fix(issue-1894): [bug]-回测时获取数据异常 (#1895)

2026-09-20 10:53:33 +08:00 · 2026-07-03 19:53:31 +08:00
parent 5a6c08b5f1
commit d70af7eafb
5 changed files with 100 additions and 8 deletions
--- a/data_provider/base.py
+++ b/data_provider/base.py
@@ -76,6 +76,7 @@ def normalize_stock_code(stock_code: str) -> str:
    - 'SH600519'    -> '600519'   (strip SH prefix)
    - 'SH.600519'   -> '600519'   (strip SH. prefix)
    - 'SZ000001'    -> '000001'   (strip SZ prefix)
+    - 'SS600519'    -> '600519'   (strip legacy Yahoo Shanghai prefix)
    - 'SZ.000001'   -> '000001'   (strip SZ. prefix)
    - 'BJ920748'    -> '920748'   (strip BJ prefix, BSE)
    - 'BJ.920748'   -> '920748'   (strip BJ. prefix, BSE)
@@ -103,15 +104,15 @@ def normalize_stock_code(stock_code: str) -> str:
        if candidate.isdigit() and 1 <= len(candidate) <= 5:
            return f"HK{candidate.zfill(5)}"

-    # Strip SH/SZ prefix (e.g. SH600519 -> 600519)
-    if upper.startswith(('SH', 'SZ')) and not upper.startswith('SH.') and not upper.startswith('SZ.'):
+    # Strip SH/SZ/SS prefix (e.g. SH600519 -> 600519, SS600519 -> 600519)
+    if upper.startswith(('SH', 'SZ', 'SS')) and not upper.startswith(('SH.', 'SZ.', 'SS.')):
        candidate = code[2:]
        # Only strip if the remainder looks like a valid numeric code
        if candidate.isdigit() and len(candidate) in (5, 6):
            return candidate

-    # Strip dotted SH/SZ prefix (e.g. SH.600519 -> 600519)
-    if upper.startswith(('SH.', 'SZ.')):
+    # Strip dotted SH/SZ/SS prefix (e.g. SH.600519 -> 600519)
+    if upper.startswith(('SH.', 'SZ.', 'SS.')):
        candidate = code[3:]
        if candidate.isdigit() and len(candidate) in (5, 6):
            return candidate
--- a/docs/CHANGELOG.md
+++ b/docs/CHANGELOG.md
@@ -37,6 +37,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/).
 - [改进] 台股报告完整消费三大法人：tw 个股报告的 `institution` 区块现会在报告中渲染三大法人净买卖超表格，并注入 LLM 分析 prompt 作为台股筹码过滤器（此前仅接入数据层，报告与 prompt 均未消费，导致报告出现「筹码结构：数据缺失」）；同时三大法人整市场抓取改用剩余 stage 预算而非较小的 per-symbol fetch 超时，避免单股/首档分析因冷抓取（~4-5s）超时而降级为 not_supported。tw-only、严格 additive、fail-open。
 - [修复] 台股财务金额币别标示：TWD 金额此前落入默认「元」(在 A 股语境易误读为人民币)，`_CURRENCY_SUFFIX` 补入 TWD→「新台币」，营业收入/归母净利润/经营现金流/每股现金分红均正确标注新台币。
 - [改进] 台股三大法人 fetcher 韧性加固：(1) 接入熔断器（复用 `realtime_types.CircuitBreaker`，按市场 twse/tpex 分流，连续失败 3 次→冷却 ~5min→半开探测），TWSE/TPEx 端点异常时快速跳过网络往返并 fail-open，避免端点故障时每档个股都付 timeout+throttle；(2) TPEx OpenAPI 仅服务最新交易日，调用方传入与服务日期不符的明确日期时改为 fail-open（返回无数据），避免静默返回错日资料。
+- [修复] 回测日线补全将 `605066.SH`、`SS605066`、`SS.605066` 等 A 股等价代码归一为裸代码抓取和写入，避免误向数据源请求 `SS605066` 导致回测数据不足。

 - [修复] 台股（tw）市场阶段（`market_phase`）新增收盘集合竞价识别：`_CLOSING_AUCTION_WINDOW_MINUTES` 缺 `tw` 键时 `.get(market, 0)` 得零宽窗口，TWSE/TPEx 13:25–13:30 的 5 分钟收盘竞价此前永远无法判定为 `closing_auction`（收盘前一刻仍 `intraday`、13:30 直接 `postmarket`）；补 `"tw": 5` 修正，附阶段边界回归测试。仅 tw 加项，cn/hk/us 与 jp/kr 行为不变。
 - [新功能] 新增 AI 建议决策风格重评估预览接口与页面预览。
--- a/src/services/backtest_service.py
+++ b/src/services/backtest_service.py
@@ -154,7 +154,7 @@ class BacktestService:
                )

                if len(forward_bars) < int(eval_window_days):
-                    for fill_code in self._ordered_candidate_codes(
+                    for fill_code in self._ordered_daily_refill_codes(
                        code_candidates=daily_code_candidates,
                        preferred_code=matched_daily_code,
                    ):
@@ -495,6 +495,33 @@ class BacktestService:
            return [normalized_preferred] + [code for code in ordered if code != normalized_preferred]
        return ordered

