app.py

import gradio as gr
from tts_module import get_voices, text_to_speech
from pexels_api import search_pexels
from moviepy.editor import (
    AudioFileClip, VideoFileClip, CompositeAudioClip,
    concatenate_audioclips, concatenate_videoclips, vfx, CompositeVideoClip
)
import asyncio
import os
import time
import requests
from googleapiclient.discovery import build
from googleapiclient.http import MediaFileUpload
import tempfile
import re
import random

# Define la carpeta de salida temporal
output_folder = "outputs"
os.makedirs(output_folder, exist_ok=True)

def clean_text_for_search(text):
    """Limpia el texto para hacer búsquedas válidas en Pexels"""
    # Eliminar caracteres especiales y limitar longitud
    text = re.sub(r'[^\w\s]', '', text).strip()
    return text

def resize_and_blur_video(clip, target_width=1920, target_height=1080):
    """
    Redimensiona y aplica desenfoque al video para mantener aspecto 16:9 con resolución objetivo.
    Los videos más pequeños se redimensionan y los verticales se convierten en horizontales con blur.
    
    Args:
        clip: VideoFileClip object
        target_width: Ancho objetivo (default 1920 para 1080p)
        target_height: Alto objetivo (default 1080 para 1080p)
    """
    try:
        w, h = clip.size
        current_ratio = w / h
        target_ratio = target_width / target_height
        
        print(f"Video original: {w}x{h}, ratio: {current_ratio}")
        
        if current_ratio < target_ratio:  # Video vertical o más cuadrado que 16:9
            # Crear un fondo desenfocado escalado
            background = (clip
                .resize(width=target_width)
                .resize(width=target_width * 2)  # Hacer el blur más suave
                .fx(vfx.blur, sigma=10)
                .resize(width=target_width))
            
            # Calcular el tamaño para el video principal
            new_height = target_height
            new_width = int(h * current_ratio)
            if new_width > target_width:
                new_width = target_width
                new_height = int(new_width / current_ratio)
            
            # Redimensionar video principal
            foreground = clip.resize(width=new_width, height=new_height)
            
            # Centrar el video
            x_center = (target_width - new_width) // 2
            y_center = (target_height - new_height) // 2
            
            final = CompositeVideoClip(
                [background,
                 foreground.set_position((x_center, y_center))],
                size=(target_width, target_height)
            )
            
            return final
            
        else:  # Video horizontal
            # Si es más ancho que 16:9, recortamos los bordes
            return clip.resize(width=target_width, height=target_height)
            
    except Exception as e:
        print(f"Error en resize_and_blur_video: {e}")
        return clip.resize(width=target_width, height=target_height)

    except Exception as e:
        print(f"Error en resize_and_blur_video: {e}")
        return clip

def concatenate_pexels_videos(keywords, num_videos_per_keyword=1, target_width=1920, target_height=1080):
    """
    Concatena videos de Pexels manteniendo una calidad y resolución consistentes.
    
    Args:
        keywords (str): Palabras clave separadas por comas
        num_videos_per_keyword (int): Número de videos por palabra clave
        target_width (int): Ancho objetivo para los videos
        target_height (int): Alto objetivo para los videos
    """
    keyword_list = [keyword.strip() for keyword in keywords.split(",") if keyword.strip()]
    if not keyword_list:
        raise ValueError("No se proporcionaron palabras clave válidas.")

    video_clips = []
    processed_keywords = []

    for keyword in keyword_list:
        try:
            print(f"Buscando videos para: '{keyword}'...")
            # Limpiar la palabra clave para búsqueda
            clean_keyword = clean_text_for_search(keyword)
            
            # Obtener videos con la nueva API
            links = search_pexels(clean_keyword, num_results=num_videos_per_keyword)
            if not links:
                print(f"No se encontraron videos para: '{keyword}'")
                continue

            for link in links:
                try:
                    print(f"Descargando video para: '{keyword}'...")
                    with tempfile.NamedTemporaryFile(delete=False, suffix=".mp4") as tmp_video:
                        video_response = requests.get(link, stream=True)
                        if video_response.status_code != 200:
                            print(f"Error descargando video: {video_response.status_code}")
                            continue
                            
                        # Guardar el video
                        for chunk in video_response.iter_content(chunk_size=8192):
                            if chunk:
                                tmp_video.write(chunk)
                        
                        tmp_video.flush()
                        
                        # Procesar el video
                        try:
                            clip = VideoFileClip(tmp_video.name)
                            
                            # Verificar duración mínima
                            if clip.duration < 3:
                                print(f"Video demasiado corto ({clip.duration}s), saltando...")
                                clip.close()
                                continue
                                
                            # Procesar y agregar el clip
                            processed_clip = resize_and_blur_video(clip, target_width, target_height)
                            if processed_clip:
                                video_clips.append(processed_clip)
                                processed_keywords.append(keyword)
                                print(f"Video procesado exitosamente para: '{keyword}'")
                            
                        except Exception as e:
                            print(f"Error procesando video: {e}")
                            if 'clip' in locals():
                                clip.close()
                            continue
                            
                finally:
                    # Limpiar archivo temporal
                    if os.path.exists(tmp_video.name):
                        try:
                            os.unlink(tmp_video.name)
                        except Exception as e:
                            print(f"Error eliminando archivo temporal: {e}")

        except Exception as e:
            print(f"Error procesando palabra clave '{keyword}': {e}")
            continue

    if not video_clips:
        raise Exception("No se pudieron obtener videos válidos para ninguna palabra clave.")

