看股功能更新

backend/app/data/fetcher.py

@@ -1,12 +1,15 @@
 """数据编排：拉取（Tushare 主 -> AKShare 兜底）+ 本地缓存。
 
-真实行情落库到 candles 表（timeframe='1d'），回测统一从库读。
+真实行情落库到 candles 表（timeframe='1d'，**不复权底座**），回测统一从库读。
+复权（qfq/hfq）在读取时按 adj_factor 表本地换算，见 api._adjust_bars。
 """
+
 from __future__ import annotations
 
 import asyncio
 
-from sqlalchemy import delete, func, select
+from sqlalchemy import func, select
+from sqlalchemy.dialects.postgresql import insert as pg_insert
 from sqlalchemy.ext.asyncio import AsyncSession
 
 from ..config import settings
@@ -40,6 +43,13 @@ async def is_cached(session: AsyncSession, symbol: str) -> bool:
     return await count_cached(session, symbol) > 0
 
 
+async def _last_cached_ts(session: AsyncSession, symbol: str):
+    res = await session.execute(
+        select(func.max(Candle.ts)).where(Candle.symbol == symbol, Candle.timeframe == "1d")
+    )
+    return res.scalar()
+
+
 async def sync_symbol(
     session: AsyncSession,
     code: str,
@@ -48,34 +58,57 @@ async def sync_symbol(
     source: str = "auto",
     force: bool = False,
 ) -> dict:
-    """拉取并缓存某标的日线。已缓存且非 force 时直接返回缓存计数。"""
-    if not force and await is_cached(session, code):
+    """增量拉取并 upsert 某标的日线（**不复权**底座）。
+
+    - 永不删除已有行：按 (symbol, timeframe, ts) 主键 upsert，
+      不会把 TDX 导入的 30 年历史冲掉；
+    - 已缓存时从最后一根的次日开始增量拉取（force 仅跳过「有缓存就返回」
+      的短路，用于缓存落后于最新交易日时的刷新）；
+    - 拉不到新行时保持原缓存不动。
+    """
+    last_ts = await _last_cached_ts(session, code)
+    if last_ts is not None and not force and not start:
         return {"symbol": code, "bars": await count_cached(session, code), "source": "cache"}
 
+    if last_ts is not None and not start:
+        # 增量：从缓存最后一根当天开始（重叠一天重新拉取，容忍数据源漏行/盘后修订）
+        start = last_ts.strftime("%Y%m%d")
     start = start or DEFAULT_START
-    adjust = settings.data_adjust
     errors: list[str] = []
     bars: list[Bar] = []
     used = None
 
     for name, fn in _providers(source):
         try:
-            # tushare/akshare 是同步网络 IO，丢到线程池避免阻塞事件循环
-            bars = await asyncio.to_thread(fn, code, start, end, adjust)
+            # tushare/akshare 是同步网络 IO，丢到线程池避免阻塞事件循环；
+            # adjust=None -> 不复权（复权在读取时按 adj_factor 换算）
+            bars = await asyncio.to_thread(fn, code, start, end, None)
             used = name
             break
         except Exception as e:  # noqa: BLE001
             errors.append(f"{name}: {e}")
 
     if not bars:
+        if last_ts is not None:
+            # 增量失败（如停牌/新股无新行）：保留缓存，不算错误
+            return {"symbol": code, "bars": await count_cached(session, code), "source": "cache"}
         raise RuntimeError("所有数据源均失败 -> " + " | ".join(errors) if errors else "无可用数据源")
 
-    # 全量替换该标的日线（避免重复主键）
-    await session.execute(delete(Candle).where(Candle.symbol == code, Candle.timeframe == "1d"))
-    for b in bars:
-        session.add(
-            Candle(symbol=code, timeframe="1d", ts=b.ts, open=b.open, high=b.high,
-                   low=b.low, close=b.close, volume=b.volume)
-        )
+    # upsert：不 delete，避免破坏既有底座（TDX 全量历史）
+    stmt = pg_insert(Candle).values([
+        {"symbol": code, "timeframe": "1d", "ts": b.ts, "open": b.open, "high": b.high,
+         "low": b.low, "close": b.close, "volume": b.volume,
+         "amount": b.amount, "turnover": b.turnover}
+        for b in bars
+    ])
+    stmt = stmt.on_conflict_do_update(
+        index_elements=["symbol", "timeframe", "ts"],
+        set_={"open": stmt.excluded.open, "high": stmt.excluded.high, "low": stmt.excluded.low,
+              "close": stmt.excluded.close, "volume": stmt.excluded.volume,
+              # 增量源缺失额/换手时保留库里的旧值（如 TDX 已回补的 30 年成交额）
+              "amount": func.coalesce(stmt.excluded.amount, Candle.amount),
+              "turnover": func.coalesce(stmt.excluded.turnover, Candle.turnover)},
+    )
+    await session.execute(stmt)
     await session.commit()
     return {"symbol": code, "bars": len(bars), "source": used}