Blame for ffmpeg.py

"""
Módulo de procesamiento multimedia con FFmpeg y sondeo con ffprobe.
Port 1 a 1 de internal/ffmpeg/ffmpeg.go con todas las validaciones y sanitizaciones.
"""

import os
import re
import math
import time
import subprocess
from typing import List, Tuple, Dict, Any, Optional, Callable

from config import logger

# Expresiones regulares idénticas a internal/ffmpeg/ffmpeg.go
TIME_REGEX = re.compile(r"time=(\d+):(\d+):(\d+\.\d+)")
SPEED_REGEX = re.compile(r"speed=\s*([\d\.]+)x")
PRESET_RES_REGEX = re.compile(r"^\d{2,4}p?$")
CUSTOM_RES_REGEX = re.compile(r"^\d{2,5}[x:]\d{2,5}$")
POS_NUM_REGEX = re.compile(r"^\d{1,3}(\.\d{1,2})?$")
BITRATE_REGEX = re.compile(r"^\d{1,4}k?$")

PAD_FILTER = "pad=ceil(iw/2)*2:ceil(ih/2)*2"

VALID_CODECS = {
    "libx264": True, "libx265": True, "h264_qsv": True, "hevc_qsv": True,
    "h264_nvenc": True, "hevc_nvenc": True, "h264_amf": True, "hevc_amf": True,
    "libvpx-vp9": True, "libaom-av1": True, "libsvtav1": True, "mpeg4": True, "copy": True,
}

VALID_PRESETS = {
    "ultrafast": True, "superfast": True, "veryfast": True, "faster": True,
    "fast": True, "medium": True, "slow": True, "slower": True, "veryslow": True,
}

SUPPORTED_SUBTITLE_CODECS = {
    "subrip": True, "srt": True, "ass": True, "ssa": True,
    "mov_text": True, "webvtt": True, "text": True,
}


def get_duration(file_path: str, timeout: float = 30.0) -> float:
    """Obtiene la duración en segundos del video usando ffprobe con timeout de 30s."""
    try:
        cmd = [
            "ffprobe",
            "-v", "error",
            "-show_entries", "format=duration",
            "-of", "default=noprint_wrappers=1:nokey=1",
            file_path,
        ]
        res = subprocess.run(cmd, capture_output=True, text=True, timeout=timeout)
        if res.returncode != 0:
            logger.warning(f"ffprobe retorno no-cero ({res.returncode}): {res.stderr}")
            return 0.0
        val_str = res.stdout.strip()
        return float(val_str)
    except Exception as e:
        logger.warning(f"Error ejecutando ffprobe para duración: {e}")
        return 0.0


def find_font_file() -> str:
    """Busca una fuente TTF válida en el sistema para drawtext."""
    candidates = [
        "/usr/share/fonts/dejavu/DejaVuSans-Bold.ttf",
        "/usr/share/fonts/dejavu/DejaVuSans.ttf",
        "/usr/share/fonts/TTF/DejaVuSans.ttf",
        "/usr/share/fonts/truetype/dejavu/DejaVuSans.ttf",
        "/usr/share/fonts/ttf-dejavu/DejaVuSans.ttf",
        "/usr/share/fonts/noto/NotoSans-Regular.ttf",
        "C:\\Windows\\Fonts\\arial.ttf",
    ]
    for c in candidates:
        if os.path.isfile(c):
            return c
    return ""


def escape_font_path(p: str) -> str:
    """Adapta una ruta de fuente al filtergraph: forward slashes y sin comillas."""
    return p.replace("\\", "/").replace("'", "")


def get_text_subtitle_streams(file_path: str, timeout: float = 30.0) -> List[int]:
    """Detecta las pistas de subtítulos de texto soportadas (srt, ass, ssa, mov_text, webvtt)."""
    try:
        cmd = [
            "ffprobe",
            "-v", "error",
            "-select_streams", "s",
            "-show_entries", "stream=index,codec_name",
            "-of", "csv=p=0",
            file_path,
        ]
        res = subprocess.run(cmd, capture_output=True, text=True, timeout=timeout)
        if res.returncode != 0:
            return []

        stream_indices = []
        for line in res.stdout.strip().splitlines():
            line = line.strip()
            if not line:
                continue
            parts = line.split(",")
            if len(parts) >= 2:
                try:
                    idx = int(parts[0].strip())
                    codec = parts[1].strip().lower()
                    if SUPPORTED_SUBTITLE_CODECS.get(codec, False):
                        stream_indices.append(idx)
                except ValueError:
                    continue
        return stream_indices
    except Exception:
        return []


def is_valid_resolution(res: str) -> bool:
    """Indica si la cadena tiene un formato de resolución aceptado."""
    res = (res or "").strip()
    if not res or res.lower() == "original":
        return True
    lower = res.lower()
    return bool(PRESET_RES_REGEX.match(lower) or CUSTOM_RES_REGEX.match(lower))


