sanguo_vnpy_v2/scripts/data_platform/daily_update_xtdata.py

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""全市场日线每日增量补全（xtdata → 主库 data/{raw,qfq}）。VPS 跑（需 miniQMT 常驻）。

设计：
- 增量窗口 LOOKBACK_DAYS 天（默认 30）：覆盖周末/节假日 gap，不重写历史，省时省 IO。
- 先 download_history_data2 批量下载增量到 xtdata 本地缓存（单线程 paced，别猛打券商后端）。
- 再逐只 get_market_data_ex 读 raw(dividend_type=none)/qfq(front)，与现有主库 parquet 按 date 去重
  合并，原子写回对应年份文件（跨年自动落到两个年份目录）。
- 轻量验证：OHLC NaN 检查；失败只 warn 不阻断。
- 幂等：重复跑同一天不重复写（drop_duplicates keep='last'）。

列与 build_daily_from_xtdata.py 完全一致：date,open,high,low,close,volume（volume×100 手→股）。
datareader._row_to_bar 只读这 6 列。

用法（VPS）：
  C:\\Python310\\python.exe -X utf8 daily_update_xtdata.py
可选环境变量：LOOKBACK_DAYS（默认 30）。
退出码：0=全部成功；1=有失败但流程完成；2=致命错误（universe 拉不到等）。
"""
import os
import sys
import time
import datetime as dt

import pandas as pd

ROOT = r"C:\sanguo_vnpy_v2"
DATA = os.path.join(ROOT, "data")
LOOKBACK = int(os.environ.get("LOOKBACK_DAYS", "30"))

from xtquant import xtdata as xd

T0 = time.time()


def log(m):
    print(f"[DAILY {time.time()-T0:.0f}s] {m}", flush=True)


def prefix_of(sym):
    return "sh" if sym[:2] in ("60", "68", "51", "56", "58") else "sz"


def today_str():
    return dt.datetime.now().strftime("%Y%m%d")


def start_str():
    return (dt.datetime.now() - dt.timedelta(days=LOOKBACK)).strftime("%Y%m%d")


def read_parquet_safe(path):
    try:
        return pd.read_parquet(path)
    except Exception:  # noqa: BLE001
        return None


def atomic_write(path, df):
    os.makedirs(os.path.dirname(path), exist_ok=True)
    tmp = path + ".tmp"
    df.to_parquet(tmp, index=False)
    os.replace(tmp, path)


def fetch(code, dividend_type):
    r = xd.get_market_data_ex([], [code], period="1d", start_time=start_str(),
                              end_time=today_str(), dividend_type=dividend_type)
    return r.get(code) if r else None


def to_pdf(df):
    return pd.DataFrame({
        "date": pd.to_datetime([str(i)[:8] for i in df.index]),
        "open": df["open"].astype(float).values,
        "high": df["high"].astype(float).values,
        "low": df["low"].astype(float).values,
        "close": df["close"].astype(float).values,
        "volume": (df["volume"].astype(float) * 100).values,  # 手→股
    })


def merge_write(code, dividend_type, kind):
    """返回 (added_rows, has_nan)。added=合并后比旧文件多的行数。"""
    sym = code.split(".")[0]
    prefix = prefix_of(sym)
    df = fetch(code, dividend_type)
    if df is None or not len(df):
        return (0, False)
    pdf = to_pdf(df)
    has_nan = bool(pdf[["open", "high", "low", "close"]].isnull().any().any())
    pdf["_y"] = pdf["date"].dt.year
    added = 0
    for yr, sub in pdf.groupby("_y"):
        sub = sub.drop(columns=["_y"]).sort_values("date")
        path = os.path.join(DATA, kind, str(int(yr)), f"{prefix}{sym}_daily.parquet")
        old = read_parquet_safe(path) if os.path.exists(path) else None
        old_n = len(old) if old is not None else 0
        if old is not None and old_n:
            merged = (pd.concat([old, sub])
                      .drop_duplicates("date", keep="last")
                      .sort_values("date"))
        else:
            merged = sub
        added += max(len(merged) - old_n, 0)
        atomic_write(path, merged)
    return (added, has_nan)


def main():
    end, start = today_str(), start_str()
    log(f"start LOOKBACK={LOOKBACK} window={start}~{end}")
    u = xd.get_stock_list_in_sector("沪深A股") or []
    if not u:
        log("FATAL: empty universe（miniQMT 未连？）")
        os._exit(2)
    log(f"universe={len(u)}")

    # 1. 批量下载增量到本地缓存（paced，200/批，sleep 1s）
    BATCH = 200
    for i in range(0, len(u), BATCH):
        chunk = u[i:i + BATCH]
        try:
            xd.download_history_data2(chunk, "1d", start, end, lambda d, p: None)
        except Exception as e:  # noqa: BLE001
            log(f"dl batch@{i} err: {e}")
        time.sleep(1.0)
    log("download phase done")

    # 2. 逐只读 + 合并写
    ok = fail = warn = 0
    bars_added = 0
    latest_dates = []
    for i, code in enumerate(u):
        try:
            a1, w1 = merge_write(code, "none", "raw")
            a2, w2 = merge_write(code, "front", "qfq")
            bars_added += a1 + a2
            if a1 + a2 > 0:
                ok += 1
                # 抽样记最新日期（前几只）
                if len(latest_dates) < 5:
                    sym = code.split(".")[0]
                    prefix = prefix_of(sym)
                    yrd = dt.datetime.now().year
                    f = os.path.join(DATA, "raw", str(yrd), f"{prefix}{sym}_daily.parquet")
                    if os.path.exists(f):
                        d = read_parquet_safe(f)
                        if d is not None and len(d):
                            latest_dates.append((code, str(d["date"].max().date())))
            if w1 or w2:
                warn += 1
        except Exception as e:  # noqa: BLE001
            fail += 1
            if fail <= 5:
                log(f"write err {code}: {e}")
        if (i + 1) % 1000 == 0:
            log(f"proc {i+1}/{len(u)} ok={ok} fail={fail} warn={warn} added={bars_added}")

    log(f"DONE ok={ok} fail={fail} warn={warn} bars_added={bars_added}")
    for c, d in latest_dates:
        log(f"sample latest_date {c} -> {d}")
    sys.stdout.flush()
    os._exit(0 if fail == 0 else 1)


if __name__ == "__main__":
    main()