perf(data): 四热路径日期区间SARGable化——substr(datetime,1,10)对索引列套函数打不进复合索引datetime列,每股扫全量日线史取短窗——08-25晨9:30生产实锤:momentum选股1/2(RPS池)174s未达<60s验收,阶段日志精确定罪(过滤段含双seek仅7.8s);病根=PanelFetcher的30天窗SQL每只股扫~5000行日线史取~22行,3226只≈1600万行=174s量级吻合。修=裸列datetime>=start AND datetime<end+1天排他上界(与按日期前10位比较在d裸日期/15m·5m时间戳两格式下语义严格等价,含end当日全部行排除次日),区间打进复合索引第4列每股只扫窗口行。四处同病同修:PanelFetcher宽表/PriceFetcher逐只get_price/limit-status近2根90天窗(原substr版90天下界同样打不进索引)/datareader CTA回测K线。+3测试:防回潮扫描(三文件钉死禁substr谓词)+双格式边界行为(d裸日期与15m时间戳end当日含次日排)+get_price同语义;portfolio+data_platform 644绿;待VPS探针终验计时 [vps]

2026-08-25 20:54:40 +08:00
parent 841ea1536e
commit 3304ff46b2
4 changed files with 101 additions and 13 deletions
@@ -242,3 +242,62 @@ class TestOptionalContract:
            fields=["close", "acc_net_value"],
        )
        assert df["acc_net_value"].isna().all()
+
+
+# ======================== SARGable 日期区间(2026-08-25 P0) ========================
+
+class TestSargableDateRange:
+    """substr(datetime,1,10) 对索引列套函数 → 日期区间打不进复合索引 datetime 列,
+    每股扫全量日线史取短窗(momentum RPS 池 3226 只 30 天窗 VPS 实测 174s;provider
+    /datareader 四热路径同病)。裸列 ``datetime>=start AND datetime<end+1天`` 与
+    substr 语义严格等价(d 裸日期/带时间戳两格式, 含 end 当日全部行、排除次日)。"""
+
+    def test_no_substr_datetime_regression(self):
+        """钉死四个热路径文件不再回潮非 SARGable 谓词。"""
+        import pathlib
+        root = pathlib.Path(__file__).resolve().parents[2]
+        for rel in (
+            "sanguo_portfolio/providers/fetchers/price.py",
+            "sanguo_portfolio/providers/local_unified_provider.py",
+            "sanguo_data/datareader.py",
+        ):
+            src = (root / rel).read_text(encoding="utf-8")
+            assert "substr(datetime" not in src, f"{rel} 回潮非 SARGable 谓词"
+
+    def test_panel_end_date_inclusive_bare_and_timestamp(self, unified_provider):
+        """end 当日全部行含(d 裸日期 + 15m 时间戳两格式), 次日排除。"""
+        conn = sqlite3.connect(unified_provider.db_path)
+        conn.executemany("INSERT INTO dbbardata VALUES(?,?,?,?,?,?,?,?,?,?,?)", [
+            # d 裸日期: end 当日含 / 次日排
+            ("600600", "SSE", "2024-06-19", "d", 100, 1e3, 0, 1.0, 1.0, 1.0, 10.0),
+            ("600600", "SSE", "2024-06-21", "d", 100, 1e3, 0, 1.0, 1.0, 1.0, 12.0),
+            # 15m 时间戳: end 当日含 / 次日排
+            ("600600", "SSE", "2024-06-20 14:35:00", "15m", 100, 1e3, 0, 1.0, 1.0, 1.0, 11.0),
+            ("600600", "SSE", "2024-06-21 09:35:00", "15m", 100, 1e3, 0, 1.0, 1.0, 1.0, 13.0),
+        ])
+        conn.commit()
+        conn.close()
+        # d 面: 窗口 06-18~06-20 → 只有裸日期 06-19 行
+        panel_d = unified_provider.get_closes_panel(
+            ["600600.XSHG"], "2024-06-18", "2024-06-20", fq="raw")
+        assert [str(d)[:10] for d in panel_d.index] == ["2024-06-19"]
+        assert panel_d.iloc[0, 0] == 10.0
+        # 15m 面: 同窗 → 只有 06-20 14:35 行(end 当日带时间戳不被右界误伤)
+        panel_15 = unified_provider.get_closes_panel(
+            ["600600.XSHG"], "2024-06-18", "2024-06-20", interval="15m", fq="raw")
+        assert [str(d)[:16] for d in panel_15.index] == ["2024-06-20 14:35"]
+        assert panel_15.iloc[0, 0] == 11.0
+
+    def test_get_price_end_date_inclusive_bare_date(self, unified_provider):
+        """PriceFetcher 逐只路径同语义: 裸日期 end 当日含、次日排。"""
+        conn = sqlite3.connect(unified_provider.db_path)
+        conn.executemany("INSERT INTO dbbardata VALUES(?,?,?,?,?,?,?,?,?,?,?)", [
+            ("600600", "SSE", "2024-06-19", "d", 100, 1e3, 0, 1.0, 1.0, 1.0, 10.0),
+            ("600600", "SSE", "2024-06-21", "d", 100, 1e3, 0, 1.0, 1.0, 1.0, 12.0),
+        ])
+        conn.commit()
+        conn.close()
+        df = unified_provider.get_price(
+            "600600.XSHG", start_date="2024-06-18", end_date="2024-06-20")
+        assert len(df) == 1
+        assert df.iloc[-1]["close"] == 10.0