Refactor course numbers

frontends/api/src/test-utils/factories/learningResources.ts

@@ -139,7 +139,7 @@ const learningResource: Factory<LearningResource> = (
         certification: faker.lorem.word(),
         offered_by: faker.lorem.word(),
         course: {
-          extra_course_numbers: maybe(() => repeat(faker.lorem.word)) ?? null,
+          course_numbers: maybe(() => repeat(faker.datatype.json)) ?? [],
         },
       }
     } else if (type === ResourceTypeEnum.Program) {

learning_resources/etl/constants.py

@@ -48,3 +48,10 @@ class ETLSource(Enum):
     ocw = "ocw"
     prolearn = "prolearn"
     podcast = "podcast"
+
+
+class CourseNumberType(Enum):
+    """Enum of course number types"""
+
+    primary = "primary"
+    cross_listed = "cross-listed"

learning_resources/etl/loaders_test.py

@@ -123,12 +123,12 @@ def test_load_program(
     platform = LearningResourcePlatformFactory.create()
 
     program = (
-        ProgramFactory.create(courses=[], platform=platform)
+        ProgramFactory.create(courses=[], platform=platform.platform)
         if program_exists
-        else ProgramFactory.build(courses=[], platform=platform)
+        else ProgramFactory.build(courses=[], platform=platform.platform)
     )
 
-    LearningResourcePlatformFactory.create(platform=platform)
+    LearningResourcePlatformFactory.create(platform=platform.platform)
 
     if program_exists:
         learning_resource = program.learning_resource
@@ -138,16 +138,16 @@ def test_load_program(
         learning_resource.save()
 
     courses = (
-        CourseFactory.create_batch(2, platform=platform)
+        CourseFactory.create_batch(2, platform=platform.platform)
         if courses_exist
-        else CourseFactory.build_batch(2, platform=platform)
+        else CourseFactory.build_batch(2, platform=platform.platform)
     )
 
     before_course_count = len(courses) if courses_exist else 0
     after_course_count = len(courses)
 
     if program_exists and has_retired_course:
-        course = CourseFactory.create(platform=platform)
+        course = CourseFactory.create(platform=platform.platform)
         before_course_count += 1
         after_course_count += 1
         program.learning_resource.resources.set(
@@ -262,9 +262,9 @@ def test_load_course(  # noqa: PLR0913
     platform = LearningResourcePlatformFactory.create()
 
     course = (
-        CourseFactory.create(runs=[], platform=platform)
+        CourseFactory.create(runs=[], platform=platform.platform)
         if course_exists
-        else CourseFactory.build(runs=[], platform=platform)
+        else CourseFactory.build(runs=[], platform=platform.platform)
     )
 
     learning_resource = course.learning_resource
@@ -283,7 +283,7 @@ def test_load_course(  # noqa: PLR0913
 
     props = {
         "readable_id": learning_resource.readable_id,
-        "platform": platform,
+        "platform": platform.platform,
         "professional": True,
         "title": learning_resource.title,
         "image": {"url": learning_resource.image.url},
@@ -375,13 +375,13 @@ def test_load_duplicate_course(
     platform = LearningResourcePlatformFactory.create()
 
     course = (
-        CourseFactory.create(runs=[], platform=platform)
+        CourseFactory.create(runs=[], platform=platform.platform)
         if course_exists
         else CourseFactory.build()
     )
 
     duplicate_course = (
-        CourseFactory.create(runs=[], platform=platform)
+        CourseFactory.create(runs=[], platform=platform.platform)
         if duplicate_course_exists
         else CourseFactory.build()
     )

learning_resources/etl/mitxonline.py

@@ -13,6 +13,7 @@
 from learning_resources.etl.constants import ETLSource
 from learning_resources.etl.utils import (
     extract_valid_department_from_id,
+    generate_course_numbers_json,
     transform_topics,
 )
 
