final quiz generation

self_hosted_env/voice_agent/gunicorn.conf.py

+# Network settings
 bind = "0.0.0.0:8000"
+
+# Worker settings
 worker_class = "uvicorn.workers.UvicornWorker"
 workers = 4 
+
+timeout = 300
+
+# Logging settings
 accesslog = "-"
 errorlog = "-"
 loglevel = "info"
\ No newline at end of file

self_hosted_env/voice_agent/main.py

@@ -162,9 +162,7 @@ def create_app() -> FastAPI:
             shutil.move(temp_csv_path, csv_dest_path)
             print(f"--- Background task: Saved new embeddings to '{csv_dest_path}' ---", flush=True)
 
-            # ==========================================================
-            # === CORRECTED LOGIC: APPEND NEW CURRICULUM TO EXISTING ===
-            # ==========================================================
+
             # --- 3. Read both JSON files ---
             print("--- Background task: Reading generated JSON structure... ---", flush=True)
             with open(temp_json_path, 'r', encoding='utf-8') as f:
@@ -324,8 +322,7 @@ def create_app() -> FastAPI:
         """
         pdf_bytes = await file.read()
         
-        # --- THIS IS THE FIX ---
-        # We now pass BOTH required arguments to the add_task method.
+
         background_tasks.add_task(
             process_pdf_curriculum_in_background, 
             pdf_bytes, 
@@ -443,6 +440,25 @@ def create_app() -> FastAPI:
         except Exception as e:
             logger.error(f"Error in get_dynamic_quiz_handler: {e}")
             raise HTTPException(status_code=500, detail=str(e))
+
+    @app.get("/quiz-interface")
+    async def serve_quiz_interface():
+        """Serve the dynamic quiz generator HTML file"""
+        try:
+            # Check for the file in a 'static' folder first
+            static_file = Path("static/dynamic_quiz_interface.html")
+            if static_file.exists():
+                return FileResponse(static_file)
+            
+            # Fallback to the root directory
+            current_file = Path("dynamic_quiz_interface.html")
+            if current_file.exists():   
+                return FileResponse(current_file)
+            
+            raise HTTPException(status_code=404, detail="Dynamic quiz interface not found")
+        except Exception as e:
+            print(f"Error serving quiz interface: {e}")
+            raise HTTPException(status_code=500, detail=f"Error serving interface: {str(e)}")
             
     @app.options("/get-audio-response")
     async def audio_response_options():

self_hosted_env/voice_agent/services/agent_service.py

@@ -361,18 +361,21 @@ class AgentService:
 
     def get_dynamic_quiz(
         self, grade: int, subject: str, unit: str, concept: str, is_arabic: bool, count: int
-    ) -> List[Dict]:
+        ) -> List[Dict]:
         """
-        Generates a dynamic quiz of 'count' questions using a hybrid approach:
+        Generates a dynamic quiz of 'count' questions using a hybrid approach with BATCHED generation:
         1. Always generates a "freshness batch" of new questions.
         2. Retrieves all questions and checks if the total meets the 'count'.
-        3. If not, generates the remaining number of questions needed.
+        3. If not, generates the remaining number of questions needed IN BATCHES.
         """
         if not self.pgvector:
             raise HTTPException(status_code=503, detail="Vector service is not available for this feature.")
-            
-        # --- PART 1: Follow the original logic to ensure freshness ---
-
+        
+        # Define maximum questions per batch to avoid token limits
+        MAX_QUESTIONS_PER_BATCH = 10
+        
+        # --- PART 1: Generate freshness questions ---
+        
         # 1. Calculate how many new questions to generate for freshness.
         num_fresh_questions = min(max(1, math.floor(count / 3)), 5)
         logger.info(f"Request for {count} questions. Step 1: Generating {num_fresh_questions} new 'freshness' questions.")
@@ -386,7 +389,7 @@ class AgentService:
         except Exception as e:
             logger.warning(f"Could not generate 'freshness' questions for the quiz due to an error: {e}")
 
