#!/usr/bin/env python3
"""3-minute candles from the 1-minute files: MASSIVE_ROOT/minute/<day>/<SYM>.json.gz → MASSIVE_ROOT/minute3/<day>/<SYM>.json.gz
(same format: timestamp of the candle start in UTC, open, high, low, close, volume). Candles start at :00, :03, :06 ...
New York time, the way the balance-method test builds them.

    python make_3min.py
"""
from __future__ import annotations

import gzip
import json
import os
from datetime import datetime, timezone
from pathlib import Path
from zoneinfo import ZoneInfo

ROOT = Path(os.environ.get("MASSIVE_ROOT", "massive_data"))
ET = ZoneInfo("America/New_York")


def main() -> None:
    files = sorted((ROOT / "minute").glob("*/*.json.gz"))
    for k, p in enumerate(files):
        out = ROOT / "minute3" / p.parent.name / p.name
        if out.exists():
            continue
        bars = json.load(gzip.open(p, "rt"))
        cs = []
        for b in bars:
            t = datetime.fromisoformat(b["timestamp"].replace("Z", "+00:00")).astimezone(ET)
            key = (t.hour * 60 + t.minute) // 3
            if cs and cs[-1]["_k"] == key:
                c = cs[-1]
                c["high"] = max(c["high"], b["high"]); c["low"] = min(c["low"], b["low"]); c["close"] = b["close"]; c["volume"] += b["volume"]
            else:
                start = t.replace(minute=(t.minute // 3) * 3, second=0).astimezone(timezone.utc)
                cs.append({"_k": key, "timestamp": start.strftime("%Y-%m-%dT%H:%M:%SZ"), "open": b["open"], "high": b["high"],
                           "low": b["low"], "close": b["close"], "volume": b["volume"]})
        out.parent.mkdir(parents=True, exist_ok=True)
        with gzip.open(out, "wt") as fh:
            json.dump([{k2: v for k2, v in c.items() if k2 != "_k"} for c in cs], fh)
        if k % 20000 == 0:
            print(f"  {k}/{len(files)}", flush=True)
    print(f"done: {len(files)} files", flush=True)


if __name__ == "__main__":
    main()