chore: establish Strategy Lab source baseline (development, not release)

1 files changed, 336 insertions, 0 deletions

diff --git a/tests/worker/test_data.py b/tests/worker/test_data.py
new file mode 100644
index 0000000..d6eeae1
--- /dev/null
+++ b/tests/worker/test_data.py
@@ -0,0 +1,336 @@
+"""Data adapter / instrument-search tests — mocked provider transport, no network."""
+import json
+import sys
+from pathlib import Path
+
+import pandas as pd
+import pytest
+
+sys.path.insert(0, str(Path(__file__).resolve().parents[1]))
+sys.path.insert(0, str(Path(__file__).resolve().parent))
+
+import worker.data as data  # noqa: E402
+from fixtures_synth import synthetic_daily, SYNTH  # noqa: E402
+
+
+# ---- live-captured provider reply shapes (see docs/search-fix.md probes) ----
+
+def _em_payload(rows):
+    return {"QuotationCodeTable": {"Data": rows, "Status": 0, "Message": "成功",
+                                   "TotalCount": len(rows)},
+            }
+
+
+def _em_row(code, name, classify, mktnum, sec_type_name):
+    return {"Code": code, "Name": name, "Classify": classify, "MktNum": mktnum,
+            "SecurityTypeName": sec_type_name, "MarketType": mktnum,
+            "QuoteID": f"{mktnum}.{code}"}
+
+
+def test_search_stock_real_shape(monkeypatch):
+    """q=600000 → real-shape eastmoney AStock row classifies as stock, identity validated."""
+    body = _em_payload([_em_row("600000", "浦发银行", "AStock", "1", "沪A")])
+    monkeypatch.setattr(data, "_http_json", lambda *a, **k: body)
+
+    def boom(*a, **k):
+        raise RuntimeError("tencent not called in this test")
+    monkeypatch.setattr(data, "_http_text", boom)
+    res = data.search_instruments("600000")
+    assert res["status"] == "ready"
+    item = res["items"][0]
+    assert item["symbol"] == "600000" and item["name"] == "浦发银行"
+    assert item["asset_type"] == "stock" and item["market"] == "cn"
+    assert item["canonical_symbol"] == "SH#600000"
+    assert item["source"] == "eastmoney"
+
+
+def test_search_etf_real_shape_with_tencent_confirmation(monkeypatch):
+    """q=510300: eastmoney Classify=Fund only counts as ETF when tencent confirms."""
+    em = _em_payload([_em_row("510300", "沪深300ETF华泰柏瑞", "Fund", "1", "基金")])
+    monkeypatch.setattr(data, "_http_json", lambda *a, **k: em)
+    tx = 'v_hint="sh~510300~\\u6caa\\u6df1300ETF\\u534e\\u6cf0\\u67cf\\u745e~hs300etfhtbr~ETF"'
+    monkeypatch.setattr(data, "_http_text", lambda *a, **k: tx)
+    res = data.search_instruments("510300")
+    etfs = [i for i in res["items"] if i["symbol"] == "510300"]
+    assert etfs and all(i["asset_type"] == "etf" for i in etfs)
+    assert {i["source"] for i in etfs} == {"eastmoney", "tencent"}
+    assert any(i["canonical_symbol"] == "SH#510300" for i in etfs)
+
+
+def test_search_etf_ambiguous_fund_excluded_without_tencent_confirmation(monkeypatch):
+    """Eastmoney Classify=Fund alone cannot distinguish ETF from LOF (probed):
+    without a tencent ETF confirmation the item must not surface, ever."""
+    em = _em_payload([_em_row("160706", "沪深300LOF", "Fund", "0", "基金")])
+    monkeypatch.setattr(data, "_http_json", lambda *a, **k: em)
+    monkeypatch.setattr(
+        data, "_http_text",
+        lambda *a, **k: 'v_hint="sz~160706~\\u6caa\\u6df1300LOF~hs300lof~LOF"')
+    res = data.search_instruments("160706")
+    assert res["status"] == "ready"
+    assert res["items"] == []  # LOF must never surface as an etf
+
+
+def test_search_lof_not_present_when_eastmoney_only(monkeypatch):
+    em = _em_payload([_em_row("160706", "沪深300LOF", "Fund", "0", "基金")])
+    monkeypatch.setattr(data, "_http_json", lambda *a, **k: em)
+
+    def boom(*a, **k):
+        raise RuntimeError("tencent down")
+    monkeypatch.setattr(data, "_http_text", boom)
+    res = data.search_instruments("160706")
+    # tencent failed but eastmoney succeeded → still ready; LOF suppressed
+    assert res["status"] == "ready"
+    assert res["items"] == []
+    assert res["providers"]["tencent"].startswith("failed")
+
+
+def test_search_index_from_chinese_query_and_class_not_code(monkeypatch):
+    """q=沪深300 → indexes come from provider class (Classify=Index / ZS tag);
+    a numeric code like 000300 is NEVER guessed to be an index by shape."""
