Update app.py

app.py

@@ -1,4 +1,5 @@
 import os
+import re
 import json
 import base64
 import gradio as gr
@@ -126,6 +127,29 @@ def load_id_answer_mapping():
     return json.loads(id_answer_mapping)
 
 
+
+def answer_matching(text):
+    if isinstance(text, list):
+        text = text[0] if text else random.choice(['A', 'B', 'C', 'D'])
+    if not isinstance(text, str):
+        return random.choice(['A', 'B', 'C', 'D'])
+    
+    patterns = [
+        r'\((A|B|C|D)\)',
+        r'^(A|B|C|D)[\s\W]*',
+        r'\b[A-D]\b',
+        r'\((a|b|c|d)\)',
+        r'\b(A|B|C|D)\.',
+    ]
+    for pattern in patterns:
+        match = re.search(pattern, text)
+        if match:
+            return match.group(1).upper()
+    
+    letters = re.findall(r'[a-zA-Z]', text)
+    return letters[0].upper() if len(letters) == 1 else random.choice(['A', 'B', 'C', 'D'])
+
+
 def evaluate_uploaded_json(user_file, model_name):
     print(f"Model Name: {model_name}")
     print(f"Uploaded File: {user_file}")
@@ -143,7 +167,8 @@ def evaluate_uploaded_json(user_file, model_name):
 
     for item in user_data:
         question_id = item["id"]
-        user_answer = item.get("model_answer")
+        raw_user_answer = item.get("model_answer", "")
+        user_answer = answer_matching(raw_user_answer)
         question_class = item.get("class", "Unknown")
 
         class_total[question_class] += 1
@@ -154,7 +179,8 @@ def evaluate_uploaded_json(user_file, model_name):
             correct += 1
 
     subclass_data = []
-    subclass_result = {}  
+    subclass_result = {}
+
     for cls in CLASS_LIST[:-5]:  
         acc = class_correct[cls] / class_total[cls] if class_total[cls] > 0 else 0
         subclass_data.append({
@@ -162,8 +188,7 @@ def evaluate_uploaded_json(user_file, model_name):
             "Accuracy": f"{acc:.2%}",
             "Correct/Total": f"{class_correct[cls]}/{class_total[cls]}"
         })
-        subclass_result[cls] = acc  
-
+        subclass_result[cls] = acc
 
     category_data = []
     for category, sub_classes in CATEGORY_MAPPING.items():
@@ -175,14 +200,13 @@ def evaluate_uploaded_json(user_file, model_name):
             "Accuracy": f"{acc:.2%}",
             "Correct/Total": f"{cat_correct}/{cat_total}"
         })
-        subclass_result[category] = acc 
+        subclass_result[category] = acc
 
     overall_accuracy = f"{correct / total:.2%} ({correct}/{total} correct)"
 
     subclass_df = pd.DataFrame(subclass_data)
     category_df = pd.DataFrame(category_data)
 
-
     save_class_accuracy_to_hf_dataset(model_name, subclass_result)
 
     return overall_accuracy, category_df, subclass_df