update extracting grade scripts

app/util/courses.py

@@ -1,13 +1,13 @@
 #!/usr/bin/env python3.11
 import json
 import csv
-from app.util.estimateGrade import getGradeFromCourseCode
+from app.util.estimateGrade import getGrade
 
 # Get all requested courses from data
 def getCourses(
   data_dir: str,
-  log: int=False,
-  log_dir: str='./output/raw/courses.json',
+  log:      int = False,
+  log_dir:  str = './output/raw/courses.json',
 ) -> dict[dict[str: any]]:
 
   courses = {}
@@ -20,13 +20,14 @@ def getCourses(
         exists = True if courses[course]["CrsNo"] == row["CrsNo"] else False
         if exists: break
       if not exists:
+        grade = getGrade(row["CrsNo"], row["Description"])
         courses[row["CrsNo"]] = {
           "CrsNo": row["CrsNo"],
           "Requests": 0,
           "Description": row["Description"],
           "Sem1": 0,
           "Sem2":0,
-          "Grade": getGradeFromCourseCode(row["CrsNo"])
+          "Grade": grade
         }
 
   if log:

app/util/estimateGrade.py

@@ -3,14 +3,36 @@ from app.util.globals import flex
 
 most_frequent = lambda l : max(set(l), key = l.count)
 
-def getGradeFromCourseCode(code: str) -> int:
-  grades = [int(s) for s in code.split("-") if s.isdigit()]
-  return None if len(grades) == 0 else most_frequent(grades)
+# extract a grade from a course code or description
+def extractGrade(string: str) -> int:
+  if string in flex: return 12
 
-def getEstimatedGrade(pupil: dict) -> int:
+  grade = None
+  for i in range(len(string)):
+    # If this is a digit like 1 and the next character is a digit like 2 we know the grade is 12
+    # or if this digit is 0 and the next digit is 9 we know the grade is 9
+    if i != len(string) - 1 and string[i].isdigit() and string[i+1].isdigit():
+      grade = int(f'{string[i]}{string[i+1]}')
+      if grade >= 8: break
+
+    # if the above condition is not met, and we know this is a single number, we can just return this single number as the grade
+    elif string[i].isdigit():
+      grade = int(string[i])
+      if grade >= 8: break
+
+  return grade if grade is not None and grade >= 8 else None
+
+# Estimage students grade based off requests
+def estimateStudentGrade(pupil: dict) -> int:
   grades = [] # List of all possible grades
   for request in (r for r in pupil["requests"] if r not in flex):
-    for extractedGrade in [int(s) for s in request["CrsNo"].split("-") if s.isdigit()]:
-      grades.append(extractedGrade)
+    extractedGrade = getGrade(request["CrsNo"], request["Description"])
+    if extractedGrade is not None: grades.append(extractedGrade)
 
   return None if len(grades) == 0 else most_frequent(grades) # Final estimate of grade
+
+# anlyse course code and description for best grade estimate
+def getGrade(crsNo: str, crsDes: str) -> int:
+  grade = extractGrade(crsDes)
+  if grade is None: grade = extractGrade(crsNo)
+  return grade

app/util/students.py

@@ -1,15 +1,15 @@
 #!/usr/bin/env python3.11
 import json
 import csv
-from app.util.estimateGrade import getEstimatedGrade
+from app.util.estimateGrade import estimateStudentGrade
 from app.util.globals import flex
 
 # sort data into usable dictionary
 def getStudents(
-  data_dir: str,
-  log: bool=False,
-  totalBlocks: int=10,
-  log_dir: str='./output/raw/students.json'
+  data_dir:    str,
+  log:         bool = False,
+  totalBlocks: int  = 10,
+  log_dir:     str  = './output/raw/students.json'
 ) -> list[ dict[ str: any ] ]:
   
   students: list[ dict[ str: any ] ] = []
@@ -50,7 +50,7 @@ def getStudents(
 
   # Estimate student grades
   for student in students:
-    student["gradelevel"] = getEstimatedGrade(student)
+    student["gradelevel"] = estimateStudentGrade(student)
 
   if log:
     with open(log_dir, "w") as outfile: