app.py_01_OK

#MisterAI/Docker_Ollama
#app.py_01
#https://huggingface.co/spaces/MisterAI/Docker_Ollama/


import logging
import requests
from pydantic import BaseModel
from langchain_community.llms import Ollama
from langchain.callbacks.manager import CallbackManager
from langchain.callbacks.streaming_stdout import StreamingStdOutCallbackHandler
import gradio as gr
import threading
import subprocess


logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)



# Cache pour stocker les modèles déjà chargés
loaded_models = {}


# Variable pour suivre l'état du bouton "Stop"
stop_flag = False


def get_model_list():
    url = "https://ollama.com/search"
    response = requests.get(url)

    # Vérifier si la requête a réussi
    if response.status_code == 200:
        # Extraire la liste des modèles depuis la page HTML
        model_list = [model.strip() for model in response.text.split('<span x-test-search-response-title>')[1:]]
        model_list = [model.split('</span>')[0] for model in model_list]
        return model_list
    else:
        logger.error(f"Erreur lors de la récupération de la liste des modèles : {response.status_code} - {response.text}")
        return []

def get_llm(model_name):
    callback_manager = CallbackManager([StreamingStdOutCallbackHandler()])
    return Ollama(model=model_name, callback_manager=callback_manager)

class InputData(BaseModel):
    model_name: str
    input: str
    max_tokens: int = 256
    temperature: float = 0.7



def pull_model(model_name):
    try:
        # Exécuter la commande pour tirer le modèle
        subprocess.run(["ollama", "pull", model_name], check=True)
        logger.info(f"Model {model_name} pulled successfully.")
    except subprocess.CalledProcessError as e:
        logger.error(f"Failed to pull model {model_name}: {e}")
        raise

def check_and_load_model(model_name):
    # Vérifier si le modèle est déjà chargé
    if model_name in loaded_models:
        logger.info(f"Model {model_name} is already loaded.")
        return loaded_models[model_name]
    else:
        logger.info(f"Loading model {model_name}...")
        # Tirer le modèle si nécessaire
        pull_model(model_name)
        llm = get_llm(model_name)
        loaded_models[model_name] = llm
        return llm

        
        
        
# Interface Gradio
def gradio_interface(model_name, input, max_tokens, temperature, stop_button=None):
    global stop_flag
    stop_flag = False
    response = None  # Initialisez la variable response ici

    def worker():
        nonlocal response  # Utilisez nonlocal pour accéder à la variable response définie dans la fonction parente
        llm = check_and_load_model(model_name)
        response = llm(input, max_tokens=max_tokens, temperature=temperature)

    thread = threading.Thread(target=worker)
    thread.start()
    thread.join()

    if stop_flag:
        return "Processing stopped by the user."
    else:
        return response  # Maintenant, response est accessible ici

model_list = get_model_list()

#with gr.Blocks(theme=gr.themes.Glass()) as demo :
#with gr.Blocks() as demo :
#    demo = gr.Interface(
demo = gr.Interface(
    fn=gradio_interface,
    inputs=[
        gr.Dropdown(model_list, label="Select Model", value="mistral"),
        gr.Textbox(label="Input"),
        gr.Slider(minimum=1, maximum=2048, step=1, label="Max Tokens", value=256),
        gr.Slider(minimum=0.0, maximum=1.0, step=0.1, label="Temperature", value=0.7),
        gr.Button(value="Stop", variant="stop")
    ],
    outputs=[
        gr.Textbox(label="Output")
#        gr.Button(value="Stop", variant="stop")
    ],        
    title="Ollama Demo"
)

def stop_processing():
    global stop_flag
    stop_flag = True



if __name__ == "__main__":
    demo.launch(server_name="0.0.0.0", server_port=7860)