diff --git a/scripts/data_platform/ak_stock_wrapper.ps1 b/scripts/data_platform/ak_stock_wrapper.ps1 index 4885ba1..a750053 100644 --- a/scripts/data_platform/ak_stock_wrapper.ps1 +++ b/scripts/data_platform/ak_stock_wrapper.ps1 @@ -1,5 +1,7 @@ -# ak_stock_wrapper.ps1 — sanguo-ak-weekly schtask wrapper (weekly 周六03:00 akshare per-stock 慢) -# northbound+share_capital+top_holders+gdhs --force; per-stock 全量慢, 周末夜间; +# ak_stock_wrapper.ps1 — sanguo-ak-weekly schtask wrapper (weekly 周六03:00 akshare per-stock 慢) +# --force 全量刷新; 顺序=gdhs,share_capital,top_holders,northbound (09-06 调序: +# 09-05 northbound 1117/2883 无数据标的连败触发断路器, 饿死后续 3 类型—— +# 最大最险的 northbound 挪最后, 即使再断路也不连累前面); # gdhs=股东户数 per-period 全市场单次 (09-02 入列, 披露守卫>90天季度末, marker 去重无新期零网络) $env:http_proxy = '' $env:https_proxy = '' @@ -9,5 +11,5 @@ $ts = Get-Date -Format 'yyyyMMdd_HHmmss' $logDir = 'C:\sanguo_vnpy_v2\data\migration_logs' if (-not (Test-Path $logDir)) { New-Item -ItemType Directory -Path $logDir -Force | Out-Null } $log = Join-Path $logDir "ak_stock_$ts.txt" -C:\Python310\python.exe -X utf8 C:\sanguo_vnpy_v2\scripts\data_platform\akshare_static_download.py --types northbound,share_capital,top_holders,gdhs --force *>> $log +C:\Python310\python.exe -X utf8 C:\sanguo_vnpy_v2\scripts\data_platform\akshare_static_download.py --types gdhs,share_capital,top_holders,northbound --force *>> $log exit $LASTEXITCODE diff --git a/scripts/data_platform/akshare_static_download.py b/scripts/data_platform/akshare_static_download.py index 2fbbfe4..4891eeb 100644 --- a/scripts/data_platform/akshare_static_download.py +++ b/scripts/data_platform/akshare_static_download.py @@ -496,10 +496,44 @@ def fetch_valuation(symbol: str) -> Optional[pd.DataFrame]: return df +NORTHBOUND_COLUMNS = [ + "持股日期", "当日收盘价", "当日涨跌幅", "持股数量", "持股市值", + "持股数量占A股百分比", "今日增持股数", "今日增持资金", "今日持股市值变化", +] + + +def _safe_northbound_em(symbol: str) -> pd.DataFrame: + """akshare stock_hsgt_individual_em 包装: 容忍无北向数据标的的崩溃响应。 + + bug 根因 (实证 akshare 1.18.x stock_hsgt_em.py:1446, 2026-09-05 ak-weekly + 断路器事故): 无北向数据标的 (301xxx 等 ~1117/2883) datacenter 返 + result=None, __stock_hsgt_zh_a_em 的 data_json["result"]["pages"] 抛 + TypeError("'NoneType' object is not subscriptable") → failed → 30 连败 + 断路器 rc=2, 饿死后续 share_capital/top_holders/gdhs。端点健康 + (600519 正常返 1683 行)。 + + 子串匹配 "not subscriptable" (稳定签名, 同 _safe_top_10_em 手法): + → 返空 df (带 NORTHBOUND_COLUMNS schema), 走「无北向持仓」empty + 写空+done (脚本头部既定语义), 不消耗重试/断路器配额; + 其他 TypeError 透传给 call_ak_with_retry 走重试。 + """ + try: + return ak.stock_hsgt_individual_em(symbol=symbol) + except TypeError as e: + if "not subscriptable" in str(e): + logger.debug( + "stock_hsgt_individual_em(%s) → 空 (无北向数据), 返空 df", symbol) + return pd.DataFrame(columns=NORTHBOUND_COLUMNS) + raise + + def fetch_northbound(symbol: str) -> Optional[pd.DataFrame]: - """stock_hsgt_individual_em(symbol='600519') — 北向持股 (~1700行/股)。""" + """stock_hsgt_individual_em(symbol='600519') — 北向持股 (~1700行/股)。 + + 通过 _safe_northbound_em 包装: 无北向数据标的 akshare 抛 NoneType + not subscriptable, 捕获返空 df (empty 语义, 不耗断路器配额)。""" df, _status = call_ak_with_retry( - ak.stock_hsgt_individual_em, f"northbound/{symbol}", symbol=symbol, + _safe_northbound_em, f"northbound/{symbol}", symbol=symbol, ) return df diff --git a/tests/data_platform/test_akshare_static_download.py b/tests/data_platform/test_akshare_static_download.py index a248211..25b6544 100644 --- a/tests/data_platform/test_akshare_static_download.py +++ b/tests/data_platform/test_akshare_static_download.py @@ -9,6 +9,7 @@ import argparse import pandas as pd +import pytest import akshare_static_download as mod # noqa: E402 (conftest 已加 sys.path) @@ -205,3 +206,34 @@ class TestHotAndGdhsFamily: ids = [u[0] for u in units] assert "20250930_gdhs" in ids assert "20251231_gdhs" not in ids + + +# ======================== ③ northbound 无数据崩溃安全包装 ======================== + +class TestSafeNorthbound: + """_safe_northbound_em (2026-09-05 ak-weekly 断路器事故修复)。 + + akshare stock_hsgt_individual_em 对无北向数据标的 (301xxx 等) 在 + data_json["result"]["pages"] 处抛 TypeError(NoneType not subscriptable) + —— 周六实测 2883 股中 1117 连败, 末段 30 连败触发断路器 rc=2, + 饿死后续 share_capital/top_holders/gdhs (gdhs 目录至今不存在)。 + 端点本身健康 (600519 正常返 1683 行, 09-06 晨复测)。 + 确定性无数据 → 空 df 带 schema → 「无北向持仓」empty 写空+done + (脚本头部既定语义), 不消耗重试/断路器配额 (同 _safe_top_10_em 手法)。 + """ + + def test_no_data_crash_returns_empty_with_schema(self, monkeypatch): + def boom(symbol): + raise TypeError("'NoneType' object is not subscriptable") + monkeypatch.setattr(mod.ak, "stock_hsgt_individual_em", boom) + df = mod._safe_northbound_em("301638") + assert df.empty + assert list(df.columns) == mod.NORTHBOUND_COLUMNS + + def test_other_typeerror_propagates(self, monkeypatch): + """非无数据签名的 TypeError 透传 (走 call_ak_with_retry 重试)。""" + def boom(symbol): + raise TypeError("connection reset by peer") + monkeypatch.setattr(mod.ak, "stock_hsgt_individual_em", boom) + with pytest.raises(TypeError): + mod._safe_northbound_em("301638")