import gradio as gr
import requests
from PIL import Image
import io

# URL de tu API
# Si estás ejecutando esto en local, suele ser http://127.0.0.1:8000
# Si la API está en Render, usa la URL de Render (ej: https://tuproyecto.onrender.com)
API_URL = "https://predictor-api-lab3.onrender.com"

def solicitar_prediccion(image_path):
    """
    Envía la imagen al endpoint /predict
    """
    if image_path is None:
        return "You must upload an image first."
    
    try:
        # Abrimos la imagen en modo binario para enviarla
        with open(image_path, "rb") as f:
            files = {"file": f}
            response = requests.post(f"{API_URL}/predict", files=files, timeout=10)
            
        response.raise_for_status()
        data = response.json()
        
        # Devolvemos la predicción
        return f"Prediction: {data.get('prediction')}"
        
    except requests.exceptions.RequestException as e:
        return f"Error connecting to the API: {str(e)}"
    except Exception as e:
        return f"Unknown error: {str(e)}"

def solicitar_resize(image_path, width, height):
    """
    Envía la imagen y dimensiones al endpoint /resize
    """
    if image_path is None:
        return None
    
    try:
        # Validar inputs
        if width <= 0 or height <= 0:
            print("Width and height must be positive.")
            return None

        payload = {"width": int(width), "height": int(height)}
        
        with open(image_path, "rb") as f:
            files = {"file": f}
            # Nota: 'data' se usa para los campos del Form (width, height)
            # y 'files' para el archivo
            response = requests.post(f"{API_URL}/resize", data=payload, files=files, timeout=10)
            
        response.raise_for_status()
        
        # La API devuelve una imagen en bytes (StreamingResponse)
        # La convertimos a objeto PIL Image para que Gradio la pueda mostrar
        image_stream = io.BytesIO(response.content)
        return Image.open(image_stream)
        
    except requests.exceptions.RequestException as e:
        print(f"Error API: {e}")
        return None

# --- Construcción de la Interfaz con Blocks ---
with gr.Blocks(title="Predictor & Resizer API Client") as demo:
    gr.Markdown("# Image API Client")
    gr.Markdown("Upload an image and choose whether you want to get a prediction or resize it.")

    with gr.Row():
        # Left Column: Input
        with gr.Column():
            gr.Markdown("### 1. Input Image")
            # Selector de imágenes. 'type="filepath"' guarda la imagen temporalmente y nos da la ruta
            input_image = gr.Image(label="Upload your image", type="filepath")

        # Right Column: Actions
        with gr.Column():
            
            # --- Prediction Section ---
            gr.Markdown("### 2. Prediction")
            predict_btn = gr.Button("🔍 Get Prediction", variant="primary")
            predict_output = gr.Textbox(label="API Result")

            # CORRECCIÓN AQUÍ: gr.HTML en mayúsculas
            gr.HTML("<hr>") 

            # --- Resize Section ---
            gr.Markdown("### 3. Resize Image")
            with gr.Row():
                w_input = gr.Number(label="Width", value=200, precision=0)
                h_input = gr.Number(label="Height", value=200, precision=0)
            
            resize_btn = gr.Button("🖼️ Resize Image")
            resize_output = gr.Image(label="Resized Image")

    # --- Connect Logic ---
    
    # Prediction Button
    predict_btn.click(
        fn=solicitar_prediccion,
        inputs=[input_image],
        outputs=predict_output
    )

    # Botón Resize
    resize_btn.click(
        fn=solicitar_resize,
        inputs=[input_image, w_input, h_input],
        outputs=resize_output
    )

# Launch the app
if __name__ == "__main__":
    demo.launch()