Remove chat rate limit

apps/ai-service/tests/test_clinical_condition_flow.py

@@ -15,6 +15,7 @@ from rag.clinical import (
     RenalContext,
 )
 from rag.models import Evidence, SourceRef
+from rag.sections import SECTION_PHRASES
 from rag.understanding import (
     LlmQueryUnderstander,
     QueryFrame,
@@ -173,12 +174,57 @@ def test_named_drug_safety_and_purpose_cues_override_noisy_relation_frames():
         replace(noisy, drugs=("paracetamol_acetaminophen",)),
         "Paracetamol có tác dụng gì?",
     )
+    golden_contraindication = _apply_named_drug_cues(
+        replace(noisy, drugs=("paracetamol_acetaminophen",)),
+        "Chống chỉ định của Paracetamol là gì?",
+    )
+    golden_contraindication_with_omitted_llm_drug = _apply_named_drug_cues(
+        replace(noisy, drugs=()),
+        "Chống chỉ định của Paracetamol là gì?",
+        resolved_turn_drug="paracetamol_acetaminophen",
+    )
+    reverse_contraindication = _apply_named_drug_cues(
+        replace(noisy, drugs=("warfarin",)),
+        "Thuốc nào chống chỉ định ở bệnh nhân đang dùng warfarin?",
+    )
 
     assert safety.turn_type == "drug_attribute"
     assert safety.attribute == "chong_chi_dinh"
     assert safety.needs_clarify is False
     assert purpose.turn_type == "drug_to_condition"
     assert purpose.attribute == "chi_dinh"
+    assert golden_contraindication.turn_type == "drug_attribute"
+    assert golden_contraindication.attribute == "chong_chi_dinh"
+    assert golden_contraindication.condition_relation == ConditionRelation.CONTRAINDICATION
+    assert golden_contraindication_with_omitted_llm_drug.turn_type == "drug_attribute"
+    assert golden_contraindication_with_omitted_llm_drug.drugs == (
+        "paracetamol_acetaminophen",
+    )
+    assert reverse_contraindication.turn_type == "condition_relation"
+
+
+def test_owned_section_phrase_routes_every_monograph_section_without_llm_help():
+    noisy = QueryFrame(
+        turn_type="condition_relation",
+        condition_relation=ConditionRelation.CONTRAINDICATION,
+        needs_clarify=True,
+        clarify_reason="Cần làm rõ",
+    )
+
+    for section_key, phrases in SECTION_PHRASES.items():
+        phrase = phrases[0]
+        corrected = _apply_named_drug_cues(
+            noisy,
+            f"{phrase} của Paracetamol là gì?",
+            resolved_turn_drug="paracetamol_acetaminophen",
+            resolved_section_key=section_key,
+            resolved_section_phrase=phrase,
+        )
+
+        assert corrected.turn_type == "drug_attribute", section_key
+        assert corrected.drugs == ("paracetamol_acetaminophen",), section_key
+        assert corrected.attribute == section_key
+        assert corrected.needs_clarify is False
 
 
 def test_ambiguous_condition_clarifies_before_retrieval():

apps/ai-service/tests/test_understanding.py

@@ -86,6 +86,30 @@ def test_drug_id_in_exact_underscore_form_resolves():
     assert frame.unknown_drugs == ()
 
 
+def test_golden_named_drug_section_overrides_a_misclassified_relation_frame():
+    """Regression for the production failure observed through the real UI."""
+    understander = LlmQueryUnderstander(_FixedLlm({
+        "turn_type": "condition_relation",
+        "drugs": [],
+        "unknown_drugs": [],
+        "attribute": None,
+        "population": None,
+        "weight_kg": None,
+        "age_text": None,
+        "indication": None,
+        "condition_relation": "contraindication",
+        "needs_clarify": False,
+        "clarify_reason": None,
+    }), CATALOG, RESOLVER)
+
+    frame = understander.understand("Chống chỉ định của Paracetamol là gì?")
+
+    assert frame.turn_type == "drug_attribute"
+    assert frame.drugs == ("paracetamol_acetaminophen",)
+    assert frame.attribute == "chong_chi_dinh"
+    assert frame.needs_clarify is False
+
+
 def test_exact_candidate_does_not_repeat_the_catalog_wide_fuzzy_scan():
     resolver = _FakeResolver({"metformin": "metformin"})
     understander = LlmQueryUnderstander(_FixedLlm({
@@ -220,6 +244,11 @@ def test_quick_replies_are_parsed_when_the_model_offers_them():
 
 
 def test_quick_replies_are_dynamic_but_bounded_before_becoming_ui_chips():
+    # 19 distinct valid entries (after "  Người   lớn  " / "người lớn" dedup
+    # and the non-string 12 are dropped) so the 18-item cap — one per
+    # monograph section, see rag/sections.py SECTION_ORDER — still trims
+    # the last one, not just the old 4-item cap.
+    extra = [f"Lựa chọn {i}" for i in range(6, 20)]
     understander = LlmQueryUnderstander(_FixedLlm({
         "turn_type": "dosing_calc", "drugs": ["paracetamol_acetaminophen"],
         "unknown_drugs": [], "attribute": None, "population": None,
@@ -228,14 +257,17 @@ def test_quick_replies_are_dynamic_but_bounded_before_becoming_ui_chips():
         "quick_replies": [
             "  Người   lớn  ", "người lớn", "Trẻ em", 12,
             "Phụ nữ có thai", "Người cao tuổi", "Lựa chọn thứ năm",
+            *extra,
         ],
     }), CATALOG, RESOLVER)
 
     frame = understander.understand("liều paracetamol")
 
-    assert frame.quick_replies == (
-        "Người lớn", "Trẻ em", "Phụ nữ có thai", "Người cao tuổi"
+    assert len(frame.quick_replies) == 18
+    assert frame.quick_replies[:5] == (
+        "Người lớn", "Trẻ em", "Phụ nữ có thai", "Người cao tuổi", "Lựa chọn thứ năm"
     )
+    assert "Lựa chọn 19" not in frame.quick_replies
 
 
 def test_string_false_does_not_turn_into_a_clarification():