app.py

import gradio as gr
import requests
from PIL import Image
from io import BytesIO
from transformers import pipeline
import torch
import soundfile as sf

# Pipelines de Transformers
image_to_text = pipeline("image-to-text", model="Salesforce/blip-image-captioning-large")
synthesiser = pipeline("text-to-speech", model="microsoft/speecht5_tts")

# Speaker embedding aleatorio (para demo)
speaker_embedding = torch.randn(1, 512)

# Obtener lista de cócteles desde la API
url = "https://www.thecocktaildb.com/api/json/v1/1/search.php?s=margarita"
response = requests.get(url)
lista = []

if response.status_code == 200:
    datos = response.json()
    drinks = datos.get("drinks", [])
    for drink in drinks:
        lista.append(drink['strDrink'])
else:
    print(f"Error: {response.status_code}")

# Función principal que actualiza la interfaz
def change_textbox(choice):
    cocktail = requests.get(f"https://www.thecocktaildb.com/api/json/v1/1/search.php?s={choice}")
    data = cocktail.json()
    dataCocktail = data.get("drinks", [])

    name = ""
    instructions = ""
    image_url = ""

    for i in dataCocktail:
        if i['strDrink'].lower() == choice.lower():
            name = i['strDrink']
            instructions = i['strInstructions']
            image_url = i['strDrinkThumb']
            break

    # Cargar imagen
    img_response = requests.get(image_url)
    image = Image.open(BytesIO(img_response.content)).convert("RGB")

    # Generar descripción de la imagen
    result = image_to_text(image)
    descripcion = result[0]['generated_text']

    # Generar audio de instrucciones
    speech = synthesiser(instructions, forward_params={"speaker_embeddings": speaker_embedding})
    sf.write("speech.wav", speech["audio"], samplerate=speech["sampling_rate"])
    audio_path = "speech.wav"

    # Generar audio de descripción de la imagen
    speech2 = synthesiser(descripcion, forward_params={"speaker_embeddings": speaker_embedding})
    sf.write("speech2.wav", speech2["audio"], samplerate=speech2["sampling_rate"])
    audio_path2 = "speech2.wav"

    return name, image, instructions, audio_path, descripcion, audio_path2

# Crear la interfaz de Gradio
with gr.Blocks() as demo:
    gr.HTML(
        """
        <style>
            body {
                background-color: #000000;
                color: #ffffff; 
                font-family: Arial, sans-serif;
                margin: 0;
                padding: 0;
                text-align: center; 
            }
            
            .gradio-container {
                background-color: #000000; 
                padding: 20px;
                border-radius: 10px;
                display: flex;
                flex-direction: column;
                align-items: center; 
                justify-content: center;
            }
           
            .gradio-container .gradio-radio {
                display: inline-block;
                margin: 10px;
                text-align: center;
            }
            h1 {
                text-align: center;
                color: #ffffff !important; 
            }
        </style>
        """
    )

    gr.Markdown("<h1>Cocktails Descriptions</h1>")
    
    radio = gr.Radio(lista, label="Choose your cocktail:")
    text = gr.Textbox(lines=1, interactive=False, show_copy_button=True, label="Cocktail Name")
    imagen = gr.Image(label="Cocktail Image")
    text2 = gr.Textbox(lines=4, interactive=False, show_copy_button=True, label="Instructions")
    audio = gr.Audio(label="Cocktail Instructions Audio")
    text3 = gr.Textbox(lines=2, interactive=False, show_copy_button=True, label="Image description")
    audio2 = gr.Audio(label="Audio image description")

    radio.change(
        fn=change_textbox, 
        inputs=radio, 
        outputs=[text, imagen, text2, audio, text3, audio2]
    )

demo.launch(share=True)