fix(data): 静态表下载两隐患根治——①北交所白烧重试跳过(东财三报表端点SH920xxx必败'NoneType'错,09-01季频洗库实测340股×3表×3retry+退避白烧≈5h;对齐top_holders BJ_PREFIXES同款跳过,valuation/financial_abstract实测BSE有数据不跳)②fetch失败语义修正:15个fetcher不再经_df_or_empty把重试耗尽(None)转空df→空parquet覆盖旧好数据+标done永不重试(09-01实证340空文件全BSE属侥幸,主板瞬时故障会真毁数据);改返None交download_one_unit既有failed分支=不写不标保旧数据复跑重试,真空df才写空+done;test_top_holders一处旧契约断言同步(失败返None);+8测试钉死 [vps]

tests/data_platform/test_akshare_static_download.py

@@ -0,0 +1,101 @@
+# -*- coding: utf-8 -*-
+"""静态表下载两大隐患修复测试 (2026-09-01 洗库事故, 修正①②)。
+
+①: 三大报表类跳过北交所前缀 (920/43/83/87) — 东财端点必败白烧重试
+   (09-01 实测 340 股 × 3 表 × 3 retry ≈ 5h), valuation 类实测 BSE 有数据不跳。
+②: fetch 失败(重试耗尽)返 None → download_one_unit 按 failed 处理,
+   不写空 parquet 覆盖好数据、不标 done (旧版 _df_or_empty 转空 df 的暗坑)。
+"""
+import argparse
+
+import pandas as pd
+
+import akshare_static_download as mod  # noqa: E402  (conftest 已加 sys.path)
+
+
+def _args(**kw):
+    base = dict(codes=None, limit=0)
+    base.update(kw)
+    return argparse.Namespace(**base)
+
+
+# ======================== ① 北交所跳过 ========================
+
+class TestStatementBseSkip:
+    ALL_CODES = [
+        ("600519", "SH"), ("000001", "SZ"),
+        ("920985", "SH"), ("430047", "SZ"),  # 北交所新/旧代码段
+    ]
+
+    def test_statements_skip_bse(self):
+        """balance/income/cashflow 三类跳北交所, 主板保留。"""
+        for t, endpoint in (("balance", "balance_sheet"),
+                            ("income", "income_sheet"),
+                            ("cashflow", "cashflow_sheet")):
+            units = mod.build_per_stock_units(
+                t, endpoint, mod.fetch_balance_sheet, self.ALL_CODES, _args())
+            codes = {uid.split(".")[0] for uid, _ in units}
+            assert codes == {"600519", "000001"}, t
+
+    def test_valuation_keeps_bse(self):
+        """valuation/financial_abstract 实测 BSE 有数据, 不跳。"""
+        for t in ("valuation", "financial_abstract"):
+            units = mod.build_per_stock_units(
+                t, t, mod.fetch_valuation, self.ALL_CODES, _args())
+            assert len(units) == 4, t
+
+    def test_codes_filter_bse_only_yields_zero(self):
+        """--codes 显式指定北交所 + 报表类 → 0 unit (语义明确)。"""
+        units = mod.build_per_stock_units(
+            "balance", "balance_sheet", mod.fetch_balance_sheet,
+            [("920985", "SH")], _args(codes="920985"))
+        assert units == []
+
+
+# ======================== ② 失败不覆盖好数据 ========================
+
+class TestFailedNoOverwrite:
+
+    def test_failed_fetch_writes_nothing(self, tmp_path, monkeypatch):
+        """fetch 返 None (重试耗尽) → failed, parquet/marker 都不落。"""
+        monkeypatch.setattr(mod, "OUT_DIR", tmp_path)
+        status, rows = mod.download_one_unit(
+            "balance", "600519.SH_balance", lambda: None, force=True)
+        assert status == "failed"
+        assert not (tmp_path / "balance" / "600519.SH_balance.parquet").exists()
+        assert not (tmp_path / "balance" / ".600519.SH_balance.akshare").exists()
+
+    def test_genuine_empty_writes_and_marks(self, tmp_path, monkeypatch):
+        """端点成功返真空 df → 写空 parquet + marker ("查过了确实无数据")。"""
+        monkeypatch.setattr(mod, "OUT_DIR", tmp_path)
+        status, _ = mod.download_one_unit(
+            "balance", "600519.SH_balance",
+            lambda: pd.DataFrame(), force=True)
+        assert status == "empty"
+        assert (tmp_path / "balance" / "600519.SH_balance.parquet").exists()
+        assert (tmp_path / "balance" / ".600519.SH_balance.akshare").exists()
+
+    def test_ok_fetch_writes_data(self, tmp_path, monkeypatch):
+        monkeypatch.setattr(mod, "OUT_DIR", tmp_path)
+        status, rows = mod.download_one_unit(
+            "balance", "600519.SH_balance",
+            lambda: pd.DataFrame({"v": [1, 2]}), force=True)
+        assert status == "ok" and rows == 2
+        assert len(pd.read_parquet(
+            tmp_path / "balance" / "600519.SH_balance.parquet")) == 2
+
+    def test_fetchers_return_none_when_retry_exhausted(self, monkeypatch):
+        """三类模式 (per-stock/per-date/per-period) 的 fetcher 失败统一返 None。"""
+        monkeypatch.setattr(
+            mod, "call_ak_with_retry",
+            lambda *a, **k: (None, "failed"))
+        assert mod.fetch_balance_sheet("SH600519") is None
+        assert mod.fetch_dragon_tiger("20260901") is None
+        assert mod.fetch_forecast("20260630") is None
+
+    def test_fetchers_return_df_on_ok(self, monkeypatch):
+        df = pd.DataFrame({"a": [1]})
+        monkeypatch.setattr(mod, "call_ak_with_retry",
+                            lambda *a, **k: (df, "ok"))
+        assert mod.fetch_income_sheet("SH600519") is df
+        assert mod.fetch_express("20260630") is df

tests/data_platform/test_top_holders_parse.py

@@ -194,8 +194,9 @@ class TestFetchTopHoldersOnePeriod:
         assert call_count["n"] == mod.AK_MAX_RETRIES, (
             f"网络异常应重试 {mod.AK_MAX_RETRIES} 次, 实际 {call_count['n']} 次"
         )
-        # 重试耗尽返空 df (无 schema, 因为确实是失败)
-        assert isinstance(df, pd.DataFrame)
+        # 重试耗尽返 None (2026-09-01 语义修正: 失败≠空数据, download_one_unit
+        # 按 failed 处理, 不写空 parquet 覆盖好数据/不标 done)
+        assert df is None
 
 
 class TestSafeTop10EmKeyErrorSdltgd: