"""
pipeline/audio.py
Extracts a 16 kHz mono WAV from the source video for AssemblyAI.
"""

from pathlib import Path

from utils import log, run_cmd


def extract_audio(video_path: Path, output_dir: Path) -> Path:
    """
    Extract audio from a video into a 16 kHz mono WAV.

    Why 16 kHz mono?
    - AssemblyAI is optimised for this format
    - Smaller file = faster upload
    - Speech intelligibility is fully preserved

    Returns
    -------
    Path to the .wav file
    """
    audio_path = output_dir / f"{video_path.stem}_audio.wav"

    cmd = [
        "ffmpeg", "-y",
        "-i", str(video_path),
        "-vn",                  # drop video stream
        "-ar", "16000",         # 16 kHz
        "-ac", "1",             # mono
        "-acodec", "pcm_s16le", # 16-bit PCM
        str(audio_path),
    ]

    run_cmd(cmd, f"Extracting audio → {audio_path.name}")

    size_mb = audio_path.stat().st_size / 1_048_576
    log("✅", f"Audio extracted ({size_mb:.1f} MB)")
    return audio_path