def build_scale_filter(res: str) -> str:
    """
    Genera el filtro de video adaptado inteligentemente a videos horizontales y verticales.
    Idéntico a BuildScaleFilter en internal/ffmpeg/ffmpeg.go.
    """
    res = (res or "").strip()
    if not res or res.lower() == "original":
        return PAD_FILTER

    lower = res.lower()
    if PRESET_RES_REGEX.match(lower):
        clean_num = lower.rstrip("p")
        try:
            h = int(clean_num)
            if 0 < h <= 4320:
                # Escala inteligente según orientación:
                # Horizontal (iw >= ih): ancho automático -2, altura H
                # Vertical / Shorts / Reels (iw < ih): ancho H, altura automática -2
                return f"scale='if(gte(iw,ih),-2,{h})':'if(gte(iw,ih),{h},-2)',{PAD_FILTER}"
        except ValueError:
            pass
        return PAD_FILTER

    if CUSTOM_RES_REGEX.match(lower):
        clean_res = lower.replace("x", ":")
        return f"scale={clean_res}:force_original_aspect_ratio=decrease,{PAD_FILTER}"

    return PAD_FILTER


def sanitize_watermark(text: str) -> str:
    """
    Deja el texto de la marca de agua en un conjunto de caracteres inofensivo
    para el filtergraph de FFmpeg (máximo 60 runas).
    """
    if not text:
        return ""
    allowed_chars = set(" -_.()!?¡¿#@&+=")
    cleaned = []
    for ch in text:
        if ch.isalnum() or ch in allowed_chars or ("\u00C0" <= ch <= "\u02AF"):
            cleaned.append(ch)
        else:
            cleaned.append(" ")
    res = "".join(cleaned).strip()
    return res[:60]


def sanitize_settings(s: Dict[str, Any]) -> Dict[str, Any]:
    """
    Sanitiza los ajustes del video contra sus dominios válidos.
    Defensa en profundidad idéntica a SanitizeSettings en Go.
    """
    sanitized = dict(s)
    sanitized["resolution"] = str(s.get("resolution", "")).strip()
    sanitized["watermark"] = sanitize_watermark(str(s.get("watermark", "")))

    fmt = str(s.get("format", "")).strip().lower()
    sanitized["format"] = "avi" if fmt == "avi" else "mp4"

    codec = str(s.get("codec", "")).strip().lower()
    sanitized["codec"] = codec if VALID_CODECS.get(codec, False) else ""

    preset = str(s.get("preset", "")).strip().lower()
    sanitized["preset"] = preset if VALID_PRESETS.get(preset, False) else ""

    crf = str(s.get("crf", "")).strip()
    if crf:
        if not POS_NUM_REGEX.match(crf):
            sanitized["crf"] = ""
        else:
            try:
                v = float(crf)
                sanitized["crf"] = crf if v <= 63 else ""
            except ValueError:
                sanitized["crf"] = ""
    else:
        sanitized["crf"] = ""

    fps = str(s.get("fps", "")).strip()
    if fps:
        if not POS_NUM_REGEX.match(fps):
            sanitized["fps"] = ""
        else:
            try:
                v = float(fps)
                sanitized["fps"] = fps if 1 <= v <= 120 else ""
            except ValueError:
                sanitized["fps"] = ""
    else:
        sanitized["fps"] = ""

    ab = str(s.get("audio_bitrate", "")).strip().lower()
    sanitized["audio_bitrate"] = ab if BITRATE_REGEX.match(ab) else ""

    return sanitized


def wm_text_position(pos: str) -> Tuple[str, str]:
    """Posición x/y de drawtext según la posición elegida."""
    p = (pos or "tr").strip().lower()
    if p == "tl":
        return "10", "10"
    elif p == "br":
        return "w-tw-10", "h-th-10"
    elif p == "bl":
        return "10", "h-th-10"
    return "w-tw-10", "10"


def wm_overlay_expr(pos: str) -> str:
    """Expresión de overlay para marca de agua en imagen."""
    p = (pos or "tr").strip().lower()
    if p == "tl":
        return "10:10"
    elif p == "br":
        return "W-w-10:H-h-10"
    elif p == "bl":
        return "10:H-h-10"
    return "W-w-10:10"


def generate_thumbnail(video_path: str, thumb_path: str, at_seconds: float = 1.0) -> bool:
    """Extrae un frame del video para usarlo como miniatura."""
    try:
        ss = f"{at_seconds:.2f}"
        cmd = ["ffmpeg", "-y", "-ss", ss, "-i", video_path, "-vframes", "1", "-q:v", "2", thumb_path]
        res = subprocess.run(cmd, capture_output=True, timeout=20)
        return res.returncode == 0
    except Exception:
        return False


