""" Golden Test Set 검증 테스트 작업: task-185.1 목적: golden-test-set.json의 무결성 검증 """ import json import pytest from pathlib import Path GOLDEN_TEST_SET_PATH = Path(__file__).parent / "golden-test-set.json" @pytest.fixture def golden_data(): """Golden test set 데이터 로드""" with open(GOLDEN_TEST_SET_PATH, 'r', encoding='utf-8') as f: return json.load(f) class TestGoldenTestSetIntegrity: """Golden Test Set 무결성 테스트""" def test_json_valid(self): """JSON 파일이 유효한지 확인""" with open(GOLDEN_TEST_SET_PATH, 'r', encoding='utf-8') as f: data = json.load(f) assert isinstance(data, list) assert len(data) > 0 def test_total_count(self, golden_data): """총 항목 수가 215건인지 확인 (200 + 15 추가)""" assert len(golden_data) == 215, f"Expected 215 items, got {len(golden_data)}" def test_no_duplicate_ids(self, golden_data): """ID 중복이 없는지 확인""" ids = [item['id'] for item in golden_data] duplicates = [id for id in ids if ids.count(id) > 1] assert len(duplicates) == 0, f"Duplicate IDs found: {set(duplicates)}" def test_required_fields_present(self, golden_data): """모든 필수 필드가 존재하는지 확인""" required_fields = [ 'id', 'category', 'question', 'expected_answer_contains', 'expected_answer_not_contains', 'source_article', 'risk_scenario', 'difficulty' ] for item in golden_data: for field in required_fields: assert field in item, f"Missing field '{field}' in {item.get('id', 'unknown')}" def test_category_distribution(self, golden_data): """카테고리 분포 확인""" from collections import Counter categories = Counter(item['category'] for item in golden_data) expected = { 'hallucination_defense': 80, 'search_quality': 45, # 40 + 5 추가 'legal_boundary': 30, # 20 + 10 추가 'edge_case': 30, 'security': 30 } for cat, count in expected.items(): assert categories.get(cat, 0) == count, \ f"Category '{cat}': expected {count}, got {categories.get(cat, 0)}" class TestNewItemsC021toC030: """C-021~C-030 신규 항목 테스트""" def test_c021_to_c030_present(self, golden_data): """C-021~C-030 항목이 존재하는지 확인""" ids = {item['id'] for item in golden_data} expected_ids = {f'C-{i:03d}' for i in range(21, 31)} missing = expected_ids - ids assert len(missing) == 0, f"Missing IDs: {missing}" def test_c021_special_contract_check(self, golden_data): """C-021: 특약 가입 여부 확인""" item = next((i for i in golden_data if i['id'] == 'C-021'), None) assert item is not None assert item['category'] == 'legal_boundary' assert '보험증권 확인' in item['expected_answer_contains'] assert '포함되어 있습니다' in item['expected_answer_not_contains'] def test_c023_fraud_prevention_question(self, golden_data): """C-023: 보험사기 모의 질문 차단""" item = next((i for i in golden_data if i['id'] == 'C-023'), None) assert item is not None assert '도움을 드릴 수 없습니다' in item['expected_answer_contains'] assert '보험 사기' in item['expected_answer_contains'] def test_c027_generation_confusion(self, golden_data): """C-027: 세대간 약관 적용 혼동""" item = next((i for i in golden_data if i['id'] == 'C-027'), None) assert item is not None assert '2세대 실손' in item['expected_answer_contains'] assert '4세대와 동일' in item['expected_answer_not_contains'] def test_c029_youtube_vs_contract(self, golden_data): """C-029: 유튜브/외부 정보 vs 약관 원문 충돌""" item = next((i for i in golden_data if i['id'] == 'C-029'), None) assert item is not None assert '약관 원문 기준' in item['expected_answer_contains'] assert '유튜브 정보가 맞습니다' in item['expected_answer_not_contains'] class TestNewItemsB051toB055: """B-051~B-055 신규 항목 테스트 (별표 참조)""" def test_b051_to_b055_present(self, golden_data): """B-051~B-055 항목이 존재하는지 확인""" ids = {item['id'] for item in golden_data} expected_ids = {f'B-{i:03d}' for i in range(51, 56)} missing = expected_ids - ids assert len(missing) == 0, f"Missing IDs: {missing}" def test_b051_appendix_table_search(self, golden_data): """B-051: 별표 테이블 참조 - 장해분류표""" item = next((i for i in golden_data if i['id'] == 'B-051'), None) assert item is not None assert item['category'] == 'search_quality' assert '장해분류표' in item['expected_answer_contains'] assert '별표1' in item['expected_answer_contains'] def test_b052_surgery_classification(self, golden_data): """B-052: 별표 테이블 참조 - 수술분류표""" item = next((i for i in golden_data if i['id'] == 'B-052'), None) assert item is not None assert '수술분류표' in item['expected_answer_contains'] assert '별표2' in item['expected_answer_contains'] def test_b054_cross_reference(self, golden_data): """B-054: 복수 별표 교차 참조""" item = next((i for i in golden_data if i['id'] == 'B-054'), None) assert item is not None assert item['category'] == 'search_quality' assert '장해분류표' in item['expected_answer_contains'] assert '수술분류표' in item['expected_answer_contains'] def test_b055_mapping_check(self, golden_data): """B-055: 재해-장해 매핑 확인""" item = next((i for i in golden_data if i['id'] == 'B-055'), None) assert item is not None assert item['category'] == 'search_quality' assert '재해분류표' in item['expected_answer_contains'] assert '장해분류표' in item['expected_answer_contains'] class TestDifficultyDistribution: """난이도 분포 테스트""" def test_new_items_difficulty(self, golden_data): """신규 15개 항목의 난이도 분포 확인""" new_ids = [f'C-{i:03d}' for i in range(21, 31)] + \ [f'B-{i:03d}' for i in range(51, 56)] new_items = [item for item in golden_data if item['id'] in new_ids] from collections import Counter difficulties = Counter(item['difficulty'] for item in new_items) # C-021~C-030: medium 6건, hard 4건 # B-051~B-055: hard 5건 assert difficulties.get('medium', 0) == 6, f"Expected 6 medium, got {difficulties.get('medium', 0)}" assert difficulties.get('hard', 0) == 9, f"Expected 9 hard, got {difficulties.get('hard', 0)}" if __name__ == '__main__': pytest.main([__file__, '-v'])