Use new OLL column for secondary topic; fix semester capitalization mismatch

learning_resources/constants.py

@@ -98,7 +98,7 @@ class PrivacyLevel(ExtendedEnum):
     unlisted = "unlisted"
 
 
-semester_mapping = {"1T": "spring", "2T": "summer", "3T": "fall"}
+semester_mapping = {"1T": "Spring", "2T": "Summer", "3T": "Fall"}
 
 
 class LearningResourceRelationTypes(TextChoices):

learning_resources/etl/oll.py

@@ -9,6 +9,7 @@
 
 import requests
 from django.conf import settings
+from django.utils.text import slugify
 
 from learning_resources.constants import (
     Availability,
@@ -28,6 +29,25 @@
 ]
 
 
+def parse_readable_id(course_data: dict, run: dict) -> str:
+    """
+    Parse the course readable_id
+
+    Args:
+        course_data (dict): course data
+        run (dict): run data
+
+    Returns:
+        str: course readable_id
+
+    """
+    if course_data["Offered by"] == "OCW":
+        semester = run.get("semester") or ""
+        year = run.get("year") or ""
+        return f"{course_data["OLL Course"]}+{slugify(semester)}_{year}"
+    return f"MITx+{course_data["OLL Course"]}"
+
+
 def extract(sheets_id: str or None = None) -> str:
     """
     Extract OLL learning_resources
@@ -82,7 +102,8 @@ def parse_topics(course_data: dict) -> list[dict]:
         {"name": topic.replace("Educational Policy", "Education Policy")}
         for topic in [
             course_data["MITxO Primary Child"],
-            course_data["MITxO Secondary Child "],
+            # Sheet/csv column title has trailing space
+            course_data["MITxO Adopted Secondary Child "],
         ]
         if topic
     ]
@@ -136,9 +157,10 @@ def transform_course(course_data: dict) -> dict:
         dict: normalized course data
 
     """
+    runs = transform_run(course_data)
     return {
         "title": course_data["title"],
-        "readable_id": f"MITx+{course_data["OLL Course"]}",
+        "readable_id": parse_readable_id(course_data, runs[0]),
         "url": course_data["url"],
         "description": course_data["description"],
         "full_description": course_data["description"],
@@ -155,7 +177,7 @@ def transform_course(course_data: dict) -> dict:
                 course_data["OLL Course"], is_ocw=False
             ),
         },
-        "runs": transform_run(course_data),
+        "runs": runs,
         "image": transform_image(course_data),
         "prices": [Decimal(0.00)],
         "etl_source": ETLSource.oll.name,

learning_resources/etl/oll_test.py

@@ -33,14 +33,14 @@ def test_oll_transform(mocker, oll_course_data):
 
     assert results[0] == {
         "title": "Introduction to Probability and Statistics",
-        "readable_id": "MITx+18.05",
+        "readable_id": "18.05+summer_2022",
         "url": "https://openlearninglibrary.mit.edu/courses/course-v1:MITx+18.05r_10+2022_Summer/about",
         "description": mocker.ANY,
         "full_description": mocker.ANY,
         "offered_by": {"code": "ocw"},
         "platform": "oll",
         "published": True,
-        "topics": [{"name": "Mathematics"}],
+        "topics": [{"name": "Mathematics"}, {"name": "Data Science"}],
         "course": {
             "course_numbers": [
                 {
@@ -70,7 +70,7 @@ def test_oll_transform(mocker, oll_course_data):
                     {"full_name": "Jennifer French Kamrin"},
                 ],
                 "availability": "Archived",
-                "semester": "summer",
+                "semester": "Summer",
                 "year": 2022,
             }
         ],
@@ -91,7 +91,7 @@ def test_oll_transform(mocker, oll_course_data):
         "offered_by": {"code": "mitx"},
         "platform": "oll",
         "published": True,
-        "topics": [{"name": "Education Policy"}],
+        "topics": [{"name": "Education Policy"}, {"name": "Digital Learning"}],
         "course": {
             "course_numbers": [
                 {
@@ -121,7 +121,7 @@ def test_oll_transform(mocker, oll_course_data):
                     {"full_name": "Elizabeth Huttner-Loan"},
                 ],
                 "availability": "Archived",
-                "semester": "spring",
+                "semester": "Spring",
                 "year": 2019,
             }
         ],

learning_resources/etl/openedx_test.py

@@ -203,7 +203,7 @@ def test_transform_course(  # noqa: PLR0913
                         "last_modified": any_instance_of(datetime),
                         "level": ["intermediate"],
                         "prices": ["0.00", "150.00"],
-                        "semester": "spring",
+                        "semester": "Spring",
                         "description": "short_description",
                         "start_date": expected_dt,
                         "title": "The Analytics Edge",

learning_resources/utils.py

@@ -95,7 +95,7 @@ def get_year_and_semester(course_run):
         )
         if match:
             year = int(match.group(1))
-            semester = match.group(2).lower()
+            semester = match.group(2)
         else:
             semester = None
             year = course_run.get("start")[:4] if course_run.get("start") else None