@@ -146,6 +147,11 @@ def _transform_course(course):
         "runs": [
             _transform_run(course_run, course) for course_run in course["courseruns"]
         ],
+        "course": {
+            "course_numbers": generate_course_numbers_json(
+                course["readable_id"], is_ocw=False
+            ),
+        },
         "published": bool(
             parse_page_attribute(course, "page_url")
         ),  # a course is only considered published if it has a page url

learning_resources/etl/mitxonline_test.py

@@ -4,13 +4,14 @@
 # pylint: disable=redefined-outer-name
 from datetime import datetime
 from itertools import chain
+from unittest.mock import ANY
 from urllib.parse import urljoin
 
 import pytest
 
 from learning_resources.constants import LearningResourceType, PlatformType
 from learning_resources.etl import mitxonline
-from learning_resources.etl.constants import ETLSource
+from learning_resources.etl.constants import CourseNumberType, ETLSource
 from learning_resources.etl.mitxonline import (
     _transform_image,
     parse_page_attribute,
@@ -194,6 +195,15 @@ def test_mitxonline_transform_programs(mock_mitxonline_programs_data):
                         }
                         for course_run_data in course_data["courseruns"]
                     ],
+                    "course": {
+                        "course_numbers": [
+                            {
+                                "value": course_data["readable_id"],
+                                "department": ANY,
+                                "listing_type": CourseNumberType.primary.value,
+                            }
+                        ]
+                    },
                 }
                 for course_data in program_data["courses"]
                 if "PROCTORED EXAM" not in course_data["title"]
@@ -271,6 +281,15 @@ def test_mitxonline_transform_courses(settings, mock_mitxonline_courses_data):
                 }
                 for course_run_data in course_data["courseruns"]
             ],
+            "course": {
+                "course_numbers": [
+                    {
+                        "value": course_data["readable_id"],
+                        "department": ANY,
+                        "listing_type": CourseNumberType.primary.value,
+                    }
+                ]
+            },
         }
         for course_data in mock_mitxonline_courses_data
         if "PROCTORED EXAM" not in course_data["title"]

learning_resources/etl/ocw.py

@@ -26,6 +26,7 @@
 from learning_resources.etl.constants import ETLSource
 from learning_resources.etl.utils import (
     extract_text_metadata,
+    generate_course_numbers_json,
     get_content_type,
 )
 from learning_resources.models import ContentFile, LearningResource
@@ -286,7 +287,7 @@ def transform_course(course_data: dict) -> dict:
     else:
         extra_course_numbers = []
 
-    readable_id = f"{course_data.get(PRIMARY_COURSE_ID)}+{slugify(course_data.get('term'))}_{course_data.get('year')}"  # noqa: E501
+    readable_id = f"{course_data[PRIMARY_COURSE_ID]}+{slugify(course_data.get('term'))}_{course_data.get('year')}"  # noqa: E501
     topics = [
         {"name": topic_name}
         for topic_name in list(
@@ -321,7 +322,11 @@ def transform_course(course_data: dict) -> dict:
         "last_modified": course_data.get("last_modified"),
         "published": True,
         "course": {
-            "extra_course_numbers": extra_course_numbers,
+            "course_numbers": generate_course_numbers_json(
+                course_data[PRIMARY_COURSE_ID],
+                extra_nums=extra_course_numbers,
+                is_ocw=True,
+            ),
         },
         "topics": topics,
         "runs": [transform_run(course_data)],

