"""Recover per-turn time boundaries for every viewer scenario from silence gaps in the merged
mp3 (turns were concatenated with ~0.35-0.5 s of silence). Writes data/viewer/turn_bounds.json:
  { sid: {"n_turns": N, "bounds": [t1, t2, ...] (N-1 seconds, midpoints of gaps), "method": "silence"|"text"} }
Fallback when the gap count doesn't match: split proportionally to turn text length.
"""
import json, subprocess, re, sys, glob, os
from concurrent.futures import ThreadPoolExecutor
V = "data/viewer"
def probe(sid, turns):
    mp3 = f"{V}/{sid}_dataset.mp3"
    if not os.path.isfile(mp3): return sid, None
    try:
        dur = float(subprocess.run(["ffprobe","-v","error","-show_entries","format=duration","-of","csv=p=0",mp3],capture_output=True,text=True).stdout.strip())
        out = subprocess.run(["ffmpeg","-hide_banner","-i",mp3,"-af","silencedetect=noise=-35dB:d=0.25","-f","null","-"],capture_output=True,text=True).stderr
    except Exception as e:
        return sid, None
    starts = [float(x) for x in re.findall(r"silence_start: ([0-9.]+)", out)]
    ends   = [float(x) for x in re.findall(r"silence_end: ([0-9.]+)", out)]
    gaps = [(s, e) for s, e in zip(starts, ends) if s > 0.15 and e < dur - 0.15]
    n = len(turns)
    if n <= 1: return sid, {"n_turns": n, "bounds": [], "duration": dur, "method": "single"}
    if len(gaps) >= n - 1:
        # take the (n-1) longest gaps, in time order
        pick = sorted(sorted(gaps, key=lambda g: g[1]-g[0], reverse=True)[:n-1])
        return sid, {"n_turns": n, "bounds": [round((s+e)/2, 3) for s, e in pick], "duration": dur, "method": "silence"}
    lens = [max(1, len(t.get("text",""))) for t in turns]; tot = sum(lens); acc = 0; b = []
    for L in lens[:-1]: acc += L; b.append(round(dur * acc / tot, 3))
    return sid, {"n_turns": n, "bounds": b, "duration": dur, "method": "text"}
jobs = []
for f in glob.glob(f"{V}/*_pred_v18m_dataset.json"):
    sid = os.path.basename(f).replace("_pred_v18m_dataset.json", "")
    try: turns = json.load(open(f)).get("turns", [])
    except Exception: continue
    jobs.append((sid, turns))
res = {}
with ThreadPoolExecutor(8) as ex:
    for sid, r in ex.map(lambda j: probe(*j), jobs):
        if r: res[sid] = r
json.dump(res, open(f"{V}/turn_bounds.json", "w"), ensure_ascii=False)
from collections import Counter
print(len(res), "scenarios;", Counter(r["method"] for r in res.values()))