feat(data): 静态表调度挪周末+vintage盘前自检——③sanguo-ak-quarter月内1号02:00改财报月首个周六02:00(/d SAT,周六02:00→周一09:30有55h跑道;旧排程7.5h夜窗连单表8h都装不下,09-01注册后首爆16h+穿整个交易日,09:32决策窗撞balance(94%Q2'26)/income(0%)vintage撕裂致value ROE归零清仓;月内1-4号读旧但一致,策略配对修复侧兼容)④新增static_vintage_check.py每日09:00自检:三报表最大报告期+覆盖率盘点(REPORT_DATE列裁剪扫描),跨表撕裂/估值断更(>48h)告警exit 3,原子落vintage_status.json供策略盘前预判(策略session协作契约);register加sanguo-ak-vintage+新wrapper;+8测试 [vps]

2026-09-01 19:03:53 +08:00
parent aa7e7b8eba
commit 8dc3f494b7
5 changed files with 328 additions and 4 deletions
@@ -0,0 +1,121 @@
+# -*- coding: utf-8 -*-
+"""静态表 vintage 一致性自检测试 (2026-09-01 洗库撕裂事故防线④)。
+
+核心契约: 跨表最大报告期不一致 → 告警 (退出码 3) + vintage_status.json
+落盘 (策略侧盘前预判数据源); 一致 → OK (退出码 0); 估值断更 → 告警。
+"""
+import json
+import os
+import time
+
+import pandas as pd
+
+import static_vintage_check as svc  # noqa: E402  (conftest 已加 sys.path)
+
+
+def _mk_pq(path, periods):
+    df = pd.DataFrame({"REPORT_DATE": list(periods), "v": range(len(periods))})
+    df.to_parquet(path, index=False)
+
+
+def _mk_table(root, name, stocks_periods):
+    """stocks_periods: {filename_stem: [报告期,...]}"""
+    d = root / name
+    d.mkdir(parents=True, exist_ok=True)
+    for stem, periods in stocks_periods.items():
+        _mk_pq(d / f"{stem}.parquet", periods)
+    return d
+
+
+# ======================== 单表扫描 ========================
+
+class TestScanStatementTable:
+
+    def test_basic_distribution(self, tmp_path):
+        d = _mk_table(tmp_path, "balance", {
+            "a": ["2025-12-31", "2026-06-30"],
+            "b": ["2026-06-30"],
+            "c": ["2026-03-31"],  # 滞后股 (未披露中报)
+        })
+        (d / "empty.parquet").write_bytes(b"x" * 10)  # <1KB 空文件
+        s = svc.scan_statement_table(d)
+        assert s["max_period"] == "2026-06-30"
+        assert s["n_valid"] == 3 and s["n_at_max"] == 2
+        assert s["n_empty"] == 1 and s["n_unreadable"] == 0
+        assert s["coverage"] == round(2 / 3, 4)  # coverage 保留 4 位小数
+
+    def test_datetime_dtype_period_normalized(self, tmp_path):
+        """真实数据 REPORT_DATE 形如 '2026-06-30 00:00:00' → 归一 YYYY-MM-DD。"""
+        _mk_table(tmp_path, "income", {
+            "a": ["2026-06-30 00:00:00"],
+        })
+        s = svc.scan_statement_table(tmp_path / "income")
+        assert s["max_period"] == "2026-06-30"
+
+    def test_missing_dir(self, tmp_path):
+        s = svc.scan_statement_table(tmp_path / "nope")
+        assert "error" in s and s["max_period"] is None
+
+
+# ======================== 跨表一致性 / 告警 ========================
+
+class TestCheckAll:
+
+    def test_vintage_tear_alert(self, tmp_path):
+        """09-01 事故签名: balance 已到 Q2'26, income 停在 Q1'26 → 撕裂告警。"""
+        _mk_table(tmp_path, "balance", {"a": ["2026-06-30"]})
+        _mk_table(tmp_path, "income", {"a": ["2026-03-31"]})
+        _mk_table(tmp_path, "cashflow", {"a": ["2026-06-30"]})
+        _, alerts = svc.check_all(tmp_path)
+        assert any("撕裂" in a and "income=2026-03-31" in a for a in alerts)
+
+    def test_consistent_ok(self, tmp_path):
+        _mk_table(tmp_path, "balance", {"a": ["2026-06-30"]})
+        _mk_table(tmp_path, "income", {"a": ["2026-06-30"]})
+        _mk_table(tmp_path, "cashflow", {"a": ["2026-06-30"]})
+        _, alerts = svc.check_all(tmp_path)
+        assert alerts == []
+
+    def test_valuation_stale_alert(self, tmp_path):
+        """估值表 (每日 ak-eod) 超 48h 未刷新 → 断更告警。"""
+        _mk_table(tmp_path, "balance", {"a": ["2026-06-30"]})
+        _mk_table(tmp_path, "income", {"a": ["2026-06-30"]})
+        _mk_table(tmp_path, "cashflow", {"a": ["2026-06-30"]})
+        d = _mk_table(tmp_path, "valuation", {"v": ["2026-08-31"]})
+        old = time.time() - 100 * 3600
+        os.utime(d / "v.parquet", (old, old))
+        _, alerts = svc.check_all(tmp_path)
+        assert any("估值表断更" in a for a in alerts)
+
+
+# ======================== main / JSON 契约 ========================
+
+class TestMain:
+
+    def test_ok_exit0_and_json_written(self, tmp_path, monkeypatch, capsys):
+        _mk_table(tmp_path, "balance", {"a": ["2026-06-30"]})
+        _mk_table(tmp_path, "income", {"a": ["2026-06-30"]})
+        _mk_table(tmp_path, "cashflow", {"a": ["2026-06-30"]})
+        jp = tmp_path / "vintage_status.json"
+        monkeypatch.setattr(
+            "sys.argv", ["x", "--root", str(tmp_path), "--json-out", str(jp)])
+        assert svc.main() == 0
+        payload = json.loads(jp.read_text(encoding="utf-8"))
+        assert set(payload["tables"]) == {"balance", "income", "cashflow"}
+        assert payload["tables"]["balance"]["max_period"] == "2026-06-30"
+        assert "OK" in capsys.readouterr().out
+        # 幂等重跑 (原子覆盖)
+        monkeypatch.setattr(
+            "sys.argv", ["x", "--root", str(tmp_path), "--json-out", str(jp)])
+        assert svc.main() == 0
+
+    def test_tear_exit3_with_alert_line(self, tmp_path, monkeypatch, capsys):
+        _mk_table(tmp_path, "balance", {"a": ["2026-06-30"]})
+        _mk_table(tmp_path, "income", {"a": ["2026-03-31"]})
+        _mk_table(tmp_path, "cashflow", {"a": ["2026-03-31"]})
+        jp = tmp_path / "v.json"
+        monkeypatch.setattr(
+            "sys.argv", ["x", "--root", str(tmp_path), "--json-out", str(jp)])
+        assert svc.main() == 3
+        assert "[vintage][ALERT]" in capsys.readouterr().out
+        assert jp.exists()  # 告警态也落 json (策略侧能看到撕裂详情)