main.py

import argparse
import os
import json 
import sys 
import subprocess 

# ======================== Spleeter setup
# Meminimalkan log TensorFlow agar tidak mengganggu output JSON
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3' 
os.environ["CUDA_DEVICE_ORDER"] = "PCI_BUS_ID"
os.environ["CUDA_VISIBLE_DEVICES"] = "-1"      

import warnings
warnings.filterwarnings("ignore", category=FutureWarning)

# setup spleeter
from multiprocessing import freeze_support
freeze_support()
# =================================================

def get_base_path():
    """Get the path where the executable or script is running from."""
    if getattr(sys, 'frozen', False):
        # The script is running in a frozen executable (e.g., PyInstaller)
        # sys.executable is the path to the executable itself
        base_path = os.path.dirname(sys.executable)
    else:
        # The script is running in a normal Python environment
        # __file__ is the path to the script file
        base_path = os.path.dirname(os.path.abspath(__file__))
    
    return base_path

def setup_ffmpeg_path(ffmpeg_bin_dir: str):
    """
    Menambahkan direktori biner FFmpeg ke variabel lingkungan PATH.
    Semua pesan diagnostik dialihkan ke sys.stderr.
    """
    if ffmpeg_bin_dir:
        # Normalize path
        ffmpeg_bin_dir = os.path.abspath(ffmpeg_bin_dir)
        
        if not os.path.isdir(ffmpeg_bin_dir):
             print(f"Warning: FFmpeg bin directory not found at '{ffmpeg_bin_dir}'. Relying on system PATH.", file=sys.stderr)
             return

        try:
            # Tambahkan ke depan PATH
            os.environ["PATH"] = ffmpeg_bin_dir + os.pathsep + os.environ["PATH"]
        except KeyError:
            # Jika PATH tidak ada, setel saja
            os.environ["PATH"] = ffmpeg_bin_dir
    else:
        # when not custom ffmpeg path, use default
        base_dir = get_base_path()
        ffmpeg_bin_dir = os.path.join(base_dir, 'ffmpeg/')
        os.environ["PATH"] = ffmpeg_bin_dir

def create_pretrained_models_link(source_dir: str):
    """
    Membuat junction (mklink /J) untuk folder 'pretrained_models'.
    Jika mklink gagal karena izin atau alasan lain, program berhenti (sys.exit(1)).
    """
    # Destination: Di mana link akan dibuat (direktori saat ini)
    DEST_LINK = os.path.join(os.getcwd(), 'pretrained_models')
    source_dir = os.path.abspath(source_dir)

    # Check 1: Apakah direktori sumber model ada?
    if not os.path.isdir(source_dir):
        # Jika direktori sumber tidak ada, kita hanya memberi tahu user dan TIDAK MENGHENTIKAN program.
        # Spleeter akan mencoba mencari model di lokasi defaultnya.
        print(f"Info: Model source directory not found at '{source_dir}'. Skipping mklink.", file=sys.stderr)
        return

    # Check 2: Apakah link/target sudah ada?
    if os.path.exists(DEST_LINK):
        return 

    # Buat perintah mklink
    command = ['mklink', '/J', DEST_LINK, source_dir]
    
    try:
        # print(f"Attempting to create model junction: {DEST_LINK} -> {source_dir}", file=sys.stderr)
        
        result = subprocess.run(
            command, 
            shell=True,
            check=False,
            capture_output=True,
            text=True
        )

        if result.returncode == 0:
            # Sukses
            pass
        elif 'You do not have sufficient privilege' in result.stderr:
            # Kegagalan karena izin, harus dihentikan
            print("\nFATAL ERROR: Model junction creation failed due to insufficient privilege.", file=sys.stderr)
            print("Please run the script as **Administrator** or manually create the junction.", file=sys.stderr)
            print(f"Command attempted: {' '.join(command)}", file=sys.stderr)
            sys.exit(1) # HENTIKAN PROGRAM
        else:
            # Kegagalan mklink lain, harus dihentikan
            print(f"\nFATAL ERROR: Model junction failed (Code {result.returncode}).", file=sys.stderr)
            print(f"Output: {result.stdout.strip()} | Error: {result.stderr.strip()}", file=sys.stderr)
            sys.exit(1) # HENTIKAN PROGRAM

    except FileNotFoundError:
        print("\nFATAL ERROR: 'mklink' command not found. Cannot create junction.", file=sys.stderr)
        sys.exit(1) # HENTIKAN PROGRAM
    except Exception as e:
        print(f"\nFATAL ERROR: An unexpected error occurred during mklink: {e}", file=sys.stderr)
        sys.exit(1) # HENTIKAN PROGRAM


def main():
    """
    Menyiapkan parser argumen, memproses argumen, dan menjalankan pemisahan audio Spleeter.
    Semua log dicetak ke sys.stderr, kecuali output JSON yang dicetak ke sys.stdout.
    """
    parser = argparse.ArgumentParser(
        description="Spleeter Standalone CLI",
        formatter_class=argparse.RawTextHelpFormatter
    )

