modules/chatbot.py

from transformers import MBartForConditionalGeneration, MBart50TokenizerFast
import torch

def initialize_chatbot():
    model = MBartForConditionalGeneration.from_pretrained("facebook/mbart-large-50-many-to-many-mmt")
    tokenizer = MBart50TokenizerFast.from_pretrained("facebook/mbart-large-50-many-to-many-mmt")
    return model, tokenizer

def get_chatbot_response(model, tokenizer, prompt, src_lang):
    tokenizer.src_lang = src_lang
    encoded_input = tokenizer(prompt, return_tensors="pt")
    generated_tokens = model.generate(**encoded_input, max_length=100)
    return tokenizer.batch_decode(generated_tokens, skip_special_tokens=True)[0]

def display_chatbot_interface(lang_code):
    translations = {
        'es': {
            'title': "AIdeaText - Chatbot Multilingüe",
            'input_placeholder': "Escribe tu mensaje aquí...",
            'send_button': "Enviar",
        },
        'en': {
            'title': "AIdeaText - Multilingual Chatbot",
            'input_placeholder': "Type your message here...",
            'send_button': "Send",
        },
        'fr': {
            'title': "AIdeaText - Chatbot Multilingue",
            'input_placeholder': "Écrivez votre message ici...",
            'send_button': "Envoyer",
        }
    }

    t = translations[lang_code]

    st.header(t['title'])

    if 'chatbot' not in st.session_state:
        st.session_state.chatbot, st.session_state.tokenizer = initialize_chatbot()

    if 'messages' not in st.session_state:
        st.session_state.messages = []

    for message in st.session_state.messages:
        with st.chat_message(message["role"]):
            st.markdown(message["content"])

    if prompt := st.chat_input(t['input_placeholder']):
        st.session_state.messages.append({"role": "user", "content": prompt})
        with st.chat_message("user"):
            st.markdown(prompt)

        with st.chat_message("assistant"):
            response = get_chatbot_response(st.session_state.chatbot, st.session_state.tokenizer, prompt, lang_code)
            st.markdown(response)
        st.session_state.messages.append({"role": "assistant", "content": response})

        # Guardar la conversación en la base de datos
        store_chat_history(st.session_state.username, st.session_state.messages)