fixes and refactor

tests/test_search_api.py

@@ -22,6 +22,7 @@ from tests.helpers import assert_validation_error  # noqa: E402
 from app.api.financial import LedgerCreate  # noqa: E402
 from app.database.base import SessionLocal  # noqa: E402
 from app.models.qdro import QDRO  # noqa: E402
+from app.config import settings  # noqa: E402
 
 
 @pytest.fixture(scope="module")
@@ -37,6 +38,8 @@ def client():
     app.dependency_overrides[get_current_user] = lambda: _User()
 
     try:
+        # Disable cache for search API tests unless explicitly testing caching
+        settings.cache_enabled = False
         yield TestClient(app)
     finally:
         app.dependency_overrides.pop(get_current_user, None)
@@ -284,3 +287,181 @@ def test_global_search_highlights_mixed_case_for_customer_file_qdro(client: Test
     assert q is not None and isinstance(q.get("highlight"), str)
     assert "<strong>" in q["highlight"]
     assert f"<strong>{token_mixed}</strong>" in q["highlight"]
+
+
+def test_file_search_whole_words_and_exact_phrase(client: TestClient):
+    token = f"FW-{uuid.uuid4().hex[:6]}"
+    owner_id = _create_customer(client, f"Owner-{token}")
+    f_exact = _create_file(client, owner_id, regarding_token="The apple pie is fresh")
+    f_plural = _create_file(client, owner_id, regarding_token="The apple pies are fresh")
+
+    # whole_words=True should match 'pie' but not 'pies'
+    payload = {
+        "query": "pie",
+        "search_types": ["file"],
+        "whole_words": True,
+        "limit": 50,
+    }
+    resp = client.post("/api/search/advanced", json=payload)
+    assert resp.status_code == 200
+    results = resp.json()["results"]
+    ids = {r["id"] for r in results}
+    assert f_exact in ids
+    assert f_plural not in ids
+
+    # exact_phrase should match the exact wording only
+    payload = {
+        "query": "apple pie",
+        "search_types": ["file"],
+        "exact_phrase": True,
+        "limit": 50,
+    }
+    resp = client.post("/api/search/advanced", json=payload)
+    assert resp.status_code == 200
+    results = resp.json()["results"]
+    ids = {r["id"] for r in results}
+    assert f_exact in ids
+    assert f_plural not in ids
+
+    # default (substring) matching should include both
+    payload = {
+        "query": "pie",
+        "search_types": ["file"],
+        "limit": 50,
+    }
+    resp = client.post("/api/search/advanced", json=payload)
+    assert resp.status_code == 200
+    results = resp.json()["results"]
+    ids = {r["id"] for r in results}
+    assert f_exact in ids and f_plural in ids
+
+
+def test_ledger_search_whole_words(client: TestClient):
+    token = f"LW-{uuid.uuid4().hex[:6]}"
+    # Create a file for ledger linkage
+    owner_id = _create_customer(client, f"Owner-{token}")
+    file_no = _create_file(client, owner_id, regarding_token=token)
+
+    # Ledger entries: 'retainer' vs 'retained'
+    resp = client.post(
+        "/api/financial/ledger/",
+        json=LedgerCreate(
+            file_no=file_no,
+            date=date.today().isoformat(),
+            t_code="NOTE",
+            t_type="2",
+            empl_num="E01",
+            quantity=0.0,
+            rate=0.0,
+            amount=0.0,
+            billed="N",
+            note="retainer fee approved",
+        ).model_dump(mode="json"),
+    )
+    assert resp.status_code == 200
+
+    resp = client.post(
+        "/api/financial/ledger/",
+        json=LedgerCreate(
+            file_no=file_no,
+            date=date.today().isoformat(),
+            t_code="NOTE",
+            t_type="2",
+            empl_num="E01",
+            quantity=0.0,
+            rate=0.0,
+            amount=0.0,
+            billed="N",
+            note="retained amount on file",
+        ).model_dump(mode="json"),
+    )
+    assert resp.status_code == 200
+
+    payload = {
+        "query": "retainer",
+        "search_types": ["ledger"],
+        "whole_words": True,
+        "limit": 50,
+    }
+    resp = client.post("/api/search/advanced", json=payload)
+    assert resp.status_code == 200
+    results = resp.json()["results"]
+    # Should contain the entry with 'retainer fee approved' and exclude 'retained amount on file'
+    texts = [r.get("description", "") for r in results]
+    assert any("retainer fee approved" in t for t in texts)
+    assert all("retained amount on file" not in t for t in texts)
+
+
+def test_qdro_search_whole_words_and_exact_phrase(client: TestClient):
+    token = f"QW-{uuid.uuid4().hex[:6]}"
+    owner_id = _create_customer(client, f"Owner-{token}")
+    file_no = _create_file(client, owner_id, regarding_token=token)
+    q1 = _create_qdro_with_form_name(file_no, form_name="Order for benefit under plan")
+    q2 = _create_qdro_with_form_name(file_no, form_name="Order benefiting alternate payee")
+
+    # whole_words=True should match 'benefit' but not 'benefiting'
+    payload = {
+        "query": "benefit",
+        "search_types": ["qdro"],
+        "whole_words": True,
+        "limit": 50,
+    }
+    resp = client.post("/api/search/advanced", json=payload)
+    assert resp.status_code == 200
+    results = resp.json()["results"]
+    ids = {r["id"] for r in results}
+    assert q1 in ids
+    assert q2 not in ids
+
+    # exact_phrase should only match the precise phrase
+    payload = {
+        "query": "Order for benefit",
+        "search_types": ["qdro"],
+        "exact_phrase": True,
+        "limit": 50,
+    }
+    resp = client.post("/api/search/advanced", json=payload)
+    assert resp.status_code == 200
+    results = resp.json()["results"]
+    ids = {r["id"] for r in results}
+    assert q1 in ids
+    assert q2 not in ids
+
+
+def test_advanced_facets_include_state_and_transaction_type(client: TestClient):
+    token = f"FAC-{uuid.uuid4().hex[:6]}"
+    # Ensure at least one TX customer
+    _ = _create_customer(client, f"Facet-{token}")
+    # Ensure at least one ledger with t_type '2'
+    owner_id = _create_customer(client, f"Owner-{token}")
+    file_no = _create_file(client, owner_id, regarding_token=token)
+    resp = client.post(
+        "/api/financial/ledger/",
+        json=LedgerCreate(
+            file_no=file_no,
+            date=date.today().isoformat(),
+            t_code="NOTE",
+            t_type="2",
+            empl_num="E01",
+            quantity=0.0,
+            rate=0.0,
+            amount=0.0,
+            billed="N",
+            note="Fee for facets token",
+        ).model_dump(mode="json"),
+    )
+    assert resp.status_code == 200
+
+    # Query can be empty; we'll aggregate facets across returned results
+    payload = {
+        "search_types": ["customer", "ledger"],
+        "limit": 200,
+    }
+    resp = client.post("/api/search/advanced", json=payload)
+    assert resp.status_code == 200
+    data = resp.json()
+    facets = data.get("facets", {})
+    assert "state" in facets and isinstance(facets["state"], dict)
+    assert any(k in ("TX", "Tx", "tx") for k in facets["state"].keys())
+    assert "transaction_type" in facets and isinstance(facets["transaction_type"], dict)
+    assert "2" in facets["transaction_type"] or 2 in facets["transaction_type"]