    print(f"Videos procesados exitosamente para las palabras: {', '.join(processed_keywords)}")
    
    # Aleatorizar el orden de los clips
    random.shuffle(video_clips)
    
    # Concatenar los clips
    try:
        final_clip = concatenate_videoclips(video_clips, method="compose")
        print(f"Video final generado: {final_clip.size}")
        return final_clip
    except Exception as e:
        raise Exception(f"Error concatenando clips: {e}")

def combine_audio_video(audio_file, video_clip, music_clip=None):
    try:
        audio_clip = AudioFileClip(audio_file)
        total_duration = audio_clip.duration + 5
        if video_clip.duration < total_duration:
            video_clip = video_clip.loop(duration=total_duration)
        video_clip = video_clip.set_duration(total_duration).fadeout(5)
        final_clip = video_clip.set_audio(audio_clip)

        if music_clip:
            if music_clip.duration < total_duration:
                repetitions = int(total_duration / music_clip.duration) + 1
                music_clips = [music_clip] * repetitions
                music_clip = concatenate_audioclips(music_clips)
            if music_clip.duration > total_duration:
                music_clip = music_clip.subclip(0, total_duration)
            music_clip = music_clip.audio_fadeout(5)
            final_clip = final_clip.set_audio(CompositeAudioClip([audio_clip, music_clip]))

        output_filename = f"final_video_{int(time.time())}.mp4"
        output_path = os.path.join(output_folder, output_filename)
        final_clip.write_videofile(output_path, codec="libx264", audio_codec="aac", fps=24)
        return output_path

    except Exception as e:
        print(f"Error combinando audio y video: {e}")
        return None

def process_input(text, txt_file, mp3_file, selected_voice, rate, pitch, keywords):
    try:
        # Validar entrada de texto
        if text.strip():
            final_text = text
        elif txt_file is not None:
            final_text = txt_file.decode("utf-8")
        else:
            # Retornar None en lugar de string de error
            return None

        # Validar voces
        voices = asyncio.run(get_voices())
        if selected_voice not in voices:
            return None

        # Generar audio
        try:
            audio_file = asyncio.run(text_to_speech(final_text, selected_voice, rate, pitch))
        except Exception as e:
            print(f"Error generando audio: {e}")
            return None

        # Procesar videos
        try:
            video_clip = concatenate_pexels_videos(keywords, num_videos_per_keyword=1)
        except Exception as e:
            print(f"Error concatenando videos: {e}")
            return None

        # Procesar música de fondo si existe
        if mp3_file is not None:
            music_clip = adjust_background_music(video_clip.duration, mp3_file.name)
        else:
            music_clip = None

        # Combinar audio y video
        final_video_path = combine_audio_video(audio_file, video_clip, music_clip)
        if final_video_path is None:
            return None

        # Subir a Google Drive
        upload_to_google_drive(final_video_path)

        # Verificar que el archivo existe antes de retornarlo
        if os.path.exists(final_video_path):
            return final_video_path
        else:
            return None

    except Exception as e:
        print(f"Error en process_input: {e}")
        return None

def upload_to_google_drive(file_path):
    try:
        api_key = os.getenv("GOOGLE_API_KEY")
        if not api_key:
            print("Error: GOOGLE_API_KEY no está definida en las variables de entorno.")
            return None

        service = build("drive", "v3", developerKey=api_key)
        file_metadata = {"name": os.path.basename(file_path)}
        media = MediaFileUpload(file_path, resumable=True)
        file = service.files().create(body=file_metadata, media_body=media, fields="id").execute()
        print(f"Archivo subido exitosamente con ID: {file.get('id')}")
        return file.get("id")
    except Exception as e:
        print(f"Error subiendo a Google Drive: {e}")
        return None

# Interfaz Gradio
with gr.Blocks() as demo:
    gr.Markdown("# Text-to-Video Generator")
    with gr.Row():
        with gr.Column():
            text_input = gr.Textbox(label="Write your text here", lines=5)
            txt_file_input = gr.File(label="Or upload a .txt file", file_types=[".txt"])
            mp3_file_input = gr.File(label="Upload background music (.mp3)", file_types=[".mp3"])
            keyword_input = gr.Textbox(label="Enter keywords separated by commas (e.g., universe, galaxy, forest, cat)")
            voices = asyncio.run(get_voices())
            voice_dropdown = gr.Dropdown(choices=list(voices.keys()), label="Select Voice")
            rate_slider = gr.Slider(minimum=-50, maximum=50, value=0, label="Speech Rate Adjustment (%)", step=1)
            pitch_slider = gr.Slider(minimum=-20, maximum=20, value=0, label="Pitch Adjustment (Hz)", step=1)
        with gr.Column():
            output_video = gr.File(label="Download Generated Video")

    btn = gr.Button("Generate Video")
     def show_error():
        return gr.Warning("Hubo un error procesando el video. Por favor, verifica los parámetros e intenta de nuevo.")
    btn.click(
        process_input,
        inputs=[text_input, txt_file_input, mp3_file_input, voice_dropdown, rate_slider, pitch_slider, keyword_input],
        outputs=output_video
    )

# Leer el puerto asignado por Hugging Face
port = int(os.getenv("PORT", 7860))

# Lanzar la aplicación
demo.launch(server_name="0.0.0.0", server_port=port, share=True)