+    @staticmethod
+    def _normalize_daily_refill_code(code: Optional[str]) -> str:
+        raw_code = str(code or "").strip()
+        if not raw_code:
+            return ""
+        return canonical_stock_code(normalize_stock_code(raw_code))
+
+    @staticmethod
+    def _ordered_daily_refill_codes(
+        *,
+        code_candidates: List[str],
+        preferred_code: Optional[str] = None,
+    ) -> List[str]:
+        ordered = BacktestService._ordered_candidate_codes(
+            code_candidates=code_candidates,
+            preferred_code=preferred_code,
+        )
+        refill_codes: List[str] = []
+        seen: set[str] = set()
+        for code in ordered:
+            refill_code = BacktestService._normalize_daily_refill_code(code)
+            if not refill_code or refill_code in seen:
+                continue
+            seen.add(refill_code)
+            refill_codes.append(refill_code)
+        return refill_codes
+
    def _get_forward_bars_by_candidates(
        self,
        *,
@@ -908,6 +935,10 @@ class BacktestService:
        return None

    def _try_fill_daily_data(self, *, code: str, analysis_date: date, eval_window_days: int) -> None:
+        refill_code = self._normalize_daily_refill_code(code)
+        if not refill_code:
+            return
+
        try:
            from data_provider.base import DataFetcherManager

@@ -915,16 +946,16 @@ class BacktestService:
            end_date = analysis_date + timedelta(days=max(eval_window_days * 2, 30))
            manager = DataFetcherManager()
            df, source = manager.get_daily_data(
-                stock_code=code,
+                stock_code=refill_code,
                start_date=analysis_date.strftime("%Y-%m-%d"),
                end_date=end_date.strftime("%Y-%m-%d"),
                days=eval_window_days * 2,
            )
            if df is None or df.empty:
                return
-            self.db.save_daily_data(df, code=code, data_source=source)
+            self.db.save_daily_data(df, code=refill_code, data_source=source)
        except Exception as exc:
-            logger.warning(f"补全日线数据失败({code}): {exc}")
+            logger.warning(f"补全日线数据失败({refill_code}): {exc}")

    def _recompute_summaries(self, *, touched_codes: List[str], eval_window_days: int, engine_version: str) -> None:
        with self.db.get_session() as session:
--- a/tests/test_a_share_fetcher_code_conversion.py
+++ b/tests/test_a_share_fetcher_code_conversion.py
@@ -117,6 +117,9 @@ class TestNormalizeStockCode(unittest.TestCase):
        self.assertEqual(normalize_stock_code("sh.600519"), "600519")
        self.assertEqual(normalize_stock_code("SZ.000001"), "000001")
        self.assertEqual(normalize_stock_code("sz.000001"), "000001")
+        self.assertEqual(normalize_stock_code("SS600519"), "600519")
+        self.assertEqual(normalize_stock_code("SS.600519"), "600519")
+        self.assertEqual(normalize_stock_code("600519.SS"), "600519")
        self.assertEqual(normalize_stock_code("BJ.920748"), "920748")


--- a/tests/test_backtest_service.py
+++ b/tests/test_backtest_service.py
@@ -13,6 +13,9 @@ import unittest
 from datetime import date, datetime
 from unittest.mock import patch

+import pandas as pd
+
+from data_provider.base import normalize_stock_code
 from src.config import Config
 from src.core.backtest_engine import OVERALL_SENTINEL_CODE
 from src.repositories.backtest_repo import BacktestRepository
@@ -558,6 +561,59 @@ class BacktestServiceTestCase(unittest.TestCase):
        us_suffix_variants = BacktestRepository._build_market_code_variants("AAPL.US", "AAPL.US")
        self.assertIn("AAPL", us_suffix_variants)

+    def test_daily_refill_codes_normalize_legacy_ss_aliases_once(self) -> None:
+        candidates = BacktestService._build_daily_code_candidates("605066.SH")
+
+        self.assertIn("SS605066", candidates)
+        self.assertEqual(normalize_stock_code("SS605066"), "605066")
+        for alias in ("605066.SH", "605066", "SH605066", "SH.605066", "605066.SS", "SS.605066"):
+            with self.subTest(alias=alias):
+                self.assertEqual(BacktestService._normalize_daily_refill_code(alias), "605066")
+        self.assertEqual(
+            BacktestService._ordered_daily_refill_codes(
+                code_candidates=candidates,
+                preferred_code="605066.SH",
+            ),
+            ["605066"],
+        )
+
+    def test_try_fill_daily_data_uses_normalized_a_share_code_for_legacy_ss_alias(self) -> None:
+        requested_codes = []
+
+        class FakeDataFetcherManager:
+            def get_daily_data(self, stock_code, start_date=None, end_date=None, days=30):
+                requested_codes.append(stock_code)
+                return (
+                    pd.DataFrame(
+                        [
+                            {
+                                "date": date(2024, 7, 1),
+                                "open": 10.0,
+                                "high": 11.0,
+                                "low": 9.0,
+                                "close": 10.5,
+                                "volume": 1000,
+                            }
+                        ]
+                    ),
+                    "FakeFetcher",
+                )
+
+        service = BacktestService(self.db)
+        with patch("data_provider.base.DataFetcherManager", FakeDataFetcherManager):
+            service._try_fill_daily_data(
+                code="SS605066",
+                analysis_date=date(2024, 7, 1),
+                eval_window_days=1,
+            )
+
+        self.assertEqual(requested_codes, ["605066"])
+        self.assertEqual(
+            [row.code for row in self.db.get_data_range("605066", date(2024, 7, 1), date(2024, 7, 1))],
+            ["605066"],
+        )
+        self.assertEqual(self.db.get_data_range("SS605066", date(2024, 7, 1), date(2024, 7, 1)), [])
+
    def test_build_market_code_variants_rejects_hk_suffix_with_6_digit_base(self) -> None:
        invalid_variants = BacktestRepository._build_market_code_variants("600519.HK", "600519.HK")
        self.assertNotIn("600519", invalid_variants)