conversion_project/core/encode_engine.py

# core/encode_engine.py
"""FFmpeg encoding engine with comprehensive logging."""

import subprocess
from pathlib import Path

from core.audio_handler import get_audio_streams, choose_audio_bitrate
from core.logger_helper import setup_logger

logger = setup_logger(Path(__file__).parent.parent / "logs")


def run_ffmpeg(input_file: Path, output_file: Path, cq: int, scale_width: int, scale_height: int,
               src_width: int, src_height: int, filter_flags: str, audio_config: dict, 
               method: str, bitrate_config: dict, subtitle_file: Path = None, audio_language: str = None):
    """
    Run FFmpeg encode with comprehensive logging.
    
    Returns tuple: (orig_size, out_size, reduction_ratio)
    """
    streams = get_audio_streams(input_file)
    
    # Log comprehensive encode settings
    header = f"\n🧩 ENCODE SETTINGS"
    logger.info(header)
    print(" ")
    
    logger.info(f"  Video:")
    logger.info(f"    • Source Resolution: {src_width}x{src_height}")
    logger.info(f"    • Target Resolution: {scale_width}x{scale_height}")
    logger.info(f"    • Encoder: av1_nvenc (preset p1, pix_fmt p010le)")
    logger.info(f"    • Scale Filter: {filter_flags}")
    logger.info(f"    • Encode Method: {method}")
    if method == "CQ":
        logger.info(f"    • CQ Value: {cq}")
    else:
        res_key = "1080" if scale_height >= 1080 or scale_width >= 1920 else "720"
        vb = bitrate_config.get(f"bitrate_{res_key}", "900k")
        maxrate = bitrate_config.get(f"maxrate_{res_key}", "1250k")
        logger.info(f"    • Bitrate: {vb}, Max: {maxrate}")
    
    logger.info(f"  Audio Streams ({len(streams)} detected):")
    print(" ")
    
    for (index, channels, avg_bitrate, src_lang, meta_bitrate) in streams:
        # Normalize to 2ch or 6ch output
        is_1080_class = scale_height >= 1080 or scale_width >= 1920
        output_channels = 6 if is_1080_class and channels >= 6 else 2
        codec, br = choose_audio_bitrate(output_channels, avg_bitrate, audio_config, is_1080_class)
        
        if codec == "copy":
            action = "COPY (preserve)"
            bitrate_display = f"{avg_bitrate}kbps"
        else:
            action = "ENCODE"
            bitrate_display = f"{br/1000:.0f}kbps"
        
        line = f"    - Stream #{index}: {channels}ch→{output_channels}ch | Lang: {src_lang} | Detected: {avg_bitrate}kbps | Action: {action} | Target: {bitrate_display}"
        print(line)
        logger.info(line)

    cmd = ["ffmpeg","-y","-i",str(input_file)]
    
    # Add subtitle input if present
    if subtitle_file:
        cmd.extend(["-i", str(subtitle_file)])
    
    cmd.extend([
        "-vf",f"scale={scale_width}:{scale_height}:flags={filter_flags}:force_original_aspect_ratio=decrease",
        "-map","0:v","-map","0:a"])
    
    # Add subtitle mapping if present
    if subtitle_file:
        cmd.extend(["-map", "1:s"])
    else:
        cmd.extend(["-map", "0:s?"])
    
    cmd.extend([
        "-c:v","av1_nvenc","-preset","p1","-pix_fmt","p010le"])

    if method=="CQ":
        cmd += ["-cq", str(cq)]
    else:
        # Use bitrate config (fallback mode)
        res_key = "1080" if scale_height >= 1080 or scale_width >= 1920 else "720"
        vb = bitrate_config.get(f"bitrate_{res_key}", "900k")
        maxrate = bitrate_config.get(f"maxrate_{res_key}", "1250k")
        bufsize = bitrate_config.get(f"bufsize_{res_key}", "1800k")
        cmd += ["-b:v", vb, "-maxrate", maxrate, "-bufsize", bufsize]

    for i, (index, channels, avg_bitrate, src_lang, meta_bitrate) in enumerate(streams):
        # Normalize to 2ch or 6ch output
        is_1080_class = scale_height >= 1080 or scale_width >= 1920
        output_channels = 6 if is_1080_class and channels >= 6 else 2
        codec, br = choose_audio_bitrate(output_channels, avg_bitrate, audio_config, is_1080_class)
        
        if codec == "copy":
            # Preserve original audio
            cmd += [f"-c:a:{i}", "copy"]
            # Only add language metadata if explicitly provided
            if audio_language:
                cmd += [f"-metadata:s:a:{i}", f"language={audio_language}"]
        else:
            # Re-encode with target bitrate
            cmd += [
                f"-c:a:{i}", codec,
                f"-b:a:{i}", str(br),
                f"-ac:{i}", str(output_channels),
                f"-channel_layout:a:{i}", "5.1" if output_channels == 6 else "stereo"
            ]
            # Only add language metadata if explicitly provided
            if audio_language:
                cmd += [f"-metadata:s:a:{i}", f"language={audio_language}"]

    # Add subtitle codec and metadata if subtitles are present
    if subtitle_file:
        cmd += ["-c:s", "srt", "-metadata:s:s:0", "language=eng"]
    else:
        cmd += ["-c:s", "copy"]
    
    cmd += [str(output_file)]

    print(f"\n🎬 Running {method} encode: {output_file.name}")
    logger.info(f"Running {method} encode: {output_file.name}")

    subprocess.run(cmd, check=True)

    orig_size = input_file.stat().st_size
    out_size = output_file.stat().st_size
    reduction_ratio = out_size / orig_size
    
    # Log comprehensive results
    logger.info(f"\n📊 ENCODE RESULTS:")
    logger.info(f"  Original Size: {orig_size/1e6:.2f} MB")
    logger.info(f"  Encoded Size:  {out_size/1e6:.2f} MB")
    logger.info(f"  Reduction:     {reduction_ratio:.1%} of original ({(1-reduction_ratio):.1%} saved)")
    logger.info(f"  Resolution:    {src_width}x{src_height} → {scale_width}x{scale_height}")
    logger.info(f"  Audio Streams: {len(streams)} streams processed")
    
    msg = f"📦 Original: {orig_size/1e6:.2f} MB → Encoded: {out_size/1e6:.2f} MB ({reduction_ratio:.1%} of original)"
    print(msg)

    return orig_size, out_size, reduction_ratio