feat: Use openedx_catalog app, backfill its CourseRuns [FC-0117]

cms/envs/common.py

@@ -898,6 +898,9 @@ def make_lms_template_path(settings):
 
     'openedx_events',
 
+    # Core models to represent courses
+    "openedx_catalog",
+
     # Core apps that power libraries
     "openedx_content",
     *openedx_content_backcompat_apps_to_install(),

lms/envs/common.py

@@ -2020,6 +2020,9 @@
 
     'openedx_events',
 
+    # Core models to represent courses
+    "openedx_catalog",
+
     # Core apps that power libraries
     "openedx_content",
     *openedx_content_backcompat_apps_to_install(),

openedx/core/djangoapps/content/course_overviews/migrations/0030_backfill_new_catalog_courseruns.py

@@ -0,0 +1,150 @@
+"""
+Data migration to populate the new CourseRun and CatalogCourse models.
+"""
+
+# Generated by Django 5.2.11 on 2026-02-13 21:47
+import logging
+
+from django.conf import settings
+from django.db import migrations
+from organizations.api import ensure_organization, exceptions as org_exceptions
+
+log = logging.getLogger(__name__)
+
+# https://github.com/openedx/openedx-platform/issues/38036
+NORMALIZE_LANGUAGE_CODES = {
+    "zh-hans": "zh-cn",
+    "zh-hant": "zh-hk",
+    "ca@valencia": "ca-es-valencia",
+}
+
+
+def backfill_openedx_catalog(apps, schema_editor):
+    """
+    Populate the new CourseRun and CatalogCourse models.
+    """
+    # CourseOverview is a cache model derived from modulestore; modulestore is the source of truth for courses, so we'll
+    # use it to get the list of "all courses on the system" to populate the new CourseRun and CatalogCourse models.
+    CourseIndex = apps.get_model("split_modulestore_django", "SplitModulestoreCourseIndex")
+    CourseOverview = apps.get_model("course_overviews", "CourseOverview")
+    CatalogCourse = apps.get_model("openedx_catalog", "CatalogCourse")
+    CourseRun = apps.get_model("openedx_catalog", "CourseRun")
+
+    created_catalog_course_ids: set[int] = set()
+    all_course_runs = CourseIndex.objects.filter(base_store="mongodb", library_version="").order_by("course_id")
+    for course_run in all_course_runs:
+        org_code: str = course_run.course_id.org
+        course_code: str = course_run.course_id.course
+        run_code: str = course_run.course_id.run
+
+        # Ensure that the Organization exists.
+        try:
+            org_data = ensure_organization(org_code)
+        except org_exceptions.InvalidOrganizationException as exc:
+            # Note: IFF the org exists among the modulestore courses but not in the Organizations database table,
+            # and if auto-create is disabled (it's enabled by default), this will raise InvalidOrganizationException. It
+            # would be up to the operator to decide how they want to resolve that.
+            raise ValueError(
+                f'The organization short code "{org_code}" exists in modulestore ({course_run.course_id}) but '
+                "not the Organizations table, and auto-creating organizations is disabled. You can resolve this by "
+                "creating the Organization manually (e.g. from the Django admin) or turning on auto-creation. "
+                "You can set active=False to prevent this Organization from being used other than for historical data. "
+            ) from exc
+        if org_data["short_name"] != org_code:
+            # On most installations, the 'short_name' database column is case insensitive (unfortunately)
+            log.warning(
+                'The course with ID "%s" does not match its Organization.short_name "%s"',
+                course_run.course_id,
+                org_data["short_name"],
+            )
+
+        # Fetch the CourseOverview if it exists
+        try:
+            course_overview = CourseOverview.objects.get(id=course_run.course_id)
+        except CourseOverview.DoesNotExist:
+            course_overview = None  # Course exists in modulestore but details aren't cached into CourseOverview yet
+        display_name: str = (course_overview.display_name if course_overview else None) or course_code
+
+        # Determine the course language.
+        # Note that in Studio, the options for course language generally came from the ALL_LANGUAGES setting, which is
+        # mostly two-letter language codes with no locale, except it uses "zh_HANS" for Mandarin and "zh_HANT" for
