services/audio_extractor.py

"""
Stage 2 — Audio Extraction & Preprocessing
Extracts audio from video, creates STT-ready WAV and stereo backup.
"""

import logging
import subprocess
from pathlib import Path
from config import AUDIO_SAMPLE_RATE, AUDIO_CHANNELS

logger = logging.getLogger(__name__)


def extract_audio_for_stt(video_path: Path, output_dir: Path) -> Path:
    """
    Extract audio from video as 16kHz mono WAV for speech-to-text.
    """
    output_path = output_dir / "audio_stt.wav"

    cmd = [
        "ffmpeg", "-y",
        "-i", str(video_path),
        "-vn",                          # No video
        "-acodec", "pcm_s16le",        # 16-bit PCM
        "-ar", str(AUDIO_SAMPLE_RATE),  # 16kHz
        "-ac", str(AUDIO_CHANNELS),     # Mono
        str(output_path)
    ]

    logger.info("Extracting audio for STT (16kHz mono WAV)...")
    result = subprocess.run(cmd, capture_output=True, text=True, timeout=600)

    if result.returncode != 0:
        raise RuntimeError(f"FFmpeg audio extraction failed: {result.stderr}")

    logger.info(f"STT audio: {output_path} ({output_path.stat().st_size / 1e6:.1f} MB)")
    return output_path


def extract_audio_stereo(video_path: Path, output_dir: Path) -> Path:
    """
    Extract original stereo audio (for background mixing later).
    """
    output_path = output_dir / "audio_original_stereo.wav"

    cmd = [
        "ffmpeg", "-y",
        "-i", str(video_path),
        "-vn",
        "-acodec", "pcm_s16le",
        "-ar", "44100",
        str(output_path)
    ]

    logger.info("Extracting stereo audio for background mixing...")
    result = subprocess.run(cmd, capture_output=True, text=True, timeout=600)

    if result.returncode != 0:
        raise RuntimeError(f"FFmpeg stereo extraction failed: {result.stderr}")

    logger.info(f"Stereo audio: {output_path}")
    return output_path


def get_audio_duration(audio_path: Path) -> float:
    """Get audio duration in seconds using ffprobe."""
    cmd = [
        "ffprobe",
        "-v", "quiet",
        "-show_entries", "format=duration",
        "-of", "csv=p=0",
        str(audio_path)
    ]
    result = subprocess.run(cmd, capture_output=True, text=True, timeout=30)
    if result.returncode != 0:
        raise RuntimeError(f"ffprobe failed: {result.stderr}")
    return float(result.stdout.strip())


def get_video_duration(video_path: Path) -> float:
    """Get video duration in seconds."""
    return get_audio_duration(video_path)  # ffprobe works on video too