+    em = _em_payload([
+        _em_row("000300", "沪深300", "Index", "1", "指数"),
+        _em_row("399300", "沪深300", "Index", "0", "指数"),
+        _em_row("510300", "沪深300ETF华泰柏瑞", "Fund", "1", "基金"),
+    ])
+    monkeypatch.setattr(data, "_http_json", lambda *a, **k: em)
+    tx = ('v_hint="sh~000300~\\u6caa\\u6df1300~hs300~ZS^sz~399300~\\u6caa\\u6df1300~hs300~ZS'
+          '^sh~510300~\\u6caa\\u6df1300ETF\\u534e\\u6cf0\\u67cf\\u745e~hs300etfhtbr~ETF^'
+          'sz~160706~\\u6caa\\u6df1300LOF~hs300lof~LOF"')
+    monkeypatch.setattr(data, "_http_text", lambda *a, **k: tx)
+    res = data.search_instruments("沪深300")
+    by_key = {(i["canonical_symbol"], i["asset_type"], i["source"]) for i in res["items"]}
+    assert ("SH#000300", "index", "eastmoney") in by_key
+    assert ("SZ#399300", "index", "eastmoney") in by_key
+    assert ("SH#000300", "index", "tencent") in by_key
+    assert all(i["asset_type"] in ("stock", "etf", "index") for i in res["items"])
+    assert not any("160706" == i["symbol"] and i["asset_type"] == "etf" for i in res["items"])
+
+
+def test_search_numeric_code_not_auto_index(monkeypatch):
+    """000300 with a provider stock-ish class (hypothetical) must be honored as
+    whatever the provider states — classification is never inferred."""
+    em = _em_payload([_em_row("000300", "某标的", "AStock", "0", "深A")])
+    monkeypatch.setattr(data, "_http_json", lambda *a, **k: em)
+    monkeypatch.setattr(
+        data, "_http_text",
+        lambda *a, **k: 'v_hint="sz~000300~\\u67d0\\u6807\\u7684~m?~GP-A"')
+    res = data.search_instruments("000300")
+    assert all(i["asset_type"] == "stock" for i in res["items"])
+
+
+def test_search_excludes_unsupported_classes_real_rows(monkeypatch):
+    """Bond/HK/OTC/LOF rows and tencent LOF tags are dropped, no guessing."""
+    em = _em_payload([
+        _em_row("160706", "20天津38", "Bond", "1", "债券"),
+        _em_row("00700", "腾讯控股", "HK", "116", "港股"),
+        _em_row("160706", "嘉实沪深300ETF联接A", "OTCFUND", "0", "场外基金"),
+        _em_row("600000", "浦发银行", "AStock", "1", "沪A"),
+    ])
+    monkeypatch.setattr(data, "_http_json", lambda *a, **k: em)
+    monkeypatch.setattr(
+        data, "_http_text",
+        lambda *a, **k: 'v_hint="sz~160706~\\u6caa\\u6df1300LOF~hs300lof~LOF"')
+    res = data.search_instruments("x")
+    assert res["status"] == "ready"
+    symbols = [i["symbol"] for i in res["items"]]
+    assert symbols == ["600000"]  # bond/HK/OTCFUND dropped; tencent LOF tag excluded and eastmoney Fund lacks ETF confirmation
+    assert res["items"][0]["asset_type"] == "stock"
+
+
+def test_search_chinese_name_query_matches_provider(monkeypatch):
+    em = _em_payload([_em_row("600000", "浦发银行", "AStock", "1", "沪A")])
+    captured = {}
+    monkeypatch.setattr(data, "_http_json", lambda url, params, **k: captured.update(params) or em)
+
+    def boom(*a, **k):
+        raise RuntimeError("tencent down")
+    monkeypatch.setattr(data, "_http_text", boom)
+    res = data.search_instruments("浦发银行")
+    assert captured["input"] == "浦发银行"  # provider receives the raw CJK query
+    assert res["items"][0]["name"] == "浦发银行"
+
+
+def test_search_provider_failure_is_explicit_failed_status(monkeypatch):
+    def boom(*a, **k):
+        raise RuntimeError("network down")
+    monkeypatch.setattr(data, "_http_json", boom)
+    monkeypatch.setattr(data, "_http_text", boom)
+    res = data.search_instruments("600000")
+    assert res["status"] == "failed"
+    assert res["items"] == []
+    assert res["error"]["code"] == "providers_unavailable"
+    assert "eastmoney" in res["error"]["message"]
+    assert "tencent" in res["error"]["message"]
+
+
+def test_search_provider_timeout_is_explicit_failed_status(monkeypatch):
+    import requests as _req
+    def timeout(*a, **k):
+        raise _req.exceptions.Timeout("read timed out")
+    monkeypatch.setattr(data, "_http_json", timeout)