    # Argumen yang Diperlukan
    parser.add_argument('-f', '--file', type=str, help='Input audio file path (e.g., input.mp3 or input.wav)', required=True)
    parser.add_argument('-o', '--output', type=str, help='Output directory path where stems will be saved', required=True)
    
    # Argumen Opsional (Stems)
    parser.add_argument('-s', '--stem', type=str, choices=['2', '4', '5'], default='2', 
        help=('Number of stems to separate into:\n[2]: Vocals / Accompaniment\n[4]: Vocals / Drums / Bass / Other\n[5]: Vocals / Drums / Bass / Piano / Other\nDefault: 2'), required=False)
    
    # Argumen Opsional (Model Path)
    parser.add_argument('-m', '--model-path', type=str, default='',
        help=('Optional: Absolute path to the directory containing the Spleeter "pretrained_models" directory. If provided, the script will attempt to create an mklink junction for it.'), required=False)
    
    # Argumen Opsional (FFmpeg Path)
    parser.add_argument('-fm', '--ffmpeg-path', type=str, default='',
        help=('Optional: Absolute path to the FFmpeg "bin" directory.\nUse this if FFmpeg is not in your system PATH.'), required=False)

    # Argumen Opsional (JSON Output Flag)
    parser.add_argument('-j', '--json-output', action='store_true', help='Output results in JSON format instead of human-readable text (Murni ke stdout).')
    
    args = parser.parse_args()

    # =======================================================
    # STEP 1: CREATE MODEL LINK (Uses parsed argument)
    # Jika args.model_path TIDAK kosong, fungsi ini akan mencoba membuat link.
    # Jika gagal (izin/kesalahan lain), program akan berhenti di dalam fungsi.
    if args.model_path:
        create_pretrained_models_link(args.model_path)
    # Jika args.model_path KOSONG, tidak ada yang terjadi, dan program berlanjut.
    # =======================================================

    # =======================================================
    # STEP 2: APPLY FFMPEG PATH
    setup_ffmpeg_path(args.ffmpeg_path)
    # =======================================================

    # Inisialisasi struktur data hasil
    result_data = {
        'input_file': args.file,
        'output_directory': args.output,
        'stems_selected': args.stem,
        'spleeter_param': f'spleeter:{args.stem}stems',
        'status': 'Error', # Status default
        'message': ''
    }
    
    # Cek file input
    if not os.path.exists(args.file):
        result_data['message'] = f"Error: Input file not found at '{args.file}'"
        
        if args.json_output:
            print(json.dumps(result_data, indent=4))
        else:
            print(result_data['message'], file=sys.stderr)
        return
    
    # --- Integrasi dengan Spleeter ---
    
    # Cetak argumen human-readable ke stderr jika tidak dalam mode JSON
    if not args.json_output:
        print("-" * 30, file=sys.stderr)
        print("Spleeter CLI Arguments:", file=sys.stderr)
        print(f"Input File: {args.file}", file=sys.stderr)
        print(f"Output Directory: {args.output}", file=sys.stderr)
        print(f"Stems Selected: {args.stem} ({result_data['spleeter_param']})", file=sys.stderr)
        if args.ffmpeg_path:
            print(f"FFmpeg Path Used: {args.ffmpeg_path}", file=sys.stderr)
        if args.model_path:
            print(f"Model Path Used: {args.model_path} (Junction attempted)", file=sys.stderr)
        print("-" * 30, file=sys.stderr)
        print("Starting separation...", file=sys.stderr)
    
    try:
        # Pustaka Spleeter dimuat di sini
        import tensorflow as tf
        tf.get_logger().setLevel('ERROR')    

        import logging
        from spleeter import separator
        logging.getLogger('spleeter').setLevel(logging.WARNING)

        # Gunakan spleeter_param
        split = separator.Separator(result_data['spleeter_param'], multiprocess=False)
        split.separate_to_file(args.file, args.output)
        
        # Update status sukses
        result_data['status'] = 'success'
        result_data['message'] = f"Successfully separated and saved stems to {args.output}"

    except Exception as e:
        # Update status error
        result_data['status'] = 'error'
        result_data['message'] = f"An error occurred during separation: {e}"
        # Jika bukan output JSON, cetak error langsung ke stderr
        if not args.json_output:
            print(result_data['message'], file=sys.stderr)

    # Logika Output Akhir
    if args.json_output:
        # HANYA mencetak output JSON ke sys.stdout.
        print(json.dumps(result_data, indent=4))
    elif result_data['status'] == 'success':
        # Mencetak pesan sukses ke sys.stderr
        print(result_data['message'], file=sys.stderr)


if __name__ == '__main__':
    main()