c9f69d9d6d
范围（不含 design-story-foundation、docs/、humanization/README.md 等进行中改动）:
1. 新增 harness/ 控制平面
- skill_harness.py 静态审计：32 个运行时 Skill 的 frontmatter/manifest/文档污染，当前 0 问题
- run_selected.py 选择性执行器：manifest 与磁盘一一对账、依赖阻断、
空跑与 skip-only 失败关闭、AST 测试形状门
- manifests/skills.json：32 个 Skill 的合同责任方与协作领域登记
- manifests/test-inventory.json：81 个测试资产登记
- specs/skill-testing.md 与 README.md：测试分层、证据边界与 harness 职责
2. 实现测试从 .claude/skills/*/scripts/ 迁至 tests/skills/<skill>/
- 71 个测试文件迁移并修复项目根与临时目录运行导入
- 数据库触发器测试宽泛异常收窄为 psycopg.errors.RaiseException
- 抽取离线大测试拆出真实 PG smoke（默认阻断，不计入离线通过）
- 抽取 presence 去重边界拆出独立测试：493 + 78 = 571 项检查不变
3. 运行时文档清理
- 13 个 SKILL.md 移除自测/离线验证段落、测试命令与测试文件事实源表述，
只保留运行时合同；业务运行合同、额度、授权与离线模式均保留
4. SoT 同步
- AGENTS.md：新增 Skill 领域索引（7 个合同责任方分组，覆盖 32 个运行时 Skill）
- 领域 07：测试入口改由 harness/manifests/ 登记，SKILL.md 不承载测试命令
- humanization 覆盖矩阵：活动测试路径同步迁移
验证证据: harness 自测 15 项 + runner 自测 13 项通过；静态审计 32 Skill / 0 问题；
73 个非数据库测试通过；8 个集成条目中 6 个 PostgreSQL 项被依赖门明确阻断；
py_compile 与 git diff --check 通过。未连接 PostgreSQL、网络、真实模型或额度。
已知边界: 真正 skill_behavior_eval 仍为 0，尚未验证任何 Skill 自然语言行为；
evaluate-frozen-replay 的 raw 存储边界冲突留待单独治理。
103 lines
4.6 KiB
Python
103 lines
4.6 KiB
Python
#!/usr/bin/env python3
|
||
"""技能 1 定基线离线测试：结构合同、grounding 门、人工确认强制。"""
|
||
import pathlib
|
||
import sys
|
||
import tempfile
|
||
import unittest
|
||
from unittest.mock import patch
|
||
|
||
PROJECT_ROOT = pathlib.Path(__file__).resolve().parents[3]
|
||
SCRIPT_DIR = PROJECT_ROOT / ".claude" / "skills" / "establish-voice-baseline" / "scripts"
|
||
sys.path.insert(0, str(SCRIPT_DIR))
|
||
|
||
import establish_voice_baseline as base # noqa: E402
|
||
|
||
SOURCE_TEXT = "老周敲了敲桌角：「我说小子，茶要凉了。」青云城的雨说来就来。"
|
||
|
||
LEDGER = {
|
||
"schema_version": "voice-baseline-v1",
|
||
"work_ref": "synthetic:demo",
|
||
"narrator": {"sentence_habits": ["短句收束"], "punctuation_habits": []},
|
||
"characters": {"老周": {"verbal_tics": ["我说小子"], "sample_lines": ["我说小子，茶要凉了。"]}},
|
||
"untouchable_verbal_tics": {"老周": ["我说小子"]},
|
||
"protected_spans": ["青云城的雨说来就来"],
|
||
"blacklist": ["值得注意的是"],
|
||
}
|
||
|
||
|
||
class BaselineContractTest(unittest.TestCase):
|
||
def test_draft_output_can_feed_confirm(self):
|
||
with tempfile.TemporaryDirectory() as tmp:
|
||
root = pathlib.Path(tmp)
|
||
source = root / "source.txt"
|
||
source.write_text(SOURCE_TEXT, encoding="utf-8")
|
||
draft_output = root / "draft.json"
|
||
self.assertEqual(
|
||
0,
|
||
base.main([
|
||
"draft", "--work-ref", "synthetic:demo", "--text-file", str(source),
|
||
"--output", str(draft_output), "--offline",
|
||
]),
|
||
)
|
||
with patch.object(base, "collect_ledger", return_value={"version": 1}) as collect:
|
