Remove preloaded quiz data and test data for full application testing

- Removed load_preloaded_concepts() function and preloaded concepts loading
- Removed create_test_quiz() function and all test quiz data
- Updated user_state initialization to start with empty data
- Removed all using_preloaded_data logic throughout the codebase
- Fixed quiz initialization to start with empty quiz instead of test data
- Now requires full document upload and processing workflow for testing

app.py

@@ -11,109 +11,20 @@ from utils.tutor import ask_guiding_question, analyze_student_approach
 from utils.exercise_generator import generate_exercise
 from utils.llm_client import call_llm
 
-# Load preloaded concepts data if available
-def load_preloaded_concepts():
-    """Load preloaded concepts from JSON file to avoid LLM calls."""
-    preloaded_file = Path(__file__).parent / "data" / "preloaded_concepts.json"
-    if preloaded_file.exists():
-        try:
-            with open(preloaded_file, 'r', encoding='utf-8') as f:
-                data = json.load(f)
-                print(f"✅ [PRELOAD] Loaded preloaded concepts: {len(data.get('concepts', []))} concepts, {len(data.get('topics', []))} topics, {len(data.get('key_points', []))} key points")
-                return data
-        except Exception as e:
-            print(f"⚠️ [PRELOAD] Failed to load preloaded concepts: {str(e)}")
-            return None
-    else:
-        print("ℹ️ [PRELOAD] No preloaded concepts file found, will use LLM extraction")
-        return None
-
-# Load preloaded data on startup
-preloaded_data = load_preloaded_concepts()
-
-# Create test quiz data for testing (option 1 is always correct)
-def create_test_quiz():
-    """Create test quiz data with option 1 as correct answer for each question."""
-    return [
-        {
-            "question": "What is the time complexity of binary search?",
-            "type": "mcq",
-            "options": [
-                "O(log n)",  # Option 1 - correct
-                "O(n)",
-                "O(n log n)",
-                "O(1)"
-            ],
-            "correct_answer": "O(log n)",
-            "explanation": "Binary search has O(log n) time complexity because it eliminates half of the search space in each iteration."
-        },
-        {
-            "question": "Which data structure follows LIFO (Last In First Out) principle?",
-            "type": "mcq",
-            "options": [
-                "Stack",  # Option 1 - correct
-                "Queue",
-                "Linked List",
-                "Tree"
-            ],
-            "correct_answer": "Stack",
-            "explanation": "A stack follows the LIFO principle where the last element added is the first one to be removed."
-        },
-        {
-            "question": "What is the result of 2^3 in binary?",
-            "type": "mcq",
-            "options": [
-                "1000",  # Option 1 - correct (8 in binary)
-                "1010",
-                "1100",
-                "1110"
-            ],
-            "correct_answer": "1000",
-            "explanation": "2^3 = 8, which is 1000 in binary representation."
-        },
-        {
-            "question": "Which sorting algorithm has the best average time complexity?",
-            "type": "mcq",
-            "options": [
-                "Merge Sort",  # Option 1 - correct
-                "Bubble Sort",
-                "Selection Sort",
-                "Insertion Sort"
-            ],
-            "correct_answer": "Merge Sort",
-            "explanation": "Merge Sort has O(n log n) average time complexity, which is optimal for comparison-based sorting algorithms."
-        },
-        {
-            "question": "What does CPU stand for?",
-            "type": "mcq",
-            "options": [
-                "Central Processing Unit",  # Option 1 - correct
-                "Computer Processing Unit",
-                "Central Program Unit",
-                "Computer Program Unit"
-            ],
-            "correct_answer": "Central Processing Unit",
-            "explanation": "CPU stands for Central Processing Unit, which is the primary component that executes instructions in a computer."
-        }
-    ]
+# Removed preloaded concepts and test quiz data for full testing
 
 # Global state
 user_state = {
-    "concepts": preloaded_data.get("concepts", []) if preloaded_data else [],
-    "topics": preloaded_data.get("topics", []) if preloaded_data else [],
-    "key_points": preloaded_data.get("key_points", []) if preloaded_data else [],
-    "document_parsed": bool(preloaded_data),  # Mark as parsed if preloaded data exists
+    "concepts": [],
+    "topics": [],
+    "key_points": [],
+    "document_parsed": False,
     "learning_style": None,
-    "parsed_data": {
-        "concepts": preloaded_data.get("concepts", []) if preloaded_data else [],
-        "topics": preloaded_data.get("topics", []) if preloaded_data else [],
-        "key_points": preloaded_data.get("key_points", []) if preloaded_data else [],
-        "text_preview": "Preloaded concepts data" if preloaded_data else ""
-    } if preloaded_data else None,
-    "current_quiz": create_test_quiz(),  # Load test quiz on startup
-    "show_quiz_modal": True,  # Show quiz modal by default for testing
-    "learning_context": "",  # User-provided context for better concept extraction
-    "using_preloaded_data": bool(preloaded_data)  # Flag to indicate preloaded data is being used
+    "parsed_data": None,
+    "current_quiz": [],  # Start with empty quiz
+    "show_quiz_modal": False,  # Don't show quiz modal by default
+    "learning_context": "",
+    "using_preloaded_data": False
 }
 
