fix(data): ak-stock top_holders KeyError 'sdltgd' 根治(_safe 捕 KeyError+匹配sdltgd,跳北交所920/83/87/43)

2026-07-27 21:08:53 +08:00
parent 8b2693423a
commit 955c05357a
2 changed files with 140 additions and 13 deletions
@@ -196,3 +196,117 @@ class TestFetchTopHoldersOnePeriod:
        )
        # 重试耗尽返空 df (无 schema, 因为确实是失败)
        assert isinstance(df, pd.DataFrame)
+
+
+class TestSafeTop10EmKeyErrorSdltgd:
+    """_safe_top_10_em 同样捕获 KeyError('sdltgd') → 返空 df。
+
+    bug 根因 (实证 akshare 新版):
+      北交所等不支持的标的, akshare 内部返缺 'sdltgd' 字段 → 抛 KeyError('sdltgd')。
+      KeyError 不是 ValueError 子类, 旧版只 except ValueError 捕不到, 透传到
+      call_ak_with_retry 走 3 次重试 (14s 退避) + 噪声日志。
+    本测试集覆盖: KeyError('sdltgd') 被吞返空 df / 其他 KeyError 不被吞 / 集成路径无重试。
+    """
+
+    def test_keyerror_sdltgd_returns_empty_df_with_schema(self):
+        """akshare 抛 KeyError('sdltgd') → 返空 df 带 8 列 schema。"""
+        from akshare_static_download import _safe_top_10_em
+
+        # akshare 实际抛 KeyError('sdltgd'), str(err) == "'sdltgd'" (含引号)
+        err = KeyError("sdltgd")
+        with patch.object(mod.ak, "stock_gdfx_free_top_10_em", side_effect=err):
+            df = _safe_top_10_em(symbol="sh920510", date="20250930")
+
+        assert isinstance(df, pd.DataFrame)
+        assert len(df) == 0
+        assert df.columns.tolist() == EXPECTED_COLS
+
+    def test_keyerror_sdltgd_does_not_raise(self):
+        """KeyError('sdltgd') 必须被吞掉 (call_ak_with_retry 才不会重试)。"""
+        from akshare_static_download import _safe_top_10_em
+
+        err = KeyError("sdltgd")
+        with patch.object(mod.ak, "stock_gdfx_free_top_10_em", side_effect=err):
+            # 不应抛
+            df = _safe_top_10_em(symbol="sh920510", date="20250930")
+            assert df.empty
+
+    def test_other_keyerror_reraises(self):
+        """非 'sdltgd' 的 KeyError 不应被吞 (让重试机制处理)。"""
+        from akshare_static_download import _safe_top_10_em
+
+        # 任意其他 KeyError (网络层错误等), 不应被吞
+        other_err = KeyError("some_other_field")
+        with patch.object(mod.ak, "stock_gdfx_free_top_10_em", side_effect=other_err):
+            with pytest.raises(KeyError, match="some_other_field"):
+                _safe_top_10_em(symbol="sh920510", date="20250930")
+
+    def test_keyerror_sdltgd_no_retry_in_fetch(self, monkeypatch):
+        """KeyError('sdltgd') 走集成层也不重试 (call_ak_with_retry 只调一次)。"""
+        monkeypatch.setattr(mod, "RETRY_BACKOFF", [0, 0, 0])
+
+        call_count = {"n": 0}
+
+        def side_effect(*args, **kwargs):
+            call_count["n"] += 1
+            raise KeyError("sdltgd")
+
+        with patch.object(mod.ak, "stock_gdfx_free_top_10_em", side_effect=side_effect):
+            df = mod.fetch_top_holders_one_period(symbol="sh920510", period="20250930")
+
+        # 关键: 只调 1 次 (KeyError 被吞, 不重试)
+        assert call_count["n"] == 1, (
+            f"KeyError('sdltgd') 应该被 _safe_top_10_em 一次吞掉, 但调了 {call_count['n']} 次"
+        )
+        assert df.empty
+        assert df.columns.tolist() == EXPECTED_COLS
+
+
+class TestBuildTopHoldersUnitsSkipBJ:
+    """build_top_holders_units 跳过北交所 920xxx / 83xxx / 87xxx / 43xxx。
+
+    根因: akshare 东财 stock_gdfx_free_top_10_em 不支持北交所。每只北交所 × 20 期 × 3 retry
+    ≈ 60 次失败调用/股 → weekly 跑不完。在 build 阶段直接跳过, 双保险。
+    """
+
+    def _make_args(self, codes: str = ""):
+        import argparse
+        return argparse.Namespace(codes=codes, limit=0)
+
+    def test_skips_920_prefix(self):
+        """920xxx (北交所新段) 不进 units。"""
+        codes = [("920510", "BJ"), ("600519", "SH")]
+        units = mod.build_top_holders_units(codes, self._make_args())
+
+        unit_ids = [uid for uid, _ in units]
+        # 920510 应该被跳, 不出现
+        assert not any("920510" in uid for uid in unit_ids), (
+            f"北交所 920510 不该进 units, 但出现了: {unit_ids}"
+        )
+        # 600519 应该在 (每报告期一个 unit)
+        assert any("600519" in uid for uid in unit_ids)
+
+    def test_skips_all_bj_prefixes(self):
+        """83xx/87xx/43xx (北交所历史段) 全跳。"""
+        codes = [
+            ("830789", "BJ"),
+            ("870866", "BJ"),
+            ("430139", "BJ"),
+            ("000001", "SZ"),
+        ]
+        units = mod.build_top_holders_units(codes, self._make_args())
+        unit_ids = [uid for uid, _ in units]
+
+        for bj_code in ("830789", "870866", "430139"):
+            assert not any(bj_code in uid for uid in unit_ids), (
+                f"北交所 {bj_code} 不该进 units"
+            )
+        # 000001 应该在
+        assert any("000001" in uid for uid in unit_ids)
+
+    def test_no_bj_codes_arg_mixed(self):
+        """--codes "920001,600519" 时 920001 跳, 600519 拉。"""
+        units = mod.build_top_holders_units([], self._make_args(codes="920001,600519"))
+        unit_ids = [uid for uid, _ in units]
+        assert not any("920001" in uid for uid in unit_ids)
+        assert any("600519" in uid for uid in unit_ids)