Upload chaty_ai_original.py

chaty_ai_original.py

@@ -0,0 +1,661 @@
+# -*- coding: utf-8 -*-
+"""Chaty AI original.ipynb
+
+Automatically generated by Colab.
+
+Original file is located at
+    https://colab.research.google.com/drive/1V1aACjqfyQoyVUSHPpKT3vEYS49J7l-8
+"""
+
+pip install diffusers transformers torch accelerate cohere langchain gradio
+
+import os
+from google.colab import userdata
+
+# Enable notebook access to secrets
+# This line ensures that the notebook can access secrets stored in Colab's secret manager.
+# No output expected, but this is a necessary setup step.
+
+# Retrieve the COHERE_API_KEY from Colab secrets
+COHERE_API_KEY = userdata.get('COHERE_API_KEY')
+
+# Set the environment variable for Cohere
+os.environ["COHERE_API_KEY"] = COHERE_API_KEY
+
+print("COHERE_API_KEY loaded successfully.")
+
+import torch
+from diffusers import StableDiffusionPipeline
+
+# 1. Specify the model ID for Stable Diffusion
+model_id = "runwayml/stable-diffusion-v1-5"
+
+# 2. Load the pre-trained Stable Diffusion model
+# Check if CUDA is available and set the device accordingly
+device = "cuda" if torch.cuda.is_available() else "cpu"
+pipeline = StableDiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.float16)
+pipeline.to(device)
+
+print(f"Stable Diffusion model '{model_id}' loaded successfully on {device}.")
+
+# 3. Define a sample text prompt for image generation
+prompt = "a photo of an astronaut riding a horse on mars"
+
+# 4. Generate an image using the loaded pipeline and the prompt
+print(f"Generating image for prompt: '{prompt}'...")
+image = pipeline(prompt).images[0]
+
+# 5. Display the generated image
+print("Image generated successfully.")
+image
+
+!pip install langchain-cohere
+print("langchain-cohere installed successfully.")
+
+from langchain_cohere import ChatCohere
+from langchain_core.prompts import ChatPromptTemplate, HumanMessagePromptTemplate, SystemMessagePromptTemplate
+from langchain_core.messages import HumanMessage, AIMessage
+
+print("Cohere and Langchain components imported successfully.")
+
+import os
+from google.colab import userdata
+
+# Retrieve the COHERE_API_KEY from Colab secrets
+COHERE_API_KEY = userdata.get('COHERE_API_KEY')
+
+# Set the environment variable for Cohere
+os.environ["COHERE_API_KEY"] = COHERE_API_KEY
+
+from langchain_cohere import ChatCohere
+
+llm = ChatCohere(cohere_api_key=COHERE_API_KEY)
+
+print("Cohere LLM initialized successfully.")
+
+system_message_template = SystemMessagePromptTemplate.from_template(
+    "You are a helpful AI assistant that can generate creative images based on user descriptions. "
+    "You can also answer questions and engage in conversation. "
+    "If a user asks for an image, extract the key elements for a Stable Diffusion prompt."
+)
+
+human_message_template = HumanMessagePromptTemplate.from_template("User: {user_input}")
+
+chat_prompt_template = ChatPromptTemplate.from_messages([
+    system_message_template,
+    human_message_template
+])
+
+print("ChatPromptTemplate created successfully.")
+
+sample_question = "What kind of images can you generate?"
+
+# Create a list of messages using the chat_prompt_template
+formatted_messages = chat_prompt_template.format_messages(user_input=sample_question)
+
+# Invoke the LLM with the formatted messages
+response = llm.invoke(formatted_messages)
+
+print(f"AI Response: {response.content}")
+
+system_message_template = SystemMessagePromptTemplate.from_template(
+    "You are a helpful AI assistant that can generate creative images based on user descriptions. "
+    "You can also answer questions and engage in conversation. "
+    "If a user asks for an image, extract the key elements for a Stable Diffusion prompt and output it in the format 'IMAGE_PROMPT: <your image prompt here>'. "
+    "Otherwise, provide a normal conversational response."
