# -*- coding: utf-8 -*-
"""QA on the finished MP4: extracts the audio already mixed with the music,
cuts it by the timeline and transcribes it. It proves the voice is
intelligible over the bed, not just that the loose WAV was fine."""
import json, os, subprocess, sys
from difflib import SequenceMatcher

ROOT = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
sys.path.insert(0, os.path.join(ROOT, "scripts"))
from narration import SCRIPTS
from verify import norm
from gen_audio import transcribe

LIMIT = 0.88


def main(names):
    worst = []
    for name in names:
        mp4 = os.path.join(ROOT, "out", f"{name}.mp4")
        tl = json.load(open(os.path.join(ROOT, "out", f"{name}_timeline.json")))
        segs = json.load(open(os.path.join(ROOT, "audio", name, "segments.json")))["segments"]
        work = os.path.join(ROOT, "tmp", name + "_fin")
        os.makedirs(work, exist_ok=True)
        integer = os.path.join(work, "todo.wav")
        subprocess.run(["ffmpeg", "-v", "error", "-y", "-i", mp4, "-vn",
                        "-ac", "1", "-ar", "16000", integer], check=True)

        chunks = []
        for bt in tl["beats"]:
            p = os.path.join(work, f"{bt['i']:02d}.wav")
            subprocess.run(["ffmpeg", "-v", "error", "-y", "-ss", f"{bt['start']:.3f}",
                            "-t", f"{segs[bt['i']]['dur'] + 0.25:.3f}", "-i", integer,
                            "-c", "copy", p], check=True)
            chunks.append(p)
        tr = transcribe(chunks, work)

        print(f"\n=== {name} (mezcla final) ===")
        for bt, p in zip(tl["beats"], chunks):
            i = bt["i"]
            got = tr[p]
            r = SequenceMatcher(None, norm(segs[i]["tts"]), norm(got)).ratio()
            if r < LIMIT:
                worst.append((name, i, r))
            print(f" {i:02d} {'OK ' if r >= LIMIT else 'REV'} {r:.2f} | {got}")
    print("\n>> a revisar:", worst if worst else "nada")


if __name__ == "__main__":
    main(sys.argv[1:] or list(SCRIPTS))