+    monkeypatch.setattr(data, "_http_text", timeout)
+    res = data.search_instruments("600000")
+    assert res["status"] == "failed"
+    assert "Timeout" in res["error"]["message"]
+
+
+def test_search_partial_failure_is_ready_with_warning(monkeypatch):
+    def boom(*a, **k):
+        raise RuntimeError("tencent down")
+    monkeypatch.setattr(
+        data, "_http_json",
+        lambda *a, **k: _em_payload([_em_row("600000", "浦发银行", "AStock", "1", "沪A")]))
+    monkeypatch.setattr(data, "_http_text", boom)
+    res = data.search_instruments("600000")
+    assert res["status"] == "ready"
+    assert res["providers"]["tencent"].startswith("failed")
+    assert res["items"][0]["symbol"] == "600000"
+
+
+def test_suggest_limits_are_bounded():
+    import inspect
+    src = inspect.getsource(data)
+    assert "SUGGEST_TIMEOUT_SECS = 4.0" in src  # per-source timeout applies to the actual calls
+    # the catalog paths are gone: every query must be a bounded provider call
+    assert "_stock_catalog" not in src
+    assert "_catalog_frame" not in src
+
+
+def test_limit_is_passed_to_provider_bounded(monkeypatch):
+    seen = {}
+    rows = [_em_row(str(600000 + i), f"n{i}", "AStock", "1", "沪A") for i in range(8)]
+    monkeypatch.setattr(data, "_http_json",
+                        lambda url, params, **k: seen.update(params) or _em_payload(rows[:params["count"]]))
+    monkeypatch.setattr(data, "_http_text", lambda *a, **k: "")
+    res = data.search_instruments("600", limit=4)
+    assert seen["count"] == 5  # bounded floor 5, well under the old full catalogs
+    assert len(res["items"]) <= 4
+
+
+def test_fetch_rejects_unknown_asset_type():
+    with pytest.raises(data.DataError):
+        data.fetch_source({"symbol": "X#1", "market": "cn", "asset_type": "crypto"},
+                          "2024-01-01", "2024-02-01", "daily", "none", ["open", "close"])
+
+
+# ---- sina listed-ETF adapter (recovery-01: eastmoney down, tencent has no
+# ETF adapter; live evidence in docs/recovery-01-plan.md). The stock-like frame
+# below is a SYNTHETIC test fixture mimicking the live-captured sina reply shape. ----
+
+def _sina_synth_frame():
+    """Synthetic fixture in the exact live-captured shape of
+    ak.fund_etf_hist_sina(symbol='sz159399') (recovered live 2026-09-17:
+    381 rows 2025-02-27..2026-09-16, columns date/open/.../amount/postVol/postAmt)."""
+    rows = [
+        {"date": "2025-12-30", "open": 1.001, "high": 1.006, "low": 0.999,
+         "close": 1.006, "volume": 695982432, "amount": 696910071.0,
+         "postVol": float("nan"), "postAmt": float("nan")},
+        {"date": "2026-01-02", "open": 1.007, "high": 1.008, "low": 0.994,
+         "close": 0.997, "volume": 505082425, "amount": 506417752.0,
+         "postVol": float("nan"), "postAmt": float("nan")},
+        {"date": "2026-09-16", "open": 1.001, "high": 1.001, "low": 0.982,
+         "close": 0.992, "volume": 144526400, "amount": 142914764.0,
+         "postVol": 300.0, "postAmt": 298.0},
+    ]
+    df = pd.DataFrame(rows)
+    df.attrs["synthetic"] = True
+    return df
+
+
+def test_sina_etf_serves_159399_with_honest_provenance(monkeypatch):
+    inst = {"symbol": "159399", "market": "cn", "asset_type": "etf"}
+    monkeypatch.setattr(data, "ak",
+                        type("M", (), {"fund_etf_hist_sina":
+                                       staticmethod(lambda **kw: _sina_synth_frame())})())
+    res = data.fetch_source(inst, "2025-12-31", "2026-09-17", "daily", "none")
+    df, endpoint, params = res
+    assert res.provider == "sina"
+    assert endpoint == "fund_etf_hist_sina"
+    assert params == {"symbol": "sz159399"}
+    assert set(df["symbol"]) == {"SZ#159399"}
+    dates = df["date"].tolist()
+    assert dates[0] >= "2025-12-31" and dates[-1] <= "2026-09-17"
+    assert "2026-09-17" not in dates  # no fabricated end-date bar
+    warns = " ".join(df.attrs.get("source_warnings", []))