+)
+
+human_message_template = HumanMessagePromptTemplate.from_template("User: {user_input}")
+
+chat_prompt_template = ChatPromptTemplate.from_messages([
+    system_message_template,
+    human_message_template
+])
+
+print("ChatPromptTemplate updated successfully with image generation instruction.")
+
+def handle_user_input(user_message: str):
+    # 2. Format the user_message using the chat_prompt_template
+    formatted_messages = chat_prompt_template.format_messages(user_input=user_message)
+
+    # 3. Invoke the LLM with the formatted messages
+    ai_response = llm.invoke(formatted_messages)
+    response_content = ai_response.content
+
+    generated_image = None
+
+    # 4. Implement logic to determine if an image generation is implied
+    IMAGE_PROMPT_PREFIX = "IMAGE_PROMPT: "
+    if response_content.startswith(IMAGE_PROMPT_PREFIX):
+        # 5. Extract the actual image description
+        image_description = response_content[len(IMAGE_PROMPT_PREFIX):].strip()
+        print(f"AI detected an image request. Generating image for: '{image_description}'")
+        try:
+            # 6. Use the pipeline to generate an image
+            generated_image = pipeline(image_description).images[0]
+            full_response = f"Image generated successfully based on your request: '{image_description}'"
+        except Exception as e:
+            full_response = f"Could not generate image for '{image_description}'. Error: {e}"
+            generated_image = None
+    else:
+        # Otherwise, it's a normal conversational response
+        full_response = response_content
+
+    # 7. Return both the AI's full textual response and the generated image object
+    return full_response, generated_image
+
+# 8. Test the handle_user_input function with a sample conversational input
+conversational_input = "Tell me a fun fact about AI."
+print(f"\nUser: {conversational_input}")
+ai_text_response, generated_image_output = handle_user_input(conversational_input)
+print(f"AI: {ai_text_response}")
+if generated_image_output:
+    print("No image expected for this conversational input.")
+
+import torch
+from diffusers import StableDiffusionPipeline
+
+# 1. Specify the model ID for Stable Diffusion
+model_id = "runwayml/stable-diffusion-v1-5"
+
+# 2. Load the pre-trained Stable Diffusion model
+# Check if CUDA is available and set the device accordingly
+device = "cuda" if torch.cuda.is_available() else "cpu"
+pipeline = StableDiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.float16)
+pipeline.to(device)
+
+print(f"Stable Diffusion model '{model_id}' loaded successfully on {device}.")
+
+image_request_input = "Generate an image of a futuristic city at sunset with flying cars."
+print(f"\nUser: {image_request_input}")
+ai_text_response, generated_image_output = handle_user_input(image_request_input)
+print(f"AI: {ai_text_response}")
+
+if generated_image_output:
+    print("Displaying generated image:")
+    display(generated_image_output)
+else:
+    print("No image was generated for this request.")
+
+import gradio as gr
+
+print("Gradio library imported successfully.")
+
+def gradio_interface_fn(user_query: str):
+    ai_text_response, generated_image_output = handle_user_input(user_query)
+    print(f"DEBUG: Type of generated_image_output: {type(generated_image_output)}")
+    print(f"DEBUG: Is generated_image_output None?: {generated_image_output is None}")
+    if generated_image_output is not None:
+        print(f"DEBUG: Generated image mode: {generated_image_output.mode}, size: {generated_image_output.size}")
+    return ai_text_response, generated_image_output
+
+print("Gradio interface function 'gradio_interface_fn' defined.")
+
+import base64
+import os
+
+image_filename = "Chaty (1).png"
+image_path_in_colab = image_filename
+
+base64_image_tag = ""
+favicon_data_uri = "" # New variable to store the base64 data URI for the favicon
+
+if os.path.exists(image_path_in_colab):
+    try:
+        with open(image_path_in_colab, "rb") as image_file:
+            encoded_string = base64.b64encode(image_file.read()).decode('utf-8')
+        base64_image_tag = f"<img src='data:image/png;base64,{encoded_string}' alt='Chaty Logo' style='height:100px;'>"
+        favicon_data_uri = f"data:image/png;base64,{encoded_string}" # Populate the favicon data URI
+        print("Image encoded to Base64 successfully and data URI for favicon set.")
