from __future__ import annotations

import argparse

import bootstrap  # noqa: F401

from signspeak.llm import generate_subtitle_and_instruction
from signspeak.pipeline import run_asl_video
from signspeak.tts import generate_tts


def main() -> None:
    parser = argparse.ArgumentParser(description="Run ASL -> llama.cpp -> Qwen3-TTS.")
    parser.add_argument("video", nargs="?", default=None, help="Video path. Creates a demo clip if omitted.")
    parser.add_argument("--gloss-override", default=None, help="Manual glosses to use when the ASL model is missing.")
    parser.add_argument("--language", default="English")
    parser.add_argument("--speaker", default="Ryan")
    args = parser.parse_args()

    intent_json, _, summary, debug_video_path = run_asl_video(args.video, args.gloss_override)
    subtitle, instruction, _ = generate_subtitle_and_instruction(intent_json)
    audio_path = generate_tts(subtitle, args.language, args.speaker, instruction)

    print(summary)
    print(f"Debug overlay: {debug_video_path}")
    print(f"Subtitle: {subtitle}")
    print(f"Voice instruction: {instruction}")
    print(f"Audio: {audio_path}")


if __name__ == "__main__":
    main()