learning_resources/etl/ocw_test.py

@@ -9,7 +9,8 @@
 from moto import mock_s3
 
 from learning_resources.conftest import OCW_TEST_PREFIX, setup_s3_ocw
-from learning_resources.etl.constants import ETLSource
+from learning_resources.constants import DEPARTMENTS
+from learning_resources.etl.constants import CourseNumberType, ETLSource
 from learning_resources.etl.ocw import (
     transform_content_files,
     transform_contentfile,
@@ -195,8 +196,26 @@ def test_transform_course(settings, legacy_uid, site_uid, expected_uid, has_extr
         assert transformed_json["image"]["alt"] == (
             "Illustration of an aircraft wing showing connections between the disciplines of the course."
         )
-        assert transformed_json["course"]["extra_course_numbers"] == (
-            ["1", "2"] if has_extra_num else []
+        assert transformed_json["course"]["course_numbers"][0] == {
+            "value": "16.01",
+            "department": {"department_id": "16", "name": DEPARTMENTS["16"]},
+            "listing_type": CourseNumberType.primary.value,
+        }
+        assert transformed_json["course"]["course_numbers"][1:] == (
+            [
+                {
+                    "value": "1",
+                    "department": {"department_id": "1", "name": DEPARTMENTS["1"]},
+                    "listing_type": CourseNumberType.cross_listed.value,
+                },
+                {
+                    "value": "2",
+                    "department": {"department_id": "2", "name": DEPARTMENTS["2"]},
+                    "listing_type": CourseNumberType.cross_listed.value,
+                },
+            ]
+            if has_extra_num
+            else []
         )
     else:
         assert transformed_json is None

learning_resources/etl/openedx.py

@@ -12,7 +12,10 @@
 
 from learning_resources.constants import LearningResourceType
 from learning_resources.etl.constants import COMMON_HEADERS
-from learning_resources.etl.utils import extract_valid_department_from_id
+from learning_resources.etl.utils import (
+    extract_valid_department_from_id,
+    generate_course_numbers_json,
+)
 from learning_resources.utils import get_year_and_semester
 
 MIT_OWNER_KEYS = ["MITx", "MITx_PRO"]
@@ -262,6 +265,9 @@ def _transform_course(config, course):
             for course_run in course.get("course_runs", [])
             if _filter_course_run(course_run)
         ],
+        "course": {
+            "course_numbers": generate_course_numbers_json(course.get("key")),
+        },
         "published": any(
             run["status"] == "published" for run in course.get("course_runs", [])
         ),

learning_resources/etl/openedx_test.py

@@ -6,7 +6,7 @@
 import pytest
 
 from learning_resources.constants import LearningResourceType
-from learning_resources.etl.constants import COMMON_HEADERS
+from learning_resources.etl.constants import COMMON_HEADERS, CourseNumberType
 from learning_resources.etl.openedx import (
     OpenEdxConfiguration,
     openedx_extract_transform_factory,
@@ -170,6 +170,18 @@ def test_transform_course(  # noqa: PLR0913
                     "published": True,
                 }
             ],
+            "course": {
+                "course_numbers": [
+                    {
+                        "value": "MITx+15.071x",
+                        "department": {
+                            "department_id": "15",
+                            "name": "Sloan School of Management",
+                        },
+                        "listing_type": CourseNumberType.primary.value,
+                    }
+                ]
+            },
         }
         if not is_course_run_deleted:
             assert transformed_courses[1]["published"] is False

learning_resources/etl/pipelines_test.py

@@ -208,7 +208,12 @@ def test_ocw_courses_etl(settings, mocker):
 
     resource = LearningResource.objects.first()
     assert resource.readable_id == "16.01+fall_2005"
-    assert resource.course.extra_course_numbers == ["16.02", "16.03", "16.04"]
+    assert [num["value"] for num in resource.course.course_numbers] == [
+        "16.01",
+        "16.02",
+        "16.03",
+        "16.04",
+    ]
     assert resource.platform.platform == PlatformType.ocw.name
     assert resource.offered_by.code == OfferedBy.ocw.name
     assert resource.departments.first().department_id == "16"

learning_resources/etl/prolearn.py

@@ -316,6 +316,9 @@ def _transform_course(
         "url": parse_url(course),
         "image": parse_image(course),
         "description": course["body"],
+        "course": {
+            "course_numbers": [],
+        },
         "published": True,
         "topics": parse_topic(course),
         "runs": _transform_runs(course),

learning_resources/etl/prolearn_test.py

@@ -204,6 +204,7 @@ def test_prolearn_transform_courses(mock_mitpe_courses_data):
                     course["start_value"], course["end_value"]
                 )
             ],
+            "course": {"course_numbers": []},
         }
         for course in extracted_data
     ]