auto-sync: 2026-05-05 11:25:53

data_platform/backfill_15min_baostock.py

@@ -92,14 +92,10 @@ def code_to_baostock(code: str) -> Tuple[str, str]:
 
 
 def is_backfilled(parquet_path: Path) -> bool:
-    """检查文件是否已经被BaoStock回补过（通过元数据标记）"""
-    if not parquet_path.exists():
-        return False
-    try:
-        df = pd.read_parquet(parquet_path)
-        return df.attrs.get("source") == "baostock"
-    except Exception:
-        return False
+    """检查文件是否已经被BaoStock回补过"""
+    # 用标记文件判断
+    marker = parquet_path.parent / f".{parquet_path.stem}.baostock"
+    return marker.exists()
 
 
 def load_progress() -> set:
@@ -167,9 +163,6 @@ def fetch_bs_15min(bs_code: str, start_date: str, end_date: str) -> Optional[pd.
     if bad_ohlc.any():
         df = df[~bad_ohlc]
 
-    # 标记数据来源
-    df.attrs["source"] = "baostock"
-
     return df if not df.empty else None
 
 
@@ -209,8 +202,10 @@ def backfill_one(code: str, start_date: str, end_date: str, force: bool = False)
     # 写入新文件
     try:
         df_new = df_new.sort_values("day").reset_index(drop=True)
-        # 保存时把attrs写入parquet metadata
         df_new.to_parquet(parquet_path, index=False)
+        # 写标记文件
+        marker = parquet_path.parent / f".{parquet_path.stem}.baostock"
+        marker.write_text(datetime.now().isoformat())
         return "ok", len(df_new)
     except Exception as e:
         logger.error("写入 %s 失败: %s", code, e)