app.py

import gradio as gr
import cv2
import os
import zipfile
from PIL import Image, ImageOps
from datetime import datetime
import hashlib

def procesar_video(video):
    try:
        if isinstance(video, dict):
            original_name = video.get("name", "video")
            video_path = video.get("file", video.get("data"))
        else:
            original_name = os.path.basename(video)
            video_path = video

        allowed_extensions = ('.mp4', '.avi', '.mov', '.mkv')
        if not original_name.lower().endswith(allowed_extensions):
            raise gr.Error("Solo se permiten archivos de video (mp4, avi, mov, mkv)")

        timestamp = datetime.now().strftime('%Y-%m-%d %H:%M:%S')
        temp_dir = f"temp_{datetime.now().strftime('%Y%m%d%H%M%S')}"
        os.makedirs(temp_dir, exist_ok=True)

        # Extracción de todos los fotogramas
        cap = cv2.VideoCapture(video_path)
        frame_count = 0
        frame_paths = []
        while True:
            ret, frame = cap.read()
            if not ret:
                break
            frame_rgb = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
            img = Image.fromarray(frame_rgb)
            img_path = os.path.join(temp_dir, f"frame_{frame_count:04d}.jpg")
            img.save(img_path)
            frame_paths.append(img_path)
            frame_count += 1
        cap.release()

        if frame_count == 0:
            raise gr.Error("No se pudieron extraer fotogramas del video")

        # Selección estratégica de 4 fotogramas equidistantes
        n_seleccion = 4
        step = max(1, frame_count // (n_seleccion + 1))
        selected_indices = [step * (i+1) for i in range(n_seleccion)]
        selected_frames = [frame_paths[min(i, len(frame_paths)-1)] for i in selected_indices]

        # Creación de collage profesional
        images = []
        for img_path in selected_frames:
            img = Image.open(img_path)
            bordered_img = ImageOps.expand(img, border=2, fill='white')  # Borde blanco
            images.append(bordered_img)

        # Configuración del diseño
        img_w, img_h = images[0].size
        margin = 30
        border_size = 20
        shadow_offset = 5
        
        collage_width = (img_w * 2) + margin + (border_size * 2)
        collage_height = (img_h * 2) + margin + (border_size * 2)
        
        collage = Image.new('RGB', 
            (collage_width, collage_height), 
            (230, 230, 230))  # Fondo gris claro
        
        # Posiciones con efecto de profundidad
        positions = [
            (border_size, border_size),
            (border_size + img_w + margin, border_size),
            (border_size, border_size + img_h + margin),
            (border_size + img_w + margin, border_size + img_h + margin)
        ]
        
        # Pegar imágenes con sombra
        for i, img in enumerate(images):
            # Sombra
            shadow = Image.new('RGBA', (img_w + shadow_offset, img_h + shadow_offset), (0,0,0,50))
            collage.paste(shadow, (positions[i][0]+shadow_offset, positions[i][1]+shadow_offset), shadow)
            
            # Imagen principal
            collage.paste(img, positions[i])

        collage_path = os.path.join(temp_dir, "collage_forense.jpg")
        collage.save(collage_path, quality=95, dpi=(300, 300))

        # Generación del ZIP con cadena de custodia
        base_name = os.path.splitext(original_name)[0]
        zip_filename = f"{base_name}_Fotogramas.zip"
        final_zip_path = os.path.join(temp_dir, zip_filename)
        
        with zipfile.ZipFile(final_zip_path, mode="w") as zipf:
            # Añadir todos los frames
            for img_path in frame_paths:
                zipf.write(img_path, os.path.basename(img_path))
            
            # Archivo TXT con formato profesional
            with open(video_path, "rb") as f:
                video_hash = hashlib.md5(f.read()).hexdigest()
            
            chain_content = (
                "=== CADENA DE CUSTODIA DIGITAL ===\r\n\r\n"
                f"• Archivo original: {original_name}\r\n"
                f"• Fecha de procesamiento: {timestamp}\r\n"
                f"• Fotogramas totales: {frame_count}\r\n"
                f"• Hash MD5 video: {video_hash}\r\n"
                f"• Fotogramas muestra: {', '.join([f'#{i+1}' for i in selected_indices])}\r\n\r\n"
                "Este documento certifica la integridad del proceso de extracción.\n"
                "Sistema Certificado por Peritos Forenses Digitales de Guatemala. \n"
                "www.forensedigital.gt"
            )
            zipf.writestr("00_CADENA_CUSTODIA.txt", chain_content)

        return collage_path, final_zip_path, temp_dir

    except Exception as e:
        raise gr.Error(f"Error en procesamiento: {str(e)}")

def limpiar_cache(temp_dir):
    if temp_dir and os.path.exists(temp_dir):
        for file in os.listdir(temp_dir):
            os.remove(os.path.join(temp_dir, file))
        os.rmdir(temp_dir)

with gr.Blocks(title="Extractior Forense de Fotogramas") as demo:
    gr.Markdown("# 📷 Extractor Forense de Fotogramas de Videos")
    gr.Markdown("**Herramienta certificada para extracción forense de fotogramas de videos**")
    gr.Markdown("Desarrollado por José R. Leonett para el Grupo de Peritos Forenses Digitales de Guatemala - [www.forensedigital.gt](https://www.forensedigital.gt)")
    
    with gr.Row():
        with gr.Column():
            video_input = gr.Video(
                label="CARGAR VIDEO",
                sources=["upload"],
                format="mp4",
                interactive=True
            )
            procesar_btn = gr.Button("🔍 INICIAR ANÁLISIS", interactive=False)
        with gr.Column():
            gr.Markdown("## Resultados:")
            gallery_output = gr.Image(label="COLLAGE DE REFERENCIA", height=400)
            download_file = gr.File(label="DESCARGAR EVIDENCIAS", visible=True)
    
    temp_dir_state = gr.State()
    zip_path_state = gr.State()
    
    def habilitar_procesado(video):
        return gr.update(interactive=True) if video else gr.update(interactive=False)
    
    video_input.change(
        fn=habilitar_procesado,
        inputs=video_input,
        outputs=procesar_btn,
        queue=False
    )
    
    def procesar_y_mostrar(video):
        if temp_dir_state.value:
            limpiar_cache(temp_dir_state.value)
        collage, zip_path, temp_dir = procesar_video(video)
        return collage, zip_path, temp_dir, zip_path
    
    procesar_btn.click(
        fn=procesar_y_mostrar,
        inputs=video_input,
        outputs=[gallery_output, download_file, temp_dir_state, zip_path_state]
    )

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