kis_trader/scripts/test_ws_tick_kis_kiwoom_gap.py

#!/usr/bin/env python3
"""
scripts/test_ws_tick_kis_kiwoom_gap.py
=====================================
한투(kis) vs 키움(kiwoom) ``ws_ticks`` 밀도·틱타임/수신갭 비교.

실 WS/주문 호출 없음. DB 조회만.
틱매도 비동기 적용 후 ~1시간 뒤, 매도 시각 전후로
「키움만 비고 한투는 쌓임」이 줄었는지 볼 때 사용.

  python3 -u scripts/test_ws_tick_kis_kiwoom_gap.py
  python3 -u scripts/test_ws_tick_kis_kiwoom_gap.py --minutes 60
  python3 -u scripts/test_ws_tick_kis_kiwoom_gap.py --start '2026-08-19 09:00:00' --end '2026-08-19 10:00:00'
  python3 -u scripts/test_ws_tick_kis_kiwoom_gap.py --code 025980 --minutes 60

로그: logs/test_ws_tick_kis_kiwoom_gap_YYYYMMDD_HHMMSS.log
"""
from __future__ import annotations

import argparse
import sys
from datetime import datetime, timedelta
from pathlib import Path
from typing import Any, Dict, List, Optional, Tuple

ROOT = Path(__file__).resolve().parents[1]
if str(ROOT) not in sys.path:
    sys.path.insert(0, str(ROOT))


def _tee_log() -> Path:
    log_dir = ROOT / "logs"
    log_dir.mkdir(exist_ok=True)
    ts = datetime.now().strftime("%Y%m%d_%H%M%S")
    path = log_dir / ("test_ws_tick_kis_kiwoom_gap_%s.log" % ts)
    fp = path.open("w", encoding="utf-8")

    class _Tee:
        def __init__(self, *files: Any) -> None:
            self.files = files

        def write(self, s: str) -> None:
            for f in self.files:
                f.write(s)
                f.flush()

        def flush(self) -> None:
            for f in self.files:
                f.flush()

    sys.stdout = _Tee(sys.__stdout__, fp)  # type: ignore[assignment]
    sys.stderr = _Tee(sys.__stderr__, fp)  # type: ignore[assignment]
    print("로그 파일:", path)
    return path


def _cols(db: Any, table: str) -> List[str]:
    rows = db.conn.execute("SHOW COLUMNS FROM `%s`" % table).fetchall()
    return [str(r["Field"]) for r in rows]


def _parse_ts(raw: Any) -> Optional[datetime]:
    s = str(raw or "").strip()
    if not s:
        return None
    for fmt in (
        "%Y-%m-%d %H:%M:%S.%f",
        "%Y-%m-%d %H:%M:%S",
        "%Y%m%d%H%M%S",
    ):
        try:
            return datetime.strptime(s[:26] if "." in s else s[:19], fmt)
        except Exception:
            continue
    try:
        return datetime.fromisoformat(s.replace("Z", ""))
    except Exception:
        return None


def _gaps(ts_list: List[datetime]) -> List[float]:
    if len(ts_list) < 2:
        return []
    ordered = sorted(ts_list)
    out: List[float] = []
    for i in range(1, len(ordered)):
        out.append((ordered[i] - ordered[i - 1]).total_seconds())
    return out


def _gap_stats(gaps: List[float]) -> Dict[str, Any]:
    if not gaps:
        return {"n": 0, "max": None, "p50": None, "gt2": 0, "gt5": 0}
    sg = sorted(gaps)
    n = len(sg)
    p50 = sg[n // 2]
    return {
        "n": n,
        "max": round(max(sg), 3),
        "p50": round(p50, 3),
        "gt2": sum(1 for x in sg if x >= 2.0),
        "gt5": sum(1 for x in sg if x >= 5.0),
    }


def _pick_time_col(cols: List[str]) -> Optional[str]:
    for c in (
        "created_at", "updated_at", "order_time", "ord_time",
        "submitted_at", "ts", "insert_time",
    ):
        if c in cols:
            return c
    return None


def main() -> int:
    ap = argparse.ArgumentParser(description="kis vs kiwoom ws_ticks 밀도·갭")
    ap.add_argument("--minutes", type=int, default=60, help="끝시각 기준 최근 N분 (start/end 없을 때)")
    ap.add_argument("--start", default="", help="YYYY-MM-DD HH:MM:SS")
    ap.add_argument("--end", default="", help="YYYY-MM-DD HH:MM:SS")
    ap.add_argument("--code", default="", help="6자리 종목. 비우면 겹치는 종목 상위")
    ap.add_argument("--limit-codes", type=int, default=15, help="겹치는 종목 리포트 개수")
    args = ap.parse_args()

