from rag.condition_evaluation import (
    ConditionEvaluationOutcome,
    summarize_condition_outcomes,
)


def test_condition_metrics_are_separate_and_unsupported_drugs_are_counted():
    rows = [
        ConditionEvaluationOutcome(
            case_id="hta",
            expected_intent="condition_to_drug",
            actual_intent="condition_to_drug",
            expected_condition="tăng huyết áp",
            actual_condition="tăng huyết áp",
            expected_clarification=False,
            actual_clarification=False,
            expected_relation="indication",
            actual_relation="indication",
            expected_drug_ids=("a", "b"),
            retrieved_drug_ids=("a", "b"),
            generated_drug_ids=("a", "d"),
            retrieved_section_keys=("chi_dinh", "chi_dinh"),
            citation_validity=(True, False),
            grounded_claims=(True, False),
            expected_patient_fields=(("renal.stage", "G4"),),
            actual_patient_fields=(("renal.stage", "G4"),),
            expected_safety_facets=("renal", "interaction"),
            retrieved_safety_facets=("renal",),
        )
    ]

    metrics = summarize_condition_outcomes(rows)

    assert metrics["intent_accuracy"] == 1.0
    assert metrics["section_correctness"] == 1.0
    assert metrics["unsupported_drug_rate"] == 0.5
    assert metrics["citation_correctness"] == 0.5
    assert metrics["patient_context_extraction_accuracy"] == 1.0
    assert metrics["safety_evidence_retrieval_accuracy"] == 0.5