+    except Exception as e:
+        base64_image_tag = ""
+        favicon_data_uri = ""
+        print(f"An error occurred while encoding the image: {e}")
+else:
+    base64_image_tag = ""
+    favicon_data_uri = ""
+    print(f"Warning: Image file '{image_filename}' not found. Please upload this file to your Colab environment if you wish to display the logo and favicon.")
+
+!pip install gTTS
+print("gTTS installed successfully.")
+
+chat_history = []
+print("Empty chat history list 'chat_history' initialized.")
+
+from langchain_core.messages import HumanMessage, AIMessage, SystemMessage
+from gtts import gTTS
+import os
+
+def handle_user_input(user_message: str, chat_history: list):
+    # Create the current human message
+    current_human_message = HumanMessage(content=user_message)
+
+    # Get the system message content from the existing template
+    system_message_content = system_message_template.format().content
+    system_message = SystemMessage(content=system_message_content)
+
+    # Combine system message, chat history, and current human message for the LLM call
+    messages_for_llm = [system_message] + chat_history + [current_human_message]
+
+    # Invoke the LLM with the combined messages
+    ai_response = llm.invoke(messages_for_llm)
+    response_content = ai_response.content
+
+    generated_image = None
+    full_response = "" # Initialize full_response
+    audio_output_path = None # Initialize audio output path
+
+    # Implement logic to determine if an image generation is implied
+    IMAGE_PROMPT_PREFIX = "IMAGE_PROMPT: "
+    if response_content.startswith(IMAGE_PROMPT_PREFIX):
+        # Extract the actual image description
+        image_description = response_content[len(IMAGE_PROMPT_PREFIX):].strip()
+        print(f"AI detected an image request. Generating image for: '{image_description}'")
+        try:
+            # Use the pipeline to generate an image
+            generated_image = pipeline(image_description).images[0]
+            full_response = f"Image generated successfully based on your request: '{image_description}'"
+        except Exception as e:
+            full_response = f"Could not generate image for '{image_description}'. Error: {e}"
+            generated_image = None
+    else:
+        # Otherwise, it's a normal conversational response
+        full_response = response_content
+
+    # Generate speech for the full_response
+    try:
+        tts = gTTS(text=full_response, lang='es') # Assuming Spanish for the voice output
+        audio_file_name = "ai_response.mp3"
+        tts.save(audio_file_name)
+        audio_output_path = audio_file_name
+        print(f"Audio generated and saved to {audio_output_path}")
+    except Exception as e:
+        print(f"Error generating audio: {e}")
+        audio_output_path = None
+
+    # Update chat history with the current user message and the AI's raw response
+    chat_history.append(current_human_message)
+    chat_history.append(AIMessage(content=ai_response.content))
+
+    # Return the AI's full textual response, the generated image object, the audio path, and the updated chat history
+    return full_response, generated_image, audio_output_path, chat_history
+
+print("handle_user_input function updated to include chat history and speech generation.")
+
+from langchain_core.prompts import MessagesPlaceholder
+
+# Re-define the system message template (it was already defined but ensures it's current)
+system_message_template = SystemMessagePromptTemplate.from_template(
+    "You are a helpful AI assistant that can generate creative images based on user descriptions. "
+    "You can also answer questions and engage in conversation. "
+    "If a user asks for an image, extract the key elements for a Stable Diffusion prompt and output it in the format 'IMAGE_PROMPT: <your image prompt here>'. "
+    "Otherwise, provide a normal conversational response."
