Update app.py

app.py

@@ -5,7 +5,6 @@ import gradio as gr
 from agents import AnalyzerAgent, CoachAgent, PredictiveAgent
 
 QUESTIONS_FILE = "questions.json"
-PREDICTIONS_CACHE = "predictions_cache.json"
 
 if not os.path.exists(QUESTIONS_FILE):
     with open(QUESTIONS_FILE, "w", encoding="utf-8") as f:
@@ -14,58 +13,26 @@ if not os.path.exists(QUESTIONS_FILE):
 with open(QUESTIONS_FILE, "r", encoding="utf-8") as f:
     QUESTION_BANK = json.load(f)
 
-if not os.path.exists(PREDICTIONS_CACHE):
-    with open(PREDICTIONS_CACHE, "w", encoding="utf-8") as f:
-        json.dump({}, f)
-
 analyzer = AnalyzerAgent()
 coach_agent = CoachAgent()
-predictor = PredictiveAgent(cache_path=PREDICTIONS_CACHE)
-
-def ensure_predictions_injected(level, subject, n=8):
-    key = f"{level}_{subject}"
-    with open(predictor.cache_path, "r", encoding="utf-8") as f:
-        cache = json.load(f)
-
-    if key in cache and cache[key].get("injected"):
-        return
-
-    preds = predictor.get_or_generate_predictions(level, subject, QUESTION_BANK, n=n)
-
-    next_id = 900000
-    existing_ids = {q["id"] for q in QUESTION_BANK}
-    while next_id in existing_ids:
-        next_id += 1
-
-    for p in preds:
-        QUESTION_BANK.append({
-            "id": next_id,
-            "subject": f"{level}_{subject}",
-            "question_type": p.get("question_type", "mcq"),
-            "text": p.get("text"),
-            "choices": p.get("choices", []),
-            "correct_answer": p.get("predicted_answer", ""),
-            "topics": [p.get("topic", "general")],
-            "difficulty": p.get("difficulty", 3),
-            "source": "predicted"
-        })
-        next_id += 1
-
-    cache[key] = {"predictions": preds, "injected": True}
-    with open(predictor.cache_path, "w", encoding="utf-8") as f:
-        json.dump(cache, f, indent=2, ensure_ascii=False)
+predictor = PredictiveAgent()
 
 def start_exam(level, subject, num_questions=10, include_predicted=True):
-    if include_predicted:
-        ensure_predictions_injected(level, subject, n=8)
-
+    # Filter only real past paper questions
     pool = [q for q in QUESTION_BANK if q.get("subject") == f"{level}_{subject}"]
 
-    if not pool:
+    # Optionally add predictions in memory (not saved)
+    predicted_questions = []
+    if include_predicted:
+        predicted_questions = predictor.generate_predictions(level, subject, n=8)
+
+    # Combine both pools
+    combined_pool = pool + predicted_questions
+    if not combined_pool:
         return [], gr.update(visible=False), gr.update(visible=False), gr.update(visible=True), []
 
-    random.shuffle(pool)
-    selected = pool[:min(num_questions, len(pool))]
+    random.shuffle(combined_pool)
+    selected = combined_pool[:min(num_questions, len(combined_pool))]
 
     exam_data = [
         {"id": q["id"], "text": q["text"], "choices": q.get("choices", []), "topics": q.get("topics", [])}
@@ -82,8 +49,14 @@ def submit_exam(answers, exam_data, level, subject):
     for q in exam_data:
         qid = str(q["id"])
         user_ans = answers.get(qid)
-        orig = next((item for item in QUESTION_BANK if item["id"] == q["id"]), None)
-        correct_ans = orig.get("correct_answer") if orig else None
+        correct_ans = None
+
+        if q["id"] < 900000:  # real past paper
+            orig = next((item for item in QUESTION_BANK if item["id"] == q["id"]), None)
+            correct_ans = orig.get("correct_answer") if orig else None
+        else:  # predicted question
+            correct_ans = q.get("correct_answer")
+
         per_question[qid] = {"user": user_ans, "correct": correct_ans, "topics": q.get("topics", [])}
         if user_ans and correct_ans and str(user_ans).strip() == str(correct_ans).strip():
             correct += 1
@@ -93,10 +66,7 @@ def submit_exam(answers, exam_data, level, subject):
     analysis = analyzer.analyze(per_question)
     coach = coach_agent.coach(analysis, level, subject)
 
-    with open(predictor.cache_path, "r", encoding="utf-8") as f:
-        cache = json.load(f)
-    pred_key = f"{level}_{subject}"
-    predictions_summary = cache.get(pred_key, {})
+    predictions_summary = predictor.summary(level, subject)
 
     return (
         f"Your Score: {score}%",
@@ -161,3 +131,4 @@ if __name__ == "__main__":
 
 
 
+