 
@@ -261,23 +172,18 @@ Just tell me what you'd like! 😊"""
 
 For example: "I want to learn about ARM assembly" or "Teach me recursion" """
             else:
-                # Generate quiz - use preloaded test quiz if available
+                # Generate quiz using LLM
                 try:
-                    # If we have preloaded data, use test quiz instead of calling LLM
-                    if user_state.get("using_preloaded_data", False):
-                        print("📝 [QUIZ] Using preloaded test quiz data (no LLM call)")
-                        questions = create_test_quiz()
-                    else:
-                        learning_context = user_state.get("learning_context", "")
-                        questions = generate_quiz(
-                            concepts=user_state["concepts"],
-                            topics=user_state.get("topics", []),
-                            num_questions=10,
-                            question_types=["mcq"],
-                            document_context=user_state.get("parsed_data", {}).get("text_preview", ""),
-                            learning_context=learning_context
-                        )
-                    
+                    learning_context = user_state.get("learning_context", "")
+                    questions = generate_quiz(
+                        concepts=user_state["concepts"],
+                        topics=user_state.get("topics", []),
+                        num_questions=10,
+                        question_types=["mcq"],
+                        document_context=user_state.get("parsed_data", {}).get("text_preview", ""),
+                        learning_context=learning_context
+                    )
+
                     if questions:
                         # Store quiz in state
                         user_state["current_quiz"] = questions
@@ -286,7 +192,7 @@ For example: "I want to learn about ARM assembly" or "Teach me recursion" """
                         user_state["quiz_trigger_value"] = user_state.get("quiz_trigger_value", 0) + 1
                         # Update the quiz_trigger State component value (this will trigger the change event)
                         # Note: We can't directly update State from here, so we'll use a workaround
-                        
+
                         response = f"✅ **Quiz Generated!** ({len(questions)} questions)\n\nI've created a quiz based on your concepts. Check the quiz panel on the right! 📝\n\nYou can:\n- Take the quiz interactively\n- Review all questions\n- Generate more questions"
                     else:
                         response = "⚠️ I had trouble generating questions. Could you try again or provide more specific concepts?"
@@ -337,39 +243,26 @@ Or tell me: "Generate exercise for {main_concept} in {detected_lang}" and I'll d
                     user_state.get("topics", [])
                 )
                 try:
-                    # Skip LLM call if using preloaded data
-                    if user_state.get("using_preloaded_data", False):
-                        response = f"""📚 **I can create a progressive learning curriculum for: {main_concept}**
+                    curriculum = generate_curriculum(
+                        concept=main_concept,
+                        language=detected_lang if detected_lang != "general" else "c",  # Default to C for assembly/low-level
+                        starting_level="beginner",
+                        num_exercises=5,
+                        generate_all=False
+                    )
 
-Since you're using preloaded data, I'll create a curriculum based on the concepts we have.
+                    curriculum_text = f"# 📚 Learning Curriculum: {main_concept}\n\n"
+                    curriculum_text += f"**Overview:** {curriculum.get('overview', '')}\n\n"
+                    curriculum_text += f"**Estimated Time:** {curriculum.get('estimated_total_time_minutes', 0)} minutes\n\n"
+                    curriculum_text += "## Learning Path:\n\n"
 
-**To generate the full curriculum:**
-1. Go to the "📚 Curriculum" tab in the sidebar
-2. The concept field is already filled: **{main_concept}**
-3. Click "🚀 Generate Curriculum"
+                    for ex in curriculum.get("learning_path", [])[:3]:
+                        curriculum_text += f"**Exercise {ex.get('exercise_number')}** ({ex.get('difficulty')}):\n"
+                        for obj in ex.get("learning_objectives", [])[:2]:
+                            curriculum_text += f"- {obj}\n"
+                        curriculum_text += "\n"
 
