dockerfile con logging 59

app.py

@@ -10,7 +10,6 @@ from datetime import datetime
 from flask import Flask, render_template, redirect, url_for, flash, request  # AGGIUNTO 'request'
 from flask_login import LoginManager, current_user
 import logging
- 
 from werkzeug.middleware.proxy_fix import ProxyFix # NUOVA RIGA 
 from flask_cors import CORS
 
@@ -142,39 +141,27 @@ def get_life_coach_model():
 
     return model_instance
 
- 
+
 def generate_response_threadsafe(prompt: str, conversation_history: list) -> str:
     """
     Generate a response using the model with thread-safe access.
-
-    Args:
-        prompt: User's input message
-        conversation_history: List of previous messages
-
-    Returns:
-        Generated response
     """
-    logger.info(f"=== GENERATE_RESPONSE_THREADSAFE CALLED ===")
-    logger.info(f"Prompt: {prompt}")
-    logger.info(f"History length: {len(conversation_history) if conversation_history else 0}")
-
+    logger.info(f"--- GENERATE_RESPONSE: Chiamata per utente {current_user.username}")
     model = get_life_coach_model()
 
-    logger.info(f"Model instance: {model}")
-    logger.info(f"Model type: {type(model)}")
-
     # Use lock to ensure only one inference at a time (GPU limitation)
     with model_lock:
-        logger.info("Calling model.generate_response()...")
+        logger.info("--- GENERATE_RESPONSE: Acquisito lock, chiamata a model.generate_response()...")
         response = model.generate_response(
             prompt=prompt,
-            max_new_tokens=256,  # Increased for more complete responses
+            max_new_tokens=256,
             conversation_history=conversation_history
         )
-        logger.info(f"Response received: {response[:100]}...")
+        logger.info(f"--- GENERATE_RESPONSE: Risposta ricevuta.")
 
     return response
 
+
 # Import models after db is initialized
 from models import User, Conversation, Message
 
@@ -213,27 +200,7 @@ from chat import chat_bp
 app.register_blueprint(auth_bp, url_prefix='/auth')
 app.register_blueprint(chat_bp, url_prefix='/chat')
 
-# ------------------------------------------------------------------
-# --- NUOVA SEZIONE: PRE-CARICAMENTO MODELLO ASINCRONO (FIX 1) ---
-# ------------------------------------------------------------------
-# Nota: La funzione get_life_coach_model() è già definita in questo file.
-def load_model_on_startup():
-    """Carica il modello in un thread separato per evitare il blocco del server."""
-    try:
-        # L'app_context è necessario per garantire la sicurezza del DB/Logging
-        with app.app_context():
-            logger.info("--- PRE-LOAD: Avvio caricamento modello in background...")
-            get_life_coach_model() # La funzione che attiva il caricamento lento
-            logger.info("--- PRE-LOAD: Caricamento modello completato.")
-    except Exception as e:
-        logger.error(f"--- PRE-LOAD ERROR: Errore nel caricamento: {e}", exc_info=True)
-
-# Avvia il thread immediatamente dopo la configurazione dell'app.
-logger.info("--- PRE-LOAD: Avvio thread per caricamento modello...")
-threading.Thread(target=load_model_on_startup).start()
-# ------------------------------------------------------------------
-# --- FINE PRE-CARICAMENTO ---
-# ------------------------------------------------------------------
+
 @app.after_request
 def add_header(response):
     """Add headers to prevent caching of static files."""