feat: canonicalize MSCI to MSE course codes

flow/importer/uw/parts/course/canonical.go

@@ -0,0 +1,19 @@
+package course
+
+import "strings"
+
+var subjectRenames = map[string]string{
+	"MSCI": "MSE",
+}
+
+func canonicalSubject(subj string) string {
+	upper := strings.ToUpper(subj)
+	if mapped, ok := subjectRenames[upper]; ok {
+		return mapped
+	}
+	return upper
+}
+
+func canonicalCourseCode(subj, num string) string {
+	return strings.ToLower(canonicalSubject(subj) + num)
+}

flow/importer/uw/parts/course/canonical_test.go

@@ -0,0 +1,23 @@
+package course
+
+import "testing"
+
+func TestCanonicalCourseCode(t *testing.T) {
+	tests := []struct {
+		subj string
+		num  string
+		want string
+	}{
+		{"MSCI", "100", "mse100"},
+		{"msci", "331", "mse331"},
+		{"MSE", "100", "mse100"},
+		{"CS", "115", "cs115"},
+	}
+
+	for _, tt := range tests {
+		got := canonicalCourseCode(tt.subj, tt.num)
+		if got != tt.want {
+			t.Fatalf("canonicalCourseCode(%s,%s) = %s, want %s", tt.subj, tt.num, got, tt.want)
+		}
+	}
+}

flow/importer/uw/parts/course/convert.go

@@ -211,7 +211,8 @@ func convertAll(
 }
 
 func convertCourse(dst *convertResult, apiCourse *apiCourse) error {
-	courseCode := strings.ToLower(apiCourse.Subject + apiCourse.Number)
+	courseCode := canonicalCourseCode(apiCourse.Subject, apiCourse.Number)
+	apiCourse.Subject = canonicalSubject(apiCourse.Subject)
 	newCourse := course{
 		Code: courseCode,
 		Name: apiCourse.Name,

flow/importer/uw/parts/course/fetch.go

@@ -37,7 +37,7 @@ func fetchAll(client *api.Client, termIds []int) ([]apiCourse, []apiClass, error
 			}
 
 			for _, class := range <-sema {
-				class.CourseCode = strings.ToLower(course.Subject + course.Number)
+				class.CourseCode = canonicalCourseCode(course.Subject, course.Number)
 				classes = append(classes, class)
 			}
 
@@ -88,7 +88,8 @@ func fetchCourses(client *api.Client, termIds []int) ([]apiCourse, error) {
 			return nil, fmt.Errorf("failed to fetch term %d: %w", termId, err)
 		}
 		for _, course := range termCourses {
-			courseCode := course.Subject + course.Number
+			course.Subject = canonicalSubject(course.Subject)
+			courseCode := canonicalCourseCode(course.Subject, course.Number)
 			if err != nil {
 				log.Warnf("skipping course with missing data")
 				continue

hasura/migrations/default/1770000000000_merge_msci_to_mse/down.sql

@@ -0,0 +1 @@
+DROP FUNCTION IF EXISTS public.merge_course(text, text);

hasura/migrations/default/1770000000000_merge_msci_to_mse/up.sql

@@ -0,0 +1,89 @@
+-- Merge MSCI course codes into MSE and make the helper reusable for future renames.
+CREATE OR REPLACE FUNCTION public.merge_course(old_code text, new_code text)
+RETURNS void AS $$
+DECLARE
+  old_id int;
+  new_id int;
+BEGIN
+  old_code := lower(old_code);
+  new_code := lower(new_code);
+
+  SELECT id INTO old_id FROM course WHERE code = old_code;
+  SELECT id INTO new_id FROM course WHERE code = new_code;
+
+  IF old_id IS NULL THEN
+    RETURN;
+  END IF;
+
+  IF new_id IS NULL THEN
+    UPDATE course SET code = new_code WHERE id = old_id;
+    RETURN;
+  END IF;
+
+  INSERT INTO course_prerequisite(course_id, prerequisite_id, is_corequisite)
+  SELECT new_id, prerequisite_id, is_corequisite
+  FROM course_prerequisite
+  WHERE course_id = old_id
+  ON CONFLICT DO NOTHING;
+
+  INSERT INTO course_prerequisite(course_id, prerequisite_id, is_corequisite)
+  SELECT course_id, new_id, is_corequisite
+  FROM course_prerequisite
+  WHERE prerequisite_id = old_id
+  ON CONFLICT DO NOTHING;
+
+  DELETE FROM course_prerequisite WHERE course_id = old_id OR prerequisite_id = old_id;
+
+  INSERT INTO course_antirequisite(course_id, antirequisite_id)
+  SELECT new_id, antirequisite_id
+  FROM course_antirequisite
+  WHERE course_id = old_id
+  ON CONFLICT DO NOTHING;
+
+  INSERT INTO course_antirequisite(course_id, antirequisite_id)
+  SELECT course_id, new_id
+  FROM course_antirequisite
+  WHERE antirequisite_id = old_id
+  ON CONFLICT DO NOTHING;
+
+  DELETE FROM course_antirequisite WHERE course_id = old_id OR antirequisite_id = old_id;
+
+  UPDATE course_section SET course_id = new_id WHERE course_id = old_id;
+
+  INSERT INTO user_course_taken(course_id, user_id, term_id, level)
+  SELECT new_id, user_id, term_id, level
+  FROM user_course_taken
+  WHERE course_id = old_id
+  ON CONFLICT DO NOTHING;
+  DELETE FROM user_course_taken WHERE course_id = old_id;
+
+  INSERT INTO user_shortlist(course_id, user_id)
+  SELECT new_id, user_id
+  FROM user_shortlist
+  WHERE course_id = old_id
+  ON CONFLICT DO NOTHING;
+  DELETE FROM user_shortlist WHERE course_id = old_id;
+
+  INSERT INTO review(course_id, prof_id, user_id, liked, course_easy, course_useful, course_comment, prof_clear, prof_engaging, prof_comment, public, legacy, created_at, updated_at)
+  SELECT new_id, prof_id, user_id, liked, course_easy, course_useful, course_comment, prof_clear, prof_engaging, prof_comment, public, legacy, created_at, updated_at
+  FROM review
+  WHERE course_id = old_id
+  ON CONFLICT (course_id, user_id) DO NOTHING;
+  DELETE FROM review WHERE course_id = old_id;
+
+  DELETE FROM course WHERE id = old_id;
+END;
+$$ LANGUAGE plpgsql;
+
+DO $$
+DECLARE
+  rec record;
+BEGIN
+  FOR rec IN SELECT code FROM course WHERE code LIKE 'msci%' LOOP
+    PERFORM merge_course(rec.code, 'mse' || substring(rec.code FROM 5));
+  END LOOP;
+END $$;
+
+REFRESH MATERIALIZED VIEW materialized.course_rating;
+REFRESH MATERIALIZED VIEW materialized.course_search_index;
+REFRESH MATERIALIZED VIEW materialized.course_review_rating;