+        # Cantonese. We normalize those to "zh-cn" and "zh-hk" for consistency with our platform UI languages /
+        # Transifex, but you can still access the "old" version using the CatalogCourse.language_short
+        # getter/setter for backwards compatbility. See https://github.com/openedx/openedx-platform/issues/38036
+        language = settings.LANGUAGE_CODE
+        if course_overview and course_overview.language:
+            language = course_overview.language.lower()
+            language = language.replace("_", "-")  # Ensure we use hyphens for consistency (`en-us` not `en_us`)
+            # Normalize this language code. The previous/non-normalized code will still be available via the
+            # "language_short" property for backwards compatibility.
+            language = NORMALIZE_LANGUAGE_CODES.get(language, language)
+            if len(language) > 2 and language[2] != "-":
+                # This seems like an invalid value; revert to the default:
+                log.warning(
+                    'The course with ID "%s" has invalid language "%s" - using default language "%s" instead.',
+                    course_run.course_id,
+                    language,
+                    settings.LANGUAGE_CODE,
+                )
+                language = settings.LANGUAGE_CODE
+
+        # Ensure that the CatalogCourse exists.
+        cc, cc_created = CatalogCourse.objects.get_or_create(
+            org_id=org_data["id"],
+            course_code=course_code,
+            defaults={
+                "display_name": display_name,
+                "language": language,
+            },
+        )
+        if cc_created:
+            created_catalog_course_ids.add(cc.pk)
+        elif cc.pk in created_catalog_course_ids:
+            # This CatalogCourse was previously created during this same migration
+            # Check if all the runs have the same display_name:
+            if (
+                course_overview
+                and course_overview.display_name
+                and course_overview.display_name != cc.display_name
+                and cc.display_name != course_code
+            ):
+                # The runs have different names, so just use the course code as the common catalog course name.
+                cc.display_name = course_code
+                cc.save(update_fields=["display_name"])
+
+        if cc.course_code != course_code:
+            raise ValueError(
+                f"The course {course_run.course_id} exists in modulestore with a different capitalization of its "
+                f'course code compared to other instances of the same run ("{course_code}" vs "{cc.course_code}"). '
+                "This really should not happen. To fix it, delete the inconsistent course runs (!). "
+            )
+
+        # Create the CourseRun
+        new_run, run_created = CourseRun.objects.get_or_create(
+            catalog_course=cc,
+            run=run_code,
+            course_id=course_run.course_id,
+            defaults={"display_name": display_name},
+        )
+
+        # Correct the "created" timestamp. Since it has auto_now_add=True, we can't set its value except using update()
+        # The CourseOverview should have the "created" date unless it's missing or the course was created before
+        # the CourseOverview model existed. In any case, it should be good enough. Otherwise use the default (now).
+        if course_overview:
+            if course_overview.created < cc.created and cc.pk in created_catalog_course_ids:
+                # Use the 'created' date from the oldest course run that we process.
+                CatalogCourse.objects.filter(pk=cc.pk).update(created=course_overview.created)
+            if run_created:
+                CourseRun.objects.filter(pk=new_run.pk).update(created=course_overview.created)
+
+
+class Migration(migrations.Migration):
+    dependencies = [
+        ("openedx_catalog", "0001_initial"),
+        ("course_overviews", "0029_alter_historicalcourseoverview_options"),
+        ("split_modulestore_django", "0003_alter_historicalsplitmodulestorecourseindex_options"),
+    ]
+
+    operations = [
+        migrations.RunPython(backfill_openedx_catalog, reverse_code=migrations.RunPython.noop),
+    ]

openedx/core/djangoapps/content/course_overviews/signals.py

@@ -2,14 +2,15 @@
 Signal handler for invalidating cached course overviews
 """
 