def normalize_thumbnail_image(src_path: str, dst_path: str) -> bool:
    """Normaliza miniatura a JPEG 320x320 con recorte centrado."""
    try:
        vf = "scale=320:320:force_original_aspect_ratio=increase:flags=lanczos,crop=320:320"
        cmd = ["ffmpeg", "-y", "-i", src_path, "-vf", vf, "-frames:v", "1", "-q:v", "3", dst_path]
        res = subprocess.run(cmd, capture_output=True, timeout=20)
        return res.returncode == 0
    except Exception:
        return False


def is_subtitle_or_mux_error(err_str: str) -> bool:
    """Comprueba si el error de stderr de FFmpeg está relacionado con subtítulos corruptos o fallos de muxing."""
    lower = err_str.lower()
    indicators = [
        "application provided duration",
        "mov_text",
        "subtitle",
        "av_interleaved_write_frame",
        "invalid argument",
        "error while writing stream",
        "pts has no value",
        "out of range for mov/mp4",
        "invalid dts",
        "invalid pts",
        "muxing",
    ]
    return any(ind in lower for ind in indicators)


def build_compress_args(
    input_path: str,
    output_path: str,
    settings: Dict[str, Any],
    wm_image_path: str,
    include_subtitles: bool = True,
) -> Tuple[List[str], bool]:
    """Genera la lista de argumentos para FFmpeg, permitiendo habilitar o deshabilitar subtítulos."""
    settings = sanitize_settings(settings)

    args = ["-y", "-i", input_path]

    use_image_wm = bool(wm_image_path and os.path.isfile(wm_image_path))
    if use_image_wm:
        args.extend(["-i", wm_image_path])
        chain = (
            f"[0:v]{build_scale_filter(settings.get('resolution', ''))}[vbase];"
            f"[1:v]scale='if(gt(iw,150),150,iw)':-1[wm];"
            f"[vbase][wm]overlay={wm_overlay_expr(settings.get('watermark_pos', ''))}[v]"
        )
        args.extend(["-filter_complex", chain, "-map", "[v]", "-map", "0:a?"])
    else:
        vf_parts = [build_scale_filter(settings.get("resolution", ""))]
        wm_text = settings.get("watermark", "")
        if wm_text:
            x_pos, y_pos = wm_text_position(settings.get("watermark_pos", ""))
            font_path = find_font_file()
            if font_path:
                esc_path = escape_font_path(font_path)
                vf_parts.append(
                    f"drawtext=fontfile='{esc_path}':text='{wm_text}':x={x_pos}:y={y_pos}:fontsize=20:fontcolor=white:box=1:boxcolor=black@0.4:boxborderw=4"
                )
            else:
                vf_parts.append(
                    f"drawtext=font='Sans':text='{wm_text}':x={x_pos}:y={y_pos}:fontsize=20:fontcolor=white"
                )
        args.extend(["-map", "0:v:0", "-map", "0:a?", "-vf", ",".join(vf_parts)])

    is_avi = settings.get("format", "").lower() == "avi" or output_path.lower().endswith(".avi")

    has_subtitles = False
    # Subtítulos de texto para contenedores no AVI
    if not is_avi and include_subtitles:
        text_sub_streams = get_text_subtitle_streams(input_path)
        for idx in text_sub_streams:
            args.extend(["-map", f"0:{idx}"])
        if text_sub_streams:
            args.extend(["-c:s", "mov_text"])
            has_subtitles = True

    codec = settings.get("codec") or "libx264"
    args.extend(["-c:v", codec])

    if settings.get("preset") and codec != "copy":
        args.extend(["-preset", settings.get("preset")])
    if settings.get("crf") and codec != "copy":
        args.extend(["-crf", str(settings.get("crf"))])
    if settings.get("fps") and codec != "copy":
        args.extend(["-r", str(settings.get("fps"))])

    audio_bitrate = settings.get("audio_bitrate") or "128k"
    clean_bitrate = audio_bitrate.lower().rstrip("k")
    ac_channels = "2"
    try:
        if int(clean_bitrate) <= 64:
            ac_channels = "1"
    except ValueError:
        pass

    audio_codec = "libmp3lame" if is_avi else "aac"
    args.extend([
        "-pix_fmt", "yuv420p",
        "-c:a", audio_codec,
        "-b:a", audio_bitrate,
        "-ac", ac_channels,
    ])

    if not is_avi:
        args.extend(["-movflags", "+faststart"])

    args.extend(["-threads", "0", output_path])
    return args, has_subtitles


def execute_ffmpeg_cmd(
    args: List[str],
    total_duration: float,
    on_cmd: Optional[Callable[[subprocess.Popen], None]] = None,
    on_progress: Optional[Callable[[Dict[str, Any]], None]] = None,
    is_cancelled: Optional[Callable[[], bool]] = None,
) -> None:
    cmd_list = ["ffmpeg"] + args

    proc = subprocess.Popen(
        cmd_list,
        stdout=subprocess.DEVNULL,
        stderr=subprocess.PIPE,
        bufsize=0,
    )

    if on_cmd:
        on_cmd(proc)

    last_report_time = 0.0
    last_stderr_lines = []
    buffer = bytearray()