-Or tell me more about what specific aspects you want to focus on! 🎓"""
-                    else:
-                        curriculum = generate_curriculum(
-                            concept=main_concept,
-                            language=detected_lang if detected_lang != "general" else "c",  # Default to C for assembly/low-level
-                            starting_level="beginner",
-                            num_exercises=5,
-                            generate_all=False
-                        )
-                    
-                        curriculum_text = f"# 📚 Learning Curriculum: {main_concept}\n\n"
-                        curriculum_text += f"**Overview:** {curriculum.get('overview', '')}\n\n"
-                        curriculum_text += f"**Estimated Time:** {curriculum.get('estimated_total_time_minutes', 0)} minutes\n\n"
-                        curriculum_text += "## Learning Path:\n\n"
-                        
-                        for ex in curriculum.get("learning_path", [])[:3]:
-                            curriculum_text += f"**Exercise {ex.get('exercise_number')}** ({ex.get('difficulty')}):\n"
-                            for obj in ex.get("learning_objectives", [])[:2]:
-                                curriculum_text += f"- {obj}\n"
-                            curriculum_text += "\n"
-                        
-                        response = curriculum_text + "\nWould you like to see the full curriculum or start with the first exercise?"
+                    response = curriculum_text + "\nWould you like to see the full curriculum or start with the first exercise?"
                 except Exception as e:
                     response = f"❌ Error generating curriculum: {str(e)}"
         
@@ -387,35 +280,23 @@ Tell me a concept or upload a document, and I'll find the best resources for you
                     user_state.get("topics", [])
                 )
                 try:
-                    # Skip LLM call if using preloaded data
-                    if user_state.get("using_preloaded_data", False):
-                        print(f"📝 [RESOURCES] Using preloaded data - skipping LLM call for: {main_concept}")
-                        # Return basic resources without LLM
-                        resources = [
-                            {
-                                "title": f"{main_concept} - Documentation",
-                                "url": f"https://www.google.com/search?q={main_concept}+documentation",
-                                "excerpt": f"Search for {main_concept} documentation and tutorials online."
-                            }
-                        ]
-                    else:
-                        print(f"🔧 [MCP TOOL] search_learning_resources_tool called (chat) - concept: {main_concept}, language: {detected_lang}")
-                        print(f"🚀 [MCP TOOL] search_learning_resources_tool - Searching for: {main_concept}")
-                        resources = search_learning_resources(
-                            concept=main_concept,
-                            language=detected_lang if detected_lang != "general" else "c",  # Default to C for assembly/low-level
-                            resource_type="all",
-                            max_results=5
-                        )
-                        print(f"✅ [MCP TOOL] search_learning_resources_tool - Found {len(resources)} resources")
-                    
+                    print(f"🔧 [MCP TOOL] search_learning_resources_tool called (chat) - concept: {main_concept}, language: {detected_lang}")
+                    print(f"🚀 [MCP TOOL] search_learning_resources_tool - Searching for: {main_concept}")
+                    resources = search_learning_resources(
+                        concept=main_concept,
+                        language=detected_lang if detected_lang != "general" else "c",  # Default to C for assembly/low-level
+                        resource_type="all",
+                        max_results=5
+                    )
+                    print(f"✅ [MCP TOOL] search_learning_resources_tool - Found {len(resources)} resources")
+
                     resources_text = f"# 🔍 Learning Resources for: {main_concept}\n\n"
                     for i, res in enumerate(resources, 1):
                         resources_text += f"**{i}. {res.get('title', 'Resource')}**\n"
                         if res.get('url'):
                             resources_text += f"🔗 {res['url']}\n"
                         resources_text += f"{res.get('excerpt', '')}\n\n"
-                    
+
                     response = resources_text
                 except Exception as e:
                     response = f"❌ Error finding resources: {str(e)}"
@@ -516,26 +397,7 @@ Provide a helpful, intelligent response. If they're asking a question, answer it
         else:
             # Initial greeting or subject prompt
             if any(word in message_lower for word in ["learn", "teach", "help", "want to", "study"]):
-                # Extract subject from message - SKIP LLM if using preloaded data
-                if user_state.get("using_preloaded_data", False):
-                    # Use preloaded concepts instead of calling LLM
-                    concepts = user_state.get("concepts", [])
-                    subject = concepts[0] if concepts else message_text
-                    response = f"""🎓 **Great! You want to learn about: {subject}**
-
-I've identified these concepts: {', '.join(concepts[:5]) if concepts else subject}
-
-**What would you like to do?**
-1. 📝 Generate a quiz to test your understanding
-2. 📚 Create a progressive learning curriculum
-3. 💻 Generate a programming exercise with starter files and tests
-4. 🔍 Find learning resources and tutorials
-5. 💬 Ask me questions (I'll guide you with questions, not answers!)
-
-**💡 Pro Tip:** Check the "💻 Programming Challenges" tab to generate complete exercise packages with starter files, tests, and hints!
-
-Just tell me what you'd like! 😊"""
-                else:
+                # Extract subject from message using LLM
                     # Extract subject from message
                     try:
                         prompt = f"""Extract the main learning subject/concept from this message: "{message_text}"
@@ -606,9 +468,8 @@ I can help you learn programming and CS concepts in several ways:
 
 **What would you like to do?** 😊"""
                 else:
-                    # Only use LLM for substantial questions, and only if not using preloaded data
-                    if not user_state.get("using_preloaded_data", False):
-                        try:
+                    # Use LLM for substantial questions
+                    try:
                             system_prompt = """You are Gnosis, a friendly and helpful AI learning tutor. You help students learn programming and computer science.
 
