Don't publish Professional Ed resources with blank run ids (#2637)

Author: mbertrand

learning_resources/etl/mitpe.py

@@ -233,10 +233,13 @@ def _transform_runs(resource_data: dict) -> list[dict]:
     duration = parse_resource_duration(resource_data.get("duration"))
     published_runs = []
     for run_data in runs_data:
+        run_id = run_data[0]
         start = parse_date(run_data[1])
         end = parse_date(run_data[2])
         enrollment_end = parse_date(run_data[3])
-        published = (not end and not enrollment_end) or (now <= (enrollment_end or end))
+        published = run_id and (
+            (not end and not enrollment_end) or (now <= (enrollment_end or end))
+        )
         if published:
             published_runs.append(
                 {
@@ -283,8 +286,8 @@ def transform_course(resource_data: dict) -> dict or None:
     Returns:
         dict: transformed course data if it has any viable runs
     """
-    runs = _transform_runs(resource_data)
-    if runs:
+    published_runs = _transform_runs(resource_data)
+    if published_runs:
         return {
             "readable_id": resource_data["uuid"],
             "offered_by": copy.deepcopy(OFFERED_BY),
@@ -303,7 +306,7 @@ def transform_course(resource_data: dict) -> dict or None:
             "delivery": transform_delivery(resource_data["learning_format"]),
             "published": True,
             "topics": parse_topics(resource_data),
-            "runs": runs,
+            "runs": published_runs,
             "format": [Format.asynchronous.name],
             "pace": [Pace.instructor_paced.name],
             "availability": Availability.dated.name,
@@ -321,8 +324,8 @@ def transform_program(resource_data: dict) -> dict or None:
     Returns:
         dict: transformed program data
     """
-    runs = _transform_runs(resource_data)
-    if runs:
+    published_runs = _transform_runs(resource_data)
+    if published_runs:
         return {
             "readable_id": resource_data["uuid"],
             "offered_by": copy.deepcopy(OFFERED_BY),
@@ -343,7 +346,7 @@ def transform_program(resource_data: dict) -> dict or None:
             "delivery": transform_delivery(resource_data["learning_format"]),
             "published": True,
             "topics": parse_topics(resource_data),
-            "runs": runs,
+            "runs": published_runs,
             "format": [Format.asynchronous.name],
             "pace": [Pace.instructor_paced.name],
             "availability": Availability.dated.name,

learning_resources/etl/mitpe_test.py

@@ -193,6 +193,22 @@ def read_json(file_path):
     )
 
 
+@pytest.fixture
+def mitpe_offeror_and_topics(mocker):
+    """Create the MIT Professional Education offeror and 2 topics"""
+    offeror = LearningResourceOfferorFactory.create(code="mitpe")
+    LearningResourceTopicMappingFactory.create(
+        offeror=offeror,
+        topic=LearningResourceTopicFactory.create(name="Product Innovation"),
+        topic_name="Technology Innovation",
+    )
+    LearningResourceTopicMappingFactory.create(
+        offeror=offeror,
+        topic=LearningResourceTopicFactory.create(name="Data Science"),
+        topic_name="Data Science",
+    )
+
+
 @pytest.mark.parametrize("prof_ed_api_url", ["http://pro_edd_api.com", None])
 def test_extract(settings, mock_fetch_data, prof_ed_api_url):
     """Test extract function"""
@@ -213,23 +229,22 @@ def test_extract(settings, mock_fetch_data, prof_ed_api_url):
 
 
 @pytest.mark.django_db
-def test_transform(mock_fetch_data):
+def test_transform(mock_fetch_data, mitpe_offeror_and_topics):
     """Test transform function, and effectively most other functions"""
-    offeror = LearningResourceOfferorFactory.create(code="mitpe")
-    LearningResourceTopicMappingFactory.create(
-        offeror=offeror,
-        topic=LearningResourceTopicFactory.create(name="Product Innovation"),
-        topic_name="Technology Innovation",
-    )
-    LearningResourceTopicMappingFactory.create(
-        offeror=offeror,
-        topic=LearningResourceTopicFactory.create(name="Data Science"),
-        topic_name="Data Science",
-    )
     extracted = mitpe.extract()
     assert len(extracted) == 3
     courses, programs = mitpe.transform(extracted)
     assert_json_equal(
         sorted(courses, key=lambda course: course["readable_id"]), EXPECTED_COURSES
     )
     assert_json_equal(programs, EXPECTED_PROGRAMS)
+
+
+@pytest.mark.django_db
+def test_transform__course_no_run_ids(mock_fetch_data, mitpe_offeror_and_topics):
+    """Resources with no run IDs should not be published"""
+    extracted = mitpe.extract()
+    assert len(extracted) == 3
+    for resource in extracted:
+        resource["run__readable_id"] = ""
+    assert mitpe.transform(extracted) == ([], [])