+)
+
+# Re-define the human message template
+human_message_template = HumanMessagePromptTemplate.from_template("User: {user_input}")
+
+# Re-define the chat_prompt_template to include MessagesPlaceholder for chat history
+chat_prompt_template = ChatPromptTemplate.from_messages([
+    system_message_template,
+    MessagesPlaceholder(variable_name="chat_history"),
+    human_message_template
+])
+
+print("ChatPromptTemplate updated successfully for history management.")
+
+def gradio_interface_fn(user_query: str, history: list):
+    ai_text_response, generated_image_output, audio_output_path, updated_chat_history = handle_user_input(user_query, history)
+    print(f"DEBUG: Type of generated_image_output: {type(generated_image_output)}")
+    print(f"DEBUG: Is generated_image_output None?: {generated_image_output is None}")
+    if generated_image_output is not None:
+        print(f"DEBUG: Generated image mode: {generated_image_output.mode}, size: {generated_image_output.size}")
+    return ai_text_response, generated_image_output, audio_output_path, updated_chat_history
+
+print("Gradio interface function 'gradio_interface_fn' updated to manage history and audio.")
+
+get_ipython().system('pip install bcrypt')
+print("bcrypt library installed successfully.")
+
+# Commented out IPython magic to ensure Python compatibility.
+# %%writefile app.py
+# import gradio as gr
+# import torch
+# from diffusers import StableDiffusionPipeline
+# from langchain_cohere import ChatCohere
+# from langchain_core.prompts import ChatPromptTemplate, HumanMessagePromptTemplate, SystemMessagePromptTemplate, MessagesPlaceholder
+# from langchain_core.messages import HumanMessage, AIMessage, SystemMessage
+# from gtts import gTTS
+# import os
+# import base64
+# import json
+# import bcrypt # Import bcrypt
+# 
+# # --- Configuration and Initialization (from your notebook) ---
+# COHERE_API_KEY = os.environ.get('COHERE_API_KEY')
+# 
+# # --- User Management Functions (from previous steps) ---
+# USERS_FILE = 'users.json'
+# CHAT_HISTORY_FILE_PATTERN = 'chat_history_{}.json' # Pattern for user-specific chat history files
+# 
+# def load_users():
+#     """Loads user data from the USERS_FILE. Returns an empty dictionary if the file doesn't exist."""
+#     if os.path.exists(USERS_FILE):
+#         with open(USERS_FILE, 'r') as f:
+#             return json.load(f)
+#     return {}
+# 
+# def save_users(users):
+#     """Saves user data to the USERS_FILE."""
+#     with open(USERS_FILE, 'w') as f:
+#         json.dump(users, f, indent=4)
+# 
+# def hash_password(password):
+#     """
+#     Hashes a password using bcrypt.
+#     The password is encoded to bytes, a salt is generated, and then the password is hashed.
+#     Returns the hashed password as a UTF-8 decoded string.
+#     """
+#     hashed = bcrypt.hashpw(password.encode('utf-8'), bcrypt.gensalt())
+#     return hashed.decode('utf-8')
+# 
+# def register_user(username, password):
+#     """
+#     Registers a new user with the given username and password.
+#     Hashes the password and saves it to users.json.
+#     Handles cases where the username already exists.
+#     """
+#     users = load_users()
+#     if username in users:
+#         return "Error: Username already exists."
+# 
+#     hashed_password = hash_password(password)
+#     users[username] = hashed_password
+#     save_users(users)
+#     # Initialize an empty chat history for the new user
+#     save_chat_history(username, [])
+#     return f"User '{username}' registered successfully."
+# 
+# def authenticate_user(username, password):
+#     """
+#     Authenticates a user against the stored hashed passwords in users.json.
+#     Returns True if authentication is successful, False otherwise.
+#     """
+#     users = load_users()
+#     stored_hashed_password = users.get(username)
+# 
+#     if stored_hashed_password:
+#         try:
+#             return bcrypt.checkpw(password.encode('utf-8'), stored_hashed_password.encode('utf-8'))
+#         except ValueError:
+#             return False
+#     return False
+# 
+# def load_chat_history(username):
+#     """
+#     Loads a user's chat history from their specific JSON file.