    now = datetime.now()
    if str(args.start or "").strip() and str(args.end or "").strip():
        start_s = str(args.start).strip()
        end_s = str(args.end).strip()
    else:
        end_dt = now
        start_dt = end_dt - timedelta(minutes=max(1, int(args.minutes or 60)))
        start_s = start_dt.strftime("%Y-%m-%d %H:%M:%S")
        end_s = end_dt.strftime("%Y-%m-%d %H:%M:%S")

    from database import TradeDB

    db = TradeDB()
    print("구간:", start_s, "~", end_s)
    print("code:", args.code or "(겹치는 종목 자동)")

    tick_cols = _cols(db, "ws_ticks")
    print("SHOW COLUMNS ws_ticks:", tick_cols)
    need = {"code", "source", "recv_ts"}
    missing = sorted(need - set(tick_cols))
    if missing:
        print("실패: ws_ticks 컬럼 없음", missing)
        return 1
    has_tick_time = "tick_time" in tick_cols

    # 전체 종목수·행수
    tot = db.conn.execute(
        """
        SELECT
          COUNT(*) n,
          SUM(source='kiwoom') n_kw,
          SUM(source='kis') n_kis,
          COUNT(DISTINCT CASE WHEN source='kiwoom' THEN code END) codes_kw,
          COUNT(DISTINCT CASE WHEN source='kis' THEN code END) codes_kis,
          COUNT(DISTINCT code) codes_either
        FROM ws_ticks
        WHERE recv_ts >= %s AND recv_ts < %s AND source IN ('kiwoom','kis')
        """,
        (start_s, end_s),
    ).fetchone()
    print("\n== 구간 합계 ==")
    print(dict(tot or {}))
    n_kw = int((tot or {}).get("n_kw") or 0)
    n_kis = int((tot or {}).get("n_kis") or 0)
    ratio = round(n_kis / n_kw, 3) if n_kw else None
    print("kis/kiwoom 행수비:", ratio, "(1에 가까울수록 밀도 비슷. 예전에 ~1.7배면 한투가 더 촘촘)")

    split = db.conn.execute(
        """
        SELECT
          SUM(srcs='kis') only_kis,
          SUM(srcs='kiwoom') only_kw,
          SUM(srcs='both') both_n
        FROM (
          SELECT code,
                 CASE
                   WHEN SUM(source='kis')>0 AND SUM(source='kiwoom')>0 THEN 'both'
                   WHEN SUM(source='kis')>0 THEN 'kis'
                   ELSE 'kiwoom'
                 END srcs
          FROM ws_ticks
          WHERE recv_ts >= %s AND recv_ts < %s AND source IN ('kiwoom','kis')
          GROUP BY code
        ) z
        """,
        (start_s, end_s),
    ).fetchone()
    print("종목 커버 키움만/한투만/둘다:", dict(split or {}))

    code_filter = str(args.code or "").strip()
    if code_filter:
        codes = [code_filter]
    else:
        rows = db.conn.execute(
            """
            SELECT code,
                   SUM(source='kiwoom') n_kw,
                   SUM(source='kis') n_kis
            FROM ws_ticks
            WHERE recv_ts >= %s AND recv_ts < %s AND source IN ('kiwoom','kis')
            GROUP BY code
            HAVING SUM(source='kiwoom') > 0 AND SUM(source='kis') > 0
            ORDER BY (SUM(source='kis') / GREATEST(SUM(source='kiwoom'), 1)) DESC
            LIMIT %s
            """,
            (start_s, end_s, int(args.limit_codes)),
        ).fetchall()
        codes = [str(r["code"]) for r in rows]
        print("\n== 겹치는 종목 kis/kw 비 상위 ==")
        for r in rows:
            d = dict(r)
            kw = int(d.get("n_kw") or 0)
            ki = int(d.get("n_kis") or 0)
            d["kis_per_kw"] = round(ki / kw, 2) if kw else None
            print(d)

