llm locale funzianante

app/llm_handling.py

@@ -1,6 +1,7 @@
 import logging
 import os
 import shutil
+from enum import Enum
 
 from openai import OpenAI
 from langchain_community.vectorstores import FAISS
@@ -13,11 +14,24 @@ from app.configs.prompts import SYSTEM_PROMPTS
 
 logging.basicConfig(level=logging.INFO)
 
+class LLMType(Enum):
+    OPENAI = "openai"
+    LOCAL = "local"
+
+# Client OpenAI standard
+openai_client = OpenAI(api_key=OPENAI_API_KEY)
+
+# Client LM Studio locale
+local_client = OpenAI(
+    base_url="http://192.168.140.5:1234/v1",
+    api_key="not-needed"
+)
+
 def get_system_prompt(prompt_type="tutor"):
     """Seleziona il prompt di sistema appropriato"""
     return SYSTEM_PROMPTS.get(prompt_type, SYSTEM_PROMPTS["tutor"])
 
-def answer_question(question, db_name, prompt_type="tutor", chat_history=None):
+def answer_question(question, db_name, prompt_type="tutor", chat_history=None, llm_type=LLMType.OPENAI):
     """
     Risponde alla domanda 'question' usando i documenti del database 'db_name'.
     Restituisce una lista di 2 messaggi in formato:
@@ -60,24 +74,30 @@ def answer_question(question, db_name, prompt_type="tutor", chat_history=None):
 
         # Prepara il contesto dai documenti
         context = "\n".join([doc.page_content for doc in relevant_docs])
+        prompt = SYSTEM_PROMPTS[prompt_type].format(context=context)
         
-        client = OpenAI(api_key=OPENAI_API_KEY)
-        
-        messages = [
-            {"role": "system", "content": SYSTEM_PROMPTS[prompt_type].format(context=context)},
-            {"role": "user", "content": question}
-        ]
-        
-        # Esegui la chiamata a OpenAI
-        response = client.chat.completions.create(
-            model="gpt-4o-mini",
-            messages=messages,
-            temperature=0.5,
-            max_tokens=3048
-        )
-        
-        answer = response.choices[0].message.content
-        
+        if llm_type == LLMType.OPENAI:
+            response = openai_client.chat.completions.create(
+                model="gpt-4-mini",
+                messages=[
+                    {"role": "system", "content": prompt},
+                    {"role": "user", "content": question}
+                ],
+                temperature=0.7
+            )
+            answer = response.choices[0].message.content
+            
+        else:  # LOCAL
+            response = local_client.chat.completions.create(
+                model="qwen2.5-coder-7b-instruct",
+                messages=[
+                    {"role": "system", "content": prompt},
+                    {"role": "user", "content": question}
+                ],
+                temperature=0.7
+            )
+            answer = response.choices[0].message.content
+            
         return [
             {"role": "user", "content": question},
             {"role": "assistant", "content": answer}

ui/chatbot_tab.py

@@ -3,10 +3,10 @@
 import gradio as gr
 from app.functions.database_handling import list_databases
 from app.configs.prompts import SYSTEM_PROMPTS  # Aggiunta importazionei
-from app.llm_handling import answer_question
+from app.llm_handling import answer_question, LLMType  # Aggiungi LLMType
 from utils.helpers import extract_text_from_files
 
-def create_chatbot_tab(create_interface=True):
+def create_chatbot_tab():
     """Crea il tab 'Chatbot' dell'interfaccia Gradio."""
     
     def chat_upload_and_respond(files, chat_history, db_name):
@@ -23,12 +23,20 @@ def create_chatbot_tab(create_interface=True):
         
         return chat_history
 
-    def respond(message, chat_history, db_name, prompt_type):
+    def respond(message, chat_history, db_name, prompt_type, llm_type):
         """Genera una risposta alla domanda dell'utente e aggiorna la chat."""
         if chat_history is None:
             chat_history = []
 
-        new_messages = answer_question(message, db_name, prompt_type)
+        # Converti stringa in enum
+        selected_llm = LLMType.LOCAL if llm_type == "local" else LLMType.OPENAI
+        
+        new_messages = answer_question(
+            message, 
+            db_name, 
+            prompt_type, 
+            llm_type=selected_llm
+        )
         chat_history.extend(new_messages)
 
         return "", chat_history
@@ -41,46 +49,52 @@ def create_chatbot_tab(create_interface=True):
     databases = list_databases()
 
     with gr.Tab("Chatbot"):
+        # Prima riga: Dropdown selettori
         with gr.Row():
-            with gr.Column(scale=2):
-                # Singolo dropdown per il database
+            with gr.Column(scale=1):
                 db_name_chat = gr.Dropdown(
                     choices=databases,
                     label="Seleziona Database",
                     value="default_db"
                 )
-
+            
+            with gr.Column(scale=1):
                 prompt_selector = gr.Dropdown(
                     choices=list(SYSTEM_PROMPTS.keys()),
                     label="Seleziona Stile Risposta",
                     value="tutor"
                 )
-
-                # Componente Chatbot
-                chatbot = gr.Chatbot(label="Conversazione", type="messages")
-                
-                # Input per la domanda
-                question_input = gr.Textbox(
-                    label="Fai una domanda",
-                    placeholder="Scrivi qui la tua domanda...",
-                    lines=2
+            
+            with gr.Column(scale=1):
+                llm_selector = gr.Dropdown(
+                    choices=["openai", "local"],
+                    label="Seleziona Modello",
+                    value="openai"
                 )
-                
-                # Bottoni per azioni
-                with gr.Row():
-                    ask_button = gr.Button("Invia")
-                    clear_button = gr.Button("Pulisci Chat") 
-                
-                # Upload file con dimensioni ridotte
-                with gr.Row():
-                    file_input = gr.File(
-                        label="Carica PDF/Docx/TXT per la conversazione", 
-                        file_types=[".pdf", ".docx", ".txt"], 
-                        file_count="multiple",
-                        height="100px",
-                        scale=3
-                    )
-                    upload_button = gr.Button("Carica Documenti", scale=1)
+        
+        # Chatbot e input
+        chatbot = gr.Chatbot(label="Conversazione", type="messages")
+        question_input = gr.Textbox(
+            label="Fai una domanda",
+            placeholder="Scrivi qui la tua domanda...",
+            lines=2
+        )
+        
+        # Bottoni per azioni
+        with gr.Row():
+            ask_button = gr.Button("Invia")
+            clear_button = gr.Button("Pulisci Chat") 
+        
+        # Upload file con dimensioni ridotte
+        with gr.Row():
+            file_input = gr.File(
+                label="Carica PDF/Docx/TXT per la conversazione", 
+                file_types=[".pdf", ".docx", ".txt"], 
+                file_count="multiple",
+                height="100px",
+                scale=3
+            )
+            upload_button = gr.Button("Carica Documenti", scale=1)
         
         # Stato della chat
         chat_state = gr.State([])
@@ -94,7 +108,7 @@ def create_chatbot_tab(create_interface=True):
 
         ask_button.click(
             fn=respond,
-            inputs=[question_input, chat_state, db_name_chat, prompt_selector],
+            inputs=[question_input, chat_state, db_name_chat, prompt_selector, llm_selector],  # Aggiungi il selettore del modello
             outputs=[question_input, chatbot]
         )