """
+        mock_client = mock.Mock()
+        mock_get_client.return_value = mock_client
+
+        # Helper to create content
+        def create_content(key_suffix):
+            course = ContentMetadataFactory(content_type=COURSE, content_key=f'course-v1:Test+Course+{key_suffix}')
+            catalog_query = CatalogQueryFactory()
+            course.catalog_queries.add(catalog_query)
+            EnterpriseCatalogFactory(catalog_query=catalog_query)
+            return course.content_key
+
+        # Warmup
+        key1 = create_content('1')
+        _reindex_algolia([key1], [], dry_run=False)
+
+        # Measure for 1 item
+        key2 = create_content('2')
+        with CaptureQueriesContext(connection) as ctx1:
+            _reindex_algolia([key2], [], dry_run=False)
+        count_1_item = len(ctx1)
+
+        # Measure for 2 items
+        key3 = create_content('3')
+        key4 = create_content('4')
+        with CaptureQueriesContext(connection) as ctx2:
+            _reindex_algolia([key3, key4], [], dry_run=False)
+        count_2_items = len(ctx2)
+
+        # The difference should be small (constant overhead), not double.
+        # Ideally count_2_items should be close to count_1_item.
+        diff = count_2_items - count_1_item
+
+        # We expect the difference to be 0 or very small (not proportional to N)
+        self.assertLess(diff, 5, f"Query count increased significantly: {count_1_item} vs {count_2_items}")