+#     Converts stored dicts back into Langchain HumanMessage/AIMessage objects.
+#     """
+#     history_file = CHAT_HISTORY_FILE_PATTERN.format(username)
+#     if os.path.exists(history_file):
+#         with open(history_file, 'r') as f:
+#             raw_history = json.load(f)
+#         chat_history = []
+#         for msg in raw_history:
+#             if msg['type'] == 'human':
+#                 chat_history.append(HumanMessage(content=msg['content']))
+#             elif msg['type'] == 'ai':
+#                 chat_history.append(AIMessage(content=msg['content']))
+#             elif msg['type'] == 'system':
+#                 chat_history.append(SystemMessage(content=msg['content']))
+#         return chat_history
+#     return []
+# 
+# def save_chat_history(username, chat_history):
+#     """
+#     Saves a user's chat history (Langchain Message objects) to their specific JSON file.
+#     Converts Langchain Message objects to dictionaries for JSON serialization.
+#     """
+#     history_file = CHAT_HISTORY_FILE_PATTERN.format(username)
+#     raw_history = []
+#     for msg in chat_history:
+#         raw_history.append({'type': msg.type, 'content': msg.content})
+#     with open(history_file, 'w') as f:
+#         json.dump(raw_history, f, indent=4)
+# 
+# def convert_langchain_to_gradio_chatbot_history(langchain_history):
+#     """
+#     Converts a list of Langchain Message objects to Gradio Chatbot format.
+#     Gradio Chatbot expects a list of lists: [[user_message, ai_message], ...]
+#     """
+#     gradio_history = []
+#     # Ensure we iterate in pairs (human, AI)
+#     for i in range(0, len(langchain_history), 2):
+#         human_msg = langchain_history[i].content if i < len(langchain_history) else ""
+#         # Check if there's a corresponding AI message
+#         ai_msg = langchain_history[i+1].content if i+1 < len(langchain_history) else ""
+#         gradio_history.append([human_msg, ai_msg])
+#     return gradio_history
+# 
+# # Initialize users.json if it doesn't exist
+# if not os.path.exists(USERS_FILE):
+#     save_users({}) # Save an empty dictionary as a valid JSON
+# 
+# 
+# # --- Stable Diffusion Model Initialization ---
+# model_id = "runwayml/stable-diffusion-v1-5"
+# device = "cuda" if torch.cuda.is_available() else "cpu"
+# pipeline = StableDiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.float16)
+# pipeline.to(device)
+# 
+# # --- Cohere LLM Initialization ---
+# llm = ChatCohere(cohere_api_key=COHERE_API_KEY)
+# 
+# # --- Prompt Templates ---
+# system_message_template = SystemMessagePromptTemplate.from_template(
+#     "You are a helpful AI assistant that can generate creative images based on user descriptions. "
+#     "You can also answer questions and engage in conversation. "
+#     "If a user asks for an image, extract the key elements for a Stable Diffusion prompt and output it in the format 'IMAGE_PROMPT: <your image prompt here>'. "
+#     "Otherwise, provide a normal conversational response."