 Be conversational, welcoming, and helpful. Explain what you can do in a friendly way."""
@@ -633,11 +494,6 @@ I can help you learn programming and CS concepts. Upload a PDF or tell me what y
                         except Exception as e:
                             response = """👋 **Hi! I'm Gnosis, your AI learning tutor!** 🎓
 
-I can help you learn programming and CS concepts. Upload a PDF or tell me what you want to learn! 😊"""
-                    else:
-                        # Using preloaded data - skip LLM
-                        response = """👋 **Hi! I'm Gnosis, your AI learning tutor!** 🎓
-
 I can help you learn programming and CS concepts. Upload a PDF or tell me what you want to learn! 😊"""
     
     # ChatInterface handles appending to history automatically
@@ -828,9 +684,8 @@ with gr.Blocks(title="Gnosis Tutor") as app:
             
             
         # Quiz Popup Modal (shown when quiz is generated) - positioned as floating panel
-        # Initialize quiz_data_state with test quiz to ensure it's always available
-        initial_quiz = create_test_quiz()
-        quiz_data_state = gr.State(value=initial_quiz)
+        # Initialize quiz_data_state with empty quiz
+        quiz_data_state = gr.State(value=[])
         current_question_idx = gr.State(value=0)
         quiz_score_state = gr.State(value={"correct": 0, "total": 0})
         quiz_trigger = gr.State(value=0)  # Trigger to update quiz modal when quiz is generated
@@ -925,7 +780,6 @@ with gr.Blocks(title="Gnosis Tutor") as app:
                     user_state["key_points"] = parsed_data.get("key_points", [])
                     user_state["document_parsed"] = True
                     user_state["parsed_data"] = parsed_data
-                    user_state["using_preloaded_data"] = False  # Override preloaded data with new extraction
                     
                     print(f"📊 [UPLOAD] Extracted: {len(user_state['concepts'])} concepts, {len(user_state['topics'])} topics, {len(user_state['key_points'])} key points")
                     
@@ -961,15 +815,7 @@ with gr.Blocks(title="Gnosis Tutor") as app:
         return {}, {}, {}, gr.update(), gr.update(), gr.update(), gr.update()
     
     def quick_generate_quiz():
-        """Generate quiz - use preloaded test quiz if available, otherwise use LLM."""
-        # If we have preloaded data, use test quiz instead of calling LLM
-        if user_state.get("using_preloaded_data", False):
-            print("📝 [QUIZ] Using preloaded test quiz data (no LLM call)")
-            questions = create_test_quiz()
-            user_state["current_quiz"] = questions
-            user_state["show_quiz_modal"] = True
-            user_state["quiz_trigger_value"] = user_state.get("quiz_trigger_value", 0) + 1
-            return f"✅ Generated {len(questions)} quiz questions from preloaded data! Check the quiz panel on the right!"
+        """Generate quiz using LLM."""
         
         concepts = user_state.get("concepts", [])
         if not concepts:
@@ -1329,16 +1175,14 @@ with gr.Blocks(title="Gnosis Tutor") as app:
     def load_quiz_to_modal():
         """Load quiz from state into modal."""
         quiz = user_state.get("current_quiz")
-        # Always ensure we have test quiz data for testing
+        # Check if we have a quiz to load
         if not quiz or len(quiz) == 0:
-            quiz = create_test_quiz()
-            user_state["current_quiz"] = quiz
-            print(f"📝 [QUIZ] Created and loaded test quiz with {len(quiz)} questions")
+            print("📝 [QUIZ] No quiz available to load")
+            return [], 0, {"correct": 0, "total": 0}, "No quiz available. Generate a quiz first!", [], gr.update(visible=False), gr.update(visible=False), gr.update(visible=False), gr.update(visible=False)
         else:
             print(f"📝 [QUIZ] Loading existing quiz with {len(quiz)} questions")
         
-        # Always show quiz (remove the empty check)
-        if False:  # Changed to False to always show quiz
+        # Show quiz if available
             return (
                 gr.update(visible=True, value="**No quiz available yet. Ask me to generate a quiz!**"),
                 gr.update(visible=False),