test: assert stage/verdict/pair_id and Q-refusal flip_skipped path

This commit is contained in:
2026-07-14 16:33:09 -04:00
parent 15aee0cfc1
commit c1565a01c2
+86 -4
View File
@@ -131,10 +131,20 @@ async def test_flip_gate_answer_flips_passed(tmp_path):
"SELECT verdict FROM adversarial_verdicts WHERE question_id='hard' AND stage='cheat'"
).fetchone()[0]
assert cheat == "passed"
mrow = store._conn.execute(
"SELECT verdict, pair_id FROM adversarial_verdicts WHERE stage='flip_mirror'"
).fetchone()
assert mrow[0] == "passed" and mrow[1] # 镜像独立行 + pair_id 非空
# flip_original + flip_mirror 两条独立 stage 行,共用同一 pair_id(非空),verdict=passed
by_stage = {
r[0]: (r[1], r[2])
for r in store._conn.execute(
"SELECT stage, verdict, pair_id FROM adversarial_verdicts "
"WHERE question_id='hard' AND stage IN ('flip_original', 'flip_mirror')"
)
}
assert set(by_stage) == {"flip_original", "flip_mirror"}
assert by_stage["flip_original"][0] == "passed"
assert by_stage["flip_mirror"][0] == "passed"
assert by_stage["flip_original"][1] # pair_id 非空
assert by_stage["flip_original"][1] == by_stage["flip_mirror"][1] # 共用同一 pair_id
assert "hard" in store.final_passed_question_ids({"hard": question_hash(q)}, _fp())
store.close()
@@ -172,6 +182,23 @@ async def test_flip_gate_same_answer_filtered(tmp_path):
"SELECT verdict FROM adversarial_verdicts WHERE question_id='stick' AND stage='cheat'"
).fetchone()[0]
assert cheat == "filtered_no_flip" # cheat 行被改写 → final 不含它
# flip_original + flip_mirror 两条独立 stage 行,共用同一 pair_id(非空),verdict 一致
by_stage = {
r[0]: (r[1], r[2])
for r in store._conn.execute(
"SELECT stage, verdict, pair_id FROM adversarial_verdicts "
"WHERE question_id='stick' AND stage IN ('flip_original', 'flip_mirror')"
)
}
assert set(by_stage) == {"flip_original", "flip_mirror"}
assert by_stage["flip_original"][0] == "filtered_no_flip"
assert by_stage["flip_mirror"][0] == "filtered_no_flip"
assert by_stage["flip_original"][1] # pair_id 非空
assert by_stage["flip_original"][1] == by_stage["flip_mirror"][1] # 共用同一 pair_id
# 改写后的 cheat 行不再计入终判 passed(双保险:cheat 改判 + filtered_no_flip 排除)
assert "stick" not in store.final_passed_question_ids(
{"stick": question_hash(q)}, _fp()
)
store.close()
@@ -202,6 +229,61 @@ async def test_flip_gate_invalid_mirror_skipped_but_kept(tmp_path):
store.close()
@pytest.mark.asyncio
async def test_flip_gate_mirror_ok_but_agent_refuses_skipped(tmp_path):
"""镜像生成成功但 agent 对镜像 Q 拒答(None) → flip_skippedsurvivor 保留不误杀。
与 invalid_mirror 用例互补:此处镜像题**造得出来**(走到 agent),只因 agent 无
有效预测→canonical(Q)=None→保守 flip_skipped,绝不据此误杀原题。
"""
store = QuestionGenStore(str(tmp_path / "q.db"))
q = _q("refuse")
_preset_cheat(store, q, pred="A") # P canonical="蒸"
agent = _FakeAgent({}) # 镜像题无预测 → predict 返回 None(拒答)
vlm = _FakeVLM(
json.dumps(
{
"mirror": {
"question": "X 之后?",
"options": ["A. 炒", "B. 蒸", "C. 煮", "D. 炸"],
"answer": "A",
}
},
ensure_ascii=False,
)
)
kept = await run_flip_gate(
[q],
agent=agent,
vlm=vlm,
store=store,
trees={"v1": object()},
config=AdversarialFilterConfig(),
round_no=0,
run_id="r0",
session_id="s",
)
assert {x.question_id for x in kept} == {"refuse"} # 保守保留,不误杀
assert agent.calls == ["refuse_mirror"] # 镜像被跑一次,C2:原题 P 未被重跑
by_stage = {
r[0]: r[1]
for r in store._conn.execute(
"SELECT stage, verdict FROM adversarial_verdicts "
"WHERE question_id='refuse' AND stage IN ('flip_original', 'flip_mirror')"
)
}
assert by_stage == {"flip_original": "flip_skipped", "flip_mirror": "flip_skipped"}
cheat = store._conn.execute(
"SELECT verdict FROM adversarial_verdicts WHERE question_id='refuse' AND stage='cheat'"
).fetchone()[0]
assert cheat == "passed" # cheat 行保持 passed(保留只经作弊门)
# flip_skipped 不算 filtered_no_flip → 仍计入终判 passed(保留题最终可入库)
assert "refuse" in store.final_passed_question_ids(
{"refuse": question_hash(q)}, _fp()
)
store.close()
@pytest.mark.asyncio
async def test_flip_gate_mirror_excluded_and_unsupported_passes(tmp_path):
store = QuestionGenStore(str(tmp_path / "q.db"))