fix: HINA 全类型缺陷修复 — SORT/CSV/ALT 3 个真实缺陷

对抗性全类型测试发现的缺陷和修复:

缺陷1: SORT/MERGE L1 关键词太严格(漏检)
  - 旧: 'SORT ON KEY' / 'MERGE ON KEY'(精确字符串)
  - COBOL 中的真实写法: SORT WORK-FILE ON ASCENDING KEY ...
  - 新: 正则 SORT(?:\s+\S+)?\s+ON\s+(?:ASCENDING|DESCENDING)?KEY

缺陷2: CSV 假阳性(STRING/INSPECT 非CSV也触发)
  - 旧: has_string=True -> CSV合并
  - 新: 要求 has_csv_merge(STRING+逗号分隔)
  - 单纯字符串拼接不再触发 CSV 分类

缺陷3: ALTERNATE RECORD KEY 被 ORGANIZATION IS 覆盖
  - 旧: 文件编成先于替代索引(同确信度先者胜)
  - 新: 替代索引放前面(更具体的分类优先)

回归: 767 passed(0 new failures)
This commit is contained in:
NB-076
2026-06-21 15:51:30 +08:00
parent 4b22c3754e
commit cdba324b5a
5 changed files with 41 additions and 20 deletions
+18 -6
View File
@@ -106,21 +106,33 @@ def resolve_csv_merge_vs_split(features: dict) -> dict:
"""区分 CSV 合并与拆分。
规则:
- STRING 语句存在 → 无换行 (合并, merge)
- INSPECT REPLACING 存在 → 有换行 (拆分, split)
- STRING 存在且含逗号分隔 → 无换行 (合并, merge)
- INSPECT REPLACING 含逗号/改行 → 有换行 (拆分, split)
单纯的 STRING 拼接/INSPECT 计数不触发(容易假阳性)。
"""
has_string = features.get("has_string", False)
has_inspect = features.get("has_inspect", False)
has_csv_merge = features.get("has_csv_merge", False) # 从源码注入
has_csv_split = features.get("has_csv_split", False) # 从源码注入
evidence: list[str] = []
if has_string:
evidence.append("STRING 语句存在 → CSV 合并 (无换行)")
if has_csv_merge:
evidence.append("STRING + 逗号分隔 → CSV 合并 (无换行)")
return {"resolved_type": "CSV合并", "confidence": 0.85, "evidence": evidence}
if has_inspect:
evidence.append("INSPECT REPLACING 存在 → CSV 拆分 (有换行)")
if has_csv_split:
evidence.append("INSPECT REPLACING 含逗号/改行 → CSV 拆分")
return {"resolved_type": "CSV拆分", "confidence": 0.85, "evidence": evidence}
# 兼容旧版:
if has_string:
evidence.append("STRING 存在但无逗号分隔 → 非CSV(低确信度)")
return {"resolved_type": "unknown", "confidence": 0.0, "evidence": evidence}
if has_inspect:
evidence.append("INSPECT 存在但无逗号/改行 → 非CSV(低确信度)")
return {"resolved_type": "unknown", "confidence": 0.0, "evidence": evidence}
evidence.append("既无 STRING 也无 INSPECT REPLACING")
return {"resolved_type": "unknown", "confidence": 0.0, "evidence": evidence}