-        # --- PART 2: Check for a shortfall and generate more if needed ---
+        # --- PART 2: Check for shortfall and generate more if needed (WITH BATCHING) ---
 
         # 3. Retrieve ALL available questions for the topic from the database.
         all_mcqs_after_freshness = self.pgvector.get_mcqs(
@@ -397,20 +400,37 @@ class AgentService:
         # 4. Calculate if there is still a shortfall.
         questions_still_needed = count - len(all_mcqs_after_freshness)
 
-        # 5. If we still need more questions, generate the exact number missing.
+        # 5. If we still need more questions, generate them IN BATCHES.
         if questions_still_needed > 0:
-            logger.info(f"After freshness batch, have {len(all_mcqs_after_freshness)} questions. Generating {questions_still_needed} more to meet count of {count}.")
+            logger.info(f"After freshness batch, have {len(all_mcqs_after_freshness)} questions. Need to generate {questions_still_needed} more to meet count of {count}.")
             
-            # Cap this second generation step as a safeguard.
-            num_to_generate_gap = min(questions_still_needed, 10)
+            total_generated = 0
+            remaining = questions_still_needed
             
-            try:
-                self.generate_and_store_mcqs(
-                    grade=grade, subject=subject, unit=unit, concept=concept,
-                    is_arabic=is_arabic, num_questions=num_to_generate_gap
-                )
-            except Exception as e:
-                logger.warning(f"Could not generate the remaining {num_to_generate_gap} questions due to an error: {e}")
+            while remaining > 0:
+                # Determine batch size (cap at MAX_QUESTIONS_PER_BATCH and also cap total at 20 per session)
+                batch_size = min(remaining, MAX_QUESTIONS_PER_BATCH)
+                
+                
+                try:
+                    logger.info(f"Generating batch {total_generated // MAX_QUESTIONS_PER_BATCH + 1} of {batch_size} questions...")
+                    self.generate_and_store_mcqs(
+                        grade=grade, subject=subject, unit=unit, concept=concept,
+                        is_arabic=is_arabic, num_questions=batch_size
+                    )
+                    total_generated += batch_size
+                    remaining -= batch_size
+                    logger.info(f"Successfully generated batch. Total gap-filling questions generated: {total_generated}")
+                    
+                except Exception as e:
+                    logger.error(f"Failed to generate batch of {batch_size} questions: {e}")
+                    # If we've generated at least some questions, continue
+                    if total_generated > 0:
+                        logger.warning(f"Continuing with {total_generated} gap-filling questions generated so far.")
+                        break
+                    else:
+                        logger.warning(f"Could not generate gap-filling questions: {e}")
+                        break
 
         # --- PART 3: Final Assembly and Return ---
 
@@ -422,10 +442,14 @@ class AgentService:
 
         if not final_pool:
             raise HTTPException(status_code=404, detail="No questions could be found or generated for this topic.")
+        
+        # Check if we have enough questions
+        if len(final_pool) < count:
+            logger.warning(f"Could only gather {len(final_pool)} questions out of {count} requested. Returning all available questions.")
             
         # 7. Randomly select the desired number of questions from the final pool.
         random.shuffle(final_pool)
-        final_quiz = final_pool[:count]
+        final_quiz = final_pool[:min(count, len(final_pool))]
         
         logger.info(f"Returning a dynamic quiz of {len(final_quiz)} questions for '{concept}'.")
         return final_quiz
\ No newline at end of file

self_hosted_env/voice_agent/start.sh

@@ -15,4 +15,4 @@ echo "MCQ table setup complete."
 
 sleep 5
 # Start the web server and keep it as the main process
-exec gunicorn -c gunicorn.conf.py main:app
\ No newline at end of file
+exec gunicorn -c gunicorn.conf.py main:app