fix(factor): load段内存峰值根治——chunks在concat后即清+transform链lazy单次物化,RSS监控实锤6.6GB被kernel周期性OOM击杀,eager逐步链3-4倍峰值收敛到~1.5倍 [vps]

sanguo_factor/universe.py

@@ -84,8 +84,11 @@ def load_universe_bars(
         return _empty_alpha_df()
 
     df = pl.concat(chunks)
+    chunks.clear()  # 立即释放逐symbol块(~1G),勿持有到函数尾
+    # transform 链 lazy 单次物化:eager 逐步链每步全量拷贝,峰值3-4倍;lazy 融合后~1.5倍
     df = (
-        df.with_columns(
+        df.lazy()
+        .with_columns(
             pl.col("dt").str.slice(0, 10).str.to_datetime("%Y-%m-%d").alias("datetime"),
             (pl.col("symbol") + "." + pl.col("exchange")).alias("vt_symbol"),
             pl.when(pl.col("volume") > 0)
@@ -97,6 +100,7 @@ def load_universe_bars(
         .with_columns(pl.int_range(pl.len()).over("vt_symbol").alias("bar_idx"))
         .select(["vt_symbol", "datetime", "open", "high", "low", "close",
                  "volume", "turnover", "open_interest", "vwap", "bar_idx"])
+        .collect()
     )
     return df