    print("\n== 종목별 recv_ts 갭 (키움 vs 한투) ==")
    stall_codes = []
    for code in codes:
        raw = db.conn.execute(
            """
            SELECT source, recv_ts, tick_time
            FROM ws_ticks
            WHERE code=%s AND recv_ts >= %s AND recv_ts < %s
              AND source IN ('kiwoom','kis')
            ORDER BY recv_ts
            """,
            (code, start_s, end_s),
        ).fetchall() if has_tick_time else db.conn.execute(
            """
            SELECT source, recv_ts
            FROM ws_ticks
            WHERE code=%s AND recv_ts >= %s AND recv_ts < %s
              AND source IN ('kiwoom','kis')
            ORDER BY recv_ts
            """,
            (code, start_s, end_s),
        ).fetchall()
        by_src: Dict[str, List[datetime]] = {"kiwoom": [], "kis": []}
        by_fid: Dict[str, List[str]] = {"kiwoom": [], "kis": []}
        for r in raw:
            src = str(r.get("source") or "")
            dtv = _parse_ts(r.get("recv_ts"))
            if src in by_src and dtv is not None:
                by_src[src].append(dtv)
            if has_tick_time and src in by_fid:
                by_fid[src].append(str(r.get("tick_time") or ""))
        kw_g = _gap_stats(_gaps(by_src["kiwoom"]))
        ki_g = _gap_stats(_gaps(by_src["kis"]))
        rec = {
            "code": code,
            "n_kw": len(by_src["kiwoom"]),
            "n_kis": len(by_src["kis"]),
            "kis_per_kw": round(len(by_src["kis"]) / len(by_src["kiwoom"]), 2) if by_src["kiwoom"] else None,
            "kw_recv_gap": kw_g,
            "kis_recv_gap": ki_g,
        }
        # 키움 max gap 이 한투보다 훨씬 크면 수신 스톨 후보
        kw_max = kw_g.get("max")
        ki_max = ki_g.get("max")
        if kw_max is not None and ki_max is not None and kw_max >= 5.0 and kw_max > (ki_max * 2.0 + 1.0):
            rec["stall_candidate"] = True
            stall_codes.append(rec)
        print(rec)

        if has_tick_time and (args.code or rec.get("stall_candidate")):
            # FID tick_time 이 같은 초로 밀집 vs recv 공백
            print("  tick_time 샘플 kw", by_fid["kiwoom"][:3], "...", by_fid["kiwoom"][-3:] if by_fid["kiwoom"] else [])
            print("  tick_time 샘플 kis", by_fid["kis"][:3], "...", by_fid["kis"][-3:] if by_fid["kis"] else [])

    print("\n== 스톨 후보 (키움 recv 갭>=5s 이고 한투 갭의 2배+1 초과) ==")
    if not stall_codes:
        print("(없음)")
    else:
        for r in stall_codes:
            print(r)

    # 매도 시각 근처
    try:
        ocols = _cols(db, "orders")
        print("\nSHOW COLUMNS orders:", ocols)
        tcol = _pick_time_col(ocols)
        side_col = "side" if "side" in ocols else ("ord_side" if "ord_side" in ocols else None)
        code_col = "code" if "code" in ocols else None
        if tcol and side_col and code_col:
            for c in (tcol, side_col, code_col):
                if not str(c).replace("_", "").isalnum():
                    raise ValueError("bad col %s" % c)
            sql = (
                "SELECT `" + tcol + "` AS ts, `" + code_col + "` AS code, `"
                + side_col + "` AS side FROM orders WHERE `" + tcol
                + "` >= %s AND `" + tcol + "` < %s"
            )
            sells = db.conn.execute(sql, (start_s, end_s)).fetchall()
            sell_rows = [dict(r) for r in sells if str(r.get("side") or "").upper() in ("SELL", "01", "sell")]
            if not sell_rows:
                sell_rows = [dict(r) for r in sells]
            print("orders 구간 행", len(sells), "매도후보", len(sell_rows), "시각컬럼", tcol)
            for r in sell_rows[:20]:
                print("  ", r)
        else:
            print("orders 에서 시각/side/code 컬럼을 못 찾음 — 매도 교차는 스킵")
    except Exception as ex:
        print("orders 조회 스킵:", ex)

    print("\n판독 힌트:")
    print("- kis/kw 행수비가 1.5~2+ 이고 키움만 max recv 갭이 크면, 예전처럼 키움 수신이 멈춘 패턴.")
    print("- 틱매도 비동기 적용 후 매도 시각 전후 키움 갭이 한투와 비슷해지면 1~5 효과.")
    print("- 한투만 종목이 많으면 구독 유니버스 차이(41 vs 100)이지 스톨이 아님.")
    print("로그 파일은 이 실행 stdout 과 동일")
    return 0


if __name__ == "__main__":
    _tee_log()
    try:
        raise SystemExit(main())
    except SystemExit:
        raise
    except Exception:
        import traceback
        traceback.print_exc()
        raise SystemExit(1)