fix(HEL-494): keep filtered reference lookups inside datahub

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

xiaobai-datahub/datahub/steward.py

@@ -202,6 +202,11 @@ def _try_published(api, api_name: str, dataset: str, params: dict[str, Any], fie
         return None
     if dataset == "stocks":
         rows = _filter_stocks(rows, params)
+        # The published master is intentionally the active list. Historical
+        # delisted/paused lookups still belong in the hub, so use its live
+        # Tushare adapter when those filters cannot be answered by the snapshot.
+        if not rows and any(params.get(key) for key in ("ts_code", "list_status", "name")):
+            return None
     if dataset == "calendar":
         rows = _filter_calendar(rows, params)
     native = _to_tushare_native(dataset, rows)
@@ -312,9 +317,12 @@ def _live_cache_key(api_name: str, params: dict[str, Any], fields: str) -> str:
 
 def _hub_query(params: dict[str, Any], **extra: Any) -> dict[str, str]:
     query = {key: str(value) for key, value in extra.items() if value not in (None, "")}
-    date = yyyymmdd(params.get("trade_date") or params.get("date") or "")
-    start = yyyymmdd(params.get("start_date") or params.get("from") or date)
-    end = yyyymmdd(params.get("end_date") or params.get("to") or date)
+    raw_date = params.get("trade_date") or params.get("date") or ""
+    date = yyyymmdd(raw_date) if raw_date else ""
+    raw_start = params.get("start_date") or params.get("from") or date
+    raw_end = params.get("end_date") or params.get("to") or date
+    start = yyyymmdd(raw_start) if raw_start else ""
+    end = yyyymmdd(raw_end) if raw_end else ""
     code = str(params.get("ts_code") or params.get("code") or "").strip()
     if code:
         query["code"] = code

xiaobai-datahub/tests/test_steward.py

@@ -54,6 +54,36 @@ class StewardQueryTests(unittest.TestCase):
         self.assertEqual(payload["meta"]["source"], "tushare")
         self.assertEqual(payload["data"][0]["ts_code"], "600000.SH")
 
+    def test_stock_filter_missing_from_active_snapshot_falls_back_inside_hub(self) -> None:
+        calls = []
+
+        def transport(api_name, params, fields):
+            calls.append((api_name, dict(params)))
+            if api_name == "stock_basic" and params.get("list_status") == "D":
+                return [
+                    {
+                        "ts_code": "000627.SZ",
+                        "symbol": "000627",
+                        "name": "退市天茂",
+                        "list_status": "D",
+                        "list_date": "19961112",
+                    }
+                ]
+            return fake_transport(api_name, params, fields)
+
+        self.hub.pipeline.adapter._transport = transport
+        payload = self.hub.api.query_api(
+            {
+                "api_name": "stock_basic",
+                "params": {"list_status": "D"},
+                "fields": "ts_code,name,list_status,list_date",
+            }
+        )
+
+        self.assertEqual(payload["meta"]["source"], "tushare")
+        self.assertEqual(payload["data"][0]["list_status"], "D")
+        self.assertIn(("stock_basic", {"list_status": "D"}), calls)
+
     def test_rt_sw_k_is_blocked(self) -> None:
         from datahub.serving import ApiError