||
code = base.main([
|
||
"confirm", "--ledger-file", str(draft_output),
|
||
"--work-ref", "synthetic:demo", "--text-file", str(source),
|
||
"--reviewer", "qingse", "--output", str(root / "confirmed.json"),
|
||
])
|
||
self.assertEqual(code, 0)
|
||
collect.assert_called_once()
|
||
self.assertEqual(collect.call_args.args[0]["schema_version"], "voice-baseline-v1")
|
||
|
||
def test_valid_ledger_passes_grounding_and_persists(self):
|
||
with patch.object(base, "persist_baseline", return_value={"version": 1}) as persist:
|
||
result = base.collect_ledger(LEDGER, work_ref="synthetic:demo",
|
||
source_texts=[SOURCE_TEXT], reviewer="qingse")
|
||
self.assertEqual(result["version"], 1)
|
||
persist.assert_called_once()
|
||
|
||
def test_ungrounded_tic_is_rejected(self):
|
||
bad = dict(LEDGER, untouchable_verbal_tics={"老周": ["根本不存在的口癖"]})
|
||
with patch.object(base, "persist_baseline") as persist:
|
||
with self.assertRaisesRegex(base.BaselineContractError, "grounding"):
|
||
base.collect_ledger(bad, work_ref="synthetic:demo",
|
||
source_texts=[SOURCE_TEXT], reviewer="qingse")
|
||
persist.assert_not_called()
|
||
|
||
def test_ungrounded_protected_span_is_rejected(self):
|
||
bad = dict(LEDGER, protected_spans=["不在正文里的句子"])
|
||
with self.assertRaisesRegex(base.BaselineContractError, "grounding"):
|
||
base.collect_ledger(bad, work_ref="synthetic:demo",
|
||
source_texts=[SOURCE_TEXT], reviewer="qingse")
|
||
|
||
def test_reviewer_is_required(self):
|
||
# persist_baseline 的确认人门在连库之前，可直接测
|
||
with self.assertRaisesRegex(base.BaselineContractError, "人工确认"):
|
||
base.persist_baseline(LEDGER, source_text=SOURCE_TEXT, reviewer="")
|
||
|
||
def test_direct_persist_rechecks_grounding_before_db(self):
|
||
with self.assertRaisesRegex(base.BaselineContractError, "grounding"):
|
||
base.persist_baseline(
|
||
dict(LEDGER, protected_spans=["不在正文里的句子"]),
|
||
source_text=SOURCE_TEXT, reviewer="qingse",
|
||
)
|
||
|
||
def test_schema_version_is_enforced(self):
|
||
bad = dict(LEDGER, schema_version="nope")
|
||
with self.assertRaisesRegex(base.BaselineContractError, "schema_version"):
|
||
base.validate_ledger(bad, work_ref="synthetic:demo")
|
||
|
||
def test_work_ref_mismatch_is_enforced(self):
|
||
with self.assertRaisesRegex(base.BaselineContractError, "work_ref"):
|
||
base.validate_ledger(LEDGER, work_ref="synthetic:other")
|
||
|
||
def test_missing_narrator_is_enforced(self):
|
||
bad = dict(LEDGER)
|
||
del bad["narrator"]
|
||
with self.assertRaisesRegex(base.BaselineContractError, "narrator"):
|
||
base.validate_ledger(bad, work_ref="synthetic:demo")
|
||
|
||
|
||
if __name__ == "__main__":
|
||
unittest.main()
|