fix: subtype resolver + comprehensive matching program test

Fix 4 remaining defects found by adversarial testing:
1. MT03 N:1 → subtype corrected to N:1 (key suffix -M/-T heuristic)
2. MT32 混合 → subtype added (項目チェック programs with WS-PREV-KEY)
3. MT33 混合异键 → WS-ALT-KEY detection → 混合(异键)
4. MT18/MT19 → subtype M:N (correct: static cannot distinguish M:N→M vs M:N→N)

Also expand subtype resolver scope: now also processes 項目チェック
classified programs with matching-like characteristics (WS-PREV-KEY),
not just マッチング.

New test: test_matching_programs.py — 10 parametrized tests covering
all 4 dimensions (category, subtype, branches, files) for every
matching program. Known limitation documented: MT18 vs MT19
requires runtime data for M:N→M vs M:N→N distinction.

Regression: 755 passed (10 new, 0 failures).
This commit is contained in:
NB-076
2026-06-21 13:40:58 +08:00
parent 6b3f526b80
commit a5939e6722
2 changed files with 103 additions and 5 deletions
@@ -0,0 +1,80 @@
"""专项测试 — 匹配程序完整识别 (10 个程序 × 4 维度)
验证所有 10 个匹配程序在以下维度上的正确性:
1. 分类正确(マッチング/二段階/項目チェック)
2. 子类型正确(1:1/1:N/N:1/M:N/二段階/混合)
3. 分支检测正确
4. 文件数检测正确
已知缺陷(静态分析固有限制):
- MT18 (M:N→M) vs MT19 (M:N→N): 运行时行为区分,静态都输出 M:N
"""
from pathlib import Path
import pytest
from cobol_testgen import extract_structure
from hina.pipeline import classify_program
FIXTURES = Path(__file__).parents[3] / "test-data" / "cobol" / "category_matching"
# (filename, expected_category, expected_subtype, min_branches, min_files)
MATCHING_TESTS = [
("MT01_1TO1.cbl", "マッチング", "1:1", 4, 2),
("MT02_1TON.cbl", "マッチング", "1:N", 4, 2),
("MT03_NTO1.cbl", "マッチング", "N:1", 4, 2),
("MT16_TWO_STAGE_1TO1.cbl", "二段階マッチング", "二段階", 4, 3),
("MT17_TWO_STAGE_NTO1.cbl", "二段階マッチング", "二段階", 4, 3),
("MT18_MN_TO_M.cbl", "マッチング", "M:N", 4, 2),
("MT19_MN_TO_N.cbl", "マッチング", "M:N", 4, 2),
("MT20_MN_TO_MXN.cbl", "マッチング", "M:N→MxN", 2, 3),
("MT32_MIXED_SAME_KEY.cbl", "項目チェック(重複含む)", "混合", 4, 2),
("MT33_MIXED_DIFF_KEY.cbl", "マッチング", "混合(异键)", 4, 2),
]
@pytest.mark.parametrize(
"filename,exp_cat,exp_subtype,min_br,min_fl",
MATCHING_TESTS,
ids=[t[0].replace('.cbl','') for t in MATCHING_TESTS],
)
def test_matching_classification(filename, exp_cat, exp_subtype, min_br, min_fl):
"""匹配程序分类 + 子类型验证"""
path = FIXTURES / filename
assert path.exists(), f"Missing: {path}"
src = path.read_text("utf-8")
# 1. extract_structure must not crash
struct = extract_structure(src)
assert struct is not None
# 2. Branch count meets minimum
assert struct["total_branches"] >= min_br, (
f"{filename}: expected >= {min_br} branches, got {struct['total_branches']}"
)
# 3. File count meets minimum
assert struct["file_count"] >= min_fl, (
f"{filename}: expected >= {min_fl} files, got {struct['file_count']}"
)
# 4. classify_program must not crash
result = classify_program(src)
assert result is not None
assert result["confidence"] > 0
# 5. Category must match (中文/日文编码问题用精确匹配)
assert result["category"] == exp_cat, (
f"{filename}: expected category '{exp_cat}', got '{result['category']}'"
)
# 6. Subtype must match
actual_st = result.get("subtype", "-")
assert actual_st == exp_subtype, (
f"{filename}: expected subtype '{exp_subtype}', got '{actual_st}'"
)
# 7. Must NOT be fallback
assert result["method"] in ("rule_engine", "keyword"), (
f"{filename}: method is '{result['method']}' (should be rule_engine or keyword)"
)