+# )
+# human_message_template = HumanMessagePromptTemplate.from_template("User: {user_input}")
+# chat_prompt_template = ChatPromptTemplate.from_messages([
+#     system_message_template,
+#     MessagesPlaceholder(variable_name="chat_history"),
+#     human_message_template
+# ])
+# 
+# # --- Core Logic Functions ---
+# def handle_user_input(user_message: str, chat_history: list, username: str):
+#     current_human_message = HumanMessage(content=user_message)
+#     system_message_content = system_message_template.format().content
+#     system_message = SystemMessage(content=system_message_content)
+#     messages_for_llm = [system_message] + chat_history + [current_human_message]
+# 
+#     ai_response = llm.invoke(messages_for_llm)
+#     response_content = ai_response.content
+# 
+#     generated_image = None
+#     full_response = ""
+#     audio_output_path = None
+# 
+#     IMAGE_PROMPT_PREFIX = "IMAGE_PROMPT: "
+#     if response_content.startswith(IMAGE_PROMPT_PREFIX):
+#         image_description = response_content[len(IMAGE_PROMPT_PREFIX):].strip()
+#         try:
+#             generated_image = pipeline(image_description).images[0]
+#             full_response = f"Image generated successfully based on your request: '{image_description}'"
+#         except Exception as e:
+#             full_response = f"Could not generate image for '{image_description}'. Error: {e}"
+#             generated_image = None
+#     else:
+#         full_response = response_content
+# 
+#     try:
+#         tts = gTTS(text=full_response, lang='es')
+#         audio_file_name = f"ai_response_{username}.mp3" if username else "ai_response_guest.mp3"
+#         tts.save(audio_file_name)
+#         audio_output_path = audio_file_name
+#     except Exception as e:
+#         audio_output_path = None
+# 
+#     chat_history.append(current_human_message)
+#     chat_history.append(AIMessage(content=ai_response.content))
+# 
+#     if username:
+#         save_chat_history(username, chat_history)
+# 
+#     return full_response, generated_image, audio_output_path, chat_history
+# 
+# 
+# def gradio_interface_fn(user_query: str, history: list, username: str):
+#     if not user_query: # Handle empty input case
+#         return "Por favor, introduce un mensaje.", None, None, history, convert_langchain_to_gradio_chatbot_history(history)
+# 
+#     ai_text_response, generated_image_output, audio_output_path, updated_chat_history_langchain = handle_user_input(user_query, history, username)
+#     gradio_chat_history_display = convert_langchain_to_gradio_chatbot_history(updated_chat_history_langchain)
+#     return ai_text_response, generated_image_output, audio_output_path, updated_chat_history_langchain, gradio_chat_history_display
+# 
+# # --- Gradio Registration Function ---
+# def register_gradio_fn(username, password):
+#     return register_user(username, password)
+# 
+# # --- Gradio Login Function ---
+# def login_gradio_fn(username, password):
+#     if authenticate_user(username, password):
+#         user_history_langchain = load_chat_history(username)
+#         gradio_chat_history_display = convert_langchain_to_gradio_chatbot_history(user_history_langchain)
+#         return (
+#             f"¡Bienvenido, {username}! Has iniciado sesión con éxito.",
+#             username,
+#             user_history_langchain,
+#             gradio_chat_history_display
+#         )
+#     else:
+#         return (
+#             "Error: Nombre de usuario o contraseña incorrectos.",
+#             '',
+#             [],
+#             []
+#         )
+# 
+# # --- Gradio Logout Function ---
+# def logout_gradio_fn():
+#     return (
+#         "Has cerrado sesión.",
+#         '',
+#         [],
+#         []
+#     )
+# 
+# # --- Gradio UI ---
+# image_filename = "Chaty (1).png"
+# base64_image_tag = ""
+# favicon_data_uri = ""
+# 
+# if os.path.exists(image_filename):
+#     try:
+#         with open(image_filename, "rb") as image_file:
+#             encoded_string = base64.b64encode(image_file.read()).decode('utf-8')
+#         base64_image_tag = f"<img src='data:image/png;base64,{encoded_string}' alt='Chaty Logo' style='height:100px;'>"
+#         favicon_data_uri = f"data:image/png;base64,{encoded_string}"
+#     except Exception as e:
+#         pass # Handle error silently in UI context