-
 import logging
 
 from django.db import transaction
 from django.db.models.signals import post_save
 from django.dispatch import Signal
 from django.dispatch.dispatcher import receiver
 
+from openedx_catalog import api as catalog_api
+from openedx_catalog.models_api import CourseRun
 from openedx.core.djangoapps.signals.signals import COURSE_CERT_DATE_CHANGE
 from xmodule.data import CertificatesDisplayBehaviors
 from xmodule.modulestore.django import SignalHandler
@@ -33,6 +34,8 @@ def _listen_for_course_publish(sender, course_key, **kwargs):  # pylint: disable
     """
     Catches the signal that a course has been published in Studio and updates the corresponding CourseOverview cache
     entry.
+
+    Also sync course data to the openedx_catalog CourseRun model.
     """
     try:
         previous_course_overview = CourseOverview.objects.get(id=course_key)
@@ -41,6 +44,51 @@ def _listen_for_course_publish(sender, course_key, **kwargs):  # pylint: disable
     updated_course_overview = CourseOverview.load_from_module_store(course_key)
     _check_for_course_changes(previous_course_overview, updated_course_overview)
 
+    # Currently, SplitModulestoreCourseIndex is the ultimate source of truth for
+    # which courses exist. When a course is published, we sync that data to
+    # CourseOverview, and from CourseOverview to CourseRun.
+
+    # In the future, CourseRun will be the "source of truth" and each CourseRun
+    # may optionally point to content and get synced to CourseOverview.
+
+    # Ensure a CourseRun exists for this course
+    try:
+        course_run = catalog_api.get_course_run(course_key)
+    except CourseRun.DoesNotExist:
+        # Presumably this is a newly-created course. Create the CourseRun.
+        course_run = catalog_api.create_course_run_for_modulestore_course_with(
+            course_id=course_key,
+            display_name=updated_course_overview.display_name,
+            language_short=updated_course_overview.language,
+        )
+
+    # Keep the CourseRun up to date as the course is edited:
+    if updated_course_overview.display_name != course_run.display_name:
+        catalog_api.sync_course_run_details(course_key, display_name=updated_course_overview.display_name)
+        # If this course is the only run in the CatalogCourse, should we update the display_name of
+        # the CatalogCourse to match the run's new name? Currently the only way to edit the name of
+        # a CatalogCourse is via the Django admin. But it's also not used anywhere yet.
+
+    if (
+        updated_course_overview.language
+        and updated_course_overview.language != course_run.catalog_course.language_short
+    ):
+        if course_run.catalog_course.runs.count() == 1:
+            # This is the only run in this CatalogCourse. Update the language of the CatalogCourse
+            catalog_api.update_catalog_course(
+                course_run.catalog_course,
+                language_short=updated_course_overview.language,
+            )
+        else:
+            LOG.warning(
+                'Course run "%s" language "%s" does not match its catalog course language, "%s"',
+                str(course_key),
+                updated_course_overview.language,
+                course_run.catalog_course.language_short,
+            )
+
+    # In the future, this will also sync schedule and other metadata to the CourseRun's related models
+
 
 @receiver(SignalHandler.course_deleted)
 def _listen_for_course_delete(sender, course_key, **kwargs):  # pylint: disable=unused-argument
@@ -56,6 +104,16 @@ def _listen_for_course_delete(sender, course_key, **kwargs):  # pylint: disable=
         sender=None,
         courserun_key=courserun_key,
     )
+    # Delete the openedx_catalog CourseRun to keep it in sync:
+    try:
+        course_run_obj = catalog_api.get_course_run(course_key)
+    except CourseRun.DoesNotExist:
+        pass
+    else:
+        catalog_course = course_run_obj.catalog_course
+        catalog_api.delete_course_run(course_key)
+        if catalog_course.runs.count() == 0:
+            catalog_api.delete_catalog_course(catalog_course)
 
 
 @receiver(post_save, sender=CourseOverview)