+    assert "股" in warns and "unadjusted" in warns
+    assert "159399" not in warns  # numbers only, identity preserved
+
+
+def test_sina_rejects_adjustment_and_range_semantics(monkeypatch):
+    inst = {"symbol": "159399", "market": "cn", "asset_type": "etf"}
+    monkeypatch.setattr(data, "ak",
+                        type("M", (), {"fund_etf_hist_sina":
+                                       staticmethod(lambda **kw: _sina_synth_frame())})())
+    with pytest.raises(data.DataError) as e:
+        data.fetch_source(inst, "2025-12-31", "2026-09-17", "daily", "qfq", source="sina")
+    assert e.value.code == "unsupported_adjustment"
+
+
+def test_sina_rejects_stock_and_index():
+    for asset in ("stock", "index"):
+        inst = {"symbol": "600000" if asset == "stock" else "000300",
+                "market": "cn", "asset_type": asset}
+        with pytest.raises(data.DataError) as e:
+            data.fetch_source(inst, "2025-12-31", "2026-09-17", "daily", "none",
+                              source="sina")
+        assert e.value.code == "source_unavailable"
+
+
+def test_auto_etf_falls_back_from_eastmoney_to_sina_honestly(monkeypatch):
+    inst = {"symbol": "159399", "market": "cn", "asset_type": "etf"}
+
+    def dis(*a, **k):
+        raise ConnectionError("RemoteDisconnected('Remote end closed connection')")
+    monkeypatch.setattr(data, "ak",
+                        type("M", (), {"fund_etf_hist_em": staticmethod(dis),
+                                       "fund_etf_hist_sina":
+                                       staticmethod(lambda **kw: _sina_synth_frame())})())
+    res, warns = data.fetch_source_with_warnings(
+        inst, "2025-12-31", "2026-09-17", "daily", "none", source="auto")
+    assert res.provider == "sina"  # actual serving provider, never disguised
+    joined = " ".join(warns)
+    assert "provider_fallback" in joined and "eastmoney" in joined and "sina" in joined
+    assert set(res.df["symbol"]) == {"SZ#159399"}
+
+
+def test_auto_etf_fails_when_no_provider_is_viable(monkeypatch):
+    inst = {"symbol": "159399", "market": "cn", "asset_type": "etf"}
+
+    def dis(*a, **k):
+        raise ConnectionError("RemoteDisconnected")
+    monkeypatch.setattr(data, "ak",
+                        type("M", (), {"fund_etf_hist_em": staticmethod(dis),
+                                       "fund_etf_hist_sina": staticmethod(dis)})())
+    with pytest.raises(data.DataError) as e:
+        data.fetch_source(inst, "2025-12-31", "2026-09-17", "daily", "none", source="auto")
+    assert e.value.code == "provider_unavailable"
+    assert "eastmoney" in str(e.value) and "sina" in str(e.value)
+
+
+def test_tencent_etf_rejection_message_unchanged():
+    inst = {"symbol": "159399", "market": "cn", "asset_type": "etf"}
+    with pytest.raises(data.DataError) as e:
+        data.fetch_source(inst, "2025-12-31", "2026-09-17", "daily", "none",
+                          source="tencent")
+    assert e.value.code == "source_unavailable"
+    assert "tencent source has no listed-ETF daily adapter" in str(e.value)
+
+
+def test_sina_registered_in_explicit_source_surface():
+    assert "sina" in data.SUPPORTED_SOURCES
+
+
+def test_fetch_columns_subset_requested():
+    req = {"symbol": "600000", "market": "cn", "asset_type": "stock"}
+    called = {}
+
+    def fake_hist(symbol, **kw):
+        called["params"] = kw
+        return synthetic_daily("SH#600000")
+    monkeypatch = pytest.MonkeyPatch()
+    monkeypatch.setattr(data, "ak", type("M", (), {"stock_zh_a_hist": staticmethod(fake_hist)})())
+    df, endpoint, params = data.fetch_source(req, "2024-01-01", "2024-02-01",
+                                             "daily", "none", ["open", "close", "volume"])
+    assert endpoint == "stock_zh_a_hist"
+    assert all(c in df.columns for c in ("date", "symbol", "open", "close", "volume"))
+    monkeypatch.undo()