+# 
+# head_content = ""
+# if favicon_data_uri:
+#     head_content = f"<link rel='icon' type='image/png' href='{favicon_data_uri}'>"
+# 
+# with gr.Blocks(
+#     title='Chaty',
+#     theme=gr.themes.Soft(),
+#     head=head_content
+# ) as demo:
+#     logged_in_user = gr.State('') # Initialize logged_in_user state
+#     chatbot_history_state = gr.State([]) # State for Langchain messages
+# 
+#     gr.Markdown(f"{base64_image_tag} Un asistente de IA interactivo que utiliza Stable Diffusion para la generaci\u00f3n de im\u00e1genes y Cohere/Langchain para la comprensi\u00f3n del lenguaje y la generaci\u00f3n de respuestas. Si pides una imagen, la generar\u00e1; de lo contrario, responder\u00e1 conversacionalmente. **\u00a1Ahora con respuestas de voz!**")
+# 
+#     # Define chatbot UI elements at a higher scope
+#     chatbot_ui_elements_column = gr.Column(visible=True)
+#     with chatbot_ui_elements_column:
+#         chatbot_display_chatbot = gr.Chatbot(label='Chaty', value=[], elem_id='chatbot', height=400)
+#         chatbot_input_textbox = gr.Textbox(lines=2, label='Escribe tu mensaje o petición de imagen aquí...', interactive=True)
+#         chatbot_submit_button = gr.Button("Enviar", interactive=True)
+#         ai_response_textbox = gr.Textbox(lines=5, label='Respuesta de la IA', interactive=False)
+#         image_output = gr.Image(label='Imagen Generada')
+#         audio_output = gr.Audio(label='Respuesta de Voz', autoplay=True)
+# 
+#     with gr.Tab("Inicio de Sesión"):
+#         gr.Markdown("## Iniciar Sesión")
+#         with gr.Column():
+#             login_username = gr.Textbox(label="Nombre de Usuario")
+#             login_password = gr.Textbox(label="Contraseña", type="password")
+#             login_button = gr.Button("Iniciar Sesión")
+#             login_output = gr.Textbox(label="Mensaje de Inicio de Sesión", interactive=False)
+#             current_logged_in_user_display = gr.Textbox(label="Usuario Actual", interactive=False, value="No has iniciado sesión.")
+#             logout_button = gr.Button("Cerrar Sesión")
+# 
+#         login_button.click(
+#             fn=login_gradio_fn,
+#             inputs=[login_username, login_password],
+#             outputs=[
+#                 login_output,
+#                 logged_in_user,
+#                 chatbot_history_state,
+#                 chatbot_display_chatbot
+#             ]
+#         )
+# 
+#         logout_button.click(
+#             fn=logout_gradio_fn,
+#             inputs=[],
+#             outputs=[
+#                 login_output,
+#                 logged_in_user,
+#                 chatbot_history_state,
+#                 chatbot_display_chatbot
+#             ]
+#         )
+# 
+#         logged_in_user.change(
+#             lambda user: f"Has iniciado sesión como: {user}" if user else "No has iniciado sesión.",
+#             inputs=logged_in_user,
+#             outputs=current_logged_in_user_display
+#         )
+# 
+#     with gr.Tab("Chatbot"):
+#         chatbot_submit_button.click(
+#             fn=gradio_interface_fn,
+#             inputs=[chatbot_input_textbox, chatbot_history_state, logged_in_user], # Removed audio_input
+#             outputs=[
+#                 ai_response_textbox,
+#                 image_output,
+#                 audio_output,
+#                 chatbot_history_state,
+#                 chatbot_display_chatbot
+#             ]
+#         ).then(
+#             lambda: "", # Clear the input textbox after submission
+#             inputs=None,
+#             outputs=chatbot_input_textbox
+#         )
+# 
+#     with gr.Tab("Registro de Usuario"):
+#         gr.Markdown("## Registrar Nuevo Usuario")
+#         with gr.Column():
+#             register_username = gr.Textbox(label="Nombre de Usuario")
+#             register_password = gr.Textbox(label="Contraseña", type="password")
+#             register_button = gr.Button("Registrar")
+#             register_output = gr.Textbox(label="Mensaje de Registro", interactive=False)
+# 
+#         register_button.click(
+#             fn=register_gradio_fn,
+#             inputs=[register_username, register_password],
+#             outputs=register_output
+#         )
+# 
+# demo.launch(debug=True, share=True)
+
+!pip install langchain-cohere
+get_ipython().system('python app.py')