fix(HEL-356): complete Shenwan industry signal

Co-authored-by: multica-agent <github@multica.ai>

xiaobai-datahub/datahub/realtime_serve.py

@@ -197,17 +197,22 @@ def fetch_sector_quote(db: HubDB, code: str, expected_date: str = "") -> dict[st
     if ts_code.isdigit():
         ts_code = f"{ts_code}.SI"
     cache_key = f"sector:{ts_code}"
+    canonical_name = _canonical_sw_name(db, ts_code)
     cached = _read_cache(db, cache_key)
-    if cached is not None:
+    if cached is not None and _sector_payload_matches(cached, canonical_name):
         return cached
     errors: list[str] = []
     try:
         row = EastmoneyAdapter().fetch_shenwan_quote(ts_code)
+        if not _sector_row_matches(row, canonical_name):
+            raise AdapterError(
+                f"industry name mismatch: expected {canonical_name}, got {row.get('name') or '--'}"
+            )
         source = str(row.get("source") or "eastmoney_sw")
     except Exception as exc:
         errors.append(f"eastmoney:{exc}")
         recovered = _load_quotes_lkg(db, cache_key, expected_date)
-        if recovered is not None:
+        if recovered is not None and _sector_payload_matches(recovered, canonical_name):
             return recovered
         raise RealtimeApiError(
             "SOURCE_UNAVAILABLE",
@@ -217,7 +222,7 @@ def fetch_sector_quote(db: HubDB, code: str, expected_date: str = "") -> dict[st
     quote_date = str(row.get("quote_date") or "")
     if want and quote_date and quote_date != want:
         recovered = _load_quotes_lkg(db, cache_key, want)
-        if recovered is not None:
+        if recovered is not None and _sector_payload_matches(recovered, canonical_name):
             return recovered
         raise RealtimeApiError("SOURCE_UNAVAILABLE", f"sector quote date {quote_date} != {want}")
     payload = _envelope(
@@ -235,6 +240,29 @@ def fetch_sector_quote(db: HubDB, code: str, expected_date: str = "") -> dict[st
     return payload
 
 
+def _canonical_sw_name(db: HubDB, ts_code: str) -> str:
+    row = db.fetchone(
+        "SELECT name FROM sector_master WHERE ts_code = ? AND family = 'sw'",
+        (ts_code,),
+    )
+    if not row:
+        row = db.fetchone(
+            "SELECT name FROM eod_sector_daily WHERE ts_code = ? AND family = 'sw' "
+            "ORDER BY trade_date DESC LIMIT 1",
+            (ts_code,),
+        )
+    return str((row or {}).get("name") or "").strip()
+
+
+def _sector_row_matches(row: dict[str, Any], canonical_name: str) -> bool:
+    return not canonical_name or str(row.get("name") or "").strip() == canonical_name
+
+
+def _sector_payload_matches(payload: dict[str, Any], canonical_name: str) -> bool:
+    data = payload.get("data") if isinstance(payload, dict) else None
+    return isinstance(data, dict) and _sector_row_matches(data, canonical_name)
+
+
 def fetch_limit_pool(db: HubDB, trade_date: str = "") -> dict[str, Any]:
     day = yyyymmdd(trade_date or now_shanghai())
     cache_key = f"limit-pool:{day}"
@@ -489,18 +517,29 @@ def warm_realtime(db: HubDB) -> dict[str, Any]:
     try:
         masters = db.fetchall(
             """
-            SELECT ts_code FROM sector_master WHERE family = 'sw'
-            UNION
-            SELECT ts_code FROM eod_sector_daily WHERE family = 'sw'
-            ORDER BY ts_code
+            SELECT ts_code, name, 0 AS priority FROM sector_master WHERE family = 'sw'
+            UNION ALL
+            SELECT ts_code, name, 1 AS priority FROM eod_sector_daily WHERE family = 'sw'
+            ORDER BY priority, ts_code
             """
         )
-        codes = [str(row.get("ts_code") or "") for row in masters if row.get("ts_code")]
-        sector_rows = _eastmoney_sector_quotes(codes)
-        for row in sector_rows:
+        canonical_names: dict[str, str] = {}
+        for master in masters:
+            master_code = str(master.get("ts_code") or "")
+            master_name = str(master.get("name") or "").strip()
+            if master_code and master_name:
+                canonical_names.setdefault(master_code, master_name)
+        codes = list(canonical_names)
+        fetched_sector_rows = _eastmoney_sector_quotes(codes)
+        for row in fetched_sector_rows:
             if _row_quote_date(row, today) != today:
                 continue
             code = str(row.get("ts_code") or "").upper()
+            if not _sector_row_matches(row, canonical_names.get(code, "")):
+                result["errors"].append(
+                    f"sector:{code}:industry name mismatch"
+                )
+                continue
             payload = _envelope(
                 row,
                 {
@@ -513,6 +552,7 @@ def warm_realtime(db: HubDB) -> dict[str, Any]:
                 },
             )
             _write_cache(db, f"sector:{code}", payload, INDEX_TTL, "eastmoney_sw")
+            sector_rows.append(row)
         result["sectors"] = len(sector_rows)
         result["rows"] += len(sector_rows)
     except Exception as exc:

xiaobai-datahub/tests/test_realtime_intraday.py

@@ -9,7 +9,7 @@ from datahub.adapters.base import AdapterError
 from datahub.adapters.eastmoney import HIS_TRENDS_URL, TRENDS_URL, EastmoneyAdapter, _normalize_market_quote
 from datahub.adapters.tencent import _tencent_symbol
 from datahub.db import HubDB
-from datahub.realtime_serve import fetch_intraday
+from datahub.realtime_serve import fetch_intraday, fetch_sector_quote
 from datahub.serving import ApiError, V1API
 from datahub.timeutil import now_shanghai, yyyymmdd
 
@@ -110,6 +110,33 @@ class IntradayLkgTests(unittest.TestCase):
         self.assertIn("intraday unavailable", str(ctx.exception))
 
 
+class SectorQuoteIdentityTests(unittest.TestCase):
+    def setUp(self) -> None:
+        self.tmp = tempfile.TemporaryDirectory()
+        self.db = HubDB(Path(self.tmp.name) / "hub.db")
+        self.db.execute(
+            "INSERT INTO sector_master(ts_code,name,family,updated_at) VALUES (?,?,?,?)",
+            ("801074.SI", "专用设备", "sw", "2026-09-09T10:00:00+08:00"),
+        )
+
+    def tearDown(self) -> None:
+        self.tmp.cleanup()
+
+    def test_rejects_same_numeric_code_from_different_taxonomy(self) -> None:
+        wrong = {
+            "ts_code": "801074.SI",
+            "name": "托育服务",
+            "change": -2.19,
+            "quote_date": "20260909",
+            "source": "eastmoney_sw",
+        }
+        with patch("datahub.realtime_serve.EastmoneyAdapter") as mocked:
+            mocked.return_value.fetch_shenwan_quote.return_value = wrong
+            with self.assertRaises(Exception) as ctx:
+                fetch_sector_quote(self.db, "801074.SI", "20260909")
+        self.assertIn("industry name mismatch", str(ctx.exception))
+
+
 class ServingIntradayDateTests(unittest.TestCase):
     def setUp(self) -> None:
         self.tmp = tempfile.TemporaryDirectory()