    # Lector que procesa tanto \r como \n (scanCRLFOrCR idéntico a Go)
    try:
        while True:
            if is_cancelled and is_cancelled():
                proc.kill()
                raise InterruptedError("compresión cancelada")

            chunk = proc.stderr.read(128)
            if not chunk:
                if buffer:
                    line = buffer.decode("utf-8", errors="replace").strip()
                    if line:
                        last_stderr_lines.append(line)
                break

            for byte in chunk:
                if byte in (10, 13):  # \n o \r
                    if buffer:
                        line = buffer.decode("utf-8", errors="replace").strip()
                        buffer.clear()
                        if line:
                            last_stderr_lines.append(line)
                            if len(last_stderr_lines) > 30:
                                last_stderr_lines.pop(0)

                            matches = TIME_REGEX.search(line)
                            if matches:
                                h = float(matches.group(1))
                                m = float(matches.group(2))
                                s = float(matches.group(3))
                                cur_time_sec = h * 3600 + m * 60 + s

                                percent = 0.0
                                if total_duration > 0:
                                    percent = min(100.0, (cur_time_sec / total_duration) * 100.0)

                                speed = "1.0x"
                                sp_match = SPEED_REGEX.search(line)
                                if sp_match:
                                    speed = sp_match.group(1) + "x"

                                eta_str = ""
                                if total_duration > 0 and cur_time_sec > 0:
                                    rem_sec = max(0.0, total_duration - cur_time_sec)
                                    try:
                                        sp_val = float(speed.rstrip("x"))
                                        if sp_val > 0:
                                            real_rem = int(rem_sec / sp_val)
                                            eta_str = f"{real_rem // 3600:02d}:{(real_rem % 3600) // 60:02d}:{real_rem % 60:02d}"
                                    except ValueError:
                                        pass

                                time_str = f"{int(h):02d}:{int(m):02d}:{int(s):02d}"

                                cur_now = time.time()
                                if cur_now - last_report_time >= 1.0:
                                    last_report_time = cur_now
                                    if on_progress:
                                        on_progress({
                                            "percent": round(percent, 1),
                                            "speed": speed,
                                            "time": time_str,
                                            "eta": eta_str,
                                        })
                else:
                    buffer.append(byte)

        proc.wait()
        if proc.returncode != 0:
            if is_cancelled and is_cancelled():
                raise InterruptedError("compresión cancelada")
            detail = "\n".join(last_stderr_lines)
            raise RuntimeError(f"ffmpeg falló con código {proc.returncode}:\n{detail}")

    finally:
        if proc.poll() is None:
            try:
                proc.kill()
            except Exception:
                pass


def compress_with_cmd(
    input_path: str,
    output_path: str,
    settings: Dict[str, Any],
    wm_image_path: str,
    total_duration: float,
    on_cmd: Optional[Callable[[subprocess.Popen], None]] = None,
    on_progress: Optional[Callable[[Dict[str, Any]], None]] = None,
    is_cancelled: Optional[Callable[[], bool]] = None,
) -> None:
    """
    Ejecuta la compresión con FFmpeg reportando el progreso en vivo.
    Si la compresión falla debido a subtítulos corruptos o timestamps inválidos,
    reintenta automáticamente descartando los subtítulos para no fallar la tarea.
    """
    args, has_subtitles = build_compress_args(input_path, output_path, settings, wm_image_path, include_subtitles=True)
    try:
        execute_ffmpeg_cmd(args, total_duration, on_cmd, on_progress, is_cancelled)
        return
    except RuntimeError as e:
        if is_cancelled and is_cancelled():
            raise
        if has_subtitles and is_subtitle_or_mux_error(str(e)):
            logger.warning(f"[ffmpeg] Falló compresión con subtítulos ({e}). Reintentando sin subtítulos...")
            if os.path.isfile(output_path):
                try:
                    os.remove(output_path)
                except Exception:
                    pass
            retry_args, _ = build_compress_args(input_path, output_path, settings, wm_image_path, include_subtitles=False)
            execute_ffmpeg_cmd(retry_args, total_duration, on_cmd, on_progress, is_cancelled)
            logger.info("[ffmpeg] Compresión completada con éxito tras descartar subtítulos corruptos.")
            return
        raise